fix(extraction): preserve symbols with colliding same-line IDs (#1349) (#2030)

Same-kind, same-name declarations on one line shared an ID and silently overwrote each other in SQLite.
Use per-extraction collision tracking to append a UTF-16 column only to later distinct declarations.
Apply the same rule across wasm, Rust language walkers, Liquid, and CFML while preserving file IDs and edge endpoints.
Add persistence, call-edge, identity-vector, and Unicode parity regressions.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

src/extraction/cfml-extractor.ts

@@ -1,6 +1,6 @@
 import type { Node as SyntaxNode } from 'web-tree-sitter';
 import { Node, Edge, ExtractionResult, ExtractionError, UnresolvedReference, Language } from '../types';
-import { generateNodeId } from './tree-sitter-helpers';
+import { generateNodeId, NodeIdAllocator } from './tree-sitter-helpers';
 import { TreeSitterExtractor } from './tree-sitter';
 import { getParser } from './grammars';
 
@@ -22,6 +22,7 @@ export class CfmlExtractor {
   private source: string;
   private language: Language;
   private nodes: Node[] = [];
+  private nodeIds = new NodeIdAllocator();
   private edges: Edge[] = [];
   private unresolvedReferences: UnresolvedReference[] = [];
   private errors: ExtractionError[] = [];
@@ -174,7 +175,7 @@ export class CfmlExtractor {
    */
   private extractComponent(openTag: SyntaxNode, containerId: string | undefined): SyntaxNode {
     const name = this.tagAttr(openTag, 'name') ?? this.componentNameFromPath();
-    const id = generateNodeId(this.filePath, 'class', name, openTag.startPosition.row + 1);
+    const id = this.nodeIds.generate(this.filePath, 'class', name, openTag.startPosition.row + 1, openTag.startPosition.column);
 
     const classNode: Node = {
       id,
@@ -261,7 +262,7 @@ export class CfmlExtractor {
     if (!name) return;
 
     const kind = parentClassId ? 'method' : 'function';
-    const id = generateNodeId(this.filePath, kind, name, tag.startPosition.row + 1);
+    const id = this.nodeIds.generate(this.filePath, kind, name, tag.startPosition.row + 1, tag.startPosition.column);
     const access = this.tagAttr(tag, 'access');
     const visibility = access === 'private' ? 'private'
       : access === 'package' ? 'internal'

src/extraction/liquid-extractor.ts

@@ -1,5 +1,5 @@
 import { Node, Edge, ExtractionResult, ExtractionError, UnresolvedReference } from '../types';
-import { generateNodeId } from './tree-sitter-helpers';
+import { generateNodeId, NodeIdAllocator } from './tree-sitter-helpers';
 
 /**
  * LiquidExtractor - Extracts relationships from Liquid template files
@@ -14,6 +14,7 @@ export class LiquidExtractor {
   private filePath: string;
   private source: string;
   private nodes: Node[] = [];
+  private nodeIds = new NodeIdAllocator();
   private edges: Edge[] = [];
   private unresolvedReferences: UnresolvedReference[] = [];
   private errors: ExtractionError[] = [];
@@ -217,7 +218,7 @@ export class LiquidExtractor {
       const line = this.getLineNumber(match.index);
 
       // Create an import node for searchability
-      const importNodeId = generateNodeId(this.filePath, 'import', snippetName!, line);
+      const importNodeId = this.nodeIds.generate(this.filePath, 'import', snippetName!, line, match.index - this.getLineStart(line));
       const importNode: Node = {
         id: importNodeId,
         kind: 'import',
@@ -242,7 +243,7 @@ export class LiquidExtractor {
       });
 
       // Create a component node for the snippet reference
-      const nodeId = generateNodeId(this.filePath, 'component', `${tagType}:${snippetName}`, line);
+      const nodeId = this.nodeIds.generate(this.filePath, 'component', `${tagType}:${snippetName}`, line, match.index - this.getLineStart(line));
 
       const node: Node = {
         id: nodeId,
@@ -289,7 +290,7 @@ export class LiquidExtractor {
       const line = this.getLineNumber(match.index);
 
       // Create an import node for searchability
-      const importNodeId = generateNodeId(this.filePath, 'import', sectionName!, line);
+      const importNodeId = this.nodeIds.generate(this.filePath, 'import', sectionName!, line, match.index - this.getLineStart(line));
       const importNode: Node = {
         id: importNodeId,
         kind: 'import',
@@ -314,7 +315,7 @@ export class LiquidExtractor {
       });
 
       // Create a component node for the section reference
-      const nodeId = generateNodeId(this.filePath, 'component', `section:${sectionName}`, line);
+      const nodeId = this.nodeIds.generate(this.filePath, 'component', `section:${sectionName}`, line, match.index - this.getLineStart(line));
 
       const node: Node = {
         id: nodeId,
@@ -378,7 +379,7 @@ export class LiquidExtractor {
       }
 
       // Create a node for the schema
-      const nodeId = generateNodeId(this.filePath, 'constant', `schema:${schemaName}`, startLine);
+      const nodeId = this.nodeIds.generate(this.filePath, 'constant', `schema:${schemaName}`, startLine, match.index - this.getLineStart(startLine));
 
       const node: Node = {
         id: nodeId,
@@ -419,7 +420,7 @@ export class LiquidExtractor {
       const line = this.getLineNumber(match.index);
 
       // Create a variable node
-      const nodeId = generateNodeId(this.filePath, 'variable', variableName!, line);
+      const nodeId = this.nodeIds.generate(this.filePath, 'variable', variableName!, line, match.index - this.getLineStart(line));
 
       const node: Node = {
         id: nodeId,

src/extraction/tree-sitter-helpers.ts

@@ -29,6 +29,23 @@ export function generateNodeId(
   return `${kind}:${hash}`;
 }
 
+/** Per-extraction identities: preserve legacy IDs unless distinct source positions collide. */
+export class NodeIdAllocator {
+  private firstColumns = new Map<string, number>();
+
+  generate(filePath: string, kind: NodeKind, name: string, line: number, column: number): string {
+    const id = generateNodeId(filePath, kind, name, line);
+    const firstColumn = this.firstColumns.get(id);
+    if (firstColumn === undefined) {
+      this.firstColumns.set(id, column);
+      return id;
+    }
+    // Columns are zero-based UTF-16 code units in both wasm and the kernel.
+    // Revisiting the same declaration must still produce the same identity.
+    return firstColumn === column ? id : `${id}:${column}`;
+  }
+}
+
 /**
  * Extract text from a syntax node
  */

src/extraction/tree-sitter.ts

@@ -16,7 +16,7 @@ import {
   UnresolvedReference,
 } from '../types';
 import { getParser, detectLanguage, isLanguageSupported, isFileLevelOnlyLanguage } from './grammars';
-import { generateNodeId, getNodeText, getChildByField, getPrecedingDocstring } from './tree-sitter-helpers';
+import { NodeIdAllocator, getNodeText, getChildByField, getPrecedingDocstring } from './tree-sitter-helpers';
 import { FN_REF_SPECS, captureFnRefCandidates, type FnRefSpec, type FnRefCandidate } from './function-ref';
 import { isGeneratedFile } from './generated-detection';
 import type { LanguageExtractor, ExtractorContext } from './tree-sitter-types';
@@ -454,6 +454,7 @@ export class TreeSitterExtractor {
   private source: string;
   private tree: Tree | null = null;
   private nodes: Node[] = [];
+  private nodeIds = new NodeIdAllocator();
   private edges: Edge[] = [];
   private unresolvedReferences: UnresolvedReference[] = [];
   // Value-reference edges (default ON; set CODEGRAPH_VALUE_REFS=0 to disable; see flushValueRefs).
@@ -1412,7 +1413,7 @@ export class TreeSitterExtractor {
       return null;
     }
 
-    const id = generateNodeId(this.filePath, kind, name, node.startPosition.row + 1);
+    const id = this.nodeIds.generate(this.filePath, kind, name, node.startPosition.row + 1, node.startPosition.column);
 
     // Some grammars (e.g. Dart) model a function/method body as a *sibling* of
     // the signature node, so the declaration node's own range is just the