297 lines
11 KiB
Python
297 lines
11 KiB
Python
"""Parse a Python file and retrieve classes and methods.
|
|
|
|
Parse enough of a Python file to recognize class and method
|
|
definitions and to find out the superclasses of a class.
|
|
|
|
The interface consists of a single function:
|
|
readmodule_ex(module [, path[, inpackage]])
|
|
module is the name of a Python module, path is an optional list of
|
|
directories where the module is to be searched. If present, path is
|
|
prepended to the system search path sys.path. (inpackage is used
|
|
internally to search for a submodule of a package.)
|
|
The return value is a dictionary. The keys of the dictionary are
|
|
the names of the classes defined in the module (including classes
|
|
that are defined via the from XXX import YYY construct). The values
|
|
are class instances of the class Class defined here.
|
|
|
|
A class is described by the class Class in this module. Instances
|
|
of this class have the following instance variables:
|
|
name -- the name of the class
|
|
super -- a list of super classes (Class instances)
|
|
methods -- a dictionary of methods
|
|
file -- the file in which the class was defined
|
|
lineno -- the line in the file on which the class statement occurred
|
|
The dictionary of methods uses the method names as keys and the line
|
|
numbers on which the method was defined as values.
|
|
If the name of a super class is not recognized, the corresponding
|
|
entry in the list of super classes is not a class instance but a
|
|
string giving the name of the super class. Since import statements
|
|
are recognized and imported modules are scanned as well, this
|
|
shouldn't happen often.
|
|
|
|
XXX describe the Function class.
|
|
|
|
BUGS
|
|
- Nested classes and functions can confuse it.
|
|
|
|
PACKAGE RELATED BUGS
|
|
- If you have a package and a module inside that or another package
|
|
with the same name, module caching doesn't work properly since the
|
|
key is the base name of the module/package.
|
|
- The only entry that is returned when you readmodule a package is a
|
|
__path__ whose value is a list which confuses certain class browsers.
|
|
- When code does:
|
|
from package import subpackage
|
|
class MyClass(subpackage.SuperClass):
|
|
...
|
|
It can't locate the parent. It probably needs to have the same
|
|
hairy logic that the import locator already does. (This logic
|
|
exists coded in Python in the freeze package.)
|
|
"""
|
|
|
|
import sys
|
|
import imp
|
|
import tokenize # Python tokenizer
|
|
from token import NAME
|
|
|
|
__all__ = ["readmodule"]
|
|
|
|
_modules = {} # cache of modules we've seen
|
|
|
|
# each Python class is represented by an instance of this class
|
|
class Class:
|
|
'''Class to represent a Python class.'''
|
|
def __init__(self, module, name, super, file, lineno):
|
|
self.module = module
|
|
self.name = name
|
|
if super is None:
|
|
super = []
|
|
self.super = super
|
|
self.methods = {}
|
|
self.file = file
|
|
self.lineno = lineno
|
|
|
|
def _addmethod(self, name, lineno):
|
|
self.methods[name] = lineno
|
|
|
|
class Function(Class):
|
|
'''Class to represent a top-level Python function'''
|
|
def __init__(self, module, name, file, lineno):
|
|
Class.__init__(self, module, name, None, file, lineno)
|
|
def _addmethod(self, name, lineno):
|
|
assert 0, "Function._addmethod() shouldn't be called"
|
|
|
|
def readmodule(module, path=[], inpackage=False):
|
|
'''Backwards compatible interface.
|
|
|
|
Like readmodule_ex() but strips Function objects from the
|
|
resulting dictionary.'''
|
|
|
|
dict = readmodule_ex(module, path, inpackage)
|
|
res = {}
|
|
for key, value in dict.items():
|
|
if not isinstance(value, Function):
|
|
res[key] = value
|
|
return res
|
|
|
|
def readmodule_ex(module, path=[], inpackage=False):
|
|
'''Read a module file and return a dictionary of classes.
|
|
|
|
Search for MODULE in PATH and sys.path, read and parse the
|
|
module and return a dictionary with one entry for each class
|
|
found in the module.'''
|
|
|
|
dict = {}
|
|
|
|
i = module.rfind('.')
|
|
if i >= 0:
|
|
# Dotted module name
|
|
package = module[:i].strip()
|
|
submodule = module[i+1:].strip()
|
|
parent = readmodule_ex(package, path, inpackage)
|
|
child = readmodule_ex(submodule, parent['__path__'], True)
|
|
return child
|
|
|
|
if module in _modules:
|
|
# we've seen this module before...
|
|
return _modules[module]
|
|
if module in sys.builtin_module_names:
|
|
# this is a built-in module
|
|
_modules[module] = dict
|
|
return dict
|
|
|
|
# search the path for the module
|
|
f = None
|
|
if inpackage:
|
|
try:
|
|
f, file, (suff, mode, type) = \
|
|
imp.find_module(module, path)
|
|
except ImportError:
|
|
f = None
|
|
if f is None:
|
|
fullpath = list(path) + sys.path
|
|
f, file, (suff, mode, type) = imp.find_module(module, fullpath)
|
|
if type == imp.PKG_DIRECTORY:
|
|
dict['__path__'] = [file]
|
|
_modules[module] = dict
|
|
path = [file] + path
|
|
f, file, (suff, mode, type) = \
|
|
imp.find_module('__init__', [file])
|
|
if type != imp.PY_SOURCE:
|
|
# not Python source, can't do anything with this module
|
|
f.close()
|
|
_modules[module] = dict
|
|
return dict
|
|
|
|
_modules[module] = dict
|
|
classstack = [] # stack of (class, indent) pairs
|
|
|
|
g = tokenize.generate_tokens(f.readline)
|
|
try:
|
|
for tokentype, token, start, end, line in g:
|
|
if token == 'def':
|
|
lineno, thisindent = start
|
|
tokentype, meth_name, start, end, line = g.next()
|
|
if tokentype != NAME:
|
|
continue # Syntax error
|
|
# close all classes indented at least as much
|
|
while classstack and \
|
|
classstack[-1][1] >= thisindent:
|
|
del classstack[-1]
|
|
if classstack:
|
|
# it's a class method
|
|
cur_class = classstack[-1][0]
|
|
cur_class._addmethod(meth_name, lineno)
|
|
else:
|
|
# it's a function
|
|
dict[meth_name] = Function(module, meth_name, file, lineno)
|
|
elif token == 'class':
|
|
lineno, thisindent = start
|
|
tokentype, class_name, start, end, line = g.next()
|
|
if tokentype != NAME:
|
|
continue # Syntax error
|
|
# close all classes indented at least as much
|
|
while classstack and \
|
|
classstack[-1][1] >= thisindent:
|
|
del classstack[-1]
|
|
# parse what follows the class name
|
|
tokentype, token, start, end, line = g.next()
|
|
inherit = None
|
|
if token == '(':
|
|
names = [] # List of superclasses
|
|
# there's a list of superclasses
|
|
level = 1
|
|
super = [] # Tokens making up current superclass
|
|
while True:
|
|
tokentype, token, start, end, line = g.next()
|
|
if token in (')', ',') and level == 1:
|
|
n = "".join(super)
|
|
if n in dict:
|
|
# we know this super class
|
|
n = dict[n]
|
|
else:
|
|
c = n.split('.')
|
|
if len(c) > 1:
|
|
# super class is of the form
|
|
# module.class: look in module for
|
|
# class
|
|
m = c[-2]
|
|
c = c[-1]
|
|
if m in _modules:
|
|
d = _modules[m]
|
|
if c in d:
|
|
n = d[c]
|
|
names.append(n)
|
|
if token == '(':
|
|
level += 1
|
|
elif token == ')':
|
|
level -= 1
|
|
if level == 0:
|
|
break
|
|
elif token == ',' and level == 1:
|
|
pass
|
|
else:
|
|
super.append(token)
|
|
inherit = names
|
|
cur_class = Class(module, class_name, inherit, file, lineno)
|
|
dict[class_name] = cur_class
|
|
classstack.append((cur_class, thisindent))
|
|
elif token == 'import' and start[1] == 0:
|
|
modules = _getnamelist(g)
|
|
for mod, mod2 in modules:
|
|
try:
|
|
# Recursively read the imported module
|
|
readmodule_ex(mod, path, inpackage)
|
|
except:
|
|
# If we can't find or parse the imported module,
|
|
# too bad -- don't die here.
|
|
pass
|
|
elif token == 'from' and start[1] == 0:
|
|
mod, token = _getname(g)
|
|
if not mod or token != "import":
|
|
continue
|
|
names = _getnamelist(g)
|
|
try:
|
|
# Recursively read the imported module
|
|
d = readmodule_ex(mod, path, inpackage)
|
|
except:
|
|
# If we can't find or parse the imported module,
|
|
# too bad -- don't die here.
|
|
continue
|
|
# add any classes that were defined in the imported module
|
|
# to our name space if they were mentioned in the list
|
|
for n, n2 in names:
|
|
if n in d:
|
|
dict[n2 or n] = d[n]
|
|
elif n == '*':
|
|
# only add a name if not already there (to mimic
|
|
# what Python does internally) also don't add
|
|
# names that start with _
|
|
for n in d:
|
|
if n[0] != '_' and not n in dict:
|
|
dict[n] = d[n]
|
|
except StopIteration:
|
|
pass
|
|
|
|
f.close()
|
|
return dict
|
|
|
|
def _getnamelist(g):
|
|
# Helper to get a comma-separated list of dotted names plus 'as'
|
|
# clauses. Return a list of pairs (name, name2) where name2 is
|
|
# the 'as' name, or None if there is no 'as' clause.
|
|
names = []
|
|
while True:
|
|
name, token = _getname(g)
|
|
if not name:
|
|
break
|
|
if token == 'as':
|
|
name2, token = _getname(g)
|
|
else:
|
|
name2 = None
|
|
names.append((name, name2))
|
|
while token != "," and "\n" not in token:
|
|
tokentype, token, start, end, line = g.next()
|
|
if token != ",":
|
|
break
|
|
return names
|
|
|
|
def _getname(g):
|
|
# Helper to get a dotted name, return a pair (name, token) where
|
|
# name is the dotted name, or None if there was no dotted name,
|
|
# and token is the next input token.
|
|
parts = []
|
|
tokentype, token, start, end, line = g.next()
|
|
if tokentype != NAME and token != '*':
|
|
return (None, token)
|
|
parts.append(token)
|
|
while True:
|
|
tokentype, token, start, end, line = g.next()
|
|
if token != '.':
|
|
break
|
|
tokentype, token, start, end, line = g.next()
|
|
if tokentype != NAME:
|
|
break
|
|
parts.append(token)
|
|
return (".".join(parts), token)
|