def save_file(self, text=None):
self.log.info(f"Saving '{self.name}' from '{self.url}'")
self.original_path = f"{DOCS_TO_PROCESS_PATH}{self.extension}/{self.name}"
self.txt_path = f"{DOCS_TO_PROCESS_PATH}{self.name}.txt"
create_file(self.original_path)
with open(self.original_path, 'wb') as f:
f.write(self.request.content)
def file_to_text(self):
provisory_file_name = f"/tmp/processor-provisory/{self.name}.txt"
create_file(provisory_file_name)
bashCommand = f"pdftotext -layout {self.original_path} {provisory_file_name}"
os.system(bashCommand)
with open(provisory_file_name, 'r') as f:
return f.read()
return ''
def __init__(self, path=None):
if path:
self.logger_path = path
create_file(self.logger_path)
logging.basicConfig(
filename=self.logger_path,
format='[%(levelname)s]: %(asctime)s - %(message)s',
level=logging.INFO)
def save_text(filename, text, probability):
"""
if the probability for the most probable language is higher than 40%
then save the file to that language folder
"""
if probability.items():
most_probable = max(probability.items(), key=lambda k: k[1])
language_slug = most_probable[0]
if probability_criteria_check(most_probable[1]):
file_path = f"{PROBABLE_DOCS_PER_LANGUAGE_PATH}{language_slug}/{filename}"
create_file(file_path)
with open(file_path, 'w') as f:
log.info(f"Saving text in '{language_slug}' probable docs folder as '{filename}'")
f.write(text)
return 1
return 0
def process_language_bootstrap_files(language):
base_path = f"{BOOTSTRAP_PATH}{language['slug']}/"
files = glob.glob(os.path.join(base_path, '*.txt'))
log.info(f"Boostraping '{language['slug']}' language with {files}")
if files:
for filename in files:
with open(filename, 'r') as f:
log.info(
f"Adding '{os.path.basename(filename)}' words to '{language['slug']}' language assets"
)
add_to_language_assets(language['slug'], f.read())
folder_path, name = os.path.split(filename)
new_path = f"{folder_path}/processed/{name}"
create_file(new_path)
os.rename(filename, new_path)
else:
log.warning(
f"There is no files to add to '{language['slug']}' language assets"
)
def save_text_file(self, text):
self.log.info(f"Converting '{self.name}' to '{self.name}.txt'")
create_file(self.txt_path)
with open(self.txt_path, 'w') as f:
text = f"{self.url}\n\n\n\n{text}"
f.write(text)
def _create_language_assets(self, path):
create_folder(path)
create_file(f"{path}dictionary.json", '{}')
create_file(f"{path}words_list")
def mark_file_as_processed(filename):
log.info(f"Marking '{os.path.basename(filename)}' as processed")
new_path = f"{DOCS_PROCESSED_PATH}{os.path.basename(filename)}"
create_file(new_path)
os.rename(filename, new_path)
