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

remove doublon

Browse Source
This commit is contained in:
Valentin CZERYBA 2023-05-21 21:12:00 +02:00
parent 769b7f43fc
commit 75772ba7f0
1 changed files with 80 additions and 40 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

118
lib/WPImport.py
View File

@ -5,6 +5,7 @@ from urllib.parse import urlparse
import requests, os, logging, re, json
from requests.adapters import HTTPAdapter
from requests.packages.urllib3.util.retry import Retry
from slugify import slugify
class WPimport:
# Constructor
@ -14,7 +15,7 @@ class WPimport:
self._wordpress = wordpress
self._logger = logger
self._parser = parser
self._headers_json = {'Content-Type': 'application/json', 'Accept':'application/json'}
self._headers_json = {'Content-Type': 'application/json; charset=utf-8', 'Accept':'application/json'}
self._request = requests.Session()
@ -487,7 +488,7 @@ class WPimport:
hour = articledate[0].text
time = dateheader[0].text.split(" ")
data = {"title":title, "content":bodyhtml, "status":"publish", "date": "{0}-{1}-{2}T{3}:00".format(time[2],month[time[1]],time[0], hour), "tags": listelement["tags"], "categories": listelement["categories"]}
params = {"search":author}
params = {"search":author, "per_page":100}
try:
self._logger.info("{0} : Search author : {1}".format(self._name, author))
page = self._request.get("http://{0}/wp-json/wp/v2/users".format(self._wordpress), auth=self._basic, params=params)
@ -503,46 +504,85 @@ class WPimport:
else:
self._logger.error("{0} : Connection error with status code for get author : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(page.content))
params = {"search": title}
try:
self._logger.info("{0} : Search post : {1}".format(self._name, title))
page = self._request.get("http://{0}/wp-json/wp/v2/posts".format(self._wordpress), auth=self._basic, params=params)
except ConnectionError as err:
self._logger.error("{0} : Connection error for search post : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for search post : {1}".format(self._name, err))
if title[len(title)-1] == " ":
title = title[:-1]
page_is_exist = False
if page.status_code == 200:
result = page.json()
self._logger.info("{0} : Number result posts : {1}".format(self._name, len(result)))
for i in result:
self._logger.info("{0} : Search title posts for {2} : {1}".format(self._name, i["title"]["rendered"], title))
if i["title"]["rendered"] == title:
page_is_exist = True
post_id = i["id"]
self._logger.debug("{0} : Data for post to update : {1}".format(self._name, result[0]))
self._logger.info("{0} : Page {1} already exist and going to update".format(self._name, title))
try:
page = self._request.post("http://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
except Exception as err:
self._logger.error("{0} : Connection error for update post : {1}".format(self._name, err))
exit(1)
if page.status_code == 200:
result = page.json()
self._logger.info("{0} : Post updated : {1}".format(self._name, title))
self._addOrUpdateComment(result["id"], comment_post, result["title"]["raw"])
self._linkImgPost(result["title"]["raw"], list_img, result["id"])
else:
self._logger.error("{0} : Post not updated due status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
else:
self._logger.error("{0} : Connection for update post error with status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
for index in range(1,10):
params = {"search": title, "per_page":100, "page": index}
try:
self._logger.info("{0} : Search post : {1}".format(self._name, title))
page = self._request.get("http://{0}/wp-json/wp/v2/posts".format(self._wordpress), auth=self._basic, params=params, headers=self._headers_json)
except ConnectionError as err:
self._logger.error("{0} : Connection error for search post : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for search post : {1}".format(self._name, err))
if page.status_code == 200:
self._logger.debug("{0} : Encoding : {1}".format(self._name, page.encoding))
page.encoding = "utf-8"
result = page.json()
if len(result) == 0:
break
self._logger.info("{0} : Number result posts : {1}".format(self._name, len(result)))
count = 0
for i in result:
if page_is_exist == False:
self._logger.info("{0} : Search title posts for |{2}| : |{1}|".format(self._name, i["title"]["rendered"], title))
title_rendered = i["title"]["rendered"].replace('’', "'")
title_rendered = title_rendered.replace('–', '-')
title_rendered = title_rendered.replace('…', '...')
title_rendered = title_rendered.replace('« ', '"')
title_rendered = title_rendered.replace(' »', '"')
self._logger.debug("{0} : SIze of title : {1} - {2}".format(self._name, len(title), len(title_rendered)))
if title_rendered == title:
page_is_exist = True
post_id = i["id"]
count = count + 1
if count > 1:
self._logger.info("{0} : Page {1} is double and going to delete".format(self._name, title))
try:
params = {"force":1}
page = self._request.delete("http://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id), auth=self._basic, headers=self._headers_json, params=params)
except ConnectionError as err:
self._logger.error("{0} : Connection error for deleted post : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for deleted post : {1}".format(self._name, err))
if page.status_code == 200:
self._logger.info("{0} : Post deleted : {1}".format(self._name, title))
else:
self._logger.error("{0} : Post not updated due status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
else:
self._logger.debug("{0} : Data for post to update : {1}".format(self._name, i))
self._logger.info("{0} : Page {1} already exist and going to update".format(self._name, title))
try:
page = self._request.post("http://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
except ConnectionError as err:
self._logger.error("{0} : Connection error for update post : {1}".format(self._name, err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for update post : {1}".format(self._name, err))
if page.status_code == 200:
result = page.json()
self._logger.info("{0} : Post updated : {1}".format(self._name, title))
self._addOrUpdateComment(result["id"], comment_post, result["title"]["raw"])
self._linkImgPost(result["title"]["raw"], list_img, result["id"])
else:
self._logger.error("{0} : Post not updated due status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
if page.status_code == 400:
self._logger.error("{0} : Connection for update post unauthorized : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
break
else:
self._logger.error("{0} : Connection for update post error with status code : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content))
if page_is_exist is False:
try:
self._logger.info("{0} : Creating posts : {1}".format(self._name, data["title"]))
page = self._request.post("http://{0}/wp-json/wp/v2/posts".format(self._wordpress), auth=self._basic, headers=self._headers_json, data=json.dumps(data))