def indent(code, spaces=4):
'''indent a block of code with whitespace (default is 4 spaces)'''
indent = indentsize(code)
if type(spaces) is int: spaces = ' '*spaces
# if '\t' is provided, will indent with a tab
nspaces = indentsize(spaces)
# blank lines (etc) need to be ignored
lines = code.split('\n')
## stq = "'''"; dtq = '"""'
## in_stq = in_dtq = False
for i in range(len(lines)):
#FIXME: works... but shouldn't indent 2nd+ lines of multiline doc
_indent = indentsize(lines[i])
if indent > _indent: continue
lines[i] = spaces+lines[i]
## #FIXME: may fail when stq and dtq in same line (depends on ordering)
## nstq, ndtq = lines[i].count(stq), lines[i].count(dtq)
## if not in_dtq and not in_stq:
## lines[i] = spaces+lines[i] # we indent
## # entering a comment block
## if nstq%2: in_stq = not in_stq
## if ndtq%2: in_dtq = not in_dtq
## # leaving a comment block
## elif in_dtq and ndtq%2: in_dtq = not in_dtq
## elif in_stq and nstq%2: in_stq = not in_stq
## else: pass
if lines[-1].strip() == '': lines[-1] = ''
return '\n'.join(lines)
def _outdent(lines, spaces=None, all=True):
'''outdent lines of code, accounting for docs and line continuations'''
indent = indentsize(lines[0])
if spaces is None or spaces > indent or spaces < 0: spaces = indent
for i in range(len(lines) if all else 1):
#FIXME: works... but shouldn't outdent 2nd+ lines of multiline doc
_indent = indentsize(lines[i])
if spaces > _indent: _spaces = _indent
else: _spaces = spaces
lines[i] = lines[i][_spaces:]
return lines
def outdent(code, spaces=None, all=True):
'''outdent a block of code (default is to strip all leading whitespace)'''
indent = indentsize(code)
if spaces is None or spaces > indent or spaces < 0: spaces = indent
#XXX: will this delete '\n' in some cases?
if not all: return code[spaces:]
return '\n'.join(_outdent(code.split('\n'), spaces=spaces, all=all))
def getcomments(pyObject):
"""Get lines of comments immediately preceding an object's source code.
Returns None when source can't be found.
"""
try:
lines, lnum = findsource(pyObject)
except (IOError, TypeError):
return None
if ismodule(pyObject):
# Look for a comment block at the top of the file.
start = 0
if lines and lines[0][:2] == '#!': start = 1
while start < len(lines) and string.strip(lines[start]) in ('', '#'):
start = start + 1
if start < len(lines) and lines[start][:1] == '#':
comments = []
end = start
while end < len(lines) and lines[end][:1] == '#':
comments.append(string.expandtabs(lines[end]))
end = end + 1
return string.join(comments, '')
# Look for a preceding block of comments at the same indentation.
elif lnum > 0:
indent = indentsize(lines[lnum])
end = lnum - 1
if end >= 0 and string.lstrip(lines[end])[:1] == '#' and \
indentsize(lines[end]) == indent:
comments = [string.lstrip(string.expandtabs(lines[end]))]
if end > 0:
end = end - 1
comment = string.lstrip(string.expandtabs(lines[end]))
while comment[:1] == '#' and indentsize(lines[end]) == indent:
comments[:0] = [comment]
end = end - 1
if end < 0: break
comment = string.lstrip(string.expandtabs(lines[end]))
while comments and string.strip(comments[0]) == '#':
comments[:1] = []
while comments and string.strip(comments[-1]) == '#':
comments[-1:] = []
return string.join(comments, '')
def _matchlambda(func, line):
"""check if lambda object 'func' matches raw line of code 'line'"""
from .detect import code as getcode
from .detect import freevars, globalvars, varnames
dummy = lambda: '__this_is_a_big_dummy_function__'
# process the line (removing leading whitespace, etc)
lhs, rhs = line.split('lambda ', 1)[-1].split(":", 1) #FIXME: if !1 inputs
try: #FIXME: unsafe
_ = eval("lambda %s : %s" % (lhs, rhs), globals(), locals())
except:
_ = dummy
# get code objects, for comparison
_, code = getcode(_).co_code, getcode(func).co_code
# check if func is in closure
_f = [line.count(i) for i in freevars(func).keys()]
if not _f: # not in closure
# check if code matches
if _ == code: return True
return False
# weak check on freevars
if not all(_f): return False #XXX: VERY WEAK
# weak check on varnames and globalvars
_f = varnames(func)
_f = [line.count(i) for i in _f[0] + _f[1]]
if _f and not all(_f): return False #XXX: VERY WEAK
_f = [line.count(i) for i in globalvars(func).keys()]
if _f and not all(_f): return False #XXX: VERY WEAK
# check if func is a double lambda
if (line.count('lambda ') > 1) and (lhs in freevars(func).keys()):
_lhs, _rhs = rhs.split('lambda ', 1)[-1].split(":",
1) #FIXME: if !1 inputs
try: #FIXME: unsafe
_f = eval("lambda %s : %s" % (_lhs, _rhs), globals(), locals())
except:
_f = dummy
# get code objects, for comparison
_, code = getcode(_f).co_code, getcode(func).co_code
if len(_) != len(code): return False
#NOTE: should be same code same order, but except for 't' and '\x88'
_ = set((i, j) for (i, j) in zip(_, code) if i != j)
if len(_) != 1: return False #('t','\x88')
return True
# check indentsize
if not indentsize(line): return False #FIXME: is this a good check???
# check if code 'pattern' matches
#XXX: or pattern match against dis.dis(code)? (or use uncompyle2?)
_ = _.split(_[0]) # 't' #XXX: remove matching values if starts the same?
_f = code.split(code[0]) # '\x88'
#NOTE: should be same code different order, with different first element
_ = dict(
re.match('([\W\D\S])(.*)', _[i]).groups() for i in range(1, len(_)))
_f = dict(
re.match('([\W\D\S])(.*)', _f[i]).groups() for i in range(1, len(_f)))
if (_.keys() == _f.keys()) and (sorted(_.values()) == sorted(_f.values())):
return True
return False
def get_internal_comments(object):
lines, lnum = inspect.findsource(object)
if len(lines) <= lnum + 1:
# object is probably an emply module.
return None
indent = inspect.indentsize(lines[lnum+1])
comments = []
for line in lines[lnum + 1:]:
comment = line.strip()
# A line is a comment if it matches the indentation the first line and
# begins with a '#'
if inspect.indentsize(line) == indent and comment.startswith('#'):
comments.append(comment)
else:
break
if len(comments) == 0:
return None
else:
return '\n'.join(comments)
def get_annotation_comment_and_line_by_source(container_object, variable_name, source_file):
try:
lines, lnum = get_source(container_object, source_file)
except (OSError, TypeError):
return None
i = 1
while indentsize(lines[lnum + i]) == indentsize(lines[lnum]):
i += 1
if len(lines) > lnum and indentsize(lines[lnum + i]) > indentsize(lines[lnum]):
this_cls_indentsize = indentsize(lines[lnum + i])
l = lnum + i
while l < len(lines):
cur_indentsize = indentsize(lines[l])
if cur_indentsize < this_cls_indentsize and lines[l].strip() != "":
break
if cur_indentsize == this_cls_indentsize:
expr = lines[l].strip().replace(" ", "")
if expr.startswith(variable_name + ":"):
lnum = l
break
l += 1
if lnum > 0:
indent = indentsize(lines[lnum])
end = lnum - 1
if end >= 0 and lines[end].lstrip()[:1] == '#' and \
indentsize(lines[end]) == indent:
comments = [lines[end].expandtabs().lstrip()]
if end > 0:
end = end - 1
comment = lines[end].expandtabs().lstrip()
while comment[:1] == '#' and indentsize(lines[end]) == indent:
comments[:0] = [comment]
end = end - 1
if end < 0: break
comment = lines[end].expandtabs().lstrip()
while comments and comments[0].strip() == '#':
comments[:1] = []
while comments and comments[-1].strip() == '#':
comments[-1:] = []
return '\n'.join(comments), lnum
return None, lnum
def _find_line_info_of_attr_in_source(
frame: Optional[FrameType], key: str,
attr: "AbstractComplexExtractor") -> _LineInfo:
if frame is None:
return _LineInfo(None, None, None, f"{key}={attr!r}")
file = frame.f_code.co_filename
firstlineno = frame.f_lineno
firstline_idx = firstlineno - 1
try:
lines, _ = inspect.findsource(frame)
except OSError:
# can't get the source code from python repl
return _LineInfo(None, None, None, f"{key}={attr!r}")
start_index = inspect.indentsize(lines[firstline_idx])
for lineno, line in enumerate(lines[firstline_idx + 1:],
start=firstlineno + 1):
# iterate line in the code block body
cur_index = inspect.indentsize(line)
if cur_index <= start_index:
# reach end of the code block,
# use code block firstlineno as SyntaxError.lineno
line = lines[firstline_idx]
lineno = firstlineno
break
if line.lstrip().startswith(key):
# find the line as SyntaxError.text
break
else:
# reach EOF,
# use code block firstlineno as SyntaxError.lineno
line = lines[firstline_idx]
lineno = firstlineno
offset = inspect.indentsize(line)
line = line.strip()
return _LineInfo(file, lineno, offset, line)
def _matchlambda(func, line):
"""check if lambda object 'func' matches raw line of code 'line'"""
from .detect import code as getcode
from .detect import freevars, globalvars, varnames
dummy = lambda : '__this_is_a_big_dummy_function__'
# process the line (removing leading whitespace, etc)
lhs,rhs = line.split('lambda ',1)[-1].split(":", 1) #FIXME: if !1 inputs
try: #FIXME: unsafe
_ = eval("lambda %s : %s" % (lhs,rhs), globals(),locals())
except: _ = dummy
# get code objects, for comparison
_, code = getcode(_).co_code, getcode(func).co_code
# check if func is in closure
_f = [line.count(i) for i in freevars(func).keys()]
if not _f: # not in closure
# check if code matches
if _ == code: return True
return False
# weak check on freevars
if not all(_f): return False #XXX: VERY WEAK
# weak check on varnames and globalvars
_f = varnames(func)
_f = [line.count(i) for i in _f[0]+_f[1]]
if _f and not all(_f): return False #XXX: VERY WEAK
_f = [line.count(i) for i in globalvars(func).keys()]
if _f and not all(_f): return False #XXX: VERY WEAK
# check if func is a double lambda
if (line.count('lambda ') > 1) and (lhs in freevars(func).keys()):
_lhs,_rhs = rhs.split('lambda ',1)[-1].split(":",1) #FIXME: if !1 inputs
try: #FIXME: unsafe
_f = eval("lambda %s : %s" % (_lhs,_rhs), globals(),locals())
except: _f = dummy
# get code objects, for comparison
_, code = getcode(_f).co_code, getcode(func).co_code
if len(_) != len(code): return False
#NOTE: should be same code same order, but except for 't' and '\x88'
_ = set((i,j) for (i,j) in zip(_,code) if i != j)
if len(_) != 1: return False #('t','\x88')
return True
# check indentsize
if not indentsize(line): return False #FIXME: is this a good check???
# check if code 'pattern' matches
#XXX: or pattern match against dis.dis(code)? (or use uncompyle2?)
_ = _.split(_[0]) # 't' #XXX: remove matching values if starts the same?
_f = code.split(code[0]) # '\x88'
#NOTE: should be same code different order, with different first element
_ = dict(re.match(r'([\W\D\S])(.*)', _[i]).groups() for i in range(1,len(_)))
_f = dict(re.match(r'([\W\D\S])(.*)', _f[i]).groups() for i in range(1,len(_f)))
if (_.keys() == _f.keys()) and (sorted(_.values()) == sorted(_f.values())):
return True
return False
def get_func_source_code(func):
if isinstance(func, types.MethodType):
if sys.version_info[0] == 3:
src_code = inspect.getsource(func.__self__.__class__)
else:
im_class = get_method_defined_class(func)
src_code = inspect.getsource(im_class)
else:
src_code = inspect.getsource(func)
indent = inspect.indentsize(src_code)
if indent > 0:
lines = src_code.split("\n")
src_code = ""
for line in lines:
src_code += line[indent:] + "\n"
return src_code
def __init__(self, o):
"""Transcode a Python function or module
In:
- ``o`` -- Python function or module to transcode
"""
if hasattr(o, '_js_name'):
# Already transcoded
self.name = o._js_name
self.javascript = o._js_code
return
src = inspect.getsource(o)
classname = ''
if isinstance(o, types.MethodType):
classname = o.im_class.__name__
o = o.im_func
if isinstance(o, types.FunctionType):
if o.__module__ == '__main__':
self.name = module = classname
else:
module = o.__module__.replace('.', '_')
if classname:
module += ('_' + classname)
self.name = module + '_'
self.name += o.func_name
indent = inspect.indentsize(src)
src = ''.join([s.expandtabs()[indent:] for s in src.splitlines(True)])
else:
if o.__name__ == '__main__':
self.name = module = ''
else:
self.name = module = o.__name__.replace('.', '_')
self.javascript = str2js(src, module)
# Keep the transcoded javascript
o._js_name = self.name
o._js_code = self.javascript
def __init__(self, o):
"""Transcode a Python function or module
In:
- ``o`` -- Python function or module to transcode
"""
if hasattr(o, '_js_name'):
# Already transcoded
self.name = o._js_name
self.javascript = o._js_code
return
src = inspect.getsource(o)
classname = ''
if isinstance(o, types.MethodType):
classname = o.im_class.__name__
o = o.im_func
if isinstance(o, types.FunctionType):
if o.__module__ == '__main__':
self.name = module = classname
else:
module = o.__module__.replace('.', '_')
if classname:
module += ('_' + classname)
self.name = module + '_'
self.name += o.func_name
indent = inspect.indentsize(src)
src = ''.join([s.expandtabs()[indent:] for s in src.splitlines(True)])
else:
if o.__name__ == '__main__':
self.name = module = ''
else:
self.name = module = o.__name__.replace('.', '_')
self.javascript = str2js(src, module)
# Keep the transcoded javascript
o._js_name = self.name
o._js_code = self.javascript
def getblocks(object, lstrip=False, enclosing=False, locate=False):
"""Return a list of source lines and starting line number for an object.
Interactively-defined objects refer to lines in the interpreter's history.
If enclosing=True, then also return any enclosing code.
If lstrip=True, ensure there is no indentation in the first line of code.
If locate=True, then also return the line number for the block of code.
DEPRECATED: use 'getsourcelines' instead
"""
lines, lnum = findsource(object)
if ismodule(object):
if lstrip: lines = _outdent(lines)
return ([lines], [0]) if locate is True else [lines]
#XXX: 'enclosing' means: closures only? or classes and files?
indent = indentsize(lines[lnum])
block = getblock(lines[lnum:]) #XXX: catch any TokenError here?
if not enclosing or not indent:
if lstrip: block = _outdent(block)
return ([block], [lnum]) if locate is True else [block]
pat1 = r'^(\s*def\s)|(.*(?<!\w)lambda(:|\s))'; pat1 = re.compile(pat1)
pat2 = r'^(\s*@)'; pat2 = re.compile(pat2)
#pat3 = r'^(\s*class\s)'; pat3 = re.compile(pat3) #XXX: enclosing class?
#FIXME: bound methods need enclosing class (and then instantiation)
# *or* somehow apply a partial using the instance
skip = 0
line = 0
blocks = []; _lnum = []
target = ''.join(block)
while line <= lnum: #XXX: repeat lnum? or until line < lnum?
# see if starts with ('def','lambda') and contains our target block
if pat1.match(lines[line]):
if not skip:
try: code = getblock(lines[line:])
except TokenError: code = [lines[line]]
if indentsize(lines[line]) > indent: #XXX: should be >= ?
line += len(code) - skip
elif target in ''.join(code):
blocks.append(code) # save code block as the potential winner
_lnum.append(line - skip) # save the line number for the match
line += len(code) - skip
else:
line += 1
skip = 0
# find skip: the number of consecutive decorators
elif pat2.match(lines[line]):
try: code = getblock(lines[line:])
except TokenError: code = [lines[line]]
skip = 1
for _line in code[1:]: # skip lines that are decorators
if not pat2.match(_line): break
skip += 1
line += skip
# no match: reset skip and go to the next line
else:
line +=1
skip = 0
if not blocks:
blocks = [block]
_lnum = [lnum]
if lstrip: blocks = [_outdent(block) for block in blocks]
# return last match
return (blocks, _lnum) if locate is True else blocks
def getblocks(object, lstrip=False, enclosing=False, locate=False):
"""Return a list of source lines and starting line number for an object.
Interactively-defined objects refer to lines in the interpreter's history.
If enclosing=True, then also return any enclosing code.
If lstrip=True, ensure there is no indentation in the first line of code.
If locate=True, then also return the line number for the block of code.
DEPRECATED: use 'getsourcelines' instead
"""
lines, lnum = findsource(object)
if ismodule(object):
if lstrip: lines = _outdent(lines)
return ([lines], [0]) if locate is True else [lines]
#XXX: 'enclosing' means: closures only? or classes and files?
indent = indentsize(lines[lnum])
block = getblock(lines[lnum:]) #XXX: catch any TokenError here?
if not enclosing or not indent:
if lstrip: block = _outdent(block)
return ([block], [lnum]) if locate is True else [block]
pat1 = r'^(\s*def\s)|(.*(?<!\w)lambda(:|\s))'
pat1 = re.compile(pat1)
pat2 = r'^(\s*@)'
pat2 = re.compile(pat2)
#pat3 = r'^(\s*class\s)'; pat3 = re.compile(pat3) #XXX: enclosing class?
#FIXME: bound methods need enclosing class (and then instantiation)
# *or* somehow apply a partial using the instance
skip = 0
line = 0
blocks = []
_lnum = []
target = ''.join(block)
while line <= lnum: #XXX: repeat lnum? or until line < lnum?
# see if starts with ('def','lambda') and contains our target block
if pat1.match(lines[line]):
if not skip:
try:
code = getblock(lines[line:])
except TokenError:
code = [lines[line]]
if indentsize(lines[line]) > indent: #XXX: should be >= ?
line += len(code) - skip
elif target in ''.join(code):
blocks.append(code) # save code block as the potential winner
_lnum.append(line - skip) # save the line number for the match
line += len(code) - skip
else:
line += 1
skip = 0
# find skip: the number of consecutive decorators
elif pat2.match(lines[line]):
try:
code = getblock(lines[line:])
except TokenError:
code = [lines[line]]
skip = 1
for _line in code[1:]: # skip lines that are decorators
if not pat2.match(_line): break
skip += 1
line += skip
# no match: reset skip and go to the next line
else:
line += 1
skip = 0
if not blocks:
blocks = [block]
_lnum = [lnum]
if lstrip: blocks = [_outdent(block) for block in blocks]
# return last match
return (blocks, _lnum) if locate is True else blocks
def __init__(cls, name: str, bases: Tuple[type], attr_dict: Dict[str,
Any]):
super().__init__(name, bases, attr_dict)
field_names: List[str] = []
__init_args = inspect.getfullargspec(getattr(cls, "__init__")).args
for key, attr in attr_dict.items():
if isinstance(type(attr), ComplexExtractorMeta):
if key in __init_args[1:]:
# Item's attribute overwrites the 'Item.__init__' parameters except first parameter.
args = [] # type: List[Any]
exc_args = None
frame = inspect.currentframe()
assert (
frame is not None
), "If running in an implementation without Python stack frame support this function returns None."
try:
outer_frame = frame.f_back
filename = outer_frame.f_code.co_filename
firstlineno = outer_frame.f_lineno
firstline_idx = firstlineno - 1
lines = None
try:
lines, _ = inspect.findsource(outer_frame)
except OSError:
# can't get the source code from python repl
pass
if lines is not None:
start_index = inspect.indentsize(
lines[firstline_idx])
for lineno, line in enumerate(
lines[firstline_idx + 1:],
start=firstlineno + 1):
# iterate line in the code block body
cur_index = inspect.indentsize(line)
if cur_index <= start_index:
# reach end of the code block, use code block firstlineno as SyntaxError.lineno
line = lines[firstline_idx]
lineno = firstlineno
break
if line.lstrip().startswith(key):
# find the line as SyntaxError.text
break
else:
# reach EOF, use code block firstlineno as SyntaxError.lineno
line = lines[firstline_idx]
lineno = firstlineno
offset = inspect.indentsize(line)
line = line.strip()
exc_args = (filename, lineno, offset, line)
else:
line = f"{key}={attr!r}"
finally:
del outer_frame
del frame
args.append(
f"{line!r} overwriten the parameter {key!r} of '{name}.__init__' method. "
f"Please using the optional parameter name={key!r} in {attr!r} to avoid overwriting parameter name."
)
if exc_args is not None:
args.append(exc_args)
raise SyntaxError(*args)
field_names.append(key)
cls._field_names = field_names
def _func_to_code(self, func):
lines = inspect.getsourcelines(func)[0]
indent = inspect.indentsize(lines[0])
code = ''.join([ i[indent:] for i in lines ])
return code
