iris-vector-graph 1.74.0__tar.gz → 1.76.0__tar.gz

{iris_vector_graph-1.74.0 → iris_vector_graph-1.76.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: iris-vector-graph
-Version: 1.74.0
+Version: 1.76.0
 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

{iris_vector_graph-1.74.0 → iris_vector_graph-1.76.0}/iris_src/src/Graph/KG/Meta.cls

@@ -62,4 +62,25 @@ ClassMethod GetKG(s1 As %String = "", s2 As %String = "", s3 As %String = "", s4
 	Return ""
 }
 
+ClassMethod IVG_PDISC(vals As %String, p As %Double) As %String [ SqlProc ]
+{
+	Set arr = ##class(%Library.DynamicArray).%FromJSON(vals)
+	Set n = arr.%Size()
+	If n = 0 { Quit "" }
+	Set i = n * p \ 1
+	If i >= n { Set i = n - 1 }
+	Quit arr.%Get(i)
+}
+
+ClassMethod IVG_PCONT(vals As %String, p As %Double) As %Double [ SqlProc ]
+{
+	Set arr = ##class(%Library.DynamicArray).%FromJSON(vals)
+	Set n = arr.%Size()
+	If n = 0 { Quit 0 }
+	Set lower = n * p \ 1
+	If lower >= n - 1 { Quit arr.%Get(n-1) + 0 }
+	Set frac = (n * p) - lower
+	Quit arr.%Get(lower) + frac * (arr.%Get(lower+1) - arr.%Get(lower))
+}
+
 }

iris_vector_graph-1.76.0/iris_src/src/IVG/Percentile.cls

@@ -0,0 +1,29 @@
+/// Percentile aggregate helpers for IVG Cypher engine.
+Class IVG.Percentile Extends %RegisteredObject [ SqlSchemaName = IVG ]
+{
+
+/// percentileDisc(vals, p) — discrete Nth percentile of JSON array.
+ClassMethod PDISC(vals As %String, p As %Double) As %String [ SqlProc ]
+{
+	Set arr = ##class(%Library.DynamicArray).%FromJSON(vals)
+	Set n = arr.%Size()
+	If n = 0 { Quit "" }
+	Set i = (n - 1) * p \ 1
+	If i >= (n - 1) { Set i = n - 1 }
+	Quit arr.%Get(i)
+}
+
+/// percentileCont(vals, p) — continuous Nth percentile (linear interpolation).
+ClassMethod PCONT(vals As %String, p As %Double) As %Double [ SqlProc ]
+{
+	Set arr = ##class(%Library.DynamicArray).%FromJSON(vals)
+	Set n = arr.%Size()
+	If n = 0 { Quit 0 }
+	Set pos = (n - 1) * p
+	Set lower = pos \ 1
+	If lower >= (n - 1) { Quit arr.%Get(n-1) + 0 }
+	Set frac = pos - lower
+	Quit arr.%Get(lower) + frac * (arr.%Get(lower+1) - arr.%Get(lower))
+}
+
+}

{iris_vector_graph-1.74.0 → iris_vector_graph-1.76.0}/iris_vector_graph/bolt_server.py

@@ -691,7 +691,15 @@ class BoltSession:
                     columns[i+2].endswith("_props") and
                     col[:-3] == columns[i+1][:-7] == columns[i+2][:-6]):
                 var_name = col[:-3]
-                node_id = str(row[i]) if row[i] is not None else ""
+                node_id = str(row[i]) if row[i] is not None else None
+
+                if node_id is None:
+                    new_cols.append(var_name)
+                    new_row.append(None)
+                    prev_node_id = None
+                    i += 3
+                    continue
+
                 labels_raw = row[i+1]
                 props_raw = row[i+2]
 
@@ -713,7 +721,8 @@ class BoltSession:
                   not col.endswith("_props") and
                   "_" not in col and
                   i + 1 < len(columns) and
-                  columns[i+1].endswith("_id")):
+                  columns[i+1].endswith("_id") and
+                  columns[i+1][:-3] == col):
                 rel_type = str(row[i]) if row[i] is not None else "RELATED_TO"
                 start_id = _node_int_id(prev_node_id) if prev_node_id else 0
                 end_node_id_str = str(row[i+1]) if i+1 < len(row) else ""

{iris_vector_graph-1.74.0 → iris_vector_graph-1.76.0}/iris_vector_graph/cypher/ast.py

@@ -210,6 +210,12 @@ class MapLiteral:
     entries: dict
 
 
+@dataclass(slots=True)
+class MapProjection:
+    variable: str
+    keys: list
+
+
 @dataclass(slots=True)
 class SubscriptExpression:
     expression: Any

{iris_vector_graph-1.74.0 → iris_vector_graph-1.76.0}/iris_vector_graph/cypher/parser.py

@@ -1140,6 +1140,24 @@ class Parser:
                         "Expected property name", prop_tok.line, prop_tok.column
                     )
                 return ast.PropertyReference(name, prop_tok.value)
+            if self.peek().kind == TokenType.LBRACE:
+                lookahead = self.lexer.peek_ahead(1)
+                if lookahead.kind == TokenType.DOT or (
+                    lookahead.kind == TokenType.IDENTIFIER and self.lexer.peek_ahead(2).kind == TokenType.COLON
+                ) or lookahead.kind == TokenType.RBRACE:
+                    self.eat()
+                    keys = []
+                    while not self.matches(TokenType.RBRACE):
+                        if self.peek().kind == TokenType.DOT:
+                            self.eat()
+                            k = self.expect(TokenType.IDENTIFIER)
+                            keys.append(("." + k.value, None))
+                        elif self.peek().kind == TokenType.IDENTIFIER:
+                            k = self.eat().value
+                            keys.append((k, None))
+                        if self.peek().kind == TokenType.COMMA:
+                            self.eat()
+                    return ast.MapProjection(variable=name, keys=keys)
             return ast.Variable(name)
 
         if tok.kind == TokenType.PARAMETER:

{iris_vector_graph-1.74.0 → iris_vector_graph-1.76.0}/iris_vector_graph/cypher/translator.py

@@ -1152,6 +1152,22 @@ def translate_to_sql(
             if stage_name in [s.split(" AS ")[0].strip() for s in context.stages]:
                 sql = sql.replace("SELECT \nFROM", f"SELECT *\nFROM", 1)
                 sql = sql.replace("SELECT DISTINCT \nFROM", f"SELECT DISTINCT *\nFROM", 1)
+        if hasattr(context, '_percentile_queries') and context._percentile_queries:
+            import re as _re
+            from_match = _re.search(r'\nFROM\s+(.*?)(?:\nWHERE|\nORDER|\nLIMIT|\nGROUP|\nHAVING|$)', sql, _re.DOTALL)
+            if from_match and len(context._percentile_queries) == 1:
+                from_clause = from_match.group(0).strip()
+                val_expr, pct_val, fn_name, var_name, alias = context._percentile_queries[0]
+                col_alias = _re.search(r'AS\s+(\w+)\s*$', sql.split('\n')[0])
+                out_alias = col_alias.group(1) if col_alias else "result"
+                proc = "PCONT" if fn_name == "percentilecont" else "PDISC"
+                inner_col = val_expr.split('.')[-1] if '.' in val_expr else val_expr
+                sql = (
+                    f"SELECT IVG.Percentile_{proc}("
+                    f"(SELECT JSON_ARRAYAGG(CAST({val_expr} AS DOUBLE)) "
+                    f"\n{from_clause}), {pct_val}) AS {out_alias}"
+                )
+                p = []
         sql = apply_pagination(sql, cypher_query, context, order_by_items)
         vl = context.var_length_paths or None
 
@@ -1326,7 +1342,13 @@ def translate_create_clause(create, context, metadata):
             node_id_expr = node.properties.get("id") or node.properties.get("node_id")
             if node_id_expr is None:
                 import uuid as _uuid
-                node_id_expr = ast.Literal(str(_uuid.uuid4()))
+                generated_id = str(_uuid.uuid4())
+                node_id_expr = ast.Literal(generated_id)
+                key = f"__create_id_{node.variable}" if node.variable else f"__create_id_anon_{id(node)}"
+                context.input_params[key] = generated_id
+                if not hasattr(context, '_anon_node_keys'):
+                    context._anon_node_keys = {}
+                context._anon_node_keys[id(node)] = generated_id
 
             var_alias = None
             if isinstance(node_id_expr, ast.Variable):
@@ -1390,8 +1412,15 @@ def translate_create_clause(create, context, metadata):
 
             def _resolve_id(id_expr, node):
                 if id_expr is None:
-                    if node.variable and node.variable in context.input_params:
-                        return context.input_params[node.variable]
+                    if node.variable:
+                        stored = context.input_params.get(f"__create_id_{node.variable}")
+                        if stored:
+                            return stored
+                        if node.variable in context.input_params:
+                            return context.input_params[node.variable]
+                    anon_id = getattr(context, '_anon_node_keys', {}).get(id(node))
+                    if anon_id:
+                        return anon_id
                     return None
                 if isinstance(id_expr, ast.Literal):
                     return id_expr.value
@@ -2454,26 +2483,29 @@ def translate_expression(expr, context, segment="select") -> str:
         return f"(({inner}) > 0)"
 
     if isinstance(expr, ast.ListComprehension):
-        source_sql = translate_expression(expr.source, context, segment=segment)
+        source_sql = translate_expression(expr.source, context, segment="inline")
         var = sanitize_identifier(expr.variable)
         alias = context.next_alias("lc")
         context.variable_aliases[expr.variable] = alias
         where_clause = ""
         if expr.predicate:
-            pred_sql = translate_expression(expr.predicate, context, segment=segment)
+            if isinstance(expr.predicate, ast.BooleanExpression):
+                pred_sql = translate_boolean_expression(expr.predicate, context)
+            else:
+                pred_sql = translate_expression(expr.predicate, context, segment="inline")
             for col in ("node_id", "p", "val", "label"):
                 pred_sql = pred_sql.replace(f"{alias}.{col}", f"{alias}.{var}")
             where_clause = f" WHERE {pred_sql}"
         select_expr = f"{alias}.{var}"
         if expr.projection:
-            proj_sql = translate_expression(expr.projection, context, segment=segment)
+            proj_sql = translate_expression(expr.projection, context, segment="inline")
             for col in ("node_id", "p", "val", "label"):
                 proj_sql = proj_sql.replace(f"{alias}.{col}", f"{alias}.{var}")
             select_expr = proj_sql
         del context.variable_aliases[expr.variable]
         return (
             f"(SELECT JSON_ARRAYAGG({select_expr}) FROM "
-            f"JSON_TABLE({source_sql}, '$[*]' COLUMNS({var} VARCHAR(1000) PATH '$')) {alias}"
+            f"JSON_TABLE({source_sql}, '$[*]' COLUMNS({var} INTEGER PATH '$')) {alias}"
             f"{where_clause})"
         )
 
@@ -2508,20 +2540,23 @@ def translate_expression(expr, context, segment="select") -> str:
         return f"(({inner}) > 0)"
 
     if isinstance(expr, ast.ListComprehension):
-        source_sql = translate_expression(expr.source, context, segment=segment)
+        source_sql = translate_expression(expr.source, context, segment="inline")
         var = sanitize_identifier(expr.variable)
         alias = context.next_alias("lc")
         where_clause = ""
         if expr.predicate:
-            pred_sql = translate_expression(expr.predicate, context, segment=segment)
+            if isinstance(expr.predicate, ast.BooleanExpression):
+                pred_sql = translate_boolean_expression(expr.predicate, context)
+            else:
+                pred_sql = translate_expression(expr.predicate, context, segment="inline")
             where_clause = f" WHERE {pred_sql.replace(var, f'{alias}.{var}')}"
         select_expr = f"{alias}.{var}"
         if expr.projection:
-            proj_sql = translate_expression(expr.projection, context, segment=segment)
+            proj_sql = translate_expression(expr.projection, context, segment="inline")
             select_expr = proj_sql.replace(var, f"{alias}.{var}")
         return (
             f"(SELECT JSON_ARRAYAGG({select_expr}) FROM "
-            f"JSON_TABLE({source_sql}, '$[*]' COLUMNS({var} VARCHAR(1000) PATH '$')) {alias}"
+            f"JSON_TABLE({source_sql}, '$[*]' COLUMNS({var} INTEGER PATH '$')) {alias}"
             f"{where_clause})"
         )
 
@@ -2618,6 +2653,20 @@ def translate_expression(expr, context, segment="select") -> str:
             f'LEFT JOIN {_table("rdf_props")} {p_alias} ON {p_alias}.s = {alias}.node_id AND {p_alias}."key" = {context.add_join_param(expr.property_name)}'
         )
         return f"{p_alias}.val"
+    if isinstance(expr, ast.MapProjection):
+        alias = context.variable_aliases.get(expr.variable, "")
+        parts = []
+        for key_spec, _ in expr.keys:
+            prop = key_spec.lstrip(".")
+            p_alias = context.next_alias("p")
+            context.join_clauses.append(
+                f"LEFT JOIN {_table('rdf_props')} {p_alias} ON {p_alias}.s = {alias}.node_id AND {p_alias}.\"key\" = {context.add_join_param(prop)}"
+            )
+            safe_prop = prop.replace("'", "''")
+            parts.append(f"'\"'||'{safe_prop}'||'\":'||COALESCE('\"'||{p_alias}.val||'\"','null')")
+        if not parts:
+            return "'{}'"
+        return "('{'||" + "||','||".join(parts) + "||'}')"
     if isinstance(expr, ast.MapLiteral):
         if not expr.entries:
             return "'{}'"
@@ -2966,10 +3015,18 @@ def translate_expression(expr, context, segment="select") -> str:
             return f"STDDEV({args[0]})" if args else "NULL"
         if fn in ("stdevp",):
             return f"STDDEV_POP({args[0]})" if args else "NULL"
-        if fn in ("percentiledisc",):
-            return f"PERCENTILE_DISC({args[1] if len(args)>1 else '0.5'}) WITHIN GROUP (ORDER BY {args[0]})" if args else "NULL"
-        if fn in ("percentilecont",):
-            return f"PERCENTILE_CONT({args[1] if len(args)>1 else '0.5'}) WITHIN GROUP (ORDER BY {args[0]})" if args else "NULL"
+        if fn in ("percentiledisc", "percentilecont"):
+            if not args:
+                return "NULL"
+            val_expr = args[0]
+            pct_expr = args[1] if len(args) > 1 else "0.5"
+            context._percentile_queries = getattr(context, "_percentile_queries", [])
+            if args_exprs and isinstance(args_exprs[0], ast.Variable):
+                var_name = args_exprs[0].name
+                alias = context.variable_aliases.get(var_name, "")
+                pct_val = float(pct_expr) if isinstance(pct_expr, str) and pct_expr.replace('.','',1).isdigit() else 0.5
+                context._percentile_queries.append((val_expr, pct_val, fn, var_name, alias))
+            return f"__PERCENTILE_PLACEHOLDER_{len(context._percentile_queries)-1 if context._percentile_queries else 0}__"
         if fn == "haversin":
             return f"(1 - COS({args[0]})) / 2" if args else "NULL"
         if fn == "e":

{iris_vector_graph-1.74.0 → iris_vector_graph-1.76.0}/iris_vector_graph/schema.py

@@ -592,10 +592,10 @@ CREATE OR REPLACE FUNCTION SQLUser.RAND() RETURNS DOUBLE LANGUAGE OBJECTSCRIPT {
 CREATE OR REPLACE FUNCTION SQLUser.NEWID() RETURNS VARCHAR(36) LANGUAGE OBJECTSCRIPT { Quit $SYSTEM.Util.CreateGUID() }
 """,
             """
-CREATE OR REPLACE FUNCTION SQLUser.LIST_REVERSE(j VARCHAR(32000)) RETURNS VARCHAR(32000) LANGUAGE OBJECTSCRIPT { Set arr = ##class(%Library.DynamicArray).%FromJSON(j), out = ##class(%Library.DynamicArray).%New(), n = arr.%Size() For i = n-1:-1:0 { Do out.%Push(arr.%Get(i)) } Quit out.%ToJSON() }
+CREATE OR REPLACE FUNCTION SQLUser.LIST_REVERSE(j VARCHAR(32000)) RETURNS VARCHAR(32000) LANGUAGE OBJECTSCRIPT { Set arr = ##class(%Library.DynamicArray).%FromJSON(j), out = ##class(%Library.DynamicArray).%New(), n = arr.%Size(), i = n-1 While i >= 0 { Do out.%Push(arr.%Get(i)) Set i = i-1 } Quit out.%ToJSON() }
 """,
             """
-CREATE OR REPLACE FUNCTION SQLUser.LIST_TAIL(j VARCHAR(32000)) RETURNS VARCHAR(32000) LANGUAGE OBJECTSCRIPT { Set arr = ##class(%Library.DynamicArray).%FromJSON(j), out = ##class(%Library.DynamicArray).%New(), n = arr.%Size() For i = 1:1:n-1 { Do out.%Push(arr.%Get(i)) } Quit out.%ToJSON() }
+CREATE OR REPLACE FUNCTION SQLUser.LIST_TAIL(j VARCHAR(32000)) RETURNS VARCHAR(32000) LANGUAGE OBJECTSCRIPT { Set arr = ##class(%Library.DynamicArray).%FromJSON(j), out = ##class(%Library.DynamicArray).%New(), n = arr.%Size(), i = 1 While i < n { Do out.%Push(arr.%Get(i)) Set i = i+1 } Quit out.%ToJSON() }
 """,
             """
 CREATE OR REPLACE FUNCTION SQLUser.REGEX_MATCH(s VARCHAR(4000), p VARCHAR(4000)) RETURNS INTEGER LANGUAGE OBJECTSCRIPT { Quit $MATCH(s, p) }

{iris_vector_graph-1.74.0 → iris_vector_graph-1.76.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "iris-vector-graph"
-version = "1.74.0"
+version = "1.76.0"
 description = "Transactional Graph + Vector retrieval system for InterSystems IRIS with hybrid search, openCypher, and GraphQL APIs"
 readme = "README.md"
 license = "MIT"