Merge branch 'master' of github.com:openaddresses/batch-machine into …

…preview-actual

CHANGELOG

@@ -4,15 +4,17 @@
 - Rewrite confrom to natively output GeoJSON+LD
 - Rewrite Preview module to expect GeoJSON+LD
 
-2023-09-06 v8.3.0
-- Update to next major alpine version
-- Remove mock library in favour of unittest.mock
-- Update all python deps to latest versions
+2023-09-09 v8.3.1
+- Use the file extension from the content-type header when it disagrees with content-disposition in https://github.com/openaddresses/batch-machine/pull/54
+
+2023-09-05 v8.3.0
+- Update Mapbox Vector Tile dependency in https://github.com/openaddresses/batch-machine/pull/52
+- Make GeoJSON+LD output a default in https://github.com/openaddresses/batch-machine/pull/51
+- Update TippeCanoe to latest version in https://github.com/openaddresses/batch-machine/pull/50
 
 2023-09-05 v8.2.4
 - Update TippeCanoe to latest version
 - Make GeoJSON+LD output a default
-
 2023-09-02 v8.2.3
 - Fix properties in geojson-ld features in https://github.com/openaddresses/batch-machine/pull/49
 - Remove extra whitespace in geojson-ld output in https://github.com/openaddresses/batch-machine/pull/48

openaddr/cache.py

@@ -207,8 +207,11 @@ def guess_url_file_extension(url):
                     if path_ext == attachment_ext:
                         _L.debug('Content-Disposition agrees: "{}"'.format(match.group('filename')))
                     else:
-                        _L.debug('Content-Disposition disagrees: "{}"'.format(match.group('filename')))
-                        path_ext = False
+                        _L.debug('Content-Disposition disagrees: "{}" says we should use "{}", using "{}" instead'.format(
+                            match.group('filename'),
+                            attachment_ext,
+                            path_ext,
+                        ))
 
         if not path_ext:
             #

openaddr/conform.py

@@ -388,7 +388,7 @@ def convert(self, source_config, source_paths, workdir):
             rc = conform_cli(source_config, source_path, dest_path)
             if rc == 0:
                 with open(dest_path) as file:
-                    addr_count = sum(1 for line in file) - 1
+                    addr_count = sum(1 for line in file)
 
                 # Success! Return the path of the output CSV
                 return dest_path, addr_count
@@ -1083,7 +1083,7 @@ def row_convert_to_out(source_config, row):
     "Convert a row from the source schema to OpenAddresses output schema"
 
     geom = row.get(GEOM_FIELDNAME.lower(), None)
-    if geom == "POINT EMPTY":
+    if geom == "POINT EMPTY" or geom == '':
         geom = None
 
     output = {

openaddr/tests/__init__.py

@@ -989,16 +989,13 @@ def test_single_tx_waco(self):
         source = join(self.src_dir, 'us/tx/city_of_waco.json')
 
         with HTTMock(self.response_content):
-            ofs = csv.field_size_limit()
-            csv.field_size_limit(1)
             state_path = process_one.process(source, self.testdir, "addresses", "default", False, False)
-            csv.field_size_limit(ofs)
 
         with open(state_path) as file:
             state = dict(zip(*json.load(file)))
 
-        self.assertEqual(state["source problem"], "Could not conform source data")
-        self.assertIsNone(state["processed"])
+        self.assertIsNone(state["source problem"])
+        self.assertEqual(state["processed"], 'out.geojson')
 
         source = join(self.src_dir, 'us/tx/city_of_waco.json')
 
@@ -1022,7 +1019,7 @@ def test_single_tx_waco(self):
             self.assertEqual(rows[0]['properties']['region'], u'TX')
             self.assertEqual(rows[0]['properties']['id'], u'')
             self.assertEqual(rows[0]['properties']['number'], u'308')
-            self.assertEqual(rows[0]['properties']['hash'], u'c782a47f7d52d99f')
+            self.assertEqual(rows[0]['properties']['hash'], u'431f816eebac0000')
             self.assertEqual(rows[0]['properties']['city'], u'Mcgregor')
             self.assertEqual(rows[0]['geometry']['coordinates'], [-97.3961768, 31.4432706]),
             self.assertEqual(rows[0]['properties']['street'], u'PULLEN ST')