logging.info("Initializing lexicon.")
lexicon = DictLexicon(input_path)
lexicon.load()
logging.info("Opening index.")
index = InvertedIndex(input_path)
index.open()
logging.info("Initializing searcher.")
searcher = Searcher(index, "sentence_id")
logging.info("Initializing storage.")
storage = LdbStorage(input_path)
storage.open_db()
if context_input is not None:
logging.info("Initializing context lexicon.")
c_lexicon = DictLexicon(context_input)
c_lexicon.load()
logging.info("Opening context index.")
c_index = InvertedIndex(context_input)
c_index.open()
logging.info("Initializing context searcher.")
if arguments.language == "rus":
c_searcher = Searcher(c_index, "ruwac_document_id")
logging.info("Output: %s" % output_path)
if os.path.exists(output_path):
shutil.rmtree(output_path)
os.makedirs(output_path)
logging.info("Initializing lexicon.")
lexicon = DictLexicon(output_path)
lexicon.load()
logging.info("Initializing storage.")
storage = LdbStorage(output_path)
storage.init_db()
storage.open_db()
logging.info("Initializing index.")
index = InvertedIndex(output_path, field_properties=[
("document_id", numpy.int32),
])
index.init_index()
index.open()
logging.info("Initializing ruwac stream and its parser.")
if arguments.language == "spa" or arguments.language == "eng":
input_path = arguments.input
output_path = arguments.output
logging.info("Input: %s" % input_path)
logging.info("Output: %s" % output_path)
if os.path.exists(output_path):
shutil.rmtree(output_path)
os.makedirs(output_path)
logging.info("Initializing lexicon.")
lexicon = DictLexicon(output_path)
logging.info("Initializing storage.")
storage = LdbStorage(output_path)
storage.init_db()
storage.open_db()
logging.info("Initializing index.")
index = InvertedIndex(output_path,
field_properties=[
("sentence_id", numpy.int32),
])
index.init_index()
index.open()
logging.info("Initializing sentence stream and its parser.")
sentence_stream = LFSentenceStream(input_path, language=arguments.language)
sentence_parser = LFSentenceParser()
sentence_indexer = LFSentenceIndexer(lexicon)
logging.info("Query: %s" % query_paths)
logging.info("Output: %s" % output_path)
logging.info("Initializing lexicon.")
lexicon = DictLexicon(input_path)
lexicon.load()
logging.info("Opening index.")
index = InvertedIndex(input_path)
index.open()
logging.info("Initializing searcher.")
searcher = Searcher(index, "sentence_id")
logging.info("Initializing storage.")
storage = LdbStorage(input_path)
storage.open_db()
if context_input is not None:
logging.info("Initializing context lexicon.")
c_lexicon = DictLexicon(context_input)
c_lexicon.load()
logging.info("Opening context index.")
c_index = InvertedIndex(context_input)
c_index.open()
logging.info("Initializing context searcher.")
if arguments.language == "rus":
c_searcher = Searcher(c_index, "ruwac_document_id")
