def parse_feature_node(s,next_index=0):
next_index = jump_over_space(s,next_index)
start_index = next_index
while True:
# Alnum and '-' are both legal characters inside a node
ch = s[next_index]
if ch.isalnum() or ch == '-' or ch == '_':
next_index += 1
else:
break
feature_node = s[start_index:next_index]
if feature_node == '':
feature_node = None
return (feature_node,next_index)
def check_symbol(s,next_index,symbol):
"""
Check whether we have got a defined symbol in the grammar. Return False if
not, or return the next_index after this symbol.
:param s: The string need to be checked
:type s: str
:param next_index: Index on the string
:type next_index: integer
:param symbol: The symbol to be checked against
:type symbol: str
"""
try:
next_index = jump_over_space(s,next_index)
if s[next_index:next_index + len(symbol)] == symbol:
return next_index + len(symbol) # We must ignore the symbol
except IndexError:
return False
else:
return False
def parse_feature_value(s,next_index=0):
"""
Parse the value of a feature entry. The value can be a plain text string
without any space or new line, or it can be several sub values separated
by '/'. In both case the function will return a list containing each sub
strings.
:param s: The string to be parsed
:type s: str
:return: A tuple containing the value as well as next_index
:rtype: tuple(list(str),integer)
"""
next_index = jump_over_space(s,next_index)
start_index = next_index
while True:
if not s[next_index].isspace():
next_index += 1
else:
break
feature_value = s[start_index:next_index]
if feature_value == '':
feature_value = None
feature_value = feature_value.split('/')
return (feature_value,next_index)
