feat(explore): find code by literal keys and flags

CHANGELOG.md

@@ -189,6 +189,8 @@ and adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
 ### New Features
 
+- `codegraph_explore` finds code containing storage keys, command-line flags, and event names; rebuild existing indexes to enable these matches.
+
 - **Codex and Astra read project guidance from `AGENTS.md`.** The canonical agent guide now lives in `AGENTS.md` (with a nested `docs/AGENTS.md` for long validation notes); `CLAUDE.md` is a thin `@AGENTS.md` wrapper for Claude Code. Codex/Astra no longer miss the old CLAUDE-only instructions.
 
 - **Next.js pages and their navigation are in the graph.** App Router pages (`app/(group)/blog/[slug]/page.tsx` → `/blog/:slug`) and Pages Router pages are routes bound to the component they export, and `<Link href>`, an internal `<a href>`, `router.push` / `router.replace` (`next/navigation` and `next/router`), `redirect()` / `permanentRedirect()` and the middleware's `NextResponse.redirect(…)` are `navigates` edges between them. `app/api/**/route.ts` exports (`GET`, `POST`, …) are endpoints bound to their functions, and `pages/api/*` handlers are `ANY /api/…`. Re-index after upgrading.

__tests__/kernel-retry-materialize.test.ts

@@ -56,7 +56,7 @@ describe.skipIf(!kernelBuilt)('kernel buffer-transport storage (#1541)', () => {
   it('storeExtractionResult persists the decoded nodes of a raw kernel result', async () => {
     const source =
       'def target_fn(root, mission_path):\n' +
-      '    return (root, mission_path)\n' +
+      '    return (root, mission_path, "adapter.mission")\n' +
       '\n' +
       'class Adapter:\n' +
       '    def adapt(self):\n' +
@@ -93,6 +93,7 @@ describe.skipIf(!kernelBuilt)('kernel buffer-transport storage (#1541)', () => {
     expect(nodes.length).toBe(raw!.counts.nodes);
     expect(nodes.map((n) => n.name)).toContain('target_fn');
     expect(nodes.map((n) => n.name)).toContain('Adapter');
+    expect(cg.findLiteralSeedIds('"adapter.mission"').map(id => cg.getNode(id)?.name)).toEqual(['target_fn']);
   });
 });
 

__tests__/literal-capture.test.ts

@@ -0,0 +1,78 @@
+import { describe, it, expect } from 'vitest';
+import type { Node } from '../src/types';
+import { captureLiterals, isSeedLiteral, seedLiteralsInQuery } from '../src/extraction/literal-capture';
+
+function node(id: string, kind: Node['kind'], startLine: number, endLine: number): Node {
+  return {
+    id, kind, name: id, qualifiedName: id, filePath: 'src/a.ts', language: 'typescript',
+    startLine, endLine, startColumn: 0, endColumn: Number.MAX_SAFE_INTEGER, updatedAt: 0,
+  };
+}
+
+describe('isSeedLiteral', () => {
+  it('keeps storage keys, flags, dotted names and paths', () => {
+    for (const v of ['bompus_custom_ds_players', '--start', '-v', 'draft.pick', 'api/v1/users', 'ns:event'])
+      expect(isSeedLiteral(v), v).toBe(v !== '-v');
+  });
+  it('drops plain words, prose, and values that start with a separator', () => {
+    for (const v of ['ready', 'Error', 'not found', './utils', '../x', '', 'a_b'])
+      expect(isSeedLiteral(v), v).toBe(false);
+  });
+});
+
+describe('seedLiteralsInQuery', () => {
+  it('finds quoted spans and bare runs, stripping surrounding punctuation', () => {
+    expect(seedLiteralsInQuery('who writes "bompus_custom_ds_players" (via --start)?'))
+      .toEqual(['bompus_custom_ds_players', '--start']);
+  });
+  it('returns nothing for a symbol-anchored question', () => {
+    expect(seedLiteralsInQuery('callers of espnPlayerKey in shared')).toEqual([]);
+  });
+});
+
+describe('captureLiterals', () => {
+  const source = [
+    `import { x } from './utils/helpers';`,            // 1: path starts with '.', never qualifies
+    `const KEY = 'bompus_custom_ds_players';`,          // 2: top level → file node
+    `export function save() {`,                         // 3
+    `  storage.set('bompus_custom_ds_players', 1);`,    // 4
+    `  log('ready');`,                                  // 5: plain word
+    `  emit(\`draft.pick\`); emit(\`draft.\${n}\`);`,    // 6: second one interpolates
+    `}`,                                                // 7
+    `export class Boot { start() { run('--start'); } }`,// 8
+  ].join('\n');
+
+  it('attributes each literal to the innermost enclosing symbol, else the file', () => {
+    const file = node('file:src/a.ts', 'file', 1, 8);
+    const save = node('save', 'function', 3, 7);
+    const boot = node('Boot', 'class', 8, 8);
+    const start = node('start', 'method', 8, 8);
+    const imp = node('./utils/helpers', 'import', 1, 1);
+    const nodes = [file, imp, save, boot, start];
+    captureLiterals(source, nodes);
+    expect(file.literals).toEqual(['bompus_custom_ds_players']);
+    expect(save.literals).toEqual(['bompus_custom_ds_players', 'draft.pick']);
+    expect(start.literals).toEqual(['--start']);
+    expect(boot.literals).toBeUndefined();
+    expect(imp.literals).toBeUndefined();
+  });
+
+  it('dedupes per node and caps at 32', () => {
+    const many = Array.from({ length: 40 }, (_, i) => `k('key_${i}'); k('key_${i}');`).join('\n');
+    const fn = node('f', 'function', 1, 40);
+    captureLiterals(many, [node('file:src/a.ts', 'file', 1, 40), fn]);
+    expect(fn.literals).toHaveLength(32);
+    expect(new Set(fn.literals).size).toBe(32);
+  });
+
+  it('uses UTF-16 columns to distinguish same-line siblings after non-ASCII source', () => {
+    const prefix = '/* café 😀 */ ';
+    const first = "function writer(){return 'cache.write';}";
+    const second = "function reader(){return 'cache.read';}";
+    const writer = { ...node('writer', 'function', 1, 1), startColumn: prefix.length, endColumn: (prefix + first).length };
+    const reader = { ...node('reader', 'function', 1, 1), startColumn: (prefix + first).length, endColumn: (prefix + first + second).length };
+    captureLiterals(prefix + first + second, [writer, reader]);
+    expect(writer.literals).toEqual(['cache.write']);
+    expect(reader.literals).toEqual(['cache.read']);
+  });
+});

__tests__/literal-compiled-index.test.ts

@@ -0,0 +1,59 @@
+import { afterEach, describe, expect, it, vi } from 'vitest';
+import * as fs from 'node:fs';
+import * as os from 'node:os';
+import * as path from 'node:path';
+import type CodeGraph from '../src';
+import type { QueryBuilder } from '../src/db/queries';
+
+const built = path.join(__dirname, '..', 'dist', 'index.js');
+const kernel = path.join(__dirname, '..', 'codegraph-kernel', 'prebuilds', `${process.platform}-${process.arch}`, 'codegraph-kernel.node');
+
+// Source-mode suites cannot exercise the compiled parse/store worker boundary.
+describe.runIf(fs.existsSync(built))('literal persistence through compiled indexing', () => {
+  let dir: string | undefined;
+  let cg: CodeGraph | undefined;
+  afterEach(() => {
+    cg?.destroy();
+    cg = undefined;
+    if (dir) fs.rmSync(dir, { recursive: true, force: true });
+    dir = undefined;
+    vi.unstubAllEnvs();
+  });
+
+  for (const mode of ['native-worker', 'native-main', 'wasm-worker']) {
+    it.runIf(mode === 'wasm-worker' || fs.existsSync(kernel))(`${mode}: fresh, unchanged, and migrated indexes retain exact owners`, async () => {
+      vi.stubEnv('CODEGRAPH_PARSE_WORKERS', '1');
+      vi.stubEnv('CODEGRAPH_KERNEL', mode === 'wasm-worker' ? '0' : '1');
+      vi.stubEnv('CODEGRAPH_NO_STORE_WORKER', mode === 'native-main' ? '1' : '0');
+      const BuiltCodeGraph: typeof CodeGraph = require(built).default;
+      dir = fs.mkdtempSync(path.join(os.tmpdir(), 'cg-literal-compiled-'));
+      fs.writeFileSync(path.join(dir, 'cache.py'), "def persist():\n    return 'cache.write'\n");
+      fs.writeFileSync(path.join(dir, 'siblings.ts'), "/* café 😀 */ export function writer(){return 'sibling.write';} export function reader(){return 'sibling.read';}\n");
+      cg = await BuiltCodeGraph.init(dir, { silent: true });
+      const names = (key: string) => cg!.findLiteralSeedIds(key).map(id => cg!.getNode(id)?.name);
+      const assertOwners = () => {
+        expect(names('cache.write')).toEqual(['persist']);
+        expect(names('sibling.write')).toEqual(['writer']);
+        expect(names('sibling.read')).toEqual(['reader']);
+      };
+      await cg.indexAll();
+      assertOwners();
+      await cg.indexAll();
+      assertOwners();
+
+      // A schema-only migration has no literal data; unchanged source still needs backfill.
+      const queries = (cg as unknown as { queries: QueryBuilder }).queries;
+      queries.replaceLiteralsForFile('cache.py', []);
+      queries.replaceLiteralsForFile('siblings.ts', []);
+      queries.setMetadata('indexed_with_extraction_version', '26');
+      expect(cg.isIndexStale()).toBe(true);
+      await cg.indexAll();
+      assertOwners();
+      expect(cg.isIndexStale()).toBe(false);
+      fs.rmSync(path.join(dir, 'cache.py'));
+      await cg.indexAll();
+      expect(names('cache.write')).toEqual([]);
+      expect(names('sibling.write')).toEqual(['writer']);
+    });
+  }
+});