User’s guide¶
Introduction: Welcome to EmPy!¶
EmPy is a powerful, robust and mature
templating system for inserting Python code in template text. EmPy
takes a source document, processes it, and produces output. This is
accomplished via expansions, which are signals to the EmPy system
where to act and are indicated with markup. Markup is set off by a
customizable prefix (by default the at sign, @
). EmPy can expand
arbitrary Python expressions, statements and control structures in
this way, as well as a variety of additional special forms. The
remaining textual data is sent to the output, allowing Python to be
used in effect as a markup language.
EmPy also supports hooks, which can intercept and modify the behavior of a running interpreter; diversions, which allow recording and playback; filters, which are dynamic and can be chained together; and a dedicated user-customizable callback markup. The system is highly configurable via command line options, configuration files, and environment variables. An extensive API is also available for embedding EmPy functionality in your own Python programs.
EmPy also has a supplemental library for additional non-essential
features (emlib
), a documentation building library used to create
this documentation (emdoc
), and an extensive help system (emhelp
)
which can be queried from the command line with the main executable
em.py
(-h/--help
, -H/--topics=TOPICS
). The base EmPy
interpreter can function with only the em.py
/em
file/module
available.
EmPy can be used in a variety of roles, including as a templating system, a text processing system (preprocessing and/or postprocessing), a simple macro processor, a frontend for a content management system, annotating documents, for literate programming, as a souped-up text encoding converter, a text beautifier (with macros and filters), and many other purposes.
Markup overview¶
Expressions are embedded in text with the @(...)
notation;
variations include conditional expressions with @(...?...!...)
and
the ability to handle thrown exceptions with @(...$...)
. As a
shortcut, simple variables and expressions can be abbreviated as
@variable
, @object.attribute
, @sequence[index]
,
@function(arguments...)
, @function{markup}{...}
and
combinations. Full-fledged statements are embedded with @{...}
.
Control flow in terms of conditional or repeated expansion is
available with @[...]
. A @
followed by any whitespace character
(including a newline) expands to nothing, allowing string
concatenations and line continuations. Line comments are indicated
with @#...
including the trailing newline. @*...*
allows inline
comments. Escapes are indicated with @\...
; diacritics with
@^...
; icons with @|...
; and emoji with @:...:
. @%...
,
@%!...
, @%%...%%
and @%%!...%%
indicate “significators,”
which are distinctive forms of variable assignment intended to specify
document metadata in a format easy to parse externally. In-place
expressions are specified with @$...$...$
. Context name and line
number changes can be made with @?...
and @!...
, respectively.
A set of markups (@((...))
, @[[...]]
, @{{...}}
, @<...>
)
are customizable by the user and can be used for any desired purpose.
@`...`
allows literal escaping of any EmPy markup. And
finally, a @@
sequence (the prefix repeated once) expands to a
single literal at sign.
The prefix defaults to @
but can be changed with
the command line option -p/--prefix=CHAR
(environment variable: EMPY_PREFIX
, configuration variable: prefix
).
Getting the software¶
The current version of EmPy is 4.1.
The official URL for this Web site is http://www.alcyone.com/software/empy/.
The latest version of the software is available in a tarball here:
http://www.alcyone.com/software/empy/empy-latest.tar.gz.
The software can be installed through PIP via this shell command:
% python3 -m pip install empy
For information about upgrading from 3.x to 4.x, see
http://www.alcyone.com/software/empy/ANNOUNCE.html#changes.
Requirements¶
EmPy works with any modern version of Python. Python version 3.x is
expected to be the default and all source file references to the
Python interpreter (e.g., the bangpath of the .py scripts) use
python3
. EmPy also has legacy support for versions of Python going
back all the way to 2.3, with special emphasis on 2.7 regardless of
its end-of-life status. It has no dependency requirements on any
third-party modules and can run directly off of a stock Python
interpreter.
EmPy will run on any operating system with a full-featured Python interpreter; this includes, but is probably not limited to, Linux, Windows, and macOS (Darwin). Using EmPy requires knowledge of the Python language.
EmPy is also compatible with several different Python implementations:
Implementation |
Supported versions |
Description |
---|---|---|
CPython |
2.3 to 2.7; 3.0 and up |
Standard implementation in C |
PyPy |
2.7; 3.2 and up |
Implementation with just-in-time compiler |
IronPython |
2.7; 3.4 and up |
Implementation for .NET CLR and Mono |
Jython |
2.7 (and up?) |
Implementation for JVM |
It’s probable that EmPy is compatible with earlier versions than those listed here (potentially going all the way back to 2.3), but this has not been tested.
Only a few .py module file(s) are needed to use EmPy; they can be
installed system-wide through a distribution package, a third-party
module/executable, or just dropped into any desired directory in the
PYTHONPATH
. A minimal installation need only install the em.py
file, either as an importable module and an executable, or both,
depending on the user’s needs.
EmPy also has optional support for several third-party modules; see Emoji markup for details.
The testing system included (the test.sh script and the tests and suites directories) is intended to run on Unix-like systems with a Bourne-like shell (e.g., sh, bash, zsh, etc.). EmPy is routinely tested with all supported versions of all available interpreters.
If you find an incompatibility with your Python interpreter or operating system, let me know.
License¶
This software is licensed under BSD (3-Clause).
Getting started¶
This section serves as a quick introduction to the EmPy system. For more details and a reference, see the sections below.
Hint
As an introduction to the terminology, the following names are used throughout:
Name |
Description |
---|---|
|
The name of the software |
|
The name of the executable and main source file |
|
The name of the main module |
|
The name of the pseudomodule, as well as the PyPI package |
|
The conventional filename extension for EmPy documents |
Starting EmPy¶
After installing EmPy (see Getting the
software), EmPy is invoked by executing the
EmPy interpreter, em.py
, on the command line (standalone mode).
If it is invoked without arguments, it will accept input from
sys.stdin
. Unless otherwise specified, the output is sent to
sys.stdout
. You can use this as an interactive session to
familiarize yourself with EmPy when starting out:
% em.py ... accepts input from stdin with results written to stdout ...
If an EmPy document is specified (which by convention has the extension .em, though this is not enforced), then that document is used as input:
% em.py document.em ... document.em is processed and results written to stdout ...
Warning
If your document filename begins with a -
, it will be interpreted as
a command line argument and cause command line option processing
errors. Either precede it with a relative path (e.g., em.py ./-weirdname.em
) or the GNU-style --
option which indicates there
are no further options (e.g., em.py -- -weirdname.em
).
Any number of command line arguments (beginning with a -
) can
precede the document name. For instance, this command writes its
output to document.out:
% em.py -o document.out document.em ... document.em is processed and results written to document.out ...
Many options are available to change the behavior of the EmPy system. This command will open the input file as UTF-8, write the output file as Latin-1, show raw errors if they occur, and delete the output file if an error occurs:
% em.py --input-encoding=utf-8 --output-encoding=latin-1 -r -d -o document.out document.em ... you get the idea ...
EmPy documents can also take arguments, which are an arbitrary
sequence of strings that follow after the document, and are analogous
to the Python interpreter arguments sys.argv
:
% em.py document.em run test ... empy.argv is ['document.em', 'run', 'test'] ...
Tip
You can create executable EmPy scripts by using a bangpath:
#!/usr/bin/env em.py
... EmPy code here ...
By default, bangpaths are treated as EmPy comments unless
--no-ignore-bangpaths
is specified.
Tip
If you wish to run EmPy under Python 2.x for some reason on a system
that also has Python 3 installed, explicitly invoke the Python 2
interpreter before running it (python2 em.py ...
). If you wish to
make this more streamlined, edit the first line (“bangpath”) of em.py
and change it to read #!/usr/bin/env python2
(or whatever your
Python 2.x interpreter is named).
Note
In some distribution packages, the EmPy interpreter may be named
empy
rather than em.py
. In the official release
tarballs, and throughout this documentation,
it is em.py
. This is to distinguish it from the pseudomodule
empy
.
See also
See the Command line options section for a list of command line options that EmPy supports.
The prefix and markup expansion¶
EmPy markup is indicated with a configurable prefix, which is by
default the at sign (@
). The character (Unicode code point)
following the prefix indicates what type of markup it is. There are a
wide variety of markups available, from comments to expression
evaluation to statement execution, and from prefixes, literals and
escapes to diacritics, icons and emojis. Converting markup into
output is referred to as expansion. Here is a long EmPy code
sample illustrating some of the more essential markups in EmPy, though
there are several not shown here:
Example 1: Markup sample
Source:
Comments:
The line below will not render.
@# This is a line comment, up to and including the newline.
If a line comment appears in the middle of a line, @# this is a comment!
the line will be continued.
Inline comments can be @*placed inline* (this phrase did not render,
but note the double space due to the spaces before and after it).
@**
* Or it can span multiple lines.
**@
Whitespace markup consumes the following space.
So two@ words becomes one word.
And this @
is a line continuation.
@* Inline comments can be used as a line comment. *@
Note the use of the trailing prefix to consume the final newline; this
is a common idiom.
Literals:
Double the prefix to render it: @@.
String literals can be used to render escaped Python strings: @
@"A is also \N{LATIN CAPITAL LETTER A}".
Escape markup can render arbitrary characters:
These are all Latin capital letter A: @
A, @\B{1000001}, @\q1001, @\o101, @\x41, @\u0041, @\U00000041, @\N{LATIN CAPITAL LETTER A}.
Backquotes can be used to escape EmPy markup.
This is not evaluated: @`@(!@#$%^&*()`.
Expressions:
Python expressions can be evaluated like this: 1 + 2 = @(1 + 2).
Expressions can be arbitrary complex: @
This is Python @('.'.join(str(x) for x in __import__('sys').version_info[:3])).
Expressions can contain builtin ternary operators:
Seven is an @(7 % 2 == 0 ? 'even' ! 'odd') number.
They can even handle exceptions: @
Division by zero is @(1/0 $ 'illegal').
Statements:
@{
print("Hello, world!")
x = 123
}@
x is now @(x), which can be simplified to @x.
Statements can execute arbitrarily complex Python code,
including defining functions and classes.
Back to expressions, they can be simplified:
@{
# Define some variables.
class Person:
def __init__(self, name):
self.name = name
a = [4, 5, 6]
p = Person('Fred')
}@
x is @x.
a[1] is @a[1].
The name of p is @p.name.
You can even call functions this way:
p's name when shouted is @p.name.upper().
Note that the parser does not try to evaluate end-of-sentence punctuation.
Control structures:
Iterate over some numbers and classify them, but stop after 5:
@[for n in range(-1, 10)]@
@[ if n > 5]@
And done.
@[ break]@
@[ end if]@
@n is @
@[ if n < 0]@
negative@
@[ elif n == 0]@
zero@
@[ elif n % 2 == 0]@
even@
@[ else # odd]@
odd@
@[ end if]@
.
@[end for]@
Note the use of indentation inside control markup and end-of-line
whitespace markup (a prefix with trailing whitespace is consumed) to
make things more clear.
You can even define your own EmPy functions:
@[def officer(name, species, rank, role)]@
@# The definition is EmPy, not Python!
@name (@species, @rank, @role)@
@[end def]@
Some of the bridge crew of the USS Enterprise (NCC-1701):
- @officer("James T. Kirk", "Human", "captain", "commanding officer")
- @officer("Spock", "Vulcan-Human hybrid", "commander", "science officer")
- @officer("Montgomery Scott", "Human", "commander", "chief engineer")
- @officer("Nyota Uhura", "Human", "lieutenant commander", "communications officer")
- @officer("Hikaru Sulu", "Human", "commander", "astrosciences/helmsman")
Diacritics: Libert@^e', @^e'galit@^e', fraternit@^e'!
Icons for curly quotes: @|"(these are curly quotes.@|")
This is an emoji: @:pile of poo:. (Of course I would choose that one.)
Output:
Comments:
The line below will not render.
If a line comment appears in the middle of a line, the line will be continued.
Inline comments can be (this phrase did not render,
but note the double space due to the spaces before and after it).
Whitespace markup consumes the following space.
So twowords becomes one word.
And this is a line continuation.
Note the use of the trailing prefix to consume the final newline; this
is a common idiom.
Literals:
Double the prefix to render it: @.
String literals can be used to render escaped Python strings: A is also A.
Escape markup can render arbitrary characters:
These are all Latin capital letter A: A, A, A, A, A, A, A, A.
Backquotes can be used to escape EmPy markup.
This is not evaluated: @(!@#$%^&*().
Expressions:
Python expressions can be evaluated like this: 1 + 2 = 3.
Expressions can be arbitrary complex: This is Python 3.10.12.
Expressions can contain builtin ternary operators:
Seven is an odd number.
They can even handle exceptions: Division by zero is illegal.
Statements:
Hello, world!
x is now 123, which can be simplified to 123.
Statements can execute arbitrarily complex Python code,
including defining functions and classes.
Back to expressions, they can be simplified:
x is 123.
a[1] is 5.
The name of p is Fred.
You can even call functions this way:
p's name when shouted is FRED.
Note that the parser does not try to evaluate end-of-sentence punctuation.
Control structures:
Iterate over some numbers and classify them, but stop after 5:
-1 is negative.
0 is zero.
1 is odd.
2 is even.
3 is odd.
4 is even.
5 is odd.
And done.
Note the use of indentation inside control markup and end-of-line
whitespace markup (a prefix with trailing whitespace is consumed) to
make things more clear.
You can even define your own EmPy functions:
Some of the bridge crew of the USS Enterprise (NCC-1701):
- James T. Kirk (Human, captain, commanding officer)
- Spock (Vulcan-Human hybrid, commander, science officer)
- Montgomery Scott (Human, commander, chief engineer)
- Nyota Uhura (Human, lieutenant commander, communications officer)
- Hikaru Sulu (Human, commander, astrosciences/helmsman)
Diacritics: Liberté, égalité, fraternité!
Icons for curly quotes: “these are curly quotes.”
This is an emoji: 💩. (Of course I would choose that one.)
Tip
If you wish to change the prefix, use -p/--prefix=CHAR
(environment variable: EMPY_PREFIX
, configuration variable: prefix
).
See also
See the Markup section for detailed specifications on all support EmPy markup.
Pseudomodule and interpreter¶
The interpreter instance is available to a running EmPy system through
the globals; by default, it is named empy
.
When it is referenced this way, it is called a pseudomodule (since
it acts like a module but it is not actually a module you can import):
Example 2: Pseudomodule sample
Source:
This version of EmPy is @empy.version.
The prefix in this interpreter is @empy.getPrefix() @
and the pseudomodule name is @empy.config.pseudomoduleName.
Do an explicit write: @empy.write("Hello, world!").
The context is currently @empy.getContext().
Adding a new global in a weird way: @empy.updateGlobals({'q': 789})@
Now q is @q!
You can do explicit expansions: @empy.expand("1 + 1 = @(1 + 1)").
q is @(empy.defined('q') ? 'defined' ! 'undefined').
Output:
This version of EmPy is 4.1.
The prefix in this interpreter is @ and the pseudomodule name is empy.
Do an explicit write: Hello, world!.
The context is currently <example 2>:5:26.
Adding a new global in a weird way: Now q is 789!
You can do explicit expansions: 1 + 1 = 2.
q is defined.
See also
See the Pseudomodule/interpreter section for details on the pseudomodule/interpreter.
Diversions, filters & hooks¶
Diversions can defer and replay output at a desired time:
Example 3: Diversions sample
Source:
This text is output normally.
@empy.startDiversion('A')@
(This text was diverted!)
@empy.stopDiverting()@
This text is back to being output normally.
Now playing the diversion:
@empy.playDiversion('A')@
And now back to normal output.
Output:
This text is output normally.
This text is back to being output normally.
Now playing the diversion:
(This text was diverted!)
And now back to normal output.
Filters can modify output before sending it to the final stream:
Example 4: Filters sample
Source:
@{
# For access to the filter classes.
import emlib
}@
This text is normal.
@empy.appendFilter(emlib.FunctionFilter(lambda x: x.upper()))@
This text is in all uppercase!
@empy.appendFilter(emlib.FunctionFilter(lambda x: '[' + x + ']'))@
Now it's also surrounded by brackets!
(Note the brackets are around output as it is sent,
not at the beginning and end of each line.)
@empy.resetFilter()@
Now it's back to normal.
Output:
This text is normal.
THIS TEXT IS IN ALL UPPERCASE!
[NOW IT'S ALSO SURROUNDED BY BRACKETS!
(NOTE THE BRACKETS ARE AROUND OUTPUT AS IT IS SENT,
NOT AT THE BEGINNING AND END OF EACH LINE.)
]Now it's back to normal.
Hooks can intercept and even alter the behavior of a running system:
Example 5: Hooks sample
Source:
@# Modify the backquote markup to prepend and append backquotes
@# (say, for a document rendering system, cough cough).
@{
import emlib
class BackquoteHook(emlib.Hook):
def __init__(self, interp):
self.interp = interp
def preBackquote(self, literal):
self.interp.write('`' + literal + '`')
return True # return true to skip the standard behavior
empy.addHook(BackquoteHook(empy))
}@
Now backquote markup will render with backquotes: @
@`this is now in backquotes`!
Output:
Now backquote markup will render with backquotes: `this is now in backquotes`!
See also
See the Diversions section, Filters section, or the Hooks section for more information.
Embedding¶
EmPy is modular and can be embedded in your Python programmers,
rather than running it standalone. Simply import the em
module and
create an Interpreter
:
import sys
import em
config = em.Configuration(...)
output = sys.stdout
with em.Interpreter(config=config, output=output) as interp:
... do things with interp ...
An exception which occurs during processing will be handled by the interpreter’s error handler.
For one-off uses, you can use the global, standalone expand
function:
import em
result = em.expand(source)
When calling this function, an ephemeral interpreter is dynamically created, used, and shutdown to perform the expansion. If an exception occurs during this processing, it will be raised to the caller, rather than handled by the ephemeral interpreter.
Important
When you create an interpreter, you must call its shutdown
method
when you are done. This is required to remove the proxy on
sys.stdout
that EmPy requires for proper operation and restore your
Python environment to the state it was before creating the
interpreter. This can be accomplished by creating the interpreter in
a with
statement — interpreters are also context managers — or by
creating it and shutting it down in a try
/finally
statement.
This is not needed when calling the expand
global function; it
creates and shuts down an ephemeral interpreter automatically.
See also
See the Embedding EmPy section section for more details on embedding EmPy in your Python programs.
Getting help¶
For basic help, use the -h/--help
option:
% em.py -h # or: --help Welcome to EmPy version 4.1. USAGE: ./em.py [<options>] [<filename, or `-` for stdin> [<argument>...]] - Options begin with `-` or `--` - Specify a filename (and arguments) to process that file as input - Specify `-` (and arguments) to process stdin with standard buffering - Specify no filename to enter interactive mode with line buffering - Specify `--` to stop processing options ...
For more help, repeat the -h/--help
option (up to three times
for the full help). For help on a particular topic, use the
-H/--topics=TOPICS
option, where TOPICS
is a comma-separated list of
topics. The list of available topics can be shown by using the topic
topics
:
% em.py -H topics # or: --topics=topics Welcome to EmPy version 4.1. TOPICS: Need more help? Add more -h options (-hh, -hhh) for more help. Use -H <topic> for help on a specific topic, or specify a comma-separated list of topics. Try `default` (-h) for default help, `more` (-hh) for more common topics, `all` (-hhh) for all help topics, or `topics` for this list. Use -V for version information, -W for version and system information, or -Z for all debug details. Available topics: usage Basic command line usage options Command line options markup Markup syntax escapes Escape sequences environ Environment variables pseudo Pseudomodule attributes and functions constructor Keyword arguments for the Interpreter constructor variables Configuration variable attributes methods Configuration methods hooks Hook methods named Named escapes diacritics Diacritic combiners icons Icons hints Usage hints topics This list of topics
Tip
Repeating the help option once (-hh
) is the same as requesting the
more
topic (-H more
). Repeating it three times (-hhh
) is the
same as requesting the all
topic (-H all
).
Warning
The builtin help system requires the presence of the emhelp
module.
If you have a minimal EmPy installation, this module may not be
available. You can get it from the release
tarball.
See also
See the rest of this document for details and specifications on all the markup and features, and see the Help topics section for the output of all the builtin help topics.
Markup¶
EmPy markup always begins with the EmPy prefix, which defaults to
@
. The character (Unicode code point) following the prefix
indicates what type of markup it is, and the different types of markup
are parsed differently.
It is legal to set the EmPy prefix to None
; then, no markup will be
parsed or expanded and EmPy will merely process filters and encoding
conversions. This can be done from the command line with the
--no-prefix
option, or by indicating a prefix that is an
empty string (''
) or the word none
.
Using a non-default prefix that is also the first character of an
existing markup will swap that markup character with the default. For
example, setting the prefix to $
would otherwise collide with the
in-place token (@$...$...$
with a default prefix). On startup
it will be adjusted so that with a $
prefix the in-place markup can
be accessed as $@...@...@
.
The following subsections list the types of markup supported by EmPy
and in which version they were introduced, organized by category.
NL
represents a newline and WS
represents any whitespace.
Important
All of the following code snippets and examples below assume that the
prefix is the default, @
. It can be changed with
-p/--prefix=CHAR
(environment variable: EMPY_PREFIX
, configuration variable: prefix
).
Markup |
Syntax |
Description |
Ver. |
---|---|---|---|
|
Consumes text up to and including newline |
1.0 |
|
|
Consumes text up to and including the final asterisk(s) |
4.0 |
|
|
Consumes the following whitespace character |
1.0 |
|
|
Produces the prefix character |
1.0 |
|
|
Produces a string from a literal |
3.1.1 |
|
|
Quotes contained markup up to final backquote(s) |
4.0 |
|
|
Render an escape character |
1.5 |
|
|
Render an escape control character by name |
4.0 |
|
|
Evaluates an expression |
1.0 |
|
|
Evaluates a simple expression |
1.0 |
|
|
Evaluates a functional expression |
4.0 |
|
|
Expression evaluation with if-else-except |
1.3 |
|
|
Copies and evaluates an expression |
1.4 |
|
|
Executes a statement or statements |
1.0 |
|
|
Control structures |
3.0 |
|
|
Branching control structure |
3.0 |
|
|
Break out of repeating control structure |
3.0 |
|
|
Continue with next iteration of repeating control structure |
3.0 |
|
|
Iterating control structure |
3.0 |
|
|
Looping control structure |
3.0 |
|
|
Looping control structure always entered once ( |
4.0 |
|
|
Exception handling, guarding |
3.0 |
|
|
Handle a context manager |
4.0 |
|
|
Structural pattern matching |
4.1 |
|
|
Branch on whether a variable is defined |
4.0 |
|
|
Define an EmPy function |
3.0 |
|
|
Render and normalize diacritic combiner(s) |
4.0 |
|
|
Render a customizable icon |
4.0 |
|
|
Render a customizable emoji |
4.0 |
|
|
Declare a significator (metadata assignment) |
1.2 |
|
|
Set the context filename |
3.0.2 |
|
|
Set the context line |
3.0.2 |
|
|
Fully-customizable markups with no set definition |
4.1 |
See also
The list of supported markup is available in the markup
help topic
and is summarized here.
Literal markups¶
Literal markups are a category of markup that evaluate to some form of themselves.
Prefix markup: @@
¶
To render the prefix character literally in the output, duplicate it
as prefix markup. For the default, @
, it will be @@
:
Example 11: Prefix literals
Source:
This becomes a single at sign: @@.
Output:
This becomes a single at sign: @.
Tip
The prefix markup is not indicated by the prefix followed by an at
sign, but rather the prefix repeated twice. So if the prefix has been
changed to $
, the prefix markup is $$
, not $@
.
Note
Prefix markup was introduced in EmPy version 1.0.
String markup: @'...'
, @"..."
, @'''...'''
, @"""..."""
¶
The interpreter prefix followed by a Python string literal
(e.g., @'...'
) evaluates the Python string literal and expands
it. All variants of string literals with single and double quotes, as
well as triple quoted string literals (with both variants) are
supported. This can be useful when you want to use Python string
escapes (not EmPy escapes) in a compact form:
Example 12: String
Source:
This is a string: @'A single-quoted string'.
This is also a string: @"A double-quoted string".
This is another string: @'''A triple single-quoted string'''.
This is yet another string: @"""A triple double-quoted string""".
This is a multiline string: @"""Triple quotes containing newlines
will be preserved."""
This is a string using escapes: @
@'Welcome to \U0001d53c\U0001d55e\u2119\U0001d56a!'.
Output:
This is a string: A single-quoted string.
This is also a string: A double-quoted string.
This is another string: A triple single-quoted string.
This is yet another string: A triple double-quoted string.
This is a multiline string: Triple quotes containing newlines
will be preserved.
This is a string using escapes: Welcome to 𝔼𝕞ℙ𝕪!.
Note
String markup was introduced in EmPy version 3.1.1.
Backquote markup: @`...`
¶
Backquote markup (@`...`
) can be used to escape any text,
including EmPy markup. Multiple opening backquotes can be used as
long as they are matched by an equal number in order to allow quoting
text which itself has backquotes in it:
Example 13: Backquote
Source:
This is literal text: @`some text`.
This is a prefix: @`@`.
This would be expanded if it were not backquoted: @`@(1 + 1)`.
This would be an error if expanded: @`@(!@#$%^&*())`.
This contains backquotes: @```here's one: ` and here's two: `` ```.
Output:
This is literal text: some text.
This is a prefix: @.
This would be expanded if it were not backquoted: @(1 + 1).
This would be an error if expanded: @(!@#$%^&*()).
This contains backquotes: here's one: ` and here's two: `` .
Warning
To use the backquote markup with content containing backquotes which
are adjacent to the start or end markup, you need to pad it with
spaces. So when quoting a single backquote, it needs to be written as
@`` ` ``
. This also means you cannot use backquote markup to
specify a completely empty string. It must always contain at least
one non-backquote character, e.g., @` `
. If you really need
backquotes without whitespace padding, you can use a hook to
intercept the backquote markup and strip it out.
Attention
Note that when markup which has starting and ending delimiters appears
alone on a line, the trailing newline will be rendered in the output.
To avoid these extra newlines, use a trailing @
to turn it
into whitespace markup which consumes that trailing newline, so e.g.
@`...`
followed by a newline becomes @`...`@
followed by a newline. This is
idiomatic for suppressing unwanted newlines. See here for
more details.
Note
Backquote markup was introduced in EmPy version 4.0.
Escape markup: @\...
¶
Escape markup allows specifying individual non-printable
characters with a special readable syntax: @\...
. It is inspired
by and extends the string literal escape codes from languages such as
C/C++ and Python.
Example 14: Escapes
Source:
@# These are all a Latin uppercase A:
Binary: @\B{1000001}
Quaternary: @\q1001, @\Q{1001}
Octal: @\o101, @\O{101}
Hexadecimal (variable bytes): @\X{41}
Hexadecimal (one-byte): @\x41
Hexadecimal (two-byte): @\u0041
Hexadecimal (eight-byte): @\U00000041
By Unicode name: @\N{LATIN CAPITAL LETTER A}
Output:
Binary: A
Quaternary: A, A
Octal: A, A
Hexadecimal (variable bytes): A
Hexadecimal (one-byte): A
Hexadecimal (two-byte): A
Hexadecimal (eight-byte): A
By Unicode name: A
The escape sequence type is indicated by the first character and then
consumes zero or more characters afterward, depending on the escape
sequence. Some sequence sequences support a variable number of
characters, delimited by curly braces ({...}
).
See also
The list of all valid escape sequences is available in the escapes
help topic and is summarized
here.
Note
Escape markup was introduced in EmPy version 1.5, and then reworked in EmPy version 4.0.
Named escape markup: @\^{...}
¶
The escape markup for controls @\^...
has an extended usage where
the character can be specified by a control code name. The resulting
named escape markup takes the form of @\^{...}
with the escape
code name between the curly braces. The name of the escape code used
in the markup is case insensitive.
The mapping of escape names to characters is specified in the
configuration variable controls
. The keys of this
dictionary must be in uppercase and the values can be integers
(Unicode code point values), lists of integers, or strings. They can
also take the form of a 2-tuple, where the first element is one of the
above values and the second element is a description string used for
displaying in help topics.
Example 15: Named escapes
Source:
Normal space: [ ]
Normal space by name: [@\^{SP}]
No-break space: [@\^{NBSP}]
Thin space: [@\^{THSP}]
En space: [@\^{ENSP}]
Em space: [@\^{EMSP}]
(Well, these would look right if it this were in a proportional font.)
Output:
Normal space: [ ]
Normal space by name: [ ]
No-break space: [ ]
Thin space: [ ]
En space: [ ]
Em space: [ ]
(Well, these would look right if it this were in a proportional font.)
See also
The list of all valid control code names is available in the
named
help topic and is summarized
here.
Note
Named escape markup was introduced in EmPy version 4.0.
Expression markup: @(...)
¶
EmPy mainly processes markups by evaluating expressions and executing
statements. Expressions are bits of Python code that return a value;
that value is then rendered into the output stream. Simple examples
of Python expressions are 1 + 2
, abs(-2)
, or "test"*3
.
In EmPy, expressions are evaluated and expanded with the expression
markup @(...)
. By default, an expression that evaluates to
None
does not print anything to the underlying output stream; it is
equivalent to it having returned ''
.
Tip
If you want to change this behavior, specify your preferred value with
--none-symbol
(configuration variable: noneSymbol
).
Example 16: Expressions
Source:
The sum of 1 and 2 is @(1 + 2).
The square of 3 is @(3**2).
The absolute value of -12 is @(abs(-12)).
This prints "test" but does not print None: @(print("test", end='')).
This, however, does: @(repr(None)).
Output:
The sum of 1 and 2 is 3.
The square of 3 is 9.
The absolute value of -12 is 12.
This prints "test" but does not print None: test.
This, however, does: None.
Attention
Note that when markup which has starting and ending delimiters appears
alone on a line, the trailing newline will be rendered in the output.
To avoid these extra newlines, use a trailing @
to turn it
into whitespace markup which consumes that trailing newline, so e.g.
@(...)
followed by a newline becomes @(...)@
followed by a newline. This is
idiomatic for suppressing unwanted newlines. See here for
more details.
Note
Expression markup was introduced in EmPy version 1.0.
Additional expression markup¶
Several expression markup variants are available.
Simple expression markup: @x
, @x.a
, @x[i]
, @x(args...)
, etc.¶
Often expressions are “simple” and unambiguous enough that needing to
use the full @(...)
syntax is unnecessary. In cases where a
single variable is being referenced unambiguously, the parentheses can
be left off to create simple expression markup:
Example 17: Simple expressions, basic
Source:
@# Set a variable to use.
@{x = 16309}@
The value of x is @x.
Output:
The value of x is 16309.
@x
is precisely the same thing as @(x)
.
This markup can be extended further. Attribute references (@x.a
),
indexing (@x[i]
), and function calls (@x(args...)
) can also be
simplified in this way. They can also be chained together
arbitrarily, so @object.attribute.subattribute
,
@object.method(arguments...)
, @object[index1][index2]
,
@object[index].attribute
, @object[index].method(arguments...)
,
etc. are all valid examples of simple expression markup. These simple
expressions can be extended arbitrarily.
Example 18: Simple expressions, chaining
Source:
@# Define some variables to use.
@{
import time
def mean(seq): # a function
return sum(seq)/len(seq)
class Person: # a class
def __init__(self, name, birth, scores):
self.name = name
self.birth = birth
self.scores = scores
def age(self):
current = time.localtime(time.time()).tm_year
return current - self.birth
person = Person("Fred", 1984, [80, 100, 70, 90]) # an instance of that class
}@
The name of person is @(person.name), or more simply @person.name.
The first letter is @(person.name[0]), or more simply @person.name[0].
He has @(len(person.scores)) scores, or more simply @len(person.scores).
His first score is @(person.scores[0]), or more simply @person.scores[0].
His average score is @(mean(person.scores)), or more simply @mean(person.scores).
His age is @(person.age()), or more simply @person.age().
Output:
The name of person is Fred, or more simply Fred.
The first letter is F, or more simply F.
He has 4 scores, or more simply 4.
His first score is 80, or more simply 80.
His average score is 85.0, or more simply 85.0.
His age is 40, or more simply 40.
Note
Final punctuation, including a period (.
), is not interpreted as an
attribute reference and thus does not result in a parse error. Thus
you can use end-of-sentence punctuation naturally after a simple
expression markup.
If you wish to concatenate an expression with immediately following text so that it will not be parsed incorrectly, either use whitespace markup or just fall back to a full expression markup:
Example 19: Simple expressions, concatenation
Source:
@# Define a variable for use.
@{thing = 'cat'}@
@# Referencing `@things` to pluralize `@thing` will not work. But:
The plural of @thing is @thing@ s.
Or: The plural of @thing is @(thing)s.
Output:
The plural of cat is cats.
Or: The plural of cat is cats.
Note
Simple expression markup was introduced in EmPy version 1.0.
Functional expression markup: @function{markup}{...}
¶
Arguments to function calls in EmPy expression markups use Python
expressions, not EmPy markup (e.g., @f(x)
calls the function f
with the variable x
). To specify EmPy markup which is expanded and
then passed in to the function, there is functional expression
markup as an extension of simple expression markup. Since each
argument to the function is expanded, the arguments are always
strings:
Example 20: Functional expressions, one argument
Source:
@{
def f(x):
return '[' + x + ']'
}@
@# Note that the argument is expanded before being passed to the function:
This will be in brackets: @f{1 + 1 is @(1 + 1)}.
Output:
This will be in brackets: [1 + 1 is 2].
Functional expressions support the application of multiple arguments
by repeating the {...}
suffix for as many arguments as is desired:
Example 21: Functional expressions, multiple arguments
Source:
@{
def f(x, y, z):
return x.lower() + ', ' + y.upper() + ', ' + z.capitalize()
}@
@# Multiple arguments are possible by repeating the pattern:
These expansions are separated by commas: @
@f{lowercase: @(1)}{uppercase: @(1 + 1)}{capitalized: @(1 + 1 + 1)}.
Output:
These expansions are separated by commas: lowercase: 1, UPPERCASE: 2, Capitalized: 3.
Warning
Functional expression markup is an extension of simple expression
markup so cannot be surrounded in parentheses. Further, it cannot be
seemlessly combined with normal function call, so @f(1){a}{b}
is
equivalent to @(f(1)('a', 'b'))
, not @(f(1, 'a', 'b'))
.
Functional argument calls will end simple expression, so
@f{a}{b}(3)
is the same as @(f('a', 'b'))(3)
, not @f('a', 'b', 3)
; that is, trailing function calls are not applied.
Note
Functional expression markup was introduced in EmPy version 4.0.
Extended expression markup: @(...?...!...$...)
¶
Expression markup has an extended expression markup form which allows more powerful manipulation of expressions.
The first form is conditional expression markup which allows for a
compact form of an @[if]
statement with a ternary operator, similar
to C/C++’s ?
and :
operators. In EmPy, however, these are
represented with ?
and !
, respectively.
Note
C/C++’s use of :
was changed to !
for EmPy since :
already has
special meaning in Python. This syntax was originally added before
Python supported the if/else
ternary expression, although EmPy’s
syntax is more general and powerful.
If a ?
is present in the expression, then the Python (not EmPy)
expression before the ?
is tested; if it is true, then the Python
expression following it is evaluated. If a !
is present afterward
and the originally expression was false, then the Python expression
following it is expanded (otherwise, nothing is). It thus acts as an
if-then-else construct:
Example 22: Extended expressions, conditional
Source:
Four is an @(4 % 2 == 0 ? 'even' ! 'odd') number.
Seven is an @(7 % 2 == 0 ? 'even' ! 'odd') number.
@# Whitespace is not required:
Eleven is an @(11 % 2 == 0?'even'!'odd') number.
Output:
Four is an even number.
Seven is an odd number.
Eleven is an odd number.
These ?
and !
sequences can be repeated indefinitely, forming an
if-else if-else chain called chained conditional expression
markup, with a !
expression serving as the conditional test for
the next ?
:
Example 23: Extended expressions, chained conditional
Source:
@# Define a variable for use.
@{x = 3}@
x is @(x == 1 ? 'one' ! x == 2 ? 'two' ! x == 3 ? 'three' ! 'unknown').
Output:
x is three.
Finally, a $
present at the end of any if-else chain represents
except expression markup: If the main expression throws an exception,
suppress it and evaluate the Python (not EmPy) except expression
instead. This can be combined with conditional expresion markup:
Example 24: Extended expressions, except
Source:
No exception: 2 + 2 = @(2 + 2 $ 'oops').
Division by zero is @(1/0 $ 'illegal').
Two divided by zero is @(2/0 % 2 == 0 ? 'even' ! 'odd' $ 'also illegal').
Output:
No exception: 2 + 2 = 4.
Division by zero is illegal.
Two divided by zero is also illegal.
Attention
Note that when markup which has starting and ending delimiters appears
alone on a line, the trailing newline will be rendered in the output.
To avoid these extra newlines, use a trailing @
to turn it
into whitespace markup which consumes that trailing newline, so e.g.
@(...)
followed by a newline becomes @(...)@
followed by a newline. This is
idiomatic for suppressing unwanted newlines. See here for
more details.
Note
Conditional expression markup was first introduced in EmPy version 1.3, updated to extended expressions (including exception handling) in EmPy version 1.4, and was expanded to support if-else chaining in 4.0.
In-place expression markup: @$...$...$
¶
Occasionally it’s desirable to designate an expression that will be
evaluated alongside its evaluation which may change, but which will be
re-evaluated with subsequent updates, or identify exactly what is
being evaluated at the same time. This is similar to the notion of
CVS or SVN keywords such as $Date ...$
. For this, there is
in-place expression markup (@$...$...$
). They consist of two
segments: first, the Python (not EmPy) expression to evaluate, and the
second, the result of that evaluation. When evaluating the markup,
the second (result) section is ignored and replaced with the
evaluation of the first and a new in-place markup is rendered. For
example:
Example 25: In-place expressions
Source:
This could be a code comment indicating the version of EmPy:
# @$empy.version$this text is replaced with the result$
Arbitrary Python expressions can be evaluated:
# @$__import__('time').asctime()$$
Output:
This could be a code comment indicating the version of EmPy:
# @$empy.version$4.1$
Arbitrary Python expressions can be evaluated:
# @$__import__('time').asctime()$Sat Apr 6 16:38:10 2024$
Note
The $
character is a common choice for an alternate prefix. If it
is chosen instead of the default @
, the in-place expression markup
will be remapped to have the form $@...@...@
; that is, the @
and $
are swapped. (This is done automatically for any prefix
collision with a markup indicator.)
Attention
Note that when markup which has starting and ending delimiters appears
alone on a line, the trailing newline will be rendered in the output.
To avoid these extra newlines, use a trailing @
to turn it
into whitespace markup which consumes that trailing newline, so e.g.
@$...$
followed by a newline becomes @$...$@
followed by a newline. This is
idiomatic for suppressing unwanted newlines. See here for
more details.
Note
In-place markup was introduced in EmPy version 1.4.
Statement markup: @{...}
¶
Again, EmPy mainly processes markups by evaluating expressions and
executing statements. Statements include assignments, control
structures (if
, for
, function and class definitions, etc.)
Statements do not yield a value; they are used for side effects,
whether that’s changing the state of the interpreter (setting or
changing variables, defining objects, calling functions, etc.) or
printing output. Statements can also consist of expressions, so an
expression (such as print("Hello, world!")
) can be used solely for
its side effects with the statement markup. Statement markup sets
off a series of statements to be executed inside the @{...}
markup. Since statements do not yield a value, they are executed but
the markup itself does not implicitly write anything. Since the
executed statements are Python, multiline statements must be formatted
and indented according to Python’s parsing rules:
Example 26: Statements
Source:
@# Note the use of whitespace markup below to consume trailing newlines.
@{x = 16309}@
x is now @x.
@{
if x > 0:
category = 'positive'
else:
category = 'non-positive'
}@
x is @category.
@{
# Since statement markup does not write anything itself, this
# statement has no effect.
x + 123
}@
Output:
x is now 16309.
x is positive.
Attention
Note that when markup which has starting and ending delimiters appears
alone on a line, the trailing newline will be rendered in the output.
To avoid these extra newlines, use a trailing @
to turn it
into whitespace markup which consumes that trailing newline, so e.g.
@{...}
followed by a newline becomes @{...}@
followed by a newline. This is
idiomatic for suppressing unwanted newlines. See here for
more details.
Note
Simple expression markup was introduced in EmPy version 1.0.
Control markups: @[...]
¶
EmPy supports a variety of control structures, analogous to the
builtin Python control structures (if
, while
, for
, etc.), with
some additional markups for convenience. This is done with control
markup indicated by @[...]
.
Since EmPy cannot rely on source indentation to delimit control
structure syntax, all primary control markups must end with an
explicit end
markup (e.g., @[if ...]...@[end if]
). The clauses
surrounded by control markup are EmPy (Python) markup and are expanded
according to the logic of each control markup; see below.
Unlike the Python control structures, the code that is expanded within
each subclause is EmPy code, not Python code. Thus, control markups
can be nested arbitrarily (e.g., @[while ...]@[for ...]@[if ...]...@[end if]@[end for]@[end while]
).
Attention
To use nested control markup that spans multiple lines and is more readable, you can rely on whitespace markup to consume the newline immediately following the control markup. As an example:
Example 27: Controls, idiom
Source:
@# Note the user of whitespace markup to consume the trailing newlines.
Counting:
@[for i, x in enumerate(range(0, 5))]@
@x is @
@[ if x % 2 == 0]@
even@
@[ else]@
odd@
@[ end if]@
.
@[end for]@
Output:
Counting:
0 is even.
1 is odd.
2 is even.
3 is odd.
4 is even.
This method of writing organizing control markup with @[...]@
all
on a single line for clarity is idiomatic EmPy. (This applies to all
markup with starting and ending delimiters.) See here for
more details.
Hint
Whitespace before the control keyword is ignored, so you can add whitespace inside the markup to simulate Python indentation for clarity, as the above example demonstrates.
Tip
Simple (“clean”) control markup which does not contain arbitrary
Python expressions — @[try]
, @[else]
, @[except ...]
,
@[finally]
, @[continue]
, @[break]
and @[end ...]
— can
include a Python-style comment for clarity:
@[if test condition]@
... many lines of EmPy code here ...
... imagine this was so long that the context could be confusing ...
... so as a reminder of which `if` control is being ended:
@[end if # test condition]@
Note
Control markups were introduced in EmPy version 3.0 unless otherwise noted below.
If control markup: @[if E]...@[end if]
¶
The simplest control markup is the if control markup. It
precisely mimics the Python if
branching control structure. The
test expressions are Python expressions. Like the native Python
control structure, it takes on the following forms:
@[if E]...@[end if]
@[if E]...@[else]...@[end if]
@[if E]...@[elif E2]...@[end if]
@[if E]...@[elif E2]...@[else]...@[end if]
@[if E]...@[elif E2]... ... @[else]...@[end if]
Thus, as with the builtin Python if
control structure, zero or more
@[elif]
clauses can be used and the @[else]
clause (only valid
at the end of the chain) is optional. If there is no @[else]
clause and all the test expressions are false, nothing will be
expanded.
Example 28: If controls
Source:
@{
def even(x):
return x % 2 == 0
}@
0 is @[if even(0)]even@[end if].
1 is @[if even(1)]even@[else]odd@[end if].
2 is @[if even(2)]even@[else]odd@[end if].
3 is @[if even(3)]even@[elif not even(3)]not even@[end if].
4 is @[if 0 == 1]wrong@[elif 1 == 2]wrong@[else]fine@[end if].
Output:
0 is even.
1 is odd.
2 is even.
3 is not even.
4 is fine.
Break and continue control markup: @[break]
, @[continue]
¶
The looping control markup structures below (@[for]
, @[while]
,
and @[dowhile]
) all support break and continue control
markup. These markups follow the native Python forms; @[break]
will exit out of the innermost looping control structure, and
@[continue]
will restart the innermost looping control structure.
They take the following forms:
@[break]
@[continue]
The following is an example using a @[while]
loop:
Example 29: Continue controls
Source:
@# Print even numbers.
@[for n in range(10)]@
@[ if n % 2 != 0]@
@[ continue]@
@[ end if]@
@n is even.
@[end for]@
Output:
0 is even.
2 is even.
4 is even.
6 is even.
8 is even.
Example 30: Break controls
Source:
@# Print numbers up to (but not including) 5.
@[for n in range(10)]@
@[ if n >= 5]@
@[ break]@
@[ end if]@
@n is less than 5.
@[end for]@
Output:
0 is less than 5.
1 is less than 5.
2 is less than 5.
3 is less than 5.
4 is less than 5.
For control markup: @[for N in E]...@[end for]
¶
A basic iteration markup is the for control markup. It precisely
mimics the Python for
looping control structure. The iterator
expression is a Python expression. Like the native Python control
structure, it takes on the following forms:
@[for N in E]...@[end for]
@[for N in E]...@[else]...@[end for]
As with the native Python control structure, an @[else]
clause is
supported; this is expanded if the loop exits without an intervening
break.
Example 31: For controls
Source:
@[for x in range(1, 6)]@
@x squared is @(x*x).
@[else]@
... and done.
@[end for]@
Output:
1 squared is 1.
2 squared is 4.
3 squared is 9.
4 squared is 16.
5 squared is 25.
... and done.
While control markup: @[while E]...@[end while]
¶
The most general looping markup is the while control markup. It
precisely mimics the Python while
looping control structure. The
test expression is a python expression. Like the native Python
control structure, it takes on the following forms:
@[while E]...@[end while]
@[while E]...@[else]...@[end while]
As with the native Python control structure, an @[else]
clause is
supported; this is invoked if the loop exits without an intervening
break.
Example 32: While controls
Source:
@{a = 1}@
@[while a <= 5]@
@a pound signs: @('#'*a).
@{a += 1}@
@[else]@
... and done.
@[end while]@
Output:
1 pound signs: #.
2 pound signs: ##.
3 pound signs: ###.
4 pound signs: ####.
5 pound signs: #####.
... and done.
Dowhile control markup: @[dowhile E]...@[end dowhile]
¶
An alternate while
control structure is provided by EmPy: dowhile
control markup. This differs from the standard while
markup only
in that the loop is always entered at least once; that is, the test
expression is not checked before the first iteration. In this way, it
is similar to the do ... while
control structure from C/C++. It
takes the following forms:
@[dowhile E]...@[end dowhile]
@[dowhile E]...@[else]...@[end dowhile]
Like the native Python while
control structure, an @[else]
clause
is supported; this is invoked if the loop exits without an intervening
break.
Example 33: Dowhile controls
Source:
@# Stop when divisible by 5, but include 0 since it's the first iteration:
@{n = 0}@
@[dowhile n % 5 != 0]@
@n works@[if n % 5 == 0] (even though it's divisible by 5)@[end if].
@{n += 1}@
@[else]@
... and done.
@[end dowhile]@
Output:
0 works (even though it's divisible by 5).
1 works.
2 works.
3 works.
4 works.
... and done.
Note
Dowhile control markup was introduced in EmPy version 4.0.
Try control markup: @[try]...@[end try]
¶
Try control markup is the EmPy equivalent of a try
statement.
As with the native Python statement, this markup can take on the
widest variety of forms. They are:
@[try]...@[except]...@[end try]
@[try]...@[except C]...@[end try]
@[try]...@[except C as N]...@[end try]
@[try]...@[except C, N]...@[end try]
@[try]...@[except (C1, C2, ...) as N]...@[end try]
@[try]...@[except C1]...@[except C2]...@[end try]
@[try]...@[except C1]...@[except C2]... ... @[end try]
@[try]...@[finally]...@[end try]
@[try]...@[except ...]...@[finally]...@[end try]
@[try]...@[except ...]...@[else]...@[end try]
@[try]...@[except ...]...@[else]...@[finally]...@[end try]
Its behavior mirrors in every way the native Python try
statement.
The try clause will be expanded, and if an exception is thrown, the
first @[except]
clause that matches the thrown exception (if there
are any) will be expanded. If a @[finally]
clause is present, that
will be expanded after any possible exception handling, regardless of
whether an exception was in fact thrown. Finally, if there is at
least one @[except]
clause, an @[else]
may be present which will
be expanded in the event that no exception is thrown (but before any
@[finally]
clause).
The argument to the @[except]
markup indicates which type(s) of
exception should be handled and with what name, if any. No argument
indicates that it will handle any exception. A simple expression will
indicate an exception class, or a tuple of exception classes, that
will be handled. The variable name of the thrown exception can be
captured and passed to the expansion with the as
keyword, or a comma
(this latter notation is invalid in modern Python versions but is
still supported in EmPy regardless of the underlying Python version).
For example:
Example 34: Try controls
Source:
Garbage is @[try]@hugalugah@[except NameError]not defined@[end try].
Division by zero is @[try]@(1/0)@[except ZeroDivisionError]illegal@[end try].
An index error is @[try]@([][3])@[except IndexError as e]@e.__class__.__name__@[end try].
And finally: @[try]@(nonexistent)@[except]oops, @[finally]something happened@[end try].
Output:
Garbage is not defined.
Division by zero is illegal.
An index error is IndexError.
And finally: oops, something happened.
Note
Try control markup was introduced in EmPy version 3.0, and was
expanded in 4.0 to include all modern valid uses of @[else]
and
@[finally]
.
With control markup: @[with E as N]...@[end with]
¶
EmPy supports a version of the with
statement, which was introduced
in Python 2.5. In EmPy, the with control markup is written as
@[with]
and mirrors the behavior of the native with
statement.
It takes the following forms:
@[with E as N]...@[end with]
@[with N]...@[end with]
@[with E]...@[end with]
All forms use context managers, just as with the native statement.
Context managers are objects which have __enter__
and __exit__
methods, and the @[with]
markup ensures that the former is called
before the markup’s contents are expanded and that the latter is
always called afterward, whether or not an exception has been thrown.
The three forms of the @[with]
markup mirror the uses of the with
keyword: The user can specify an expression and a variable name with
the as
keyword, or just a variable name, or just an expression (it
will be entered and exited, but the name of the resulting object will
not be available). For example:
Example 35: With controls
Source:
@{
import os, sys
# Create a test file to use with the @[with ...] markup.
with open('/tmp/with.txt', 'w') as f:
print("Hello, world!", file=f)
}@
@[with open('/tmp/with.txt') as f]@f.read()@[end with]@
Output:
Hello, world!
Note
Although the with
keyword was only introduced in Python 2.5, the
@[with]
markup will work in any supported version of Python.
Note
With control markup was introduced in EmPy version 4.0.
Match control markup: @[match E]@[case C]...@[end match]
¶
Python 3.10 introduces structural pattern matching with the
match
/case
control structure. The analog to this in EmPy is the
match control markup. It takes the following forms:
@[match E]@[case C1]...@[end match]
@[match E]@[case C1]...@[case C2]...@[end match]
@[match E]@[case C1]...@[case C2]...@[else]...@[end match]
The control markup behaves the same as the native Python control
structure: The first @[case]
clause which matches is expanded
and the control finishes. An optional @[else]
clause can appear at
the end of the chain of cases and will be expanded if no previous
@[case]
clause matches; @[else]
is identical to @[case _]
.
Markup that is present between the @[match]
markup and the first
@[case]
is unconditionally expanded. Typically this will be markup
that expands to nothing, but it can involve a preamble if desired.
Example 36: Match controls
Source:
@# Markup here is expanded unconditionally.
@[match (100, 100)]@
(100, 100) is @
@[ case (0, 0)]@
the origin.
@[ case (0, y)]@
Y=@y.
@[ case (x, 0)]@
X=@x.
@[ case (x, y) if x == y]@
X=Y=@x.
@[ case (x, y)]@
X=@x, y=@y.
@[ else]@
not a point.
@[end match]@
Output:
(100, 100) is X=Y=100.
Warning
Since the @[match]
markup relies on the underlying Python
functionality, using this markup with a version of Python before 3.10
will result in a CompatibilityError
being raised.
Note
Match control markup was introduced in EmPy version 4.1.
Defined control markup: @[defined N]...@[end defined]
¶
Sometimes it’s useful to know whether a name is defined in either the locals or globals dictionaries. EmPy provides a dedicated markup for this purpose: defined control markup. It takes the following forms:
@[defined N]...@[end defined]
@[defined N]...@[else]...@[end defined]
When provided a name, it will expand the contained markup if that name
is defined in either the locals or globals. @[defined NAME]...@[end defined]
is equivalent to @[if 'NAME' in globals() or 'NAME' in locals()]...@[end if]
. An @[else]
clause is also supported; if
present, this will be expanded if the name does not appear in the
locals or globals. If no @[else]
clause is present and the name is
not defined, nothing will be expanded.
Example 37: Defined controls
Source:
@{cat = 'Boots'}@
Cat is @[defined cat]@cat@[else]not defined@[end defined].
Dog is @[defined dog]@dog@[else]not defined@[end defined].
Output:
Cat is Boots.
Dog is not defined.
Note
Defined control markup was introduced in EmPy version 4.0.
Def control markup: @[def F(...)]...@[end def]
¶
EmPy supports defining functions which expand EmPy code, not Python
code as with the standard def
Python statement. This is called
def control markup. It takes on the following form:
@[def F(...)]...@[end def]
Def control markup involves specifying the signature of the resulting
function (such as with the standard Python def
statement) and
encloses the EmPy code that the function should expand. It is then
defined in the interpreter’s globals/locals and can be called like any
other Python function.
It is best demonstrated with a simple example:
Example 38: Def controls
Source:
@# Define an EmPy-native function.
@[def element(name, symbol, atomicNumber, group)]@
Element @name (symbol @symbol, atomic number @atomicNumber) is a @group@
@[end def]@
@# Now use it.
@element('hydrogen', 'H', 1, 'reactive nonmetal').
@element('helium', 'He', 2, 'noble gas').
@element('lithium', 'Li', 3, 'alkali metal').
@element('beryllium', 'Be', 4, 'alkaline earth metal').
@element('boron', 'B', 5, 'metalloid').
@element('carbon', 'C', 6, 'reactive nonmetal').
Output:
Element hydrogen (symbol H, atomic number 1) is a reactive nonmetal.
Element helium (symbol He, atomic number 2) is a noble gas.
Element lithium (symbol Li, atomic number 3) is a alkali metal.
Element beryllium (symbol Be, atomic number 4) is a alkaline earth metal.
Element boron (symbol B, atomic number 5) is a metalloid.
Element carbon (symbol C, atomic number 6) is a reactive nonmetal.
Hint
The markup @[def FUNC(...)]DEFN@[end def]
is equivalent to the
following Python code:
def FUNC(...):
r"""DEFN"""
return empy.expand(r"""DEFN""", locals())
It simply defines a Python function with the provided signature, a
docstring indicating its EmPy definition, and the function calls the
expand
method on the pseudomodule/interpreter with the definition
and returns the results.
Tip
Functions defined with def control markup are callable Python objects
like any other. They can be called through any mechanism, whether
Python (f(...)
), through EmPy markup (@f(...)
), or even via
functional expression markup
(@f{...}
).
Diacritic markup: @^ CHAR DIACRITIC(S)
¶
EmPy provides a quick and convenient way to combine diacritics
(accents) to characters with diacritic markup. Diacritic markup
consists of the prefix @^
, followed by the base character, and then
either a single character representing the accent to apply or a
sequence of such characters enclosed in curly braces ({...}
).
The first character is the base character to combine diacritics with,
and the remaining characters (possibly more than one if the curly
braces form is used) are diacritic codes corresponding to Unicode
combining characters that can be combined (or just appended) to the
base character. These combining diacritics are simpler, more easily
entered characters that (at least in some cases) resemble the actual
desired combining character. For instance, '
(apostrophe)
represents the acute accent ◌́; `
(backquote) represents the grave accent ◌̀; ^
represents the circumflex
accent ◌̂, and so on:
Example 39: Diacritics
Source:
French: Voil@^a`, c'est ici que @^c,a s'arr@^e^te.
Spanish: Necesito ir al ba@^n~o ahora mismo.
Portuguese: Informa@^c,@^a~o @^e' poder.
Swedish: Hur m@^aonga kockar kr@^a:vs f@^o:r att koka vatten?
Vietnamese: Ph@^o{h?} b@^o` vi@^e^n ngon qu@^a'!
Esperanto: E@^h^o@^s^an@^g^e @^c^iu@^j^a@^u(de!
Shakespearean: All are punish@^e`d.
Output:
French: Voilà, c'est ici que ça s'arrête.
Spanish: Necesito ir al baño ahora mismo.
Portuguese: Informação é poder.
Swedish: Hur många kockar krävs för att koka vatten?
Vietnamese: Phở bò viên ngon quá!
Esperanto: Eĥoŝanĝe ĉiuĵaŭde!
Shakespearean: All are punishèd.
Tip
Curly braces can enclose zero or more characters representing
diacritics. If they enclose zero, the diacritic markup has no effect
(@^e{}
is no different from e
). If they enclose one, the results
are no different from not using curly braces (@^e{'}
and @^e'
are have identical results). Only when applying more than one
diacritic are curly braces required.
By default, the base character and diacritics will be combined with
NFKC normalization — this will, when possible, replace the base
character and its combiners with a single Unicode character
representing the combination, if one exists. Normalization is not
required (and may sometimes fail when a suitable combined form does
not exist); in these cases, your system’s Unicode renderer will cope
as best it can. To change the normalization type, use
-z/--normalization-form=F
(configuration variable: normalizationForm
). To disable normalization,
set it to the empty string.
The dictionary mapping all diacritic codes to combining characters is
stored in the diacritics
configuration variable.
This can be modified or completely replaced as desired. The values
can be integers (Unicode code point values), lists of integers, or
strings. They can also take the form of a 2-tuple, where the first
element is one of the above values and the second element is a
description string used for displaying in help topics.
See also
The list of all default diacritic codes is available in the
diacritics
help topic and is summarized
here.
Note
Diacritic markup was introduced in EmPy version 4.0.
Icon markup: @|...
¶
A customizable brief way to map “icon” keys — short, user-specified
strings — to arbitrary Unicode strings exists in the form of icon
markup. Icon markup is set off with @|...
and then followed by
an unambiguous, arbitrary-length sequence of characters (Unicode code
points) corresponding to one of its keys.
The icon keys can be any set of distinct strings, and are specified in
the icons
configuration variable whose keys are the
string keys as used in the markup, as well as all possible key
prefixes (more on this in a bit). The values can be integers (Unicode
code point values), lists of integers, strings, or objects which
implement a __str__
method. They can also take the form of a
2-tuple, where the first element is one of the above values and the
second element is a description string used for displaying in help
topics. An (arbitrary) default set are provided by default, but these
can be modified or completely replaced as desired.
Keys can be arbitrary length and can consist of whatever characters
are desired (including letters, numbers, punctuation, or even Unicode
characters). They are not delimited by whitespace; however, they must
be unambiguous, so if more than one key exists with the same prefixes
(say, #+
and #-
), a key cannot be defined as the common prefix
(#
) as this would be found first and would hide the longer prefixes.
Such a common prefix key should be set to the value None
(which
indicates to the parser that the icon is potentially valid but not yet
complete).
This validation is done automatically when the icon markup is first
used: The dictionary of icons is traversed and any common prefixes not
defined in the dictionary are set to None
. In the event that this
auto-validation may be expensive and the user wishes to do it manually
to avoid this step, specify the
--no-auto-validate-icons
command line option
(autoValidateIcons
configuration variable) to disable
it.
To customize icons, modify or replace the icons
configuration variable:
Tip
If you’re finding problems with icons being ambiguous, you can add
delimiters at the end of the icon key to ensure that they are
unambiguous. For example, the icons !
, !!
and !?
would normally
be ambiguous. However, wrapping them in, say, curly braces, will
remove the ambiguity: {!}
, {!!}
, {!?}
are unambiguous and can be
used as icon keys.
See also
The list of all valid icon keys is available in the icons
help topic
and is summarized here.
Warning
The default set of icons were chosen by the author for his convenience and to demonstrate what icon markup can do. It is expected that users using icon markup will modify (or more likely completely replace) the icons dictionary to their liking. Thus the default icons are subject to change.
Note
Icon markup was introduced in EmPy version 4.0.
Emoji markup: @:...:
¶
A dedicated emoji markup is available to translate Unicode emoji
names, and Unicode names more generally, into Unicode glyphs. Using
the markup is simple: Use the @:...:
syntax and put the name of the
emoji character between the colons. Since EmPy is often used in
wrapped text, any newlines in the emoji name will be replaced with
spaces.
By default it uses the builtin unicodedata.lookup
function call
which allow the lookup of any Unicode code point by name, not just
emoji. Whether names are case sensitive or not, or whether words are
separated by spaces or underscores or either, is module-dependent.
The builtin unicodedata
module (the fallback if no emoji-specific
modules are installed) is case insensitive and requires spaces, not
underscores:
Example 42: Emojis
Source:
Latin capital letter A: @:LATIN CAPITAL LETTER A:
Latin small letter O with diaeresis: @:latin small letter o with diaeresis:
White heavy check mark: @:WHITE HEAVY CHECK MARK:
Volcano: @:VOLCANO:
Output:
Latin capital letter A: A
Latin small letter O with diaeresis: ö
White heavy check mark: ✅
Volcano: 🌋
User-specified emoji can also be assigned to the
info.variable(‘emojis’) configuration variable; this will be checked
before any emoji modules are queried. The values of this dictionary
can be any string or object which implement a __str__
method, not
necessarily just a single Unicode code point. Emojis in the emojis
dictionary are case sensitive:
Example 43: Emojis, custom
Source:
@# What's with this guy and cats?
@{
empy.config.emojis['kittycat'] = '\U0001f408'
}@
This is a kitty cat: @:kittycat:
Output:
This is a kitty cat: 🐈
The emoji markup can also use third-party emoji modules if they are
present. These can be installed in the usual way with PyPI (e.g.,
python3 -m pip install emoji
) or any other preferred method. The
following emoji modules are supported:
Module |
Function |
Parameter |
Capitalization |
Spaces or underscores |
---|---|---|---|---|
|
|
|
lowercase |
underscores |
|
|
|
lowercase |
underscores |
|
|
|
lowercase |
underscores |
|
|
|
both |
spaces |
On first usage, each module is checked to see if it is present and is
then registered in the order listed above. When a lookup on a name is
performed, each module which is present is queried in order, and if it
finds the given name, that is used as output. If no modules find the
name, by default an error is generated, but this behavior can be
changed with the --ignore-emoji-not-found
command line
option.
The order in which modules are queried is also customizable with the
--emoji-modules
command line option; specify the
sequence of emoji module names to test separated by commas. Use the
--no-emoji-modules
command line option to only enable
the builtin unicodedata
module lookup, deactivating the use of any
custom modules which may be installed. And use
--disable-emoji-modules
to disable all emoji module
lookup; only the emojis
configuration variable will be consulted.
If you’re aware of other third-party emoji modules you’d like to see supported, contact the author.
Tip
It’s expected that the typical EmPy user will have at most one
third-party module installed, so no effort has been put in place to
avoid conflicts or redundancies regarding emoji names between them
other than specifying this desired lookup order. Choose a third-party
module that works for you, or just rely on the builtin unicodedata
lookup table.
If you’re relying on a third-party module to be present, you might want to have your EmPy code explicitly import that module so that if it’s missing, the dependency will be more clear.
Attention
Note that when markup which has starting and ending delimiters appears
alone on a line, the trailing newline will be rendered in the output.
To avoid these extra newlines, use a trailing @
to turn it
into whitespace markup which consumes that trailing newline, so e.g.
@:...:
followed by a newline becomes @:...:@
followed by a newline. This is
idiomatic for suppressing unwanted newlines. See here for
more details.
Note
Emoji markup was introduced in EmPy version 4.0.
Significator markup: @%[!]... NL
, @%%[!]...%% NL
¶
Significators are ways to perform distinctive assignments within
an EmPy system which are easily parsed externally; for instance, for
specifying metadata for an EmPy source document. In its simplest
form, it defines in a variable in the globals with the evaluation of a
Python value. The significator @%KEY VALUE
is equivalent to the
Python assignment statement __KEY__ = VALUE
.
The name of the assigned variable is preceded and ended with a double
underscore (__
). (This behavior can be changed with
configurations.) Note that the value assigned can be any Python
expression, not just a string literal:
Example 44: Significators, basics
Source:
@%title "A Tale of Two Cities"
@%author 'Charles Dickens'
@%year 1859
@%version '.'.join([str(x) for x in __import__('sys').version_info[0:2]])
The book is _@(__title__)_ (@__year__) by @__author__.
This version of Python is @__version__.
Output:
The book is _A Tale of Two Cities_ (1859) by Charles Dickens.
This version of Python is 3.10.
Whitespace is allowed between the @%
markup introducer and the key,
and any (non-newline) whitespace is allowed between the key and the
value. The ending newline is always consumed.
A variant of significator markup can span multiple lines. Instead of
using @%
and a newline to delimit the significator, use @%%
and
%%
followed by a newline:
Example 45: Significators, multiline
Source:
@%%longName "This is a potentially very long \
name which can span multiple lines." %%
@%%longerName """This is a triple quoted string
which itself contains newlines. Note the newlines
are preserved.""" %%
@%%longExpression [[1, 2, 3],
[4, 5, 6],
[7, 8, 9]] %%
Long name: @__longName__
Longer name: @__longerName__
Long expression: @__longExpression__
Output:
Long name: This is a potentially very long name which can span multiple lines.
Longer name: This is a triple quoted string
which itself contains newlines. Note the newlines
are preserved.
Long expression: [[1, 2, 3], [4, 5, 6], [7, 8, 9]]
Note
When using multiline significators, the value must still be a valid Python expression. So the significator
@%%bad 1 + 2 + 3 +
4 + 5 + 6 %%
is a syntax error due to the intervening newline. To correct this,
use a backslash character (\
) to escape the newline or enclose the
value expression in parentheses:
@%%good1 1 + 2 + 3 + \
4 + 5 + 6 %%
@%%good2 (1 + 2 + 3 +
4 + 5 + 6) %%
Two more subvariants of significator markup exists, one for each of
these two variants. Frequently significator values will just be
string literals and for uniformity users may wish to not deal with
full Python expressions. For these purposes, significator values can
be stringized, or treated merely as strings with no Python
evaluation. Simply insert a !
after the @%
or @%%
markup
introducer and before the name of the key:
Example 46: Significators, stringized
Source:
@# These values are all implicit strings.
@%!single This is on a single line.
@%%!multi This is on
multiple lines. %%
Single line: @__single__
xMultiple lines: @__multi__
Output:
Single line: This is on a single line.
xMultiple lines: This is on
multiple lines.
Finally, the values for both single and multiline significator markups
are optional. If the markup is not stringized, the value will be
None
; if stringized, it will be the empty string (''
):
Example 47: Significators, optional values
Source:
@%none
@%!empty
This is a None: @repr(__none__).
This is an empty string: @repr(__empty__).
Output:
This is a None: None.
This is an empty string: ''.
Hint
Significators can appear anywhere in an EmPy document, but typically are used at the beginning.
Tip
A compiled regular expression object is returned by the
significatorRe
configuration method and can be used to
systematically find all the significators in a given text.
The values of a non-stringinized significators can be any Python expression, so can include side effects from prior EmPy expansions. It’s best practice, however, to only have significator values depend on the value of previous significators, so that trimmed down processors can evaluate them without having to expand the entire document.
Note
Significator markup was introduced in EmPy version 1.2. Stringified and multiline variants were introduced in version 4.0.
Context markups¶
Contexts are objects which track the current progress of an EmPy interpreter through its source document(s) for the purposes of error reporting. This is handled automatically by the EmPy system, but they can be modified through the API or with context markup.
Note
Context markups were introduced in EmPy version 3.0.2.
Context name markup: @?... NL
¶
The context name markup can be used to change the current context
name with @?... NL
; it uses as the new name what follows on the
same line and consumes everything up to and including the newline.
Whitespace surrounding the context name is ignored.
Example 48: Context names
Source:
@?Test
This context is now: @empy.getContext().
Output:
This context is now: Test:2:22.
Context line markup: @!... NL
¶
The context line markup can be used to change the current context
line with @!... NL
; it uses as the new line what follows on the
same line and consumes everything up to and including the newline.
Whitespace surrounding the context name is ignored. If the remaining
text is not parseable as an integer, it is a parse error.
Example 49: Context lines
Source:
@!1000
This context is now: @empy.getContext().
Note that the line is 1001 since it's the next line after the markup.
Output:
This context is now: <example 49>:1001:22.
Note that the line is 1001 since it's the next line after the markup.
Extension markup: @((...))
, @[[...]]
, @{{...}}
, @<...>
, …¶
Markup can be provided with customizable user-defined meanings with extension markup. Use of these markups calls a method on an extension instance installed with the running interpreter. Once installed, an extension cannot be uninstalled. Each of these methods has the following signature:
method(contents: str, depth: int, locals: Optional[dict]) -> str
It takes the following arguments:
Argument |
Type |
Description |
---|---|---|
|
|
The contents inside the markup |
|
|
The number of opening and closing markup characters |
|
|
The locals dictionary or |
The return value is a string to serialize to the output. Of course, these methods can perform any other desired side effects. Encountering extension markup with no installed extension, or with an extension that has no correspondingly defined method is an error.
The following extension are available by default:
Markup |
Start |
Extension method name |
Depth |
---|---|---|---|
|
|
|
2+ |
|
|
|
2+ |
|
|
|
2+ |
|
|
|
1+ |
Extension markup is parsed so that the number of starting and ending
characters can be of any depth but the number of starting and ending
characters must match. For angle brackets extension markup, any depth
can be used. For the others, a depth of 2 or more is needed (since
only one character, e.g., @(...)
, is a different markup). For
additional user-specified markups, it is up to the user.
Example 50: Extensions
Source:
@{
import em
class Extension(em.Extension):
def parentheses(self, contents, depth, locals):
return '[{}] "{}" (depth {})'.format('parentheses', contents, depth)
def square_brackets(self, contents, depth, locals):
return '[{}] "{}" (depth {})'.format('square_brackets', contents, depth)
def curly_braces(self, contents, depth, locals):
return '[{}] "{}" (depth {})'.format('curly_braces', contents, depth)
def angle_brackets(self, contents, depth, locals):
return '[{}] "{}" (depth {})'.format('angle_brackets', contents, depth)
empy.installExtension(Extension())
}@
Parentheses: @((This is a test.))
Parentheses: @(((This is a test.)))
Parentheses: @((((This is a test.))))
Square brackets: @[[This is a test.]]
Square brackets: @[[[This is a test.]]]
Square brackets: @[[[[This is a test.]]]]
Curly braces: @{{This is a test.}}
Curly braces: @{{{This is a test.}}}
Curly braces: @{{{{This is a test.}}}}
Angle brackets: @<This is a test.>
Angle brackets: @<<This is a test.>>
Angle brackets: @<<<This is a test.>>>
Angle brackets: @<<<<This is a test.>>>>
Output:
Parentheses: [parentheses] "This is a test." (depth 2)
Parentheses: [parentheses] "This is a test." (depth 3)
Parentheses: [parentheses] "This is a test." (depth 4)
Square brackets: [square_brackets] "This is a test." (depth 2)
Square brackets: [square_brackets] "This is a test." (depth 3)
Square brackets: [square_brackets] "This is a test." (depth 4)
Curly braces: [curly_braces] "This is a test." (depth 2)
Curly braces: [curly_braces] "This is a test." (depth 3)
Curly braces: [curly_braces] "This is a test." (depth 4)
Angle brackets: [angle_brackets] "This is a test." (depth 1)
Angle brackets: [angle_brackets] "This is a test." (depth 2)
Angle brackets: [angle_brackets] "This is a test." (depth 3)
Angle brackets: [angle_brackets] "This is a test." (depth 4)
Extensions can be used to define entirely new markup in this way. The
first optional argument to the Extension
constructor is a dict or
list or 2-tuples mapping first characters to method names. If it is a
list, it will be added to the default mapping:
Example 51: Extensions, additions
Source:
@{
import em
class Extension(em.Extension):
def __init__(self):
super().__init__([('/', 'slashes')])
def parentheses(self, contents, depth, locals):
return '[{}] "{}" (depth {})'.format('parentheses', contents, depth)
def square_brackets(self, contents, depth, locals):
return '[{}] "{}" (depth {})'.format('square_brackets', contents, depth)
def curly_braces(self, contents, depth, locals):
return '[{}] "{}" (depth {})'.format('curly_braces', contents, depth)
def angle_brackets(self, contents, depth, locals):
return '[{}] "{}" (depth {})'.format('angle_brackets', contents, depth)
def slashes(self, contents, depth, locals):
return '[{}] "{}" (depth {})'.format('slashes', contents, depth)
empy.installExtension(Extension())
}@
Parentheses: @((This is a test.))
Square brackets: @[[This is a test.]]
Curly braces: @{{This is a test.}}
Angle brackets: @<This is a test.>
Slashes: @/This is a test./
Slashes: @//This is a test.//
Slashes: @///This is a test.///
Slashes: @////This is a test.////
Output:
Parentheses: [parentheses] "This is a test." (depth 2)
Square brackets: [square_brackets] "This is a test." (depth 2)
Curly braces: [curly_braces] "This is a test." (depth 2)
Angle brackets: [angle_brackets] "This is a test." (depth 1)
Slashes: [slashes] "This is a test." (depth 1)
Slashes: [slashes] "This is a test." (depth 2)
Slashes: [slashes] "This is a test." (depth 3)
Slashes: [slashes] "This is a test." (depth 4)
Warning
If the opening and closing characters are the same, the markup must
contain at least one other character in order to be recognized. For
instance, if an extension is added for @/.../
, the character
sequence @//
is not an extension markup with a depth of one
containing the empty string, it is the unfinished start of an
extension markup of at least depth two. @/ /
, with a space in
between, on the other hand, would be valid.
The first constructor argument can also be a dict, in which case the mapping is completely replaced:
Example 52: Extensions, replacement
Source:
@{
import em
class Extension(em.Extension):
def __init__(self):
super().__init__({
'((': 'parens',
'[[': 'brackets',
'{{': 'braces',
'<': 'angles',
'/': 'slashes',
})
def parens(self, contents, depth, locals):
return '[{}] "{}" (depth {})'.format('parens', contents, depth)
def brackets(self, contents, depth, locals):
return '[{}] "{}" (depth {})'.format('brackets', contents, depth)
def braces(self, contents, depth, locals):
return '[{}] "{}" (depth {})'.format('braces', contents, depth)
def angles(self, contents, depth, locals):
return '[{}] "{}" (depth {})'.format('angles', contents, depth)
def slashes(self, contents, depth, locals):
return '[{}] "{}" (depth {})'.format('slashes', contents, depth)
empy.installExtension(Extension())
}@
Parentheses: @((This is a test.))
Square brackets: @[[This is a test.]]
Curly braces: @[[This is a test.]]
Angle brackets: @<This is a test.>
Slashes: @/This is a test./
Output:
Parentheses: [parens] "This is a test." (depth 2)
Square brackets: [brackets] "This is a test." (depth 2)
Curly braces: [brackets] "This is a test." (depth 2)
Angle brackets: [angles] "This is a test." (depth 1)
Slashes: [slashes] "This is a test." (depth 1)
Note that the markup opening can either be one character or two. This
is so that some extension markup can exist alongside of existing
markup (e.g., @((...))
is extension markup whereas @(...)
is
expression markup).
Finally, you can manually install token class factories corresponding to your extension:
Example 53: Extensions, manual
Source:
@{
import em
class Extension(em.Extension):
def __init__(self):
super().__init__({})
def the_colons(self, contents, depth, locals):
return '[{}] "{}" (depth {})'.format('the_colons', contents, depth)
empy.installExtension(Extension())
factory = empy.config.getFactory()
factory.addToken(empy.config.createExtensionToken(';', 'the_colons', ':'))
}@
The colons: @;This is a test.:
Output:
The colons: [the_colons] "This is a test." (depth 1)
Attention
Note that when markup which has starting and ending delimiters appears
alone on a line, the trailing newline will be rendered in the output.
To avoid these extra newlines, use a trailing @
to turn it
into whitespace markup which consumes that trailing newline, so e.g.
@<...>
followed by a newline becomes @<...>@
followed by a newline. This is
idiomatic for suppressing unwanted newlines. See here for
more details.
Note
Extension markup was introduced in EmPy version 4.1.
Features¶
Various additional features are available in a running EmPy system.
Pseudomodule/interpreter¶
The pseudomodule/interpreter can be accessed by a running EmPy system by
referencing its name (which defaults to
empy
) in the globals:
Example 2: Pseudomodule sample
Source:
This version of EmPy is @empy.version.
The prefix in this interpreter is @empy.getPrefix() @
and the pseudomodule name is @empy.config.pseudomoduleName.
Do an explicit write: @empy.write("Hello, world!").
The context is currently @empy.getContext().
Adding a new global in a weird way: @empy.updateGlobals({'q': 789})@
Now q is @q!
You can do explicit expansions: @empy.expand("1 + 1 = @(1 + 1)").
q is @(empy.defined('q') ? 'defined' ! 'undefined').
Output:
This version of EmPy is 4.1.
The prefix in this interpreter is @ and the pseudomodule name is empy.
Do an explicit write: Hello, world!.
The context is currently <example 2>:5:26.
Adding a new global in a weird way: Now q is 789!
You can do explicit expansions: 1 + 1 = 2.
q is defined.
Important
The pseudomodule and interpreter are one and the same object; the
terms pseudomodule and interpreter are used interchangeably. The
interpreter exposes itself as the pseudomodule
empy
in a running EmPy system; this
pseudomodule is never imported explicitly.
Note
The pseudomodule was introduced in EmPy version 1.0.
Interpreter attributes and methods¶
Interpreter attributes¶
The following attributes are set on the pseudomodule after it is initialized.
version: str
The version of EmPy.
compat: list[str]
A list of strings indicating the “compatibility features” that were automatically enabled to support earlier versions of Python. Possible strings are:
Feature |
Description |
---|---|
|
No |
|
No |
|
|
|
Substituted an implementation of |
|
Substituted an implementation of |
|
Python was built with narrow Unicode (strings natively stored as UTF-16) |
executable: str
The path to the EmPy interpreter that is being used by the system (analogous to
sys.executable
).argv: list[str]
The arguments (analogous to
sys.argv
) used to start the interpreter. The first element is the EmPy document filename and the remaining elements are the arguments, if any. If no EmPy document was specified,<->
is used.config: Configuration
The configuration instance that the interpreter is using.
core: Core
The core used by this interpreter.
extension: Optional[Extension]
The extension, if any, used by this interpreter.
ok: bool
A Boolean indicating whether or not the interpreter is still active.
error: Optional[Error]
If an error occurs, the instance of it will be assigned to this attribute. When using
invoke
, this will determine whether or not a failure exit code is returned. No error is indicated byNone
.
Interpreter methods¶
These methods involve the interpreter directly.
Note
Most interpreter methods return None
so they can be called from
EmPy expression markup.
__init__(**kwargs)
The constructor. It takes the following keyword arguments (listed in alphabetical order), all of which have reasonable (or obvious) defaults:
Argument
Meaning
Default
argv
The system arguments to use
['<->']
callback
A custom callback to register [deprecated]
None
config
The configuration instance to use
default
core
The interpreter core to use
Core()
definerFunc
Interpreter
@[def]
definition function [deprecated]default (Python)
dispatcher
Dispatch errors or raise to caller?
True
evalFunc
Interpreter expression evaluation [deprecated]
eval
execFunc
Interpreter statement execution [deprecated]
exec
executable
The path to the EmPy executable
".../em.py"
extension
The extension to install on this interpreter
None
filespec
A 3-tuple of the input filename, output mode, and buffering
None
filters
The list of filters to install
[]
finalizers
The list of finalizers to install
[]
globals
The globals dictionary to use
{}
handler
The error handler to use
default
hooks
The list of hooks to install
[]
ident
The identifier of the interpreter (used for debugging)
None
immediately
Declare the interpreter ready immediately after initialization?
True
input
The input file to use for interactivity
sys.stdin
matcherFunc
Interpreter
@[match]
matcher function [deprecated]default (Python)
output
The output file to use
sys.stdout
root
The root interpreter context filename
'<root>'
serializerFunc
Interpreter serializer function [deprecated]
str
The ordering of the arguments does not matter. Missing arguments have reasonable defaults and unrecognized arguments are ignored.
Important
The order of the
Interpreter
constructor arguments has changed over time and is subject to change in the future, so you must use keyword arguments to prevent any ambiguity, e.g.:import em myConfig = em.Configuration(...) myGlobals = {...} myOutput = open(...) interp = em.Interpreter( config=myConfig, globals=myGlobals, output=myOutput, ...)
The allowed arguments are:
argv: Optional[list[str]]
The list of EmPy arguments. The first element is always the EmPy executable, with the remaining elements the actual arguments. If not specified, a reasonable default is used with no arguments.
callback: Optional[Callable]
The custom callback to register. Defaults to no callback.
Warning
Custom callbacks are deprecated in favor of extensions but registering them is supported for backward compatibility providing that extensions are not also being used.
config: Optional[Configuration]
The configuration to use for this interpreter. If not specified, a default configuration will be created and used.
Note
The current configuration of an interpreter can be modified while the interpreter is running and the changes will be effective as they occur. Configurations can also be shared between multiple interpreters if desired.
core: Core
The core to use with this interpreter. See Cores for more information.
dispatcher: bool | Callable
The dispatcher to use when an error is encountered. Dispatchers determine whether the error handler will be called (
True
), whether the error will be reraised to the caller (False
), or something else (a custom callable). See Error dispatchers for more information.executable: str
A string representing the path to the EmPy executable.
filespec: Optional[tuple[str, str, int | str]]
An optional 3-tuple of the filename, the file open mode, and the buffering mode of the EmPy script to be loaded. When using the command line arguments, this will be handled automatically.
filters: list[Filter]
A list of filters to install at startup. Defaults to none.
globals: Optional[dict]
The globals dictionary to use for this interpreter. If not specified, an empty dictionary will be created and used.
handler: Optional[Callable]
The error handler to set. If not specified, use the default handler.
hooks: list[Hook]
A list of hooks to install at startup. Defaults to none.
ident: Optional[str]
The name of the interpreter, printed when calling
repr
on the interpreter/pseudomodule object. Used only for debugging; defaults toNone
.immediately: bool
A Boolean which indicates whether or not the
ready
method will be called before the constructor exits. This is only relevant for hooks which implement theatReady
method. Defaults to true.input: file
The input file to use for interactive mode and pausing at end. Defaults to
sys.stdin
.output: file
The output file to use. Defaults to
sys.stdout
.root: str | (tuple[str] | tuple[str, int] | tuple[str, int, int]) | Context
The root context to use, which appears at the bottom of every Python error traceback (
-r/--raw-errors
). This can be a string, representing the filename; a tuple with between 1 and 3 parameters, with the full 3-tuple consisting of the name, the line number, and the column number; or an instance of theContext
class, which will be cloned.
Warning
The five
...Func
arguments (evalFunc
,execFunc
,definerFunc
,matcherFunc
andserializerFunc
) collectively define the behavior of the underlying interpreter (which defaults to Python). Alternates can be specified if desired; however, specifying these arguments in the constructor in this way is now deprecated. Instead, use cores. These arguments to the constructor are still supported for backward compatibility, however. Using cores is as simple as passing the functions to aCore
constructor and using that:import em core = em.Core( evaluate=evalFunc, execute=execFunc, serialize=serializerFunc, define=definerFunc, match=matcherFunc, ) interp = em.Interpreter(core=core, ...)
See also
The list of
Interpreter
constructor arguments is available in theconstructor
help topic and is summarized here.
__enter__()
/__exit__(*exc)
The interpreter presents a context manager interface and so can be used with the
with
Python control structure, e.g.:import em with em.Interpreter(...) as interp: ... manipulate interp here ...
reset([clearStacks])
Reset the interpreter to a pristine state. If
clearStacks
is true, reset the stacks as well.
ready()
Declare the interpreter ready for processing. This calls the
atReady
hook. By default this is called before the constructor exits, but the user can do this explicitly by passingFalse
to theimmediately
constructor argument and calling it when they wish to declare the interpreter ready.
shutdown()
Shutdown the interpreter. No further expansion must be done. This method is idempotent.
Important
When you create an interpreter, you must call its
shutdown
method when you are done. This is required to remove the proxy onsys.stdout
that EmPy requires for proper operation and restore your Python environment to the state it was before creating the interpreter. This can be accomplished by creating the interpreter in awith
statement — interpreters are also context managers — or by creating it and shutting it down in atry
/finally
statement.This is not needed when calling the
expand
global function; it creates and shuts down an ephemeral interpreter automatically.
Interpreter file-like methods¶
These methods mimic a file so the interpreter can be treated as a file-like object in APIs.
write(data: str)
Write the string data to the output stream.
writelines(lines: list[str])
Write the sequence of strings to the output stream.
flush()
Flush the output stream.
close()
Close the output stream. Note this will never close the fundamental output stream; it will only flush it, and it will only close other streams when they are not at the bottom of the stream stack.
serialize(thing: object)
Write a string version of the object to the output stream. This will reference
--none-symbol
(configuration variable:noneSymbol
) if the object isNone
.
Interpreter context methods¶
These methods manipulate the interpreter’s context stack.
identify() -> tuple
Get a 4-tuple of the current context, consisting of the filename, the line number, the column number, and the number of characters (Unicode code points) processed.
getContext() -> Context
Get the current context object.
newContext([name: str, [line: int, [column: int]]]) -> Context
Create a new context and return it.
pushContext(context: str | tuple | Context)
Push the given context on top of the context stack.
popContext()
Pop the top context off the context stack; do not return it.
setContext(context: str | tuple | Context)
Replace the context on the top of the context stack with the given context.
setContextName(name: str)
Set the top context’s name to the given value.
setContextLine(line: int)
Set the top context’s line to the given value.
setContextColumn(column: int)
Set the top context’s column to the given value.
setContextData([name: str, [line: int, [column: int]]])
Set the top context’s name, line, and/or column to the given value(s).
restoreContext(oldContext: str | tuple | Context)
Restore the top context on the stack to the given context.
Interpreter finalizer methods¶
These methods manipulate the interpreter’s finalizers.
clearFinalizers()
Clear all finalizers from this interpreter.
appendFinalizer(finalizer: Callable)
/atExit(finalizer: Callable)
Append the given finalizer to the finalizers list for this interpreter.
atExit
is an alias for backward compatibility.prependFinalizer(finalizer: Callable)
Prepend the given finalizer to the finalizers list for this interpreter.
Interpreter globals methods¶
These methods manipulate the interpreter’s globals.
getGlobals() -> dict
Get the current globals dictionary.
setGlobals(globals: dict)
Set the current globals dictionary,.
updateGlobals(moreGlobals: dict)
Update the current globals dictionary, adding this dictionary’s entries to it.
clearGlobals()
Clear the current globals dictionary completely.
saveGlobals([deep: bool])
Save a copy of the globals off to on the history stack. If deep is true, do a deep copy (defaults to false).
restoreGlobals([destructive: bool])
Restore the globals dictionary on the top of the globals history stack. If destructive is true (default), pop it off when done.
flattenGlobals([skipKeys: list[str]])
Flatten the interpreter namespace into the globals. If
skipKeys
is specified, skip over those keys; otherwise, use the defaults from the configuration.
Interpreter expansion methods¶
These methods are involved with markup expansion.
include(fileOrFilename, [locals, [name]])
Include the given EmPy (not Python) document (or filename, which is opened) and process it with the given optional locals dictionary and context name.
expand(data, [locals, [name], [dispatcher]]) -> str
Create a new context and stream to evaluate the EmPy data, with the given optional locals and context name and return the result. If the expansion raises an exception, by default (
False
) it will be raised up to the caller; setdispatcher
to true to have the interpreter handle it with its formal error handler mechanism. Setdispatcher
to another callable to do some custom dispatching. See Error dispatchers for more information.defined(name, [locals]) -> bool
Return a Boolean indicating whether the given name is present in the interpreter globals (or the optional locals, if provided).
lookup(name, [locals]) -> object
Lookup the value of a name in the globals (and optionally the locals, if provided) and return the value.
evaluate(expression, [locals, [write]]) -> object
Evaluate the given Python expression in the interpreter, with the given optional locals dictionary. If write is true, write it to the output stream, otherwise return it (defaults to false).
execute(statements, [locals])
Execute the given Python statements in the interpreter, with the given optional locals dictionary.
single(source, [locals]) -> Optional[object]
Execute the given Python expression or statement, with the given optional locals dictionary. This compiles the code with the
single
Python compilation mode which supports either. Return the result orNone
. This method is not used internally by the EmPy system and is not used by interpreter cores but is available for embedding.atomic(name, value, [locals])
Do an atomic assignment of the given name and value in the interpreter globals. If the optional locals dictionary is provided, set it in the locals instead.
assign(name, value, [locals])
Do a potentially complex assignment of the given name “lvalue” and “rvalue.” Unlike
atomic
,assign
can support tuple assignment.significate(key, [value, [locals]])
Declare a significator with the given key and optional value (if not specified, defaults to
None
). If the optional locals dictionary is provided, set it in the locals instead.quote(string) -> str
Given an EmPy string, return it quoted.
escape(string) -> str
Given an EmPy string, escape non-ASCII characters in it and return.
getPrefix() -> str
Get this interpreter’s prefix.
setPrefix(char)
Set this interpreter’s prefix.
Interpreter diversion methods¶
These methods manipulate the interpreter’s diversions.
stopDiverting()
Stop any current diversion.
createDiversion(name)
Create a new diversion with the given name but do not start diverting to it.
retrieveDiversion(name) -> Diversion
Get the diversion with the given name.
startDiversion(name)
Start diverting to a diversion with the given name, creating if it necessary.
playDiversion(name, [drop])
Play the diversion with the given name, optionally dropping it (default is true).
replayDiversion(name, [drop])
Play the diversion with the given name, optionally dropping it (default is false).
dropDiversion(name)
Drop the diversion with the given name without playing it.
playAllDiversions()
Play all diversions in sorted order by name, dropping them.
replayAllDiversions()
Replay all diversions in sorted order by name, leaving them in place.
dropAllDiversions()
Drop all diversions without playing them.
getCurrentDiversionName() -> Optional[str]
Get the name of the current diversion or
None
if there is no current diversion.getAllDiversionNames() -> list[str]
Get a list of the names of all diversions in sorted order.
isExistingDiversionName(name) -> bool
Is the given name the name of an existing diversion?
Interpreter filter methods¶
These methods manipulate the interpreter’s filters.
resetFilter()
Reset the filtering system so there are no filters.
getFilter() -> Filter
Get the top-most filter.
getLastFilter() -> Filter
Get the bottom-most filter.
setFilter(*filters)
Set the top-most filter(s) to the given filter chain, replacing any current chain. More than one filter can be specified as separate arguments.
prependFilter(filter)
Prepend the given filter to the current filter chain.
appendFilter(filter)
Append the given filter to the current filter chain.
setFilterChain(filters)
Set the filter chain to the given list of filters, replacing any current chain.
Interpreter core methods¶
These methods involve cores, an optional EmPy feature for overriding the core behavior of an EmPy interpreter. See cores for details.
hasCore() -> bool
Does this interpreter currently have a core installed? (A core is installed by the time initialized ends, regardless of whether or not a custom core has been specified.)
getCore() -> Core
Get the core currently inserted into this interpreter.
insertCore(core: Optional[Core])
Insert a core on this interpreter. If
core
isNone
, a default core will be created and used. This will detach any previous core.ejectCore()
Eject this interpreter’s core. Since cores may have a reference back to the interpreter, this will cut any cyclical link that may be present. This method is automatically called by the interpreter’s
shutdown
method. Note that once this method is called, no further expansion can be performed by this interpreter until a new core is installed. This method is idempotent.resetCore()
Reset the core for this interpreter to the default (Python).
Interpreter extension methods¶
hasExtension() -> bool
Does this interpreter already have an extension installed?
getExtension() -> Optional[Extension]
Get the extension installed in this interpreter, or
None
if there is no such extension.installExtension(Extension)
Install an extension in this interpreter. This can only be done once per interpreter.
callExtension(name: str, contents: str, depth: int)
Calls the extension as if extension markup had been encountered and serializes the result.
Interpreter hook methods¶
These methods manipulate the interpreter’s hooks.
invokeHook(_name, **kwargs)
Invoke the hooks associated with the given name and keyword arguments dictionary. This is the primary method called when hook events are invoked.
areHooksEnabled() -> bool
Are hooks currently enabled?
enableHooks()
Enable hooks.
disableHooks()
Disable hooks. Any existing hooks will not be called until
enableHooks
is called.getHooks() -> list[Hook]
Get the current list of hooks.
prependHook(hook)
Prepend the given hook to the list of hooks.
appendHook(hook)
Append the given hook to the list of hooks.
removeHook(hook)
Remove the given hook from the list of hooks.
clearHooks()
Clear the list of hooks.
Interpreter callback methods¶
These methods manipulate the interpreter’s custom callback. A callback is a callable object which takes one argument: the content to process.
hasCallback() -> bool
Does this interpreter have a custom callback registered?
getCallback() -> Optional[Callable]
Return the interpreter’s registered custom callback or
None
if none is registered.registerCallback(callback)
Register the given callback with the interpreter, replacing any existing callback.
deregisterCallback()
Remove the current interpreter’s registered callback, if any.
invokeCallback(contents)
Manually invoke the interpreter’s custom callback as if the custom markup
@<...>
were expanded.getExitCode() -> int
Get the exit code that will be returned by the process given the current state of the
error
attribute.
Warning
Custom callbacks are deprecated in favor of extensions but registering them and querying them is supported for backward compatibility, providing that extensions are not also being used.
Interpreter error handler methods¶
These methods manipulate the interpreter’s error handler. A handler is a callable object which takes three arguments: the type of the error, the error instance itself, and a traceback object.
defaultHandler(type, error, traceback)
The default EmPy error handler. This can be called manually by custom error handlers if desired.
getHandler() -> Callable
Get the current error handler, or
None
for the default.setHandler(handler, [exitOnError])
Set the error handler. If
exitOnError
is notNone
(defaults to false), also set the interpreter’s configuration’sexitOnError
configuration variable. This default is so that custom error handlers do not automatically exit which is usually the intent.invokeHandler(*exc)
Manually invoke the error handler. The arguments should be the 3-tuple of the return value of
sys.exc_info
as a single argument or as variable arguments, e.g.:interp.invokeHandler(sys.exc_info()) # or: *sys.exc_info()
Interpreter emoji methods¶
initializeEmojiModules([moduleNames])
Initialize the allowed emoji modules to use by name. If the names list is not specified, use the defaults.
getEmojiModule(moduleName) -> Module
Get the initialized module abstraction corresponding to the given module name.
getEmojiModuleNames() -> list[str]
Return the list of available emoji modules by name in their proper order.
substituteEmoji(text) -> str
Use the emoji facilities to lookup the given emoji name and return the result as if the emoji markup
@:...:
were expanded.
See also
The list of pseudomodule/interpreter attributes in methods is
available in the pseudo
help topic and is summarized
here.
Cores¶
By default, of course, EmPy expands expressions and statements through the underlying Python interpreter. But even this behavior is configurable, with interpreter cores. Cores can be used to completely change the underlying language that EmPy expands to (provided, naturally, that it can be implemented with Python).
Cores (represented with the base class Core
in the module em
)
support any language/system that can be encapsulated with a set of
globals, optional locals, and with the four following actions (methods
on the core object):
As with the Python languge, expressions which are evaluated must
return a value (or None
); statements which are executed do not
return a value. Beyond this definition, the meaning and purpose of
these operations in a custom core are completely configurable.
Cores can be created by either providing a set of callables to the
Core
constructor or by deriving a subclass from Core
and providing
overriding methods. The default implementation (Core()
) provides
the implementation for Python.
The default Core
class has the following constructor:
__init__(**kargs)
Create a core with optional keyword arguments. The arguments can be callables representing the different possible operations on a core, or an interpreter to attach to immediately after initialization:
Argument
Description
Default
evaluate
Evaluate an expression; return the result
eval
execute
Execute a statement; no return result
exec
serialize
Serialize an object; return the result
str
define
Functionality of
@[def ...]
markup; no return result(Python-specific)
match
Functionality of
@[match ...]
markup; no return result(Python-specific)
An optional
interp
argument is also allowed which will automatically attach this core to the given interpreter after intiialization.
The expected signatures of these methods are as follows:
evaluate(code: str, globals: dict, [locals: dict]) -> object
Evaluate the string expression given the current globals (and optional locals) and return the result.
execute(code: str, globals: dict, [locals: dict])
Execute the string statements given the current globals (and optional locals). The return value is ignored.
serialize(thing: object) -> str
Take a custom object and render it as a string for output.
define(signature: str, definition: str, [globals: dict, [locals: dict]])
Implement the behavior of the
def
control markup@[def SIGNATURE]DEFINITION@[end def]
, given the globals dictionary (defaults to the interpreter’s globals), and an optional locals dictionary. The meaning of the signature and the definition are arbitrary.match(expression: str, casePairs: list[tuple[str, list[Token]]], globals: dict, [locals: dict])
Implement the behavior of the
match
control markup@[match EXPRESSION]...@[end match]
. The first argument is the string expression to test.casePairs
is a list of 2-tuples representing thecase
statements. The first element of each tuple it the string representing the case test, and the second element is a list of tokens representing the markup to expand in the case it’s selected. Also passed are the globals dictionary to use (defaults to the interpreter’s globals), and an optional locals dictionary.
These methods are used by the interpreter to manage its core and need not be overridden by the user:
attach(interp: Interpreter)
Attach this core to the given interpreter. This is called automatically by the
attachCore
method in theInterpreter
class.detach()
Detach this core from any interpreter it might be attached to. This is called automatically by the
resetCore
andejectCore
methods on theInterpreter
class.
Cores are installed in an interpreter by either providing
one as the core
constructor argument or by
calling the attachCore
interpreter method. An interpreter’s core
can be reset to the default (Python implementation core) with the
resetCore
method, or can be manually removed by calling the
ejectCore
method (though this will leave the interpeter
non-functional until a new core is installed). An interpreter’s core
can also be modified dynamically while an interpreter is running.
Cores must be attached to the interpreter in order to make sure that
cores have a reference to the interpreter (needed for the default
define
method). This is done automatically when calling
attachCore
.
As mentioned earlier, cores can be created in one of two ways. First,
a subclass of Core
in the em
module can be instantiated and
provided to the interpreter:
import em
class MyCore(em.Core):
def evaluate(self, code, globals, locals=None): ...
def execute(self, code, globals, locals=None): ...
def serialize(self, thing): ...
def define(self, signature, definition, globals, locals=None): ...
def match(self, expression, casePairs, globals, locals=None): ...
core = MyCore(...)
interp = em.Interpreter(core=core)
Alternatively, the core methods can be implemented as standalone
callables and passed to the Core
constructor:
import em
def myEvaluate(code, globals, locals=None): ...
def myExecute(code, globals, locals=None): ...
def mySerialize(thing): ...
def myDefine(signature, definition, globals, locals=None): ...
def myMatch(expression, casePairs, globals, locals=None): ...
core = em.Core(
evaluate=myEvaluate,
execute=myExecute,
serialize=mySerialize,
define=myDefine,
match=myMatch,
)
interp = em.Interpreter(core=core)
For methods which are not defined in a subclass or specified as a callable in the constructor, the default (Python implementation core) will be used. The source code for the default definitions can be used as a guide.
Tip
The define
method and match
method correspond to the @[def ...]
and @[match ...]
markup, respectively. If these markups will not
be used by the user, these methods can be left unimplemented as they
will never be called. It’s good form, however, to have them raise
NotImplementedError
in that case just to make the intent clear.
Note
Interpreter cores were introduced in EmPy version 4.1.
Callbacks¶
Warning
Custom callbacks are deprecated in favor of extensions but registering them and querying them is supported for backward compatibility providing that extensions are not also being used.
Before the introduction of extensions, only one markup
was available for customization: the custom markup, @<...>
.
This meaning of this markup was provided with the use of a Python
callable object referred to as a custom callback, or just
“callback,” which can be set or queried using pseudomodule functions.
At most one custom callback can be registered, and once registered, it
cannot be deregistered or replaced.
When the custom markup @<...>
is encountered, the contents inside
the markup are passed to the current custom callback. Its return
value, if not None
, is then written to the output stream. The
custom callback may also perform operations with side effects
resulting in output, as well. If there are multiple opening angle
brackets, an equal number of closing angle brackets will be required
to match. This allows the embedding of <
and >
in the contents
passed to the callback.
The custom callback is a callable object which, when invoked, is
passed a single argument: a string representing the contents of what
was found inside the custom markup @<...>
. Only one custom
callback can be registered at a time.
To register a callback, call empy.registerCallback
. To see if there
is a callback registered, use empy.hasCallback
. To retrieve the
callback registered with the interpreter, use empy.getCallback
; if
no callback is registered, None
will be returned. Finally, to
invoke the callback explicitly just as if the custom markup had been
encountered, call empy.invokeCallback
. For instance, @<This text>
would be equivalent to the call @empy.invokeCallback("This text")
.
Invoking a callback (either explicitly with empy.invokeCallback
or
by processing a @<...>
custom markup) when no callback has been
registered is an error.
Example 54: Custom markup
Source:
@{
def callback(contents):
return contents.upper()
empy.registerCallback(callback)
}@
This will be in uppercase: @<This is a test>.
This will also contain angle brackets: @<<This is <also a> test>>.
Output:
This will be in uppercase: THIS IS A TEST.
This will also contain angle brackets: THIS IS <ALSO A> TEST.
Note
Custom markup was introduced in EmPy version 3.3 and transformed into extension markup in EmPy version 4.1.
Finalizers¶
Every interpreter must shutdown (by the shutdown
method being called
either implicitly or explicitly). As part of this process, the
interpreter runs any finalizers that may have been registered with
the interpreter.
Finalizers are callable objects which take zero arguments. The interpreter contains a list of finalizers which are called sequentially in order. If a finalizer raises an exception, any remaining finalizers will not be called that exception will be propagated up.
To add a finalizer, call either the appendFinalizer
or
preprendFinalizer
interpreter methods. To clear all finalizers,
call the clearFinalizers
method.
Note
atExit
is an alias for appendFinalizer
for backward compatibility.
Example 55: Finalizers
Source:
@# These are printed in reverse order.
@empy.appendFinalizer(lambda: empy.write("This is the last line.\n"))@
@empy.appendFinalizer(lambda empy=empy: empy.write("This is the penultimate line.\n"))@
@{
import em
class Finalizer:
def __init__(self, interp):
self.interp = interp
def __call__(self):
self.interp.write("This is the third to last line.\n")
finalizer = Finalizer(empy)
empy.appendFinalizer(finalizer)
}@
This is the first line.
Output:
This is the first line.
This is the third to last line.
This is the penultimate line.
This is the last line.
Tip
The finalizers are callables which take no arguments when they are called, so if they need a reference to the interpreter to do their work (as the example above), they need to arrange it with a closure, an implicit argument, or by implementing it as an instance which contains a reference to the interpreter. All three of these approaches are illustrated in the example above.
Warning
When using the expand
method or standalone function, the StringIO
object which is used to capture the output of the ephemeral
interpreter is processed before finalizers are handled. Thus
attempting to use finalizers to generate output will not work as
expected. Instead, create a dedicated interpreter:
import em
source = ... # the EmPy source to expand
with em.StringIO() as file:
with em.Interpreter(output=file, ...) as interp:
interp.string(source)
output = file.getvalue()
Note
Finalizers were introduced in EmPy version 2.1.
Diversions¶
EmPy supports an extended form of diversions, which are a mechanism for deferring and playing back output on demand, similar to the functionality included in m4. Multiple “streams” of output can be diverted (deferred) and played back (undiverted) in this manner. A diversion is identified with a name, which is any immutable object such an integer or string. Diversions can be played back multiple times (“replayed”) if desired. When recalled, diverted code is not resent through the EmPy interpreter (although a filter could be set up to do this).
By default, no diversions take place. When no diversion is in effect,
processing output goes directly to the specified output file. This
state can be explicitly requested at any time by calling the
empy.stopDiverting
function. It is always legal to call this
function, even when there is currently no active diversion.
When diverted, however, output goes to a deferred location which can
then be recalled later. Output is diverted with the
empy.startDiversion
function, which takes an argument that is the
name of the diversion. If there is no diversion by that name, a new
diversion is created and output will be sent to that diversion; if the
diversion already exists, output will be appended to that preexisting
diversion.
Output send to diversions can be recalled in two ways. The first is
through the empy.playDiversion
function, which takes the name of the
diversion as an argument. This plays back the named diversion, sends
it to the output, and then erases that diversion. A variant of this
behavior is the empy.replayDiversion
, which plays back the named
diversion but does not eliminate it afterwards; empy.replayDiversion
can be repeatedly called with the same diversion name, and will replay
that diversion repeatedly. empy.createDiversion
will create a
diversion without actually diverting to it, for cases where you want
to make sure a diversion exists but do not yet want to send anything
to it.
The diversion object itself can be retrieved with
empy.retrieveDiversion
. Diversions act as writable file-objects,
supporting the usual write
, writelines
, flush
, and close
methods. The data that has been diverted to them can be manually
retrieved in one of two ways; either through the asString
method,
which returns the entire contents of the diversion as a single string,
or through the asFile
method, which returns the contents of the
diversion as a readable (not writable) file-like object.
Diversions can also be explicitly deleted without playing them back
with the empy.dropDiversion
function, which takes the desired
diversion name as an argument.
Additionally there are three functions which will apply the above
operations to all existing diversions: empy.playAllDiversions
,
empy.replayAllDiversions
, and empy.dropAllDiversions
. The
diversions are handled in lexicographical order by their name. Also,
all three will do the equivalent of a empy.stopDiverting
call before
they do their thing.
The name of the current diversion can be requested with the
empy.getCurrentDiversionName
function; also, the names of all
existing diversions (in sorted order) can be retrieved with
empy.getAllDiversionNames
. empy.isExistingDiversionName
will
return whether or not a diversion with the given name exists.
When all processing is finished, the equivalent of a call to
empy.playAllDiversions
is done. This can be disabled with the
--no-auto-play-diversions
(configuration variable: autoPlayDiversions
) option.
Example 3: Diversions sample
Source:
This text is output normally.
@empy.startDiversion('A')@
(This text was diverted!)
@empy.stopDiverting()@
This text is back to being output normally.
Now playing the diversion:
@empy.playDiversion('A')@
And now back to normal output.
Output:
This text is output normally.
This text is back to being output normally.
Now playing the diversion:
(This text was diverted!)
And now back to normal output.
Note
Diversions were introduced in EmPy version 1.0.
Filters¶
EmPy also supports dynamic filters. Filters are put in place immediately before the final output file, and so are only invoked after all other processing has taken place (including interpreting and diverting). Filters take input, remap it, and then send it to the output. They can be chained together where a series of filters point to each other in series and then finally to the output file.
The current top-level filter can be retrieved with empy.getFilter
(or empy.getFirstFilter
). The last filter in the chain (the one
just before the underlying file) can be retrieved with
empy.getLastFilter
. The filter can be set with empy.setFilter
(which allows multiple arguments to constitute a chain). To append a
filter at the end of the chain (inserting it just before the
underlying output file), use empy.appendFilter
. To prepend it to
the top of the chain, use empy.prependFilter
. A filter chain can be
set directly with empy.setFilterChain
. And a filter chain can be
reset with empy.resetFilter
, removing all filters.
Filters are, at their core, simply file-like objects (minimally
supporting write
, flush
, and close
methods that behave in the
usual way) which, after performing whatever processing they need to
do, send their work to the next file-like object or filter in line,
called that filter’s “sink.” That is to say, filters can be “chained”
together; the action of each filter takes place in sequence, with the
output of one filter being the input of the next. The final sink of
the filter chain will be the output file. Additionally, filters
support a _flush
method (note the leading underscore) which will
always flush the filter’s underlying sink; this method should be not
overridden.
Filters also support two additional methods, not part of the
traditional file interface: attach
, which takes as an argument a
file-like object (perhaps another filter) and sets that as the
filter’s “sink” — that is, the next filter/file-like object in line.
detach
(which takes no arguments) is another method which flushes
the filter and removes its sink, leaving it isolated. Finally,
next
, if present, is an attribute which references the filter’s sink
— or None
, if the filter does not yet have a sink attached.
To create your own filter, you can create an object which supports the
above described interface, or simply derive from the Filter
class
(or one of its subclasses) in the emlib
module and override the
relevant methods.
Example 4: Filters sample
Source:
@{
# For access to the filter classes.
import emlib
}@
This text is normal.
@empy.appendFilter(emlib.FunctionFilter(lambda x: x.upper()))@
This text is in all uppercase!
@empy.appendFilter(emlib.FunctionFilter(lambda x: '[' + x + ']'))@
Now it's also surrounded by brackets!
(Note the brackets are around output as it is sent,
not at the beginning and end of each line.)
@empy.resetFilter()@
Now it's back to normal.
Output:
This text is normal.
THIS TEXT IS IN ALL UPPERCASE!
[NOW IT'S ALSO SURROUNDED BY BRACKETS!
(NOTE THE BRACKETS ARE AROUND OUTPUT AS IT IS SENT,
NOT AT THE BEGINNING AND END OF EACH LINE.)
]Now it's back to normal.
Note
Filters were introduced in EmPy version 1.3.
Hooks¶
The EmPy system allows for the registration of hooks with a running EmPy interpreter. Hooks are objects, registered with an interpreter, whose methods represent specific hook events. Any number of hook objects can be registered with an interpreter, and when a hook is invoked, the associated method on each one of those hook objects will be called by the interpreter in sequence. The method name indicates the type of hook, and it is called with a keyword list of arguments corresponding the event arguments.
To use a hook, derive a class from emlib.Hook
and override the
desired methods (with the same signatures as they appear in the base
class). Create an instance of that subclass, and then register it
with a running interpreter with the empy.addHook
function. A hook
instance can be removed with the empy.removeHook
function.
More than one hook instance can be registered with an interpreter; in
such a case, the appropriate methods are invoked on each instance in
the order in which they were appended. To adjust this behavior, an
optional prepend
argument to the empy.addHook
function can be used
dictate that the new hook should placed at the beginning of the
sequence of hooks, rather than at the end (which is the default).
Also there are explicit empy.appendHook
and empy.prependHook
functions.
All hooks can be enabled and disabled entirely for a given
interpreter; this is done with the empy.enableHooks
and
empy.disableHooks
functions. By default hooks are enabled, but
obviously if no hooks have been registered no hooks will be called.
Whether hooks are enabled or disabled can be determined by calling
empy.areHooksEnabled
. To get the list of registered hooks, call
empy.getHooks
. All the hooks can be removed with empy.clearHooks
.
Finally, to invoke a hook manually, use empy.invokeHook
.
For a list of supported hooks, see Hook methods or
the Hook
class definition in the emlib
module. (There is also an
AbstractHook
class in this module which does not have blank stubs
for existing hook methods in case a user wishes to create them
dynamically.)
For example:
Example 5: Hooks sample
Source:
@# Modify the backquote markup to prepend and append backquotes
@# (say, for a document rendering system, cough cough).
@{
import emlib
class BackquoteHook(emlib.Hook):
def __init__(self, interp):
self.interp = interp
def preBackquote(self, literal):
self.interp.write('`' + literal + '`')
return True # return true to skip the standard behavior
empy.addHook(BackquoteHook(empy))
}@
Now backquote markup will render with backquotes: @
@`this is now in backquotes`!
Output:
Now backquote markup will render with backquotes: `this is now in backquotes`!
Note
Hooks were originally introduced in EmPy version 2.0, much improved in version 3.2, and revamped again in version 4.0.
Hook methods¶
Hook at...
methods¶
These hooks are called when a self-contained event occurs.
atInstallProxy(proxy, new)
A
sys.stdout
proxy was installed. The Boolean valuenew
indicates whether or not the proxy was preexisting.atUninstallProxy(proxy, done)
A
sys.stdout
proxy was uninstalled. The Boolean valuedone
indicates whether the reference count went to zero (and so the proxy has been completely removed).atStartup()
The interpreter has started up.
atReady()
The interpreter has declared itself ready for processing.
atFinalize()
The interpreter is finalizing.
atShutdown()
The interpreter is shutting down.
atParse(scanner, locals)
The interpreter is initiating a parse action with the given scanner and locals dictionary (which may be
None
).atToken(token)
The interpreter is expanding a token.
atHandle(info, fatal, contexts)
The interpreter has encountered an error. The
info
parameter is a 3-tuple error (error type, error, traceback) returned fromsys.exc_info
,fatal
is a Boolean indicating whether the interpreter should exit afterwards, andcontexts
is the context stack.atInteract()
The interpreter is going interactive.
Hook context methods¶
pushContext(context)
This context is being pushed.
popContext(context)
This context has been popped.
setContext(context)
This context has been set or modified.
restoreContext(context)
This context has been restored.
Hook pre...
/post...
methods¶
The pre...
hooks are invoked before a token expands. The hook can
return a true value to indicate that it has intercepted the expansion
and the token should cancel native expansion. Not explicitly
returning anything, as in standard Python, is equivalent to returning
None
, which is a false value, which continues expansion:
Example 56: Hook pre...
methods
Source:
@{
import emlib
import sys
class Hook(emlib.Hook):
def __init__(self, interp):
self.interp = interp
def preString(self, string):
self.interp.write('[' + string + ']')
return True
empy.addHook(Hook(empy))
}@
@# Now test it:
@"Hello, world!"
Output:
["Hello, world!"]
Tip
It’s typical to want to have an instance of the interpreter/pseudomodule available to the hook, but it is neither done automatically nor is it required.
The post...
hooks are invoked after a non-intercepted token finishes
expanding. Not all pre...
hooks have a corresponding post...
hook. The post...
hooks take at most one argument (the result of
the token expansion, if applicable) and their return value is ignored.
preLineComment(comment)
,postLineComment()
The line comment
@#... NL
with the given text.preInlineComment(comment)
,postInlineComment()
The inline comment
@*...*
with the given text.preWhitespace(whitespace)
The whitespace token
@ WS
with the given whitespace.prePrefix()
The prefix token
@@
.preString(string)
,postString()
The string token
@'...'
, etc. with the given string.preBackquote(literal)
,postBackquote(result)
The backquote token
@`...`
with the given literal.preSignificator(key, value, stringized)
,postSignificator()
The significator token
@%... NL
, etc. with the given key, value and a Boolean indicating whether the significator is stringized.preContextName(name)
,postContentName()
The context name token
@?...
with the given name.preContextLine(line)
,postContextLine()
The context line token
@!...
with the given line.preExpression(pairs, except, locals)
,postExpression(result)
The expression token
@(...)
with the given if-then run pairs, the except run, and the locals dictionary (which may beNone
).preSimple(code, subtokens, locals)
,postSimple(result)
The simple expression token
@word
(etc.) with the given code, subtokens and locals.preInPlace(code, locals)
,postInPlace(result)
The in-place expression token
@$...$...$
with the given code (first section) and locals (which may beNone
).preStatement(code, locals)
,postStatement()
The statement token
@{...}
with the given code and locals (which may beNone
).preControl(type, rest, locals)
,postControl()
The control token
@[...]
of the given type, with the rest run and locals (which may be None).preEscape(code)
,postEscape()
The control token
@\...
with the resulting code.preDiacritic(code)
,postDiacritic()
The diacritic token
@^...
with the resulting code.preIcon(code)
,postIcon()
The icon token
@|...
with the resulting code.preEmoji(name)
,postEmoji()
The emoji token
@:...:
with the given name.preExtension(name, contents, depth)
,postExtension(result)
An extension with the given name, contents and depth was invoked.
preCustom(contents)
,postCustom()
The custom token
@<...>
with the given contents.
Hook before...
/after...
methods¶
The before...
and after...
hooks are invoked before and after (go
figure) mid-level expansion activities are performed. Any locals
argument indicates the locals dictionary, which may be None
.
If the expansion returns something relevant, it is passed as a
result
argument to the corresponding after...
method.
beforeProcess(command, n)
,afterProcess()
The given command (with index number) is being processed.
beforeInclude(file, locals, name)
,afterInclude()
The given file is being processed with the given name.
beforeExpand(string, locals, name)
,afterExpand(result)
empy.expand
is being called with the given string and name.beforeTokens(tokens, locals)
,afterTokens(result)
The given list of tokens is being processed.
beforeFileLines(file, locals)
,afterFileLines()
The given file is being read by lines.
beforeFileChunks(file, locals)
,afterFileChunks()
The given file is being read by buffered chunks.
beforeFileFull(file, locals)
,afterFileFull()
The given file is being read fully.
beforeString(string, locals)
,afterString()
The given string is being processed.
beforeQuote(string)
,afterQuote(result)
The given string is being quoted.
beforeEscape(string)
,afterEscape(result)
The given string is being escaped.
beforeSignificate(key, value, locals)
,afterSignificate()
The given key/value pair is being processed.
beforeCallback(contents)
,afterCallback()
The custom callback is being processed with the given contents.
beforeAtomic(name, value, locals)
,afterAtomic()
The given atomic variable setting with the name and value is being processed.
beforeMulti(names, values, locals)
,afterMulti()
The given complex variable setting with the names and values is being processed.
beforeImport(name, locals)
,afterImport()
A module with the given name is being imported.
beforeFunctional(code, lists, locals)
,afterFunctional(result)
A functional markup is with the given code and argument lists (of EmPy code) is being processed.
beforeEvaluate(expression, locals, write)
,afterEvaluate(result)
An evaluation markup is being processed with the given code and a Boolean indicating whether or not the results are being written directly to the output stream or returned.
beforeExecute(statements, locals)
,afterExecute()
A statement markup is being processed.
beforeSingle(source, locals)
,afterSingle(result)
A “single” source (either an expression or a statement) is being compiled and processed.
beforeFinalizer(final)
,afterFinalizer()
The given finalizer is being processed. If the
beforeFinalizer
hook returns true for a particular finalizer, then that finalizer will not be called.
See also
The list of hook methods is available in the hooks
help topic and is
summarized here.
Customization¶
The behavior of an EmPy system can be customized in various ways.
Command line options¶
EmPy uses a standard GNU-style command line options processor with
both short and long options (e.g., -p
or --prefix
). Short
options can be combined into one word, and options can have values
either in the next word or in the same word separated by an =
. An
option consisting of only --
indicates that no further option
processing should be performed.
EmPy supports the following options:
-V/--version
Print version information exit. Repeat the option for more details (see below).
-W/--info
Print additional information, including the operating system, Python implementation and Python version number.
-Z/--details
Print all additional details about the running environment, including interpreter, system, platform, and operating system release details.
-h/--help
Print basic help and exit. Repeat the option for more extensive help. Specifying
-h
once is equivalent to-H default
; twice to-H more
, and three or more times to-H all
(see below).-H/--topics=TOPICS
Print extended help by topic(s). Topics are a comma-separated list of the following choices:
Topic
Description
usage
Basic command line usage
options
Command line options
markup
Markup syntax
escapes
Escape sequences
environ
Environment variables
pseudo
Pseudomodule attributes and functions
variables
Configuration variable attributes
methods
Configuration methods
hook
Hook methods
controls
Named escapes (control codes)
diacritics
Diacritic combiners
icons
Icons
emojis
User-specified emojis (optional)
hints
Usage hints
topics
This list of topics
default
usage,options,markup,hints
andtopics
more
usage,options,markup,escapes,environ,hints
andtopics
all
usage,options,markup,escapes,environ,pseudo,config,controls,diacritics,icons,hints
As a special case,
-H
with no topic argument is treated as-H all
rather than error.-v/--verbose
The EmPy system will print debugging information to
sys.stderr
as it is doing its processing.-p/--prefix=CHAR
(environment variable:EMPY_PREFIX
, configuration variable:prefix
)Specify the desired EmPy prefix. It must consistent of a single Unicode code point (or character), or an empty string for no prefix (see below). Defaults to
@
.--no-prefix
Specify that EmPy use no prefix. In this mode, will only process text and perform no markup expansion. This is equivalent to specyfing
-p ''
.-q/--no-output
Use a null file for the output file.
-m/--pseudomodule=NAME
(environment variable:EMPY_PSEUDO
, configuration variable:pseudomoduleName
)Specify the name of the EmPy pseudomodule/interpreter. Defaults to
empy
.-f/--flatten
(environment variable:EMPY_FLATTEN
, configuration variable:doFlatten
)Before processing, move the contents of the
empy
pseudomodule into the globals, just as ifempy.flattenGlobals()
were executed immediately after starting the interpreter. This is the equivalent of executingfrom empy import *
(though since the pseudomodule is not a real module that statement is invalid). e.g.,empy.include
can be referred to simply asinclude
when this flag is specified on the command line.-k/--keep-going
Don’t exit immediately when an error occurs. Execute the error handler but continue processing EmPy tokens.
-e/--ignore-errors
Ignore errors completely. No error dispatcher or handler is executed and token processing continues indefinitely. Implies
-k/--keep-going
.-r/--raw-errors
(environment variable:EMPY_RAW_ERRORS
, configuration variable:rawErrors
)After logging an EmPy error, show the full Python traceback that caused it. Useful for debugging.
--brief-errors
When printing an EmPy error, show only its arguments and not its keyword arguments. This is in contrast to the default (verbose) where keyword arguments are shown.
--verbose-errors
When printing an EmPy error, show both its arguments and its (sorted) keyword arguments. This is the default.
-i/--interactive
(environment variable:EMPY_INTERACTIVE
, configuration variable:goInteractive
)Enter interactive mode after processing is complete by continuing to process EmPy markup from the input file, which is by default
sys.stdin
); this can be changed with theinput
interpreter attribute. This is helpful for inspecting the state of the interpreter after processing.-d/--delete-on-error
(environment variable:EMPY_DELETE_ON_ERROR
, configuration variable:deleteOnError
)If an error occurs, delete the output file; requires the use of the one of the output options such as
-o/--output=FILENAME
. This is useful when running EmPy under a build systemn such as GNU Make. If this option is not selected and an error occurs, the output file will stop when the error is encountered.-n/--no-proxy
(environment variable:EMPY_NO_PROXY
, configuration variable:useProxy
)Do not install a proxy in
sys.stdout
. This will make EmPy thread safe but writing tosys.stdout
will not be captured or processed in any way.--config=STATEMENTS
Perform the given configuration variable assignments. This option can be specified multiple times.
-c/--config-file=FILENAME
(environment variable:EMPY_CONFIG
)Read and process the given configuration file(s), separated by the platform-specific path delimiter (
;
on Windows,:
on other operating systems). This option can be specified multiple times.--config-variable=NAME
(configuration variable:configVariableName
)Specify the variable name corresponding to the current configuration when configuration files are processed. Defaults to
_
.-C/--ignore-missing-config
Ignore missing files while reading and processing configurations. By default, a missing file is an error.
-o/--output=FILENAME
Specify the file to write output to. If this argument is not used, final output is written to the underlying
sys.stdout
.-a/--append=FILENAME
Specify the file to append output to. If this argument is not used, final output is appended to the underlying
sys.stdout
.-O/--output-binary=FILENAME
Specify the file to write output to and open it as binary.
-A/--append-binary=FILENAME
Specify the file to append output to and open it as binary.
--output-mode=MODE
Specify the output mode to use.
--input-mode=MODE
Specify the input mode to use. Defaults to
'r'
.-b/--buffering
(environment variable:EMPY_BUFFERING
, configuration variable:buffering
)Specify the buffering to use. Use an integer to specify the maximum number of bytes to read per block or one of the following string values:
Name
Value
Description
full
-1
Use full buffering
none
0
Use no buffering
line
1
Use line buffering
default
16384
Default buffering
If the choice of buffering is incompatible with other settings, a
ConfigurationError
is raised. This option has no effect on interactive mode, assys.stdin
is already open. Defaults to 16384.--default-buffering
Use default buffering.
-N/--no-buffering
Use no buffering.
-L/--line-buffering
Use line buffering.
-B/--full-buffering
Use full buffering.
-P/--preprocess=FILENAME
Process the given EmPy (not Python) file before main document processing begins.
-Q/--postprocess=FILENAME
Process the given EmPy (not Python) file after main document processing begins.
-I/--import=MODULES
Import the given Python (not EmPy) module(s) into the interpreter globals before main document processing begins.
-D/--define=DEFN
Define the given variable into the interpreter globals before main document processing begins. This is executed as a Python assignment statement (
variable = ...
); if it does not contain a=
character, then the variable is defined in the globals with the valueNone
.-S/--string=STR
Define the given string variable into the interpreter globals before main document processing begins. The value is always treated as a string and never evaluated; if it does not contain a
=
character, then the variable is defined as the empty string (''
).-E/--execute=STATEMENT
Execute the given arbitrary Python (not EmPy) statement before main document processing begins.
-F/--file=FILENAME
Execute the given Python (not EmPy) file before main document processing begins.
-G/--postfile=FILENAME
Execute the given Python (not EmPy) file after main document processing begins.
-w/--pause-at-end
Prompt for a line of input after all processing is complete. Useful for systems where the window running EmPy would automatically disappear after EmPy exits (e.g., Windows). By default, the input file used is
sys.stdin
, so this will not work when redirecting stdin to an EmPy process. This can be changed with theinput
interpreter attribute.-l/--relative-path
(configuration variable:relativePath
)Prepend the location of the EmPy script to Python’s
sys.path
. This is useful when the EmPy scripts themselves import Python .py modules in that same directory.--replace-newlines
Replace newlines in (Python) expressions before evaluation.
--no-replace-newlines
Don’t replace newlines in (Python) expressions before evaluations. This is the default.
--ignore-bangpaths
Treat bangpaths as comments. By default, bangpaths (starting lines that begin with the characters
#!
) are treated as comments and ignored.--no-ignore-bangpaths
Do not treat bangpaths as comments. This is the opposite of the default.
--none-symbol
(configuration variable:noneSymbol
)The string to write when expanding the value
None
. Defaults toNone
, which will result in no output.--no-none-symbol
Do not write any preset string when expanding
None
; equivalent to settingnoneSymbol
toNone
.--expand-user
Expand user constructions (
~user
) in configuration file pathnames. This is the default.--no-expand-user
Do not expand user constructions (
~user
) in configuration file pathnames. By default they are expanded.--auto-validate-icons
Auto-validate icons when an icon markup is first used. This is the default. See below.
--no-auto-validate-icons
Do not auto-validate icons when an icon markup is first used. See below.
--starting-line
(configuration variable:startingLine
)Specify an integer representing the default starting line for contexts. Default is 1.
--starting-column
(configuration variable:startingColumn
)Specify an integer representing the default starting column for contexts. Default is 1.
--emoji-modules
(configuration variable:emojiModuleNames
)A comma-separated list of emoji modules to try to use for the emoji markup (
@:...:
). See below. Defaults toemoji,emojis,emoji_data_python,unicodedata
.--no-emoji-modules
Only use
unicodedata
as an emoji module; disable all other emoji modules.--disable-emoji-modules
Disable all emoji module usage; just rely on the
emojis
attribute of the configuration. See below.--ignore-emoji-not-found
When using emoji markup (
@:...:
), do not raise an error when an emoji is not found; just pass the:...:
text through.-u/--binary/--unicode
(environment variable:EMPY_BINARY
, configuration variable:useBinary
)Operate in binary mode; open files in binary mode and use the
codecs
module for Unicode support. This is necessary in older versions of Python 2.x.-x/--encoding=E
Specify both input and output Unicode encodings. Requires specifying both an input and an output file.
--input-encoding=E
(environment variable:EMPY_INPUT_ENCODING
, configuration variable:inputEncoding
)Specify the input Unicode encoding. Requires specifying an input file rather than
sys.stdout
.Note
Specifying a non-default encoding when using interactive mode (
sys.stdin
) raises aConfigurationError
.--output-encoding=E
(environment variable:EMPY_OUTPUT_ENCODING
, configuration variable:outputEncoding
)Specify the output Unicode encoding. Requires specifying an output file rather than
sys.stdout
.Note
Specifying a non-default encoding when using
sys.stdout
raises aConfigurationError
.-y/--errors=E
Specify both input and output Unicode error handlers.
--input-errors=E
(environment variable:EMPY_INPUT_ERRORS
, configuration variable:inputErrors
)Specify the input Unicode error handler.
Note
Specifying a non-default error handler when using interactive mode (
sys.stdin
) raises aConfigurationError
.--output-errors=E
(environment variable:EMPY_OUTPUT_ERRORS
, configuration variable:outputErrors
)Specify the output Unicode error handler.
Note
Specifying a non-default error handler when using
sys.stdout
raises aConfigurationError
.-z/--normalization-form=F
(configuration variable:normalizationForm
)Specify the Unicode normalization to perform when using the diacritics markup (
@^...
). Specify an empty string (''
) to skip normalization. Defaults toNFKC
for modern versions of Python and''
for very old versions of Python 2.x.--no-auto-play-diversions
(configuration variable:autoPlayDiversions
)Before exiting, do not automatically play back any remaining diversions. By default such diversions are played back.
--no-check-variables
(configuration variable:checkVariables
)When modifying configuration variables, normally the existence and types of these variables is checked and if it doesn’t exist or it is attempting to be assigned to an incompatible type, it will raise a
ConfigurationError
. To override this behavior, use this flag.--context-format
(configuration variable:contextFormat
)Specify the format for printing contexts. See below.
--success-code=N
(configuration variable:successCode
)Specify the exit code for the Python interpreter on success. Defaults to 0.
--failure-code=N
(configuration variable:failureCode
)Specify the exit code for the Python interpreter when a processing error occurs. Defaults to 1.
--unknown-code=N
(configuration variable:unknownCode
)Specify the exit code for the Python interpreter when an invalid configuration (such as unknown command line options) is encountered. Defaults to 2.
See also
The list of command line options is available in the options
help
topic and is summarized
here.
Environment variables¶
The following environment variables are supported:
EMPY_OPTIONS
Specify additional command line options to be used. These are in effect added to the start of the command line and parsed before any explicit command line options and processing begins.
For example, this will run the EmPy interpreter as if the
-r
and-d
command line options were specified:% export EMPY_OPTIONS='-r -d'; em.py ...
EMPY_CONFIG
(command line option:-c/--config-file=FILENAME
)Specify the configuration file(s) to process before main document processing begins.
EMPY_PREFIX
(command line option:-p/--prefix=CHAR
, configuration variable:prefix
)Specify the prefix to use when processing.
EMPY_PSEUDO
(command line option:-m/--pseudomodule=NAME
, configuration variable:pseudomoduleName
)Specify the name of the pseudomodule/interpreter to use when processing.
EMPY_FLATTEN
(command line option:-f/--flatten
, configuration variable:doFlatten
)If defined, flatten the globals before processing.
EMPY_RAW_ERRORS
(command line option:-r/--raw-errors
, configuration variable:rawErrors
)If defined, after an error occurs, show the full Python tracebacks of the exception.
EMPY_INTERACTIVE
(command line option:-i/--interactive
, configuration variable:goInteractive
)If defined, enter interactive mode by processing markup from
sys.stdin
after main document processing is complete.EMPY_DELETE_ON_ERROR
(command line option:-d/--delete-on-error
, configuration variable:deleteOnError
)If defined, when an error occurs, delete the corresponding output file.
EMPY_NO_PROXY
(command line option:-n/--no-proxy
, configuration variable:useProxy
)If defined, do not install a
sys.stdout
proxy.EMPY_BUFFERING
(command line option:-b/--buffering
, configuration variable:buffering
)Specify the desired file buffering.
EMPY_BINARY
(command line option:-u/--binary/--unicode
, configuration variable:useBinary
)If defined, use binary mode.
EMPY_ENCODING
Specify the desired input and output Unicode encodings.
EMPY_INPUT_ENCODING
(command line option:--input-encoding=E
, configuration variable:inputEncoding
)Specify the desired input Unicode encoding only.
EMPY_OUTPUT_ENCODING
(command line option:--output-encoding=E
, configuration variable:outputEncoding
)Specify the desired output Unicode encoding only.
EMPY_ERRORS
Specify the desired input and output Unicode error handler.
EMPY_INPUT_ERRORS
(command line option:--input-errors=E
, configuration variable:inputErrors
)Specify the desired input Unicode error handler.
EMPY_OUTPUT_ERRORS
(command line option:--output-errors=E
, configuration variable:outputErrors
)Specify the desired output Unicode error handler.
See also
The list of environment variables is available in the environ
help
topic and is summarized
here.
Note
Environment variables were first introduced in EmPy version 2.2, and revamped in version 4.0.
Configuration¶
Configurations are objects which determine the behavior of an EmPy
interpreter. They can be created with an instance of the
Configuration
class and have a set of attributes (configuration
variables) which can be modified. Most configuration variables
correspond to a command line option. The configuration instance also
contains supporting methods which are used by the interpreter which
can be overridden.
When configuration variables are modified, they are by default checked
to make sure have a known name and that they have the correct type; if
not, a ConfigurationError
will be raised. This behavior can be
disabled with --no-check-variables
(configuration variable: checkVariables
).
When a configuration is assigned to an interpreter, it exists as a
config
attribute of the empy
pseudomodule and can be modified by a
running EmPy system. Configurations can be shared between multiple
interpreters if desired.
Example 57: Configuration instances
Source:
@{
empy.config.prefix = '$'
}$
${
print("The EmPy prefix is now $, not @!")
}$
Output:
The EmPy prefix is now $, not @!
Tip
This example shows a quirk of changing configurations in the middle of
processing an EmPy document; the prefix changes from a @
to a $
by the end of the first statement markup, so a $
and a newline is
required to suppress the trailing newline; a @
would have been sent
to the output unchanged since it is no longer the prefix. Use
command line options, environment variables or configuration
files to avoid
issues like this, as they are processed before any EmPy document.
For changing the prefix, use -p/--prefix=CHAR
(environment variable: EMPY_PREFIX
, configuration variable: prefix
).
Configuration files¶
Configuration files are snippets of Python (not EmPy) code which
can be executed under an EmPy system to modify the current
configuration. By convention they have the extension .conf. though
this is not a requirement. Configuration files are processed before
any expansion begins and are specified with the -c/--config-file=FILENAME
(environment variable: EMPY_CONFIG
) command line option; a list of configuration files can be
specified with a :
delimiter (;
on Windows); the delimiter can be
specified with --path-separator
(configuration variable: pathSeparator
). A nonexistent
configuration file specified in this way is an error unless
-C/--ignore-missing-config
is specified.
When a configuration file is processed, its contents are executed in a Python (not EmPy) interpreter and then any resulting variable assignments are assigned to the configuration instance. So:
prefix = '$'
is a simple configuration file which will change the EmPy prefix to
$
.
Any resulting variable beginning with an underscore will be ignored. Thus these variables can be used as auxiliary variables in the configuration file. For example, this configuration file will define custom emojis for the numbered keycaps:
emojis = {}
for _x in range(10):
emojis[str(_x)] = '{}\ufe0f\u20e3'.format(_x)
Finally, when a configuration file is processed, the current
configuration instance is presented as a variable named _
(this can
be changed with --config-variable=NAME
(configuration variable: configVariableName
)). The
following example does the same as the previous example but uses the
dedicated variable:
_.emojis.update(((str(_x), '{}\ufe0f\u20e3'.format(_x)) for _x in range(10)))
Tip
To make a set of configuration files automatic loaded by EmPy, use the
EMPY_CONFIG
environment variable in your startup
shell:
% export EMPY_CONFIG=~/path/to/default.conf
To make a more general set of options available, set EMPY_OPTIONS
.
Configuration variables¶
The following configuration variables exist with the given types and
their corresponding command line options and environment variables.
Default values are shown after a =
sign. When a corresponding
command line option exists, See the command line
options for more detailed information.
name: str = 'default'
The name of this configuration. It is for organizational purposes and is not used directly by the EmPy system.
notes = None
Arbitrary data about this configuration. It can be anything from an integer to a string to a dictionary to a class instance, or its default,
None
. It is for organizational purposes and is not used directly by the EmPy system.prefix: str = '@'
(command line option:-p/--prefix=CHAR
, environment variable:EMPY_PREFIX
)The prefix the interpreter is using to delimit EmPy markup. Must be a single Unicode code point (character).
pseudomoduleName: str = 'empy'
(command line option:-m/--pseudomodule=NAME
, environment variable:EMPY_PSEUDO
)The name of the pseudomodule for this interpreter.
verbose: bool = False
If true, print debugging information before processing each EmPy token.
rawErrors: bool = False
(command line option:-r/--raw-errors
, environment variable:EMPY_RAW_ERRORS
)If true, print a Python traceback for every exception that is thrown.
exitOnError: bool = True
If true, exit the EmPy interpreter after an error occurs. If false, processing will continue despite the error.
ignoreErrors: bool = False
If true, all errors are ignored by the EmPy interpreter. Setting this to true also implies
exitOnError
is false.contextFormat: str = '%(name)s:%(line)d:%(column)d'
(command line option:--context-format
)The string format to use to render contexts. EmPy will automatically determine whether or not it should use the
%
operator or thestr.format
method with this format. See Context formatting for more details.goInteractive: bool = False
(command line option:-i/--interactive
, environment variable:EMPY_INTERACTIVE
)When done processing the main EmPy document (if any), go into interactive mode by running a REPL loop with
sys.stdin
. If such document is specified (i.e., EmPy is invoked with no arguments), go into interactive mode as well.deleteOnError: bool = False
(command line option:-d/--delete-on-error
, environment variable:EMPY_DELETE_ON_ERROR
)If an output file is chosen (e.g., with
-o/--output=FILENAME
or one of the other such options) and an error occurs, delete the output file. If this is set to true with output set tosys.stdout
, a ConfigurationError will be raised.doFlatten: bool = False
(command line option:-f/--flatten
, environment variable:EMPY_FLATTEN
)Flatten the contents of the
empy
pseudomodule into the globals rather than having them all under the pseudomodule name.useProxy: bool = True
(command line option:-n/--no-proxy
, environment variable:EMPY_NO_PROXY
)If true, install a proxy object for
sys.stdout
. This should be true if any output is being done viaprint
orsys.stdout.write
.relativePath: bool = False
(command line option:-l/--relative-path
)If true, the directory of the EmPy script’s path will be prepended to Python’s
sys.path
.buffering: int = 16384
(command line option:-b/--buffering
, environment variable:EMPY_BUFFERING
)Specify the buffering for the input and output files.
replaceNewlines: bool = False
If true, newlines in emoji names, Unicode character name escape markup, and code evaluation will be changed to spaces. This can help when writing EmPy with a word-wrapping editor.
ignoreBangpaths: bool = True
If true, a bangpath (the first line of a file which starts with
#!
) will be treated as an EmPy comment, allowing the creation of EmPy executable scripts. If false, it will not be treated specially and will be rendered to the output.noneSymbol: Optional[str] = None
(command line option:--none-symbol
)When an EmPy expansion evaluates to None (e.g.,
@(None)
), this is the string that will be rendered to the output stream. If set toNone
(the default), no output will be rendered.missingConfigIsError: bool = True
If a configuration file is specified with
-c/--config-file=FILENAME
but does not exist, if this variable is true an error will be raised.pauseAtEnd: bool = False
When done processing EmPy files, read a line from
sys.stdin
before exiting the interpreter. This can be useful when testing under consoles on Windows.startingLine: int = 1
(command line option:--starting-line
)The line to start with in contexts.
startingColumn: int = 1
(command line option:--starting-column
)The column to start with in contexts.
significatorDelimiters: tuple = ('__', '__')
A 2-tuple of strings representing the prefix and suffix to add to significator names in order to determine what name to give them in the globals.
emptySignificator: object = None
The default value to use for non-stringized significators.
autoValidateIcons: bool = True
When icons are used with a custom dictionary, a preprocessing phase needs to be done to make sure that all icon starting substrings are marked in the
icons
dictionary. If this variable is false, this extra processing step will not be done; this is provided if the user wants to specify their own properly-validated icons dictionary and wishes to avoid a redundant step.emojiModuleNames: list[str] = ['emoji', 'emojis', 'emoji_data_python', 'unicodedata']
(command line option:--emoji-modules
)The list of names of supported emoji modules that the EmPy system will attempt t use at startup.
emojiNotFoundIsError: bool = True
If true, a non-existing emoji is an error.
useBinary: bool = False
(command line option:-u/--binary/--unicode
, environment variable:EMPY_BINARY
)If true, open files in binary mode.
inputEncoding: str = 'utf-8'
(command line option:--input-encoding=E
, environment variable:EMPY_INPUT_ENCODING
)The file input encoding to use. This needs to be set before files are opened to take effect.
outputEncoding: str = 'utf-8'
(command line option:--output-encoding=E
, environment variable:EMPY_OUTPUT_ENCODING
)The file output encoding to use. This needs to be set before files are opened to take effect.
inputErrors: str = 'strict'
(command line option:--input-errors=E
, environment variable:EMPY_INPUT_ERRORS
)the file input error handler to use. This needs to be set before files are opened to take effect.
outputErrors: str = 'strict'
(command line option:--output-errors=E
, environment variable:EMPY_OUTPUT_ERRORS
)The file output error handler to use. This needs to be set before files are opened to take effect.
normalizationForm: str = 'NFKC'
(command line option:-z/--normalization-form=F
)The normalization form to use when applying diacritic combiners. Set to
None
or''
in order to skip normalization.autoPlayDiversions: bool = True
(command line option:--no-auto-play-diversions
)If diversions are extant when an interpreter is ready to exist, if this variable is true then those diversions will be undiverted to the output stream in lexicographical order by name.
expandUserConstructions: bool = True
If true, when processing configuration files, call
os.path.expanduser
on each filename to expand~
and~user
constructions.configVariableName: str = '_'
(command line option:--config-variable=NAME
)When processing configuration files, the existing configuration object can be referenced as a variable. This indicates its name.
successCode: int = 0
(command line option:--success-code=N
)The exit code to return when a processing is successful.
failureCode: int = 1
(command line option:--failure-code=N
)The exit code to return when an error occurs during processing.
unknownCode: int = 2
(command line option:--unknown-code=N
)The exit code to return when a configuration error is found (and processing never starts).
checkVariables: bool = True
(command line option:--no-check-variables
)If true, configuration variables will be checked to make sure they are known variables and have the proper type on assignment.
pathSeparator: str = ';'
(Windows) or':'
(others)--path-separator
)The path separator to use when specifying multiple filenames with
-c/--config-file=FILENAME
. Defaults to;
on Windows and:
on other platforms.duplicativeFirsts: list[str] = ['(', '[', '{']
The list of first markup characters that may be duplicated to indicate variants. For instance,
@(...)
is expression markup, but@((...))
is parenthesis extension markup.controls: dict = {...}
The controls dictionary used by the named escape markup.
diacritics: dict = {...}
The diacritic combiners dictionary used by the diacritic markup.
icons: dict = {...}
The icons dictionary used by the icon markup.
emojis: dict = {...}
The custom emojis dictionary which is referenced first by the emoji markup. Defaults to an empty dictionary.
See also
The list of configuration variables is available in the variables
help
topic and is summarized
here.
Note
Configuration objects were introduced in EmPy version 4.0; previously an underused options dictionary was introduced in version 2.2.2.
Configuration methods¶
The following methods are supported by configuration instances:
__init__(**kwargs)
The constructor. Takes a set of keyword arguments that are then set as attributes in the configuration instance. So
config = em.Configuration(prefix='$')
is a shorter form of
config = em.Configuration() config.prefix = '$'
isInitialized() -> bool
Has this instance been initialized? Before initialization, no typechecking is done even if
checkVariables
is set.check(inputFilename, outputFilename)
Check the file settings against these filenames and raise a
ConfigurationError
is there appears to be an inconsistency.has(name) -> bool
Is this name an existing configuration variable?
get(name, default=None) -> bool
Get the value of this configuration variable or return this default if it does not exist.
set(name, value)
Set the configuration variable to the given value.
update(**kwargs)
Set a series of configuration variables via a set of keyword arguments.
clone(deep=False) -> Configuration
Clone this configuration and return it. If
deep
is true, make it a deep copy.run(statements)
Execute a series of configuration commands.
load(filename, required=None)
Load and execute a configuration file. If
required
is true, raise an exception; if false, ignore; ifNone
, use the default for this configuration.path(path, required=None)
Load and execute one or more configuration files separated by the path separator.
required
argument is the same as forload
above.hasEnvironment(name) -> bool
Is the given environment variable defined, regardless of its value?
environment(name, default=None, blank=None)
Get the value of the environment variable. If it is not defined, return
default
; if it is defined but is empty, returnblank
.hasDefaultPrefix() -> bool
Is the
prefix
configuration variable set to the default?has{Full|No|Line|Fixed}Buffering() -> bool
Is buffering set to full, none, line, or some fixed value, respectively?
createFactory([tokens]) -> Factory
Create a token factory from the list of token classes and return it. If
tokens
is not specified, use the default list.adjustFactory()
Adjust an existing factory to take into account a non-default prefix.
getFactory([tokens], [force])
Get a factory, creating one if one has not yet been created, with the given
tokens
list (if not specified, a default list will be used). Ifforce
is true, then create a new one even if one already exists.resetFactory()
Clear the current factory, if any.
createExtensionToken(first: str, name: str, [last: str])
Create a new extension token class with the first character
first
and with method namename
. Iflast
is specified, use that as the last character; otherwise, guess for a closed form with a default offirst
.hasBinary() -> bool
Is binary (formerly called Unicode) support enabled?
enableBinary([major, minor])
Enable binary support, conditionally if
major
andminor
(the major and minor versions of Python) are specified and binary support is needed for this version of Python.disableBinary()
Turn off binary/Unicode support.
isDefaultEncodingErrors([encoding, errors, asInput]) -> bool
Are both the file encoding and file error handler the default? Check for input if
asInput
is true, otherwise check for output.getDefaultEncoding([default]) -> str
Get the current default encoding, overriding with
default
if desired.open(filename, mode=None, buffering=-1, encoding=None, errors=None, expand=None) -> file
The main wrapper around the
open
/codecs.open
call, allowing for seamless file opening in both binary and non-binary mode across all supported Python versions.significatorReString() -> str
Return a regular expression string that will match significators in EmPy code with this configuration’s prefix.
Hint
It can be used in Python like this:
data = open('script.em', 'r').read() for result in empy.config.significatorRe().findall(data): string2, key2, value2, string1, key1, value1 = result if key1: print("Single line significator: {} = {}{}".format( key1, value1, ' (stringized)' if string1 else '')) else: # key2 print("Multi-line significator: {} = {}{}".format( key2, value2, ' (stringized)' if string2 else ''))
significatorRe([flags]) -> re.Pattern
Return a compiled regular expression pattern object for this configuration’s prefix. Override the
re
flags
if desired.significatorFor(key) -> str
Return the significator variable name for this significator key.
setContextFormat(rawFormat)
Set the context format for this configuration. See context formatting.
renderContext(context) -> str
Render the given context using the existing context format string.
calculateIconsSignature() -> tuple
Calculate the icons signature to try to detect any accidental changes.
signIcons()
Calculate the icons signature and update the configuration with it.
transmogrifyIcons([icons])
Process the icons dictionary and make sure any keys’ prefixes are backfilled with
None
values. This is necessary for the functioning of the icon markup. This method will be called automatically unlessautoValidateIcons
is false.validateIcons([icons])
Check whether the icons have possibly changed and transmogrify them if necessary.
initializeEmojiModules([moduleNames])
Scan for existing emoji modules and set up the appropriate internal data structures. Use the list of module names in the configuration if
moduleNames
is not specified.substituteEmoji(text) -> str
Perform emoji substitution with the detected emoji modules.
isSuccessCode(code) -> bool
Is this exit code a success code?
isExitError(error) -> bool
Is this exception instance an exit error rather than a real error?
errorToExitCode(error) -> int
Return an appropriate exit code for this error.
isNotAnError(error) -> bool
Does this exception instance not represent an actual error?
formatError(error[, prefix, suffix]) -> str
Return a string representing the details of the given exception instance, with an optional prefix and suffix.
See also
The list of configuration methods is available in the methods
help
topic and is summarized
here.
Error handling¶
Error dispatchers¶
When an error occurs in an EmPy system, first an error dispatcher is invoked. The purpose of the dispatcher is to determine at a high-level what should be done about the error. A dispatcher is a zero-argument callable which primarily determines whether the error should be handled by the running interpreter, whether it should be raise to the parent caller rather than handled by the interpreter, or some other custom behavior.
When specified in the Interpreter
constructor or one
of the high-level interpreter methods (e.g., file
or string
), it
can take on a few special values:
Value |
Meaning |
Corresponding method |
---|---|---|
|
Use interpreter default |
— |
|
Interpreter should handle error |
|
|
Interpreter should reraise error |
|
Note
For standalone interpreters and its high-level methods, the default
dispatcher is True
(dispatch
); that is, the interpeter will handle
the error itself. When calling the expand
interpreter method or the
global expand
function, the dispatcher is False
(reraise
); in
other words, calls to expand
will result in any occurring errors
being raised to the caller rather than handled by the interpteter.
Note
Error dispatchers were introduced in EmPy version 4.0.1.
Error handlers¶
Once an error is dispatched to the interpteter, it is handled by an
error handler. An error handler is a callable object that will
respond to the error and take any necessary action. If no
user-specified error handler is set, the default error handler is
used, which prints a formatted EmPy error message to sys.stderr
.
An error handler is a callable object with the following signature:
handler(type, error, traceback) -> bool
It takes the error type, the error instance, and the traceback object
corresponding to an exception (a tuple of which is the return value of
sys.exc_info()
) and returns an optional Boolean. If the return
value is true, the default handler will also be invoked after the
error handler is called. (Not explicitly returning anything will
implicitly return None
, which is a false value.)
The current error that the interpreter has encountered is set in the
interpreter’s error
attribute (with None
indicating no error).
The error handler can manually set this attribute to None
to clear
the error if desired.
After the error handler(s) have been called, the interpreter will then
decide how to resolve the error. If the error
attribute of the
interpreter is still non-None
and the configuration variable
exitOnError
is true (option: -k/--keep-going
), the
interpreter will exit. If the error
attribute is None
, it will
continue running.
If the ignoreErrors
configuration variable (option:
-e/--ignore-errors
) is true, then no error dispatchers or error
handlers will be called.
Note
Error handlers were introduced in EmPy version 4.0.
Error classses¶
The following error classes are used by EmPy:
Class |
Base class |
Meaning |
---|---|---|
|
|
Base error class |
|
|
An error involving inconsistent settings has occurred |
|
|
An error involving diversions has occurred |
|
|
An error involving filters has occurred |
|
|
An error involving cores has occurred |
|
|
An error involving extensions has occurred |
|
|
A stack has underflowed (internal error) |
|
|
An unknown emoji was requested |
|
|
An old-style string error (used internally) |
|
|
An error invoking the interpreter has occurred |
|
|
An error involving a bad configuration has occurred |
|
|
An error involving backward compatibility has occurred |
|
|
A requested configuration file was missing |
|
|
Invalid EmPy syntax was encountered |
|
|
Invalid EmPy syntax was encountered (but may be resolved by reading further data) |
Reference¶
The following reference material is available:
Getting version and debugging information¶
To print the version of EmPy you have installed, run:
% em.py -V # or: --version Welcome to EmPy version 4.1.
To print additional information including the Python implementation and version, operating system, and machine type, run:
% em.py -W # or: --info Welcome to EmPy version 4.1, in CPython/3.10.12, on Linux (POSIX), with x86_64.
For diagnostic details (say, to report a potential problem to the developer), run:
% em.py -Z # or: --details Welcome to EmPy version 4.1, in CPython/3.10.12, on Linux (POSIX), with x86_64. Details: - basic/implementation: CPython - basic/machine: x86_64 - basic/os: POSIX - basic/system: Linux - basic/version: 3.10.12 ...
Examples and testing¶
For quick examples of EmPy code, check out the examples throughout this document. For a more expansive tour of examples illustrating EmPy features, check out tests/sample/sample.em. For a real-world example, check out README.md.em, which is the EmPy source file from which this documentation is generated.
EmPy has an extensive testing system. (If you have EmPy installed via an operating system package that does not include the test system and you wish to use it, download the tarball.)
EmPy’s testing system consists of the shell script test.sh and two directories: tests and suites. The tests directory contains the unit/system tests, and the suites directory contains files with lists of tests to run. The test.sh shell script will run with any modern Bourne-like shell.
Tests can be run changing to the directory where test.sh and both the
tests and suites directories are located, and then executing
./test.sh
followed by the tests desired to be run following on the
command line. For example, this runs a quick test:
% ./test.sh tests/sample/sample.em tests/sample/sample.em (python3) [PASS] PASSES: 1/1 All tests passed (python3).
Specifying a directory will run all the tests contained in that directory and all its subdirectories:
% ./test.sh tests/common/trivial tests/common/trivial/empty.em (python3) [PASS] tests/common/trivial/long.em (python3) [PASS] tests/common/trivial/medium.em (python3) [PASS] tests/common/trivial/short.em (python3) [PASS] tests/common/trivial/short_no_newline.em (python3) [PASS] PASSES: 5/5 All tests passed (python3).
Warning
The tests directory contains a superset of all tests for Python
versions, so running all the tests with ./test.sh tests
will
generate test failures.
Suites can be run by using the @
character before the filename. A
suite is a list of tests, one per line, to run. Blank lines and lines
starting with #
are ignored:
% cat suites/default # Run tests for Python versions from 3.0 up. tests/common tests/modern tests/python3 tests/sample
% ./test.sh @suites/default tests/common/callbacks/deregister.em (python3) [PASS] tests/common/callbacks/get_none.em (python3) [PASS] tests/common/callbacks/get_one.em (python3) [PASS] ... PASSES: 433/433 All tests passed (python3).
To test a version of Python other than the default (that is, other
than a Python 3.x interpreter named python3
), specify it with the
-p
option to the test script and use that version’s test suite. To
test CPython 2.7, for instance:
% ./test.sh -p python2.7 @suites/python2.7 tests/common/callbacks/deregister.em (python2.7) [PASS] tests/common/callbacks/get_none.em (python2.7) [PASS] tests/common/callbacks/get_one.em (python2.7) [PASS] ...
Suites for all supported interpreters and versions are provided. For example, if you have PyPy 3.10 installed:
% ./test.sh -p pypy3.10 @suites/pypy3.10 tests/common/callbacks/deregister.em (pypy3.10) [PASS] tests/common/callbacks/get_none.em (pypy3.10) [PASS] tests/common/callbacks/get_one.em (pypy3.10) [PASS] ...
To only report errors (“quiet mode”), use the -q
option:
% ./test.sh -q @suites/default PASSES: 433/433 All tests passed (python3).
For more information about the testing tool, run:
% ./test.sh -h # or: --help Usage: ./test.sh [<option>...] [--] (<file> | <directory> | @<suite>)... Run one or more EmPy tests, comparing the results to exemplars, and return an exit code indicating whether all tests succeeded or whether there were some failures. If no tests are specified, this help is displayed. Test filenames, directory names, and suite names cannot contain spaces. ...
Note
A simple benchmark test system was introduced in EmPy version 2.1, and was expanded to a full unit and system test suites for all supported versions of Python in EmPy version 4.0.
Embedding EmPy¶
EmPy can be easily embedded into your Python programs. Simply ensure
that the em.py file is available in the PYTHONPATH
and import em
as a module:
import em
print(em)
To embed an interpreter, create an instance of the Interpreter
class. The interpreter constructor requires keyword arguments, all
with reasonable defaults; see here for the list. One
important argument to an interpreter is a
configuration, which, if needed, should be
constructed first and then passed into the interpreter. If no
configuration is specified, a default instance will be created and
used:
import em
config = em.Configuration(...)
interp = em.Interpreter(config=config, ...)
Then call interpreter methods on it such as write
, string
,
evaluate
, execute
, expand
, and so on. The full list of
interpreter methods is here. Exceptions that
occur during processing will be handled by the interpreter’s error
handler.
Important
When you create an interpreter, you must call its shutdown
method
when you are done. This is required to remove the proxy on
sys.stdout
that EmPy requires for proper operation and restore your
Python environment to the state it was before creating the
interpreter. This can be accomplished by creating the interpreter in
a with
statement — interpreters are also context managers — or by
creating it and shutting it down in a try
/finally
statement.
This is not needed when calling the expand
global function; it
creates and shuts down an ephemeral interpreter automatically.
Calling the interpreter’s shutdown
can be handled with either with a
try
/finally
statement or a with
statement:
import em
interp = em.Interpreter(...)
try:
... do some things with the interpreter ...
finally:
interp.shutdown()
# or ...
with em.Interpreter(...) as interp:
... do other things with the interpreter ...
Warning
If you receive a ConsistencyError
mentioning the proxy when quitting
your program, you are likely not calling the shutdown
method on the
interpreter. Make sure to call shutdown
so the interpreter can
clean up after itself.
There is also a global expand
function which will expand a single
string, creating and destroying an ephemeral interpreter to do so.
You can use this function to do a one-off expansion of, say, a large
file:
import em
data = open('tests/sample/sample.em').read()
print(em.expand(data))
If an exception occurs during expand
processing, the exception will
be raised to the caller.
Modules¶
A fully-functional EmPy system contains the following modules and files.
empy
pseudomodule¶
The pseudomodule is not an actual module, but rather the instance of the running EmPy interpreter exposed to the EmPy system. It is automatically placed into the interpreter’s globals and cannot be imported explicitly. See Pseudomodule/interpreter for details.
em
module¶
The primary EmPy module. It contains the Configuration
and
Interpreter
classes as well as all supporting logic. An EmPy system
can be functional with only this module present if needed.
It also includes the following global functions:
details(level, [prelim, postlim, file])
Write details about the running system to the given file, which defaults to
sys.stdout
. Thelevel
parameter is an attribute of theem.Version
class (effectively an enum).prelim
andpostlim
indicate preliminary and postliminary text to output before and after the details (and have reasonable defaults).Note
This function requires the
emlib
to be installed to function most effectively.
expand(data, **kwargs) -> str
Create a ephemeral interpreter with the given kwargs, expand data, shut the interpreter down, and then return the expansion. The function takes the same keyword arguments as the
Interpreter
constructor, with the following additions:Argument
Meaning
Default
dispatcher
Dispatch errors or raise to caller?
False
locals
The locals dictionary
{}
name
The context filename
"<expand>"
If the markup that is being expanded causes an exception to be raised, by default the exception will be let through to the caller.
Important
As with the
Interpreter
constructor, the order of theexpand
arguments has changed over time and is subject to change in the future, so you must use keyword arguments to prevent any ambiguity, e.g.:myConfig = em.Configuration(...) myGlobals = {...} myOutput = open(...) result = em.expand(source, config=myConfig, globals=myGlobals, ...)
Attempts have been made to make the
expand
function as backward compatible (to 3.x) as feasible, but some usages are ambiguous or do not have direct mappings to configurations. ACompatibilityError
will be raised in these cases; if you encounter this, redesign your use ofexpand
to be compatible with the modern usage. In particular, in 3.x, additional keyword arguments were used to indicate the locals dictionary; in 4.x, keyword arguments are used for all arguments so the locals dictionary must be specified as a distinctlocals
keyword argument:myGlobals = {...} myLocals = {...} result = em.expand(source, globals=myGlobals, locals=myLocals)
Warning
Not all of the
Interpreter
constructor arguments are compatible with theexpand
function. Thefilters
,handler
,input
andoutput
arguments are immediately overridden by the inherent nature of the ephemeral interpreter and so would not behave as expected. Thus, if they are specified, aConfigurationError
will be raised. For more detailed configuration of an interpreter, it’s better to create one yourself rather than rely onexpand
.
invoke(args, **kwargs)
Invoke the EmPy system with the given command line arguments (
sys.argv[1:]
, notsys.argv
) and optional string settings. This is the entry point used by the main EmPy function. The remaining keyword arguments correspond to theInterpreter
constructor arguments.Warning
Since the
invoke
function configures and manages the lifetime of anInterpreter
, not all of the constructor arguments are compatible with it. Specifically, thefilespec
andimmediately
arguments need to be managed by the function and so specifying a starting value is nonsensical. Thus, if they are specified, aConfigurationError
will be raised.
emlib
module¶
The EmPy supporting library. It contains various support classes,
including the base classes Filter
and Hook
to assist in creating
this supporting functionality.
emhelp
module¶
The EmPy help system. It can be accessed from the main executable
with the -h/--help
and -H/--topics=TOPICS
command
line options. If the emlib module is not available to the executable,
the help system will return an error.
emdoc
module¶
The EmPy documentation system, used to create this document.
Note
Unlike the other EmPy modules, emdoc
requires a modern Python 3.x
interpreter.
Using EmPy with build tools¶
If you’re using EmPy to process documents within a build system such
as GNU Make or Ninja, you’ll want to use the -o/--output=FILENAME
(or
-a/--append=FILENAME
) and -d/--delete-on-error
options together. This
will guarantee that a file will be output (or appended) to a file
without shell redirection, and that the file will be deleted if an
error occurs. This will prevent errors from leaving a partial file
around which subsequent invocations of the build system will mistake
as being up to date. The invocation of EmPy should look like this
(the --
is not required if the input filename never starts with a
dash):
em.py -d -o $output -- $input
For GNU Make:
EMPY ?= em.py
EMPY_OPTIONS ?= -d
%: %.em
$(EMPY) $(EMPY_OPTIONS) -o $@ -- $<
For Ninja:
empy = em.py
empy_options = -d
rule empy
command = $empy $empy_options -o $out -- $in
Context formatting¶
Contexts are objects which contain the filename, the line number, the column number, and the character (Unicode code point) number to record the location of an EmPy error during processing.
These are formatted into human-readable strings with a context
format, a string specifiable with --context-format
(configuration variable: contextFormat
). A few different mechanisms for formatting contexts are
available:
Mechanism |
Description |
Example |
---|---|---|
format |
Use the |
|
operator |
Use the |
|
variable |
Use |
|
The default context format is %(name)s:%(line)d:%(column)d
and
uses the operator mechanism for backward compatibility.
When a context format is set, EmPy will attempt to detect which of the above mechanisms is needed:
Mechanism |
Criteria |
---|---|
format |
string begins with |
operator |
string begins with |
variable |
string begins with |
Data flow¶
input ⟶ interpreter ⟶ diversions ⟶ filters ⟶ output
Here, in summary, is how data flows through a working EmPy system:
Input comes from a source, such as an .em file on the command line,
sys.stdin
, or via anempy.include
statement.The interpreter processes this material as it comes in, processing EmPy expansions as it goes.
After interpretation, data is then sent through the diversion layer, which may allow it directly through (if no diversion is in progress) or defer it temporarily. Diversions that are recalled initiate from this point.
Any filters in place are then used to filter the data and produce filtered data as output.
Finally, any material surviving this far is sent to the output stream. That stream is
sys.stdout
by default, but can be changed with the-o/--output=FILENAME
or-a/--append=FILENAME
options.If an error occurs, execute the error handler (which by default prints an EmPy error) If the
-r/--raw-errors
option is specified, then print a full Python traceback. If-k/--keep-going
is specified, continue processing rather than exit; otherwise halt.On unsuccessful exit, if
-d/--delete-on-error
is specified, delete any specified output file.
Glossary¶
The following terms with their definitions are used by EmPy:
- callback
The user-provided callback which is called when the custom markup
@<...>
is encountered.- command
A processing step which is performed before or after main document processing. Examples are
-D/--define=DEFN
,-F/--file=FILENAME
or-P/--preprocess=FILENAME
.- configuration
An object encapsulating all the configurable behavior of an interpreter which passed into interpreter on creation. Configurations can be shared between multiple interpreters.
- context
An object which tracks the location of the parser in an EmPy file for tracking and error reporting purposes.
- control markup
A markup used to direct high-level control flow within an EmPy session. Control markups are expressed with the
@[...]
notation.- core
An interpreter core which determines how the underlying language is evaluated, executed, serialized, and how the
@[def ...]
control markup works. By default, the underlying language is Python.- custom
The custom markup invokes a callback which is provided by the user, allowing any desired behavior. Custom markup is
@<...>
.- diacritic
A markup which joins together a letter and one or more combining characters from a dictionary in the configuration and outputs it. Diacritic markup is
@^...
.- dispatcher
An error dispatcher determines whether to dispatch the error to the interpreter’s error handler (
True
), to reraise the error to the caller (False
), or something else.- diversion
A process by which output is deferred, and can be recalled later on demand, multiple times if desired.
- document
An EmPy file containing EmPy markup to expand.
- embedding
Using an EmPy system by importing the
em
module and using the API to create and manipulate interpreters programmatically, as opposed to standalone.- emoji
A markup which looks up a Unicode code point by name via a customizable set of installable emoji modules, or via a dictionary in the configuration. Emoji markup is
@:...:
.- error
An exception thrown by a running EmPy system. When these occur, they are dispatched by an error dispatcher and then (possibly) passed to an error handler.
- escape
A markup designed to expand to a single (often non-printable) character, similar to escape sequences in C or other languages. Escape markup is
@\...
.- expansion
The process of processing EmPy markups and producing output.
- expression
An expression markup represents a Python expression to be evaluated, and replaced with the
str
of its value. Expression markup is@(...)
.- file
An object which exhibits a file-like interface (methods such as
write
andclose
).- filter
A file-like object which can be chained to other filters or the final stream, and can buffer, alter, or manipulate in any way the data sent. Filters can be chained together in arbitrary order.
- finalizer
A function which is called when an interpreter exits. Multiple finalizers can be added to each interpreter.
- globals
The dictionary (or dictionary-like object) which resides inside the interpreter and holds the currently-defined variables.
- handler
An error handler which is called whenever an error occurs in the EmPy system. The default error handler prints details about the error to
sys.stderr
.- hook
A callable object that can be registered in a dictionary, and which will be invoked before, during, or after certain internal operations, identified by name with a string. Some types of hooks can override the behavior of the EmPy interpreter.
- icon
A markup which looks up a variable-length abbreviation for a string from a lookup table in the configuration. Icon markup is
@|...
.- interpreter
The application (or class instance) which processes EmPy markup.
- locals
Along with the globals, a locals dictionary can be passed into individual EmPy API calls.
- markup
EmPy substitutions set off with a prefix (by default
@
) and appropriate delimiters.- named escape
A control character referenced by name in an escape markup,
@\^{...}
.- output
The final destination of the result of processing an EmPy file.
- prefix
The Unicode code point (character) used to set off an expansions. By default, the prefix is
@
. If set toNone
, no markup will be processed.- processor
An extensible system which processes a group of EmPy files, usually arranged in a filesystem, and scans them for significators.
- proxy
An object which replaces the
sys.stdout
file object and allows the EmPy system to intercept any indirect output tosys.stdout
(say, by theprint
function).- pseudomodule
The module-like object named
empy
(by default) which is exposed as a global inside every EmPy system. The pseudomodule and the interpreter are in fact the same object, an instance of theInterpreter
class.- significator
A special form of an assignment markup in EmPy which can be easily parsed externally, primarily designed for representing uniform assignment across a collection of files. Significator markup is
@%[!]... NL
and@%%[!]...%% NL
.- standalone
Using the EmPy system by running the
em.py
executable from the command line.- statement
A line of code that needs to be executed; statements do not have return values. Statement markup is
@{...}
.- stream
A file-like object which manages diversion and filtering. A stack of these is used by the interpreter with the top one being active.
- system
A running EmPy environment.
- token
An element of EmPy parsing. Tokens are parsed and then processed one at a time.
Statistics¶
% wc emdoc.py emhelp.py emlib.py em.py test.sh LEGACY.md LICENSE.md README.md ANNOUNCE.md HELP.md README.md.em 505 1399 17316 emdoc.py 990 4408 42749 emhelp.py 1128 3354 35545 emlib.py 6324 22547 235116 em.py 773 2759 19803 test.sh 2701 15082 100991 LEGACY.md 14 230 1520 LICENSE.md 6554 29326 203221 README.md 767 4160 29317 ANNOUNCE.md 914 5106 46231 HELP.md 6686 31391 219427 README.md.em 27356 119762 951236 total
% sha1sum emdoc.py emhelp.py emlib.py em.py test.sh LEGACY.md LICENSE.md README.md ANNOUNCE.md HELP.md README.md.em 0a9d4545f1c4ef49e12811fe57e780c28166f369 emdoc.py f3c625c51a1425e1bf84b2aefae7a74fae21deed emhelp.py e965e967e34f58f7280c64a406f46bc7b3915ced emlib.py 9500c6e6029e2bb63396246a1d967597c06d3673 em.py 7f55200b3b67770ada2240c77074952c972ab207 test.sh 2c09a1fc1345f3dd637cd397cc674696f16fdc60 LEGACY.md cd2b06faf3ef9188ef67c38818a0c6076f757801 LICENSE.md 595ebd968db1f64fbeb095204cc913f8fcf1e364 README.md dcc21725b020d597ba2e4096402c511923eb242c ANNOUNCE.md 652c8fca03f510bae15afd867b59195b251b2b59 HELP.md 0afeba9e1d35c975ac5775c4e74ff17f69632c5e README.md.em
End notes¶
Acknowledgements¶
Questions, suggestions, bug reports, evangelism, and even complaints from many people over the years have helped make EmPy what it is today. Some, but by no means all, of these people are (in alphabetical order by surname):
Biswapesh Chattopadhyay
Beni Cherniavsky
Dr. S. Candelaria de Ram
Eric Eide
Dinu Gherman
Grzegorz Adam Hankiewicz
Robert Kroeger
Bohdan Kushnir
Kouichi Takahashi
Ville Vainio
Known issues and caveats¶
A running EmPy system is just an alternate form of a Python interpreter; EmPy code is just as powerful as any Python code. Thus it is vitally important that an EmPy system not expand EmPy markup from an untrusted source; this is just as unsafe and potentially dangerous as executing untrusted Python code.
To function properly, EmPy must override
sys.stdout
with a proxy file object, so that it can capture output of side effects and support diversions for each interpreter instance. It is important that code executed in an environment not rebindsys.stdout
, although it is perfectly legal to reference it explicitly (e.g.,@sys.stdout.write("Hello world\n")
). If one really needs to access the “true” stdout, then usesys.__stdout__
instead (which should also not be rebound). EmPy uses the standard Python error handlers when exceptions are raised in EmPy code, which print tosys.stderr
.sys.stderr
,sys.__stdout__
, andsys.__stderr__
are never overridden by the interpreter; onlysys.stdout
is.If you are using multiple interpreters with distinct output files and are using the low-level interpreter methods (the ones not documented here) to perform expansion and output, the
sys.stdout
proxy will not be reliable. Only the high-level interpreter methods (evaluate
,execute
,string
,expand
properly use the protected stream stack on thesys.stdout
proxy to guarantee valid output. Either only use a single interpreter instance at a time (creating and shutting it down with itsshutdown
method), use the-n/--no-proxy
option and only perform output with thewrite
method on the interpreter (i.e., do not use anyprint
statements in your code), or only use the high-level interpreter methods documented here.The
empy
“module” exposed through the EmPy interface (e.g.,@empy
) is an artificial module. It is automatically exposed in the globals of a running interpreter and it cannot be manually imported with theimport
statement (nor should it be — it is an artifact of the EmPy processing system and does not correspond directly to any .py file).For an EmPy statement expansion all alone on a line, e.g.,
@{a = 1}
, will include a blank line due to the newline following the closing curly brace. To suppress this blank line, use the symmetric convention@{a = 1}@
, where the final@
markup precedes the newline, making it whitespace markup and thus consumed. For instance:@{a = 1} There will be an extra newline above (following the closing brace). Compare this to: @{a = 1}@ There will be no extra newline above.
See here for more details.
Errors generated from within nested control structures (e.g.,
@[for ...]@[if ...]...@[end if]@[end for]
will report a context of the start of the top-level control structure markup, not the innermost markup, which would be much more helpful. This issue is not new to 4.0 and will be addressed in a future release.Errors are very literal and could be made more useful to find the underlying cause.
Contexts (such as
empy.identify
) track the context of executed EmPy code, not Python code. This means, for instance, that blocks of code delimited with@{
and}
will identify themselves as appearing on the line at which the@{
appears. If you’re tracking errors and want more information about the location of the errors from the Python code, use the-r/--raw-errors
option, which will provide you with the full Python traceback.The
@[for ...]
variable specification supports tuples for tuple unpacking, even recursive tuples. However, it is limited in that the names included may only be valid Python identifiers, not arbitrary Python “lvalues.” Since this is something of an accidental Python feature that is very unlikely to be relied on in practice, this is not thought to be a significant limitation. As a concrete example:a = [None] for a[0] in range(5): print(a)
is valid (but strange) Python, but the EmPy equivalent with
@[for a[0] in range(5)]...
is invalid.The
:=
assignment expression syntax (“walrus operator”) forwhile
loops andif
statements, introduced in Python 3.8, is not supported in the EmPy equivalent control markups@[while]
and@[if]
. This may be supported in the future.As of Python 3.10, the
with
control structure supports specifying multiple context managers separated by commas. This is not yet supported by EmPy, but may be in a future version. For now, just use nested@[with]
control markups.
For package maintainers¶
EmPy is available as a system package in most major Linux distributions, though many have not updated to EmPy 4.0 yet.
EmPy can be made available as an operating system/distribution package
in several different ways. Regardless of the high-level organization,
the installed .py Python files must be made available as importable
Python modules, with the additional requirement that em.py must be
made available as an executable in the default PATH
. If necessary,
this executable may also be named empy
, but em.py
is preferred —
and either way it is still important that the em.py file be available
for importing as a Python module (em
).
Note
Since EmPy 4.0 is not fully compatible with EmPy 3.x, I suggest making both EmPy 3.x and 4.0 packages available side by side until 4.0 becomes more fully adopted by the community.
Here is a breakdown of the contents of a release tarball:
File |
Description |
---|---|
em.py |
Main EmPy module and executable |
emhelp.py |
Help subsystem module |
emlib.py |
Supplementary EmPy facilities module |
emdoc.py |
Documentation subsystem module |
ANNOUNCE.md |
EmPy 4.0 release announcement |
HELP.md |
Help topic summaries |
LICENSE.md |
Software license |
README.md |
README (this file) |
README.md.em |
README source file |
doc |
HTML documentation directory hierarchy |
test.sh |
Test shell script |
tests |
Tests directory hierarchy |
suites |
Test suites directory hierarchy |
They can either be bundled up into a single, monolithic package, or divided into a series of subpackages. Here’s a suggestion for a fleshed-out series of EmPy subpackages:
empy-minimal
Just the em.py file, available as a Python module as well as an executable. Note that this will not allow the use of the EmPy help subsystem, unless the module emhelp.py is also included.
empy-basic
The .md files, all the .py files (em.py, emhelp.py, emlib.py, emdoc.py) available as Python modules, with the em.py file also available as an executable.
empy-doc
Just the contents of the README source file README.md.em and the docs directory hierarchy.
empy-test
The test script test.sh, the tests directory, and the suites directory.
empy
All of the above.
Reporting bugs¶
If you find a bug in EmPy, please follow these steps:
Whittle a reproducible test case down to the smallest standalone example which demonstrates the issue, the smaller the better;
Collect the output of
em.py -Z
(this will provide detailed diagnostic details about your environment), or at leastem.py -W
(which provides only basic details);Send me an email with EmPy in the Subject line including both files and a description of the problem.
Thank you!
Release history¶
- 4.1 (2024 Mar 24)
Add support for extension markup
@((...))
,@[[...]]
,@{{...}}
,@<...>
, etc., with custom callbacks retained for backward compatibility; add@[match]
control support; add interpreter cores for overriding interpreter behavior; add more command line option toggles; add notion of verbose/brief errors; more uniform error message formatting; various documentation updates.
- 4.0.1 (2023 Dec 24)
Add root context argument, serializers, and idents to interpreter; fix
setContext...
methods so they also modify the currents stack; better backward compatibility forexpand
function andCompatibilityError
; fix inconsistent stack usage withexpand
method; add error dispatchers, cleaner error handling andignoreErrors
; haveexpand
method/function raise exceptions to caller; eliminate need forFullContext
class distinct fromContext
; support comments in “clean” controls; add--no-none-symbol
option; add clearer errors for removed literal markup; addContainer
support class inemlib
; hide non-standard proxy attributes and methods; support string errors (why not); update and expand tests; help subsystem and documentation updates.- 4.0 (2023 Nov 29)
A major revamp, refresh, and modernization. Major new features include inline comments
@*...*
; backquote literals@`...`
; chained if-then-else expressions; functional expressions@f{...}
; full support for@[try]
,@[while ...]
and@[with ...]
control markup;@[defined ...]
control markup; stringized and multiline significators; named escapes@\^{...}
; diacritics@^...
; icons@|...
; emojis@:...:
; configurations; full Unicode and file buffering support; proxy now reference counted; hooks can override behavior; many bug fixes; an extensive builtin help system (emhelp
); and rewritten and expanded documentation in addition to a dedicated module (emdoc
). Changes include relicensing to BSD, interpreter constructor now requires keyword arguments,-d/--delete-on-error
instead of “fully buffered files”; cleaned up environment variables; “repr” markup replaced with emoji markup; remove literal markups@)
,@]
,@}
; context line markup@!...
no longer pre-adjusts line; custom markup@<...>
now parsed more sensibly; filter shortcuts removed; context now track column and character count; auxiliary classes moved toemlib
module; useargv
instead ofargc
for interpreter arguments. See Full list of changes between EmPy 3.x and 4.0 for a more comprehensive list.
- 3.3.4a (2021 Nov 19)
Fix an error in setup.py in the downloadable tarball (did not affect PIP downloads).
- 3.3.4 (2019 Feb 26)
Minor fix for a Python 3.x compatibility issue.
- 3.3.3 (2017 Feb 12)
Fix for
empy.defined
interpreter method.- 3.3.2 (2014 Jan 24)
Additional fix for source compatibility between 2.x and 3.0.
- 3.3.1 (2014 Jan 22)
Source compatibility for 2.x and 3.0; 1.x compatibility dropped.
- 3.3 (2003 Oct 27)
Custom markup
@<...>
; remove separate pseudomodule instance for greater transparency; deprecateInterpreter
attribute of pseudomodule; deprecate auxiliary class name attributes associated with pseudomodule in preparation for separate support library in 4.0; add--no-callback-error
[defunct] and--no-bangpath-processing
[now--no-ignore-bangpaths
] command line options; addatToken
hook.- 3.2 (2003 Oct 7)
Reengineer hooks support to use hook instances; add
-v/--verbose
and-l/--relative-path
option; reversed PEP 317 style; modify Unicode support to give less confusing errors in the case of unknown encodings and error handlers; relicensed under LGPL.- 3.1.1 (2003 Sep 20)
Add string literal
@"..."
markup; add-w/--pause-at-end
option; fix improper globals collision error via thesys.stdout
proxy.- 3.1 (2003 Aug 8)
Unicode support (Python 2.0 and above); add Document and Processor helper classes for processing significators [later moved to
emlib
]; add--no-prefix
option for suppressing all markups.- 3.0.4 (2003 Aug 7)
Implement somewhat more robust “lvalue” parsing for
@[for]
construct (thanks to Beni Cherniavsky for inspiration).- 3.0.3 (2003 Jul 9)
Fix bug regarding recursive tuple unpacking using
@[for]
; addempy.saveGlobals
,empy.restoreGlobals
, andempy.defined
functions.- 3.0.2 (2003 Jun 19)
@?
and@!
markups for changing the current context name and line, respectively; addupdate
method to interpreter; new and renamed context operations,empy.setContextName
,empy.setContextLine
,empy.pushContext
,empy.popContext
.- 3.0.1 (2003 Jun 9)
Fix simple bug preventing command line preprocessing directives (
-I/--import=MODULES
,-D/--define=DEFN
,-E/--execute=STATEMENT
,-F/--file=FILENAME
,-P/--preprocess=FILENAME
) from executing properly; defensive PEP 317 compliance [defunct].- 3.0 (2003 Jun 1)
Replace substitution markup with control markup
@[...]
; support@(...?...!...)
for conditional expressions; add acknowledgements and glossary sections to documentation; rename buffering option back to-b/--buffering
; add-m/--pseudomodule=NAME
and-n/--no-proxy
for suppressingsys.stdout
proxy; rename main error class toError
; add standaloneexpand
function; add--binary
and--chunk-size
options [defunct]; reengineer parsing system to use tokens for easy extensibility; safeguard curly braces in simple expressions [now used by functional expressions]; fix bug involving customInterpreter
instances ignoring globals argument;distutils
[nowsetuptools
] support.- 2.3 (2003 Feb 20)
Proper and full support for concurrent and recursive interpreters; protection from closing the true stdout file object; detect edge cases of interpreter globals or
sys.stdout
proxy collisions; add globals manipulation functionsempy.getGlobals
,empy.setGlobals
, andempy.updateGlobals
which properly preserve theempy
pseudomodule; separate usage info out into easily accessible lists for easier presentation; have-h
option show simple usage and-H
show extended usage [defunct]; addNullFile
utility class.- 2.2.6 (2003 Jan 30)
Fix a bug in the
Filter.detach
method (which would not normally be called anyway).- 2.2.5 (2003 Jan 9)
Strip carriage returns out of executed code blocks for DOS/Windows compatibility.
- 2.2.4 (2002 Dec 23)
Abstract Filter interface to use methods only; add
@[noop: ...]
substitution for completeness and block commenting [defunct].- 2.2.3 (2002 Dec 16)
Support compatibility with Jython by working around a minor difference between CPython and Jython in string splitting.
- 2.2.2 (2002 Dec 14)
Include better docstrings for pseudomodule functions; segue to a dictionary-based options system for interpreters; add
empy.clearAllHooks
andempy.clearGlobals
; include a short documentation section on embedding interpreters; fix a bug in significator regular expression.- 2.2.1 (2002 Nov 30)
Tweak test script to avoid writing unnecessary temporary file; add
Interpreter.single
method; exposeevaluate
,execute
,substitute
[defunct], andsingle
methods to the pseudomodule; add (rather obvious)EMPY_OPTIONS
environment variable support; addempy.enableHooks
andempy.disableHooks
; include optimization to transparently disable hooks until they are actually used.- 2.2 (2002 Nov 21)
Switched to
-V/--version
option for version information;empy.createDiversion
for creating initially empty diversion; direct access to diversion objects withempy.retrieveDiversion
; environment variable support; removed--raw
long argument (use-r/--raw-errors
instead); added quaternary escape code (well, why not).- 2.1 (2002 Oct 18)
empy.atExit
[nowempy.appendFinalizer
] registration separate from hooks to allow for normal interpreter support; include a benchmark sample and test.sh verification script; exposeempy.string
directly;-D/--define=DEFN
option for explicit defines on command line; remove ill-conceived support for@else:
separator in@[if ...]
substitution [defunct]; handle nested substitutions properly [defunct];@[macro ...]
substitution for creating recallable expansions [defunct]; add support for finalizers withempy.atExit
[nowempy.appendFinalizer
].- 2.0.1 (2002 Oct 8)
Fix missing usage information; fix
after_evaluate
hook not getting called [defunct].- 2.0 (2002 Sep 30)
Parsing system completely revamped and simplified, eliminating a whole class of context-related bugs; builtin support for buffered filters; support for registering hooks; support for command line arguments; interactive mode with
-i/--interactive
; significator value extended to be any valid Python expression.- 1.5.1 (2002 Sep 24)
Allow
@]
to represent unbalanced close brackets in@[...]
markups [defunct].- 1.5 (2002 Sep 18)
Escape codes (
@\...
); conditional and repeated expansion substitutions [defunct]; replaced with control markups]; fix a few bugs involving files which do not end in newlines.- 1.4 (2002 Sep 7)
Add in-place markup
@:...:...:
[now@$...$...$
]; fix bug with triple quotes; collapse conditional and protected expression syntaxes into the single generalized@(...)
notation;empy.setName
andempy.setLine
functions [nowempy.setContextName
andempy.setContextLine
]; true support for multiple concurrent interpreters with improvedsys.stdout
proxy; proper support forempy.expand
to return a string evaluated in a subinterpreter as intended; reorganized parser class hierarchy.- 1.3 (2002 Aug 24)
Pseudomodule as true instance; move toward more verbose (and clear) pseudomodule functions; fleshed out diversions model; filters; conditional expressions; protected expressions; preprocessing with
-P/--preprocess=FILENAME
(in preparation for possible support for command line arguments).- 1.2 (2002 Aug 16)
Treat bangpaths as comments;
empy.quote
for the opposite process ofempy.expand
; significators (@%...
sequences); add-I/--import=MODULES
and-f/--flatten
options; much improved documentation.- 1.1.5 (2002 Aug 15)
Add a separate
invoke
function that can be called multiple times with arguments to simulate multiple runs.- 1.1.4 (2002 Aug 12)
Handle strings thrown as exceptions properly; use
getopt
to process command line arguments; cleanup file buffering with AbstractFile; very slight documentation and code cleanup.- 1.1.3 (2002 Aug 9)
Support for changing the prefix from within the
empy
pseudomodule.- 1.1.2 (2002 Aug 5)
Renamed buffering option [defunct], added
-F/--file=FILENAME
option for interpreting Python files from the command line, fixed improper handling of exceptions from command line options (-E/--execute=STATEMENT
,-F/--file=FILENAME
).- 1.1.1 (2002 Aug 4)
Typo bugfixes; documentation clarification.
- 1.1 (2002 Aug 4)
Added option for fully buffering output (including file opens), executing commands through the command line; some documentation errors fixed.
- 1.0 (2002 Jul 23)
Renamed project to EmPy. Documentation and sample tweaks; added
empy.flatten
[nowempy.flattenGlobals
]; added-a/--append=FILENAME
option. First official release.- 0.3 (2002 Apr 14)
Extended “simple expression” syntax, interpreter abstraction, proper context handling, better error handling, explicit file inclusion, extended samples.
- 0.2 (2002 Apr 13)
Bugfixes, support non-expansion of
None
s, allow choice of alternate prefix.- 0.1.1 (2002 Apr 12)
Bugfixes, support for Python 1.5.x [defunct], add
-r/--raw-errors
option.- 0.1 (2002 Apr 12)
Initial early access release.
Contact¶
This software was written by Erik Max Francis. If you use this software, have suggestions for future releases, or bug reports or problems with this documentation, I’d love to hear about it.
Even if you try out EmPy for a project and find it unsuitable, I’d like to know what stumbling blocks you ran into so they can potentially be addressed in a future version.
I hope you enjoy using EmPy! ℰ
About this document¶
This document was generated with EmPy itself using the emdoc
module.
Both the source (README.md.em) and the resulting Markdown text
(README.md) are included in the release tarball, as is the HTML
directory hierarchy generated with Sphinx (doc).
This documentation for EmPy version 4.1 was generated from README.md.em (SHA1 0afeba9e1d35c975ac5775c4e74ff17f69632c5e
, 219427 bytes) at 2024-04-03 16:52:31 using EmPy version 4.1, in CPython/3.10.12, on Linux (POSIX), with x86_64.
Comment markup¶
Comment markup consumes its contents and performs no output. A few variants of comment markup are available.
Line comment markup:
@#... NL
¶Line comment markup consists of a starting
@#
and consumes up until (and including) the following newline. Note that if the markup appears in the middle of a line, that line will be continued since it consumes the ending newline.Example 6: Line comments
Source:
Output:
Note
Line comment markup was introduced in EmPy version 1.0.
Inline comment markup:
@*...*
¶Inline comment markup (
@*...*
) is a form of comment markup that can appear anywhere in text and can even span multiple lines. It consumes everything up to and including the final asterisk(s).Example 7: Inline comments, basic
Source:
Output:
Multiple asterisks can be used as long as they are matched with the end of the markup. This allows asterisks to appear in the comment:
Example 8: Inline comments, advanced
Source:
Output:
Attention
Note that when markup which has starting and ending delimiters appears alone on a line, the trailing newline will be rendered in the output. To avoid these extra newlines, use a trailing
@
to turn it into whitespace markup which consumes that trailing newline, so e.g.@*...*
followed by a newline becomes@*...*@
followed by a newline. This is idiomatic for suppressing unwanted newlines. See here for more details.Note
Inline comment markup was introduced in EmPy version 4.0.
Whitespace markup:
@ WS
¶While not quite a comment, whitespace markup is sufficiently common and useful that it warrants introduction early on. The interpreter prefix followed by any whitespace character, including a newline, is consumed. This allows a way to concatenate two strings, create a line continuation, or create a line separator:
Example 9: Whitespace, basic
Source:
Output:
Tip
A trailing prefix after markup which has beginning and ending delimiters — for instance, inline comment (
@*...*
), expression (@(...)
), statement (@{...}
) and control (@[...]
) — is idiomatic for suppressing the newline when there is nothing at the end of the line after the markup. The trailing prefix will consume the final newline, eliminating unwanted newlines.For example, using a statement markup (see below) on a whole line will result in a seemingly spurious newline:
Example 10: Whitespace, idiom
Source:
Output:
Note
Whitespace markup was introduced in EmPy version 1.0.