add diff ratio to the heuristic

core/tests/test_page.py

@@ -43,15 +43,17 @@ class TestEditPage:
         user = baker.make(User, is_superuser=True)
         page = baker.prepare(Page)
         page.save(force_lock=True)
-        first_rev = baker.make(PageRev, author=user, page=page, date=now())
+        first_rev = baker.make(
+            PageRev, author=user, page=page, date=now(), content="Hello World"
+        )
         client.force_login(user)
         url = reverse("core:page_edit", kwargs={"page_name": page._full_name})
-        client.post(url, data={"content": "Hello World"})
+        client.post(url, data={"content": "Hello World!"})
         assert page.revisions.count() == 1
         assert page.revisions.last() == first_rev
         first_rev.refresh_from_db()
         assert first_rev.author == user
-        assert first_rev.content == "Hello World"
+        assert first_rev.content == "Hello World!"
 
     def test_pagerev_not_reused(self, client):
         """Test that a new revision is created if too much time
@@ -71,6 +73,7 @@ class TestEditPage:
 
 @pytest.mark.django_db
 def test_page_revision(client: Client):
+    """Test the GET to request to a specific revision page."""
     page = baker.prepare(Page)
     page.save(force_lock=True)
     page.view_groups.add(settings.SITH_GROUP_SUBSCRIBERS_ID)

core/views/forms.py

@@ -20,6 +20,7 @@
 # Place - Suite 330, Boston, MA 02111-1307, USA.
 #
 #
+import difflib
 import re
 from datetime import date, datetime, timedelta
 from io import BytesIO
@@ -385,29 +386,43 @@ class PageRevisionForm(forms.ModelForm):
 
     Notes:
         Saving this form won't always result in a new revision.
-        If the previous revision on the same page was made less
+        If the previous revision on the same page was made :
-        than 20 minutes ago by the same author,
+
+        - less than 20 minutes ago
+        - by the same author
+        - with a diff ratio higher than 20%
+
         then the latter will be edited and the new revision won't be created.
     """
 
-    UPDATE_THRESHOLD = timedelta(minutes=20)
+    TIME_THRESHOLD = timedelta(minutes=20)
+    DIFF_THRESHOLD = 0.2
 
     class Meta:
         model = PageRev
         fields = ["title", "content"]
         widgets = {"content": MarkdownInput}
 
-    def __init__(self, *args, author: User, page: Page, **kwargs):
+    def __init__(
-        super().__init__(*args, **kwargs)
+        self, *args, author: User, page: Page, instance: PageRev | None = None, **kwargs
+    ):
+        super().__init__(*args, instance=instance, **kwargs)
         self.author = author
         self.page = page
+        self.initial_content = instance.content if instance else ""
+
+    def diff_ratio(self, new_str: str) -> float:
+        return difflib.SequenceMatcher(
+            None, self.initial_content, new_str
+        ).quick_ratio()
 
     def save(self, commit=True):  # noqa FBT002
         revision: PageRev = self.instance
         if (
             revision._state.adding
             or revision.author != self.author
-            or revision.date + self.UPDATE_THRESHOLD < now()
+            or revision.date + self.TIME_THRESHOLD < now()
+            or self.diff_ratio(revision.content) < (1 - self.DIFF_THRESHOLD)
         ):
             revision.author = self.author
             revision.page = self.page