web_scrap/insert_wordpress.py

#!/usr/bin/python3
from bs4 import BeautifulSoup
from urllib.parse import urlparse
from requests.auth import HTTPBasicAuth
from getpass import getpass
import requests, os, argparse, logging


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"}
    parser = argparse.ArgumentParser()
    parser.add_argument("--user", help="wordpress user", required=True)
    parser.add_argument("--file", help="HTML file", required=True)
    args = parser.parse_args()
    password = getpass()
    if len(password) == 0:
        print("No password error !!! ")
        exit(1)
    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:
        contents = f.read()

        soup = BeautifulSoup(contents, 'html.parser')
        articletitle = soup.find_all("h2", class_="articletitle")
        articlebody = soup.find_all("div", class_="articlebody")
        articledate = soup.find_all("span", class_="articledate")
        dateheader = soup.find_all("div", class_="dateheader")
        itemfooter = soup.find_all("div", class_="itemfooter")
        a = itemfooter[0].find_all("a", {"rel": True})
        tag = []
        for i in a:
            rel = i.get("rel")
            if rel[0] == 'tag':
                tag.append(i.text)
        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
        body = articlebody[0]
        hour = articledate[0].text
        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": listtag}
        print(data)
        exit(0)

    page = requests.post("http://localhost:8080/wp-json/wp/v2/posts", auth=basic, data=data)
    print(page.status_code)
    if page.status_code == 201:
        print(page.content)