Spaces Data
Minimal test - lines (285, 1273)
path: .spaces[11].metrics.mi.mi_sei
old: -56.70445274425955
new: -56.733169013583094
path: .spaces[11].metrics.mi.mi_original
old: -22.144053616693412
new: -22.176847150158935
path: .spaces[11].metrics.loc.cloc
old: 298.0
new: 299.0
path: .spaces[11].metrics.loc.blank
old: 102.0
new: 103.0
path: .spaces[11].metrics.loc.sloc
old: 987.0
new: 989.0
Code
class LRParser:
def __init__(self, lrtab, errorf):
self.productions = lrtab.lr_productions
self.action = lrtab.lr_action
self.goto = lrtab.lr_goto
self.errorfunc = errorf
self.set_defaulted_states()
self.errorok = True
def errok(self):
self.errorok = True
def restart(self):
del self.statestack[:]
del self.symstack[:]
sym = YaccSymbol()
sym.type = '$end'
self.symstack.append(sym)
self.statestack.append(0)
# Defaulted state support.
# This method identifies parser states where there is only one possible reduction action.
# For such states, the parser can make a choose to make a rule reduction without consuming
# the next look-ahead token. This delayed invocation of the tokenizer can be useful in
# certain kinds of advanced parsing situations where the lexer and parser interact with
# each other or change states (i.e., manipulation of scope, lexer states, etc.).
#
# See: http://www.gnu.org/software/bison/manual/html_node/Default-Reductions.html#Default-Reductions
def set_defaulted_states(self):
self.defaulted_states = {}
for state, actions in self.action.items():
rules = list(actions.values())
if len(rules) == 1 and rules[0] < 0:
self.defaulted_states[state] = rules[0]
def disable_defaulted_states(self):
self.defaulted_states = {}
def parse(self, input=None, lexer=None, debug=False, tracking=False, tokenfunc=None):
if debug or yaccdevel:
if isinstance(debug, int):
debug = PlyLogger(sys.stderr)
return self.parsedebug(input, lexer, debug, tracking, tokenfunc)
elif tracking:
return self.parseopt(input, lexer, debug, tracking, tokenfunc)
else:
return self.parseopt_notrack(input, lexer, debug, tracking, tokenfunc)
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# parsedebug().
#
# This is the debugging enabled version of parse(). All changes made to the
# parsing engine should be made here. Optimized versions of this function
# are automatically created by the ply/ygen.py script. This script cuts out
# sections enclosed in markers such as this:
#
# #--! DEBUG
# statements
# #--! DEBUG
#
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
def parsedebug(self, input=None, lexer=None, debug=False, tracking=False, tokenfunc=None):
#--! parsedebug-start
lookahead = None # Current lookahead symbol
lookaheadstack = [] # Stack of lookahead symbols
actions = self.action # Local reference to action table (to avoid lookup on self.)
goto = self.goto # Local reference to goto table (to avoid lookup on self.)
prod = self.productions # Local reference to production list (to avoid lookup on self.)
defaulted_states = self.defaulted_states # Local reference to defaulted states
pslice = YaccProduction(None) # Production object passed to grammar rules
errorcount = 0 # Used during error recovery
#--! DEBUG
debug.info('PLY: PARSE DEBUG START')
#--! DEBUG
# If no lexer was given, we will try to use the lex module
if not lexer:
from . import lex
lexer = lex.lexer
# Set up the lexer and parser objects on pslice
pslice.lexer = lexer
pslice.parser = self
# If input was supplied, pass to lexer
if input is not None:
lexer.input(input)
if tokenfunc is None:
# Tokenize function
get_token = lexer.token
else:
get_token = tokenfunc
# Set the parser() token method (sometimes used in error recovery)
self.token = get_token
# Set up the state and symbol stacks
statestack = [] # Stack of parsing states
self.statestack = statestack
symstack = [] # Stack of grammar symbols
self.symstack = symstack
pslice.stack = symstack # Put in the production
errtoken = None # Err token
# The start state is assumed to be (0,$end)
statestack.append(0)
sym = YaccSymbol()
sym.type = '$end'
symstack.append(sym)
state = 0
while True:
# Get the next symbol on the input. If a lookahead symbol
# is already set, we just use that. Otherwise, we'll pull
# the next token off of the lookaheadstack or from the lexer
#--! DEBUG
debug.debug('')
debug.debug('State : %s', state)
#--! DEBUG
if state not in defaulted_states:
if not lookahead:
if not lookaheadstack:
lookahead = get_token() # Get the next token
else:
lookahead = lookaheadstack.pop()
if not lookahead:
lookahead = YaccSymbol()
lookahead.type = '$end'
# Check the action table
ltype = lookahead.type
t = actions[state].get(ltype)
else:
t = defaulted_states[state]
#--! DEBUG
debug.debug('Defaulted state %s: Reduce using %d', state, -t)
#--! DEBUG
#--! DEBUG
debug.debug('Stack : %s',
('%s . %s' % (' '.join([xx.type for xx in symstack][1:]), str(lookahead))).lstrip())
#--! DEBUG
if t is not None:
if t > 0:
# shift a symbol on the stack
statestack.append(t)
state = t
#--! DEBUG
debug.debug('Action : Shift and goto state %s', t)
#--! DEBUG
symstack.append(lookahead)
lookahead = None
# Decrease error count on successful shift
if errorcount:
errorcount -= 1
continue
if t < 0:
# reduce a symbol on the stack, emit a production
p = prod[-t]
pname = p.name
plen = p.len
# Get production function
sym = YaccSymbol()
sym.type = pname # Production name
sym.value = None
#--! DEBUG
if plen:
debug.info('Action : Reduce rule [%s] with %s and goto state %d', p.str,
'['+','.join([format_stack_entry(_v.value) for _v in symstack[-plen:]])+']',
goto[statestack[-1-plen]][pname])
else:
debug.info('Action : Reduce rule [%s] with %s and goto state %d', p.str, [],
goto[statestack[-1]][pname])
#--! DEBUG
if plen:
targ = symstack[-plen-1:]
targ[0] = sym
#--! TRACKING
if tracking:
t1 = targ[1]
sym.lineno = t1.lineno
sym.lexpos = t1.lexpos
t1 = targ[-1]
sym.endlineno = getattr(t1, 'endlineno', t1.lineno)
sym.endlexpos = getattr(t1, 'endlexpos', t1.lexpos)
#--! TRACKING
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# The code enclosed in this section is duplicated
# below as a performance optimization. Make sure
# changes get made in both locations.
pslice.slice = targ
try:
# Call the grammar rule with our special slice object
del symstack[-plen:]
self.state = state
p.callable(pslice)
del statestack[-plen:]
#--! DEBUG
debug.info('Result : %s', format_result(pslice[0]))
#--! DEBUG
symstack.append(sym)
state = goto[statestack[-1]][pname]
statestack.append(state)
except SyntaxError:
# If an error was set. Enter error recovery state
lookaheadstack.append(lookahead) # Save the current lookahead token
symstack.extend(targ[1:-1]) # Put the production slice back on the stack
statestack.pop() # Pop back one state (before the reduce)
state = statestack[-1]
sym.type = 'error'
sym.value = 'error'
lookahead = sym
errorcount = error_count
self.errorok = False
continue
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
else:
#--! TRACKING
if tracking:
sym.lineno = lexer.lineno
sym.lexpos = lexer.lexpos
#--! TRACKING
targ = [sym]
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# The code enclosed in this section is duplicated
# above as a performance optimization. Make sure
# changes get made in both locations.
pslice.slice = targ
try:
# Call the grammar rule with our special slice object
self.state = state
p.callable(pslice)
#--! DEBUG
debug.info('Result : %s', format_result(pslice[0]))
#--! DEBUG
symstack.append(sym)
state = goto[statestack[-1]][pname]
statestack.append(state)
except SyntaxError:
# If an error was set. Enter error recovery state
lookaheadstack.append(lookahead) # Save the current lookahead token
statestack.pop() # Pop back one state (before the reduce)
state = statestack[-1]
sym.type = 'error'
sym.value = 'error'
lookahead = sym
errorcount = error_count
self.errorok = False
continue
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
if t == 0:
n = symstack[-1]
result = getattr(n, 'value', None)
#--! DEBUG
debug.info('Done : Returning %s', format_result(result))
debug.info('PLY: PARSE DEBUG END')
#--! DEBUG
return result
if t is None:
#--! DEBUG
debug.error('Error : %s',
('%s . %s' % (' '.join([xx.type for xx in symstack][1:]), str(lookahead))).lstrip())
#--! DEBUG
# We have some kind of parsing error here. To handle
# this, we are going to push the current token onto
# the tokenstack and replace it with an 'error' token.
# If there are any synchronization rules, they may
# catch it.
#
# In addition to pushing the error token, we call call
# the user defined p_error() function if this is the
# first syntax error. This function is only called if
# errorcount == 0.
if errorcount == 0 or self.errorok:
errorcount = error_count
self.errorok = False
errtoken = lookahead
if errtoken.type == '$end':
errtoken = None # End of file!
if self.errorfunc:
if errtoken and not hasattr(errtoken, 'lexer'):
errtoken.lexer = lexer
self.state = state
tok = call_errorfunc(self.errorfunc, errtoken, self)
if self.errorok:
# User must have done some kind of panic
# mode recovery on their own. The
# returned token is the next lookahead
lookahead = tok
errtoken = None
continue
else:
if errtoken:
if hasattr(errtoken, 'lineno'):
lineno = lookahead.lineno
else:
lineno = 0
if lineno:
sys.stderr.write('yacc: Syntax error at line %d, token=%s\n' % (lineno, errtoken.type))
else:
sys.stderr.write('yacc: Syntax error, token=%s' % errtoken.type)
else:
sys.stderr.write('yacc: Parse error in input. EOF\n')
return
else:
errorcount = error_count
# case 1: the statestack only has 1 entry on it. If we're in this state, the
# entire parse has been rolled back and we're completely hosed. The token is
# discarded and we just keep going.
if len(statestack) <= 1 and lookahead.type != '$end':
lookahead = None
errtoken = None
state = 0
# Nuke the pushback stack
del lookaheadstack[:]
continue
# case 2: the statestack has a couple of entries on it, but we're
# at the end of the file. nuke the top entry and generate an error token
# Start nuking entries on the stack
if lookahead.type == '$end':
# Whoa. We're really hosed here. Bail out
return
if lookahead.type != 'error':
sym = symstack[-1]
if sym.type == 'error':
# Hmmm. Error is on top of stack, we'll just nuke input
# symbol and continue
#--! TRACKING
if tracking:
sym.endlineno = getattr(lookahead, 'lineno', sym.lineno)
sym.endlexpos = getattr(lookahead, 'lexpos', sym.lexpos)
#--! TRACKING
lookahead = None
continue
# Create the error symbol for the first time and make it the new lookahead symbol
t = YaccSymbol()
t.type = 'error'
if hasattr(lookahead, 'lineno'):
t.lineno = t.endlineno = lookahead.lineno
if hasattr(lookahead, 'lexpos'):
t.lexpos = t.endlexpos = lookahead.lexpos
t.value = lookahead
lookaheadstack.append(lookahead)
lookahead = t
else:
sym = symstack.pop()
#--! TRACKING
if tracking:
lookahead.lineno = sym.lineno
lookahead.lexpos = sym.lexpos
#--! TRACKING
statestack.pop()
state = statestack[-1]
continue
# Call an error function here
raise RuntimeError('yacc: internal parser error!!!\n')
#--! parsedebug-end
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# parseopt().
#
# Optimized version of parse() method. DO NOT EDIT THIS CODE DIRECTLY!
# This code is automatically generated by the ply/ygen.py script. Make
# changes to the parsedebug() method instead.
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
def parseopt(self, input=None, lexer=None, debug=False, tracking=False, tokenfunc=None):
#--! parseopt-start
lookahead = None # Current lookahead symbol
lookaheadstack = [] # Stack of lookahead symbols
actions = self.action # Local reference to action table (to avoid lookup on self.)
goto = self.goto # Local reference to goto table (to avoid lookup on self.)
prod = self.productions # Local reference to production list (to avoid lookup on self.)
defaulted_states = self.defaulted_states # Local reference to defaulted states
pslice = YaccProduction(None) # Production object passed to grammar rules
errorcount = 0 # Used during error recovery
# If no lexer was given, we will try to use the lex module
if not lexer:
from . import lex
lexer = lex.lexer
# Set up the lexer and parser objects on pslice
pslice.lexer = lexer
pslice.parser = self
# If input was supplied, pass to lexer
if input is not None:
lexer.input(input)
if tokenfunc is None:
# Tokenize function
get_token = lexer.token
else:
get_token = tokenfunc
# Set the parser() token method (sometimes used in error recovery)
self.token = get_token
# Set up the state and symbol stacks
statestack = [] # Stack of parsing states
self.statestack = statestack
symstack = [] # Stack of grammar symbols
self.symstack = symstack
pslice.stack = symstack # Put in the production
errtoken = None # Err token
# The start state is assumed to be (0,$end)
statestack.append(0)
sym = YaccSymbol()
sym.type = '$end'
symstack.append(sym)
state = 0
while True:
# Get the next symbol on the input. If a lookahead symbol
# is already set, we just use that. Otherwise, we'll pull
# the next token off of the lookaheadstack or from the lexer
if state not in defaulted_states:
if not lookahead:
if not lookaheadstack:
lookahead = get_token() # Get the next token
else:
lookahead = lookaheadstack.pop()
if not lookahead:
lookahead = YaccSymbol()
lookahead.type = '$end'
# Check the action table
ltype = lookahead.type
t = actions[state].get(ltype)
else:
t = defaulted_states[state]
if t is not None:
if t > 0:
# shift a symbol on the stack
statestack.append(t)
state = t
symstack.append(lookahead)
lookahead = None
# Decrease error count on successful shift
if errorcount:
errorcount -= 1
continue
if t < 0:
# reduce a symbol on the stack, emit a production
p = prod[-t]
pname = p.name
plen = p.len
# Get production function
sym = YaccSymbol()
sym.type = pname # Production name
sym.value = None
if plen:
targ = symstack[-plen-1:]
targ[0] = sym
#--! TRACKING
if tracking:
t1 = targ[1]
sym.lineno = t1.lineno
sym.lexpos = t1.lexpos
t1 = targ[-1]
sym.endlineno = getattr(t1, 'endlineno', t1.lineno)
sym.endlexpos = getattr(t1, 'endlexpos', t1.lexpos)
#--! TRACKING
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# The code enclosed in this section is duplicated
# below as a performance optimization. Make sure
# changes get made in both locations.
pslice.slice = targ
try:
# Call the grammar rule with our special slice object
del symstack[-plen:]
self.state = state
p.callable(pslice)
del statestack[-plen:]
symstack.append(sym)
state = goto[statestack[-1]][pname]
statestack.append(state)
except SyntaxError:
# If an error was set. Enter error recovery state
lookaheadstack.append(lookahead) # Save the current lookahead token
symstack.extend(targ[1:-1]) # Put the production slice back on the stack
statestack.pop() # Pop back one state (before the reduce)
state = statestack[-1]
sym.type = 'error'
sym.value = 'error'
lookahead = sym
errorcount = error_count
self.errorok = False
continue
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
else:
#--! TRACKING
if tracking:
sym.lineno = lexer.lineno
sym.lexpos = lexer.lexpos
#--! TRACKING
targ = [sym]
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# The code enclosed in this section is duplicated
# above as a performance optimization. Make sure
# changes get made in both locations.
pslice.slice = targ
try:
# Call the grammar rule with our special slice object
self.state = state
p.callable(pslice)
symstack.append(sym)
state = goto[statestack[-1]][pname]
statestack.append(state)
except SyntaxError:
# If an error was set. Enter error recovery state
lookaheadstack.append(lookahead) # Save the current lookahead token
statestack.pop() # Pop back one state (before the reduce)
state = statestack[-1]
sym.type = 'error'
sym.value = 'error'
lookahead = sym
errorcount = error_count
self.errorok = False
continue
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
if t == 0:
n = symstack[-1]
result = getattr(n, 'value', None)
return result
if t is None:
# We have some kind of parsing error here. To handle
# this, we are going to push the current token onto
# the tokenstack and replace it with an 'error' token.
# If there are any synchronization rules, they may
# catch it.
#
# In addition to pushing the error token, we call call
# the user defined p_error() function if this is the
# first syntax error. This function is only called if
# errorcount == 0.
if errorcount == 0 or self.errorok:
errorcount = error_count
self.errorok = False
errtoken = lookahead
if errtoken.type == '$end':
errtoken = None # End of file!
if self.errorfunc:
if errtoken and not hasattr(errtoken, 'lexer'):
errtoken.lexer = lexer
self.state = state
tok = call_errorfunc(self.errorfunc, errtoken, self)
if self.errorok:
# User must have done some kind of panic
# mode recovery on their own. The
# returned token is the next lookahead
lookahead = tok
errtoken = None
continue
else:
if errtoken:
if hasattr(errtoken, 'lineno'):
lineno = lookahead.lineno
else:
lineno = 0
if lineno:
sys.stderr.write('yacc: Syntax error at line %d, token=%s\n' % (lineno, errtoken.type))
else:
sys.stderr.write('yacc: Syntax error, token=%s' % errtoken.type)
else:
sys.stderr.write('yacc: Parse error in input. EOF\n')
return
else:
errorcount = error_count
# case 1: the statestack only has 1 entry on it. If we're in this state, the
# entire parse has been rolled back and we're completely hosed. The token is
# discarded and we just keep going.
if len(statestack) <= 1 and lookahead.type != '$end':
lookahead = None
errtoken = None
state = 0
# Nuke the pushback stack
del lookaheadstack[:]
continue
# case 2: the statestack has a couple of entries on it, but we're
# at the end of the file. nuke the top entry and generate an error token
# Start nuking entries on the stack
if lookahead.type == '$end':
# Whoa. We're really hosed here. Bail out
return
if lookahead.type != 'error':
sym = symstack[-1]
if sym.type == 'error':
# Hmmm. Error is on top of stack, we'll just nuke input
# symbol and continue
#--! TRACKING
if tracking:
sym.endlineno = getattr(lookahead, 'lineno', sym.lineno)
sym.endlexpos = getattr(lookahead, 'lexpos', sym.lexpos)
#--! TRACKING
lookahead = None
continue
# Create the error symbol for the first time and make it the new lookahead symbol
t = YaccSymbol()
t.type = 'error'
if hasattr(lookahead, 'lineno'):
t.lineno = t.endlineno = lookahead.lineno
if hasattr(lookahead, 'lexpos'):
t.lexpos = t.endlexpos = lookahead.lexpos
t.value = lookahead
lookaheadstack.append(lookahead)
lookahead = t
else:
sym = symstack.pop()
#--! TRACKING
if tracking:
lookahead.lineno = sym.lineno
lookahead.lexpos = sym.lexpos
#--! TRACKING
statestack.pop()
state = statestack[-1]
continue
# Call an error function here
raise RuntimeError('yacc: internal parser error!!!\n')
#--! parseopt-end
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# parseopt_notrack().
#
# Optimized version of parseopt() with line number tracking removed.
# DO NOT EDIT THIS CODE DIRECTLY. This code is automatically generated
# by the ply/ygen.py script. Make changes to the parsedebug() method instead.
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
def parseopt_notrack(self, input=None, lexer=None, debug=False, tracking=False, tokenfunc=None):
#--! parseopt-notrack-start
lookahead = None # Current lookahead symbol
lookaheadstack = [] # Stack of lookahead symbols
actions = self.action # Local reference to action table (to avoid lookup on self.)
goto = self.goto # Local reference to goto table (to avoid lookup on self.)
prod = self.productions # Local reference to production list (to avoid lookup on self.)
defaulted_states = self.defaulted_states # Local reference to defaulted states
pslice = YaccProduction(None) # Production object passed to grammar rules
errorcount = 0 # Used during error recovery
# If no lexer was given, we will try to use the lex module
if not lexer:
from . import lex
lexer = lex.lexer
# Set up the lexer and parser objects on pslice
pslice.lexer = lexer
pslice.parser = self
# If input was supplied, pass to lexer
if input is not None:
lexer.input(input)
if tokenfunc is None:
# Tokenize function
get_token = lexer.token
else:
get_token = tokenfunc
# Set the parser() token method (sometimes used in error recovery)
self.token = get_token
# Set up the state and symbol stacks
statestack = [] # Stack of parsing states
self.statestack = statestack
symstack = [] # Stack of grammar symbols
self.symstack = symstack
pslice.stack = symstack # Put in the production
errtoken = None # Err token
# The start state is assumed to be (0,$end)
statestack.append(0)
sym = YaccSymbol()
sym.type = '$end'
symstack.append(sym)
state = 0
while True:
# Get the next symbol on the input. If a lookahead symbol
# is already set, we just use that. Otherwise, we'll pull
# the next token off of the lookaheadstack or from the lexer
if state not in defaulted_states:
if not lookahead:
if not lookaheadstack:
lookahead = get_token() # Get the next token
else:
lookahead = lookaheadstack.pop()
if not lookahead:
lookahead = YaccSymbol()
lookahead.type = '$end'
# Check the action table
ltype = lookahead.type
t = actions[state].get(ltype)
else:
t = defaulted_states[state]
if t is not None:
if t > 0:
# shift a symbol on the stack
statestack.append(t)
state = t
symstack.append(lookahead)
lookahead = None
# Decrease error count on successful shift
if errorcount:
errorcount -= 1
continue
if t < 0:
# reduce a symbol on the stack, emit a production
p = prod[-t]
pname = p.name
plen = p.len
# Get production function
sym = YaccSymbol()
sym.type = pname # Production name
sym.value = None
if plen:
targ = symstack[-plen-1:]
targ[0] = sym
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# The code enclosed in this section is duplicated
# below as a performance optimization. Make sure
# changes get made in both locations.
pslice.slice = targ
try:
# Call the grammar rule with our special slice object
del symstack[-plen:]
self.state = state
p.callable(pslice)
del statestack[-plen:]
symstack.append(sym)
state = goto[statestack[-1]][pname]
statestack.append(state)
except SyntaxError:
# If an error was set. Enter error recovery state
lookaheadstack.append(lookahead) # Save the current lookahead token
symstack.extend(targ[1:-1]) # Put the production slice back on the stack
statestack.pop() # Pop back one state (before the reduce)
state = statestack[-1]
sym.type = 'error'
sym.value = 'error'
lookahead = sym
errorcount = error_count
self.errorok = False
continue
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
else:
targ = [sym]
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# The code enclosed in this section is duplicated
# above as a performance optimization. Make sure
# changes get made in both locations.
pslice.slice = targ
try:
# Call the grammar rule with our special slice object
self.state = state
p.callable(pslice)
symstack.append(sym)
state = goto[statestack[-1]][pname]
statestack.append(state)
except SyntaxError:
# If an error was set. Enter error recovery state
lookaheadstack.append(lookahead) # Save the current lookahead token
statestack.pop() # Pop back one state (before the reduce)
state = statestack[-1]
sym.type = 'error'
sym.value = 'error'
lookahead = sym
errorcount = error_count
self.errorok = False
continue
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
if t == 0:
n = symstack[-1]
result = getattr(n, 'value', None)
return result
if t is None:
# We have some kind of parsing error here. To handle
# this, we are going to push the current token onto
# the tokenstack and replace it with an 'error' token.
# If there are any synchronization rules, they may
# catch it.
#
# In addition to pushing the error token, we call call
# the user defined p_error() function if this is the
# first syntax error. This function is only called if
# errorcount == 0.
if errorcount == 0 or self.errorok:
errorcount = error_count
self.errorok = False
errtoken = lookahead
if errtoken.type == '$end':
errtoken = None # End of file!
if self.errorfunc:
if errtoken and not hasattr(errtoken, 'lexer'):
errtoken.lexer = lexer
self.state = state
tok = call_errorfunc(self.errorfunc, errtoken, self)
if self.errorok:
# User must have done some kind of panic
# mode recovery on their own. The
# returned token is the next lookahead
lookahead = tok
errtoken = None
continue
else:
if errtoken:
if hasattr(errtoken, 'lineno'):
lineno = lookahead.lineno
else:
lineno = 0
if lineno:
sys.stderr.write('yacc: Syntax error at line %d, token=%s\n' % (lineno, errtoken.type))
else:
sys.stderr.write('yacc: Syntax error, token=%s' % errtoken.type)
else:
sys.stderr.write('yacc: Parse error in input. EOF\n')
return
else:
errorcount = error_count
# case 1: the statestack only has 1 entry on it. If we're in this state, the
# entire parse has been rolled back and we're completely hosed. The token is
# discarded and we just keep going.
if len(statestack) <= 1 and lookahead.type != '$end':
lookahead = None
errtoken = None
state = 0
# Nuke the pushback stack
del lookaheadstack[:]
continue
# case 2: the statestack has a couple of entries on it, but we're
# at the end of the file. nuke the top entry and generate an error token
# Start nuking entries on the stack
if lookahead.type == '$end':
# Whoa. We're really hosed here. Bail out
return
if lookahead.type != 'error':
sym = symstack[-1]
if sym.type == 'error':
# Hmmm. Error is on top of stack, we'll just nuke input
# symbol and continue
lookahead = None
continue
# Create the error symbol for the first time and make it the new lookahead symbol
t = YaccSymbol()
t.type = 'error'
if hasattr(lookahead, 'lineno'):
t.lineno = t.endlineno = lookahead.lineno
if hasattr(lookahead, 'lexpos'):
t.lexpos = t.endlexpos = lookahead.lexpos
t.value = lookahead
lookaheadstack.append(lookahead)
lookahead = t
else:
sym = symstack.pop()
statestack.pop()
state = statestack[-1]
continue
# Call an error function here
raise RuntimeError('yacc: internal parser error!!!\n')
#--! parseopt-notrack-end
Minimal test - lines (695, 991)
path: .spaces[11].spaces[7].metrics.mi.mi_visual_studio
old: 14.750568775532573
new: 14.686557152200027
path: .spaces[11].spaces[7].metrics.mi.mi_sei
old: 1.2764412093476525
new: 1.191549450015586
path: .spaces[11].spaces[7].metrics.mi.mi_original
old: 25.223472606160698
new: 25.114012730262047
path: .spaces[11].spaces[7].metrics.loc.blank
old: 29.0
new: 30.0
path: .spaces[11].spaces[7].metrics.loc.sloc
old: 295.0
new: 297.0
path: .spaces[11].spaces[7].metrics.loc.cloc
old: 83.0
new: 84.0
Code
def parseopt(self, input=None, lexer=None, debug=False, tracking=False, tokenfunc=None):
#--! parseopt-start
lookahead = None # Current lookahead symbol
lookaheadstack = [] # Stack of lookahead symbols
actions = self.action # Local reference to action table (to avoid lookup on self.)
goto = self.goto # Local reference to goto table (to avoid lookup on self.)
prod = self.productions # Local reference to production list (to avoid lookup on self.)
defaulted_states = self.defaulted_states # Local reference to defaulted states
pslice = YaccProduction(None) # Production object passed to grammar rules
errorcount = 0 # Used during error recovery
# If no lexer was given, we will try to use the lex module
if not lexer:
from . import lex
lexer = lex.lexer
# Set up the lexer and parser objects on pslice
pslice.lexer = lexer
pslice.parser = self
# If input was supplied, pass to lexer
if input is not None:
lexer.input(input)
if tokenfunc is None:
# Tokenize function
get_token = lexer.token
else:
get_token = tokenfunc
# Set the parser() token method (sometimes used in error recovery)
self.token = get_token
# Set up the state and symbol stacks
statestack = [] # Stack of parsing states
self.statestack = statestack
symstack = [] # Stack of grammar symbols
self.symstack = symstack
pslice.stack = symstack # Put in the production
errtoken = None # Err token
# The start state is assumed to be (0,$end)
statestack.append(0)
sym = YaccSymbol()
sym.type = '$end'
symstack.append(sym)
state = 0
while True:
# Get the next symbol on the input. If a lookahead symbol
# is already set, we just use that. Otherwise, we'll pull
# the next token off of the lookaheadstack or from the lexer
if state not in defaulted_states:
if not lookahead:
if not lookaheadstack:
lookahead = get_token() # Get the next token
else:
lookahead = lookaheadstack.pop()
if not lookahead:
lookahead = YaccSymbol()
lookahead.type = '$end'
# Check the action table
ltype = lookahead.type
t = actions[state].get(ltype)
else:
t = defaulted_states[state]
if t is not None:
if t > 0:
# shift a symbol on the stack
statestack.append(t)
state = t
symstack.append(lookahead)
lookahead = None
# Decrease error count on successful shift
if errorcount:
errorcount -= 1
continue
if t < 0:
# reduce a symbol on the stack, emit a production
p = prod[-t]
pname = p.name
plen = p.len
# Get production function
sym = YaccSymbol()
sym.type = pname # Production name
sym.value = None
if plen:
targ = symstack[-plen-1:]
targ[0] = sym
#--! TRACKING
if tracking:
t1 = targ[1]
sym.lineno = t1.lineno
sym.lexpos = t1.lexpos
t1 = targ[-1]
sym.endlineno = getattr(t1, 'endlineno', t1.lineno)
sym.endlexpos = getattr(t1, 'endlexpos', t1.lexpos)
#--! TRACKING
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# The code enclosed in this section is duplicated
# below as a performance optimization. Make sure
# changes get made in both locations.
pslice.slice = targ
try:
# Call the grammar rule with our special slice object
del symstack[-plen:]
self.state = state
p.callable(pslice)
del statestack[-plen:]
symstack.append(sym)
state = goto[statestack[-1]][pname]
statestack.append(state)
except SyntaxError:
# If an error was set. Enter error recovery state
lookaheadstack.append(lookahead) # Save the current lookahead token
symstack.extend(targ[1:-1]) # Put the production slice back on the stack
statestack.pop() # Pop back one state (before the reduce)
state = statestack[-1]
sym.type = 'error'
sym.value = 'error'
lookahead = sym
errorcount = error_count
self.errorok = False
continue
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
else:
#--! TRACKING
if tracking:
sym.lineno = lexer.lineno
sym.lexpos = lexer.lexpos
#--! TRACKING
targ = [sym]
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# The code enclosed in this section is duplicated
# above as a performance optimization. Make sure
# changes get made in both locations.
pslice.slice = targ
try:
# Call the grammar rule with our special slice object
self.state = state
p.callable(pslice)
symstack.append(sym)
state = goto[statestack[-1]][pname]
statestack.append(state)
except SyntaxError:
# If an error was set. Enter error recovery state
lookaheadstack.append(lookahead) # Save the current lookahead token
statestack.pop() # Pop back one state (before the reduce)
state = statestack[-1]
sym.type = 'error'
sym.value = 'error'
lookahead = sym
errorcount = error_count
self.errorok = False
continue
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
if t == 0:
n = symstack[-1]
result = getattr(n, 'value', None)
return result
if t is None:
# We have some kind of parsing error here. To handle
# this, we are going to push the current token onto
# the tokenstack and replace it with an 'error' token.
# If there are any synchronization rules, they may
# catch it.
#
# In addition to pushing the error token, we call call
# the user defined p_error() function if this is the
# first syntax error. This function is only called if
# errorcount == 0.
if errorcount == 0 or self.errorok:
errorcount = error_count
self.errorok = False
errtoken = lookahead
if errtoken.type == '$end':
errtoken = None # End of file!
if self.errorfunc:
if errtoken and not hasattr(errtoken, 'lexer'):
errtoken.lexer = lexer
self.state = state
tok = call_errorfunc(self.errorfunc, errtoken, self)
if self.errorok:
# User must have done some kind of panic
# mode recovery on their own. The
# returned token is the next lookahead
lookahead = tok
errtoken = None
continue
else:
if errtoken:
if hasattr(errtoken, 'lineno'):
lineno = lookahead.lineno
else:
lineno = 0
if lineno:
sys.stderr.write('yacc: Syntax error at line %d, token=%s\n' % (lineno, errtoken.type))
else:
sys.stderr.write('yacc: Syntax error, token=%s' % errtoken.type)
else:
sys.stderr.write('yacc: Parse error in input. EOF\n')
return
else:
errorcount = error_count
# case 1: the statestack only has 1 entry on it. If we're in this state, the
# entire parse has been rolled back and we're completely hosed. The token is
# discarded and we just keep going.
if len(statestack) <= 1 and lookahead.type != '$end':
lookahead = None
errtoken = None
state = 0
# Nuke the pushback stack
del lookaheadstack[:]
continue
# case 2: the statestack has a couple of entries on it, but we're
# at the end of the file. nuke the top entry and generate an error token
# Start nuking entries on the stack
if lookahead.type == '$end':
# Whoa. We're really hosed here. Bail out
return
if lookahead.type != 'error':
sym = symstack[-1]
if sym.type == 'error':
# Hmmm. Error is on top of stack, we'll just nuke input
# symbol and continue
#--! TRACKING
if tracking:
sym.endlineno = getattr(lookahead, 'lineno', sym.lineno)
sym.endlexpos = getattr(lookahead, 'lexpos', sym.lexpos)
#--! TRACKING
lookahead = None
continue
# Create the error symbol for the first time and make it the new lookahead symbol
t = YaccSymbol()
t.type = 'error'
if hasattr(lookahead, 'lineno'):
t.lineno = t.endlineno = lookahead.lineno
if hasattr(lookahead, 'lexpos'):
t.lexpos = t.endlexpos = lookahead.lexpos
t.value = lookahead
lookaheadstack.append(lookahead)
lookahead = t
else:
sym = symstack.pop()
#--! TRACKING
if tracking:
lookahead.lineno = sym.lineno
lookahead.lexpos = sym.lexpos
#--! TRACKING
statestack.pop()
state = statestack[-1]
continue
# Call an error function here
raise RuntimeError('yacc: internal parser error!!!\n')
#--! parseopt-end
Minimal test - lines (348, 685)
path: .spaces[11].spaces[6].metrics.mi.mi_original
old: 21.035894272552483
new: 20.939751557131487
path: .spaces[11].spaces[6].metrics.mi.mi_sei
old: -3.155980740420077
new: -3.2424030802285344
path: .spaces[11].spaces[6].metrics.mi.mi_visual_studio
old: 12.301692557048234
new: 12.245468746860515
path: .spaces[11].spaces[6].metrics.loc.cloc
old: 103.0
new: 104.0
path: .spaces[11].spaces[6].metrics.loc.sloc
old: 336.0
new: 338.0
path: .spaces[11].spaces[6].metrics.loc.blank
old: 30.0
new: 31.0
Code
def parsedebug(self, input=None, lexer=None, debug=False, tracking=False, tokenfunc=None):
#--! parsedebug-start
lookahead = None # Current lookahead symbol
lookaheadstack = [] # Stack of lookahead symbols
actions = self.action # Local reference to action table (to avoid lookup on self.)
goto = self.goto # Local reference to goto table (to avoid lookup on self.)
prod = self.productions # Local reference to production list (to avoid lookup on self.)
defaulted_states = self.defaulted_states # Local reference to defaulted states
pslice = YaccProduction(None) # Production object passed to grammar rules
errorcount = 0 # Used during error recovery
#--! DEBUG
debug.info('PLY: PARSE DEBUG START')
#--! DEBUG
# If no lexer was given, we will try to use the lex module
if not lexer:
from . import lex
lexer = lex.lexer
# Set up the lexer and parser objects on pslice
pslice.lexer = lexer
pslice.parser = self
# If input was supplied, pass to lexer
if input is not None:
lexer.input(input)
if tokenfunc is None:
# Tokenize function
get_token = lexer.token
else:
get_token = tokenfunc
# Set the parser() token method (sometimes used in error recovery)
self.token = get_token
# Set up the state and symbol stacks
statestack = [] # Stack of parsing states
self.statestack = statestack
symstack = [] # Stack of grammar symbols
self.symstack = symstack
pslice.stack = symstack # Put in the production
errtoken = None # Err token
# The start state is assumed to be (0,$end)
statestack.append(0)
sym = YaccSymbol()
sym.type = '$end'
symstack.append(sym)
state = 0
while True:
# Get the next symbol on the input. If a lookahead symbol
# is already set, we just use that. Otherwise, we'll pull
# the next token off of the lookaheadstack or from the lexer
#--! DEBUG
debug.debug('')
debug.debug('State : %s', state)
#--! DEBUG
if state not in defaulted_states:
if not lookahead:
if not lookaheadstack:
lookahead = get_token() # Get the next token
else:
lookahead = lookaheadstack.pop()
if not lookahead:
lookahead = YaccSymbol()
lookahead.type = '$end'
# Check the action table
ltype = lookahead.type
t = actions[state].get(ltype)
else:
t = defaulted_states[state]
#--! DEBUG
debug.debug('Defaulted state %s: Reduce using %d', state, -t)
#--! DEBUG
#--! DEBUG
debug.debug('Stack : %s',
('%s . %s' % (' '.join([xx.type for xx in symstack][1:]), str(lookahead))).lstrip())
#--! DEBUG
if t is not None:
if t > 0:
# shift a symbol on the stack
statestack.append(t)
state = t
#--! DEBUG
debug.debug('Action : Shift and goto state %s', t)
#--! DEBUG
symstack.append(lookahead)
lookahead = None
# Decrease error count on successful shift
if errorcount:
errorcount -= 1
continue
if t < 0:
# reduce a symbol on the stack, emit a production
p = prod[-t]
pname = p.name
plen = p.len
# Get production function
sym = YaccSymbol()
sym.type = pname # Production name
sym.value = None
#--! DEBUG
if plen:
debug.info('Action : Reduce rule [%s] with %s and goto state %d', p.str,
'['+','.join([format_stack_entry(_v.value) for _v in symstack[-plen:]])+']',
goto[statestack[-1-plen]][pname])
else:
debug.info('Action : Reduce rule [%s] with %s and goto state %d', p.str, [],
goto[statestack[-1]][pname])
#--! DEBUG
if plen:
targ = symstack[-plen-1:]
targ[0] = sym
#--! TRACKING
if tracking:
t1 = targ[1]
sym.lineno = t1.lineno
sym.lexpos = t1.lexpos
t1 = targ[-1]
sym.endlineno = getattr(t1, 'endlineno', t1.lineno)
sym.endlexpos = getattr(t1, 'endlexpos', t1.lexpos)
#--! TRACKING
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# The code enclosed in this section is duplicated
# below as a performance optimization. Make sure
# changes get made in both locations.
pslice.slice = targ
try:
# Call the grammar rule with our special slice object
del symstack[-plen:]
self.state = state
p.callable(pslice)
del statestack[-plen:]
#--! DEBUG
debug.info('Result : %s', format_result(pslice[0]))
#--! DEBUG
symstack.append(sym)
state = goto[statestack[-1]][pname]
statestack.append(state)
except SyntaxError:
# If an error was set. Enter error recovery state
lookaheadstack.append(lookahead) # Save the current lookahead token
symstack.extend(targ[1:-1]) # Put the production slice back on the stack
statestack.pop() # Pop back one state (before the reduce)
state = statestack[-1]
sym.type = 'error'
sym.value = 'error'
lookahead = sym
errorcount = error_count
self.errorok = False
continue
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
else:
#--! TRACKING
if tracking:
sym.lineno = lexer.lineno
sym.lexpos = lexer.lexpos
#--! TRACKING
targ = [sym]
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# The code enclosed in this section is duplicated
# above as a performance optimization. Make sure
# changes get made in both locations.
pslice.slice = targ
try:
# Call the grammar rule with our special slice object
self.state = state
p.callable(pslice)
#--! DEBUG
debug.info('Result : %s', format_result(pslice[0]))
#--! DEBUG
symstack.append(sym)
state = goto[statestack[-1]][pname]
statestack.append(state)
except SyntaxError:
# If an error was set. Enter error recovery state
lookaheadstack.append(lookahead) # Save the current lookahead token
statestack.pop() # Pop back one state (before the reduce)
state = statestack[-1]
sym.type = 'error'
sym.value = 'error'
lookahead = sym
errorcount = error_count
self.errorok = False
continue
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
if t == 0:
n = symstack[-1]
result = getattr(n, 'value', None)
#--! DEBUG
debug.info('Done : Returning %s', format_result(result))
debug.info('PLY: PARSE DEBUG END')
#--! DEBUG
return result
if t is None:
#--! DEBUG
debug.error('Error : %s',
('%s . %s' % (' '.join([xx.type for xx in symstack][1:]), str(lookahead))).lstrip())
#--! DEBUG
# We have some kind of parsing error here. To handle
# this, we are going to push the current token onto
# the tokenstack and replace it with an 'error' token.
# If there are any synchronization rules, they may
# catch it.
#
# In addition to pushing the error token, we call call
# the user defined p_error() function if this is the
# first syntax error. This function is only called if
# errorcount == 0.
if errorcount == 0 or self.errorok:
errorcount = error_count
self.errorok = False
errtoken = lookahead
if errtoken.type == '$end':
errtoken = None # End of file!
if self.errorfunc:
if errtoken and not hasattr(errtoken, 'lexer'):
errtoken.lexer = lexer
self.state = state
tok = call_errorfunc(self.errorfunc, errtoken, self)
if self.errorok:
# User must have done some kind of panic
# mode recovery on their own. The
# returned token is the next lookahead
lookahead = tok
errtoken = None
continue
else:
if errtoken:
if hasattr(errtoken, 'lineno'):
lineno = lookahead.lineno
else:
lineno = 0
if lineno:
sys.stderr.write('yacc: Syntax error at line %d, token=%s\n' % (lineno, errtoken.type))
else:
sys.stderr.write('yacc: Syntax error, token=%s' % errtoken.type)
else:
sys.stderr.write('yacc: Parse error in input. EOF\n')
return
else:
errorcount = error_count
# case 1: the statestack only has 1 entry on it. If we're in this state, the
# entire parse has been rolled back and we're completely hosed. The token is
# discarded and we just keep going.
if len(statestack) <= 1 and lookahead.type != '$end':
lookahead = None
errtoken = None
state = 0
# Nuke the pushback stack
del lookaheadstack[:]
continue
# case 2: the statestack has a couple of entries on it, but we're
# at the end of the file. nuke the top entry and generate an error token
# Start nuking entries on the stack
if lookahead.type == '$end':
# Whoa. We're really hosed here. Bail out
return
if lookahead.type != 'error':
sym = symstack[-1]
if sym.type == 'error':
# Hmmm. Error is on top of stack, we'll just nuke input
# symbol and continue
#--! TRACKING
if tracking:
sym.endlineno = getattr(lookahead, 'lineno', sym.lineno)
sym.endlexpos = getattr(lookahead, 'lexpos', sym.lexpos)
#--! TRACKING
lookahead = None
continue
# Create the error symbol for the first time and make it the new lookahead symbol
t = YaccSymbol()
t.type = 'error'
if hasattr(lookahead, 'lineno'):
t.lineno = t.endlineno = lookahead.lineno
if hasattr(lookahead, 'lexpos'):
t.lexpos = t.endlexpos = lookahead.lexpos
t.value = lookahead
lookaheadstack.append(lookahead)
lookahead = t
else:
sym = symstack.pop()
#--! TRACKING
if tracking:
lookahead.lineno = sym.lineno
lookahead.lexpos = sym.lexpos
#--! TRACKING
statestack.pop()
state = statestack[-1]
continue
# Call an error function here
raise RuntimeError('yacc: internal parser error!!!\n')
#--! parsedebug-end
Minimal test - lines (1474, 1507)
path: .spaces[17].spaces[0].metrics.mi.mi_sei
old: 90.99927297875892
new: 91.64046290596512
path: .spaces[17].spaces[0].metrics.loc.cloc
old: 18.0
new: 19.0
path: .spaces[17].spaces[0].metrics.loc.blank
old: 2.0
new: 1.0
Code
def __init__(self, terminals):
self.Productions = [None] # A list of all of the productions. The first
# entry is always reserved for the purpose of
# building an augmented grammar
self.Prodnames = {} # A dictionary mapping the names of nonterminals to a list of all
# productions of that nonterminal.
self.Prodmap = {} # A dictionary that is only used to detect duplicate
# productions.
self.Terminals = {} # A dictionary mapping the names of terminal symbols to a
# list of the rules where they are used.
for term in terminals:
self.Terminals[term] = []
self.Terminals['error'] = []
self.Nonterminals = {} # A dictionary mapping names of nonterminals to a list
# of rule numbers where they are used.
self.First = {} # A dictionary of precomputed FIRST(x) symbols
self.Follow = {} # A dictionary of precomputed FOLLOW(x) symbols
self.Precedence = {} # Precedence rules for each terminal. Contains tuples of the
# form ('right',level) or ('nonassoc', level) or ('left',level)
self.UsedPrecedence = set() # Precedence rules that were actually used by the grammer.
# This is only used to provide error checking and to generate
# a warning about unused precedence rules.
self.Start = None # Starting symbol for the grammar
Minimal test - lines (1001, 1273)
path: .spaces[11].spaces[8].metrics.mi.mi_visual_studio
old: 16.54961214899427
new: 16.479952392418138
path: .spaces[11].spaces[8].metrics.mi.mi_sei
old: 5.076770954215533
new: 4.98727741341402
path: .spaces[11].spaces[8].metrics.mi.mi_original
old: 28.299836774780204
new: 28.180718591035017
path: .spaces[11].spaces[8].metrics.loc.sloc
old: 271.0
new: 273.0
path: .spaces[11].spaces[8].metrics.loc.cloc
old: 75.0
new: 76.0
path: .spaces[11].spaces[8].metrics.loc.blank
old: 29.0
new: 30.0
Code
def parseopt_notrack(self, input=None, lexer=None, debug=False, tracking=False, tokenfunc=None):
#--! parseopt-notrack-start
lookahead = None # Current lookahead symbol
lookaheadstack = [] # Stack of lookahead symbols
actions = self.action # Local reference to action table (to avoid lookup on self.)
goto = self.goto # Local reference to goto table (to avoid lookup on self.)
prod = self.productions # Local reference to production list (to avoid lookup on self.)
defaulted_states = self.defaulted_states # Local reference to defaulted states
pslice = YaccProduction(None) # Production object passed to grammar rules
errorcount = 0 # Used during error recovery
# If no lexer was given, we will try to use the lex module
if not lexer:
from . import lex
lexer = lex.lexer
# Set up the lexer and parser objects on pslice
pslice.lexer = lexer
pslice.parser = self
# If input was supplied, pass to lexer
if input is not None:
lexer.input(input)
if tokenfunc is None:
# Tokenize function
get_token = lexer.token
else:
get_token = tokenfunc
# Set the parser() token method (sometimes used in error recovery)
self.token = get_token
# Set up the state and symbol stacks
statestack = [] # Stack of parsing states
self.statestack = statestack
symstack = [] # Stack of grammar symbols
self.symstack = symstack
pslice.stack = symstack # Put in the production
errtoken = None # Err token
# The start state is assumed to be (0,$end)
statestack.append(0)
sym = YaccSymbol()
sym.type = '$end'
symstack.append(sym)
state = 0
while True:
# Get the next symbol on the input. If a lookahead symbol
# is already set, we just use that. Otherwise, we'll pull
# the next token off of the lookaheadstack or from the lexer
if state not in defaulted_states:
if not lookahead:
if not lookaheadstack:
lookahead = get_token() # Get the next token
else:
lookahead = lookaheadstack.pop()
if not lookahead:
lookahead = YaccSymbol()
lookahead.type = '$end'
# Check the action table
ltype = lookahead.type
t = actions[state].get(ltype)
else:
t = defaulted_states[state]
if t is not None:
if t > 0:
# shift a symbol on the stack
statestack.append(t)
state = t
symstack.append(lookahead)
lookahead = None
# Decrease error count on successful shift
if errorcount:
errorcount -= 1
continue
if t < 0:
# reduce a symbol on the stack, emit a production
p = prod[-t]
pname = p.name
plen = p.len
# Get production function
sym = YaccSymbol()
sym.type = pname # Production name
sym.value = None
if plen:
targ = symstack[-plen-1:]
targ[0] = sym
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# The code enclosed in this section is duplicated
# below as a performance optimization. Make sure
# changes get made in both locations.
pslice.slice = targ
try:
# Call the grammar rule with our special slice object
del symstack[-plen:]
self.state = state
p.callable(pslice)
del statestack[-plen:]
symstack.append(sym)
state = goto[statestack[-1]][pname]
statestack.append(state)
except SyntaxError:
# If an error was set. Enter error recovery state
lookaheadstack.append(lookahead) # Save the current lookahead token
symstack.extend(targ[1:-1]) # Put the production slice back on the stack
statestack.pop() # Pop back one state (before the reduce)
state = statestack[-1]
sym.type = 'error'
sym.value = 'error'
lookahead = sym
errorcount = error_count
self.errorok = False
continue
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
else:
targ = [sym]
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# The code enclosed in this section is duplicated
# above as a performance optimization. Make sure
# changes get made in both locations.
pslice.slice = targ
try:
# Call the grammar rule with our special slice object
self.state = state
p.callable(pslice)
symstack.append(sym)
state = goto[statestack[-1]][pname]
statestack.append(state)
except SyntaxError:
# If an error was set. Enter error recovery state
lookaheadstack.append(lookahead) # Save the current lookahead token
statestack.pop() # Pop back one state (before the reduce)
state = statestack[-1]
sym.type = 'error'
sym.value = 'error'
lookahead = sym
errorcount = error_count
self.errorok = False
continue
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
if t == 0:
n = symstack[-1]
result = getattr(n, 'value', None)
return result
if t is None:
# We have some kind of parsing error here. To handle
# this, we are going to push the current token onto
# the tokenstack and replace it with an 'error' token.
# If there are any synchronization rules, they may
# catch it.
#
# In addition to pushing the error token, we call call
# the user defined p_error() function if this is the
# first syntax error. This function is only called if
# errorcount == 0.
if errorcount == 0 or self.errorok:
errorcount = error_count
self.errorok = False
errtoken = lookahead
if errtoken.type == '$end':
errtoken = None # End of file!
if self.errorfunc:
if errtoken and not hasattr(errtoken, 'lexer'):
errtoken.lexer = lexer
self.state = state
tok = call_errorfunc(self.errorfunc, errtoken, self)
if self.errorok:
# User must have done some kind of panic
# mode recovery on their own. The
# returned token is the next lookahead
lookahead = tok
errtoken = None
continue
else:
if errtoken:
if hasattr(errtoken, 'lineno'):
lineno = lookahead.lineno
else:
lineno = 0
if lineno:
sys.stderr.write('yacc: Syntax error at line %d, token=%s\n' % (lineno, errtoken.type))
else:
sys.stderr.write('yacc: Syntax error, token=%s' % errtoken.type)
else:
sys.stderr.write('yacc: Parse error in input. EOF\n')
return
else:
errorcount = error_count
# case 1: the statestack only has 1 entry on it. If we're in this state, the
# entire parse has been rolled back and we're completely hosed. The token is
# discarded and we just keep going.
if len(statestack) <= 1 and lookahead.type != '$end':
lookahead = None
errtoken = None
state = 0
# Nuke the pushback stack
del lookaheadstack[:]
continue
# case 2: the statestack has a couple of entries on it, but we're
# at the end of the file. nuke the top entry and generate an error token
# Start nuking entries on the stack
if lookahead.type == '$end':
# Whoa. We're really hosed here. Bail out
return
if lookahead.type != 'error':
sym = symstack[-1]
if sym.type == 'error':
# Hmmm. Error is on top of stack, we'll just nuke input
# symbol and continue
lookahead = None
continue
# Create the error symbol for the first time and make it the new lookahead symbol
t = YaccSymbol()
t.type = 'error'
if hasattr(lookahead, 'lineno'):
t.lineno = t.endlineno = lookahead.lineno
if hasattr(lookahead, 'lexpos'):
t.lexpos = t.endlexpos = lookahead.lexpos
t.value = lookahead
lookaheadstack.append(lookahead)
lookahead = t
else:
sym = symstack.pop()
statestack.pop()
state = statestack[-1]
continue
# Call an error function here
raise RuntimeError('yacc: internal parser error!!!\n')
#--! parseopt-notrack-end