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

add file tmp

Browse Source
This commit is contained in:
Valentin CZERYBA 2023-06-19 23:58:59 +02:00
parent 703cc8922a
commit 7dace5bdb7
3 changed files with 20 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
import_export_canalblog.py
View File

@ -5,7 +5,7 @@ from urllib.parse import urlparse
from concurrent import futures
from concurrent.futures import as_completed, wait, ALL_COMPLETED
import argparse, logging, threading
import argparse, logging, threading, os, glob
from lib.WPImport import WPimport
from lib.WPExport import WPExport
from lib.WPRemove import WPRemove
@ -40,8 +40,9 @@ def remove(index, number, args, basic, logger, ssl_wordpress):
del removeWp
def download(name_thread, max_thread, url, logger, parser, directory, html, img, ssl_canalblog):
def download(name_thread, max_thread, url, logger, parser, directory, html, img, ssl_canalblog, revert, tmp):
exportWp = WPExport(name="Thread-{0}".format(int(name_thread) + 1), url=url, logger=logger, parser=parser, directory=directory, ssl_canalblog=ssl_canalblog)
if not revert:
webpage = exportWp.getUrlPage(name_thread, max_thread)
for i in ["article", "page"]:
for j in ["publications", "principal"]:
@ -54,7 +55,7 @@ def download(name_thread, max_thread, url, logger, parser, directory, html, img,
def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, basic, serial, ssl_wordpress, ssl_canalblog, create, update, image):
def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, basic, serial, ssl_wordpress, ssl_canalblog, create, update, image, revert, tmp):
canalblog = canalblog.split(",")
wordpress = wordpress.split(",")
name = "Thread-{0}".format(int(name_thread) + 1)
@ -71,6 +72,7 @@ def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, bas
logger.error("{0} : parsing error : {1}".format(name, err))
exit(1)
exportWp = WPExport(name="Thread-{0}".format(int(name_thread) + 1), url=url, logger=logger, parser=parser, ssl_canalblog=ssl_canalblog)
if not revert:
webpage = exportWp.getUrlPage(name_thread, max_thread)
del exportWp
for j in wordpress:
@ -93,6 +95,7 @@ def importUrl(name_thread, max_thread, canalblog, logger, parser, wordpress, bas
logger.error("parsing error : {0}".format(err))
exit(1)
exportWp = WPExport(name=name, url=url, logger=logger, parser=parser, ssl_canalblog=ssl_canalblog)
if not revert:
webpage = exportWp.getUrlPage(name_thread, max_thread)
del exportWp
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)
@ -127,6 +130,7 @@ def importDirectory(name_thread, max_thread, directory, logger, parser, wordpres
if __name__ == '__main__':
TMP = "/tmp/import_export_canablog"
parser = argparse.ArgumentParser()
parser.add_argument("--debug", help="Verbosity", action="store_true")
parser.add_argument("--logfile", help="Log file", default="")
@ -134,6 +138,7 @@ if __name__ == '__main__':
parser.add_argument("--parser", help="Parser content", default="html.parser")
parser.add_argument("--parallel", help="Define number thread (default : 1)", default=1)
parser.add_argument("--no-ssl", help="No ssl for canalblog and/or wordpress (example wordpress,canalblog)", dest="ssl", default="")
parser.add_argument("--revert", help="Restart a work from stopping work", action="store_true")
subparsers = parser.add_subparsers(dest="command")
@ -188,6 +193,8 @@ if __name__ == '__main__':
help="File")
if not os.path.exists(TMP):
os.mkdir(TMP)
args = parser.parse_args()
@ -256,7 +263,7 @@ if __name__ == '__main__':
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_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)
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, TMP)
for i in range(0, int(args.parallel))
]
@ -294,7 +301,7 @@ if __name__ == '__main__':
try:
with futures.ThreadPoolExecutor(max_workers=int(args.parallel)) as ex:
wait_for = [
ex.submit(download, i, int(args.parallel), url, logger, args.parser, args.directory, args.html, args.img, ssl_canalblog)
ex.submit(download, i, int(args.parallel), url, logger, args.parser, args.directory, args.html, args.img, ssl_canalblog, args.revert, TMP)
for i in range(0, int(args.parallel))
]
except Exception as err:

4
lib/WPExport.py
View File

@ -6,7 +6,7 @@ from requests.adapters import HTTPAdapter
from requests.packages.urllib3.util.retry import Retry
class WPExport:
def __init__(self, name = "Thread-0", url = "", logger = None, parser = "html.parser", directory = "backup", ssl_canalblog=True):
def __init__(self, name = "Thread-0", url = "", logger = None, parser = "html.parser", directory = "backup", ssl_canalblog=True, tmp="/tmp/import_export_canablog"):
self._url = url
self._logger = logger
self._parser = parser
@ -22,7 +22,7 @@ class WPExport:
status_forcelist=[429, 500, 502, 503, 504], backoff_factor=2)
self._request.mount('{0}://'.format(self._protocol), HTTPAdapter(max_retries=retries))
self._tmp = tmp
# Destructor
def __del__(self):

3
lib/WPImport.py
View File

@ -8,7 +8,7 @@ from requests.packages.urllib3.util.retry import Retry
class WPimport:
# Constructor
def __init__(self, name="Thread-0", basic=None, wordpress="", logger=None, parser="html.parser", ssl_wordpress=True, no_create=False, no_update=False, no_image=False):
def __init__(self, name="Thread-0", basic=None, wordpress="", logger=None, parser="html.parser", ssl_wordpress=True, no_create=False, no_update=False, no_image=False, tmp="/tmp/import_export_canablog"):
self._name = name
self._basic = basic
self._wordpress = wordpress
@ -28,6 +28,7 @@ class WPimport:
self._no_create = no_create
self._no_update = no_update
self._no_image = no_image
self._tmp = tmp
# Destructor
def __del__(self):