warn-scraper 1.2.92__tar.gz → 1.2.94__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.
Files changed (135) hide show
  1. {warn_scraper-1.2.92/warn_scraper.egg-info → warn_scraper-1.2.94}/PKG-INFO +1 -1
  2. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/co.py +6 -1
  3. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/utils.py +5 -3
  4. {warn_scraper-1.2.92 → warn_scraper-1.2.94/warn_scraper.egg-info}/PKG-INFO +1 -1
  5. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/.devcontainer/devcontainer.json +0 -0
  6. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/.github/dependabot.yml.disabled +0 -0
  7. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/.github/workflows/continuous-deployment.yml +0 -0
  8. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/.gitignore +0 -0
  9. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/.pre-commit-config.yaml +0 -0
  10. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/LICENSE +0 -0
  11. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/MANIFEST.in +0 -0
  12. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/Makefile +0 -0
  13. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/Pipfile +0 -0
  14. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/Pipfile.lock +0 -0
  15. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/README.md +0 -0
  16. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/Makefile +0 -0
  17. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/_static/R42693.pdf +0 -0
  18. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/_static/gao-03-1003.pdf +0 -0
  19. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/_static/releasing-actions-finished.png +0 -0
  20. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/_static/releasing-actions-start.png +0 -0
  21. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/_static/releasing-changelog-button.png +0 -0
  22. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/_static/releasing-changelog-entered.png +0 -0
  23. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/_static/releasing-draft-button.png +0 -0
  24. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/_static/releasing-name-release.png +0 -0
  25. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/_static/releasing-name-tag.png +0 -0
  26. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/_static/releasing-publish-button.png +0 -0
  27. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/_static/releasing-pypi.png +0 -0
  28. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/_static/releasing-release-published.png +0 -0
  29. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/_static/releasing-releases-button.png +0 -0
  30. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/_static/releasing-tag-button.png +0 -0
  31. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/_templates/sources.md.tmpl +0 -0
  32. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/conf.py +0 -0
  33. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/contributing.rst +0 -0
  34. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/index.rst +0 -0
  35. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/make.bat +0 -0
  36. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/reference.rst +0 -0
  37. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/releasing.md +0 -0
  38. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/requirements.txt +0 -0
  39. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/al.md +0 -0
  40. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/az.md +0 -0
  41. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/ca.md +0 -0
  42. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/co.md +0 -0
  43. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/dc.md +0 -0
  44. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/de.md +0 -0
  45. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/ia.md +0 -0
  46. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/in.md +0 -0
  47. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/job_center.md +0 -0
  48. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/ks.md +0 -0
  49. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/md.md +0 -0
  50. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/me.md +0 -0
  51. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/mo.md +0 -0
  52. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/ny.md +0 -0
  53. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/ok.md +0 -0
  54. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/or.md +0 -0
  55. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/sc.md +0 -0
  56. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/tx.md +0 -0
  57. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/ut.md +0 -0
  58. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/va.md +0 -0
  59. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/vt.md +0 -0
  60. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/scrapers/wi.md +0 -0
  61. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/sources.md +0 -0
  62. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/docs/usage.md +0 -0
  63. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/setup.cfg +0 -0
  64. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/setup.py +0 -0
  65. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/tests/__init__.py +0 -0
  66. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/tests/cassettes/test_cached_detail_pages.yaml +0 -0
  67. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/tests/cassettes/test_cached_search_results.yaml +0 -0
  68. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/tests/cassettes/test_missing_detail_page_values.yaml +0 -0
  69. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/tests/cassettes/test_no_results.yaml +0 -0
  70. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/tests/cassettes/test_paged_results.yaml +0 -0
  71. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/tests/cassettes/test_scrape_integration.yaml +0 -0
  72. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/tests/conftest.py +0 -0
  73. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/tests/fixtures/2021_page_1.html +0 -0
  74. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/tests/fixtures/2021_page_2.html +0 -0
  75. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/tests/test_cache.py +0 -0
  76. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/tests/test_delete.py +0 -0
  77. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/tests/test_job_center.py +0 -0
  78. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/tests/test_job_center_cache.py +0 -0
  79. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/tests/test_openpyxl.py +0 -0
  80. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/__init__.py +0 -0
  81. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/cache.py +0 -0
  82. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/cli.py +0 -0
  83. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/platforms/__init__.py +0 -0
  84. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/platforms/job_center/__init__.py +0 -0
  85. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/platforms/job_center/cache.py +0 -0
  86. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/platforms/job_center/site.py +0 -0
  87. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/platforms/job_center/urls.py +0 -0
  88. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/platforms/job_center/utils.py +0 -0
  89. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/runner.py +0 -0
  90. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/__init__.py +0 -0
  91. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/ak.py +0 -0
  92. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/al.py +0 -0
  93. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/az.py +0 -0
  94. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/ca.py +0 -0
  95. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/ct.py +0 -0
  96. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/dc.py +0 -0
  97. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/de.py +0 -0
  98. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/fl.py +0 -0
  99. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/ga.py +0 -0
  100. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/hi.py +0 -0
  101. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/ia.py +0 -0
  102. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/id.py +0 -0
  103. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/il.py +0 -0
  104. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/in.py +0 -0
  105. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/ks.py +0 -0
  106. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/ky.py +0 -0
  107. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/la.py +0 -0
  108. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/md.py +0 -0
  109. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/me.py +0 -0
  110. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/mi.py +0 -0
  111. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/mo.py +0 -0
  112. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/mt.py +0 -0
  113. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/ne.py +0 -0
  114. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/nj.py +0 -0
  115. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/nm.py +0 -0
  116. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/ny.py +0 -0
  117. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/oh.py +0 -0
  118. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/ok.py +0 -0
  119. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/or.py +0 -0
  120. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/ri.py +0 -0
  121. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/sc.py +0 -0
  122. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/sd.py +0 -0
  123. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/tn.py +0 -0
  124. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/tx.py +0 -0
  125. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/ut.py +0 -0
  126. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/va.py +0 -0
  127. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/vt.py +0 -0
  128. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/wa.py +0 -0
  129. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn/scrapers/wi.py +0 -0
  130. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn_scraper.egg-info/SOURCES.txt +0 -0
  131. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn_scraper.egg-info/dependency_links.txt +0 -0
  132. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn_scraper.egg-info/entry_points.txt +0 -0
  133. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn_scraper.egg-info/not-zip-safe +0 -0
  134. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn_scraper.egg-info/requires.txt +0 -0
  135. {warn_scraper-1.2.92 → warn_scraper-1.2.94}/warn_scraper.egg-info/top_level.txt +0 -0
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: warn-scraper
3
- Version: 1.2.92
3
+ Version: 1.2.94
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
@@ -157,7 +157,12 @@ def scrape(
157
157
  for key, value in row.items():
158
158
  standardized_key = header_crosswalk[key]
159
159
  row_dict[standardized_key] = value
160
- standardized_data.append(row_dict)
160
+ if len(row_dict["company"]) < 3 and row_dict["letter"] == "Avis Budget Group":
161
+ row_dict["company"] = "Avis Budget Group"
162
+ if len(row_dict["company"]) < 3: # or len(row_dict['naics']) <5:
163
+ logger.debug(f"Dropping row of questionable quality: {row_dict}")
164
+ else:
165
+ standardized_data.append(row_dict)
161
166
 
162
167
  # Set the path to the final CSV
163
168
  output_csv = data_dir / "co.csv"
@@ -86,7 +86,7 @@ def save_if_good_url(filename, url, **kwargs):
86
86
  success_flag = False
87
87
  content = False
88
88
  else:
89
- with open(filename, "wb", encoding="utf-8") as outfile:
89
+ with open(filename, "wb") as outfile:
90
90
  outfile.write(response.content)
91
91
  success_flag = True
92
92
  content = response.content
@@ -109,7 +109,9 @@ def write_rows_to_csv(output_path: Path, rows: list, mode="w"):
109
109
  writer.writerows(rows)
110
110
 
111
111
 
112
- def write_dict_rows_to_csv(output_path, headers, rows, mode="w", extrasaction="raise"):
112
+ def write_dict_rows_to_csv(
113
+ output_path, headers, rows, mode="w", extrasaction="raise", encoding="utf-8"
114
+ ):
113
115
  """Write the provided dictionary to the provided path as comma-separated values.
114
116
 
115
117
  Args:
@@ -121,7 +123,7 @@ def write_dict_rows_to_csv(output_path, headers, rows, mode="w", extrasaction="r
121
123
  """
122
124
  create_directory(output_path, is_file=True)
123
125
  logger.debug(f"Writing {len(rows)} rows to {output_path}")
124
- with open(output_path, mode, newline="") as f:
126
+ with open(output_path, mode, newline="", encoding=encoding) as f:
125
127
  # Create the writer object
126
128
  writer = csv.DictWriter(f, fieldnames=headers, extrasaction=extrasaction)
127
129
  # If we are writing a new row ...
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: warn-scraper
3
- Version: 1.2.92
3
+ Version: 1.2.94
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