Mock mimetype test when run in github action. Python image for github action does not support mimetype

Author: patkyn

.github/workflows/publish-release.yml

@@ -35,7 +35,7 @@ jobs:
         run: |
           echo ${{ github.workspace }}
           cd ${{ github.workspace }}/tests
-          poetry run pytest
+          poetry run pytest --github-action-run=True
       - name: Build
         id: build-step
         run: |

.github/workflows/publish-test.yml

@@ -33,7 +33,7 @@ jobs:
         run: |
           echo ${{ github.workspace }}
           cd ${{ github.workspace }}/tests
-          poetry run pytest
+          poetry run pytest --github-action-run=True
       - name: Build
         id: build-step
         run: |

.github/workflows/run-tests.yml

@@ -48,5 +48,5 @@ jobs:
       run: |
         echo ${{ github.workspace }}
         cd ${{ github.workspace }}/tests
-        poetry run pytest
+        poetry run pytest --cov=dwcahandler --github-action-run=True
 

poetry.lock

@@ -14,6 +14,7 @@ pandas = "^2.2.0"
 requests = "^2.32.0"
 pytest = "^8.2.0"
 pytest-mock = "^3.12.0"
+pytest-cov = "^5.0.0"
 metapype = "^0.0.26"
 flake8 = "^7.1.1"
 

pyproject.toml

@@ -678,15 +678,16 @@ def get_media_type(media_format: str):
         def get_multimedia_format_type(row: dict):
             url = row['identifier']
             media_format = None
-            try:
-                mime_type = mimetypes.guess_type(url)
-                if mime_type and len(mime_type) > 0 and mime_type[0]:
-                    media_format = mime_type[0]
-            except Exception as error:
-                log.error("Error getting mimetype from url %s: %s", url, error)
+            if url:
+                try:
+                    mime_type = mimetypes.guess_type(url)
+                    if mime_type and len(mime_type) > 0 and mime_type[0]:
+                        media_format = mime_type[0]
+                except Exception as error:
+                    log.error("Error getting mimetype from url %s: %s", url, error)
 
             media_type = ''
-            if 'type' not in row or not row['type']:
+            if 'type' not in row or not row['type'] or row['type'] is nan:
                 media_type = get_media_type(media_format)
             else:
                 media_type = row['type']

src/dwcahandler/dwca/core_dwca.py

@@ -0,0 +1,5 @@
+
+def pytest_addoption(parser):
+    parser.addoption(
+        "--github-action-run", action="store", default=False, help="Set this to True if it's been called from github action"
+    )

tests/conftest.py

@@ -1,9 +1,15 @@
 import pandas as pd
+import dwcahandler
 from dwcahandler.dwca import CsvFileType, CoreOrExtType
 from dwcahandler.dwca.core_dwca import Dwca
 from operator import attrgetter
+import logging
+import pytest
 
-MIMETYPE_IMAGE_URL = 'https://www.gstatic.com/webp/gallery/1.webp'
+logging.basicConfig(level=logging.DEBUG)
+log = logging.getLogger("test_multimedia_content")
+
+MIMETYPE_IMAGE_URL = "https://www.gstatic.com/webp/gallery/1.webp"
 INVALID_MIMETYPE_URL = "https://invalid.url.jpeg"
 IMAGE_URL = "https://images.ala.org.au/image/proxyImageThumbnailLarge?imageId=a36b5634-0277-47c7-b4e3-383e24ce8d1a"
 AUDIO_URL = "https://images.ala.org.au/image/proxyImage?imageId=480f5f5e-e96c-4ae3-8230-c53a37bc542e"
@@ -20,13 +26,26 @@
                         keys=['occurrenceID'])
 
 
+def mock_guess_type(url):
+    if url == MIMETYPE_IMAGE_URL:
+        return ('image/webp', None)
+    elif url == INVALID_MIMETYPE_URL:
+        return ('image/jpeg', None)
+    return (None, None)
+
+
+@pytest.fixture
+def mock_mime_types(monkeypatch, request):
+    if request.config.getoption("--github-action-run"):
+        monkeypatch.setattr(dwcahandler.dwca.core_dwca.mimetypes, "guess_type", mock_guess_type)
+
+
 class TestMultimediaExtension:
 
     def test_extract_associate_media(self):
         """
         Test for associated media to be expanded into multimedia extension
         """
-
         occ_associated_media_df = pd.DataFrame(data=[["1", "species1", IMAGE_URL],
                                                      ["2", "species2", AUDIO_URL],
                                                      ["3", "species3", f"{VIDEO_URL}|{MIMETYPE_IMAGE_URL}"]],
@@ -60,7 +79,7 @@ def test_extract_associate_media(self):
         assert sorted(list(map(attrgetter('field_name'), dwca.meta_content.meta_elements[1].fields))) == \
                sorted(['coreid', 'identifier'])
 
-    def test_fill_additional_multimedia_info(self):
+    def test_fill_additional_multimedia_info(self, mock_mime_types):
         """
         Test for fill additional multimedia info if format and type is not provided
         :return:
@@ -88,12 +107,14 @@ def test_fill_additional_multimedia_info(self):
                                                     ["3", MIMETYPE_IMAGE_URL, 'image/webp', 'StillImage']],
                                               columns=['occurrenceID', 'identifier', 'format', 'type'])
 
+        dwca.ext_content[0].df_content.fillna('', inplace=True)
+        expected_multimedia_df.fillna('', inplace=True)
+
         # Test that the multimedia extension will now contain the format and type
         pd.testing.assert_frame_equal(dwca.ext_content[0].df_content.drop(
                                       columns=['coreid']), expected_multimedia_df)
 
-
-    def test_fill_multimedia_info_with_format_type_partially_supplied(self):
+    def test_fill_multimedia_info_with_format_type_partially_supplied(self, mock_mime_types):
         """
         Test fill_additional_multimedia_info if format or type is already present.
         Calling fill_additional_multimedia_info should not change the existing values in the content
@@ -156,8 +177,7 @@ def test_fill_multimedia_info_with_format_type_partially_supplied(self):
         pd.testing.assert_frame_equal(dwca.ext_content[0].df_content.drop(
             columns=['coreid']), expected_multimedia_df)
 
-
-    def test_fill_multimedia_info_type_from_forrmat(self):
+    def test_fill_multimedia_info_type_from_forrmat(self, mock_mime_types):
         """
         Test fill_additional_multimedia_info if only format is already present.
         Calling fill_additional_multimedia_info should not change the existing values in the content