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

add tags

Browse Source
This commit is contained in:
Valentin CZERYBA 2023-03-23 23:49:42 +01:00
parent eae95d5671
commit 3622e37942
1 changed files with 26 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

31
insert_wordpress.py
View File

@ -7,9 +7,8 @@ import requests, os, argparse, logging
if __name__ == '__main__': if __name__ == '__main__':
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"}
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)
@ -20,6 +19,13 @@ if __name__ == '__main__':
exit(1) exit(1)
basic = HTTPBasicAuth(args.user, password) basic = HTTPBasicAuth(args.user, password)
page = requests.get("http://localhost:8080/wp-json/wp/v2/tags")
if page.status_code == 200:
tags = page.json()
print(tags)
with open(args.file, 'r') as f: with open(args.file, 'r') as f:
contents = f.read() contents = f.read()
@ -35,13 +41,28 @@ if __name__ == '__main__':
rel = i.get("rel") rel = i.get("rel")
if rel[0] == 'tag': if rel[0] == 'tag':
tag.append(i.text) tag.append(i.text)
print(tag) listtag = []
for i in tag:
tag_exist = False
for j in tags:
if j["name"] == i:
tag_exist = True
listtag.append(j["id"])
if tag_exist is False:
data = {"name": i}
page = requests.post("http://localhost:8080/wp-json/wp/v2/tags", auth=basic, data=data)
if page.status_code == 201:
result = page.json()
listtag.append(result["id"])
title = articletitle[0].text title = articletitle[0].text
body = articlebody[0] body = articlebody[0]
hour = articledate[0].text hour = articledate[0].text
time = dateheader[0].text.split(" ") time = dateheader[0].text.split(" ")
data = {"title":title, "content":body, "status":"publish", "date": "{0}-{1}-{2}T{3}:00".format(time[2],month[time[1]],time[0], hour), "tags": [5]} data = {"title":title, "content":body, "status":"publish", "date": "{0}-{1}-{2}T{3}:00".format(time[2],month[time[1]],time[0], hour), "tags": listtag}
#print(data) print(data)
exit(0) exit(0)
page = requests.post("http://localhost:8080/wp-json/wp/v2/posts", auth=basic, data=data) page = requests.post("http://localhost:8080/wp-json/wp/v2/posts", auth=basic, data=data)