def skipCodeBlock (self,s,i,kind):
trace = False ; verbose = True
# if trace: g.trace('***',g.callers())
startIndent = self.startSigIndent
if trace: g.trace('startIndent',startIndent)
assert startIndent is not None
i = start = g.skip_ws_and_nl(s,i)
parenCount = 0
underIndentedStart = None # The start of trailing underindented blank or comment lines.
while i < len(s):
progress = i
ch = s[i]
if g.is_nl(s,i):
if trace and verbose: g.trace(g.get_line(s,i))
backslashNewline = (i > 0 and g.match(s,i-1,'\\\n'))
if backslashNewline:
# An underindented line, including docstring,
# does not end the code block.
i += 1 # 2010/11/01
else:
i = g.skip_nl(s,i)
j = g.skip_ws(s,i)
if g.is_nl(s,j):
pass # We have already made progress.
else:
i,underIndentedStart,breakFlag = self.pythonNewlineHelper(
s,i,parenCount,startIndent,underIndentedStart)
if breakFlag: break
elif ch == '#':
i = g.skip_to_end_of_line(s,i)
elif ch == '"' or ch == '\'':
i = g.skip_python_string(s,i)
elif ch in '[{(':
i += 1 ; parenCount += 1
# g.trace('ch',ch,parenCount)
elif ch in ']})':
i += 1 ; parenCount -= 1
# g.trace('ch',ch,parenCount)
else: i += 1
assert(progress < i)
# The actual end of the block.
if underIndentedStart is not None:
i = underIndentedStart
if trace: g.trace('***backtracking to underindent range')
if trace: g.trace(g.get_line(s,i))
if 0 < i < len(s) and not g.match(s,i-1,'\n'):
g.trace('Can not happen: Python block does not end in a newline.')
g.trace(g.get_line(s,i))
return i,False
# 2010/02/19: Include all following material
# until the next 'def' or 'class'
i = self.skipToTheNextClassOrFunction(s,i,startIndent)
if (trace or self.trace) and s[start:i].strip():
g.trace('%s returns\n' % (kind) + s[start:i])
return i,True
def skipCodeBlock(self, s, i, kind):
trace = False
verbose = True
# if trace: g.trace('***',g.callers())
startIndent = self.startSigIndent
if trace: g.trace('startIndent', startIndent)
assert startIndent is not None
i = start = g.skip_ws_and_nl(s, i)
parenCount = 0
underIndentedStart = None # The start of trailing underindented blank or comment lines.
while i < len(s):
progress = i
ch = s[i]
if g.is_nl(s, i):
if trace and verbose: g.trace(g.get_line(s, i))
backslashNewline = (i > 0 and g.match(s, i - 1, '\\\n'))
if backslashNewline:
# An underindented line, including docstring,
# does not end the code block.
i += 1 # 2010/11/01
else:
i = g.skip_nl(s, i)
j = g.skip_ws(s, i)
if g.is_nl(s, j):
pass # We have already made progress.
else:
i, underIndentedStart, breakFlag = self.pythonNewlineHelper(
s, i, parenCount, startIndent, underIndentedStart)
if breakFlag: break
elif ch == '#':
i = g.skip_to_end_of_line(s, i)
elif ch == '"' or ch == '\'':
i = g.skip_python_string(s, i)
elif ch in '[{(':
i += 1
parenCount += 1
# g.trace('ch',ch,parenCount)
elif ch in ']})':
i += 1
parenCount -= 1
# g.trace('ch',ch,parenCount)
else:
i += 1
assert (progress < i)
# The actual end of the block.
if underIndentedStart is not None:
i = underIndentedStart
if trace: g.trace('***backtracking to underindent range')
if trace: g.trace(g.get_line(s, i))
if 0 < i < len(s) and not g.match(s, i - 1, '\n'):
g.trace('Can not happen: Python block does not end in a newline.')
g.trace(g.get_line(s, i))
return i, False
# 2010/02/19: Include all following material
# until the next 'def' or 'class'
i = self.skipToTheNextClassOrFunction(s, i, startIndent)
if (trace or self.trace) and s[start:i].strip():
g.trace('%s returns\n' % (kind) + s[start:i])
return i, True
def pythonNewlineHelper(self, s, i, parenCount, startIndent,
underIndentedStart):
trace = False
breakFlag = False
j, indent = g.skip_leading_ws_with_indent(s, i, self.tab_width)
if trace:
g.trace('startIndent', startIndent, 'indent', indent, 'parenCount',
parenCount, 'line', repr(g.get_line(s, j)))
if indent <= startIndent and parenCount == 0:
# An underindented line: it ends the block *unless*
# it is a blank or comment line or (2008/9/1) the end of a triple-quoted string.
if g.match(s, j, '#'):
if trace: g.trace('underindent: comment')
if underIndentedStart is None: underIndentedStart = i
i = j
elif g.match(s, j, '\n'):
if trace: g.trace('underindent: blank line')
# Blank lines never start the range of underindented lines.
i = j
else:
if trace: g.trace('underindent: end of block')
breakFlag = True # The actual end of the block.
else:
if underIndentedStart and g.match(s, j, '\n'):
# Add the blank line to the underindented range.
if trace:
g.trace(
'properly indented blank line extends underindent range'
)
elif underIndentedStart and g.match(s, j, '#'):
# Add the (properly indented!) comment line to the underindented range.
if trace:
g.trace(
'properly indented comment line extends underindent range'
)
elif underIndentedStart is None:
pass
else:
# A properly indented non-comment line.
# Give a message for all underindented comments in underindented range.
if trace:
g.trace(
'properly indented line generates underindent errors')
s2 = s[underIndentedStart:i]
lines = g.splitlines(s2)
for line in lines:
if line.strip():
junk, indent = g.skip_leading_ws_with_indent(
line, 0, self.tab_width)
if indent <= startIndent:
if j not in self.errorLines: # No error yet given.
self.errorLines.append(j)
self.underindentedComment(line)
underIndentedStart = None
if trace:
g.trace('breakFlag', breakFlag, 'returns', i, 'underIndentedStart',
underIndentedStart)
return i, underIndentedStart, breakFlag
def skipInterface(self, s, i):
"""Skip from the opening delim to *past* the matching closing delim.
If no matching is found i is set to len(s)"""
trace = False
start = i
delim2 = "end."
level = 0
start = i
startIndent = self.startSigIndent
if trace:
g.trace("***", "startIndent", startIndent, g.callers())
while i < len(s):
progress = i
if g.is_nl(s, i):
backslashNewline = i > 0 and g.match(s, i - 1, "\\\n")
i = g.skip_nl(s, i)
if not backslashNewline and not g.is_nl(s, i):
j, indent = g.skip_leading_ws_with_indent(s, i, self.tab_width)
line = g.get_line(s, j)
if trace:
g.trace("indent", indent, line)
if indent < startIndent and line.strip():
# An non-empty underindented line.
# Issue an error unless it contains just the closing bracket.
if level == 1 and g.match(s, j, delim2):
pass
else:
if j not in self.errorLines: # No error yet given.
self.errorLines.append(j)
self.underindentedLine(line)
elif s[i] in (" ", "\t"):
i += 1 # speed up the scan.
elif self.startsComment(s, i):
i = self.skipComment(s, i)
elif self.startsString(s, i):
i = self.skipString(s, i)
elif g.match(s, i, delim2):
i += len(delim2)
if trace:
g.trace("returns\n", repr(s[start:i]))
return i
else:
i += 1
assert progress < i
self.error("no interface")
if 1:
g.pr("** no interface **")
i, j = g.getLine(s, start)
g.trace(i, s[i:j])
else:
if trace:
g.trace("** no interface")
return start
def skipInterface(self, s, i):
'''Skip from the opening delim to *past* the matching closing delim.
If no matching is found i is set to len(s)'''
trace = False
start = i
delim2 = 'end.'
level = 0
start = i
startIndent = self.startSigIndent
if trace: g.trace('***', 'startIndent', startIndent, g.callers())
while i < len(s):
progress = i
if g.is_nl(s, i):
backslashNewline = i > 0 and g.match(s, i - 1, '\\\n')
i = g.skip_nl(s, i)
if not backslashNewline and not g.is_nl(s, i):
j, indent = g.skip_leading_ws_with_indent(
s, i, self.tab_width)
line = g.get_line(s, j)
if trace: g.trace('indent', indent, line)
if indent < startIndent and line.strip():
# An non-empty underindented line.
# Issue an error unless it contains just the closing bracket.
if level == 1 and g.match(s, j, delim2):
pass
else:
if j not in self.errorLines: # No error yet given.
self.errorLines.append(j)
self.underindentedLine(line)
elif s[i] in (
' ',
'\t',
):
i += 1 # speed up the scan.
elif self.startsComment(s, i):
i = self.skipComment(s, i)
elif self.startsString(s, i):
i = self.skipString(s, i)
elif g.match(s, i, delim2):
i += len(delim2)
if trace: g.trace('returns\n', repr(s[start:i]))
return i
else:
i += 1
assert progress < i
self.error('no interface')
if 1:
g.pr('** no interface **')
i, j = g.getLine(s, start)
g.trace(i, s[i:j])
else:
if trace: g.trace('** no interface')
return start
def pythonNewlineHelper (self,s,i,parenCount,startIndent,underIndentedStart):
trace = False
breakFlag = False
j, indent = g.skip_leading_ws_with_indent(s,i,self.tab_width)
if trace: g.trace(
'startIndent',startIndent,'indent',indent,'parenCount',parenCount,
'line',repr(g.get_line(s,j)))
if indent <= startIndent and parenCount == 0:
# An underindented line: it ends the block *unless*
# it is a blank or comment line or (2008/9/1) the end of a triple-quoted string.
if g.match(s,j,'#'):
if trace: g.trace('underindent: comment')
if underIndentedStart is None: underIndentedStart = i
i = j
elif g.match(s,j,'\n'):
if trace: g.trace('underindent: blank line')
# Blank lines never start the range of underindented lines.
i = j
else:
if trace: g.trace('underindent: end of block')
breakFlag = True # The actual end of the block.
else:
if underIndentedStart and g.match(s,j,'\n'):
# Add the blank line to the underindented range.
if trace: g.trace('properly indented blank line extends underindent range')
elif underIndentedStart and g.match(s,j,'#'):
# Add the (properly indented!) comment line to the underindented range.
if trace: g.trace('properly indented comment line extends underindent range')
elif underIndentedStart is None:
pass
else:
# A properly indented non-comment line.
# Give a message for all underindented comments in underindented range.
if trace: g.trace('properly indented line generates underindent errors')
s2 = s[underIndentedStart:i]
lines = g.splitlines(s2)
for line in lines:
if line.strip():
junk, indent = g.skip_leading_ws_with_indent(line,0,self.tab_width)
if indent <= startIndent:
if j not in self.errorLines: # No error yet given.
self.errorLines.append(j)
self.underindentedComment(line)
underIndentedStart = None
if trace: g.trace('breakFlag',breakFlag,'returns',i,'underIndentedStart',underIndentedStart)
return i,underIndentedStart,breakFlag
def scanFunction(self, parent, s, i):
'''scan function(args) { body } and then rescan body.'''
trace = True and not g.unitTesting
# k1, k2 = g.getLine(s,i)
# g.trace(s[k1:k2])
i1 = i
# Scan backward for '('
i2 = i - 1
while 0 <= i2 and s[i2].isspace():
i2 -= 1
in_expr = s[i2] == '('
assert g.match(s, i, 'function')
i += len('function')
i = g.skip_ws_and_nl(s, i)
i = self.skipId(s, i)
i = g.skip_ws_and_nl(s, i)
# Skip the argument list.
if not g.match(s, i, '('):
if trace: g.trace('syntax error: no argument list', i)
return i
i = self.skipBlock(s, i, '(', ')')
# skipBlock skips the ')'
if i == len(s):
g.trace('no args', g.get_line(s, i))
return i
# Skip the body.
i = g.skip_ws_and_nl(s, i)
if not g.match(s, i, '{'):
if trace: g.trace('no function body', i)
return i
block_i1 = i
block_i2 = i = self.skipBlock(s, i, '{', '}')
j = g.skip_ws_and_nl(s, i)
if g.match(s, j, '('):
i = g.skip_parens(s, i)
if in_expr:
j = g.skip_ws_and_nl(s, i)
if g.match(s, j, ')'):
i = j + 1
assert i > i1
block_s = s[block_i1 + 1:block_i2 - 1].strip()
if trace: g.trace('*** rescanning ***\n\n%s\n\n' % block_s)
self.scanHelper(parent, block_s)
return i
def scanFunction(self, parent, s, i):
'''scan function(args) { body } and then rescan body.'''
trace = True and not g.unitTesting
# k1, k2 = g.getLine(s,i)
# g.trace(s[k1:k2])
i1 = i
# Scan backward for '('
i2 = i-1
while 0 <= i2 and s[i2].isspace():
i2 -= 1
in_expr = s[i2] == '('
assert g.match(s, i, 'function')
i += len('function')
i = g.skip_ws_and_nl(s, i)
i = self.skipId(s, i)
i = g.skip_ws_and_nl(s, i)
# Skip the argument list.
if not g.match(s, i, '('):
if trace: g.trace('syntax error: no argument list',i)
return i
i = self.skipBlock(s, i,'(', ')')
# skipBlock skips the ')'
if i == len(s):
g.trace('no args', g.get_line(s, i))
return i
# Skip the body.
i = g.skip_ws_and_nl(s, i)
if not g.match(s, i, '{'):
if trace: g.trace('no function body', i)
return i
block_i1 = i
block_i2 = i = self.skipBlock(s, i, '{', '}')
j = g.skip_ws_and_nl(s,i)
if g.match(s, j, '('):
i = g.skip_parens(s, i)
if in_expr:
j = g.skip_ws_and_nl(s,i)
if g.match(s, j, ')'):
i = j + 1
assert i > i1
block_s = s[block_i1+1:block_i2-1].strip()
if trace: g.trace('*** rescanning ***\n\n%s\n\n' % block_s)
self.scanHelper(parent, block_s)
return i
def startsHelper(self,s,i,kind,tags,tag=None):
'''return True if s[i:] starts a class or function.
Sets sigStart, sigEnd, sigId and codeEnd ivars.'''
trace = (False and kind == 'class') # and not g.unitTesting
verbose = True
self.codeEnd = self.sigEnd = self.sigId = None
# Underindented lines can happen in any language, not just Python.
# The skipBlock method of the base class checks for such lines.
self.startSigIndent = self.getLeadingIndent(s,i)
# Get the tag that starts the class or function.
if not g.match(s,i,'<'): return False
self.sigStart = i
i += 1
sigIdStart = j = g.skip_ws_and_nl(s,i)
i = self.skipId(s,j)
# Fix bug 501636: Leo's import code should support non-ascii xml tags.
# The call to g.toUnicode only needed on Python 2.x.
self.sigId = theId = g.toUnicode(s[j:i].lower())
# Set sigId ivar 'early' for error messages.
# Bug fix: html case does not matter.
if not theId: return False
if theId not in tags:
if trace and verbose:
g.trace('**** %s theId: %s not in tags: %s' % (
kind,theId,tags))
return False
if trace and verbose: g.trace(theId)
classId = ''
sigId = theId
# Complete the opening tag.
i, ok, complete = self.skipToEndOfTag(s,i,start=sigIdStart)
if not ok:
if trace and verbose: g.trace('no tail',g.get_line(s,i))
return False
sigEnd = i
# Bug fix: 2011/11/05.
# For xml/html, make sure the signature includes any trailing whitespace.
if not g.match(s,sigEnd,'\n') and not g.match(s,sigEnd-1,'\n'):
# sigEnd = g.skip_line(s,sigEnd)
sigEnd = g.skip_ws(s,sigEnd)
if not complete:
i,ok = self.skipToMatchingTag(s,i,theId,tags,start=sigIdStart)
if not ok:
if trace and verbose: g.trace('no matching tag:',theId)
return False
# Success: set the ivars.
# Not used in xml/html.
# self.sigStart = self.adjustDefStart(s,self.sigStart)
self.codeEnd = i
self.sigEnd = sigEnd
self.sigId = sigId
self.classId = classId
# Scan to the start of the next tag.
done = False
while not done and i < len(s):
progress = i
if self.startsComment(s,i):
i = self.skipComment(s,i)
elif self.startsString(s,i):
i = self.skipString(s,i)
elif s[i] == '<':
start = i
i += 1
if i < len(s) and s[i] == '/':
i += 1
j = g.skip_ws_and_nl(s,i)
if self.startsId(s,j):
i = self.skipId(s,j)
word = s[j:i].lower()
if word in tags:
self.codeEnd = start
done = True
break
else:
i = j
else:
i += 1
assert done or progress < i,'i: %d, ch: %s' % (i,repr(s[i]))
if trace: g.trace(repr(s[self.sigStart:self.codeEnd]))
return True
def startsHelper(self, s, i, kind, tags, tag=None):
'''return True if s[i:] starts a class or function.
Sets sigStart, sigEnd, sigId and codeEnd ivars.'''
trace = (False and kind == 'class') # and not g.unitTesting
verbose = True
self.codeEnd = self.sigEnd = self.sigId = None
# Underindented lines can happen in any language, not just Python.
# The skipBlock method of the base class checks for such lines.
self.startSigIndent = self.getLeadingIndent(s, i)
# Get the tag that starts the class or function.
if not g.match(s, i, '<'): return False
self.sigStart = i
i += 1
sigIdStart = j = g.skip_ws_and_nl(s, i)
i = self.skipId(s, j)
# Fix bug 501636: Leo's import code should support non-ascii xml tags.
# The call to g.toUnicode only needed on Python 2.x.
self.sigId = theId = g.toUnicode(s[j:i].lower())
# Set sigId ivar 'early' for error messages.
# Bug fix: html case does not matter.
if not theId: return False
if theId not in tags:
if trace and verbose:
g.trace('**** %s theId: %s not in tags: %s' %
(kind, theId, tags))
return False
if trace and verbose: g.trace(theId)
classId = ''
sigId = theId
# Complete the opening tag.
i, ok, complete = self.skipToEndOfTag(s, i, start=sigIdStart)
if not ok:
if trace and verbose: g.trace('no tail', g.get_line(s, i))
return False
sigEnd = i
# Bug fix: 2011/11/05.
# For xml/html, make sure the signature includes any trailing whitespace.
if not g.match(s, sigEnd, '\n') and not g.match(s, sigEnd - 1, '\n'):
# sigEnd = g.skip_line(s,sigEnd)
sigEnd = g.skip_ws(s, sigEnd)
if not complete:
i, ok = self.skipToMatchingTag(s, i, theId, tags, start=sigIdStart)
if not ok:
if trace and verbose: g.trace('no matching tag:', theId)
return False
# Success: set the ivars.
# Not used in xml/html.
# self.sigStart = self.adjustDefStart(s,self.sigStart)
self.codeEnd = i
self.sigEnd = sigEnd
self.sigId = sigId
self.classId = classId
# Scan to the start of the next tag.
done = False
while not done and i < len(s):
progress = i
if self.startsComment(s, i):
i = self.skipComment(s, i)
elif self.startsString(s, i):
i = self.skipString(s, i)
elif s[i] == '<':
start = i
i += 1
if i < len(s) and s[i] == '/':
i += 1
j = g.skip_ws_and_nl(s, i)
if self.startsId(s, j):
i = self.skipId(s, j)
word = s[j:i].lower()
if word in tags:
self.codeEnd = start
done = True
break
else:
i = j
else:
i += 1
assert done or progress < i, 'i: %d, ch: %s' % (i, repr(s[i]))
if trace: g.trace(repr(s[self.sigStart:self.codeEnd]))
return True