Merge pull request #619 from ZeitOnline/ZO-4683_article_tts_reference

ZO-4683: Check date_last_modified instead of checksum

core/docs/changelog/ZO-4683.change

@@ -0,0 +1 @@
+ZO-4683: Only publish article after referencing tts audio if it is unchanged

core/src/zeit/connector/testcontent/online/2022/08/trockenheit

@@ -22,6 +22,7 @@
     <attribute ns="http://namespaces.zeit.de/CMS/workflow" name="corrected">yes</attribute>
     <attribute ns="http://namespaces.zeit.de/CMS/document" name="date_first_released">2022-08-09T06:11:30+00:00</attribute>
     <attribute ns="http://namespaces.zeit.de/CMS/document" name="date_last_checkout">2022-08-10T21:35:24.546250+00:00</attribute>
+    <attribute ns="http://namespaces.zeit.de/CMS/document" name="date_last_modified">2022-08-10T21:35:24.546250+00:00</attribute>
     <attribute ns="http://namespaces.zeit.de/CMS/workflow" name="date_last_published">2022-08-10T21:35:23.950553+00:00</attribute>
     <attribute ns="http://namespaces.zeit.de/CMS/workflow" name="date_last_published_semantic">2022-08-09T06:11:30+00:00</attribute>
     <attribute ns="http://namespaces.zeit.de/CMS/workflow" name="edited">yes</attribute>

core/src/zeit/speech/connection.py

@@ -12,10 +12,10 @@
 from zeit.cms.repository.interfaces import IFolder
 from zeit.cms.workflow.interfaces import IPublish, IPublishInfo
 from zeit.connector.search import SearchVar
-from zeit.content.article.interfaces import IArticle, ISpeechbertChecksum
+from zeit.content.article.interfaces import IArticle
 from zeit.content.audio.audio import AUDIO_SCHEMA_NS, Audio
 from zeit.content.audio.interfaces import IAudio, IAudioReferences, ISpeechInfo
-from zeit.speech.errors import ChecksumMismatchError
+from zeit.speech.errors import AudioReferenceError
 import zeit.cms.interfaces
 import zeit.cms.repository.folder
 import zeit.speech.interfaces
@@ -57,7 +57,7 @@ def _create(self, data: dict) -> IAudio:
         speech.audio_type = 'tts'
         folder = self._get_target_folder(data['uuid'])
         folder[data['uuid']] = speech
-        log.info('Text-to-speech was created for article uuid %s', data['uuid'])
+        log.info('Created %s for article uuid %s', speech, data['uuid'])
         self._update(data, folder[data['uuid']])
         return folder[data['uuid']]
 
@@ -73,15 +73,15 @@ def _update(self, data: dict, speech: IAudio):
                 elif audio['type'] == 'PREVIEW_TTS':
                     ISpeechInfo(co).preview_url = audio_entry['url']
             ISemanticChange(co).last_semantic_change = datetime.now(pytz.UTC)
-        log.info('Text-to-speech %s was updated for article uuid %s', speech.uniqueId, data['uuid'])
+        log.info('Updated %s for article uuid %s', speech, data['uuid'])
 
     def _find(self, article_uuid: str) -> Optional[IAudio]:
         connector = zope.component.getUtility(zeit.connector.interfaces.IConnector)
         result = list(connector.search([AUDIO_ID], (AUDIO_ID == str(article_uuid))))
         if not result:
             return None
         content = zeit.cms.interfaces.ICMSContent(result[0][0])
-        log.debug('Text-to-speech %s found for %s.', content.uniqueId, article_uuid)
+        log.debug('%s found for article uuid %s.', content, article_uuid)
         return content
 
     def update(self, data: dict):
@@ -94,36 +94,39 @@ def update(self, data: dict):
         self._add_audio_reference(speech)
 
     def _add_audio_reference(self, speech: IAudio):
-        article = self._assert_checksum_matches(speech)
         IPublish(speech).publish(background=False)
+
+        article = self._assert_article_unchanged(speech)
         if speech in IAudioReferences(article).items:
+            log.debug('%s already references %s', article, speech)
             return
         with checked_out(article, raise_if_error=True) as co:
             references = IAudioReferences(co)
             references.add(speech)
+        log.info('Added reference from %s to %s', article, speech)
         IPublish(article).publish(background=False)
 
     def _article(self, speech: IAudio) -> IArticle:
         return zeit.cms.interfaces.ICMSContent(
             zeit.cms.content.interfaces.IUUID(ISpeechInfo(speech).article_uuid), None
         )
 
-    def _assert_checksum_matches(self, speech: IAudio) -> IArticle:
+    def _assert_article_unchanged(self, speech: IAudio) -> IArticle:
         article = self._article(speech)
-        article_checksum = ISpeechbertChecksum(article).calculate()
-        if article_checksum != ISpeechInfo(speech).checksum:
-            raise ChecksumMismatchError(
-                'Speechbert checksum mismatch for article %s and speech %s',
-                article.uniqueId,
-                speech.uniqueId,
-            )
-        return article
+        pub_status = zeit.cms.workflow.interfaces.IPublicationStatus(article).published
+        if pub_status == 'published':
+            return article
+        raise AudioReferenceError(
+            '%s was modified after publish. Skipped adding reference %s.',
+            article,
+            speech,
+        )
 
     def _remove_reference_from_article(self, speech: IAudio):
         article = self._article(speech)
         if not article:
             log.warning(
-                'No article found for Text-to-speech %s. ' 'Maybe it was already deleted?',
+                'No article found for %s. Maybe it was already deleted?',
                 speech,
             )
             return
@@ -135,12 +138,12 @@ def delete(self, data: dict):
         speech = self._find(data['article_uuid'])
         if not speech:
             log.warning(
-                'No Text-to-speech found for article uuid %s. '
+                'No audio object found for article uuid %s. '
                 'Maybe it was already deleted?' % data['article_uuid'],
             )
             return
         self._remove_reference_from_article(speech)
         IPublish(speech).retract(background=False)
         unique_id = speech.uniqueId
         del speech.__parent__[speech.__name__]
-        log.info('Text-to-speech %s successfully deleted.', unique_id)
+        log.info('Deleted %s', unique_id)

core/src/zeit/speech/errors.py

@@ -1,2 +1,5 @@
-class ChecksumMismatchError(Exception):
-    """An exception raised when the checksum of the article and the speech do not match."""
+class AudioReferenceError(Exception):
+    """
+    An exception raised if the article was updated after publish, to avoid publishing it with
+    unreviewed changes
+    """

core/src/zeit/speech/tests/test_connection.py

@@ -10,7 +10,7 @@
 from zeit.content.audio.interfaces import IAudioReferences, ISpeechInfo
 from zeit.content.audio.testing import AudioBuilder
 from zeit.speech.connection import Speech
-from zeit.speech.errors import ChecksumMismatchError
+from zeit.speech.errors import AudioReferenceError
 from zeit.speech.testing import TTS_CREATED, TTS_DELETED, FunctionalTestCase
 import zeit.cms.checkout.interfaces
 import zeit.cms.workflow.mock
@@ -87,23 +87,29 @@ def test_update_audio_without_touching_the_article(self):
         assert zeit.cms.workflow.mock._publish_count[article.uniqueId] == 2
         assert zeit.cms.workflow.mock._publish_count[audio.uniqueId] == 2
 
-    def test_if_checksum_does_not_match_do_not_add_reference(self):
-        tts_msg = self.setup_speech_message('checksum', 'cake')
-        with pytest.raises(ChecksumMismatchError):
-            self.create_audio(tts_msg)
+    def test_if_article_changed_do_not_add_reference(self):
+        IPublish(self.article).publish(background=False)
+        with checked_out(self.article) as co:
+            paragraph = co.body.create_item('p')
+            paragraph.text = 'the article has changed'
+        with pytest.raises(AudioReferenceError):
+            self.create_audio(TTS_CREATED)
         article = ICMSContent(self.article_uid)
         reference = IAudioReferences(article)
         assert not reference.items
 
-    def test_update_audio_fails_because_checksum_is_not_matching(self):
+    def test_update_audio_fails_if_article_changed(self):
         audio = self.create_audio(TTS_CREATED)
-        assert zeit.cms.workflow.mock._publish_count[audio.uniqueId] == 1
-
-        tts_msg = self.setup_speech_message('checksum', 'cake')
+        article = ICMSContent(self.article_uid)
+        reference = IAudioReferences(article)
+        assert audio in reference.items
+        with checked_out(self.article) as co:
+            paragraph = co.body.create_item('p')
+            paragraph.text = 'the article has changed'
         self.repository.connector.search_result = [(self.article.uniqueId)]
         with mock.patch('zeit.speech.connection.Speech._find', return_value=audio):
-            with pytest.raises(ChecksumMismatchError):
-                Speech().update(tts_msg)
+            with pytest.raises(AudioReferenceError):
+                Speech().update(TTS_CREATED)
 
     def test_handle_delete_event(self):
         audio = self.create_audio(TTS_CREATED)
@@ -137,5 +143,5 @@ def test_unable_to_remove_anything_because_article_is_missing(self):
         self.repository.connector.search_result = []
         with mock.patch('zeit.speech.connection.Speech._find', return_value=audio):
             Speech().delete(TTS_DELETED)
-        assert f'No article found for Text-to-speech {audio}' in self.caplog.text
-        assert not self.repository.has_key('audio')
+        assert f'No article found for {audio}' in self.caplog.text
+        assert 'audio' not in self.repository