wip: date changes

src/caselawclient/models/documents.py

@@ -33,6 +33,7 @@
 )
 
 MINIMUM_ENRICHMENT_TIME = datetime.timedelta(minutes=20)
+MINIMUM_REPARSE_TIME = datetime.timedelta(minutes=20)
 
 
 class UnparsableDate(Warning):
@@ -285,6 +286,24 @@ def enrichment_datetime(self) -> Optional[datetime.datetime]:
         """When was this document successfully enriched (date from XML)"""
         return self.get_latest_manifestation_datetime("tna-enriched")
 
+    @cached_property
+    def request_reparse_datetime(self) -> Optional[datetime.datetime]:
+        """When was a reparse request last sent to TDR?"""
+        date_as_string = self.api_client.get_property(self.uri, "last_sent_to_parser")
+        try:
+            return datetime.datetime.fromisoformat(date_as_string)
+        except ValueError:
+            return None
+
+    @cached_property
+    def request_enrich_datetime(self) -> Optional[datetime.datetime]:
+        """When was a reparse request last sent to TDR?"""
+        date_as_string = self.api_client.get_property(self.uri, "last_sent_to_parser")
+        try:
+            return datetime.datetime.fromisoformat(date_as_string)
+        except ValueError:
+            return None
+
     @cached_property
     def is_published(self) -> bool:
         return self.api_client.get_published(self.uri)
@@ -525,9 +544,9 @@ def can_enrich(self) -> bool:
     @cached_property
     def enriched_recently(self) -> bool:
         """
-        Has this document been enriched recently?
+        Has this document been sent for enrichment recently?
         """
-        last_enrichment = self.enrichment_datetime
+        last_enrichment = self.request_enrich_datetime
         now = datetime.datetime.now(tz=datetime.timezone.utc)
         if last_enrichment and now - last_enrichment < MINIMUM_ENRICHMENT_TIME:
             return True
@@ -645,7 +664,18 @@ def can_reparse(self) -> bool:
         """
         Is it sensible to reparse this document?
         """
-        if self.docx_exists():
+        if (self.reparsed_recently is False) and self.docx_exists():
+            return True
+        return False
+
+    @cached_property
+    def reparsed_recently(self) -> bool:
+        """
+        Has this document been sent to reparsing recently?
+        """
+        last_reparse = self.request_reparse_datetime
+        now = datetime.datetime.now(tz=datetime.timezone.utc)
+        if last_reparse and now - last_reparse < MINIMUM_REPARSE_TIME:
             return True
         return False
 

tests/models/test_documents.py

@@ -427,7 +427,7 @@ def test_unpublish(
         )
 
 
-class TestDocumentEnrichedRecently:
+class TestDocumentEnrichedOrReparsedRecently:
     def test_enriched_recently_returns_false_when_never_enriched(self, mock_api_client):
         document = Document("test/1234", mock_api_client)
         mock_api_client.get_property.return_value = ""
@@ -436,20 +436,42 @@ def test_enriched_recently_returns_false_when_never_enriched(self, mock_api_clie
 
     def test_enriched_recently_returns_true_within_cooldown(self, mock_api_client):
         document = Document("test/1234", mock_api_client)
-        document.enrichment_datetime = datetime.datetime.now(
+        document.request_enrich_datetime = datetime.datetime.now(
             tz=datetime.timezone.utc
         ) - datetime.timedelta(seconds=30)
 
         assert document.enriched_recently is True
 
     def test_enriched_recently_returns_false_outside_cooldown(self, mock_api_client):
         document = Document("test/1234", mock_api_client)
-        document.enrichment_datetime = datetime.datetime.now(
+        document.request_enrich_datetime = datetime.datetime.now(
             tz=datetime.timezone.utc
         ) - datetime.timedelta(days=2)
 
         assert document.enriched_recently is False
 
+    def test_reparsed_recently_returns_false_when_never_enriched(self, mock_api_client):
+        document = Document("test/1234", mock_api_client)
+        mock_api_client.get_property.return_value = ""
+
+        assert document.reparsed_recently is False
+
+    def test_reparsed_recently_returns_true_within_cooldown(self, mock_api_client):
+        document = Document("test/1234", mock_api_client)
+        document.request_reparse_datetime = datetime.datetime.now(
+            tz=datetime.timezone.utc
+        ) - datetime.timedelta(seconds=30)
+
+        assert document.reparsed_recently is True
+
+    def test_reparsed_recently_returns_false_outside_cooldown(self, mock_api_client):
+        document = Document("test/1234", mock_api_client)
+        document.request_reparse_datetime = datetime.datetime.now(
+            tz=datetime.timezone.utc
+        ) - datetime.timedelta(days=2)
+
+        assert document.reparsed_recently is False
+
 
 class TestCanEnrich:
     @pytest.mark.parametrize(
@@ -493,6 +515,46 @@ def test_returns_true_when_enriched_recently_is_true_and_validates_against_schem
                 assert document.can_enrich is can_enrich
 
 
+class TestCanReparse:
+    @pytest.mark.parametrize(
+        "reparsed_recently, docx_exists, can_reparse",
+        [
+            (
+                True,
+                True,
+                False,
+            ),  # Reparsed recently and docx exists - Can't reparse
+            (
+                True,
+                False,
+                False,
+            ),  # Reparsed recently and no docx- Can't reparse
+            (
+                False,
+                False,
+                False,
+            ),  # Not reparsed recently and no docx - Can't reparse
+            (
+                False,
+                True,
+                True,
+            ),  # Not reparsed recently and docx exists - Can reparse
+        ],
+    )
+    def test_can_reparse_logic(
+        self, mock_api_client, reparsed_recently, docx_exists, can_reparse
+    ):
+        document = Document("test/1234", mock_api_client)
+        with patch.object(
+            Document, "reparsed_recently", new_callable=PropertyMock
+        ) as mock_reparsed_recently:
+            with patch.object(Document, "docx_exists") as mock_docx_exists:
+                mock_reparsed_recently.return_value = reparsed_recently
+                mock_docx_exists.return_value = docx_exists
+
+                assert document.can_reparse is can_reparse
+
+
 class TestDocumentEnrich:
     @time_machine.travel(datetime.datetime(1955, 11, 5, 6))
     @patch("caselawclient.models.documents.announce_document_event")