def get_language_words(search, language):
# Retourne la liste des mots candidat correspondant au langage
# Les mots sont stockés dans un fichier de ce type : ./language_words/[language].words
language_file = path.join(epp.getcwd(), "language_words", language + ".words")
if path.exists(language_file):
epp.log("language file Ok : %s" % language_file)
candidates = Candidate.Candidates()
r1 = get_regex(search)
# Parcours du fichier du langage pour trouver les candidats
with open(language_file, "r") as fileh:
for match in [r1.findall(l) for l in fileh]:
candidates.extend(match, is_language_word=True)
return set(candidates.get_list())
else:
return set()
def get_language(forced_language, filetype):
if forced_language != None:
epp.log("Use options.language: {0}".format(forced_language))
return forced_language
else:
epp.log("Scan filetypes for extensions {0}".format(filetype))
language = None
config_file = path.join(epp.getcwd(), "language_words", "filetypes.txt")
if filetype != '' and path.exists(config_file):
r = re.compile(r"^(?:\w+;)*{0}\S*\t(?P<language>\w+)".format(filetype), re.IGNORECASE)
with open(config_file, "r") as f:
for l in f:
match = r.search(l)
if match:
language = match.group("language")
epp.log("language found: {0}".format(language))
break
return language