Files
openide/python/helpers/generator3/module_redeclarator.py
Mikhail Golubev 383a76d223 [python] Make paths to binary modules in skeleton generator tests OS-independent
On Windows, we added paths containing backslashes to skeleton headers, while the expected test
data contains unix-style paths. Now in tests these paths are normalized to forward-slashes
regardless of the platform. It wasn't noticed before 575f96b1d5f72c5159eac8a6a368806f34d9734e
because earlier we didn't test modules containing more than one component in their relative
path to the corresponding test root.

GitOrigin-RevId: 6ad3436c26d4d8e7e693e37dcfa5d3ecdab59e83
2023-07-13 10:14:53 +00:00

1140 lines
52 KiB
Python

from generator3.core import OriginType
from generator3.util_methods import *
from generator3.util_methods import get_portable_test_module_path
from generator3.docstring_parsing import *
class emptylistdict(dict):
"""defaultdict not available before 2.5; simplest reimplementation using [] as default"""
def __getitem__(self, item):
if item in self:
return dict.__getitem__(self, item)
else:
it = []
self.__setitem__(item, it)
return it
class Buf(object):
"""Buffers data in a list, can write to a file. Indentation is provided externally."""
def __init__(self, indenter):
self.data = []
self.indenter = indenter
def put(self, data):
if data:
self.data.append(ensureUnicode(data))
def out(self, indent, *what):
"""Output the arguments, indenting as needed, and adding an eol"""
self.put(self.indenter.indent(indent))
for item in what:
self.put(item)
self.put("\n")
def flush_bytes(self, outfile):
for data in self.data:
outfile.write(data.encode(OUT_ENCODING, "replace"))
def flush_str(self, outfile):
for data in self.data:
outfile.write(data)
if version[0] < 3:
flush = flush_bytes
else:
flush = flush_str
def isEmpty(self):
return len(self.data) == 0
class ClassBuf(Buf):
def __init__(self, name, indenter):
super(ClassBuf, self).__init__(indenter)
self.name = name
#noinspection PyBroadException
class ModuleRedeclarator(object):
def __init__(self, module, mod_qname, mod_filename, cache_dir, indent_size=4, doing_builtins=False):
"""
@param module: module object
@param mod_qname: module qualified name
@param mod_filename: filename of binary module (the .dll or .so). Can be None for modules
that don't have corresponding binary files (e.g. builtins)
@param cache_dir: per-binary cache directory where the generated stub will be stored.
Normally, it's "<IDE system dir>/python_stubs/cache/<sha256 digest>/".
@param indent_size: amount of space characters per indent
"""
import generator3.core
self.test_mode = generator3.core.is_test_mode()
self.gen_version = generator3.core.version()
self.module = module
self.qname = mod_qname
self.cache_dir = cache_dir
self.mod_filename = mod_filename
# we write things into buffers out-of-order
self.header_buf = Buf(self)
self.imports_buf = Buf(self)
self.functions_buf = Buf(self)
self.classes_buf = Buf(self)
self.classes_buffs = list()
self.footer_buf = Buf(self)
self.indent_size = indent_size
self._indent_step = " " * self.indent_size
self.split_modules = False
#
self.imported_modules = {"": the_builtins} # explicit module imports: {"name": module}
self.hidden_imports = {} # {'real_mod_name': 'alias'}; we alias names with "__" since we don't want them exported
# ^ used for things that we don't re-export but need to import, e.g. certain base classes in gnome.
self._defined = {} # stores True for every name defined so far, to break circular refs in values
self.doing_builtins = doing_builtins
self.ret_type_cache = {}
self.used_imports = emptylistdict() # qual_mod_name -> [imported_names,..]: actually used imported names
def _initializeQApp4(self):
try: # QtGui should be imported _before_ QtCore package.
# This is done for the QWidget references from QtCore (such as QSignalMapper). Known bug in PyQt 4.7+
# Causes "TypeError: C++ type 'QWidget*' is not supported as a native Qt signal type"
import PyQt4.QtGui
except ImportError:
pass
# manually instantiate and keep reference to singleton QCoreApplication (we don't want it to be deleted during the introspection)
# use QCoreApplication instead of QApplication to avoid blinking app in Dock on Mac OS
try:
from PyQt4.QtCore import QCoreApplication
self.app = QCoreApplication([])
return
except ImportError:
pass
def _initializeQApp5(self):
try:
from PyQt5.QtCore import QCoreApplication
self.app = QCoreApplication([])
return
except ImportError:
pass
def indent(self, level):
"""Return indentation whitespace for given level."""
return self._indent_step * level
def flush(self):
qname_parts = self.qname.split('.')
if self.split_modules:
last_pkg_dir = build_pkg_structure(self.cache_dir, self.qname)
with fopen(os.path.join(last_pkg_dir, "__init__.py"), "w") as init:
for buf in (self.header_buf, self.imports_buf, self.functions_buf, self.classes_buf):
buf.flush(init)
data = ""
for buf in self.classes_buffs:
with fopen(os.path.join(last_pkg_dir, buf.name) + '.py', "w") as dummy:
self.header_buf.flush(dummy)
self.imports_buf.flush(dummy)
buf.flush(dummy)
data += self.create_local_import(buf.name)
init.write(data)
self.footer_buf.flush(init)
else:
last_pkg_dir = build_pkg_structure(self.cache_dir, '.'.join(qname_parts[:-1]))
# In some rare cases submodules of a binary might have been generated earlier than the module
# for the binary itself. For instance, it happens for "pyexpat" built-in module which
# submodules "pyexpat.errors" and "pyexpat.model" are processed together with "_elementtree"
# and "pickle" before "pyexpat" and thus empty pyexpat/__init__.py for them should be replaced
# with the skeleton for the main module itself later on.
existing_pkg_init = os.path.join(last_pkg_dir, qname_parts[-1], '__init__.py')
if os.path.exists(existing_pkg_init):
skeleton_path = existing_pkg_init
else:
skeleton_path = os.path.join(last_pkg_dir, qname_parts[-1] + '.py')
with fopen(skeleton_path, "w") as mod:
for buf in (self.header_buf, self.imports_buf, self.functions_buf, self.classes_buf):
buf.flush(mod)
for buf in self.classes_buffs:
buf.flush(mod)
self.footer_buf.flush(mod)
# Some builtin classes effectively change __init__ signature without overriding it.
# This callable serves as a placeholder to be replaced via REDEFINED_BUILTIN_SIGS
def fake_builtin_init(self):
pass # just a callable, sig doesn't matter
fake_builtin_init.__doc__ = object.__init__.__doc__ # this forces class's doc to be used instead
def create_local_import(self, name):
if len(name.split(".")) > 1: return ""
data = "from "
if version[0] >= 3:
data += "."
data += name + " import " + name + "\n"
return data
def find_imported_name(self, item):
"""
Finds out how the item is represented in imported modules.
@param item what to check
@return qualified name (like "sys.stdin") or None
"""
# TODO: return a pair, not a glued string
if not isinstance(item, SIMPLEST_TYPES):
for mname in self.imported_modules:
m = self.imported_modules[mname]
for inner_name in m.__dict__:
suspect = getattr(m, inner_name)
if suspect is item:
if mname:
mname += "."
elif self.module is the_builtins: # don't short-circuit builtins
return None
return mname + inner_name
return None
_initializers = (
(dict, "{}"),
(tuple, "()"),
(list, "[]"),
)
def invent_initializer(self, a_type):
"""
Returns an innocuous initializer expression for a_type, or "None"
"""
for initializer_type, r in self._initializers:
if initializer_type == a_type:
return r
# NOTE: here we could handle things like defaultdict, sets, etc if we wanted
return "None"
def fmt_value(self, out, p_value, indent, prefix="", postfix="", as_name=None, seen_values=None):
"""
Formats and outputs value (it occupies an entire line or several lines).
@param out function that does output (a Buf.out)
@param p_value the value.
@param indent indent level.
@param prefix text to print before the value
@param postfix text to print after the value
@param as_name hints which name are we trying to print; helps with circular refs.
@param seen_values a list of keys we've seen if we're processing a dict
"""
SELF_VALUE = "<value is a self-reference, replaced by this string>"
ERR_VALUE = "<failed to retrieve the value>"
if isinstance(p_value, SIMPLEST_TYPES):
out(indent, prefix, reliable_repr(p_value), postfix)
else:
if sys.platform == "cli":
imported_name = None
else:
imported_name = self.find_imported_name(p_value)
if imported_name:
out(indent, prefix, imported_name, postfix)
# TODO: kind of self.used_imports[imported_name].append(p_value) but split imported_name
# else we could potentially return smth we did not otherwise import. but not likely.
else:
if isinstance(p_value, (list, tuple)):
if not seen_values:
seen_values = [p_value]
if len(p_value) == 0:
out(indent, prefix, repr(p_value), postfix)
else:
if isinstance(p_value, list):
lpar, rpar = "[", "]"
else:
lpar, rpar = "(", ")"
out(indent, prefix, lpar)
for value in p_value:
if value in seen_values:
value = SELF_VALUE
elif not isinstance(value, SIMPLEST_TYPES):
seen_values.append(value)
self.fmt_value(out, value, indent + 1, postfix=",", seen_values=seen_values)
out(indent, rpar, postfix)
elif isinstance(p_value, dict):
if len(p_value) == 0:
out(indent, prefix, repr(p_value), postfix)
else:
if not seen_values:
seen_values = [p_value]
out(indent, prefix, "{")
keys = list(p_value.keys())
try:
keys.sort()
except TypeError:
pass # unsortable keys happen, e,g, in py3k _ctypes
for k in keys:
value = p_value[k]
try:
is_seen = value in seen_values
except:
is_seen = False
value = ERR_VALUE
if is_seen:
value = SELF_VALUE
elif not isinstance(value, SIMPLEST_TYPES):
seen_values.append(value)
if isinstance(k, SIMPLEST_TYPES):
self.fmt_value(out, value, indent + 1, prefix=repr(k) + ": ", postfix=",",
seen_values=seen_values)
else:
# both key and value need fancy formatting
self.fmt_value(out, k, indent + 1, postfix=": ", seen_values=seen_values)
self.fmt_value(out, value, indent + 2, seen_values=seen_values)
out(indent + 1, ",")
out(indent, "}", postfix)
else: # something else, maybe representable
# look up this value in the module.
if sys.platform == "cli":
out(indent, prefix, "None", postfix)
return
found_name = ""
for inner_name in self.module.__dict__:
if self.module.__dict__[inner_name] is p_value:
found_name = inner_name
break
if self._defined.get(found_name, False):
out(indent, prefix, found_name, postfix)
elif hasattr(self, "app"):
return
else:
# a forward / circular declaration happens
notice = ""
try:
representation = repr(p_value)
except Exception:
import traceback
traceback.print_exc(file=sys.stderr)
return
if not self.test_mode:
real_value = cleanup(representation)
else:
# Don't rely on repr() output in tests, as it may contain memory layout dependent id
real_value = ''
if found_name:
if found_name == as_name:
notice = " # (!) real value is %r" % real_value
real_value = "None"
else:
notice = " # (!) forward: %s, real value is %r" % (found_name, real_value)
if SANE_REPR_RE.match(real_value) and is_valid_expr(real_value):
out(indent, prefix, real_value, postfix, notice)
else:
if not found_name:
notice = " # (!) real value is %r" % real_value
out(indent, prefix, "None", postfix, notice)
def get_ret_type(self, attr):
"""
Returns a return type string as given by T_RETURN in tokens, or None
"""
if attr:
ret_type = RET_TYPE.get(attr, None)
if ret_type:
return ret_type
thing = getattr(self.module, attr, None)
if thing:
if not isinstance(thing, type) and is_callable(thing): # a function
return None # TODO: maybe divinate a return type; see pygame.mixer.Channel
return attr
# adds no noticeable slowdown, I did measure. dch.
for im_name, im_module in self.imported_modules.items():
cache_key = (im_name, attr)
cached = self.ret_type_cache.get(cache_key, None)
if cached:
return cached
ret_type = getattr(im_module, attr, None)
if ret_type:
if isinstance(ret_type, type):
# detect a constructor
constr_args = detect_constructor(ret_type)
if constr_args is None:
constr_args = "*(), **{}" # a silly catch-all constructor
reference = "%s(%s)" % (attr, constr_args)
elif is_callable(ret_type): # a function, classes are ruled out above
return None
else:
reference = attr
if im_name:
result = "%s.%s" % (im_name, reference)
else: # built-in
result = reference
self.ret_type_cache[cache_key] = result
return result
# TODO: handle things like "[a, b,..] and (foo,..)"
return None
SIG_DOC_NOTE = "restored from __doc__"
SIG_DOC_UNRELIABLY = "NOTE: unreliably restored from __doc__ "
def restore_by_docstring(self, signature_string, class_name, deco=None, ret_hint=None):
"""
@param signature_string: parameter list extracted from the doc string.
@param class_name: name of the containing class, or None
@param deco: decorator to use
@param ret_hint: return type hint, if available
@return (reconstructed_spec, return_type, note) or (None, _, _) if failed.
"""
action("restoring func %r of class %r", signature_string, class_name)
# parse
parsing_failed = False
ret_type = None
try:
# strict parsing
tokens = paramSeqAndRest.parseString(signature_string, True)
ret_name = None
if tokens:
ret_t = tokens[-1]
if ret_t[0] is T_RETURN:
ret_name = ret_t[1]
ret_type = self.get_ret_type(ret_name) or self.get_ret_type(ret_hint)
except ParseException:
# it did not parse completely; scavenge what we can
parsing_failed = True
tokens = []
try:
# most unrestrictive parsing
tokens = paramSeq.parseString(signature_string, False)
except ParseException:
pass
#
seq = transform_seq(tokens)
# add safe defaults for unparsed
if parsing_failed:
doc_node = self.SIG_DOC_UNRELIABLY
starred = None
double_starred = None
for one in seq:
if type(one) is str:
if one.startswith("**"):
double_starred = one
elif one.startswith("*"):
starred = one
if not starred:
seq.append("*args")
if not double_starred:
seq.append("**kwargs")
else:
doc_node = self.SIG_DOC_NOTE
# add 'self' if needed YYY
if class_name and (not seq or seq[0] != 'self'):
first_param = propose_first_param(deco)
if first_param:
seq.insert(0, first_param)
seq = make_names_unique(seq)
return (seq, ret_type, doc_node)
def parse_func_doc(self, func_doc, func_id, func_name, class_name, deco=None, sip_generated=False):
"""
@param func_doc: __doc__ of the function.
@param func_id: name to look for as identifier of the function in docstring
@param func_name: name of the function.
@param class_name: name of the containing class, or None
@param deco: decorator to use
@return (reconstructed_spec, return_literal, note) or (None, _, _) if failed.
"""
if sip_generated:
overloads = []
for part in func_doc.split('\n'):
signature = func_id + '('
i = part.find(signature)
if i >= 0:
overloads.append(part[i + len(signature):])
if len(overloads) > 1:
docstring_results = [self.restore_by_docstring(overload, class_name, deco) for overload in overloads]
ret_types = []
for result in docstring_results:
rt = result[1]
if rt and rt not in ret_types:
ret_types.append(rt)
if ret_types:
ret_literal = " or ".join(ret_types)
else:
ret_literal = None
param_lists = [result[0] for result in docstring_results]
spec = build_signature(func_name, restore_parameters_for_overloads(param_lists))
return (spec, ret_literal, "restored from __doc__ with multiple overloads")
# find the first thing to look like a definition
prefix_re = re.compile(r"\s*(?:(\w+)[ \t]+)?" + func_id + r"\s*\(") # "foo(..." or "int foo(..."
match = prefix_re.search(func_doc) # Note: this and previous line may consume up to 35% of time
# parse the part that looks right
if match:
ret_hint = match.group(1)
params, ret_literal, doc_note = self.restore_by_docstring(func_doc[match.end():], class_name, deco, ret_hint)
spec = func_name + flatten(params)
return (spec, ret_literal, doc_note)
else:
return (None, None, None)
def is_predefined_builtin(self, module_name, class_name, func_name):
return self.doing_builtins and module_name == BUILTIN_MOD_NAME and (
class_name, func_name) in PREDEFINED_BUILTIN_SIGS
def redo_function(self, out, p_func, p_name, indent, p_class=None, p_modname=None, classname=None, seen=None):
"""
Restore function argument list as best we can.
@param out output function of a Buf
@param p_func function or method object
@param p_name function name as known to owner
@param indent indentation level
@param p_class the class that contains this function as a method
@param p_modname module name
@param seen {id(func): name} map of functions already seen in the same namespace;
id() because *some* functions are unhashable (eg _elementtree.Comment in py2.7)
"""
action("redoing func %r of class %r", p_name, p_class)
if seen is not None:
other_func = seen.get(id(p_func), None)
if other_func and getattr(other_func, "__doc__", None) is getattr(p_func, "__doc__", None):
# _bisect.bisect == _bisect.bisect_right in py31, but docs differ
out(indent, p_name, " = ", seen[id(p_func)])
out(indent, "")
return
else:
seen[id(p_func)] = p_name
# real work
if classname is None:
classname = p_class and p_class.__name__ or None
if p_class and hasattr(p_class, '__mro__'):
sip_generated = [base_t for base_t in p_class.__mro__ if 'sip.simplewrapper' in str(base_t)]
else:
sip_generated = False
deco = None
deco_comment = ""
mod_class_method_tuple = (p_modname, classname, p_name)
ret_literal = None
is_init = False
# any decorators?
action("redoing decos of func %r of class %r", p_name, p_class)
if self.doing_builtins and p_modname == BUILTIN_MOD_NAME:
deco = KNOWN_DECORATORS.get((classname, p_name), None)
if deco:
deco_comment = " # known case"
elif p_class and p_name in p_class.__dict__:
# detect native methods declared with METH_CLASS flag
descriptor = p_class.__dict__[p_name]
if p_name != "__new__" and type(descriptor).__name__.startswith('classmethod'):
# 'classmethod_descriptor' in Python 2.x and 3.x, 'classmethod' in Jython
deco = "classmethod"
elif type(p_func).__name__.startswith('staticmethod'):
deco = "staticmethod"
if p_name == "__new__":
deco = "staticmethod"
deco_comment = " # known case of __new__"
action("redoing innards of func %r of class %r", p_name, p_class)
if deco and HAS_DECORATORS:
out(indent, "@", deco, deco_comment)
if inspect and inspect.isfunction(p_func):
out(indent, "def ", p_name, restore_by_inspect(p_func), ": # reliably restored by inspect", )
out_doc_attr(out, p_func, indent + 1, p_class)
elif self.is_predefined_builtin(*mod_class_method_tuple):
spec, sig_note = restore_predefined_builtin(classname, p_name)
out(indent, "def ", spec, ": # ", sig_note)
out_doc_attr(out, p_func, indent + 1, p_class)
elif sys.platform == 'cli' and is_clr_type(p_class):
is_static, spec, sig_note = restore_clr(p_name, p_class)
if is_static:
out(indent, "@staticmethod")
if not spec: return
if sig_note:
out(indent, "def ", spec, ": #", sig_note)
else:
out(indent, "def ", spec, ":")
if not p_name in ['__gt__', '__ge__', '__lt__', '__le__', '__ne__', '__reduce_ex__', '__str__']:
out_doc_attr(out, p_func, indent + 1, p_class)
elif mod_class_method_tuple in PREDEFINED_MOD_CLASS_SIGS:
sig, ret_literal = PREDEFINED_MOD_CLASS_SIGS[mod_class_method_tuple]
if classname:
ofwhat = "%s.%s.%s" % mod_class_method_tuple
else:
ofwhat = "%s.%s" % (p_modname, p_name)
out(indent, "def ", p_name, sig, ": # known case of ", ofwhat)
out_doc_attr(out, p_func, indent + 1, p_class)
else:
# __doc__ is our best source of arglist
sig_note = "real signature unknown"
spec = ""
is_init = (p_name == "__init__" and p_class is not None)
funcdoc = None
if is_init and hasattr(p_class, "__doc__"):
if hasattr(p_func, "__doc__"):
funcdoc = p_func.__doc__
if funcdoc == object.__init__.__doc__:
funcdoc = p_class.__doc__
elif hasattr(p_func, "__doc__"):
funcdoc = p_func.__doc__
sig_restored = False
action("parsing doc of func %r of class %r", p_name, p_class)
if isinstance(funcdoc, STR_TYPES):
(spec, ret_literal, more_notes) = self.parse_func_doc(funcdoc, p_name, p_name, classname, deco,
sip_generated)
if spec is None and p_name == '__init__' and classname:
(spec, ret_literal, more_notes) = self.parse_func_doc(funcdoc, classname, p_name, classname, deco,
sip_generated)
sig_restored = spec is not None
if more_notes:
if sig_note:
sig_note += "; "
sig_note += more_notes
if not sig_restored:
# use an allow-all declaration
decl = []
if p_class:
first_param = propose_first_param(deco)
if first_param:
decl.append(first_param)
decl.append("*args")
decl.append("**kwargs")
spec = p_name + "(" + ", ".join(decl) + ")"
out(indent, "def ", spec, ": # ", sig_note)
# to reduce size of stubs, don't output same docstring twice for class and its __init__ method
if not is_init or funcdoc != p_class.__doc__:
out_docstring(out, funcdoc, indent + 1)
# body
if ret_literal and not is_init:
out(indent + 1, "return ", ret_literal)
else:
out(indent + 1, "pass")
if deco and not HAS_DECORATORS:
out(indent, p_name, " = ", deco, "(", p_name, ")", deco_comment)
out(0, "") # empty line after each item
def redo_class(self, out, p_class, p_name, indent, p_modname=None, seen=None, inspect_dir=False):
"""
Restores a class definition.
@param out output function of a relevant buf
@param p_class the class object
@param p_name class name as known to owner
@param indent indentation level
@param p_modname name of module
@param seen {class: name} map of classes already seen in the same namespace
"""
action("redoing class %r of module %r", p_name, p_modname)
if seen is not None:
if p_class in seen:
out(indent, p_name, " = ", seen[p_class])
out(indent, "")
return
else:
seen[p_class] = p_name
bases = get_bases(p_class)
base_def = ""
skipped_bases = []
if bases:
skip_qualifiers = [p_modname, BUILTIN_MOD_NAME, 'exceptions']
skip_qualifiers.extend(KNOWN_FAKE_REEXPORTERS.get(p_modname, ()))
bases_list = [] # what we'll render in the class decl
for base in bases:
if [1 for (cls, mdl) in KNOWN_FAKE_BASES if cls == base and mdl != self.module]:
# our base is a wrapper and our module is not its defining module
skipped_bases.append(str(base))
continue
# somehow import every base class
base_name = base.__name__
qual_module_name = qualifier_of(base, skip_qualifiers)
got_existing_import = False
if qual_module_name:
if qual_module_name in self.used_imports:
import_list = self.used_imports[qual_module_name]
if base in import_list:
bases_list.append(base_name) # unqualified: already set to import
got_existing_import = True
if not got_existing_import:
mangled_qualifier = "__" + qual_module_name.replace('.', '_') # foo.bar -> __foo_bar
bases_list.append(mangled_qualifier + "." + base_name)
self.hidden_imports[qual_module_name] = mangled_qualifier
else:
bases_list.append(base_name)
base_def = "(" + ", ".join(bases_list) + ")"
if self.split_modules:
for base in bases_list:
local_import = self.create_local_import(base)
if local_import:
out(indent, local_import)
out(indent, "class ", p_name, base_def, ":",
skipped_bases and " # skipped bases: " + ", ".join(skipped_bases) or "")
out_doc_attr(out, p_class, indent + 1)
# inner parts
methods = {}
properties = {}
others = {}
we_are_the_base_class = p_modname == BUILTIN_MOD_NAME and p_name == "object"
field_source = {}
try:
if hasattr(p_class, "__dict__") and not inspect_dir:
field_source = p_class.__dict__
field_keys = field_source.keys() # Jython 2.5.1 _codecs fail here
else:
field_keys = dir(p_class) # this includes unwanted inherited methods, but no dict + inheritance is rare
except:
field_keys = ()
for item_name in field_keys:
item_qname = p_modname + '.' + p_name + '.' + item_name
if item_name in ("__doc__", "__module__"):
if we_are_the_base_class:
item = "" # must be declared in base types
else:
continue # in all other cases must be skipped
elif keyword.iskeyword(item_name): # for example, PyQt4 contains definitions of methods named 'exec'
continue
elif item_qname in CLASS_ATTR_BLACKLIST:
note('skipping blacklisted attribute ' + item_qname)
item = field_source.get(item_name)
else:
try:
item = getattr(p_class, item_name) # let getters do the magic
except AttributeError:
item = field_source.get(item_name) # have it raw
if item is None:
continue
except Exception:
continue
if is_callable(item) and not isinstance(item, type):
methods[item_name] = item
elif is_property(item):
properties[item_name] = item
else:
others[item_name] = item
#
if we_are_the_base_class:
others["__dict__"] = {} # force-feed it, for __dict__ does not contain a reference to itself :)
# add fake __init__s to have the right sig
if p_class in FAKE_BUILTIN_INITS:
methods["__init__"] = self.fake_builtin_init
note("Faking init of %s", p_name)
elif '__init__' not in methods:
init_method = getattr(p_class, '__init__', None)
if init_method:
methods['__init__'] = init_method
#
seen_funcs = {}
for item_name in sorted_no_case(methods.keys()):
item = methods[item_name]
try:
self.redo_function(out, item, item_name, indent + 1, p_class, p_modname, classname=p_name, seen=seen_funcs)
except:
handle_error_func(item_name, out)
#
known_props = KNOWN_PROPS.get(p_modname, {})
a_setter = "lambda self, v: None"
a_deleter = "lambda self: None"
for item_name in sorted_no_case(properties.keys()):
item = properties[item_name]
prop_docstring = getattr(item, '__doc__', None)
prop_key = (p_name, item_name)
if prop_key in known_props:
prop_descr = known_props.get(prop_key, None)
if prop_descr is None:
continue # explicitly omitted
acc_line, getter_and_type = prop_descr
if getter_and_type:
getter, prop_type = getter_and_type
else:
getter, prop_type = None, None
out(indent + 1, item_name,
" = property(", format_accessors(acc_line, getter, a_setter, a_deleter), ")"
)
if prop_type:
if prop_docstring:
out(indent + 1, '"""', prop_docstring)
out(0, "")
out(indent + 1, ':type: ', prop_type)
out(indent + 1, '"""')
else:
out(indent + 1, '""":type: ', prop_type, '"""')
out(0, "")
else:
out(indent + 1, item_name, " = property(lambda self: object(), lambda self, v: None, lambda self: None) # default")
if prop_docstring:
out(indent + 1, '"""', prop_docstring, '"""')
out(0, "")
if properties:
out(0, "") # empty line after the block
#
for item_name in sorted_no_case(others.keys()):
item = others[item_name]
self.fmt_value(out, item, indent + 1, prefix=item_name + " = ")
if p_name == "object":
out(indent + 1, "__module__ = ''")
if others:
out(0, "") # empty line after the block
#
if not methods and not properties and not others:
out(indent + 1, "pass")
def redo_simple_header(self, p_name):
"""Puts boilerplate code on the top"""
out = self.header_buf.out # 1st class methods rule :)
out(0, "# encoding: %s" % OUT_ENCODING) # line 1
# NOTE: maybe encoding should be selectable
if hasattr(self.module, "__name__"):
self_name = self.module.__name__
if self_name != p_name:
mod_name = " calls itself " + self_name
else:
mod_name = ""
else:
mod_name = " does not know its name"
out(0, "# module ", p_name, mod_name) # line 2
origin_type = OriginType.FILE
import generator3.core
if generator3.core.is_pregeneration_mode():
origin = origin_type = OriginType.PREGENERATED
elif self.mod_filename:
origin = self.mod_filename
elif p_name in sys.builtin_module_names:
origin = origin_type = OriginType.BUILTIN
else:
try:
origin = getattr(self.module, "__file__")
except AttributeError:
origin = origin_type = OriginType.BUILTIN
if self.test_mode and origin_type == OriginType.FILE:
origin = get_portable_test_module_path(origin, self.qname)
out(0, "# from %s" % origin) # line 3
out(0, "# by generator %s" % self.gen_version) # line 4
if p_name == BUILTIN_MOD_NAME and version[0] == 2 and version[1] >= 6:
out(0, "from __future__ import print_function")
out_doc_attr(out, self.module, 0)
def redo_imports(self):
module_type = type(sys)
for item_name in self.module.__dict__.keys():
try:
item = self.module.__dict__[item_name]
except:
continue
if type(item) is module_type: # not isinstance, py2.7 + PyQt4.QtCore on windows have a bug here
self.imported_modules[item_name] = item
self.add_import_header_if_needed()
ref_notice = getattr(item, "__file__", str(item)) if not self.test_mode else '<ref>'
if hasattr(item, "__name__"):
self.imports_buf.out(0, "import ", item.__name__, " as ", item_name, " # ", ref_notice)
else:
self.imports_buf.out(0, item_name, " = None # ??? name unknown; ", ref_notice)
def add_import_header_if_needed(self):
if self.imports_buf.isEmpty():
self.imports_buf.out(0, "")
self.imports_buf.out(0, "# imports")
def redo(self, p_name, inspect_dir):
"""
Restores module declarations.
Intended for built-in modules and thus does not handle import statements.
@param p_name name of module
"""
action("redoing header of module %r %r", p_name, str(self.module))
if "pyqt4" in p_name.lower(): # qt4 specific patch
self._initializeQApp4()
elif "pyqt5" in p_name.lower(): # qt5 specific patch
self._initializeQApp5()
self.redo_simple_header(p_name)
# find whatever other self.imported_modules the module knows; effectively these are imports
action("redoing imports of module %r %r", p_name, str(self.module))
try:
self.redo_imports()
except:
pass
action("redoing innards of module %r %r", p_name, str(self.module))
module_type = type(sys)
# group what we have into buckets
vars_simple = {}
vars_complex = {}
funcs = {}
classes = {}
module_dict = self.module.__dict__
if inspect_dir:
module_dict = dir(self.module)
for item_name in module_dict:
note("looking at %s", item_name)
# Python/C API can declare a symbol with an arbitrary name
if not is_identifier(item_name): # noqa
continue
if item_name in (
"__dict__", "__doc__", "__module__", "__file__", "__name__", "__builtins__", "__package__"):
continue # handled otherwise
if self.test_mode and item_name in ('__loader__', '__spec__', '__cached__'):
continue
try:
item = getattr(self.module, item_name) # let getters do the magic
except AttributeError:
if not item_name in self.module.__dict__: continue
item = self.module.__dict__[item_name] # have it raw
# check if it has percolated from an imported module
except NotImplementedError:
if not item_name in self.module.__dict__: continue
item = self.module.__dict__[item_name] # have it raw
# unless we're adamantly positive that the name was imported, we assume it is defined here
mod_name = None # module from which p_name might have been imported
# IronPython has non-trivial reexports in System module, but not in others:
skip_modname = sys.platform == "cli" and p_name != "System"
surely_not_imported_mods = KNOWN_FAKE_REEXPORTERS.get(p_name, ())
## can't figure weirdness in some modules, assume no reexports:
#skip_modname = skip_modname or p_name in self.KNOWN_FAKE_REEXPORTERS
if not skip_modname:
try:
mod_name = getattr(item, '__module__', None)
except:
pass
# we assume that module foo.bar never imports foo; foo may import foo.bar. (see pygame and pygame.rect)
maybe_import_mod_name = mod_name if isinstance(mod_name, type(p_name)) else ''
import_is_from_top = len(p_name) > len(maybe_import_mod_name) and p_name.startswith(maybe_import_mod_name)
note("mod_name = %s, prospective = %s, from top = %s", mod_name, maybe_import_mod_name, import_is_from_top)
want_to_import = False
if (mod_name
and mod_name != BUILTIN_MOD_NAME
and mod_name != p_name
and mod_name not in surely_not_imported_mods
and not import_is_from_top
):
# import looks valid, but maybe it's a .py file? we're certain not to import from .py
# e.g. this rules out _collections import collections and builtins import site.
try:
imported = __import__(mod_name) # ok to repeat, Python caches for us
if imported:
qualifiers = mod_name.split(".")[1:]
for qual in qualifiers:
imported = getattr(imported, qual, None)
if not imported:
break
imported_path = (getattr(imported, '__file__', False) or "").lower()
want_to_import = not (imported_path.endswith('.py') or imported_path.endswith('.pyc'))
imported_name = getattr(imported, "__name__", None)
if imported_name == p_name:
want_to_import = False
note("path of %r is %r, want? %s", mod_name, imported_path, want_to_import)
except ImportError:
want_to_import = False
# NOTE: if we fail to import, we define 'imported' names here lest we lose them at all
if want_to_import:
import_list = self.used_imports[mod_name]
if item_name not in import_list:
import_list.append(item_name)
if not want_to_import:
if isinstance(item, type) or type(item).__name__ == 'classobj':
classes[item_name] = item
elif is_callable(item): # some classes are callable, check them before functions
funcs[item_name] = item
elif isinstance(item, module_type):
continue # self.imported_modules handled above already
else:
if isinstance(item, SIMPLEST_TYPES):
vars_simple[item_name] = item
else:
vars_complex[item_name] = item
# sort and output every bucket
action("outputting innards of module %r %r", p_name, str(self.module))
#
omitted_names = OMIT_NAME_IN_MODULE.get(p_name, [])
if vars_simple:
out = self.functions_buf.out
prefix = "" # try to group variables by common prefix
PREFIX_LEN = 2 # default prefix length if we can't guess better
out(0, "# Variables with simple values")
for item_name in sorted_no_case(vars_simple.keys()):
if item_name in omitted_names:
out(0, "# definition of " + item_name + " omitted")
continue
item = vars_simple[item_name]
# track the prefix
if len(item_name) >= PREFIX_LEN:
prefix_pos = string.rfind(item_name, "_") # most prefixes end in an underscore
if prefix_pos < 1:
prefix_pos = PREFIX_LEN
beg = item_name[0:prefix_pos]
if prefix != beg:
out(0, "") # space out from other prefix
prefix = beg
else:
prefix = ""
# output
replacement = REPLACE_MODULE_VALUES.get((p_name, item_name), None)
if replacement is not None:
out(0, item_name, " = ", replacement, " # real value of type ", str(type(item)), " replaced")
elif is_skipped_in_module(p_name, item_name):
t_item = type(item)
out(0, item_name, " = ", self.invent_initializer(t_item), " # real value of type ", str(t_item),
" skipped")
else:
self.fmt_value(out, item, 0, prefix=item_name + " = ")
self._defined[item_name] = True
out(0, "") # empty line after vars
#
if funcs:
out = self.functions_buf.out
out(0, "# functions")
out(0, "")
seen_funcs = {}
for item_name in sorted_no_case(funcs.keys()):
if item_name in omitted_names:
out(0, "# definition of ", item_name, " omitted")
continue
item = funcs[item_name]
try:
self.redo_function(out, item, item_name, 0, p_modname=p_name, seen=seen_funcs)
except:
handle_error_func(item_name, out)
else:
self.functions_buf.out(0, "# no functions")
#
if classes:
self.classes_buf.out(0, "# classes")
self.classes_buf.out(0, "")
seen_classes = {}
# sort classes so that inheritance order is preserved
cls_list = [] # items are (class_name, mro_tuple)
for cls_name in sorted_no_case(classes.keys()):
cls = classes[cls_name]
ins_index = len(cls_list)
for i in range(ins_index):
maybe_child_bases = cls_list[i][1]
if cls in maybe_child_bases:
ins_index = i # we could not go farther than current ins_index
break # ...and need not go fartehr than first known child
cls_list.insert(ins_index, (cls_name, get_mro(cls)))
self.split_modules = self.mod_filename and len(cls_list) >= 30
for item_name in [cls_item[0] for cls_item in cls_list]:
buf = ClassBuf(item_name, self)
self.classes_buffs.append(buf)
out = buf.out
if item_name in omitted_names:
out(0, "# definition of ", item_name, " omitted")
continue
item = classes[item_name]
self.redo_class(out, item, item_name, 0, p_modname=p_name, seen=seen_classes, inspect_dir=inspect_dir)
self._defined[item_name] = True
out(0, "") # empty line after each item
if self.doing_builtins and p_name == BUILTIN_MOD_NAME and version[0] < 3:
# classobj still supported
txt = classobj_txt
self.classes_buf.out(0, txt)
if self.doing_builtins and p_name == BUILTIN_MOD_NAME:
txt = create_generator()
self.classes_buf.out(0, txt)
txt = create_async_generator()
self.classes_buf.out(0, txt)
txt = create_function()
self.classes_buf.out(0, txt)
txt = create_method()
self.classes_buf.out(0, txt)
txt = create_coroutine()
self.classes_buf.out(0, txt)
# Fake <type 'namedtuple'>
if version[0] >= 3 or (version[0] == 2 and version[1] >= 6):
namedtuple_text = create_named_tuple()
self.classes_buf.out(0, namedtuple_text)
else:
self.classes_buf.out(0, "# no classes")
#
if vars_complex:
out = self.footer_buf.out
out(0, "# variables with complex values")
out(0, "")
for item_name in sorted_no_case(vars_complex.keys()):
if item_name in omitted_names:
out(0, "# definition of " + item_name + " omitted")
continue
item = vars_complex[item_name]
if str(type(item)) == "<type 'namespace#'>":
continue # this is an IronPython submodule, we mustn't generate a reference for it in the base module
replacement = REPLACE_MODULE_VALUES.get((p_name, item_name), None)
if replacement is not None:
out(0, item_name + " = " + replacement + " # real value of type " + str(type(item)) + " replaced")
elif is_skipped_in_module(p_name, item_name):
t_item = type(item)
out(0, item_name + " = " + self.invent_initializer(t_item) + " # real value of type " + str(
t_item) + " skipped")
else:
self.fmt_value(out, item, 0, prefix=item_name + " = ", as_name=item_name)
self._defined[item_name] = True
out(0, "") # empty line after each item
values_to_add = ADD_VALUE_IN_MODULE.get(p_name, None)
if values_to_add:
self.footer_buf.out(0, "# intermittent names")
for value in values_to_add:
self.footer_buf.out(0, value)
# imports: last, because previous parts could alter used_imports or hidden_imports
self.output_import_froms()
if self.imports_buf.isEmpty():
self.imports_buf.out(0, "# no imports")
self.imports_buf.out(0, "") # empty line after imports
def output_import_froms(self):
"""Mention all imported names known within the module, wrapping as per PEP."""
out = self.imports_buf.out
if self.used_imports:
self.add_import_header_if_needed()
for mod_name in sorted_no_case(self.used_imports.keys()):
import_names = self.used_imports[mod_name]
if import_names:
self._defined[mod_name] = True
right_pos = 0 # tracks width of list to fold it at right margin
import_heading = "from % s import (" % mod_name
right_pos += len(import_heading)
names_pack = [import_heading]
indent_level = 0
import_names = list(import_names)
import_names.sort()
for n in import_names:
self._defined[n] = True
len_n = len(n)
if right_pos + len_n >= 78:
out(indent_level, *names_pack)
names_pack = [n, ", "]
if indent_level == 0:
indent_level = 1 # all but first line is indented
right_pos = self.indent_size + len_n + 2
else:
names_pack.append(n)
names_pack.append(", ")
right_pos += (len_n + 2)
# last line is...
if indent_level == 0: # one line
names_pack[0] = names_pack[0][:-1] # cut off lpar
names_pack[-1] = "" # cut last comma
else: # last line of multiline
names_pack[-1] = ")" # last comma -> rpar
out(indent_level, *names_pack)
out(0, "") # empty line after group
if self.hidden_imports:
self.add_import_header_if_needed()
for mod_name in sorted_no_case(self.hidden_imports.keys()):
out(0, 'import ', mod_name, ' as ', self.hidden_imports[mod_name])
out(0, "") # empty line after group