warn-scraper 1.2.105__py3-none-any.whl → 1.2.107__py3-none-any.whl

warn/scrapers/dc.py

@@ -48,13 +48,15 @@ def scrape(
         url = f"https://does.dc.gov/page/industry-closings-and-layoffs-warn-notifications-{current_year - 1}"
         success, content = utils.save_if_good_url(targetfile, url)
 
-    root_html = content
-    #    r = utils.get_url(url)
-    #    r.encoding = "utf-8"
-    #    root_html = r.text
+    root_html = cache.read(targetfile)  # Explicitly re-read as text for regex to work
 
-    # Save it to the cache
-    #    cache.write(targetfile, root_html)
+    # A June 2025 entry includes a weird table inside a table cell.
+    # This is an ugly patch.
+    weirdtable = r"(\s+<table>\s+<tbody>\s+<tr>\s+<td>)(.*)(</td>\s+</tr>\s+</tbody>\s+</table>\s+)"
+    after = re.subn(weirdtable, r"\2", root_html)  #
+    if after[1] > 0:
+        logger.debug(f"{after[1]} changes made to {url}")
+        root_html = after[0]
 
     # Parse the list of links
     soup = BeautifulSoup(root_html, "html5lib")
@@ -84,11 +86,18 @@ def scrape(
         r = utils.get_url(href)
         r.encoding = "utf-8"
         html = r.text
-
         # Save it to the cache
         cache_key = uuid.uuid5(uuid.NAMESPACE_URL, href)
         cache.write(f"dc/{cache_key}.html", html)
 
+        # A June 2025 entry includes a weird table inside a table cell.
+        # This is an ugly patch.
+        weirdtable = r"(\s+<table>\s+<tbody>\s+<tr>\s+<td>)(.*)(</td>\s+</tr>\s+</tbody>\s+</table>\s+)"
+        after = re.subn(weirdtable, r"\2", html)
+        if after[1] > 0:
+            logger.debug(f"{after[1]} changes made to {href}")
+            html = after[0]
+
         # Add it to the list
         html_list.append(html)
 

{warn_scraper-1.2.105.dist-info → warn_scraper-1.2.107.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: warn-scraper
-Version: 1.2.105
+Version: 1.2.107
 Summary: Command-line interface for downloading WARN Act notices of qualified plant closings and mass layoffs from state government websites
 Home-page: https://github.com/biglocalnews/warn-scraper
 Author: Big Local News

{warn_scraper-1.2.105.dist-info → warn_scraper-1.2.107.dist-info}/RECORD

@@ -31,7 +31,7 @@ warn/scrapers/az.py,sha256=elGbue01Gjf_DQ66Wy9qqGIOJsiY-KIKJOVeft8pCXg,1447
 warn/scrapers/ca.py,sha256=VQOfjHXPCc-jYwh-EPGVVfnzvXB7pdmCt2uJ6QnMPRM,8600
 warn/scrapers/co.py,sha256=hUfqrzlhXQBkP4vxewVRrMZrgInoLer5S2MZlyYIQE4,7878
 warn/scrapers/ct.py,sha256=PKeZtlB0-z2wCmYmGl_WYoVo2gzwKV36upZcJVaJxjM,4852
-warn/scrapers/dc.py,sha256=_sHLnVqK_W90QqJb_W88yDlgPjoMl63LYZP3CJfdN9g,4484
+warn/scrapers/dc.py,sha256=C0JwgGX7A4JMxlahTrfzbFKpmyPxF7y6wsnUf-sE3OU,5120
 warn/scrapers/de.py,sha256=GyM92A-lFwZAfRxgbO-sIWhRfmBEKirzchaPIv-u0o4,1364
 warn/scrapers/fl.py,sha256=YJ6Qt-jJZ7_iUKlHDaQuaV2gRmae8AJKS5dwwChadBE,9563
 warn/scrapers/ga.py,sha256=o_OF4zPQ3vJM8USQPD7l_ThyRWAzUZkwrwLHCvWmHMI,7429
@@ -65,9 +65,9 @@ warn/scrapers/va.py,sha256=hOPuiAjnTmtXCOdnBM_jAJuz9_u6oCxtbm2F-9m3ot0,10732
 warn/scrapers/vt.py,sha256=d-bo4WK2hkrk4BhCCmLpEovcoZltlvdIUB6O0uaMx5A,1186
 warn/scrapers/wa.py,sha256=UXdVtHZo_a-XfoiyOooTRfTb9W3PErSZdKca6SRORgs,4282
 warn/scrapers/wi.py,sha256=ClEzXkwZbop0W4fkQgsb5oHAPUrb4luUPGV-jOKwkcg,4855
-warn_scraper-1.2.105.dist-info/licenses/LICENSE,sha256=ZV-QHyqPwyMuwuj0lI05JeSjV1NyzVEk8Yeu7FPtYS0,585
-warn_scraper-1.2.105.dist-info/METADATA,sha256=vXzxe3HPaiZ9IjDDSA__ROWblz2T4jWKgjP_50v6XYQ,2385
-warn_scraper-1.2.105.dist-info/WHEEL,sha256=SmOxYU7pzNKBqASvQJ7DjX3XGUF92lrGhMb3R6_iiqI,91
-warn_scraper-1.2.105.dist-info/entry_points.txt,sha256=poh_oSweObGlBSs1_2qZmnTodlOYD0KfO7-h7W2UQIw,47
-warn_scraper-1.2.105.dist-info/top_level.txt,sha256=gOhHgNEkrUvajlzoKkVOo-TlQht9MoXnKOErjzqLGHo,11
-warn_scraper-1.2.105.dist-info/RECORD,,
+warn_scraper-1.2.107.dist-info/licenses/LICENSE,sha256=ZV-QHyqPwyMuwuj0lI05JeSjV1NyzVEk8Yeu7FPtYS0,585
+warn_scraper-1.2.107.dist-info/METADATA,sha256=NYe0Bakge_0cvILQQr7jVI7xnpeHzZ2dDRYhJAMOK4Q,2385
+warn_scraper-1.2.107.dist-info/WHEEL,sha256=SmOxYU7pzNKBqASvQJ7DjX3XGUF92lrGhMb3R6_iiqI,91
+warn_scraper-1.2.107.dist-info/entry_points.txt,sha256=poh_oSweObGlBSs1_2qZmnTodlOYD0KfO7-h7W2UQIw,47
+warn_scraper-1.2.107.dist-info/top_level.txt,sha256=gOhHgNEkrUvajlzoKkVOo-TlQht9MoXnKOErjzqLGHo,11
+warn_scraper-1.2.107.dist-info/RECORD,,