warn-scraper 1.2.76__py3-none-any.whl → 1.2.78__py3-none-any.whl

warn/scrapers/ca.py

@@ -138,23 +138,24 @@ def _extract_excel_data(wb_path):
     # Loop through all the rows
     payload = []
     for row in rows:
-        first_cell = row[0].value.strip().lower()
-        # Exit if we've reached summary row at bottom
-        if first_cell == "report summary":
-            break
-
-        data = {
-            "county": row[0].value.strip(),
-            "notice_date": _convert_date(row[1].value),
-            "received_date": _convert_date(row[2].value),
-            "effective_date": _convert_date(row[3].value),
-            "company": row[4].value.strip(),
-            "layoff_or_closure": row[5].value.strip(),
-            "num_employees": row[num_employees_index].value,
-            "address": row[address_index].value.strip(),
-            "source_file": str(wb_path).split("/")[-1],
-        }
-        payload.append(data)
+        if row[0].value:
+            first_cell = row[0].value.strip().lower()
+            # Exit if we've reached summary row at bottom
+            if first_cell == "report summary":
+                break
+
+            data = {
+                "county": row[0].value.strip(),
+                "notice_date": _convert_date(row[1].value),
+                "received_date": _convert_date(row[2].value),
+                "effective_date": _convert_date(row[3].value),
+                "company": row[4].value.strip(),
+                "layoff_or_closure": row[5].value.strip(),
+                "num_employees": row[num_employees_index].value,
+                "address": row[address_index].value.strip(),
+                "source_file": str(wb_path).split("/")[-1],
+            }
+            payload.append(data)
     return payload
 
 

warn/scrapers/id.py

@@ -53,7 +53,7 @@ def scrape(
     # Start finding the link before "Who to contact"
     html = r.text
     localizedhtml = html.split("<h2>Who to contact")[0]
-    soup = BeautifulSoup(localizedhtml, features="lxml")
+    soup = BeautifulSoup(localizedhtml, features="html5lib")
     last_url = soup.find_all("a")[-1]["href"]
     pdf_url = f"{base_url}{last_url}"
 

{warn_scraper-1.2.76.dist-info → warn_scraper-1.2.78.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: warn-scraper
-Version: 1.2.76
+Version: 1.2.78
 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-1.2.76.dist-info → warn_scraper-1.2.78.dist-info}/RECORD

@@ -28,7 +28,7 @@ warn/scrapers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 warn/scrapers/ak.py,sha256=h7BYMTV0whwWAPhbzVDVKMMoVCFphKly70aiTHabPq4,1847
 warn/scrapers/al.py,sha256=D0rT9GQ0vwfkRuveVAt-Po-T6b2TI1EPGeLOBy2m3_M,2240
 warn/scrapers/az.py,sha256=elGbue01Gjf_DQ66Wy9qqGIOJsiY-KIKJOVeft8pCXg,1447
-warn/scrapers/ca.py,sha256=rBTB-6LmNIlbGCqrCtI3O-w2e_0kcVSFxvjvh4EHBlk,8511
+warn/scrapers/ca.py,sha256=VQOfjHXPCc-jYwh-EPGVVfnzvXB7pdmCt2uJ6QnMPRM,8600
 warn/scrapers/co.py,sha256=g076Zqe8XA8tbW03HP6-03mJV8fft1niHfa5Sy6me9A,7388
 warn/scrapers/ct.py,sha256=HLMmBSFhT5Y3vZQUwRyCTxiG5BMQXTfG3SEj5rkQEL4,4771
 warn/scrapers/dc.py,sha256=_sHLnVqK_W90QqJb_W88yDlgPjoMl63LYZP3CJfdN9g,4484
@@ -37,7 +37,7 @@ warn/scrapers/fl.py,sha256=YJ6Qt-jJZ7_iUKlHDaQuaV2gRmae8AJKS5dwwChadBE,9563
 warn/scrapers/ga.py,sha256=iuvXoP5nPEnjm8wiEWhtBwLnkdmElhJILin3No5HIyo,7412
 warn/scrapers/hi.py,sha256=pSplAP15_ZBfQtcywyErmvNcrk7u55TjZj_F0Nqw9L8,5660
 warn/scrapers/ia.py,sha256=zOncaA9M0d6paT4pB7UU_4D_yxUgeUiGRcnpKi9DsRA,1999
-warn/scrapers/id.py,sha256=rJBoJGpREP72ovZnxeulkdX_umEi4aEbcZUW3v0VqlA,5853
+warn/scrapers/id.py,sha256=FldJHv8sqBVKIdq7YlOK_O3Uh2Xj9i0WL7wQgszBgTA,5857
 warn/scrapers/il.py,sha256=sygdvsNuB_Gvu3o_HidtpSP4FLz0szKb1zEHqGxVtlI,1563
 warn/scrapers/in.py,sha256=dAT40ROhhKiwLcwa_YJ6EyhsYBLe0IX2rOWXmNa6JMs,2026
 warn/scrapers/ks.py,sha256=F_3biEMF7zgCX2XVuUACR74Vyzapta4SaM9SY3EuZCU,1266
@@ -65,9 +65,9 @@ warn/scrapers/va.py,sha256=13lhkQrSkPGHEiWUuf1qiS890PWYE5gV-TgISpoiQnc,1711
 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-1.2.76.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-warn_scraper-1.2.76.dist-info/METADATA,sha256=Adb8KydFcC46yyPoAb06yPV33sEgYxMiQRmx6Z2k9jA,2025
-warn_scraper-1.2.76.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
-warn_scraper-1.2.76.dist-info/entry_points.txt,sha256=poh_oSweObGlBSs1_2qZmnTodlOYD0KfO7-h7W2UQIw,47
-warn_scraper-1.2.76.dist-info/top_level.txt,sha256=gOhHgNEkrUvajlzoKkVOo-TlQht9MoXnKOErjzqLGHo,11
-warn_scraper-1.2.76.dist-info/RECORD,,
+warn_scraper-1.2.78.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+warn_scraper-1.2.78.dist-info/METADATA,sha256=6GsM9rlqqDjF5QQ5eVR_MJXNH8JJd3lji3jFbDeudYc,2025
+warn_scraper-1.2.78.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
+warn_scraper-1.2.78.dist-info/entry_points.txt,sha256=poh_oSweObGlBSs1_2qZmnTodlOYD0KfO7-h7W2UQIw,47
+warn_scraper-1.2.78.dist-info/top_level.txt,sha256=gOhHgNEkrUvajlzoKkVOo-TlQht9MoXnKOErjzqLGHo,11
+warn_scraper-1.2.78.dist-info/RECORD,,