warn-scraper 1.2.115__tar.gz → 1.2.117__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {warn_scraper-1.2.115/warn_scraper.egg-info → warn_scraper-1.2.117}/PKG-INFO +1 -1
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/co.py +16 -2
- {warn_scraper-1.2.115 → warn_scraper-1.2.117/warn_scraper.egg-info}/PKG-INFO +1 -1
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/.devcontainer/devcontainer.json +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/.github/dependabot.yml.disabled-for-sanity +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/.github/workflows/continuous-deployment.yml +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/.github/workflows/continuous-deployment.yml.broken-tests +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/.gitignore +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/.pre-commit-config.yaml +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/LICENSE +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/MANIFEST.in +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/Makefile +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/Pipfile +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/Pipfile.lock +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/README.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/Makefile +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/_static/R42693.pdf +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/_static/gao-03-1003.pdf +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/_static/releasing-actions-finished.png +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/_static/releasing-actions-start.png +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/_static/releasing-changelog-button.png +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/_static/releasing-changelog-entered.png +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/_static/releasing-draft-button.png +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/_static/releasing-name-release.png +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/_static/releasing-name-tag.png +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/_static/releasing-publish-button.png +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/_static/releasing-pypi.png +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/_static/releasing-release-published.png +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/_static/releasing-releases-button.png +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/_static/releasing-tag-button.png +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/_templates/sources.md.tmpl +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/conf.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/contributing.rst +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/index.rst +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/make.bat +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/reference.rst +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/releasing.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/requirements.txt +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/al.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/az.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/ca.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/co.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/dc.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/de.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/ia.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/in.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/job_center.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/ks.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/md.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/me.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/mo.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/ny.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/ok.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/or.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/sc.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/tx.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/ut.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/va.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/vt.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/scrapers/wi.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/sources.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/docs/usage.md +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/setup.cfg +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/setup.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/tests/__init__.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/tests/cassettes/test_cached_detail_pages.yaml +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/tests/cassettes/test_cached_search_results.yaml +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/tests/cassettes/test_missing_detail_page_values.yaml +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/tests/cassettes/test_no_results.yaml +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/tests/cassettes/test_paged_results.yaml +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/tests/cassettes/test_scrape_integration.yaml +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/tests/conftest.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/tests/fixtures/2021_page_1.html +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/tests/fixtures/2021_page_2.html +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/tests/test_cache.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/tests/test_delete.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/tests/test_job_center.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/tests/test_job_center_cache.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/tests/test_openpyxl.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/__init__.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/cache.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/cli.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/platforms/__init__.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/platforms/job_center/__init__.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/platforms/job_center/cache.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/platforms/job_center/site.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/platforms/job_center/urls.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/platforms/job_center/utils.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/runner.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/__init__.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/ak.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/al.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/az.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/ca.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/ct.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/dc.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/de.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/fl.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/ga.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/hi.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/ia.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/id.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/il.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/in.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/ks.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/ky.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/la.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/md.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/me.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/mi.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/mo.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/mt.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/ne.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/nj.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/nm.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/ny.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/oh.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/ok.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/or.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/ri.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/sc.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/sd.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/tn.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/tx.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/ut.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/va.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/vt.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/wa.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/scrapers/wi.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn/utils.py +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn_scraper.egg-info/SOURCES.txt +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn_scraper.egg-info/dependency_links.txt +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn_scraper.egg-info/entry_points.txt +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn_scraper.egg-info/not-zip-safe +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn_scraper.egg-info/requires.txt +0 -0
- {warn_scraper-1.2.115 → warn_scraper-1.2.117}/warn_scraper.egg-info/top_level.txt +0 -0
@@ -1,6 +1,6 @@
|
|
1
1
|
Metadata-Version: 2.4
|
2
2
|
Name: warn-scraper
|
3
|
-
Version: 1.2.
|
3
|
+
Version: 1.2.117
|
4
4
|
Summary: Command-line interface for downloading WARN Act notices of qualified plant closings and mass layoffs from state government websites
|
5
5
|
Home-page: https://github.com/biglocalnews/warn-scraper
|
6
6
|
Author: Big Local News
|
@@ -184,6 +184,7 @@ def scrape(
|
|
184
184
|
"@dropdown": "dropdown",
|
185
185
|
"Received": "received_date",
|
186
186
|
"Notes": "notes",
|
187
|
+
"12/1/25`": "company",
|
187
188
|
# Only add new matches above here, not below here.
|
188
189
|
}
|
189
190
|
|
@@ -263,6 +264,13 @@ def scrape(
|
|
263
264
|
for row in cleaned_data:
|
264
265
|
row_dict = {}
|
265
266
|
mangled = []
|
267
|
+
|
268
|
+
line = {}
|
269
|
+
for key in row:
|
270
|
+
if not key.startswith("blank_cell_"):
|
271
|
+
line[key] = row[key]
|
272
|
+
row = line
|
273
|
+
|
266
274
|
for key in row:
|
267
275
|
if (
|
268
276
|
key not in header_crosswalk and key not in header_garbage
|
@@ -321,13 +329,19 @@ def scrape_google_sheets(table, header_list=None):
|
|
321
329
|
# Parse the header row into a list,
|
322
330
|
# preserving its order in the sheet
|
323
331
|
header_list = []
|
332
|
+
blanks = 0
|
324
333
|
for cellindex, cell in enumerate(header_soup.find_all("td")):
|
325
334
|
cell_text = cell.text.strip()
|
326
|
-
#
|
335
|
+
# Handle empty header cells
|
327
336
|
if cell_text:
|
328
337
|
header_list.append(cell_text)
|
329
|
-
|
338
|
+
elif not cell_text and cellindex == 0:
|
330
339
|
header_list.append("Company Name")
|
340
|
+
elif not cell_text and cellindex > 0:
|
341
|
+
blanks += 1
|
342
|
+
cell_text = f"blank_cell_{blanks}"
|
343
|
+
logger.debug(f"Adding {cell_text}")
|
344
|
+
header_list.append(cell_text)
|
331
345
|
|
332
346
|
# Loop through all the data rows, which start
|
333
347
|
# after the header and the little bar
|
@@ -1,6 +1,6 @@
|
|
1
1
|
Metadata-Version: 2.4
|
2
2
|
Name: warn-scraper
|
3
|
-
Version: 1.2.
|
3
|
+
Version: 1.2.117
|
4
4
|
Summary: Command-line interface for downloading WARN Act notices of qualified plant closings and mass layoffs from state government websites
|
5
5
|
Home-page: https://github.com/biglocalnews/warn-scraper
|
6
6
|
Author: Big Local News
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
{warn_scraper-1.2.115 → warn_scraper-1.2.117}/tests/cassettes/test_cached_search_results.yaml
RENAMED
File without changes
|
{warn_scraper-1.2.115 → warn_scraper-1.2.117}/tests/cassettes/test_missing_detail_page_values.yaml
RENAMED
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|