v4l3n71n/web_scrap
1
0
Fork 0
Code Issues Pull Requests Projects Releases 10 Wiki Activity

fix file tmp for directory

Browse Source
This commit is contained in:
Valentin CZERYBA 2023-06-28 23:03:27 +02:00
parent 55d62cebfb
commit 9f87f38347
2 changed files with 23 additions and 18 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

31
import_export_canalblog.py
View File

@ -11,6 +11,18 @@ from lib.WPExport import WPExport
from lib.WPRemove import WPRemove from lib.WPRemove import WPRemove
from lib.WPChange import WPChange from lib.WPChange import WPChange
def errorRevert(logger, revert, tmp):
if revert is True:
files_tmp = glob.glob("{0}/*.json".format(tmp))
if len(files_tmp) == 0:
logger.error("Error revert, because files not found")
exit(1)
if len(files_tmp) != int(args.parallel):
for file_r in files_tmp:
os.remove(file_r)
logger.error("Error revert, because number files tmp is incompatible with parallel number")
exit(1)
def change(index, number, args, logger): def change(index, number, args, logger):
changeWp = WPChange(logger=logger, index_name=index, number_thread=number) changeWp = WPChange(logger=logger, index_name=index, number_thread=number)
changeWp.fromDirectory(args.directory) changeWp.fromDirectory(args.directory)
@ -106,7 +118,7 @@ def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, bas
del importWp del importWp
def importDirectory(name_thread, max_thread, directory, logger, parser, wordpress, basic, serial, ssl_wordpress, create, update, image): def importDirectory(name_thread, max_thread, directory, logger, parser, wordpress, basic, serial, ssl_wordpress, create, update, image, revert):
name = "Thread-{0}".format(int(name_thread) + 1) name = "Thread-{0}".format(int(name_thread) + 1)
directory = directory.split(",") directory = directory.split(",")
wordpress = wordpress.split(",") wordpress = wordpress.split(",")
@ -114,7 +126,7 @@ def importDirectory(name_thread, max_thread, directory, logger, parser, wordpres
for i in wordpress: for i in wordpress:
importWp = WPimport(name=name, basic=basic, wordpress=i, logger=logger, parser=parser, ssl_wordpress=ssl_wordpress, no_create=create, no_update=update, no_image=image) importWp = WPimport(name=name, basic=basic, wordpress=i, logger=logger, parser=parser, ssl_wordpress=ssl_wordpress, no_create=create, no_update=update, no_image=image)
for j in directory: for j in directory:
importWp.fromDirectory(j, name_thread, max_thread) importWp.fromDirectory(j, name_thread, max_thread, revert)
del importWp del importWp
else: else:
@ -123,7 +135,7 @@ def importDirectory(name_thread, max_thread, directory, logger, parser, wordpres
exit(1) exit(1)
for i in range(0, len(wordpress)-1): for i in range(0, len(wordpress)-1):
importWp = WPimport(name=name, basic=basic, wordpress=wordpress[i], logger=logger, parser=parser, ssl_wordpress=ssl_wordpress, no_create=create, no_update=update, no_image=image) importWp = WPimport(name=name, basic=basic, wordpress=wordpress[i], logger=logger, parser=parser, ssl_wordpress=ssl_wordpress, no_create=create, no_update=update, no_image=image)
importWp.fromDirectory(directory[i]) importWp.fromDirectory(directory[i], name_thread, max_thread, revert)
del importWp del importWp
@ -249,8 +261,9 @@ if __name__ == '__main__':
with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex: with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex:
wait_for = [ ex.submit(remove, i, args.parallel, args, basic, logger, ssl_wordpress) for i in range(0, int(args.parallel)) ] wait_for = [ ex.submit(remove, i, args.parallel, args, basic, logger, ssl_wordpress) for i in range(0, int(args.parallel)) ]
wait(wait_for, return_when=ALL_COMPLETED) wait(wait_for, return_when=ALL_COMPLETED)
errorRevert(logger, args.revert, args.tmp)
wait_for = [ wait_for = [
ex.submit(importDirectory, i, int(args.parallel), args.directory, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress, args.create, args.update, args.image) ex.submit(importDirectory, i, int(args.parallel), args.directory, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress, args.create, args.update, args.image, args.revert)
for i in range(0, int(args.parallel)) for i in range(0, int(args.parallel))
] ]
except Exception as err: except Exception as err:
@ -260,15 +273,7 @@ if __name__ == '__main__':
with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex: with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex:
wait_for = [ ex.submit(remove, i, args.parallel, args, basic, logger, ssl_wordpress) for i in range(0, int(args.parallel)) ] wait_for = [ ex.submit(remove, i, args.parallel, args, basic, logger, ssl_wordpress) for i in range(0, int(args.parallel)) ]
wait(wait_for, return_when=ALL_COMPLETED) wait(wait_for, return_when=ALL_COMPLETED)
if args.revert is True: errorRevert(logger, args.revert, args.tmp)
files_tmp = glob.glob("{0}/*.json".format(args.tmp))
if len(files_tmp) == 0:
logger.error("Error revert, because files not found")
exit(1)
if len(files_tmp) != int(args.parallel):
for file_r in files_tmp:
os.remove(file_r)
wait_for = [ wait_for = [
ex.submit(importUrl, i, int(args.parallel), args.canalblog, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress, ssl_canalblog, args.create, args.update, args.image, args.revert, args.tmp) ex.submit(importUrl, i, int(args.parallel), args.canalblog, logger, args.parser, args.wordpress, basic, args.serial, ssl_wordpress, ssl_canalblog, args.create, args.update, args.image, args.revert, args.tmp)
for i in range(0, int(args.parallel)) for i in range(0, int(args.parallel))

10
lib/WPImport.py
View File

@ -80,7 +80,7 @@ class WPimport:
directories = self._getDirectories([], "{0}".format(directory)) directories = self._getDirectories([], "{0}".format(directory))
if len(directories) > 0: if len(directories) > 0:
files = self._getFiles(directories) files = self._getFiles(directories)
if args.revert is False: if revert is False:
self._tmpFiles(files=files, number_thread=number_thread, max_thread=max_thread) self._tmpFiles(files=files, number_thread=number_thread, max_thread=max_thread)
self._fromFileTmp() self._fromFileTmp()
else: else:
@ -90,7 +90,7 @@ class WPimport:
def fromFile(self, files=[]): def fromFile(self, files=[]):
for i in range(0, len(files)): for i in range(0, len(files)):
if os.path.exists(files[i]): if os.path.exists(files[i]):
self._logger.info("{0} : ({1}/{2}) File is being processed : {3}".format(self._name, i+1, currentRangeFiles + 1, files[i])) self._logger.info("{0} : ({1}/{2}) File is being processed : {3}".format(self._name, i+1, len(files), files[i]))
with open(files[i], 'r') as f: with open(files[i], 'r') as f:
content = f.read() content = f.read()
self._logger.debug("{0} : Size of article : {1}".format(self._name, len(content))) self._logger.debug("{0} : Size of article : {1}".format(self._name, len(content)))
@ -110,10 +110,10 @@ class WPimport:
try: try:
with open("{0}/{1}.json".format(self._tmp, self._name)) as file: with open("{0}/{1}.json".format(self._tmp, self._name)) as file:
files = json.loads(file.read()) files = json.loads(file.read())
self._logger.debug("{0} : size of webpage : {1}".format(self._name, len(webpage))) self._logger.debug("{0} : size of webpage : {1}".format(self._name, len(files)))
for i in range(0, len(files)): for i in range(0, len(files)):
if os.path.exists(files[i]): if os.path.exists(files[i]):
self._logger.info("{0} : ({1}/{2}) File is being processed : {3}".format(self._name, i+1, currentRangeFiles + 1, files[i])) self._logger.info("{0} : ({1}/{2}) File is being processed : {3}".format(self._name, i+1, len(files), files[i]))
with open(files[i], 'r') as f: with open(files[i], 'r') as f:
content = f.read() content = f.read()
self._logger.debug("{0} : Size of article : {1}".format(self._name, len(content))) self._logger.debug("{0} : Size of article : {1}".format(self._name, len(content)))
@ -128,7 +128,7 @@ class WPimport:
self._logger.error("{0} : Read file json from tmp : {1}".format(self._name, ex)) self._logger.error("{0} : Read file json from tmp : {1}".format(self._name, ex))
def _tmpFiles(self, number_thread=1, max_thread=1): def _tmpFiles(self, files=[], number_thread=1, max_thread=1):
divFiles = int(len(files) / max_thread) divFiles = int(len(files) / max_thread)
currentRangeFiles = int(divFiles * (number_thread+1)) currentRangeFiles = int(divFiles * (number_thread+1))
firstRange = int(currentRangeFiles - divFiles) firstRange = int(currentRangeFiles - divFiles)