From 188c81a89cfd8b80be859bd447076237b7894fb5 Mon Sep 17 00:00:00 2001
From: Sameer6305 <sskadam6305@gmail.com>
Date: Sun, 31 May 2026 15:51:05 +0530
Subject: [PATCH] fix(cli): wire deduplicate CLI through graph store and
 EntityMerger

---
 semantica/cli.py           | 53 ++++++++++++++++++++++++++++++---
 tests/test_cli_commands.py | 60 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 109 insertions(+), 4 deletions(-)

diff --git a/semantica/cli.py b/semantica/cli.py
index 9f475549..831e696d 100644
--- a/semantica/cli.py
+++ b/semantica/cli.py
@@ -1038,19 +1038,64 @@ def deduplicate(
     """
     cli_ctx = _require_ctx(cli_ctx)
 
+    strategy_map = {
+        "blocking": "blocking_v2",
+        "semantic": "legacy",
+        "hybrid": "hybrid_v2",
+    }
+
+    def _load_entities() -> List[Dict[str, Any]]:
+        from .graph_store import get_nodes
+        from .graph_store.config import graph_store_config
+
+        graph_db = dict(cli_ctx.config.to_dict().get("graph_db", {}))
+        backend = cli_ctx.store_backend or graph_db.pop("backend", None)
+        previous_graph_config = graph_store_config.get_all()
+        graph_store_config.update(graph_db)
+        if backend:
+            graph_store_config.set("default_backend", backend)
+        try:
+            return get_nodes(limit=sys.maxsize)
+        finally:
+            graph_store_config.update(previous_graph_config)
+
     def _action() -> None:
         if _is_dry(cli_ctx, local_dry):
             _dry(cli_ctx, "deduplicate", json_out=_is_json(cli_ctx, local_json),
                  strategy=strategy, dedup_action=dedup_action)
             return
         try:
-            from .deduplication import detect_duplicates, merge_entities
+            from .deduplication import detect_duplicates
+            from .deduplication.entity_merger import EntityMerger
+
+            entities = _load_entities()
+            candidate_strategy = strategy_map.get(strategy, strategy)
+            detector_sort_by = "similarity_score" if sort_by == "similarity" else "confidence"
+            detection_kwargs: Dict[str, Any] = {
+                "candidate_strategy": candidate_strategy,
+                "sort_by": detector_sort_by,
+            }
             if dedup_action == "detect":
-                result = detect_duplicates(strategy=strategy, min_similarity=min_similarity)
+                result = detect_duplicates(
+                    entities,
+                    method="group",
+                    similarity_threshold=min_similarity,
+                    **detection_kwargs,
+                )
             elif dedup_action == "merge":
-                result = merge_entities(strategy=strategy, min_similarity=min_similarity)
+                merger = EntityMerger()
+                result = merger.merge_duplicates(
+                    entities,
+                    threshold=min_similarity,
+                    **detection_kwargs,
+                )
             else:
-                result = detect_duplicates(strategy=strategy, min_similarity=min_similarity)
+                result = detect_duplicates(
+                    entities,
+                    method="group",
+                    similarity_threshold=min_similarity,
+                    **detection_kwargs,
+                )
         except ImportError as exc:
             raise click.ClickException(f"Deduplication module not available: {exc}") from exc
         if output:
diff --git a/tests/test_cli_commands.py b/tests/test_cli_commands.py
index 311588c1..f375479e 100644
--- a/tests/test_cli_commands.py
+++ b/tests/test_cli_commands.py
@@ -544,6 +544,66 @@ class TestDeduplicate:
         data = _json_output(result)
         assert data["dry_run"] is True
 
+    def test_detect_runtime_path(self, runner, monkeypatch):
+        entities = [
+            {"id": "e1", "name": "Alice", "type": "Person"},
+            {"id": "e2", "name": "Alice", "type": "Person"},
+            {"id": "e3", "name": "Bob", "type": "Person"},
+        ]
+
+        class FakeStore:
+            def get_nodes(self, labels=None, properties=None, limit=100, **options):
+                return entities
+
+        monkeypatch.setattr("semantica.graph_store.methods._get_store", lambda: FakeStore())
+        monkeypatch.setattr("semantica.graph_store.methods.get_nodes", lambda **kwargs: entities)
+
+        result = runner.invoke(
+            main,
+            ["deduplicate", "--action", "detect", "--min-similarity", "0.1", "--json"],
+        )
+
+        _ok(result)
+        assert "Alice" in result.output
+        assert "Bob" not in result.output or "entities" in result.output
+
+    def test_merge_runtime_path(self, runner, monkeypatch):
+        entities = [
+            {"id": "e1", "name": "Alice", "type": "Person"},
+            {"id": "e2", "name": "Alice", "type": "Person"},
+        ]
+        captured = {}
+
+        class FakeStore:
+            def get_nodes(self, labels=None, properties=None, limit=100, **options):
+                return entities
+
+        monkeypatch.setattr("semantica.graph_store.methods._get_store", lambda: FakeStore())
+        monkeypatch.setattr("semantica.graph_store.methods.get_nodes", lambda **kwargs: entities)
+
+        def fake_merge(self, loaded_entities, **kwargs):
+            captured["entities"] = loaded_entities
+            captured["kwargs"] = kwargs
+            return [{"merged": True, "count": len(loaded_entities)}]
+
+        monkeypatch.setattr(
+            "semantica.deduplication.entity_merger.EntityMerger.merge_duplicates",
+            fake_merge,
+        )
+
+        result = runner.invoke(
+            main,
+            ["deduplicate", "--action", "merge", "--json"],
+        )
+
+        _ok(result)
+        data = _json_output(result)
+        assert data == [{"merged": True, "count": 2}]
+        assert captured["entities"] == entities
+        assert captured["kwargs"]["threshold"] == pytest.approx(0.7)
+        assert captured["kwargs"]["candidate_strategy"] == "hybrid_v2"
+        assert captured["kwargs"]["sort_by"] == "similarity_score"
+
     def test_global_dry_run_triggers_dry(self, runner):
         result = runner.invoke(main, ["--dry-run", "--json", "deduplicate"])
         _ok(result)