iris-vector-graph 1.63.0__tar.gz → 1.63.2__tar.gz

{iris_vector_graph-1.63.0 → iris_vector_graph-1.63.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: iris-vector-graph
-Version: 1.63.0
+Version: 1.63.2
 Summary: Transactional Graph + Vector retrieval system for InterSystems IRIS with hybrid search, openCypher, and GraphQL APIs
 Project-URL: Homepage, https://github.com/intersystems-community/iris-vector-graph
 Project-URL: Documentation, https://github.com/intersystems-community/iris-vector-graph/tree/main/docs
@@ -23,14 +23,14 @@ Classifier: Topic :: Database
 Classifier: Topic :: Scientific/Engineering
 Classifier: Topic :: Scientific/Engineering :: Bio-Informatics
 Requires-Python: >=3.10
-Requires-Dist: intersystems-irispython>=3.2.0
+Requires-Dist: intersystems-iris>=1.0.0
 Provides-Extra: biodata
 Requires-Dist: biopython>=1.81; extra == 'biodata'
 Requires-Dist: bioservices>=1.11.0; extra == 'biodata'
 Requires-Dist: mygene>=3.2.0; extra == 'biodata'
 Requires-Dist: obonet>=1.0.0; extra == 'biodata'
 Provides-Extra: core
-Requires-Dist: intersystems-irispython>=3.2.0; extra == 'core'
+Requires-Dist: intersystems-iris>=1.0.0; extra == 'core'
 Provides-Extra: demo
 Requires-Dist: python-fasthtml>=0.12.0; extra == 'demo'
 Provides-Extra: dev
@@ -655,6 +655,14 @@ anchors = engine.get_kg_anchors(icd_codes=["J18.0", "E11.9"])
 
 ## Changelog
 
+### v1.63.2 (2026-04-25)
+- fix: `MATCH (a)-[r*1..N]-(b)` undirected BFS now traverses `^KG("in",...)` for inbound edges (was outbound-only)
+- fix: `MATCH (a)<-[r*1..N]-(b)` inbound-only BFS now works  
+- fix: `initialize_schema()` ObjectScript LoadDir tries Docker `/tmp/src/` before Mac path — fixes silent compile failure in test containers
+- 4 E2E tests: directed-out, undirected, multihop undirected, directed-in all passing
+- Arno BFSJson falls back gracefully to BFSFastJson for graphs >3.5MB adjacency string (299K+ long-ID edges); per-seed export is spec 079 future work
+
+
 ### v1.63.0 (2026-04-25)
 - feat: Arno/Rust fast path for BFS (`_execute_var_length_cypher`) — when `libarno_callout.so` is loaded with `Graph.KG.NKGAccel.BFSJson`, var-length Cypher queries use Rust BFS over `^NKG` integer adjacency instead of ObjectScript `BFSFastJson`. Projected 128ms → <30ms p50 for 6K+ result BFS at 10K/50K scale. Falls back transparently to `BFSFastJson` when Arno not loaded. (spec 079, arno spec 035)
 

{iris_vector_graph-1.63.0 → iris_vector_graph-1.63.2}/README.md

@@ -583,6 +583,14 @@ anchors = engine.get_kg_anchors(icd_codes=["J18.0", "E11.9"])
 
 ## Changelog
 
+### v1.63.2 (2026-04-25)
+- fix: `MATCH (a)-[r*1..N]-(b)` undirected BFS now traverses `^KG("in",...)` for inbound edges (was outbound-only)
+- fix: `MATCH (a)<-[r*1..N]-(b)` inbound-only BFS now works  
+- fix: `initialize_schema()` ObjectScript LoadDir tries Docker `/tmp/src/` before Mac path — fixes silent compile failure in test containers
+- 4 E2E tests: directed-out, undirected, multihop undirected, directed-in all passing
+- Arno BFSJson falls back gracefully to BFSFastJson for graphs >3.5MB adjacency string (299K+ long-ID edges); per-seed export is spec 079 future work
+
+
 ### v1.63.0 (2026-04-25)
 - feat: Arno/Rust fast path for BFS (`_execute_var_length_cypher`) — when `libarno_callout.so` is loaded with `Graph.KG.NKGAccel.BFSJson`, var-length Cypher queries use Rust BFS over `^NKG` integer adjacency instead of ObjectScript `BFSFastJson`. Projected 128ms → <30ms p50 for 6K+ result BFS at 10K/50K scale. Falls back transparently to `BFSFastJson` when Arno not loaded. (spec 079, arno spec 035)
 

{iris_vector_graph-1.63.0 → iris_vector_graph-1.63.2}/iris_src/src/Graph/KG/Traversal.cls

@@ -151,8 +151,9 @@ ClassMethod TraverseAllPredicates(s As %String, hop As %Integer, maxHops As %Int
 }
 
 /// Optimized BFS using Process-Private Globals and $ListBuild to bypass object bottleneck
+/// @param direction "out" (default), "in", or "both" for undirected BFS
 /// @returns Number of edges found. Results are in ^||BFS.Results(i) = $lb(s, p, o, w, step)
-ClassMethod BFSFast(srcId As %String, preds As %DynamicArray = "", maxHops As %Integer = 2, dstLabel As %String = "") As %Integer
+ClassMethod BFSFast(srcId As %String, preds As %DynamicArray = "", maxHops As %Integer = 2, dstLabel As %String = "", direction As %String = "out") As %Integer
 {
     Kill ^||BFS.Results
     If (srcId = "") || (maxHops <= 0) Return 0
@@ -171,10 +172,20 @@ ClassMethod BFSFast(srcId As %String, preds As %DynamicArray = "", maxHops As %I
             Set s = $Order(frontier(s))
             If s = "" Quit
             
-            If fixedP '= "" {
-                Do ..TraverseWithPredicateFast(s, fixedP, hop, maxHops, nextP, dstLabel, .count, .nextFrontier, .seen)
-            } Else {
-                Do ..TraverseAllPredicatesFast(s, hop, maxHops, nextP, dstLabel, .count, .nextFrontier, .seen)
+            If direction = "out" || (direction = "both") {
+                If fixedP '= "" {
+                    Do ..TraverseWithPredicateFast(s, fixedP, hop, maxHops, nextP, dstLabel, .count, .nextFrontier, .seen)
+                } Else {
+                    Do ..TraverseAllPredicatesFast(s, hop, maxHops, nextP, dstLabel, .count, .nextFrontier, .seen)
+                }
+            }
+            
+            If direction = "in" || (direction = "both") {
+                If fixedP '= "" {
+                    Do ..TraverseWithPredicateFastInbound(s, fixedP, hop, .count, .nextFrontier, .seen)
+                } Else {
+                    Do ..TraverseAllPredicatesFastInbound(s, hop, .count, .nextFrontier, .seen)
+                }
             }
         }
         
@@ -216,6 +227,30 @@ ClassMethod TraverseAllPredicatesFast(s As %String, hop As %Integer, maxHops As
     }
 }
 
+/// Inbound BFS: traverse ^KG("in", 0, s, p, o) where s is the target, o is the source.
+/// Result stored as (o, p, s) to preserve original edge direction.
+ClassMethod TraverseWithPredicateFastInbound(s As %String, p As %String, hop As %Integer, ByRef count As %Integer, ByRef nextFrontier, ByRef seen) [ Internal, Private ]
+{
+    Set o = ""
+    For {
+        Set o = $Order(^KG("in", 0, s, p, o))
+        If o = "" Quit
+        Set count = count + 1
+        Set ^||BFS.Results(count) = $ListBuild(s, p, o, $Get(^KG("out", 0, o, p, s), 1.0), hop)
+        If '$Data(seen(o)) Set nextFrontier(o) = ""
+    }
+}
+
+ClassMethod TraverseAllPredicatesFastInbound(s As %String, hop As %Integer, ByRef count As %Integer, ByRef nextFrontier, ByRef seen) [ Internal, Private ]
+{
+    Set p = ""
+    For {
+        Set p = $Order(^KG("in", 0, s, p))
+        If p = "" Quit
+        Do ..TraverseWithPredicateFastInbound(s, p, hop, .count, .nextFrontier, .seen)
+    }
+}
+
 /// Main BFS implementation
 ClassMethod BFS(srcId As %String, preds As %DynamicArray = "", maxHops As %Integer = 2, dstLabel As %String = "") As %DynamicArray
 {
@@ -287,7 +322,7 @@ ClassMethod BFSJSON(srcId As %String, preds As %DynamicArray, maxHops As %Intege
 /// @param maxHops  Maximum hops (default 2)
 /// @param dstLabel Filter destination nodes by label (empty = no filter)
 /// @returns JSON array: [{"s":..,"p":..,"o":..,"w":..,"step":..}, ...] in traversal order
-ClassMethod BFSFastJson(srcId As %String, predsJson As %String = "", maxHops As %Integer = 2, dstLabel As %String = "") As %String
+ClassMethod BFSFastJson(srcId As %String, predsJson As %String = "", maxHops As %Integer = 2, dstLabel As %String = "", direction As %String = "out") As %String
 {
     // Parse predicates JSON into a %DynamicArray (null/"" means all predicates)
     Set preds = ""
@@ -300,7 +335,7 @@ ClassMethod BFSFastJson(srcId As %String, predsJson As %String = "", maxHops As
     }
 
     // Run BFSFast -- writes to ^||BFS.Results(i) = $ListBuild(s,p,o,w,step)
-    Set count = ..BFSFast(srcId, preds, maxHops, dstLabel)
+    Set count = ..BFSFast(srcId, preds, maxHops, dstLabel, direction)
 
     If count = 0 Return "[]"
 

{iris_vector_graph-1.63.0 → iris_vector_graph-1.63.2}/iris_vector_graph/bulk_loader.py

@@ -273,8 +273,8 @@ class BulkLoader:
         Must be called after load_edges(use_noindex=True).
         Also rebuilds bitmap extent indexes for correct COUNT(*).
         """
-        import intersystems_iris
-        iris_obj = intersystems_iris.createIRIS(self.conn)
+        import iris
+        iris_obj = iris.createIRIS(self.conn)
         results = {}
 
         for cls in ["Graph.KG.rdfedges", "Graph.KG.rdflabels", "Graph.KG.rdfprops", "Graph.KG.nodes"]:
@@ -301,8 +301,8 @@ class BulkLoader:
         Returns True if successful.
         """
         try:
-            import intersystems_iris
-            iris_obj = intersystems_iris.createIRIS(self.conn)
+            import iris
+            iris_obj = iris.createIRIS(self.conn)
             logger.info("Building ^KG + ^NKG globals from SQL tables...")
             t0 = time.time()
             iris_obj.classMethodVoid("Graph.KG.Traversal", "BuildKG")
@@ -422,7 +422,7 @@ def main():
     logger.info(f"Graph: {G.number_of_nodes():,} nodes, {G.number_of_edges():,} edges")
 
 
-    from intersystems_iris.dbapi._DBAPI import connect
+    from iris.dbapi._DBAPI import connect  # intersystems_iris.dbapi = iris.dbapi
     conn = connect(args.host, args.port, args.namespace, args.user, args.password)
     logger.info(f"Connected to IRIS {args.host}:{args.port}/{args.namespace}")
 

{iris_vector_graph-1.63.0 → iris_vector_graph-1.63.2}/iris_vector_graph/cypher/translator.py

@@ -1744,6 +1744,9 @@ def translate_relationship_pattern(
         def _resolve_id_param(node):
             id_val = node.properties.get("id")
             if id_val is None:
+                if node.variable and node.variable in context.input_params:
+                    val = context.input_params[node.variable]
+                    return f"${node.variable}" if isinstance(val, str) else None
                 return None
             if isinstance(id_val, ast.Variable):
                 return f"${id_val.name}"
@@ -1756,7 +1759,7 @@ def translate_relationship_pattern(
         src_id_param = _resolve_id_param(source_node)
         dst_id_param = _resolve_id_param(target_node)
 
-        direction_str = "both" if rel.direction == ast.Direction.BOTH else "out"
+        direction_str = "both" if rel.direction == ast.Direction.BOTH else ("in" if rel.direction == ast.Direction.INCOMING else "out")
 
         context.var_length_paths.append(
             {

{iris_vector_graph-1.63.0 → iris_vector_graph-1.63.2}/iris_vector_graph/engine.py

@@ -1025,6 +1025,31 @@ class IRISGraphEngine:
         source_id = _resolve(vl.get("src_id_param"))
         target_id = _resolve(vl.get("dst_id_param"))
 
+        if source_id is None and parameters:
+            src_var = vl.get("source_var")
+            if src_var and src_var in parameters:
+                source_id = str(parameters[src_var])
+            else:
+                source_id = next(
+                    (str(v) for v in parameters.values() if isinstance(v, str)), None
+                )
+
+        if target_id is None and parameters:
+            dst_var = vl.get("target_var")
+            if dst_var and dst_var in parameters:
+                target_id = str(parameters[dst_var])
+            else:
+                vals = [str(v) for v in parameters.values() if isinstance(v, str)]
+                target_id = vals[1] if len(vals) > 1 else None
+
+        if source_id is None or target_id is None:
+            sql_params = sql_query.parameters[0] if sql_query.parameters else []
+            str_params = [p for p in sql_params if isinstance(p, str) and not p.startswith("Graph_KG")]
+            if source_id is None and len(str_params) >= 1:
+                source_id = str_params[0]
+            if target_id is None and len(str_params) >= 2:
+                target_id = str_params[1]
+
         if source_id is None or target_id is None:
             raise ValueError(
                 "shortestPath requires both source and target node IDs to be bound. "
@@ -1131,7 +1156,11 @@ class IRISGraphEngine:
                 source_id = item
                 break
         if source_id is None and parameters:
-            source_id = next(iter(parameters.values()), None)
+            src_var = vl.get("source_var")
+            if src_var and src_var in parameters:
+                source_id = str(parameters[src_var])
+            else:
+                source_id = next(iter(parameters.values()), None)
 
         if source_id is None:
             return {
@@ -1177,6 +1206,7 @@ class IRISGraphEngine:
                     predicates_json,
                     max_hops,
                     "",
+                    vl.get("direction", "out"),
                 )
                 bfs_results = _json.loads(str(bfs_json)) if bfs_json else []
             except Exception as e:
@@ -1214,9 +1244,14 @@ class IRISGraphEngine:
                 seen.add(oid)
                 target_ids.append(oid)
 
+        import re as _re
+        sql_str = sql_query.sql if isinstance(sql_query.sql, str) else ""
+        alias_match = _re.search(r'SELECT\s+DISTINCT\s+\S+\s+AS\s+(\w+)|SELECT\s+\S+\s+AS\s+(\w+)', sql_str, _re.IGNORECASE)
+        col_name = (alias_match.group(1) or alias_match.group(2)) if alias_match else "b_id"
+
         if not target_ids:
             return {
-                "columns": ["b_id", "b_labels", "b_props"],
+                "columns": [col_name, "b_labels", "b_props"],
                 "rows": [],
                 "sql": "",
                 "params": [],
@@ -1236,7 +1271,7 @@ class IRISGraphEngine:
             )
 
         return {
-            "columns": ["b_id", "b_labels", "b_props"],
+            "columns": [col_name, "b_labels", "b_props"],
             "rows": [list(r) for r in rows],
             "sql": f"BFSFastJson({source_id}, {predicates_json}, {max_hops})",
             "params": [],
@@ -4853,7 +4888,7 @@ class IRISGraphEngine:
             try:
                 iris_obj = self._iris_obj()
                 result = iris_obj.classMethodValue(
-                    "Graph.KG.Traversal", "BFSFastJson", seed, "", hops, ""
+                    "Graph.KG.Traversal", "BFSFastJson", seed, "", hops, "", "out"
                 )
                 if result:
                     edges = json.loads(str(result))
@@ -4916,9 +4951,9 @@ class IRISGraphEngine:
 
             return iris.createIRIS(self.conn)
         except (TypeError, AttributeError):
-            import intersystems_iris
+            import iris
 
-            return intersystems_iris.createIRIS(self.conn)
+            return iris.createIRIS(self.conn)
 
     def vec_create_index(
         self,

{iris_vector_graph-1.63.0 → iris_vector_graph-1.63.2}/iris_vector_graph/schema.py

@@ -17,43 +17,18 @@ logger = logging.getLogger(__name__)
 
 
 def _call_classmethod(conn_or_cursor, class_name: str, method_name: str, *args) -> Any:
-    """Call an IRIS ObjectScript class method using the native API.
-
-    Works with both the ``iris`` package (iris.createIRIS) and the
-    ``intersystems_iris`` package (intersystems_iris.createIRIS).  Tries
-    ``iris.createIRIS`` first because it accepts the connection objects
-    returned by ``iris.connect()`` (the standard test-fixture connection).
-
-    Resolves the connection from either a connection object directly or from
-    ``cursor._connection`` if a cursor is passed.
-
-    Returns the method's return value, or raises if the class/method does not
-    exist or the native API is unavailable.
-    """
-    # Accept either a connection or a cursor
     if hasattr(conn_or_cursor, "cursor"):
-        conn = conn_or_cursor  # it's a connection
+        conn = conn_or_cursor
     elif hasattr(conn_or_cursor, "_connection"):
-        conn = conn_or_cursor._connection  # it's a cursor
+        conn = conn_or_cursor._connection
     else:
-        conn = conn_or_cursor  # best-effort fallback
-
-    # Try iris.createIRIS first (accepts iris.IRISConnection from iris.connect())
-    try:
-        import iris as _iris_pkg  # type: ignore[import]
-
-        iris_obj = _iris_pkg.createIRIS(conn)
-        return iris_obj.classMethodValue(class_name, method_name, *args)
-    except (ImportError, AttributeError, TypeError):
-        pass
-
-    # Fall back to intersystems_iris.createIRIS (accepts intersystems_iris.IRISConnection)
-    import intersystems_iris as _iris_pkg2  # type: ignore[import]
-
-    iris_obj = _iris_pkg2.createIRIS(conn)
+        conn = conn_or_cursor
+    import iris as _iris_pkg
+    iris_obj = _iris_pkg.createIRIS(conn)
     return iris_obj.classMethodValue(class_name, method_name, *args)
 
 
+
 class GraphSchema:
     """Domain-agnostic RDF-style graph schema management"""
 
@@ -782,7 +757,7 @@ LANGUAGE OBJECTSCRIPT
 
         src_dir = str(iris_src_path / "src") if iris_src_path else ""
         deployed = False
-        for candidate_dir in [src_dir, "/tmp/src/", "/irisdev/app/iris_src/src/"]:
+        for candidate_dir in ["/tmp/src/", "/irisdev/app/iris_src/src/", src_dir]:
             if not candidate_dir:
                 continue
             try:

{iris_vector_graph-1.63.0 → iris_vector_graph-1.63.2}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "iris-vector-graph"
-version = "1.63.0"
+version = "1.63.2"
 description = "Transactional Graph + Vector retrieval system for InterSystems IRIS with hybrid search, openCypher, and GraphQL APIs"
 readme = "README.md"
 license = "MIT"
@@ -38,12 +38,12 @@ requires-python = ">=3.10"
 
 # Core dependencies
 dependencies = [
-    "intersystems-irispython>=3.2.0",
+    "intersystems-iris>=1.0.0",
 ]
 
 [project.optional-dependencies]
 core = [
-    "intersystems-irispython>=3.2.0",
+    "intersystems-iris>=1.0.0",
 ]
 full = [
     "numpy>=1.24.0",

{iris_vector_graph-1.63.0 → iris_vector_graph-1.63.2}/tests/unit/test_cypher_e2e_new_features.py

@@ -457,3 +457,98 @@ class TestCypherRound2E2E:
         q = "MATCH (n) WHERE n.id = $id CALL ivg.vector.search('Gene','embedding',[0.1,0.2,0.3,0.4],5) YIELD node, score RETURN n.id, node, score"
         r = translate_to_sql(parse_query(q), {"id": "x"})
         assert r.var_length_paths is not None or r.sql is not None
+
+
+@pytest.mark.skipif(SKIP_IRIS_TESTS, reason="SKIP_IRIS_TESTS=true")
+class TestUndirectedBFSE2E:
+
+    @pytest.fixture(autouse=True)
+    def setup(self, iris_connection):
+        from iris_vector_graph.engine import IRISGraphEngine
+        self.conn = iris_connection
+        self.engine = IRISGraphEngine(iris_connection, embedding_dimension=4)
+        self.engine.initialize_schema()
+        self._run = uuid.uuid4().hex[:8]
+        self._nodes = []
+        yield
+        cursor = self.conn.cursor()
+        for nid in self._nodes:
+            try:
+                cursor.execute("DELETE FROM Graph_KG.rdf_edges WHERE s=? OR o_id=?", [nid, nid])
+                cursor.execute("DELETE FROM Graph_KG.rdf_labels WHERE s=?", [nid])
+                cursor.execute("DELETE FROM Graph_KG.nodes WHERE node_id=?", [nid])
+            except Exception:
+                pass
+        try:
+            self.conn.commit()
+        except Exception:
+            self.conn.rollback()
+
+    def _node(self, suffix):
+        nid = f"ubfs_{self._run}_{suffix}"
+        self._nodes.append(nid)
+        self.engine.create_node(nid, labels=["Gene"])
+        return nid
+
+    def _edge(self, s, p, o):
+        self.engine.create_edge(s, p, o)
+
+    def test_directed_out_returns_outbound_only(self):
+        a = self._node("A")
+        b = self._node("B")
+        c = self._node("C")
+        self._edge(a, "REL", b)
+        self._edge(c, "REL", a)
+
+        result = self.engine.execute_cypher(
+            "MATCH (x)-[r*1..1]->(y) WHERE x.id = $id RETURN y.id",
+            {"id": a},
+        )
+        ids = {row[0] for row in result["rows"]}
+        assert b in ids
+        assert c not in ids
+
+    def test_undirected_returns_both_inbound_and_outbound(self):
+        a = self._node("hub")
+        b = self._node("outbound")
+        c = self._node("inbound")
+        self._edge(a, "REL", b)
+        self._edge(c, "REL", a)
+
+        result = self.engine.execute_cypher(
+            "MATCH (x)-[r*1..1]-(y) WHERE x.id = $id RETURN y.id",
+            {"id": a},
+        )
+        ids = {row[0] for row in result["rows"]}
+        assert b in ids, f"outbound neighbor missing: {ids}"
+        assert c in ids, f"inbound neighbor missing: {ids}"
+
+    def test_undirected_multihop(self):
+        a = self._node("mh_a")
+        b = self._node("mh_b")
+        c = self._node("mh_c")
+        self._edge(b, "REL", a)
+        self._edge(b, "REL", c)
+
+        result = self.engine.execute_cypher(
+            "MATCH (x)-[r*1..2]-(y) WHERE x.id = $id RETURN y.id",
+            {"id": a},
+        )
+        ids = {row[0] for row in result["rows"]}
+        assert b in ids
+        assert c in ids
+
+    def test_directed_in_returns_inbound_only(self):
+        a = self._node("tgt")
+        b = self._node("src")
+        c = self._node("other")
+        self._edge(b, "REL", a)
+        self._edge(a, "REL", c)
+
+        result = self.engine.execute_cypher(
+            "MATCH (x)<-[r*1..1]-(y) WHERE x.id = $id RETURN y.id",
+            {"id": a},
+        )
+        ids = {row[0] for row in result["rows"]}
+        assert b in ids, f"inbound src missing: {ids}"
+        assert c not in ids, f"outbound dst should not appear: {ids}"