def _load_list(self):
self._list = []
filename = common.get_list_file_name()
path = os.path.join(self._path_dir, filename)
with open(path) as f:
write = False #no doc has been read yet: there's nothing to write
while True:
line = f.readline()
if line == '':
self._add_to_list(release_id, date, title,
url) #flush the last doc
break
if line.startswith('LOTW-ID: '): #found new doc
if write: #write if there is a previous doc to write
self._add_to_list(release_id, date, title, url)
release_id = line[9:].strip()
write = True
elif line.startswith('LOTW-DATE: '):
date = line[11:].strip()
elif line.startswith('LOTW-TITLE: '):
title = line[12:].strip()
elif line.startswith('LOTW-URL: '):
url = line[10:].strip()
def _load(self, company_id):
cfr = ConfigReader()
p1 = cfr.get('ROOT_ORIGINAL')
p2 = cfr.get('FORMATTED_PR')
p3 = os.path.join(p1, p2)
metafile = common.get_list_file_name(company_id)
path_text = os.path.join(p3, str(company_id))
path_meta = os.path.join(p3, metafile)
pr_text = self._load_text(path_text)
self._load_meta(path_meta, pr_text)
def _write_files(self):
list_file = common.get_list_file_name(self._company_id)
path_list = os.path.join(self._output_path, list_file)
path_text = os.path.join(self._output_path, self._company_id)
dc = DataCleaner('utf-8')
clean_list = dc.clean(''.join(self._sb_list))
clean_text = dc.clean(''.join(self._sb_text))
with open(path_list, 'w') as f:
f.write(clean_list)
with open(path_text, 'w') as f:
f.write(clean_text)
def _write_files(self, sb_list, sb_text):
list_file = common.get_list_file_name(self._company_id)
path_list = os.path.join(self._output_path, list_file) #refactor
path_text = os.path.join(self._output_path, self._company_id)
clean_list = self._dc.clean(''.join(sb_list))
clean_text = ''.join(sb_text) #this has been cleaned in _format_body()
with open(path_list, 'w') as f:
f.write(clean_list)
with open(path_text, 'w') as f:
f.write(clean_text)
def _write_list(self):
path = os.path.join(self._path_dir, common.get_list_file_name())
with open(path, 'w') as f:
f.write(''.join(self._sb))