aboutsummaryrefslogtreecommitdiff
path: root/extract-event-url
diff options
context:
space:
mode:
authordiogo464 <[email protected]>2025-07-24 16:55:50 +0100
committerdiogo464 <[email protected]>2025-07-24 16:55:50 +0100
commit5c3fe9aac767ee44f271ea8d1086b28336c88334 (patch)
tree0a2c6a165c0ee429b4caf5c6a80430a1aebe3361 /extract-event-url
parent6138d43d3ff8736bbcb95e4e34cdbb3f79f4d7cc (diff)
added extract-event-url
Diffstat (limited to 'extract-event-url')
-rwxr-xr-xextract-event-url25
1 files changed, 25 insertions, 0 deletions
diff --git a/extract-event-url b/extract-event-url
new file mode 100755
index 000000000..1bd86898d
--- /dev/null
+++ b/extract-event-url
@@ -0,0 +1,25 @@
1#!/usr/bin/env python3
2import os
3import re
4import sys
5
6pattern = r'<a[^>]*class=[\'"][^\'"]*evcal_evdata_row evo_clik_row[^\'"]*[\'"][^>]*href=[\'"]([^\'"]+)[\'"][^>]*>'
7
8for slug in sys.argv[1:]:
9 page_path = os.path.join("events", slug, "page.html")
10 event_url_path = os.path.join("events", slug, "event-url")
11 if os.path.exists(event_url_path):
12 continue
13
14 with open(page_path, "r") as f:
15 page_content = f.read()
16
17 match = re.search(pattern, page_content)
18 if not match:
19 print(f"event {slug} has no event url")
20 continue
21
22 event_page_url = match.group(1)
23 with open(event_url_path, "w") as f:
24 f.write(event_page_url)
25