Automatically add tags to bookmarks based on URL pattern (#736)

* [WIP] DSL * upd * upd * upd * upd * upd * upd * upd * upd * upd * upd * upd * dsl2 * full feature * upd * upd * upd * upd * rename to auto_tagging_rules * update migration after rebase * add REST API tests * improve settings view --------- Co-authored-by: Sascha Ißbrücker <sascha.issbruecker@gmail.com>
2025-08-13 13:39:27 +02:00 · 2024-05-17 10:39:46 +03:00
parent e03f536925
commit fa5f78cf71
9 changed files with 369 additions and 0 deletions
--- a/bookmarks/tests/test_auto_tagging.py
+++ b/bookmarks/tests/test_auto_tagging.py
@@ -0,0 +1,179 @@
+from bookmarks.services import auto_tagging
+from django.test import TestCase
+
+
+class AutoTaggingTestCase(TestCase):
+    def test_auto_tag_by_domain(self):
+        script = """
+            example.com example
+            test.com test
+        """
+        url = "https://example.com/"
+
+        tags = auto_tagging.get_tags(script, url)
+
+        self.assertEqual(tags, set(["example"]))
+
+    def test_auto_tag_by_domain_ignores_case(self):
+        script = """
+            EXAMPLE.com example
+        """
+        url = "https://example.com/"
+
+        tags = auto_tagging.get_tags(script, url)
+
+        self.assertEqual(tags, set(["example"]))
+
+    def test_auto_tag_by_domain_should_add_all_tags(self):
+        script = """
+            example.com one two three
+        """
+        url = "https://example.com/"
+
+        tags = auto_tagging.get_tags(script, url)
+
+        self.assertEqual(tags, set(["one", "two", "three"]))
+
+    def test_auto_tag_by_domain_work_with_idn_domains(self):
+        script = """
+            रजिस्ट्री.भारत tag1
+        """
+        url = "https://www.xn--81bg3cc2b2bk5hb.xn--h2brj9c/"
+
+        tags = auto_tagging.get_tags(script, url)
+
+        self.assertEqual(tags, set(["tag1"]))
+
+        script = """
+            xn--81bg3cc2b2bk5hb.xn--h2brj9c tag1
+        """
+        url = "https://www.रजिस्ट्री.भारत/"
+
+        tags = auto_tagging.get_tags(script, url)
+
+        self.assertEqual(tags, set(["tag1"]))
+
+    def test_auto_tag_by_domain_and_path(self):
+        script = """
+            example.com/one one
+            example.com/two two
+            test.com test
+        """
+        url = "https://example.com/one/"
+
+        tags = auto_tagging.get_tags(script, url)
+
+        self.assertEqual(tags, set(["one"]))
+
+    def test_auto_tag_by_domain_and_path_ignores_case(self):
+        script = """
+            example.com/One one
+        """
+        url = "https://example.com/one/"
+
+        tags = auto_tagging.get_tags(script, url)
+
+        self.assertEqual(tags, set(["one"]))
+
+    def test_auto_tag_by_domain_and_path_matches_path_ltr(self):
+        script = """
+            example.com/one one
+            example.com/two two
+            test.com test
+        """
+        url = "https://example.com/one/two"
+
+        tags = auto_tagging.get_tags(script, url)
+
+        self.assertEqual(tags, set(["one"]))
+
+    def test_auto_tag_by_domain_ignores_domain_in_path(self):
+        script = """
+            example.com example
+        """
+        url = "https://test.com/example.com"
+
+        tags = auto_tagging.get_tags(script, url)
+
+        self.assertEqual(tags, set([]))
+
+    def test_auto_tag_by_domain_includes_subdomains(self):
+        script = """
+            example.com example
+            test.example.com test
+            some.example.com some
+        """
+        url = "https://test.example.com/"
+
+        tags = auto_tagging.get_tags(script, url)
+
+        self.assertEqual(tags, set(["example", "test"]))
+
+    def test_auto_tag_by_domain_matches_domain_rtl(self):
+        script = """
+            example.com example
+        """
+        url = "https://example.com.bad-website.com/"
+
+        tags = auto_tagging.get_tags(script, url)
+
+        self.assertEqual(tags, set([]))
+
+    def test_auto_tag_by_domain_ignores_schema(self):
+        script = """
+            https://example.com/ https
+            http://example.com/ http
+        """
+        url = "http://example.com/"
+
+        tags = auto_tagging.get_tags(script, url)
+
+        self.assertEqual(tags, set(["https", "http"]))
+
+    def test_auto_tag_by_domain_ignores_lines_with_no_tags(self):
+        script = """
+            example.com
+        """
+        url = "https://example.com/"
+
+        tags = auto_tagging.get_tags(script, url)
+
+        self.assertEqual(tags, set([]))
+
+    def test_auto_tag_by_domain_path_and_qs(self):
+        script = """
+            example.com/page?a=b tag1     # true, matches a=b
+            example.com/page?a=c&c=d tag2 # true, matches both a=c and c=d
+            example.com/page?c=d&l=p tag3 # false, l=p doesn't exists
+            example.com/page?a=bb tag4    # false bb != b
+            example.com/page?a=b&a=c tag5 # true, matches both a=b and a=c
+            example.com/page?a=B tag6     # true, matches a=b because case insensitive
+            example.com/page?A=b tag7     # true, matches a=b because case insensitive
+        """
+        url = "https://example.com/page/some?z=x&a=b&v=b&c=d&o=p&a=c"
+
+        tags = auto_tagging.get_tags(script, url)
+
+        self.assertEqual(tags, set(["tag1", "tag2", "tag5", "tag6", "tag7"]))
+
+    def test_auto_tag_by_domain_path_and_qs_with_empty_value(self):
+        script = """
+            example.com/page?a= tag1
+            example.com/page?b= tag2
+        """
+        url = "https://example.com/page/some?a=value"
+
+        tags = auto_tagging.get_tags(script, url)
+
+        self.assertEqual(tags, set(["tag1"]))
+
+    def test_auto_tag_by_domain_path_and_qs_works_with_encoded_url(self):
+        script = """
+            example.com/page?a=йцу tag1
+            example.com/page?a=%D0%B9%D1%86%D1%83 tag2
+        """
+        url = "https://example.com/page?a=%D0%B9%D1%86%D1%83"
+
+        tags = auto_tagging.get_tags(script, url)
+
+        self.assertEqual(tags, set(["tag1", "tag2"]))
--- a/bookmarks/tests/test_bookmarks_api.py
+++ b/bookmarks/tests/test_bookmarks_api.py
@@ -440,6 +440,20 @@ class BookmarksApiTestCase(LinkdingApiTestCase, BookmarkFactoryMixin):
        bookmark = Bookmark.objects.get(url=data["url"])
        self.assertFalse(bookmark.shared)

+    def test_create_bookmark_should_add_tags_from_auto_tagging(self):
+        tag1 = self.setup_tag()
+        tag2 = self.setup_tag()
+
+        self.authenticate()
+        profile = self.get_or_create_test_user().profile
+        profile.auto_tagging_rules = f"example.com {tag2.name}"
+        profile.save()
+
+        data = {"url": "https://example.com/", "tag_names": [tag1.name]}
+        self.post(reverse("bookmarks:bookmark-list"), data, status.HTTP_201_CREATED)
+        bookmark = Bookmark.objects.get(url=data["url"])
+        self.assertCountEqual(bookmark.tags.all(), [tag1, tag2])
+
    def test_get_bookmark(self):
        self.authenticate()
        bookmark = self.setup_bookmark()
@@ -512,6 +526,22 @@ class BookmarksApiTestCase(LinkdingApiTestCase, BookmarkFactoryMixin):
        updated_bookmark = Bookmark.objects.get(id=bookmark.id)
        self.assertEqual(updated_bookmark.shared, True)

+    def test_update_bookmark_adds_tags_from_auto_tagging(self):
+        bookmark = self.setup_bookmark()
+        tag1 = self.setup_tag()
+        tag2 = self.setup_tag()
+
+        self.authenticate()
+        profile = self.get_or_create_test_user().profile
+        profile.auto_tagging_rules = f"example.com {tag2.name}"
+        profile.save()
+
+        data = {"url": "https://example.com/", "tag_names": [tag1.name]}
+        url = reverse("bookmarks:bookmark-detail", args=[bookmark.id])
+        self.put(url, data, expected_status_code=status.HTTP_200_OK)
+        updated_bookmark = Bookmark.objects.get(id=bookmark.id)
+        self.assertCountEqual(updated_bookmark.tags.all(), [tag1, tag2])
+
    def test_patch_bookmark(self):
        self.authenticate()
        bookmark = self.setup_bookmark()
@@ -583,6 +613,22 @@ class BookmarksApiTestCase(LinkdingApiTestCase, BookmarkFactoryMixin):
        self.assertEqual(updated_bookmark.description, bookmark.description)
        self.assertListEqual(updated_bookmark.tag_names, bookmark.tag_names)

+    def test_patch_bookmark_adds_tags_from_auto_tagging(self):
+        bookmark = self.setup_bookmark()
+        tag1 = self.setup_tag()
+        tag2 = self.setup_tag()
+
+        self.authenticate()
+        profile = self.get_or_create_test_user().profile
+        profile.auto_tagging_rules = f"example.com {tag2.name}"
+        profile.save()
+
+        data = {"tag_names": [tag1.name]}
+        url = reverse("bookmarks:bookmark-detail", args=[bookmark.id])
+        self.patch(url, data, expected_status_code=status.HTTP_200_OK)
+        updated_bookmark = Bookmark.objects.get(id=bookmark.id)
+        self.assertCountEqual(updated_bookmark.tags.all(), [tag1, tag2])
+
    def test_delete_bookmark(self):
        self.authenticate()
        bookmark = self.setup_bookmark()
--- a/bookmarks/tests/test_bookmarks_service.py
+++ b/bookmarks/tests/test_bookmarks_service.py
@@ -130,6 +130,18 @@ class BookmarkServiceTestCase(TestCase, BookmarkFactoryMixin):

            mock_create_html_snapshot.assert_not_called()

+    def test_create_should_add_tags_from_auto_tagging(self):
+        tag1 = self.setup_tag()
+        tag2 = self.setup_tag()
+        profile = self.get_or_create_test_user().profile
+        profile.auto_tagging_rules = f"example.com {tag2.name}"
+        profile.save()
+
+        bookmark_data = Bookmark(url="https://example.com")
+        bookmark = create_bookmark(bookmark_data, tag1.name, self.user)
+
+        self.assertCountEqual(bookmark.tags.all(), [tag1, tag2])
+
    def test_update_should_create_web_archive_snapshot_if_url_did_change(self):
        with patch.object(
            tasks, "create_web_archive_snapshot"
@@ -201,6 +213,18 @@ class BookmarkServiceTestCase(TestCase, BookmarkFactoryMixin):

            mock_create_html_snapshot.assert_not_called()

+    def test_update_should_add_tags_from_auto_tagging(self):
+        tag1 = self.setup_tag()
+        tag2 = self.setup_tag()
+        profile = self.get_or_create_test_user().profile
+        profile.auto_tagging_rules = f"example.com {tag2.name}"
+        profile.save()
+        bookmark = self.setup_bookmark(url="https://example.com")
+
+        update_bookmark(bookmark, tag1.name, self.user)
+
+        self.assertCountEqual(bookmark.tags.all(), [tag1, tag2])
+
    def test_archive_bookmark(self):
        bookmark = Bookmark(
            url="https://example.com",
--- a/bookmarks/tests/test_settings_general_view.py
+++ b/bookmarks/tests/test_settings_general_view.py
@@ -42,6 +42,7 @@ class SettingsGeneralViewTestCase(TestCase, BookmarkFactoryMixin):
            "display_remove_bookmark_action": True,
            "permanent_notes": False,
            "custom_css": "",
+            "auto_tagging_rules": "",
        }

        return {**form_data, **overrides}
@@ -102,6 +103,7 @@ class SettingsGeneralViewTestCase(TestCase, BookmarkFactoryMixin):
            "permanent_notes": True,
            "default_mark_unread": True,
            "custom_css": "body { background-color: #000; }",
+            "auto_tagging_rules": "example.com tag",
        }
        response = self.client.post(reverse("bookmarks:settings.general"), form_data)
        html = response.content.decode()
@@ -168,6 +170,9 @@ class SettingsGeneralViewTestCase(TestCase, BookmarkFactoryMixin):
            self.user.profile.default_mark_unread, form_data["default_mark_unread"]
        )
        self.assertEqual(self.user.profile.custom_css, form_data["custom_css"])
+        self.assertEqual(
+            self.user.profile.auto_tagging_rules, form_data["auto_tagging_rules"]
+        )
        self.assertSuccessMessage(html, "Profile updated")

    def test_update_profile_should_not_be_called_without_respective_form_action(self):