server: implement similar posts as a search term, without sort

2 files changed, 59 insertions, 0 deletions

diff --git a/server/szurubooru/search/configs/post_search_config.py b/server/szurubooru/search/configs/post_search_config.py
index eda4084..c8eed18 100644
--- a/server/szurubooru/search/configs/post_search_config.py
+++ b/server/szurubooru/search/configs/post_search_config.py
@@ -122,6 +122,31 @@ def _pool_filter(
     )(query, criterion, negated)
 
 
+# includes the given post itself
+def _similar_filter(
+    query: SaQuery, criterion: Optional[criteria.BaseCriterion], negated: bool
+) -> SaQuery:
+    assert criterion
+    # subquery for tags of the given post (post id in criterion)
+    filter_func_tag = search_util.create_num_filter(model.PostTag.post_id)
+    tag_query = db.session.query(model.PostTag.tag_id)
+    tag_query = filter_func_tag(tag_query, criterion, False)
+    tag_query = tag_query.subquery("source_tags")
+
+    # subquery for posts with matching tags
+    pt_alias = sa.orm.aliased(model.PostTag)
+    subquery = (
+        db.session.query(pt_alias.post_id)
+        .filter(pt_alias.tag_id.in_(tag_query))
+        .group_by(pt_alias.post_id)
+        .subquery("similar_posts")
+    )
+    expr = model.Post.post_id.in_(subquery)
+    if negated:
+        expr = ~expr
+    return query.filter(expr)
+
+
 def _create_metric_num_filter(name: str):
     def wrapper(
         query: SaQuery,
@@ -414,6 +439,7 @@ class PostSearchConfig(BaseSearchConfig):
                     ),
                 ),
                 (["pool"], _pool_filter),
+                (["similar"], _similar_filter),
             ]
         ))
         return filters
diff --git a/server/szurubooru/tests/search/configs/test_post_search_config.py b/server/szurubooru/tests/search/configs/test_post_search_config.py
index 84e1601..3a347de 100644
--- a/server/szurubooru/tests/search/configs/test_post_search_config.py
+++ b/server/szurubooru/tests/search/configs/test_post_search_config.py
@@ -965,3 +965,36 @@ def test_around_query(
     db.session.add_all([tag_a, tag_b, tag_c, post1, post2, post3])
     db.session.flush()
     verify_around(input, 2, expected_prev_id, expected_next_id)
+
+
+@pytest.mark.parametrize("input,expected_post_ids", [
+    ("similar:1", [6, 4, 1]),
+    ("similar:2", [6, 5, 4, 2]),
+    ("similar:3", [6, 5, 3]),
+    ("similar:4", [6, 5, 4, 2, 1]),
+    ("similar:5", [6, 5, 4, 3, 2]),
+    ("similar:6", [6, 5, 4, 3, 2, 1]),
+    ("-similar:1", [5, 3, 2]),
+    ("-similar:2", [3, 1]),
+    ("-similar:3", [4, 2, 1]),
+    ("-similar:4", [3]),
+    ("-similar:5", [1]),
+    ("-similar:6", []),
+])
+def test_filter_by_similar(
+    post_factory, tag_factory, verify_unpaged, input, expected_post_ids
+):
+    tagA = tag_factory(names=["a"])
+    tagB = tag_factory(names=["b"])
+    tagC = tag_factory(names=["c"])
+    postA = post_factory(id=1, tags=[tagA])
+    postB = post_factory(id=2, tags=[tagB])
+    postC = post_factory(id=3, tags=[tagC])
+    postAB = post_factory(id=4, tags=[tagA, tagB])
+    postBC = post_factory(id=5, tags=[tagB, tagC])
+    postABC = post_factory(id=6, tags=[tagA, tagB, tagC])
+    db.session.add_all(
+        [tagA, tagB, tagC, postA, postB, postC, postAB, postBC, postABC]
+    )
+    db.session.flush()
+    verify_unpaged(input, expected_post_ids, True)