def lexemes_from_file(filename):
line_num = 0
lexeme_file = open(filename, "r")
lexemes = []
words = {}
curr_cat = None
for line in lexeme_file:
if not line.strip():
continue
l = line.split()
line_num += 1
if l[0] == "--":
curr_cat = l[1]
else:
try:
if line.strip() not in words.keys():
words[line.strip()] = []
words[line.strip()].append(curr_cat)
except:
raise BadFileError("Lexeme file is invalid (line %d) '%s'" % (line_num, line))
for word, categories in words.items():
lexemes.append(lexeme(word, categories))
return lexemes
curr_cat = l[1]
else:
try:
if line.strip() not in words.keys():
words[line.strip()] = []
words[line.strip()].append(curr_cat)
except:
raise BadFileError("Lexeme file is invalid (line %d) '%s'" % (line_num, line))
for word, categories in words.items():
lexemes.append(lexeme(word, categories))
return lexemes
if __name__ == "__main__":
lexemes = [
lexeme("a", ["Art"]),
lexeme("the", ["Art"]),
lexeme("child", ["N"]),
lexeme("man", ["N"]),
lexeme("laughs", ["V"]),
lexeme("eats", ["V"]),
]
lex = lexicon(lexemes)
print lex.by_lex
print lex.by_cat
