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

add class + wordpress

Browse Source
This commit is contained in:
Valentin CZERYBA 2023-03-29 22:31:35 +02:00
parent 19229bc65b
commit f250637912
1 changed files with 54 additions and 47 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

95
insert_wordpress.py
View File

@ -5,27 +5,33 @@ from requests.auth import HTTPBasicAuth
from getpass import getpass from getpass import getpass
import requests, os, argparse, logging, re import requests, os, argparse, logging, re
def insertWordpress(file, basic): class WPimport:
tags = []
month = {"janvier":"01", "février": "02", "mars": "03", "avril":"04", "mai": "05", "juin": "06", "juillet": "07", "août": "08", "septembre": "09", "octobre": "10", "novembre": "11", "décembre": "12"}
liste = ["categories", "tags"] def __init__(self, basic, wordpress):
elements = {} self.basic = basic
element = {} self.wordpress = wordpress
listelement = {}
for i in liste: def fromFile(self, file):
page = requests.get("http://localhost:8080/wp-json/wp/v2/{0}".format(i)) with open(file, 'r') as f:
if page.status_code == 200: contents = f.read()
elements[i] = page.json() self.insertWordpress(contents)
element[i] = []
listelement[i] = []
def insertWordpress(self, content):
tags = []
month = {"janvier":"01", "février": "02", "mars": "03", "avril":"04", "mai": "05", "juin": "06", "juillet": "07", "août": "08", "septembre": "09", "octobre": "10", "novembre": "11", "décembre": "12"}
liste = ["categories", "tags"]
elements = {}
element = {}
listelement = {}
with open(file, 'r') as f: for i in liste:
contents = f.read() page = requests.get("http://{0}/wp-json/wp/v2/{1}".format(self.wordpress,i))
if page.status_code == 200:
elements[i] = page.json()
element[i] = []
listelement[i] = []
soup = BeautifulSoup(contents, 'html.parser') soup = BeautifulSoup(content, 'html.parser')
articletitle = soup.find_all("h2", class_="articletitle") articletitle = soup.find_all("h2", class_="articletitle")
articlebody = soup.find_all("div", class_="articlebody") articlebody = soup.find_all("div", class_="articlebody")
articledate = soup.find_all("span", class_="articledate") articledate = soup.find_all("span", class_="articledate")
@ -63,14 +69,11 @@ def insertWordpress(file, basic):
array = listelement[i].append(k["id"]) array = listelement[i].append(k["id"])
if element_exist is False: if element_exist is False:
data = {"name": j} data = {"name": j}
page = requests.post("http://localhost:8080/wp-json/wp/v2/{0}".format(i), auth=basic, data=data) page = requests.post("http://{0}/wp-json/wp/v2/{1}".format(self.wordpress, i), auth=self.basic, data=data)
if page.status_code == 201: if page.status_code == 201:
result = page.json() result = page.json()
listelement[i].append(result["id"]) listelement[i].append(result["id"])
title = articletitle[0].text title = articletitle[0].text
author = articleacreator[0].text.lower() author = articleacreator[0].text.lower()
body = articlebody[0].find_all("p") body = articlebody[0].find_all("p")
@ -86,34 +89,36 @@ def insertWordpress(file, basic):
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"]} 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}
page = requests.get("http://localhost:8080/wp-json/wp/v2/users", auth=basic, params=params) page = requests.get("http://{0}/wp-json/wp/v2/users".format(self.wordpress), auth=self.basic, params=params)
if page.status_code == 200: if page.status_code == 200:
result = page.json() result = page.json()
data["author"] = result[0]["id"] data["author"] = result[0]["id"]
params = {"search":title}
params = {"search":title} page = requests.get("http://{0}/wp-json/wp/v2/posts".format(self.wordpress), auth=self.basic, params=params)
page = requests.get("http://localhost:8080/wp-json/wp/v2/posts", auth=basic, params=params) page_exist = True
page_exist = True if page.status_code:
if page.status_code:
result = page.json()
if len(result) == 0:
page_exist = False
else:
print("La page {0} existe deja".format(title))
if page_exist == False:
page = requests.post("http://localhost:8080/wp-json/wp/v2/posts", auth=basic, data=data)
if page.status_code == 201:
result = page.json() result = page.json()
print("Article ajoute : {0}".format(result["title"]["raw"])) if len(result) == 0:
print(comment_post) page_exist = False
for i in comment_post: else:
data = {"post": result["id"], "content": i["content"], "date": i["date"], "author_name": i["author"]} print("La page {0} existe deja".format(title))
page = requests.post("http://localhost:8080/wp-json/wp/v2/comments", auth=basic, data=data)
print(page.status_code) if page_exist == False:
if page.status_code == 201: page = requests.post("http://{0}/wp-json/wp/v2/posts".format(self.wordpress), auth=self.basic, data=data)
print("Commentaire ajoute pour {0}".format(result["title"]["raw"])) if page.status_code == 201:
result = page.json()
print("Article ajoute : {0}".format(result["title"]["raw"]))
print(comment_post)
for i in comment_post:
data = {"post": result["id"], "content": i["content"], "date": i["date"], "author_name": i["author"]}
page = requests.post("http://{0}/wp-json/wp/v2/comments".format(self.wordpress), auth=self.basic, data=data)
print(page.status_code)
if page.status_code == 201:
print("Commentaire ajoute pour {0}".format(result["title"]["raw"]))
@ -121,6 +126,7 @@ if __name__ == '__main__':
parser = argparse.ArgumentParser() parser = argparse.ArgumentParser()
parser.add_argument("--user", help="wordpress user", required=True) parser.add_argument("--user", help="wordpress user", required=True)
parser.add_argument("--file", help="HTML file", required=True) parser.add_argument("--file", help="HTML file", required=True)
parser.add_argument("--wordpress", help="URL Wordpress", required=True)
args = parser.parse_args() args = parser.parse_args()
password = getpass() password = getpass()
if len(password) == 0: if len(password) == 0:
@ -128,4 +134,5 @@ if __name__ == '__main__':
exit(1) exit(1)
basic = HTTPBasicAuth(args.user, password) basic = HTTPBasicAuth(args.user, password)
insertWordpress(args.file, basic) importWp = WPimport(basic)
importWp.fromFile(args.file)