Merge pull request #5700 from openstates/ga-bill-verison-urls-special

GA: fix bill version URLs for special session 2026

Author: jessemortenson

scrapers/ga/bills.py

@@ -4,7 +4,14 @@
 
 from openstates.scrape import Scraper, Bill, VoteEvent
 
-from .util import get_client, get_url, backoff, get_token, SESSION_SITE_IDS
+from .util import (
+    get_client,
+    get_url,
+    backoff,
+    get_token,
+    SESSION_SITE_IDS,
+    SPECIAL_SESSION_SITE_URL_PARTS,
+)
 
 #         Methods (7):
 #            GetLegislationDetail(xs:int LegislationId, )
@@ -365,11 +372,13 @@ def scrape(self, session=None, chamber=None):
                     url = (
                         f"https://www.legis.ga.gov/api/legislation/document/{bill_bit}"
                     )
-                if session == "2023_ss":
-                    # gets url as 2023EX220573.pdf
-                    # so would be easier to put together using session & version id
-                    # to get https://www.legis.ga.gov/api/legislation/document/2023EX/220573
-                    url = f"https://www.legis.ga.gov/api/legislation/document/2023EX/{doc_id}"
+                if session in ["2023_ss", "2026_ss"]:
+                    # Last couple special sessions the url provided in instrument["Versions"]["DocumentDescription"]
+                    # is wrong, so it'll look like http://www.legis.ga.gov/Legislation/2026EX/249298.pdf
+                    # but that URL does not resolve (gets 404 on GA page)
+                    # the web UI actually gets to you to a path that uses a varient session ID and the doc_id
+                    # like https://www.legis.ga.gov/api/legislation/document/2026EX/249298
+                    url = f"https://www.legis.ga.gov/api/legislation/document/{SPECIAL_SESSION_SITE_URL_PARTS[session]}/{doc_id}"
                 link = bill.add_version_link(name, url, media_type="application/pdf")
                 link["extras"] = {
                     "_internal_document_id": doc_id,

scrapers/ga/util.py

@@ -67,6 +67,10 @@ def _():
     "2011_ss": 22,
     "2011_12": 21,
 }
+SPECIAL_SESSION_SITE_URL_PARTS = {
+    "2026_ss": "2026EX",
+    "2023_ss": "2023EX",
+}
 
 
 def get_key(timestamp):