PGŽ Sport Platform — Round 1+2 baseline (sport2.html + API)

scrapers/rijeka_sport_scraper.py

@@ -0,0 +1,20 @@
+#!/usr/bin/env python3
+# rijeka_sport_scraper.py — sport.rijeka.hr + rijeka.hr/sport
+import sys, os
+sys.path.insert(0, '/opt/pgz-sport/scrapers')
+from pgz_sport_deep import harvest as base_harvest, fetch, extract_text, find_links, find_pdf_links
+import logging
+logging.getLogger().handlers.clear()
+logging.basicConfig(level=logging.INFO, format='%(asctime)s [rijeka_sport] %(message)s')
+
+# Override roots
+import pgz_sport_deep
+pgz_sport_deep.ROOTS = [
+    "https://www.rijeka.hr/teme-za-gradane/sport-i-rekreacija/",
+    "https://www.rijeka.hr/sport/",
+    "https://sport.rijeka.hr",
+    "https://www.rijekasport.hr",
+]
+
+if __name__ == "__main__":
+    pgz_sport_deep.harvest()