sophie/sheets
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

a

Browse source
This commit is contained in:
Eduard Prigoana 2025-08-19 08:15:02 +03:00
parent 650f6fde39
commit f2eec0afd9
3 changed files with 12 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

3
config.py
View file

@ -15,7 +15,10 @@ exclude_names = {
"K4HKn0", "K4HKn0",
"AI Models", "AI Models",
"🎹 BPM & Key Tracker", "🎹 BPM & Key Tracker",
"🎹Comps & Edits"
"🎹 Worst Comps & Edits", "🎹 Worst Comps & Edits",
"🎹Worst Comps & Edits",
"🎹 Yedits",
"Allegations", "Allegations",
"Rap Disses Timeline", "Rap Disses Timeline",
"Underground Artists", "Underground Artists",

10
info/status.json
View file

@ -1,15 +1,15 @@
{ {
"last_updated": "2025-08-11T17:26:51.365776Z", "last_updated": "2025-08-19T05:14:17.304876Z",
"files": { "files": {
"Artists.html": { "Artists.html": {
"hash": "5c794c69a86112ac86a59c25f62409754b692b08ab57c4555afa9e45721ed054", "hash": "b48e3d341500e82e179f8813443e5fed899313b48ef79b83bd87388189c49a90",
"last_archived": "2025-08-11T17:26:51.365790Z" "last_archived": "2025-08-19T05:14:17.304886Z"
}, },
"artists.csv": { "artists.csv": {
"hash": "f65b4bac779810e9f6775f3d98505d876e2692233c2581bd4a53471c8cf188ac" "hash": "609ab1ad6adf62bc1ab8e1b433e1676ab6787e65c760e22d7da7a8e26bbdf33e"
}, },
"artists.xlsx": { "artists.xlsx": {
"hash": "f41e3527911e55863c3cc4e0cec778bb689d8d6d5086c53fe964df963583b910" "hash": "101b6cdef091774668b030bd1bed27a78cb227b487f5f04883c7a2f4bb184b5e"
} }
} }
} }

5
parser.py
View file

@ -19,8 +19,11 @@ def generate_csv():
if len(cells) < 4: if len(cells) < 4:
continue continue
# Always take the artist name from the column text
artist_name_raw = cells[0].get_text(strip=True)
# Only use the <a> for the URL (if it exists)
link_tag = cells[0].find("a") link_tag = cells[0].find("a")
artist_name_raw = link_tag.get_text(strip=True) if link_tag else cells[0].get_text(strip=True)
artist_url = link_tag["href"] if link_tag else "" artist_url = link_tag["href"] if link_tag else ""
if not artist_url: if not artist_url:
continue continue