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

private variable

Browse Source
This commit is contained in:
Valentin CZERYBA 2023-04-02 13:14:52 +02:00
parent f07f8c040f
commit cc33ab34df
1 changed files with 14 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

28
insert_wordpress.py
View File

@ -8,22 +8,22 @@ import requests, os, argparse, logging, re, json
class WPimport: class WPimport:
def __init__(self, basic, wordpress): def __init__(self, basic, wordpress):
self.basic = basic self._basic = basic
self.wordpress = wordpress self._wordpress = wordpress
def fromFile(self, file): def fromFile(self, file):
with open(file, 'r') as f: with open(file, 'r') as f:
contents = f.read() contents = f.read()
self.insertWordpress(contents) self._insertWordpress(contents)
def _linkImgPost(self, title, list_img, post_id): def _linkImgPost(self, title, list_img, post_id):
for i in list_img: for i in list_img:
data = {"post": post_id} data = {"post": post_id}
r = requests.post("http://{0}/wp-json/wp/v2/media/{1}".format(self.wordpress, i["id"]), auth=self.basic, data=data) r = requests.post("http://{0}/wp-json/wp/v2/media/{1}".format(self._wordpress, i["id"]), auth=self._basic, data=data)
if r.status_code == 200: if r.status_code == 200:
print("Association d'une image à l'article {0}".format(title)) print("Association d'une image à l'article {0}".format(title))
def insertWordpress(self, content): def _insertWordpress(self, content):
tags = [] 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"} 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"] liste = ["categories", "tags"]
@ -32,7 +32,7 @@ class WPimport:
listelement = {} listelement = {}
for i in liste: for i in liste:
page = requests.get("http://{0}/wp-json/wp/v2/{1}".format(self.wordpress,i)) page = requests.get("http://{0}/wp-json/wp/v2/{1}".format(self._wordpress,i))
if page.status_code == 200: if page.status_code == 200:
elements[i] = page.json() elements[i] = page.json()
element[i] = [] element[i] = []
@ -67,7 +67,7 @@ class WPimport:
split_fileimg = href_img.split("/") split_fileimg = href_img.split("/")
img_name = split_fileimg[len(split_fileimg)-1] img_name = split_fileimg[len(split_fileimg)-1]
params = { "search": img_name} params = { "search": img_name}
r = requests.get("http://{0}/wp-json/wp/v2/media".format(self.wordpress), auth=self.basic, params=params) r = requests.get("http://{0}/wp-json/wp/v2/media".format(self._wordpress), auth=self._basic, params=params)
if r.status_code == 200: if r.status_code == 200:
res = r.json() res = r.json()
if len(res) == 0: if len(res) == 0:
@ -76,7 +76,7 @@ class WPimport:
if img_name.split(".")[1]: if img_name.split(".")[1]:
img_type = "image/jpg" img_type = "image/jpg"
headers={ 'Content-Type': img_type,'Content-Disposition' : 'attachment; filename={0}'.format(img_name)} headers={ 'Content-Type': img_type,'Content-Disposition' : 'attachment; filename={0}'.format(img_name)}
r = requests.post("http://{0}/wp-json/wp/v2/media".format(self.wordpress), auth=self.basic, headers=headers, data=data) r = requests.post("http://{0}/wp-json/wp/v2/media".format(self._wordpress), auth=self._basic, headers=headers, data=data)
if r.status_code == 201: if r.status_code == 201:
res = r.json() res = r.json()
id_res = res["id"] id_res = res["id"]
@ -120,7 +120,7 @@ class WPimport:
listelement[i].append(k["id"]) 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://{0}/wp-json/wp/v2/{1}".format(self.wordpress, i), auth=self.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"])
@ -143,13 +143,13 @@ class WPimport:
time = dateheader[0].text.split(" ") 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"]} 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://{0}/wp-json/wp/v2/users".format(self.wordpress), auth=self.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://{0}/wp-json/wp/v2/posts".format(self._wordpress), auth=self._basic, params=params)
page_exist = True page_exist = True
headers = {'Content-Type': 'application/json', 'Accept':'application/json'} headers = {'Content-Type': 'application/json', 'Accept':'application/json'}
if page.status_code == 200: if page.status_code == 200:
@ -159,7 +159,7 @@ class WPimport:
else: else:
print("La page {0} existe deja et mis à jour".format(title)) print("La page {0} existe deja et mis à jour".format(title))
post_id = result[0]["id"] post_id = result[0]["id"]
page = requests.post("http://{0}/wp-json/wp/v2/posts/{1}".format(self.wordpress, post_id), auth=self.basic, headers=headers, data=json.dumps(data)) page = requests.post("http://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id), auth=self._basic, headers=headers, data=json.dumps(data))
if page.status_code == 200: if page.status_code == 200:
result = page.json() result = page.json()
print("Article mis à jour : {0}".format(result["title"]["raw"])) print("Article mis à jour : {0}".format(result["title"]["raw"]))
@ -168,13 +168,13 @@ class WPimport:
if page_exist == False: if page_exist == False:
page = requests.post("http://{0}/wp-json/wp/v2/posts".format(self.wordpress), auth=self.basic, headers=headers, data=json.dumps(data)) page = requests.post("http://{0}/wp-json/wp/v2/posts".format(self._wordpress), auth=self._basic, headers=headers, data=json.dumps(data))
if page.status_code == 201: if page.status_code == 201:
result = page.json() result = page.json()
print("Article ajoute : {0}".format(result["title"]["raw"])) print("Article ajoute : {0}".format(result["title"]["raw"]))
for i in comment_post: for i in comment_post:
data = {"post": result["id"], "content": i["content"], "date": i["date"], "author_name": i["author"]} 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) page = requests.post("http://{0}/wp-json/wp/v2/comments".format(self._wordpress), auth=self._basic, data=data)
if page.status_code == 201: if page.status_code == 201:
print("Commentaire ajoute pour {0}".format(result["title"]["raw"])) print("Commentaire ajoute pour {0}".format(result["title"]["raw"]))
self._linkImgPost(result["title"]["raw"], list_img, result["id"]) self._linkImgPost(result["title"]["raw"], list_img, result["id"])