def main(workers=0): import os import loacore.utils.db as db import loacore.process.file_process as file_process # backup = db.database_backup() # backup.write_backup() db.download_db(db_name="new") file_process.add_files([os.path.join(os.path.dirname(__file__), "test_file.txt")], workers=workers)
def add_all(): import loacore.process.file_process as file_process import os from loacore.conf import DATA_PATH file_paths = [] for dirpath, dirnames, filenames in os.walk( os.path.abspath(os.path.join(DATA_PATH, "raw", "imdb"))): for name in filenames: file_paths.append(os.path.join(dirpath, name)) file_process.add_files(file_paths, encoding="utf8", lang="en")
def process(workers): import loacore.process.file_process as file_process file_paths = [] for dirpath, dirnames, filenames in os.walk( os.path.abspath(os.path.join(loacore.conf.DATA_PATH, "raw", "uci"))): for name in filenames: file_paths.append(os.path.join(dirpath, name)) file_process.add_files(file_paths, encoding="utf8", lang="en", workers=workers)
def add_test_imdb_pos(workers): import loacore.process.file_process as file_process import loacore.load.file_load as file_load from loacore.conf import DATA_PATH import os # file_load.clean_db() file_process.add_files([ os.path.join(DATA_PATH, 'raw', 'imdb', 'test', 'test_imdb_labelled_pos.txt') ], encoding='utf8', lang='en', workers=workers)
def add_files_to_database(): from loacore.conf import DATA_PATH file_paths = [] # for dirpath, dirnames, filenames in os.walk(os.path.join(DATA_PATH, 'raw', 'TempAlta')): # for name in filenames: # print(name) # file_paths.append(os.path.join(dirpath, name)) # # for dirpath, dirnames, filenames in os.walk(os.path.join(DATA_PATH, 'raw', 'TempBaja')): # for name in filenames: # print(name) # file_paths.append(os.path.join(dirpath, name)) # # file_process.add_files(file_paths, lang='es') # file_paths.clear() file_load.clean_db() for dirpath, dirnames, filenames in os.walk( os.path.abspath(os.path.join(DATA_PATH, 'raw', 'uci'))): for name in filenames: print(name) file_paths.append(os.path.join(dirpath, name)) file_process.add_files(file_paths, encoding='utf8', lang='en') file_paths.clear() for dirpath, dirnames, filenames in os.walk( os.path.abspath(os.path.join(DATA_PATH, 'raw', 'corrected'))): for name in filenames: print(name) file_paths.append(os.path.join(dirpath, name)) file_process.add_files(file_paths, encoding='UTF-16LE', lang='es') file_paths.clear()
from loacore.process.file_process import add_files from loacore.conf import DATA_PATH import os print(DATA_PATH) print(os.path.join(DATA_PATH, "raw/imdb/test/test_imdb_labelled_neg.txt")) add_files([os.path.join(DATA_PATH, "raw/imdb/test/test_imdb_labelled_neg.txt")], workers=2)