VocaDB Wiki revision history.py

''' Output (sorted by date):
λ "VocaDB Wiki revision history.py"
Found 77 pages

2024-01-03 10:22:51+00:00: wiki/59 'Content policy' (version 28) edit by Shiroizu
2024-01-01 13:54:17+00:00: wiki/86 'Album entry editing' (version 11) edit by Shiroizu
2024-01-01 13:54:02+00:00: wiki/83 'Artist entry editing' (version 19) edit by Shiroizu
2024-01-01 13:53:30+00:00: wiki/89 'Song entry editing ' (version 19) edit by Shiroizu
...
2023-02-16 03:32:05+00:00: wiki/53 'Management guidelines / golden rules' (version 24) edit by andreoda
2023-02-16 03:31:50+00:00: wiki/29 'License' (version 14) edit by andreoda
2023-02-16 01:56:24+00:00: wiki/22 'VocaDB domains & login process' (version 6) edit by andreoda
'''

import requests
import time
from bs4 import BeautifulSoup
from datetime import datetime

all_wiki_pages_url = "https://wiki.vocadb.net/pages/allpages"
page_edits = []  # edit_date, wiki_page_title, wiki_page_id, version, editor

r = requests.get(all_wiki_pages_url)
all_wiki_pages = BeautifulSoup(r.text, "html.parser").select("table td.pagename a")

print(f"Found {len(all_wiki_pages)} pages\n")

for wiki_page_link in all_wiki_pages:
    # <a href="/wiki/62/artist-merging-and-splitting-guidelines"> Artist merging and splitting guidelines</a>
    wiki_page_title = wiki_page_link.text
    wiki_page_id = str(wiki_page_link).split("/wiki/")[1].split("/")[0]
    wiki_page_history_url = f"https://wiki.vocadb.net/pages/history/{wiki_page_id}"
    r = requests.get(wiki_page_history_url)
    most_recent_edit = BeautifulSoup(r.text, "html.parser").select(
        "tbody tr:first-child td"
    )
    version, edit_date, editor, _ = most_recent_edit
    edit_date = datetime.fromisoformat(edit_date.text)
    # print(f"Version {version.text}, {edit_date}, by {editor.text}")
    page_edits.append(
        [edit_date, wiki_page_title, wiki_page_id, version.text, editor.text]
    )
    time.sleep(0.5)

page_edits.sort(key=lambda x: x[0], reverse=True)
for page_edit in page_edits:
    edit_date, wiki_page_title, wiki_page_id, version, editor = page_edit
    print(
        f"{edit_date}: wiki/{wiki_page_id} '{wiki_page_title}' (version {version}) edit by {editor}"
    )