def do(sh):
"""Transforms an expression of the form [a-z0-9A-Z] into a NumberSet of
code points that corresponds to the characters and character ranges mentioned.
"""
assert sh.__class__.__name__ == "StringIO" \
or sh.__class__.__name__ == "file"
def __check_letter(stream, letter):
position = stream.tell()
if stream.read(1) == letter: return True
else:
stream.seek(position)
return False
# check, if the set is thought to be inverse (preceeded by '^')
tracker = Tracker()
if __check_letter(sh, "^"): tracker.negation_f = True
char_code = None
while char_code != 0xFF:
char_code = utf8.__read_one_utf8_code_from_stream(sh)
if char_code == ord("-"):
raise RegularExpressionException(
"Character range operator '-' requires a preceding character as in 'a-z'."
)
elif char_code == 0xFF:
raise RegularExpressionException(
"Missing closing ']' in character range expression.")
elif char_code == ord("]"):
break
elif char_code == ord("\\"):
char_code = snap_backslashed_character.do(sh)
if not __check_letter(sh, "-"):
# (*) Normal character
tracker.consider_letter(char_code)
else:
# (*) Character range: 'character0' '-' 'character1'
char_code_2 = utf8.__read_one_utf8_code_from_stream(sh)
if char_code_2 in [0xFF, ord(']')]:
raise RegularExpressionException(
"Character range: '-' requires a character following '-'.")
elif char_code == ord("-"):
raise RegularExpressionException(
"Character range operator '-' followed by '-'.")
elif char_code_2 == ord("\\"):
char_code_2 = snap_backslashed_character.do(sh)
# value denotes 'end', i.e first character outside the interval => add 1
tracker.consider_interval(char_code, char_code_2 + 1)
if tracker.negation_f: return tracker.match_set.inverse()
else: return tracker.match_set
def do(sh):
"""Transforms an expression of the form [a-z0-9A-Z] into a NumberSet of
code points that corresponds to the characters and character ranges mentioned.
"""
assert sh.__class__.__name__ == "StringIO" \
or sh.__class__.__name__ == "file"
def __check_letter(stream, letter):
position = stream.tell()
if stream.read(1) == letter: return True
else: stream.seek(position); return False
# check, if the set is thought to be inverse (preceeded by '^')
tracker = Tracker()
if __check_letter(sh, "^"): tracker.negation_f = True
char_code = None
while char_code != 0xFF:
char_code = utf8.__read_one_utf8_code_from_stream(sh)
if char_code == ord("-"):
raise RegularExpressionException("Character range operator '-' requires a preceding character as in 'a-z'.")
elif char_code == 0xFF:
raise RegularExpressionException("Missing closing ']' in character range expression.")
elif char_code == ord("]"):
break
elif char_code == ord("\\"):
char_code = snap_backslashed_character.do(sh)
if not __check_letter(sh, "-"):
# (*) Normal character
tracker.consider_letter(char_code)
else:
# (*) Character range: 'character0' '-' 'character1'
char_code_2 = utf8.__read_one_utf8_code_from_stream(sh)
if char_code_2 in [0xFF, ord(']')]:
raise RegularExpressionException("Character range: '-' requires a character following '-'.")
elif char_code == ord("-"):
raise RegularExpressionException("Character range operator '-' followed by '-'.")
elif char_code_2 == ord("\\"):
char_code_2 = snap_backslashed_character.do(sh)
# value denotes 'end', i.e first character outside the interval => add 1
tracker.consider_interval(char_code, char_code_2 + 1)
if tracker.negation_f: return tracker.match_set.inverse()
else: return tracker.match_set
def get_character_code_sequence(sh):
assert sh.__class__.__name__ == "StringIO" \
or sh.__class__.__name__ == "file"
# Only \" is a special character '"', any other backslashed character
# remains as the sequence 'backslash' + character
sequence = []
while 1 + 1 == 2:
char_code = utf8.__read_one_utf8_code_from_stream(sh)
if char_code == 0xFF:
raise RegularExpressionException(
"End of file reached while parsing quoted string.")
elif char_code == ord("\\"):
char_code = snap_backslashed_character.do(sh)
if char_code == None:
raise RegularExpressionException(
"Unidentified backslash-sequence in quoted string.")
elif char_code == ord('"'):
break
sequence.append(char_code)
return sequence
def read_character_code(fh):
# NOTE: This function is tested with the regeression test for feature request 2251359.
# See directory $QUEX_PATH/TEST/2251359.
pos = fh.tell()
start = fh.read(1)
if start == "":
seek(pos); return -1
elif start == "'":
# read an utf-8 char an get the token-id
# Example: '+'
if check(fh, "\\"):
# snap_backslashed_character throws an exception if 'backslashed char' is nonsense.
character_code = snap_backslashed_character.do(fh, ReducedSetOfBackslashedCharactersF=True)
else:
character_code = __read_one_utf8_code_from_stream(fh)
if character_code == 0xFF:
error_msg("Missing utf8-character for definition of character code by character.", fh)
elif fh.read(1) != '\'':
error_msg("Missing closing ' for definition of character code by character.", fh)
return character_code
if start == "U":
if fh.read(1) != "C": seek(pos); return -1
# read Unicode Name
# Example: UC MATHEMATICAL_MONOSPACE_DIGIT_FIVE
skip_whitespace(fh)
ucs_name = __read_token_identifier(fh)
if ucs_name == "": seek(pos); return -1
# Get the character set related to the given name. Note, the size of the set
# is supposed to be one.
character_code = ucs_property_db.get_character_set("Name", ucs_name)
if type(character_code) in [str, unicode]:
verify_word_in_list(ucs_name, ucs_property_db["Name"].code_point_db,
"The string %s\ndoes not identify a known unicode character." % ucs_name,
fh)
elif type(character_code) not in [int, long]:
error_msg("%s relates to more than one character in unicode database." % ucs_name, fh)
return character_code
fh.seek(pos)
character_code = read_integer(fh)
if character_code != None: return character_code
# Try to interpret it as something else ...
fh.seek(pos)
return -1
def get_character_code_sequence(sh):
assert sh.__class__.__name__ == "StringIO" \
or sh.__class__.__name__ == "file"
# Only \" is a special character '"', any other backslashed character
# remains as the sequence 'backslash' + character
sequence = []
while 1 + 1 == 2:
char_code = utf8.__read_one_utf8_code_from_stream(sh)
if char_code == 0xFF:
raise RegularExpressionException("End of file reached while parsing quoted string.")
elif char_code == ord("\\"):
char_code = snap_backslashed_character.do(sh)
if char_code == None:
raise RegularExpressionException("Unidentified backslash-sequence in quoted string.")
elif char_code == ord('"'):
break
sequence.append(char_code)
return sequence
def snap_primary(stream, PatternDict):
"""primary: " non_double_quote * " = character string
[ non_rect_bracket_close ] = set of characters
{ identifier } = pattern replacement
( expression )
non_control_character+ = lonely characters
primary repetition_cmd
"""
__debug_entry("primary", stream)
x = stream.read(1)
lookahead = stream.read(1)
if x != "" and lookahead != "": stream.seek(-1, 1)
if x == "": return __debug_exit(None, stream)
# -- 'primary' primary
if x == "\"": result = snap_character_string.do(stream)
elif x == "[":
stream.seek(-1, 1)
result = character_set_expression.do(stream, PatternDict)
elif x == "{":
result = snap_replacement(stream, PatternDict)
elif x == ".":
result = create_ALL_BUT_NEWLINE_state_machine()
elif x == "(":
result = snap_bracketed_expression(stream, PatternDict)
elif x.isspace():
# a lonestanding space ends the regular expression
stream.seek(-1, 1)
return __debug_exit(None, stream)
elif x in ["*", "+", "?"]:
raise RegularExpressionException(
"lonely operator '%s' without expression proceeding." % x)
elif x == "\\":
if lookahead == "C":
stream.read(1)
result = snap_case_folded_pattern(stream, PatternDict)
else:
stream.seek(-1, 1)
trigger_set = character_set_expression.snap_property_set(stream)
if trigger_set == None:
stream.seek(
1, 1) # snap_property_set() leaves tream right before '\\'
char_code = snap_backslashed_character.do(stream)
if char_code == None:
raise RegularExpressionException(
"Backslash followed by unrecognized character code.")
trigger_set = char_code
result = StateMachine()
result.add_transition(result.init_state_index,
trigger_set,
AcceptanceF=True)
elif x not in CONTROL_CHARACTERS:
# NOTE: The '\' is not inside the control characters---for a reason.
# It is used to define for example character codes using '\x' etc.
stream.seek(-1, 1)
result = snap_non_control_character(stream, PatternDict)
else:
# NOTE: This includes the '$' sign which means 'end of line'
# because the '$' sign is in CONTROL_CHARACTERS, but is not checked
# against. Thus, it it good to leave here on '$' because the
# '$' sign is handled on the very top level.
# this is not a valid primary
stream.seek(-1, 1)
return __debug_exit(None, stream)
# -- optional repetition command?
result_repeated = __snap_repetition_range(result, stream)
## print "##imr:", result.get_string(NormalizeF=False)
if result_repeated != None: result = result_repeated
return __debug_exit(construct.beautify(result), stream)
def do(sh):
"""Transforms an expression of the form [a-z0-9A-Z] into a NumberSet of
code points that corresponds to the characters and character ranges mentioned.
"""
assert sh.__class__.__name__ == "StringIO" \
or sh.__class__.__name__ == "file"
def __check_letter(stream, letter):
position = stream.tell()
if stream.read(1) == letter: return True
else: stream.seek(position); return False
# check, if the set is thought to be inverse (preceeded by '^')
tracker = Tracker()
if __check_letter(sh, "^"): tracker.negation_f = True
char_code = None
quote_checker = DoubleQuoteChecker() # Checks for " appearing twice. Some users did use
# # constructs such as "-" and ended up in confusing behavior.
while char_code != 0xFF:
char_code = utf8.__read_one_utf8_code_from_stream(sh)
quote_checker.do(char_code)
if char_code == ord("-"):
raise RegularExpressionException("Character range operator '-' requires a preceding character as in 'a-z'.")
elif char_code == 0xFF:
raise RegularExpressionException("Missing closing ']' in character range expression.")
elif char_code == ord("]"):
break
elif char_code == ord("\\"):
char_code = snap_backslashed_character.do(sh)
if not __check_letter(sh, "-"):
# (*) Normal character
tracker.consider_letter(char_code)
else:
# (*) Character range: 'character0' '-' 'character1'
char_code_2 = utf8.__read_one_utf8_code_from_stream(sh)
quote_checker.do(char_code_2)
if char_code_2 in [0xFF, ord(']')]:
raise RegularExpressionException("Character range: '-' requires a character following '-'.")
elif char_code == ord("-"):
raise RegularExpressionException("Character range operator '-' followed by '-'.")
elif char_code_2 == ord("\\"):
char_code_2 = snap_backslashed_character.do(sh)
# value denotes 'end', i.e first character outside the interval => add 1
if char_code == char_code_2:
utf8_string = utf8.map_unicode_to_utf8(char_code)
raise RegularExpressionException("Character range '%s-%s' has only one element.\n" \
% (utf8_string, utf8_string) + \
"In this case avoid range expression for clarity.")
tracker.consider_interval(char_code, char_code_2 + 1)
if tracker.negation_f:
result = tracker.match_set.inverse()
if Setup.get_character_value_limit() != sys.maxint:
result.intersect_with(Interval(0, Setup.get_character_value_limit()))
return result
else:
return tracker.match_set
def do(sh):
"""Transforms an expression of the form [a-z0-9A-Z] into a NumberSet of
code points that corresponds to the characters and character ranges mentioned.
"""
assert sh.__class__.__name__ == "StringIO" \
or sh.__class__.__name__ == "file"
def __check_letter(stream, letter):
position = stream.tell()
if stream.read(1) == letter: return True
else:
stream.seek(position)
return False
# check, if the set is thought to be inverse (preceeded by '^')
tracker = Tracker()
if __check_letter(sh, "^"): tracker.negation_f = True
char_code = None
quote_checker = DoubleQuoteChecker(
) # Checks for " appearing twice. Some users did use
# # constructs such as "-" and ended up in confusing behavior.
while char_code != 0xFF:
char_code = utf8.__read_one_utf8_code_from_stream(sh)
quote_checker.do(char_code)
if char_code == ord("-"):
raise RegularExpressionException(
"Character range operator '-' requires a preceding character as in 'a-z'."
)
elif char_code == 0xFF:
raise RegularExpressionException(
"Missing closing ']' in character range expression.")
elif char_code == ord("]"):
break
elif char_code == ord("\\"):
char_code = snap_backslashed_character.do(sh)
if not __check_letter(sh, "-"):
# (*) Normal character
tracker.consider_letter(char_code)
else:
# (*) Character range: 'character0' '-' 'character1'
char_code_2 = utf8.__read_one_utf8_code_from_stream(sh)
quote_checker.do(char_code_2)
if char_code_2 in [0xFF, ord(']')]:
raise RegularExpressionException(
"Character range: '-' requires a character following '-'.")
elif char_code == ord("-"):
raise RegularExpressionException(
"Character range operator '-' followed by '-'.")
elif char_code_2 == ord("\\"):
char_code_2 = snap_backslashed_character.do(sh)
# value denotes 'end', i.e first character outside the interval => add 1
if char_code == char_code_2:
utf8_string = utf8.map_unicode_to_utf8(char_code)
raise RegularExpressionException("Character range '%s-%s' has only one element.\n" \
% (utf8_string, utf8_string) + \
"In this case avoid range expression for clarity.")
tracker.consider_interval(char_code, char_code_2 + 1)
if tracker.negation_f:
result = tracker.match_set.inverse()
if Setup.get_character_value_limit() != sys.maxint:
result.intersect_with(
Interval(0, Setup.get_character_value_limit()))
return result
else:
return tracker.match_set
def read_character_code(fh):
# NOTE: This function is tested with the regeression test for feature request 2251359.
# See directory $QUEX_PATH/TEST/2251359.
pos = fh.tell()
start = fh.read(1)
if start == "":
seek(pos)
return -1
elif start == "'":
# read an utf-8 char an get the token-id
# Example: '+'
if check(fh, "\\"):
# snap_backslashed_character throws an exception if 'backslashed char' is nonsense.
character_code = snap_backslashed_character.do(
fh, ReducedSetOfBackslashedCharactersF=True)
else:
character_code = __read_one_utf8_code_from_stream(fh)
if character_code == 0xFF:
error_msg(
"Missing utf8-character for definition of character code by character.",
fh)
elif fh.read(1) != '\'':
error_msg(
"Missing closing ' for definition of character code by character.",
fh)
return character_code
if start == "U":
if fh.read(1) != "C":
seek(pos)
return -1
# read Unicode Name
# Example: UC MATHEMATICAL_MONOSPACE_DIGIT_FIVE
skip_whitespace(fh)
ucs_name = __read_token_identifier(fh)
if ucs_name == "":
seek(pos)
return -1
# Get the character set related to the given name. Note, the size of the set
# is supposed to be one.
character_code = ucs_property_db.get_character_set("Name", ucs_name)
if type(character_code) in [str, unicode]:
verify_word_in_list(
ucs_name, ucs_property_db["Name"].code_point_db,
"The string %s\ndoes not identify a known unicode character." %
ucs_name, fh)
elif type(character_code) not in [int, long]:
error_msg(
"%s relates to more than one character in unicode database." %
ucs_name, fh)
return character_code
fh.seek(pos)
character_code = read_integer(fh)
if character_code != None: return character_code
# Try to interpret it as something else ...
fh.seek(pos)
return -1
def snap_non_control_characters(stream):
"""Snaps any 'non_control_character' using UTF8 encoding from the given string. Note, that
in UTF8 a character may consist of more than one byte. Creates a state machine
that contains solely one trigger for each character to a acceptance state.
This function **concatinates** incoming characters, but **repetition** has preceedence
over concatination, so it checks after each character whether it is followed by
a repetition ('*', '+', '?', '{..}'). In such a case, the repetition of the character
is appended.
"""
__debug_entry("non-control characters", stream)
result = StateMachine()
state_index = result.init_state_index
# (*) read first character
position = stream.tell()
char_code = utf8.__read_one_utf8_code_from_stream(stream)
while char_code != 0xFF:
# (1) check against occurence of control characters
# this needs to come **before** the backslashed character interpretation.
# NOTE: A backslashed character can be a whitespace (for example '\n').
# (check against 0xFF to avoid overflow in function 'chr()')
if char_code < 0xFF \
and (chr(char_code) in CONTROL_CHARACTERS or chr(char_code).isspace()):
stream.seek(-1, 1)
break
# (2) treat backslashed characters
if char_code == ord('\\'):
stream.seek(-1, 1)
trigger_set = character_set_expression.snap_property_set(stream)
if trigger_set == None:
stream.seek(1, 1) # snap_property_set() leaves tream right before '\\'
char_code = snap_backslashed_character.do(stream)
if char_code == None:
raise RegularExpressionException("Backslash followed by unrecognized character code.")
trigger_set = char_code
else:
trigger_set = char_code
# (3) read next character
position = stream.tell()
next_char_code = utf8.__read_one_utf8_code_from_stream(stream)
# -- check for repetition (repetition has preceedence over concatination)
if next_char_code in [ord("+"), ord("*"), ord("?"), ord("{")]:
# (*) create state machine that consist of a single transition
tmp = StateMachine()
tmp.add_transition(tmp.init_state_index, trigger_set, AcceptanceF=True)
# -- repeat the single character state machine
stream.seek(position)
tmp_repeated = __snap_repetition_range(tmp, stream)
# -- append it to the result (last state must be set to acceptance for concatenation)
result.states[state_index].set_acceptance()
result = sequentialize.do([result, tmp_repeated], MountToFirstStateMachineF=True)
# as soon as there is repetition there might be more than one acceptance
# state and thus simple concatination via 'add_transition' fails.
# let us return and check treat the remaining chars
# at the next call to this function.
return __debug_exit(result, stream)
else:
# (*) add new transition from current state to a new state triggering
# on the given character.
state_index = result.add_transition(state_index, trigger_set)
char_code = next_char_code
# last character in the chain triggers an 'acceptance state'
result.states[state_index].set_acceptance()
return __debug_exit(result, stream)
def snap_non_control_characters(stream):
"""Snaps any 'non_control_character' using UTF8 encoding from the given string. Note, that
in UTF8 a character may consist of more than one byte. Creates a state machine
that contains solely one trigger for each character to a acceptance state.
This function **concatinates** incoming characters, but **repetition** has preceedence
over concatination, so it checks after each character whether it is followed by
a repetition ('*', '+', '?', '{..}'). In such a case, the repetition of the character
is appended.
"""
__debug_entry("non-control characters", stream)
result = StateMachine()
state_index = result.init_state_index
# (*) read first character
position = stream.tell()
char_code = utf8.__read_one_utf8_code_from_stream(stream)
while char_code != 0xFF:
# (1) check against occurence of control characters
# this needs to come **before** the backslashed character interpretation.
# NOTE: A backslashed character can be a whitespace (for example '\n').
# (check against 0xFF to avoid overflow in function 'chr()')
if char_code < 0xFF \
and (chr(char_code) in CONTROL_CHARACTERS or chr(char_code).isspace()):
stream.seek(-1, 1)
break
# (2) treat backslashed characters
if char_code == ord('\\'):
stream.seek(-1, 1)
trigger_set = character_set_expression.snap_property_set(stream)
if trigger_set == None:
stream.seek(
1, 1) # snap_property_set() leaves tream right before '\\'
char_code = snap_backslashed_character.do(stream)
if char_code == None:
raise RegularExpressionException(
"Backslash followed by unrecognized character code.")
trigger_set = char_code
else:
trigger_set = char_code
# (3) read next character
position = stream.tell()
next_char_code = utf8.__read_one_utf8_code_from_stream(stream)
# -- check for repetition (repetition has preceedence over concatination)
if next_char_code in [ord("+"), ord("*"), ord("?"), ord("{")]:
# (*) create state machine that consist of a single transition
tmp = StateMachine()
tmp.add_transition(tmp.init_state_index,
trigger_set,
AcceptanceF=True)
# -- repeat the single character state machine
stream.seek(position)
tmp_repeated = __snap_repetition_range(tmp, stream)
# -- append it to the result (last state must be set to acceptance for concatenation)
result.states[state_index].set_acceptance()
result = sequentialize.do([result, tmp_repeated],
MountToFirstStateMachineF=True)
# as soon as there is repetition there might be more than one acceptance
# state and thus simple concatination via 'add_transition' fails.
# let us return and check treat the remaining chars
# at the next call to this function.
return __debug_exit(result, stream)
else:
# (*) add new transition from current state to a new state triggering
# on the given character.
state_index = result.add_transition(state_index, trigger_set)
char_code = next_char_code
# last character in the chain triggers an 'acceptance state'
result.states[state_index].set_acceptance()
return __debug_exit(result, stream)