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
@@ -94,7 +94,8 @@ def update(self, data: dict):
         self._add_audio_reference(speech)
 
     def _add_audio_reference(self, speech: IAudio):
-        article = self._assert_checksum_matches(speech)
+        article = self._assert_article_unchanged(speech)
+
         IPublish(speech).publish(background=False)
         if speech in IAudioReferences(article).items:
             return
@@ -108,12 +109,13 @@ def _article(self, speech: IAudio) -> IArticle:
             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',
+        last_modified = zeit.cms.workflow.interfaces.IModified(article).date_last_modified
+        last_published = zeit.cms.workflow.interfaces.IPublishInfo(article).date_last_published
+        if last_modified > last_published:
+            raise AudioReferenceError(
+                'Article %s was modified after publish. Speech %s is not referenced.',
                 article.uniqueId,
                 speech.uniqueId,
             )

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)