SET allows labeling of path variables (openCypher violation)

[zihaozeng0021]

Configuration

FalkorDB Version: falkordb/falkordb:latest (pulled 2025-12-15)

Operating System: WSL2 – Ubuntu 24.04.1 LTS

Installation Method: Docker

---

Steps to reproduce

1. Start falkorDB via Docker

docker run --rm -p 6379:6379 falkordb/falkordb:latest

2. Connect with redis-cli

redis-cli -h 127.0.0.1 -p 6379

3. Run the following queries

GRAPH.DELETE g

GRAPH.QUERY g "CREATE (:X {k:10})-[:S {k:10}]->(:Y {k:10})"

GRAPH.QUERY g "MATCH p=(a:X {k:10})-[r:S {k:10}]->(b:Y {k:10}) SET p:L RETURN r"

---

Expected behaviour

The query should fail with a semantic/type error.

In Cypher Query Language Reference, Version 9

• A label is a token that is assigned to nodes only. (See page 5)
• The SET clause is used to update labels on nodes and properties on nodes and relationships. (See page 107)

p is a named path, so SET p:L is invalid and should be rejected during semantic analysis (e.g. “expected Node but was Path”).

Actual behaviour

The query executes successfully with no error and returns the relationship r.
The invalid SET p:L clause is silently accepted.

---

This repo is using Opire - what does it mean? 👇

💵 Everyone can add rewards for this issue commenting /reward 100 (replace 100 with the amount).
🕵️‍♂️ If someone starts working on this issue to earn the rewards, they can comment /try to let everyone know!
🙌 And when they open the PR, they can comment /claim #1468 either in the PR description or in a PR's comment.

🪙 Also, everyone can tip any user commenting /tip 20 @zihaozeng0021 (replace 20 with the amount, and @zihaozeng0021 with the user to tip).

📖 If you want to learn more, check out our documentation.

[coderabbitai]

📝 CodeRabbit Plan Mode

Generate an implementation plan and prompts that you can use with your favorite coding agent.

• Create Plan

Examples

• Example 1
• Example 2

---

🔗 Similar Issues

Possible Duplicates

• ISSUE-1394

Related Issues

• SET incorrectly allows assigning properties to a Path variable (SET p.k1 = ... should fail) #1394

👤 Suggested Assignees

• swf2000

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord or schedule a call!

[overcut-ai]

Completed Working on "Automated Ticket Triage "

✅ Automated triage of issue #1468 completed: labels bug, component:query-engine, p2 applied; triage summary comment posted; no duplicates found.

---

👉 View complete log

[overcut-ai]

Triage Summary

• Component: component:query-engine (confidence 0.82) – semantic/type validation for SET clauses is handled during query planning/execution.
• Category: bug (confidence 0.95) – behavior violates the Cypher spec and silently accepts an invalid query shape.
• Priority: p2 (confidence 0.78) – correctness issue affecting standards compliance, but no crash or data loss, so medium severity.
• Duplicates: None found after scanning currently open issues; related report SET incorrectly allows assigning properties to a Path variable (SET p.k1 = ... should fail) #1394 covers property assignment on paths, which is similar but distinct (properties vs. labels), so no linkage made.

Actions Taken

1. Added labels: bug, component:query-engine, p2.
2. No duplicate labels or comments required.

Suggested Next Steps for Engineering

• Extend semantic analysis so SET targets are limited to nodes (for labels) or nodes/relationships (for properties); explicitly reject path variables with a descriptive error.
• Add regression coverage (e.g., unit + flow test) verifying SET p:Label yields a semantic/type error message.
• Audit other SET variants (properties, relationship labels) to ensure consistent node/relationship-only enforcement.

Please let me know if further context is needed!

[overcut-ai]

Completed Working on "Auto Root Cause Analysis"

✅ Workflow completed successfully.

---

👉 View complete log

[overcut-ai]

🛠 Root Cause Analysis (RCA)

Summary (openCypher compliance):
SET clauses are meant to modify node labels or properties. FalkorDB currently allows the clause SET p:L where p is a named path. Because the semantic analyser never verifies that the identifier on the left‐hand side is a node alias, the query reaches execution. At runtime the update is silently ignored when the alias cannot be resolved to a node, so the query completes without error – violating the openCypher specification.

Evidence

File	Line(s)	Note
src/ast/ast_validations.c	885-913	Node aliases are stored with type T_NODE, named paths are stored with NULL, so their kind is lost.
src/ast/ast_validations.c	1580-1614 (_Validate_set_property)	Only checks LHS is an identifier – no type check that it is a node alias.
src/ast/ast_build_op_contexts.c	118-125	Update contexts for CYPHER_AST_SET_LABELS accept any alias, including paths.
src/execution_plan/ops/shared/update_functions.c	382-401 (EvalEntityUpdates)	If Record_GetType returns REC_TYPE_UNKNOWN (alias isn’t a node), function returns silently – explains lack of runtime error.

Root Cause:
Semantic validation does not reject SET … label operations whose LHS resolves to anything other than a node alias. Named path identifiers therefore bypass validation, and execution treats the update as a no-op instead of raising an error.

Impacted Areas:
Any SET / REMOVE clause that targets an alias which is not a node (e.g. paths, edges, scalars) will be silently ignored, potentially hiding user mistakes and diverging from openCypher behaviour.

Confidence: 0.63

---

💡 Fix Suggestion

• Description: Extend semantic validation to (a) store named-path aliases with a distinct type (e.g., T_PATH), and (b) reject SET label items whose alias type is not T_NODE. This surfaces a clear compile-time error instead of silently ignoring the clause.

diff --git a/src/errors/error_msgs.h b/src/errors/error_msgs.h
@@
-#define EMSG_SET_LHS_NON_ALIAS "FalkorDB does not currently support non-alias references on the left-hand side of SET expressions"
+#define EMSG_SET_LHS_NON_ALIAS "FalkorDB does not currently support non-alias references on the left-hand side of SET expressions"
+#define EMSG_SET_LABELS_ON_NON_NODE "Only node variables may receive labels (alias '%s')"

diff --git a/src/ast/ast_validations.c b/src/ast/ast_validations.c
@@
-    const char *alias = cypher_ast_identifier_get_name(alias_node);
-    _IdentifierAdd(vctx, alias, NULL);
+    /* register named path alias explicitly as path type */
+    const char *alias = cypher_ast_identifier_get_name(alias_node);
+    _IdentifierAdd(vctx, alias, (void*)T_PATH);
@@
     vctx->clause = cypher_astnode_type(n);
-    return VISITOR_RECURSE;
+    /* Extra validation: ensure label updates target only nodes */
+    uint nitems = cypher_ast_set_nitems(n);
+    for(uint i = 0; i < nitems; i++) {
+        const cypher_astnode_t *item = cypher_ast_set_get_item(n, i);
+        if(cypher_astnode_type(item) != CYPHER_AST_SET_LABELS) continue;
+
+        const cypher_astnode_t *id = cypher_ast_set_labels_get_identifier(item);
+        const char *alias = cypher_ast_identifier_get_name(id);
+        void *alias_type = _IdentifiersFind(vctx, alias);
+
+        if(alias_type == (void*)T_EDGE || alias_type == (void*)T_PATH) {
+            ErrorCtx_SetError(EMSG_SET_LABELS_ON_NON_NODE, alias);
+            return VISITOR_BREAK;
+        }
+    }
+    return VISITOR_RECURSE;
 }

(≈35 LOC added/modified.)

Risk Assessment

Low → Moderate. Changes are limited to compile-time validation; however, changing how path aliases are stored (T_PATH vs NULL) touches identifier lookup logic. Run the full test-suite and pay attention to any code that assumed NULL for path aliases.

Test Recommendations

1. ❌ Invalid label on path should error:

   MATCH p=(a:X {k:10})-[r:S {k:10}]->(b:Y {k:10})
   SET p:L

   Expect: semantic error "Only node variables may receive labels".
2. ✔️ Valid node labeling still works:

   MATCH (a:X {k:10})
   SET a:L
   RETURN a

3. ✔️ Other invalid cases also error (edge alias):

   MATCH (a)-[r]->(b)
   SET r:RELTYPE

---

ℹ️ Let us know if further clarification or additional reproduction details are needed.