Related
I am running into an error/warning when inheriting from collections.deque. Sphinx complains:
utils.py:docstring of utils.Test.count:1: WARNING: py:class reference target not found: integer -- return number of occurrences of value
For this example, utils.py only contains
from collections import deque
class Test(deque):
pass
My best guess is that this might be caused by deque being implemented in C, since integer is not a valid Python type. Looking around a bit, I found this line in the CPython source of collections, which could be exactly that part, should my hypothesis be correct. But I have no clue about CPython, the inner workings of Sphinx autodoc and the boundary between those worlds.
Is there a good solution to correcting this, whether in the annotation of CPython or the autodoc plugin?
It seems I was more or less on the money. integer is used elsewhere and seems to work, but more importantly Sphinx expects docstrings to follow the example outlined in PEP 7, which suggests signatures to be placed on the first line (followed by a line break before any other documentation). So it looks for a class named by the whole string between -> and \n, in my case being integer -- return number of occurrences of value.
The workaround is to put all those strings into nitpick_ignore:
# Wrong docstrings in the CPython source of collections.queue
("py:class", "-- reverse *IN PLACE*"),
("py:class", "integer -- return first index of value."),
("py:class", "integer -- return number of occurrences of value"),
("py:class", "-- insert object before index"),
("py:class", "-- remove first occurrence of value."),
("py:class", "-- size of D in memory, in bytes"),
("py:class", "-- return a reverse iterator over the deque"),
A proper fix would be to change the documentation in the CPython code accordingly.
I was about to do that, when I found out my coworker beat me to it by 2 weeks. But on the issue he opened, a core developer argues that this scheme is nothing but an example and that the documentation is only required to be human readable as opposed to specifying the exact types. So it doesn't look like there is going to be a fix anytime soon.
I have a module called test.go that contains two simple Go functions which accept string types:
package main
import (
"fmt"
"C"
)
//export TestConcat
func TestConcat(testArg string, testArg2 string) (string) {
retval := testArg + testArg2
return retval
}
//export TestHello
func TestHello(testArg string) {
fmt.Println("%v\n", testArg)
}
func main(){}
I compile it as a shared library with go build -o test.so -buildmode=c-shared test.go
Then I have a Python module called test.py
import ctypes
from ctypes import cdll
test_strings = [
"teststring1",
"teststring2"
]
if __name__ == '__main__':
lib = cdll.LoadLibrary("./test.so")
lib.TestConcat.argtypes = [ctypes.c_wchar_p, ctypes.c_wchar_p]
lib.TestHello.argtypes = [ctypes.c_wchar_p]
for test_string in test_strings:
print(
lib.TestConcat("hello", test_string)
)
lib.TestHello(test_string)
Then I run test.py and get a nasty segfault
runtime: out of memory: cannot allocate 279362762964992-byte block (66781184 in use)
fatal error: out of memory
I've tried wrapping the arguments in ctypes.c_wchar_p to no avail.
What am I doing wrong here? And specifically, how does one interact with Go functions that accept string arguments in Python?
The Go's string type is actually something like
type string {
ptr *byte
size int
}
so that is what the Test{Hello|Concat} actually expect—not a pair of pointers but a pair of struct-typed values.
In other words, cgo performs just enough magic to gateway calls from Go to C and back, but it does not perform automatic conversions of values.
You have two options:
Explicitly work with this from your ctypes bindings, if possible.
When compiling your package, cgo generates a header file which contains a C definition for the struct representing a Go string; you could use it right away.
Make the functions exported to C compatible with the C's "type system".
For this, cgo offers helper functions C.CString and C.GoString.
Basically, you can define your API like this:
func TestHello(a, b *C.char) *C.char {
testArg1, testArg2 := C.GoString(a), C.GoString(b)
return C.CString(testArg + TestArg2)
}
Note few caveats here:
Both of these helpers copy the memory of their argument, so the silly example above would work just fine but it would first duplicate the memory blocks pointed to by a and b, then eat up twice as much memory to produce the concatenated string and then copy the memory of the resulting string once again to produce the returned pointer.
IOW, this approach is fine if you're trying to export to C some big chunk of Go code so that these allocations are dwarfed by whatever that chunk does.
Using *C.char is the same as *char in C, so the string is expected to be NUL-terminated; if it's not, use C.GoStringN.
Every memory block allocated by C.CString have to be freed by a call to C.free. And here's a twist: C.free is basically a thin shim to call free() from the linked in libc, so if you can guarantee the complete product (the code fully loaded into memory and (inter)linked using the dymanic linker) has only a single copy of libc linked in, you can call free() from the non-Go code on the memory blocks produced by calls to C.Cstring in the Go code.
A few more random pointers:
I'm not well-versed in Python's ctypes but I'd speculate using ctypes.c_wchar_p is not correct: in C (and C++, FWIW) wchar_t is a type to denote a single fixed-sized "wide character", which is usually a UCS-2/UTF-16 code point, and Go's strings are not composed of these—they may contain arbitrary bytes, and when they are used to contain Unicode text, they are encoded using UTF-8 which is a multi-byte ecoding (a single Unicode code point may be represented by 1 to 4 bytes in the string).
In either case, wchar_t cannot be used for UTF-8 (and actually many seasoned devs beleive it's an abomination).
Please read the docs on cmd/cgo completely before embarking on this project. Really, please do!
Can you add new statements (like print, raise, with) to Python's syntax?
Say, to allow..
mystatement "Something"
Or,
new_if True:
print "example"
Not so much if you should, but rather if it's possible (short of modifying the python interpreters code)
You may find this useful - Python internals: adding a new statement to Python, quoted here:
This article is an attempt to better understand how the front-end of Python works. Just reading documentation and source code may be a bit boring, so I'm taking a hands-on approach here: I'm going to add an until statement to Python.
All the coding for this article was done against the cutting-edge Py3k branch in the Python Mercurial repository mirror.
The until statement
Some languages, like Ruby, have an until statement, which is the complement to while (until num == 0 is equivalent to while num != 0). In Ruby, I can write:
num = 3
until num == 0 do
puts num
num -= 1
end
And it will print:
3
2
1
So, I want to add a similar capability to Python. That is, being able to write:
num = 3
until num == 0:
print(num)
num -= 1
A language-advocacy digression
This article doesn't attempt to suggest the addition of an until statement to Python. Although I think such a statement would make some code clearer, and this article displays how easy it is to add, I completely respect Python's philosophy of minimalism. All I'm trying to do here, really, is gain some insight into the inner workings of Python.
Modifying the grammar
Python uses a custom parser generator named pgen. This is a LL(1) parser that converts Python source code into a parse tree. The input to the parser generator is the file Grammar/Grammar[1]. This is a simple text file that specifies the grammar of Python.
[1]: From here on, references to files in the Python source are given relatively to the root of the source tree, which is the directory where you run configure and make to build Python.
Two modifications have to be made to the grammar file. The first is to add a definition for the until statement. I found where the while statement was defined (while_stmt), and added until_stmt below [2]:
compound_stmt: if_stmt | while_stmt | until_stmt | for_stmt | try_stmt | with_stmt | funcdef | classdef | decorated
if_stmt: 'if' test ':' suite ('elif' test ':' suite)* ['else' ':' suite]
while_stmt: 'while' test ':' suite ['else' ':' suite]
until_stmt: 'until' test ':' suite
[2]: This demonstrates a common technique I use when modifying source code I’m not familiar with: work by similarity. This principle won’t solve all your problems, but it can definitely ease the process. Since everything that has to be done for while also has to be done for until, it serves as a pretty good guideline.
Note that I've decided to exclude the else clause from my definition of until, just to make it a little bit different (and because frankly I dislike the else clause of loops and don't think it fits well with the Zen of Python).
The second change is to modify the rule for compound_stmt to include until_stmt, as you can see in the snippet above. It's right after while_stmt, again.
When you run make after modifying Grammar/Grammar, notice that the pgen program is run to re-generate Include/graminit.h and Python/graminit.c, and then several files get re-compiled.
Modifying the AST generation code
After the Python parser has created a parse tree, this tree is converted into an AST, since ASTs are much simpler to work with in subsequent stages of the compilation process.
So, we're going to visit Parser/Python.asdl which defines the structure of Python's ASTs and add an AST node for our new until statement, again right below the while:
| While(expr test, stmt* body, stmt* orelse)
| Until(expr test, stmt* body)
If you now run make, notice that before compiling a bunch of files, Parser/asdl_c.py is run to generate C code from the AST definition file. This (like Grammar/Grammar) is another example of the Python source-code using a mini-language (in other words, a DSL) to simplify programming. Also note that since Parser/asdl_c.py is a Python script, this is a kind of bootstrapping - to build Python from scratch, Python already has to be available.
While Parser/asdl_c.py generated the code to manage our newly defined AST node (into the files Include/Python-ast.h and Python/Python-ast.c), we still have to write the code that converts a relevant parse-tree node into it by hand. This is done in the file Python/ast.c. There, a function named ast_for_stmt converts parse tree nodes for statements into AST nodes. Again, guided by our old friend while, we jump right into the big switch for handling compound statements and add a clause for until_stmt:
case while_stmt:
return ast_for_while_stmt(c, ch);
case until_stmt:
return ast_for_until_stmt(c, ch);
Now we should implement ast_for_until_stmt. Here it is:
static stmt_ty
ast_for_until_stmt(struct compiling *c, const node *n)
{
/* until_stmt: 'until' test ':' suite */
REQ(n, until_stmt);
if (NCH(n) == 4) {
expr_ty expression;
asdl_seq *suite_seq;
expression = ast_for_expr(c, CHILD(n, 1));
if (!expression)
return NULL;
suite_seq = ast_for_suite(c, CHILD(n, 3));
if (!suite_seq)
return NULL;
return Until(expression, suite_seq, LINENO(n), n->n_col_offset, c->c_arena);
}
PyErr_Format(PyExc_SystemError,
"wrong number of tokens for 'until' statement: %d",
NCH(n));
return NULL;
}
Again, this was coded while closely looking at the equivalent ast_for_while_stmt, with the difference that for until I've decided not to support the else clause. As expected, the AST is created recursively, using other AST creating functions like ast_for_expr for the condition expression and ast_for_suite for the body of the until statement. Finally, a new node named Until is returned.
Note that we access the parse-tree node n using some macros like NCH and CHILD. These are worth understanding - their code is in Include/node.h.
Digression: AST composition
I chose to create a new type of AST for the until statement, but actually this isn't necessary. I could've saved some work and implemented the new functionality using composition of existing AST nodes, since:
until condition:
# do stuff
Is functionally equivalent to:
while not condition:
# do stuff
Instead of creating the Until node in ast_for_until_stmt, I could have created a Not node with an While node as a child. Since the AST compiler already knows how to handle these nodes, the next steps of the process could be skipped.
Compiling ASTs into bytecode
The next step is compiling the AST into Python bytecode. The compilation has an intermediate result which is a CFG (Control Flow Graph), but since the same code handles it I will ignore this detail for now and leave it for another article.
The code we will look at next is Python/compile.c. Following the lead of while, we find the function compiler_visit_stmt, which is responsible for compiling statements into bytecode. We add a clause for Until:
case While_kind:
return compiler_while(c, s);
case Until_kind:
return compiler_until(c, s);
If you wonder what Until_kind is, it's a constant (actually a value of the _stmt_kind enumeration) automatically generated from the AST definition file into Include/Python-ast.h. Anyway, we call compiler_until which, of course, still doesn't exist. I'll get to it an a moment.
If you're curious like me, you'll notice that compiler_visit_stmt is peculiar. No amount of grep-ping the source tree reveals where it is called. When this is the case, only one option remains - C macro-fu. Indeed, a short investigation leads us to the VISIT macro defined in Python/compile.c:
#define VISIT(C, TYPE, V) {\
if (!compiler_visit_ ## TYPE((C), (V))) \
return 0; \
It's used to invoke compiler_visit_stmt in compiler_body. Back to our business, however...
As promised, here's compiler_until:
static int
compiler_until(struct compiler *c, stmt_ty s)
{
basicblock *loop, *end, *anchor = NULL;
int constant = expr_constant(s->v.Until.test);
if (constant == 1) {
return 1;
}
loop = compiler_new_block(c);
end = compiler_new_block(c);
if (constant == -1) {
anchor = compiler_new_block(c);
if (anchor == NULL)
return 0;
}
if (loop == NULL || end == NULL)
return 0;
ADDOP_JREL(c, SETUP_LOOP, end);
compiler_use_next_block(c, loop);
if (!compiler_push_fblock(c, LOOP, loop))
return 0;
if (constant == -1) {
VISIT(c, expr, s->v.Until.test);
ADDOP_JABS(c, POP_JUMP_IF_TRUE, anchor);
}
VISIT_SEQ(c, stmt, s->v.Until.body);
ADDOP_JABS(c, JUMP_ABSOLUTE, loop);
if (constant == -1) {
compiler_use_next_block(c, anchor);
ADDOP(c, POP_BLOCK);
}
compiler_pop_fblock(c, LOOP, loop);
compiler_use_next_block(c, end);
return 1;
}
I have a confession to make: this code wasn't written based on a deep understanding of Python bytecode. Like the rest of the article, it was done in imitation of the kin compiler_while function. By reading it carefully, however, keeping in mind that the Python VM is stack-based, and glancing into the documentation of the dis module, which has a list of Python bytecodes with descriptions, it's possible to understand what's going on.
That's it, we're done... Aren't we?
After making all the changes and running make, we can run the newly compiled Python and try our new until statement:
>>> until num == 0:
... print(num)
... num -= 1
...
3
2
1
Voila, it works! Let's see the bytecode created for the new statement by using the dis module as follows:
import dis
def myfoo(num):
until num == 0:
print(num)
num -= 1
dis.dis(myfoo)
Here's the result:
4 0 SETUP_LOOP 36 (to 39)
>> 3 LOAD_FAST 0 (num)
6 LOAD_CONST 1 (0)
9 COMPARE_OP 2 (==)
12 POP_JUMP_IF_TRUE 38
5 15 LOAD_NAME 0 (print)
18 LOAD_FAST 0 (num)
21 CALL_FUNCTION 1
24 POP_TOP
6 25 LOAD_FAST 0 (num)
28 LOAD_CONST 2 (1)
31 INPLACE_SUBTRACT
32 STORE_FAST 0 (num)
35 JUMP_ABSOLUTE 3
>> 38 POP_BLOCK
>> 39 LOAD_CONST 0 (None)
42 RETURN_VALUE
The most interesting operation is number 12: if the condition is true, we jump to after the loop. This is correct semantics for until. If the jump isn't executed, the loop body keeps running until it jumps back to the condition at operation 35.
Feeling good about my change, I then tried running the function (executing myfoo(3)) instead of showing its bytecode. The result was less than encouraging:
Traceback (most recent call last):
File "zy.py", line 9, in
myfoo(3)
File "zy.py", line 5, in myfoo
print(num)
SystemError: no locals when loading 'print'
Whoa... this can't be good. So what went wrong?
The case of the missing symbol table
One of the steps the Python compiler performs when compiling the AST is create a symbol table for the code it compiles. The call to PySymtable_Build in PyAST_Compile calls into the symbol table module (Python/symtable.c), which walks the AST in a manner similar to the code generation functions. Having a symbol table for each scope helps the compiler figure out some key information, such as which variables are global and which are local to a scope.
To fix the problem, we have to modify the symtable_visit_stmt function in Python/symtable.c, adding code for handling until statements, after the similar code for while statements [3]:
case While_kind:
VISIT(st, expr, s->v.While.test);
VISIT_SEQ(st, stmt, s->v.While.body);
if (s->v.While.orelse)
VISIT_SEQ(st, stmt, s->v.While.orelse);
break;
case Until_kind:
VISIT(st, expr, s->v.Until.test);
VISIT_SEQ(st, stmt, s->v.Until.body);
break;
[3]: By the way, without this code there’s a compiler warning for Python/symtable.c. The compiler notices that the Until_kind enumeration value isn’t handled in the switch statement of symtable_visit_stmt and complains. It’s always important to check for compiler warnings!
And now we really are done. Compiling the source after this change makes the execution of myfoo(3) work as expected.
Conclusion
In this article I've demonstrated how to add a new statement to Python. Albeit requiring quite a bit of tinkering in the code of the Python compiler, the change wasn't difficult to implement, because I used a similar and existing statement as a guideline.
The Python compiler is a sophisticated chunk of software, and I don't claim being an expert in it. However, I am really interested in the internals of Python, and particularly its front-end. Therefore, I found this exercise a very useful companion to theoretical study of the compiler's principles and source code. It will serve as a base for future articles that will get deeper into the compiler.
References
I used a few excellent references for the construction of this article. Here they are, in no particular order:
PEP 339: Design of the CPython compiler - probably the most important and comprehensive piece of official documentation for the Python compiler. Being very short, it painfully displays the scarcity of good documentation of the internals of Python.
"Python Compiler Internals" - an article by Thomas Lee
"Python: Design and Implementation" - a presentation by Guido van Rossum
Python (2.5) Virtual Machine, A guided tour - a presentation by Peter Tröger
original source
One way to do things like this is to preprocess the source and modify it, translating your added statement to python. There are various problems this approach will bring, and I wouldn't recommend it for general usage, but for experimentation with language, or specific-purpose metaprogramming, it can occassionally be useful.
For instance, lets say we want to introduce a "myprint" statement, that instead of printing to the screen instead logs to a specific file. ie:
myprint "This gets logged to file"
would be equivalent to
print >>open('/tmp/logfile.txt','a'), "This gets logged to file"
There are various options as to how to do the replacing, from regex substitution to generating an AST, to writing your own parser depending on how close your syntax matches existing python. A good intermediate approach is to use the tokenizer module. This should allow you to add new keywords, control structures etc while interpreting the source similarly to the python interpreter, thus avoiding the breakage crude regex solutions would cause. For the above "myprint", you could write the following transformation code:
import tokenize
LOGFILE = '/tmp/log.txt'
def translate(readline):
for type, name,_,_,_ in tokenize.generate_tokens(readline):
if type ==tokenize.NAME and name =='myprint':
yield tokenize.NAME, 'print'
yield tokenize.OP, '>>'
yield tokenize.NAME, "open"
yield tokenize.OP, "("
yield tokenize.STRING, repr(LOGFILE)
yield tokenize.OP, ","
yield tokenize.STRING, "'a'"
yield tokenize.OP, ")"
yield tokenize.OP, ","
else:
yield type,name
(This does make myprint effectively a keyword, so use as a variable elsewhere will likely cause problems)
The problem then is how to use it so that your code is usable from python. One way would just be to write your own import function, and use it to load code written in your custom language. ie:
import new
def myimport(filename):
mod = new.module(filename)
f=open(filename)
data = tokenize.untokenize(translate(f.readline))
exec data in mod.__dict__
return mod
This requires you handle your customised code differently from normal python modules however. ie "some_mod = myimport("some_mod.py")" rather than "import some_mod"
Another fairly neat (albeit hacky) solution is to create a custom encoding (See PEP 263) as this recipe demonstrates. You could implement this as:
import codecs, cStringIO, encodings
from encodings import utf_8
class StreamReader(utf_8.StreamReader):
def __init__(self, *args, **kwargs):
codecs.StreamReader.__init__(self, *args, **kwargs)
data = tokenize.untokenize(translate(self.stream.readline))
self.stream = cStringIO.StringIO(data)
def search_function(s):
if s!='mylang': return None
utf8=encodings.search_function('utf8') # Assume utf8 encoding
return codecs.CodecInfo(
name='mylang',
encode = utf8.encode,
decode = utf8.decode,
incrementalencoder=utf8.incrementalencoder,
incrementaldecoder=utf8.incrementaldecoder,
streamreader=StreamReader,
streamwriter=utf8.streamwriter)
codecs.register(search_function)
Now after this code gets run (eg. you could place it in your .pythonrc or site.py) any code starting with the comment "# coding: mylang" will automatically be translated through the above preprocessing step. eg.
# coding: mylang
myprint "this gets logged to file"
for i in range(10):
myprint "so does this : ", i, "times"
myprint ("works fine" "with arbitrary" + " syntax"
"and line continuations")
Caveats:
There are problems to the preprocessor approach, as you'll probably be familiar with if you've worked with the C preprocessor. The main one is debugging. All python sees is the preprocessed file which means that text printed in the stack trace etc will refer to that. If you've performed significant translation, this may be very different from your source text. The example above doesn't change line numbers etc, so won't be too different, but the more you change it, the harder it will be to figure out.
Yes, to some extent it is possible. There is a module out there that uses sys.settrace() to implement goto and comefrom "keywords":
from goto import goto, label
for i in range(1, 10):
for j in range(1, 20):
print i, j
if j == 3:
goto .end # breaking out from nested loop
label .end
print "Finished"
Short of changing and recompiling the source code (which is possible with open source), changing the base language is not really possible.
Even if you do recompile the source, it wouldn't be python, just your hacked-up changed version which you need to be very careful not to introduce bugs into.
However, I'm not sure why you'd want to. Python's object-oriented features makes it quite simple to achieve similar results with the language as it stands.
General answer: you need to preprocess your source files.
More specific answer: install EasyExtend, and go through following steps
i) Create a new langlet ( extension language )
import EasyExtend
EasyExtend.new_langlet("mystmts", prompt = "my> ", source_ext = "mypy")
Without additional specification a bunch of files shall be created under EasyExtend/langlets/mystmts/ .
ii) Open mystmts/parsedef/Grammar.ext and add following lines
small_stmt: (expr_stmt | print_stmt | del_stmt | pass_stmt | flow_stmt |
import_stmt | global_stmt | exec_stmt | assert_stmt | my_stmt )
my_stmt: 'mystatement' expr
This is sufficient to define the syntax of your new statement. The small_stmt non-terminal is part of the Python grammar and it's the place where the new statement is hooked in. The parser will now recognize the new statement i.e. a source file containing it will be parsed. The compiler will reject it though because it still has to be transformed into valid Python.
iii) Now one has to add semantics of the statement. For this one has to edit
msytmts/langlet.py and add a my_stmt node visitor.
def call_my_stmt(expression):
"defines behaviour for my_stmt"
print "my stmt called with", expression
class LangletTransformer(Transformer):
#transform
def my_stmt(self, node):
_expr = find_node(node, symbol.expr)
return any_stmt(CST_CallFunc("call_my_stmt", [_expr]))
__publish__ = ["call_my_stmt"]
iv) cd to langlets/mystmts and type
python run_mystmts.py
Now a session shall be started and the newly defined statement can be used:
__________________________________________________________________________________
mystmts
On Python 2.5.1 (r251:54863, Apr 18 2007, 08:51:08) [MSC v.1310 32 bit (Intel)]
__________________________________________________________________________________
my> mystatement 40+2
my stmt called with 42
Quite a few steps to come to a trivial statement, right? There isn't an API yet that lets one define simple things without having to care about grammars. But EE is very reliable modulo some bugs. So it's just a matter of time that an API emerges that lets programmers define convenient stuff like infix operators or small statements using just convenient OO programming. For more complex things like embedding whole languages in Python by means of building a langlet there is no way of going around a full grammar approach.
Here's a very simple but crappy way to add new statements, in interpretive mode only. I'm using it for little 1-letter commands for editing gene annotations using only sys.displayhook, but just so I could answer this question I added sys.excepthook for the syntax errors as well. The latter is really ugly, fetching the raw code from the readline buffer. The benefit is, it's trivially easy to add new statements this way.
jcomeau#intrepid:~/$ cat demo.py; ./demo.py
#!/usr/bin/python -i
'load everything needed under "package", such as package.common.normalize()'
import os, sys, readline, traceback
if __name__ == '__main__':
class t:
#staticmethod
def localfunction(*args):
print 'this is a test'
if args:
print 'ignoring %s' % repr(args)
def displayhook(whatever):
if hasattr(whatever, 'localfunction'):
return whatever.localfunction()
else:
print whatever
def excepthook(exctype, value, tb):
if exctype is SyntaxError:
index = readline.get_current_history_length()
item = readline.get_history_item(index)
command = item.split()
print 'command:', command
if len(command[0]) == 1:
try:
eval(command[0]).localfunction(*command[1:])
except:
traceback.print_exception(exctype, value, tb)
else:
traceback.print_exception(exctype, value, tb)
sys.displayhook = displayhook
sys.excepthook = excepthook
>>> t
this is a test
>>> t t
command: ['t', 't']
this is a test
ignoring ('t',)
>>> ^D
I've found a guide on adding new statements:
https://troeger.eu/files/teaching/pythonvm08lab.pdf
Basically, to add new statements, you must edit Python/ast.c (among other things) and recompile the python binary.
While it's possible, don't. You can achieve almost everything via functions and classes (which wont require people to recompile python just to run your script..)
It's possible to do this using EasyExtend:
EasyExtend (EE) is a preprocessor
generator and metaprogramming
framework written in pure Python and
integrated with CPython. The main
purpose of EasyExtend is the creation
of extension languages i.e. adding
custom syntax and semantics to Python.
It's not exactly adding new statements to the language syntax, but macros are a powerful tool: https://github.com/lihaoyi/macropy
Some things can be done with decorators. Let's e.g. assume, Python had no with statement. We could then implement a similar behaviour like this:
# ====== Implementation of "mywith" decorator ======
def mywith(stream):
def decorator(function):
try: function(stream)
finally: stream.close()
return decorator
# ====== Using the decorator ======
#mywith(open("test.py","r"))
def _(infile):
for l in infile.readlines():
print(">>", l.rstrip())
It is a pretty unclean solution however as done here. Especially the behaviour where the decorator calls the function and sets _ to None is unexpected. For clarification: This decorator is equivalent to writing
def _(infile): ...
_ = mywith(open(...))(_) # mywith returns None.
and decorators are normally expected to modify, not to execute, functions.
I used such a method before in a script where I had to temporarily set the working directory for several functions.
OUTDATED:
The Logix project is now deprecated and no longer developed, per the Logix website.
There is a language based on python called Logix with which you CAN do such things. It hasn't been under development for a while, but the features that you asked for do work with the latest version.
Not without modifying the interpreter. I know a lot of languages in the past several years have been described as "extensible", but not in the way you're describing. You extend Python by adding functions and classes.
Ten years ago you couldn't, and I doubt that's changed. However, it wasn't that hard to modify the syntax back then if you were prepared to recompile python, and I doubt that's changed, either.
I have started playing with Sage recently, and I've come to suspect that the standard Python int is wrapped in a customized class called Integer in Sage. If I type in type(1) in Python, I get <type 'int'>, however, if I type in the same thing in the sage prompt I get <type 'sage.rings.integer.Integer'>.
If I wanted to replace Python int (or list or dict) with my own custom class, how might it be done? How difficult would it be (e.g. could I do it entirely in Python)?
As an addendum to the other answers: when running any code, Sage has a preprocessing step which converts the Sage-Python to true Python (which is then executed). This is done by the preparse function, e.g.
sage: preparse('a = 1')
'a = Integer(1)'
sage: preparse('2^40')
'Integer(2)**Integer(40)'
sage: preparse('F.<x> = PolynomialRing(ZZ)')
"F = PolynomialRing(ZZ, names=('x',)); (x,) = F._first_ngens(1)"
This step is precisely what allows the transparent use of Integers (in place of ints) and the other non-standard syntax (like the polynomial ring example above and [a..b] etc).
As far as I understand, this is the only way to completely transparently use replacements for the built-in types in Python.
You are able to subclass all of Python's built-in types. For example:
class MyInt(int):
pass
i = MyInt(2)
#i is now an instance of MyInt, but still will behave entirely like an integer.
However, you need to explicitly say each integer is a member of MyInt. So type(1) will still be int, you'll need to do type(MyInt(1)).
Hopefully that's close to what you're looking for.
In the case of Sage, it's easy. Sage has complete control of its own REPL (read-evaluate-print loop), so it can parse the commands you give it and make the parts of your expression into whatever classes it wants. It is not so easy to have standard Python automatically use your integer type for integer literals, however. Simply reassigning the built-in int() to some other type won't do it. You could probably do it with an import filter, that scans each file imported for (say) integer literals and replaces them with MyInt(42) or whatever.
Can you add new statements (like print, raise, with) to Python's syntax?
Say, to allow..
mystatement "Something"
Or,
new_if True:
print "example"
Not so much if you should, but rather if it's possible (short of modifying the python interpreters code)
You may find this useful - Python internals: adding a new statement to Python, quoted here:
This article is an attempt to better understand how the front-end of Python works. Just reading documentation and source code may be a bit boring, so I'm taking a hands-on approach here: I'm going to add an until statement to Python.
All the coding for this article was done against the cutting-edge Py3k branch in the Python Mercurial repository mirror.
The until statement
Some languages, like Ruby, have an until statement, which is the complement to while (until num == 0 is equivalent to while num != 0). In Ruby, I can write:
num = 3
until num == 0 do
puts num
num -= 1
end
And it will print:
3
2
1
So, I want to add a similar capability to Python. That is, being able to write:
num = 3
until num == 0:
print(num)
num -= 1
A language-advocacy digression
This article doesn't attempt to suggest the addition of an until statement to Python. Although I think such a statement would make some code clearer, and this article displays how easy it is to add, I completely respect Python's philosophy of minimalism. All I'm trying to do here, really, is gain some insight into the inner workings of Python.
Modifying the grammar
Python uses a custom parser generator named pgen. This is a LL(1) parser that converts Python source code into a parse tree. The input to the parser generator is the file Grammar/Grammar[1]. This is a simple text file that specifies the grammar of Python.
[1]: From here on, references to files in the Python source are given relatively to the root of the source tree, which is the directory where you run configure and make to build Python.
Two modifications have to be made to the grammar file. The first is to add a definition for the until statement. I found where the while statement was defined (while_stmt), and added until_stmt below [2]:
compound_stmt: if_stmt | while_stmt | until_stmt | for_stmt | try_stmt | with_stmt | funcdef | classdef | decorated
if_stmt: 'if' test ':' suite ('elif' test ':' suite)* ['else' ':' suite]
while_stmt: 'while' test ':' suite ['else' ':' suite]
until_stmt: 'until' test ':' suite
[2]: This demonstrates a common technique I use when modifying source code I’m not familiar with: work by similarity. This principle won’t solve all your problems, but it can definitely ease the process. Since everything that has to be done for while also has to be done for until, it serves as a pretty good guideline.
Note that I've decided to exclude the else clause from my definition of until, just to make it a little bit different (and because frankly I dislike the else clause of loops and don't think it fits well with the Zen of Python).
The second change is to modify the rule for compound_stmt to include until_stmt, as you can see in the snippet above. It's right after while_stmt, again.
When you run make after modifying Grammar/Grammar, notice that the pgen program is run to re-generate Include/graminit.h and Python/graminit.c, and then several files get re-compiled.
Modifying the AST generation code
After the Python parser has created a parse tree, this tree is converted into an AST, since ASTs are much simpler to work with in subsequent stages of the compilation process.
So, we're going to visit Parser/Python.asdl which defines the structure of Python's ASTs and add an AST node for our new until statement, again right below the while:
| While(expr test, stmt* body, stmt* orelse)
| Until(expr test, stmt* body)
If you now run make, notice that before compiling a bunch of files, Parser/asdl_c.py is run to generate C code from the AST definition file. This (like Grammar/Grammar) is another example of the Python source-code using a mini-language (in other words, a DSL) to simplify programming. Also note that since Parser/asdl_c.py is a Python script, this is a kind of bootstrapping - to build Python from scratch, Python already has to be available.
While Parser/asdl_c.py generated the code to manage our newly defined AST node (into the files Include/Python-ast.h and Python/Python-ast.c), we still have to write the code that converts a relevant parse-tree node into it by hand. This is done in the file Python/ast.c. There, a function named ast_for_stmt converts parse tree nodes for statements into AST nodes. Again, guided by our old friend while, we jump right into the big switch for handling compound statements and add a clause for until_stmt:
case while_stmt:
return ast_for_while_stmt(c, ch);
case until_stmt:
return ast_for_until_stmt(c, ch);
Now we should implement ast_for_until_stmt. Here it is:
static stmt_ty
ast_for_until_stmt(struct compiling *c, const node *n)
{
/* until_stmt: 'until' test ':' suite */
REQ(n, until_stmt);
if (NCH(n) == 4) {
expr_ty expression;
asdl_seq *suite_seq;
expression = ast_for_expr(c, CHILD(n, 1));
if (!expression)
return NULL;
suite_seq = ast_for_suite(c, CHILD(n, 3));
if (!suite_seq)
return NULL;
return Until(expression, suite_seq, LINENO(n), n->n_col_offset, c->c_arena);
}
PyErr_Format(PyExc_SystemError,
"wrong number of tokens for 'until' statement: %d",
NCH(n));
return NULL;
}
Again, this was coded while closely looking at the equivalent ast_for_while_stmt, with the difference that for until I've decided not to support the else clause. As expected, the AST is created recursively, using other AST creating functions like ast_for_expr for the condition expression and ast_for_suite for the body of the until statement. Finally, a new node named Until is returned.
Note that we access the parse-tree node n using some macros like NCH and CHILD. These are worth understanding - their code is in Include/node.h.
Digression: AST composition
I chose to create a new type of AST for the until statement, but actually this isn't necessary. I could've saved some work and implemented the new functionality using composition of existing AST nodes, since:
until condition:
# do stuff
Is functionally equivalent to:
while not condition:
# do stuff
Instead of creating the Until node in ast_for_until_stmt, I could have created a Not node with an While node as a child. Since the AST compiler already knows how to handle these nodes, the next steps of the process could be skipped.
Compiling ASTs into bytecode
The next step is compiling the AST into Python bytecode. The compilation has an intermediate result which is a CFG (Control Flow Graph), but since the same code handles it I will ignore this detail for now and leave it for another article.
The code we will look at next is Python/compile.c. Following the lead of while, we find the function compiler_visit_stmt, which is responsible for compiling statements into bytecode. We add a clause for Until:
case While_kind:
return compiler_while(c, s);
case Until_kind:
return compiler_until(c, s);
If you wonder what Until_kind is, it's a constant (actually a value of the _stmt_kind enumeration) automatically generated from the AST definition file into Include/Python-ast.h. Anyway, we call compiler_until which, of course, still doesn't exist. I'll get to it an a moment.
If you're curious like me, you'll notice that compiler_visit_stmt is peculiar. No amount of grep-ping the source tree reveals where it is called. When this is the case, only one option remains - C macro-fu. Indeed, a short investigation leads us to the VISIT macro defined in Python/compile.c:
#define VISIT(C, TYPE, V) {\
if (!compiler_visit_ ## TYPE((C), (V))) \
return 0; \
It's used to invoke compiler_visit_stmt in compiler_body. Back to our business, however...
As promised, here's compiler_until:
static int
compiler_until(struct compiler *c, stmt_ty s)
{
basicblock *loop, *end, *anchor = NULL;
int constant = expr_constant(s->v.Until.test);
if (constant == 1) {
return 1;
}
loop = compiler_new_block(c);
end = compiler_new_block(c);
if (constant == -1) {
anchor = compiler_new_block(c);
if (anchor == NULL)
return 0;
}
if (loop == NULL || end == NULL)
return 0;
ADDOP_JREL(c, SETUP_LOOP, end);
compiler_use_next_block(c, loop);
if (!compiler_push_fblock(c, LOOP, loop))
return 0;
if (constant == -1) {
VISIT(c, expr, s->v.Until.test);
ADDOP_JABS(c, POP_JUMP_IF_TRUE, anchor);
}
VISIT_SEQ(c, stmt, s->v.Until.body);
ADDOP_JABS(c, JUMP_ABSOLUTE, loop);
if (constant == -1) {
compiler_use_next_block(c, anchor);
ADDOP(c, POP_BLOCK);
}
compiler_pop_fblock(c, LOOP, loop);
compiler_use_next_block(c, end);
return 1;
}
I have a confession to make: this code wasn't written based on a deep understanding of Python bytecode. Like the rest of the article, it was done in imitation of the kin compiler_while function. By reading it carefully, however, keeping in mind that the Python VM is stack-based, and glancing into the documentation of the dis module, which has a list of Python bytecodes with descriptions, it's possible to understand what's going on.
That's it, we're done... Aren't we?
After making all the changes and running make, we can run the newly compiled Python and try our new until statement:
>>> until num == 0:
... print(num)
... num -= 1
...
3
2
1
Voila, it works! Let's see the bytecode created for the new statement by using the dis module as follows:
import dis
def myfoo(num):
until num == 0:
print(num)
num -= 1
dis.dis(myfoo)
Here's the result:
4 0 SETUP_LOOP 36 (to 39)
>> 3 LOAD_FAST 0 (num)
6 LOAD_CONST 1 (0)
9 COMPARE_OP 2 (==)
12 POP_JUMP_IF_TRUE 38
5 15 LOAD_NAME 0 (print)
18 LOAD_FAST 0 (num)
21 CALL_FUNCTION 1
24 POP_TOP
6 25 LOAD_FAST 0 (num)
28 LOAD_CONST 2 (1)
31 INPLACE_SUBTRACT
32 STORE_FAST 0 (num)
35 JUMP_ABSOLUTE 3
>> 38 POP_BLOCK
>> 39 LOAD_CONST 0 (None)
42 RETURN_VALUE
The most interesting operation is number 12: if the condition is true, we jump to after the loop. This is correct semantics for until. If the jump isn't executed, the loop body keeps running until it jumps back to the condition at operation 35.
Feeling good about my change, I then tried running the function (executing myfoo(3)) instead of showing its bytecode. The result was less than encouraging:
Traceback (most recent call last):
File "zy.py", line 9, in
myfoo(3)
File "zy.py", line 5, in myfoo
print(num)
SystemError: no locals when loading 'print'
Whoa... this can't be good. So what went wrong?
The case of the missing symbol table
One of the steps the Python compiler performs when compiling the AST is create a symbol table for the code it compiles. The call to PySymtable_Build in PyAST_Compile calls into the symbol table module (Python/symtable.c), which walks the AST in a manner similar to the code generation functions. Having a symbol table for each scope helps the compiler figure out some key information, such as which variables are global and which are local to a scope.
To fix the problem, we have to modify the symtable_visit_stmt function in Python/symtable.c, adding code for handling until statements, after the similar code for while statements [3]:
case While_kind:
VISIT(st, expr, s->v.While.test);
VISIT_SEQ(st, stmt, s->v.While.body);
if (s->v.While.orelse)
VISIT_SEQ(st, stmt, s->v.While.orelse);
break;
case Until_kind:
VISIT(st, expr, s->v.Until.test);
VISIT_SEQ(st, stmt, s->v.Until.body);
break;
[3]: By the way, without this code there’s a compiler warning for Python/symtable.c. The compiler notices that the Until_kind enumeration value isn’t handled in the switch statement of symtable_visit_stmt and complains. It’s always important to check for compiler warnings!
And now we really are done. Compiling the source after this change makes the execution of myfoo(3) work as expected.
Conclusion
In this article I've demonstrated how to add a new statement to Python. Albeit requiring quite a bit of tinkering in the code of the Python compiler, the change wasn't difficult to implement, because I used a similar and existing statement as a guideline.
The Python compiler is a sophisticated chunk of software, and I don't claim being an expert in it. However, I am really interested in the internals of Python, and particularly its front-end. Therefore, I found this exercise a very useful companion to theoretical study of the compiler's principles and source code. It will serve as a base for future articles that will get deeper into the compiler.
References
I used a few excellent references for the construction of this article. Here they are, in no particular order:
PEP 339: Design of the CPython compiler - probably the most important and comprehensive piece of official documentation for the Python compiler. Being very short, it painfully displays the scarcity of good documentation of the internals of Python.
"Python Compiler Internals" - an article by Thomas Lee
"Python: Design and Implementation" - a presentation by Guido van Rossum
Python (2.5) Virtual Machine, A guided tour - a presentation by Peter Tröger
original source
One way to do things like this is to preprocess the source and modify it, translating your added statement to python. There are various problems this approach will bring, and I wouldn't recommend it for general usage, but for experimentation with language, or specific-purpose metaprogramming, it can occassionally be useful.
For instance, lets say we want to introduce a "myprint" statement, that instead of printing to the screen instead logs to a specific file. ie:
myprint "This gets logged to file"
would be equivalent to
print >>open('/tmp/logfile.txt','a'), "This gets logged to file"
There are various options as to how to do the replacing, from regex substitution to generating an AST, to writing your own parser depending on how close your syntax matches existing python. A good intermediate approach is to use the tokenizer module. This should allow you to add new keywords, control structures etc while interpreting the source similarly to the python interpreter, thus avoiding the breakage crude regex solutions would cause. For the above "myprint", you could write the following transformation code:
import tokenize
LOGFILE = '/tmp/log.txt'
def translate(readline):
for type, name,_,_,_ in tokenize.generate_tokens(readline):
if type ==tokenize.NAME and name =='myprint':
yield tokenize.NAME, 'print'
yield tokenize.OP, '>>'
yield tokenize.NAME, "open"
yield tokenize.OP, "("
yield tokenize.STRING, repr(LOGFILE)
yield tokenize.OP, ","
yield tokenize.STRING, "'a'"
yield tokenize.OP, ")"
yield tokenize.OP, ","
else:
yield type,name
(This does make myprint effectively a keyword, so use as a variable elsewhere will likely cause problems)
The problem then is how to use it so that your code is usable from python. One way would just be to write your own import function, and use it to load code written in your custom language. ie:
import new
def myimport(filename):
mod = new.module(filename)
f=open(filename)
data = tokenize.untokenize(translate(f.readline))
exec data in mod.__dict__
return mod
This requires you handle your customised code differently from normal python modules however. ie "some_mod = myimport("some_mod.py")" rather than "import some_mod"
Another fairly neat (albeit hacky) solution is to create a custom encoding (See PEP 263) as this recipe demonstrates. You could implement this as:
import codecs, cStringIO, encodings
from encodings import utf_8
class StreamReader(utf_8.StreamReader):
def __init__(self, *args, **kwargs):
codecs.StreamReader.__init__(self, *args, **kwargs)
data = tokenize.untokenize(translate(self.stream.readline))
self.stream = cStringIO.StringIO(data)
def search_function(s):
if s!='mylang': return None
utf8=encodings.search_function('utf8') # Assume utf8 encoding
return codecs.CodecInfo(
name='mylang',
encode = utf8.encode,
decode = utf8.decode,
incrementalencoder=utf8.incrementalencoder,
incrementaldecoder=utf8.incrementaldecoder,
streamreader=StreamReader,
streamwriter=utf8.streamwriter)
codecs.register(search_function)
Now after this code gets run (eg. you could place it in your .pythonrc or site.py) any code starting with the comment "# coding: mylang" will automatically be translated through the above preprocessing step. eg.
# coding: mylang
myprint "this gets logged to file"
for i in range(10):
myprint "so does this : ", i, "times"
myprint ("works fine" "with arbitrary" + " syntax"
"and line continuations")
Caveats:
There are problems to the preprocessor approach, as you'll probably be familiar with if you've worked with the C preprocessor. The main one is debugging. All python sees is the preprocessed file which means that text printed in the stack trace etc will refer to that. If you've performed significant translation, this may be very different from your source text. The example above doesn't change line numbers etc, so won't be too different, but the more you change it, the harder it will be to figure out.
Yes, to some extent it is possible. There is a module out there that uses sys.settrace() to implement goto and comefrom "keywords":
from goto import goto, label
for i in range(1, 10):
for j in range(1, 20):
print i, j
if j == 3:
goto .end # breaking out from nested loop
label .end
print "Finished"
Short of changing and recompiling the source code (which is possible with open source), changing the base language is not really possible.
Even if you do recompile the source, it wouldn't be python, just your hacked-up changed version which you need to be very careful not to introduce bugs into.
However, I'm not sure why you'd want to. Python's object-oriented features makes it quite simple to achieve similar results with the language as it stands.
General answer: you need to preprocess your source files.
More specific answer: install EasyExtend, and go through following steps
i) Create a new langlet ( extension language )
import EasyExtend
EasyExtend.new_langlet("mystmts", prompt = "my> ", source_ext = "mypy")
Without additional specification a bunch of files shall be created under EasyExtend/langlets/mystmts/ .
ii) Open mystmts/parsedef/Grammar.ext and add following lines
small_stmt: (expr_stmt | print_stmt | del_stmt | pass_stmt | flow_stmt |
import_stmt | global_stmt | exec_stmt | assert_stmt | my_stmt )
my_stmt: 'mystatement' expr
This is sufficient to define the syntax of your new statement. The small_stmt non-terminal is part of the Python grammar and it's the place where the new statement is hooked in. The parser will now recognize the new statement i.e. a source file containing it will be parsed. The compiler will reject it though because it still has to be transformed into valid Python.
iii) Now one has to add semantics of the statement. For this one has to edit
msytmts/langlet.py and add a my_stmt node visitor.
def call_my_stmt(expression):
"defines behaviour for my_stmt"
print "my stmt called with", expression
class LangletTransformer(Transformer):
#transform
def my_stmt(self, node):
_expr = find_node(node, symbol.expr)
return any_stmt(CST_CallFunc("call_my_stmt", [_expr]))
__publish__ = ["call_my_stmt"]
iv) cd to langlets/mystmts and type
python run_mystmts.py
Now a session shall be started and the newly defined statement can be used:
__________________________________________________________________________________
mystmts
On Python 2.5.1 (r251:54863, Apr 18 2007, 08:51:08) [MSC v.1310 32 bit (Intel)]
__________________________________________________________________________________
my> mystatement 40+2
my stmt called with 42
Quite a few steps to come to a trivial statement, right? There isn't an API yet that lets one define simple things without having to care about grammars. But EE is very reliable modulo some bugs. So it's just a matter of time that an API emerges that lets programmers define convenient stuff like infix operators or small statements using just convenient OO programming. For more complex things like embedding whole languages in Python by means of building a langlet there is no way of going around a full grammar approach.
Here's a very simple but crappy way to add new statements, in interpretive mode only. I'm using it for little 1-letter commands for editing gene annotations using only sys.displayhook, but just so I could answer this question I added sys.excepthook for the syntax errors as well. The latter is really ugly, fetching the raw code from the readline buffer. The benefit is, it's trivially easy to add new statements this way.
jcomeau#intrepid:~/$ cat demo.py; ./demo.py
#!/usr/bin/python -i
'load everything needed under "package", such as package.common.normalize()'
import os, sys, readline, traceback
if __name__ == '__main__':
class t:
#staticmethod
def localfunction(*args):
print 'this is a test'
if args:
print 'ignoring %s' % repr(args)
def displayhook(whatever):
if hasattr(whatever, 'localfunction'):
return whatever.localfunction()
else:
print whatever
def excepthook(exctype, value, tb):
if exctype is SyntaxError:
index = readline.get_current_history_length()
item = readline.get_history_item(index)
command = item.split()
print 'command:', command
if len(command[0]) == 1:
try:
eval(command[0]).localfunction(*command[1:])
except:
traceback.print_exception(exctype, value, tb)
else:
traceback.print_exception(exctype, value, tb)
sys.displayhook = displayhook
sys.excepthook = excepthook
>>> t
this is a test
>>> t t
command: ['t', 't']
this is a test
ignoring ('t',)
>>> ^D
I've found a guide on adding new statements:
https://troeger.eu/files/teaching/pythonvm08lab.pdf
Basically, to add new statements, you must edit Python/ast.c (among other things) and recompile the python binary.
While it's possible, don't. You can achieve almost everything via functions and classes (which wont require people to recompile python just to run your script..)
It's possible to do this using EasyExtend:
EasyExtend (EE) is a preprocessor
generator and metaprogramming
framework written in pure Python and
integrated with CPython. The main
purpose of EasyExtend is the creation
of extension languages i.e. adding
custom syntax and semantics to Python.
It's not exactly adding new statements to the language syntax, but macros are a powerful tool: https://github.com/lihaoyi/macropy
Some things can be done with decorators. Let's e.g. assume, Python had no with statement. We could then implement a similar behaviour like this:
# ====== Implementation of "mywith" decorator ======
def mywith(stream):
def decorator(function):
try: function(stream)
finally: stream.close()
return decorator
# ====== Using the decorator ======
#mywith(open("test.py","r"))
def _(infile):
for l in infile.readlines():
print(">>", l.rstrip())
It is a pretty unclean solution however as done here. Especially the behaviour where the decorator calls the function and sets _ to None is unexpected. For clarification: This decorator is equivalent to writing
def _(infile): ...
_ = mywith(open(...))(_) # mywith returns None.
and decorators are normally expected to modify, not to execute, functions.
I used such a method before in a script where I had to temporarily set the working directory for several functions.
OUTDATED:
The Logix project is now deprecated and no longer developed, per the Logix website.
There is a language based on python called Logix with which you CAN do such things. It hasn't been under development for a while, but the features that you asked for do work with the latest version.
Not without modifying the interpreter. I know a lot of languages in the past several years have been described as "extensible", but not in the way you're describing. You extend Python by adding functions and classes.
Ten years ago you couldn't, and I doubt that's changed. However, it wasn't that hard to modify the syntax back then if you were prepared to recompile python, and I doubt that's changed, either.