add thread for directory import

This commit is contained in:
2023-04-28 23:37:13 +02:00
parent 5a4bdbb420
commit 7a1286c4e2
2 changed files with 33 additions and 22 deletions

View File

@@ -48,18 +48,21 @@ class WPimport:
self._logger.debug("{0} : {1}".format(self._name, r.content))
def fromDirectory(self, directory):
def fromDirectory(self, directory="", number_thread=1, max_thread=1):
directory = "{0}/archives".format(directory)
directories = self._getDirectories([], "{0}".format(directory))
files = self._getFiles(directories)
self.fromFile(files)
def fromFile(self, files):
for file in files:
if os.path.exists(file):
self._logger.info("{0} : File is being processed : {1}".format(self._name, file))
with open(file, 'r') as f:
def fromFile(self, files=[], number_thread=1, max_thread=1):
divFiles = int(len(files) / max_thread)
currentRangeFiles = int(divFiles * number_thread)
firstRange = int(currentRangeFiles - divFiles)
for i in range(firstRange, currentRangeFiles):
if os.path.exists(files[i]):
self._logger.info("{0} : File is being processed : {1}".format(self._name, files[i]))
with open(files[i], 'r') as f:
content = f.read()
soup = BeautifulSoup(content, self._parser)
articlebody = soup.find_all("div", class_="articlebody")