opencode-codebase-index 0.5.1 → 0.5.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +38 -7
- package/commands/call-graph.md +24 -0
- package/dist/cli.cjs +1232 -40
- package/dist/cli.cjs.map +1 -1
- package/dist/cli.js +1232 -40
- package/dist/cli.js.map +1 -1
- package/dist/index.cjs +1233 -41
- package/dist/index.cjs.map +1 -1
- package/dist/index.js +1233 -41
- package/dist/index.js.map +1 -1
- package/native/codebase-index-native.darwin-arm64.node +0 -0
- package/native/codebase-index-native.darwin-x64.node +0 -0
- package/native/codebase-index-native.linux-arm64-gnu.node +0 -0
- package/native/codebase-index-native.linux-x64-gnu.node +0 -0
- package/native/codebase-index-native.win32-x64-msvc.node +0 -0
- package/package.json +2 -3
- package/skill/SKILL.md +13 -1
package/dist/index.cjs
CHANGED
|
@@ -785,9 +785,15 @@ function getDefaultSearchConfig() {
|
|
|
785
785
|
minScore: 0.1,
|
|
786
786
|
includeContext: true,
|
|
787
787
|
hybridWeight: 0.5,
|
|
788
|
+
fusionStrategy: "rrf",
|
|
789
|
+
rrfK: 60,
|
|
790
|
+
rerankTopN: 20,
|
|
788
791
|
contextLines: 0
|
|
789
792
|
};
|
|
790
793
|
}
|
|
794
|
+
function isValidFusionStrategy(value) {
|
|
795
|
+
return value === "weighted" || value === "rrf";
|
|
796
|
+
}
|
|
791
797
|
function getDefaultDebugConfig() {
|
|
792
798
|
return {
|
|
793
799
|
enabled: false,
|
|
@@ -842,6 +848,9 @@ function parseConfig(raw) {
|
|
|
842
848
|
minScore: typeof rawSearch.minScore === "number" ? rawSearch.minScore : defaultSearch.minScore,
|
|
843
849
|
includeContext: typeof rawSearch.includeContext === "boolean" ? rawSearch.includeContext : defaultSearch.includeContext,
|
|
844
850
|
hybridWeight: typeof rawSearch.hybridWeight === "number" ? Math.min(1, Math.max(0, rawSearch.hybridWeight)) : defaultSearch.hybridWeight,
|
|
851
|
+
fusionStrategy: isValidFusionStrategy(rawSearch.fusionStrategy) ? rawSearch.fusionStrategy : defaultSearch.fusionStrategy,
|
|
852
|
+
rrfK: typeof rawSearch.rrfK === "number" ? Math.max(1, Math.floor(rawSearch.rrfK)) : defaultSearch.rrfK,
|
|
853
|
+
rerankTopN: typeof rawSearch.rerankTopN === "number" ? Math.min(200, Math.max(0, Math.floor(rawSearch.rerankTopN))) : defaultSearch.rerankTopN,
|
|
845
854
|
contextLines: typeof rawSearch.contextLines === "number" ? Math.min(50, Math.max(0, rawSearch.contextLines)) : defaultSearch.contextLines
|
|
846
855
|
};
|
|
847
856
|
const rawDebug = input.debug && typeof input.debug === "object" ? input.debug : {};
|
|
@@ -2977,6 +2986,9 @@ function hashContent(content) {
|
|
|
2977
2986
|
function hashFile(filePath) {
|
|
2978
2987
|
return native.hashFile(filePath);
|
|
2979
2988
|
}
|
|
2989
|
+
function extractCalls(content, language) {
|
|
2990
|
+
return native.extractCalls(content, language);
|
|
2991
|
+
}
|
|
2980
2992
|
var VectorStore = class {
|
|
2981
2993
|
inner;
|
|
2982
2994
|
dimensions;
|
|
@@ -3308,6 +3320,12 @@ var Database = class {
|
|
|
3308
3320
|
getChunksByFile(filePath) {
|
|
3309
3321
|
return this.inner.getChunksByFile(filePath);
|
|
3310
3322
|
}
|
|
3323
|
+
getChunksByName(name) {
|
|
3324
|
+
return this.inner.getChunksByName(name);
|
|
3325
|
+
}
|
|
3326
|
+
getChunksByNameCi(name) {
|
|
3327
|
+
return this.inner.getChunksByNameCi(name);
|
|
3328
|
+
}
|
|
3311
3329
|
deleteChunksByFile(filePath) {
|
|
3312
3330
|
return this.inner.deleteChunksByFile(filePath);
|
|
3313
3331
|
}
|
|
@@ -3351,6 +3369,73 @@ var Database = class {
|
|
|
3351
3369
|
getStats() {
|
|
3352
3370
|
return this.inner.getStats();
|
|
3353
3371
|
}
|
|
3372
|
+
// ── Symbol methods ──────────────────────────────────────────────
|
|
3373
|
+
upsertSymbol(symbol) {
|
|
3374
|
+
this.inner.upsertSymbol(symbol);
|
|
3375
|
+
}
|
|
3376
|
+
upsertSymbolsBatch(symbols) {
|
|
3377
|
+
if (symbols.length === 0) return;
|
|
3378
|
+
this.inner.upsertSymbolsBatch(symbols);
|
|
3379
|
+
}
|
|
3380
|
+
getSymbolsByFile(filePath) {
|
|
3381
|
+
return this.inner.getSymbolsByFile(filePath);
|
|
3382
|
+
}
|
|
3383
|
+
getSymbolByName(name, filePath) {
|
|
3384
|
+
return this.inner.getSymbolByName(name, filePath) ?? null;
|
|
3385
|
+
}
|
|
3386
|
+
getSymbolsByName(name) {
|
|
3387
|
+
return this.inner.getSymbolsByName(name);
|
|
3388
|
+
}
|
|
3389
|
+
getSymbolsByNameCi(name) {
|
|
3390
|
+
return this.inner.getSymbolsByNameCi(name);
|
|
3391
|
+
}
|
|
3392
|
+
deleteSymbolsByFile(filePath) {
|
|
3393
|
+
return this.inner.deleteSymbolsByFile(filePath);
|
|
3394
|
+
}
|
|
3395
|
+
// ── Call Edge methods ────────────────────────────────────────────
|
|
3396
|
+
upsertCallEdge(edge) {
|
|
3397
|
+
this.inner.upsertCallEdge(edge);
|
|
3398
|
+
}
|
|
3399
|
+
upsertCallEdgesBatch(edges) {
|
|
3400
|
+
if (edges.length === 0) return;
|
|
3401
|
+
this.inner.upsertCallEdgesBatch(edges);
|
|
3402
|
+
}
|
|
3403
|
+
getCallers(targetName, branch) {
|
|
3404
|
+
return this.inner.getCallers(targetName, branch);
|
|
3405
|
+
}
|
|
3406
|
+
getCallersWithContext(targetName, branch) {
|
|
3407
|
+
return this.inner.getCallersWithContext(targetName, branch);
|
|
3408
|
+
}
|
|
3409
|
+
getCallees(symbolId, branch) {
|
|
3410
|
+
return this.inner.getCallees(symbolId, branch);
|
|
3411
|
+
}
|
|
3412
|
+
deleteCallEdgesByFile(filePath) {
|
|
3413
|
+
return this.inner.deleteCallEdgesByFile(filePath);
|
|
3414
|
+
}
|
|
3415
|
+
resolveCallEdge(edgeId, toSymbolId) {
|
|
3416
|
+
this.inner.resolveCallEdge(edgeId, toSymbolId);
|
|
3417
|
+
}
|
|
3418
|
+
// ── Branch Symbol methods ────────────────────────────────────────
|
|
3419
|
+
addSymbolsToBranch(branch, symbolIds) {
|
|
3420
|
+
this.inner.addSymbolsToBranch(branch, symbolIds);
|
|
3421
|
+
}
|
|
3422
|
+
addSymbolsToBranchBatch(branch, symbolIds) {
|
|
3423
|
+
if (symbolIds.length === 0) return;
|
|
3424
|
+
this.inner.addSymbolsToBranchBatch(branch, symbolIds);
|
|
3425
|
+
}
|
|
3426
|
+
getBranchSymbolIds(branch) {
|
|
3427
|
+
return this.inner.getBranchSymbolIds(branch);
|
|
3428
|
+
}
|
|
3429
|
+
clearBranchSymbols(branch) {
|
|
3430
|
+
return this.inner.clearBranchSymbols(branch);
|
|
3431
|
+
}
|
|
3432
|
+
// ── GC methods for symbols/edges ─────────────────────────────────
|
|
3433
|
+
gcOrphanSymbols() {
|
|
3434
|
+
return this.inner.gcOrphanSymbols();
|
|
3435
|
+
}
|
|
3436
|
+
gcOrphanCallEdges() {
|
|
3437
|
+
return this.inner.gcOrphanCallEdges();
|
|
3438
|
+
}
|
|
3354
3439
|
};
|
|
3355
3440
|
|
|
3356
3441
|
// src/git/index.ts
|
|
@@ -3458,6 +3543,29 @@ function getHeadPath(repoRoot) {
|
|
|
3458
3543
|
}
|
|
3459
3544
|
|
|
3460
3545
|
// src/indexer/index.ts
|
|
3546
|
+
var CALL_GRAPH_LANGUAGES = /* @__PURE__ */ new Set(["typescript", "tsx", "javascript", "jsx", "python", "go", "rust"]);
|
|
3547
|
+
var CALL_GRAPH_SYMBOL_CHUNK_TYPES = /* @__PURE__ */ new Set([
|
|
3548
|
+
"function_declaration",
|
|
3549
|
+
"function",
|
|
3550
|
+
"arrow_function",
|
|
3551
|
+
"method_definition",
|
|
3552
|
+
"class_declaration",
|
|
3553
|
+
"interface_declaration",
|
|
3554
|
+
"type_alias_declaration",
|
|
3555
|
+
"enum_declaration",
|
|
3556
|
+
"function_definition",
|
|
3557
|
+
"class_definition",
|
|
3558
|
+
"decorated_definition",
|
|
3559
|
+
"method_declaration",
|
|
3560
|
+
"type_declaration",
|
|
3561
|
+
"type_spec",
|
|
3562
|
+
"function_item",
|
|
3563
|
+
"impl_item",
|
|
3564
|
+
"struct_item",
|
|
3565
|
+
"enum_item",
|
|
3566
|
+
"trait_item",
|
|
3567
|
+
"mod_item"
|
|
3568
|
+
]);
|
|
3461
3569
|
function float32ArrayToBuffer(arr) {
|
|
3462
3570
|
const float32 = new Float32Array(arr);
|
|
3463
3571
|
return Buffer.from(float32.buffer);
|
|
@@ -3482,6 +3590,892 @@ function isRateLimitError(error) {
|
|
|
3482
3590
|
return message.includes("429") || message.toLowerCase().includes("rate limit") || message.toLowerCase().includes("too many requests");
|
|
3483
3591
|
}
|
|
3484
3592
|
var INDEX_METADATA_VERSION = "1";
|
|
3593
|
+
var RANKING_TOKEN_CACHE_LIMIT = 4096;
|
|
3594
|
+
var rankingQueryTokenCache = /* @__PURE__ */ new Map();
|
|
3595
|
+
var rankingNameTokenCache = /* @__PURE__ */ new Map();
|
|
3596
|
+
var rankingPathTokenCache = /* @__PURE__ */ new Map();
|
|
3597
|
+
var rankingTextTokenCache = /* @__PURE__ */ new Map();
|
|
3598
|
+
var STOPWORDS = /* @__PURE__ */ new Set([
|
|
3599
|
+
"the",
|
|
3600
|
+
"and",
|
|
3601
|
+
"for",
|
|
3602
|
+
"with",
|
|
3603
|
+
"from",
|
|
3604
|
+
"that",
|
|
3605
|
+
"this",
|
|
3606
|
+
"into",
|
|
3607
|
+
"using",
|
|
3608
|
+
"where",
|
|
3609
|
+
"what",
|
|
3610
|
+
"when",
|
|
3611
|
+
"why",
|
|
3612
|
+
"how",
|
|
3613
|
+
"are",
|
|
3614
|
+
"was",
|
|
3615
|
+
"were",
|
|
3616
|
+
"be",
|
|
3617
|
+
"been",
|
|
3618
|
+
"being",
|
|
3619
|
+
"find",
|
|
3620
|
+
"show",
|
|
3621
|
+
"get",
|
|
3622
|
+
"run",
|
|
3623
|
+
"use",
|
|
3624
|
+
"code",
|
|
3625
|
+
"function",
|
|
3626
|
+
"implementation",
|
|
3627
|
+
"retrieve",
|
|
3628
|
+
"results",
|
|
3629
|
+
"result",
|
|
3630
|
+
"search",
|
|
3631
|
+
"pipeline",
|
|
3632
|
+
"top",
|
|
3633
|
+
"in",
|
|
3634
|
+
"on",
|
|
3635
|
+
"of",
|
|
3636
|
+
"to",
|
|
3637
|
+
"by",
|
|
3638
|
+
"as",
|
|
3639
|
+
"or",
|
|
3640
|
+
"an",
|
|
3641
|
+
"a"
|
|
3642
|
+
]);
|
|
3643
|
+
var TEST_PATH_SEGMENTS = [
|
|
3644
|
+
"tests/",
|
|
3645
|
+
"__tests__/",
|
|
3646
|
+
"/test/",
|
|
3647
|
+
"fixtures/",
|
|
3648
|
+
"benchmark",
|
|
3649
|
+
"README",
|
|
3650
|
+
"ARCHITECTURE",
|
|
3651
|
+
"docs/"
|
|
3652
|
+
];
|
|
3653
|
+
var IMPLEMENTATION_EXCLUDE_PATH_SEGMENTS = [
|
|
3654
|
+
"tests/",
|
|
3655
|
+
"__tests__/",
|
|
3656
|
+
"/test/",
|
|
3657
|
+
"fixtures/",
|
|
3658
|
+
"benchmark",
|
|
3659
|
+
"readme",
|
|
3660
|
+
"architecture",
|
|
3661
|
+
"docs/",
|
|
3662
|
+
"examples/",
|
|
3663
|
+
"example/",
|
|
3664
|
+
".github/",
|
|
3665
|
+
"/scripts/",
|
|
3666
|
+
"/migrations/",
|
|
3667
|
+
"/generated/"
|
|
3668
|
+
];
|
|
3669
|
+
var SOURCE_INTENT_HINTS = /* @__PURE__ */ new Set([
|
|
3670
|
+
"implement",
|
|
3671
|
+
"implementation",
|
|
3672
|
+
"function",
|
|
3673
|
+
"method",
|
|
3674
|
+
"class",
|
|
3675
|
+
"logic",
|
|
3676
|
+
"algorithm",
|
|
3677
|
+
"pipeline",
|
|
3678
|
+
"indexer",
|
|
3679
|
+
"where"
|
|
3680
|
+
]);
|
|
3681
|
+
var DOC_TEST_INTENT_HINTS = /* @__PURE__ */ new Set([
|
|
3682
|
+
"test",
|
|
3683
|
+
"tests",
|
|
3684
|
+
"fixture",
|
|
3685
|
+
"fixtures",
|
|
3686
|
+
"benchmark",
|
|
3687
|
+
"readme",
|
|
3688
|
+
"docs",
|
|
3689
|
+
"documentation"
|
|
3690
|
+
]);
|
|
3691
|
+
var DOC_INTENT_HINTS = /* @__PURE__ */ new Set([
|
|
3692
|
+
"readme",
|
|
3693
|
+
"docs",
|
|
3694
|
+
"documentation",
|
|
3695
|
+
"guide",
|
|
3696
|
+
"usage"
|
|
3697
|
+
]);
|
|
3698
|
+
function setBoundedCache(cache, key, value) {
|
|
3699
|
+
if (cache.size >= RANKING_TOKEN_CACHE_LIMIT) {
|
|
3700
|
+
const oldest = cache.keys().next().value;
|
|
3701
|
+
if (oldest !== void 0) {
|
|
3702
|
+
cache.delete(oldest);
|
|
3703
|
+
}
|
|
3704
|
+
}
|
|
3705
|
+
cache.set(key, value);
|
|
3706
|
+
}
|
|
3707
|
+
function tokenizeTextForRanking(text) {
|
|
3708
|
+
if (!text) {
|
|
3709
|
+
return /* @__PURE__ */ new Set();
|
|
3710
|
+
}
|
|
3711
|
+
const lowered = text.toLowerCase();
|
|
3712
|
+
const cache = rankingQueryTokenCache.get(lowered) ?? rankingTextTokenCache.get(lowered);
|
|
3713
|
+
if (cache) {
|
|
3714
|
+
return cache;
|
|
3715
|
+
}
|
|
3716
|
+
const tokens = new Set(
|
|
3717
|
+
lowered.replace(/[^\w\s]/g, " ").split(/\s+/).filter((token) => token.length > 1 && !STOPWORDS.has(token))
|
|
3718
|
+
);
|
|
3719
|
+
setBoundedCache(rankingQueryTokenCache, lowered, tokens);
|
|
3720
|
+
setBoundedCache(rankingTextTokenCache, lowered, tokens);
|
|
3721
|
+
return tokens;
|
|
3722
|
+
}
|
|
3723
|
+
function splitPathTokens(filePath) {
|
|
3724
|
+
const lowered = filePath.toLowerCase();
|
|
3725
|
+
const cache = rankingPathTokenCache.get(lowered);
|
|
3726
|
+
if (cache) {
|
|
3727
|
+
return cache;
|
|
3728
|
+
}
|
|
3729
|
+
const normalized = lowered.replace(/[^a-z0-9/._-]/g, " ").split(/[/._-]+/).filter((token) => token.length > 1);
|
|
3730
|
+
const tokens = new Set(normalized);
|
|
3731
|
+
setBoundedCache(rankingPathTokenCache, lowered, tokens);
|
|
3732
|
+
return tokens;
|
|
3733
|
+
}
|
|
3734
|
+
function splitNameTokens(name) {
|
|
3735
|
+
if (!name) {
|
|
3736
|
+
return /* @__PURE__ */ new Set();
|
|
3737
|
+
}
|
|
3738
|
+
const lowered = name.toLowerCase();
|
|
3739
|
+
const cache = rankingNameTokenCache.get(lowered);
|
|
3740
|
+
if (cache) {
|
|
3741
|
+
return cache;
|
|
3742
|
+
}
|
|
3743
|
+
const tokens = new Set(
|
|
3744
|
+
lowered.replace(/[^\w\s]/g, " ").split(/\s+/).filter((token) => token.length > 1)
|
|
3745
|
+
);
|
|
3746
|
+
setBoundedCache(rankingNameTokenCache, lowered, tokens);
|
|
3747
|
+
return tokens;
|
|
3748
|
+
}
|
|
3749
|
+
function chunkTypeBoost(chunkType) {
|
|
3750
|
+
switch (chunkType) {
|
|
3751
|
+
case "function":
|
|
3752
|
+
case "function_declaration":
|
|
3753
|
+
case "method":
|
|
3754
|
+
case "method_definition":
|
|
3755
|
+
case "class":
|
|
3756
|
+
case "class_declaration":
|
|
3757
|
+
return 0.2;
|
|
3758
|
+
case "interface":
|
|
3759
|
+
case "type":
|
|
3760
|
+
case "enum":
|
|
3761
|
+
case "struct":
|
|
3762
|
+
case "impl":
|
|
3763
|
+
case "trait":
|
|
3764
|
+
case "module":
|
|
3765
|
+
return 0.1;
|
|
3766
|
+
default:
|
|
3767
|
+
return 0;
|
|
3768
|
+
}
|
|
3769
|
+
}
|
|
3770
|
+
function isTestOrDocPath(filePath) {
|
|
3771
|
+
return TEST_PATH_SEGMENTS.some((segment) => filePath.includes(segment));
|
|
3772
|
+
}
|
|
3773
|
+
function isLikelyImplementationPath(filePath) {
|
|
3774
|
+
const lowered = filePath.toLowerCase();
|
|
3775
|
+
if (IMPLEMENTATION_EXCLUDE_PATH_SEGMENTS.some((segment) => lowered.includes(segment))) {
|
|
3776
|
+
return false;
|
|
3777
|
+
}
|
|
3778
|
+
const ext = lowered.split(".").pop() ?? "";
|
|
3779
|
+
if (["md", "mdx", "txt", "rst", "adoc", "snap", "json", "yaml", "yml", "lock"].includes(ext)) {
|
|
3780
|
+
return false;
|
|
3781
|
+
}
|
|
3782
|
+
return true;
|
|
3783
|
+
}
|
|
3784
|
+
function classifyQueryIntent(tokens) {
|
|
3785
|
+
const sourceIntentHits = tokens.filter((t) => SOURCE_INTENT_HINTS.has(t)).length;
|
|
3786
|
+
const docTestIntentHits = tokens.filter((t) => DOC_TEST_INTENT_HINTS.has(t)).length;
|
|
3787
|
+
return sourceIntentHits >= docTestIntentHits ? "source" : "doc_test";
|
|
3788
|
+
}
|
|
3789
|
+
function classifyQueryIntentRaw(query) {
|
|
3790
|
+
const lowerQuery = query.toLowerCase();
|
|
3791
|
+
const docTestRawHits = Array.from(DOC_TEST_INTENT_HINTS).filter(
|
|
3792
|
+
(hint) => new RegExp(`\\b${hint}\\b`).test(lowerQuery)
|
|
3793
|
+
).length;
|
|
3794
|
+
const sourceRawHits = [
|
|
3795
|
+
"implement",
|
|
3796
|
+
"implementation",
|
|
3797
|
+
"implements",
|
|
3798
|
+
"function",
|
|
3799
|
+
"method",
|
|
3800
|
+
"class",
|
|
3801
|
+
"logic",
|
|
3802
|
+
"algorithm",
|
|
3803
|
+
"pipeline",
|
|
3804
|
+
"indexer"
|
|
3805
|
+
].filter((hint) => new RegExp(`\\b${hint}\\b`).test(lowerQuery)).length;
|
|
3806
|
+
if (docTestRawHits > sourceRawHits) {
|
|
3807
|
+
return "doc_test";
|
|
3808
|
+
}
|
|
3809
|
+
if (sourceRawHits > docTestRawHits) {
|
|
3810
|
+
return "source";
|
|
3811
|
+
}
|
|
3812
|
+
const hasWhereIsPattern = /\bwhere\s+is\b/.test(lowerQuery);
|
|
3813
|
+
const hasIdentifierHints = extractIdentifierHints(query).length > 0;
|
|
3814
|
+
if (hasWhereIsPattern && hasIdentifierHints && docTestRawHits === 0) {
|
|
3815
|
+
return "source";
|
|
3816
|
+
}
|
|
3817
|
+
const queryTokens = Array.from(tokenizeTextForRanking(query));
|
|
3818
|
+
return classifyQueryIntent(queryTokens);
|
|
3819
|
+
}
|
|
3820
|
+
function classifyDocIntent(tokens) {
|
|
3821
|
+
const docHits = tokens.filter((t) => DOC_INTENT_HINTS.has(t)).length;
|
|
3822
|
+
const testHits = tokens.filter((t) => ["test", "tests", "fixture", "fixtures", "benchmark"].includes(t)).length;
|
|
3823
|
+
if (docHits > 0 && testHits === 0) return "docs";
|
|
3824
|
+
if (testHits > 0 && docHits === 0) return "test";
|
|
3825
|
+
if (testHits > 0 || docHits > 0) return "mixed";
|
|
3826
|
+
return "none";
|
|
3827
|
+
}
|
|
3828
|
+
function isImplementationChunkType(chunkType) {
|
|
3829
|
+
return [
|
|
3830
|
+
"export_statement",
|
|
3831
|
+
"function",
|
|
3832
|
+
"function_declaration",
|
|
3833
|
+
"method",
|
|
3834
|
+
"method_definition",
|
|
3835
|
+
"class",
|
|
3836
|
+
"class_declaration",
|
|
3837
|
+
"interface",
|
|
3838
|
+
"type",
|
|
3839
|
+
"enum",
|
|
3840
|
+
"module"
|
|
3841
|
+
].includes(chunkType);
|
|
3842
|
+
}
|
|
3843
|
+
function extractIdentifierHints(query) {
|
|
3844
|
+
const identifiers = query.match(/[A-Za-z_][A-Za-z0-9_]*/g) ?? [];
|
|
3845
|
+
return identifiers.filter((id) => id.length >= 3).filter((id) => {
|
|
3846
|
+
const lower = id.toLowerCase();
|
|
3847
|
+
if (STOPWORDS.has(lower)) return false;
|
|
3848
|
+
return /[A-Z]/.test(id) || id.includes("_") || id.endsWith("Results") || id.endsWith("Result");
|
|
3849
|
+
}).map((id) => id.toLowerCase());
|
|
3850
|
+
}
|
|
3851
|
+
function extractCodeTermHints(query) {
|
|
3852
|
+
const terms = query.match(/[A-Za-z_][A-Za-z0-9_]*/g) ?? [];
|
|
3853
|
+
return terms.map((term) => term.toLowerCase()).filter((term) => term.length >= 3).filter((term) => !STOPWORDS.has(term));
|
|
3854
|
+
}
|
|
3855
|
+
function normalizeIdentifierVariants(identifier) {
|
|
3856
|
+
const lower = identifier.toLowerCase();
|
|
3857
|
+
const compact = lower.replace(/[^a-z0-9]/g, "");
|
|
3858
|
+
const snake = compact.replace(/([a-z])([A-Z])/g, "$1_$2").toLowerCase();
|
|
3859
|
+
const kebab = snake.replace(/_/g, "-");
|
|
3860
|
+
const variants = [lower, compact, snake, kebab].filter((value) => value.length > 0);
|
|
3861
|
+
return Array.from(new Set(variants));
|
|
3862
|
+
}
|
|
3863
|
+
function scoreIdentifierMatch(name, filePath, hints) {
|
|
3864
|
+
const nameLower = (name ?? "").toLowerCase();
|
|
3865
|
+
const pathLower = filePath.toLowerCase();
|
|
3866
|
+
let best = 0;
|
|
3867
|
+
for (const hint of hints) {
|
|
3868
|
+
const variants = normalizeIdentifierVariants(hint);
|
|
3869
|
+
for (const variant of variants) {
|
|
3870
|
+
if (nameLower === variant) {
|
|
3871
|
+
best = Math.max(best, 1);
|
|
3872
|
+
} else if (nameLower.includes(variant)) {
|
|
3873
|
+
best = Math.max(best, 0.8);
|
|
3874
|
+
} else if (pathLower.includes(variant)) {
|
|
3875
|
+
best = Math.max(best, 0.6);
|
|
3876
|
+
}
|
|
3877
|
+
}
|
|
3878
|
+
}
|
|
3879
|
+
return best;
|
|
3880
|
+
}
|
|
3881
|
+
function extractPrimaryIdentifierQueryHint(query) {
|
|
3882
|
+
const identifiers = extractIdentifierHints(query);
|
|
3883
|
+
if (identifiers.length > 0) {
|
|
3884
|
+
return identifiers[0] ?? null;
|
|
3885
|
+
}
|
|
3886
|
+
const codeTerms = extractCodeTermHints(query);
|
|
3887
|
+
const best = codeTerms.find((term) => term.length >= 6);
|
|
3888
|
+
return best ?? null;
|
|
3889
|
+
}
|
|
3890
|
+
var FILE_PATH_HINT_EXTENSIONS = [
|
|
3891
|
+
"ts",
|
|
3892
|
+
"tsx",
|
|
3893
|
+
"js",
|
|
3894
|
+
"jsx",
|
|
3895
|
+
"mjs",
|
|
3896
|
+
"cjs",
|
|
3897
|
+
"mts",
|
|
3898
|
+
"cts",
|
|
3899
|
+
"py",
|
|
3900
|
+
"rs",
|
|
3901
|
+
"go",
|
|
3902
|
+
"java",
|
|
3903
|
+
"kt",
|
|
3904
|
+
"kts",
|
|
3905
|
+
"swift",
|
|
3906
|
+
"rb",
|
|
3907
|
+
"php",
|
|
3908
|
+
"c",
|
|
3909
|
+
"h",
|
|
3910
|
+
"cc",
|
|
3911
|
+
"cpp",
|
|
3912
|
+
"cxx",
|
|
3913
|
+
"hpp",
|
|
3914
|
+
"cs",
|
|
3915
|
+
"scala",
|
|
3916
|
+
"lua",
|
|
3917
|
+
"sh",
|
|
3918
|
+
"bash",
|
|
3919
|
+
"zsh",
|
|
3920
|
+
"json",
|
|
3921
|
+
"yaml",
|
|
3922
|
+
"yml",
|
|
3923
|
+
"toml"
|
|
3924
|
+
];
|
|
3925
|
+
var FILE_PATH_HINT_SUFFIX_REGEX = new RegExp(
|
|
3926
|
+
"\\s+\\bin\\s+[\"'`]?((?:\\.\\/)?(?:[A-Za-z0-9._-]+\\/)+[A-Za-z0-9._-]+\\.(?:" + FILE_PATH_HINT_EXTENSIONS.join("|") + "))[\"'`]?[\\])}>.,;!?]*\\s*$",
|
|
3927
|
+
"i"
|
|
3928
|
+
);
|
|
3929
|
+
function normalizeFilePathForHintMatch(filePath) {
|
|
3930
|
+
return filePath.replace(/\\/g, "/").toLowerCase().replace(/^\.\//, "");
|
|
3931
|
+
}
|
|
3932
|
+
function pathMatchesHint(filePath, hint) {
|
|
3933
|
+
const normalizedPath = normalizeFilePathForHintMatch(filePath);
|
|
3934
|
+
const normalizedHint = normalizeFilePathForHintMatch(hint);
|
|
3935
|
+
return normalizedPath.endsWith(normalizedHint) || normalizedPath.includes(`/${normalizedHint}`) || normalizedPath.includes(normalizedHint);
|
|
3936
|
+
}
|
|
3937
|
+
function extractFilePathHint(query) {
|
|
3938
|
+
const match = query.match(FILE_PATH_HINT_SUFFIX_REGEX);
|
|
3939
|
+
const rawPath = match?.[1];
|
|
3940
|
+
if (!rawPath) {
|
|
3941
|
+
return null;
|
|
3942
|
+
}
|
|
3943
|
+
return rawPath.replace(/^\.\//, "");
|
|
3944
|
+
}
|
|
3945
|
+
function stripFilePathHint(query) {
|
|
3946
|
+
const stripped = query.replace(FILE_PATH_HINT_SUFFIX_REGEX, "").trim();
|
|
3947
|
+
return stripped.length > 0 ? stripped : query;
|
|
3948
|
+
}
|
|
3949
|
+
function buildDeterministicIdentifierPass(query, candidates, limit) {
|
|
3950
|
+
if (classifyQueryIntentRaw(query) !== "source") {
|
|
3951
|
+
return [];
|
|
3952
|
+
}
|
|
3953
|
+
const primary = extractPrimaryIdentifierQueryHint(query);
|
|
3954
|
+
if (!primary) {
|
|
3955
|
+
return [];
|
|
3956
|
+
}
|
|
3957
|
+
const filePathHint = extractFilePathHint(query);
|
|
3958
|
+
const primaryVariants = normalizeIdentifierVariants(primary);
|
|
3959
|
+
const hints = [primary, ...extractIdentifierHints(query), ...extractCodeTermHints(query)].map((value) => value.toLowerCase()).filter((value, idx, arr) => value.length >= 3 && arr.indexOf(value) === idx).slice(0, 8);
|
|
3960
|
+
const deterministic = candidates.filter(
|
|
3961
|
+
(candidate) => isLikelyImplementationPath(candidate.metadata.filePath) && isImplementationChunkType(candidate.metadata.chunkType)
|
|
3962
|
+
).map((candidate) => {
|
|
3963
|
+
const nameLower = (candidate.metadata.name ?? "").toLowerCase();
|
|
3964
|
+
const pathLower = candidate.metadata.filePath.toLowerCase();
|
|
3965
|
+
let maxMatch = 0;
|
|
3966
|
+
const nameMatchesPrimary = primaryVariants.some(
|
|
3967
|
+
(variant) => nameLower === variant || nameLower.replace(/[^a-z0-9]/g, "") === variant.replace(/[^a-z0-9]/g, "")
|
|
3968
|
+
);
|
|
3969
|
+
const pathMatchesFileHint = filePathHint ? pathMatchesHint(candidate.metadata.filePath, filePathHint) : false;
|
|
3970
|
+
for (const hint of hints) {
|
|
3971
|
+
const variants = normalizeIdentifierVariants(hint);
|
|
3972
|
+
for (const variant of variants) {
|
|
3973
|
+
if (nameLower === variant) {
|
|
3974
|
+
maxMatch = Math.max(maxMatch, 1);
|
|
3975
|
+
} else if (nameLower.includes(variant)) {
|
|
3976
|
+
maxMatch = Math.max(maxMatch, 0.85);
|
|
3977
|
+
} else if (pathLower.includes(variant)) {
|
|
3978
|
+
maxMatch = Math.max(maxMatch, 0.7);
|
|
3979
|
+
}
|
|
3980
|
+
}
|
|
3981
|
+
}
|
|
3982
|
+
if (pathMatchesFileHint && nameMatchesPrimary) {
|
|
3983
|
+
maxMatch = Math.max(maxMatch, 1);
|
|
3984
|
+
}
|
|
3985
|
+
return {
|
|
3986
|
+
candidate,
|
|
3987
|
+
maxMatch,
|
|
3988
|
+
pathMatchesFileHint,
|
|
3989
|
+
nameMatchesPrimary
|
|
3990
|
+
};
|
|
3991
|
+
}).filter((entry) => entry.maxMatch >= 0.7).sort((a, b) => {
|
|
3992
|
+
const aAnchored = a.pathMatchesFileHint && a.nameMatchesPrimary ? 1 : 0;
|
|
3993
|
+
const bAnchored = b.pathMatchesFileHint && b.nameMatchesPrimary ? 1 : 0;
|
|
3994
|
+
if (aAnchored !== bAnchored) return bAnchored - aAnchored;
|
|
3995
|
+
if (b.maxMatch !== a.maxMatch) return b.maxMatch - a.maxMatch;
|
|
3996
|
+
if (b.candidate.score !== a.candidate.score) return b.candidate.score - a.candidate.score;
|
|
3997
|
+
return a.candidate.id.localeCompare(b.candidate.id);
|
|
3998
|
+
}).slice(0, Math.max(limit * 2, 12));
|
|
3999
|
+
return deterministic.map((entry) => ({
|
|
4000
|
+
id: entry.candidate.id,
|
|
4001
|
+
score: entry.pathMatchesFileHint && entry.nameMatchesPrimary ? 0.995 : Math.min(1, 0.9 + entry.maxMatch * 0.09),
|
|
4002
|
+
metadata: entry.candidate.metadata
|
|
4003
|
+
}));
|
|
4004
|
+
}
|
|
4005
|
+
function fuseResultsWeighted(semanticResults, keywordResults, keywordWeight, limit) {
|
|
4006
|
+
const semanticWeight = 1 - keywordWeight;
|
|
4007
|
+
const fusedScores = /* @__PURE__ */ new Map();
|
|
4008
|
+
for (const r of semanticResults) {
|
|
4009
|
+
fusedScores.set(r.id, {
|
|
4010
|
+
score: r.score * semanticWeight,
|
|
4011
|
+
metadata: r.metadata
|
|
4012
|
+
});
|
|
4013
|
+
}
|
|
4014
|
+
for (const r of keywordResults) {
|
|
4015
|
+
const existing = fusedScores.get(r.id);
|
|
4016
|
+
if (existing) {
|
|
4017
|
+
existing.score += r.score * keywordWeight;
|
|
4018
|
+
} else {
|
|
4019
|
+
fusedScores.set(r.id, {
|
|
4020
|
+
score: r.score * keywordWeight,
|
|
4021
|
+
metadata: r.metadata
|
|
4022
|
+
});
|
|
4023
|
+
}
|
|
4024
|
+
}
|
|
4025
|
+
const results = Array.from(fusedScores.entries()).map(([id, data]) => ({
|
|
4026
|
+
id,
|
|
4027
|
+
score: data.score,
|
|
4028
|
+
metadata: data.metadata
|
|
4029
|
+
}));
|
|
4030
|
+
results.sort((a, b) => b.score - a.score || a.id.localeCompare(b.id));
|
|
4031
|
+
return results.slice(0, limit);
|
|
4032
|
+
}
|
|
4033
|
+
function fuseResultsRrf(semanticResults, keywordResults, rrfK, limit) {
|
|
4034
|
+
const maxPossibleRaw = 2 / (rrfK + 1);
|
|
4035
|
+
const rankByIdSemantic = /* @__PURE__ */ new Map();
|
|
4036
|
+
const rankByIdKeyword = /* @__PURE__ */ new Map();
|
|
4037
|
+
const metadataById = /* @__PURE__ */ new Map();
|
|
4038
|
+
semanticResults.forEach((result, index) => {
|
|
4039
|
+
rankByIdSemantic.set(result.id, index + 1);
|
|
4040
|
+
metadataById.set(result.id, result.metadata);
|
|
4041
|
+
});
|
|
4042
|
+
keywordResults.forEach((result, index) => {
|
|
4043
|
+
rankByIdKeyword.set(result.id, index + 1);
|
|
4044
|
+
if (!metadataById.has(result.id)) {
|
|
4045
|
+
metadataById.set(result.id, result.metadata);
|
|
4046
|
+
}
|
|
4047
|
+
});
|
|
4048
|
+
const allIds = /* @__PURE__ */ new Set([...rankByIdSemantic.keys(), ...rankByIdKeyword.keys()]);
|
|
4049
|
+
const fused = [];
|
|
4050
|
+
for (const id of allIds) {
|
|
4051
|
+
const semanticRank = rankByIdSemantic.get(id);
|
|
4052
|
+
const keywordRank = rankByIdKeyword.get(id);
|
|
4053
|
+
const semanticScore = semanticRank ? 1 / (rrfK + semanticRank) : 0;
|
|
4054
|
+
const keywordScore = keywordRank ? 1 / (rrfK + keywordRank) : 0;
|
|
4055
|
+
const metadata = metadataById.get(id);
|
|
4056
|
+
if (!metadata) continue;
|
|
4057
|
+
fused.push({
|
|
4058
|
+
id,
|
|
4059
|
+
score: maxPossibleRaw > 0 ? (semanticScore + keywordScore) / maxPossibleRaw : 0,
|
|
4060
|
+
metadata
|
|
4061
|
+
});
|
|
4062
|
+
}
|
|
4063
|
+
fused.sort((a, b) => b.score - a.score || a.id.localeCompare(b.id));
|
|
4064
|
+
return fused.slice(0, limit);
|
|
4065
|
+
}
|
|
4066
|
+
function rerankResults(query, candidates, rerankTopN, options) {
|
|
4067
|
+
if (rerankTopN <= 0 || candidates.length <= 1) {
|
|
4068
|
+
return candidates;
|
|
4069
|
+
}
|
|
4070
|
+
const topN = Math.min(rerankTopN, candidates.length);
|
|
4071
|
+
const queryTokens = tokenizeTextForRanking(query);
|
|
4072
|
+
if (queryTokens.size === 0) {
|
|
4073
|
+
return candidates;
|
|
4074
|
+
}
|
|
4075
|
+
const queryTokenList = Array.from(queryTokens);
|
|
4076
|
+
const intent = classifyQueryIntentRaw(query);
|
|
4077
|
+
const docIntent = classifyDocIntent(queryTokenList);
|
|
4078
|
+
const preferSourcePaths = options?.prioritizeSourcePaths ?? intent === "source";
|
|
4079
|
+
const identifierHints = extractIdentifierHints(query);
|
|
4080
|
+
const head = candidates.slice(0, topN).map((candidate, idx) => {
|
|
4081
|
+
const pathTokens = splitPathTokens(candidate.metadata.filePath);
|
|
4082
|
+
const nameTokens = splitNameTokens(candidate.metadata.name ?? "");
|
|
4083
|
+
const chunkTypeTokens = tokenizeTextForRanking(candidate.metadata.chunkType);
|
|
4084
|
+
let exactOrPrefixNameHits = 0;
|
|
4085
|
+
let pathOverlap = 0;
|
|
4086
|
+
let chunkTypeHits = 0;
|
|
4087
|
+
for (const token of queryTokenList) {
|
|
4088
|
+
if (nameTokens.has(token)) {
|
|
4089
|
+
exactOrPrefixNameHits += 1;
|
|
4090
|
+
} else {
|
|
4091
|
+
for (const nameToken of nameTokens) {
|
|
4092
|
+
if (nameToken.startsWith(token) || token.startsWith(nameToken)) {
|
|
4093
|
+
exactOrPrefixNameHits += 1;
|
|
4094
|
+
break;
|
|
4095
|
+
}
|
|
4096
|
+
}
|
|
4097
|
+
}
|
|
4098
|
+
if (pathTokens.has(token)) {
|
|
4099
|
+
pathOverlap += 1;
|
|
4100
|
+
}
|
|
4101
|
+
if (chunkTypeTokens.has(token)) {
|
|
4102
|
+
chunkTypeHits += 1;
|
|
4103
|
+
}
|
|
4104
|
+
}
|
|
4105
|
+
const likelyTestOrDoc = isTestOrDocPath(candidate.metadata.filePath);
|
|
4106
|
+
const lowerPath = candidate.metadata.filePath.toLowerCase();
|
|
4107
|
+
const lowerName = (candidate.metadata.name ?? "").toLowerCase();
|
|
4108
|
+
const hasIdentifierMatch = identifierHints.some((id) => lowerPath.includes(id) || lowerName.includes(id));
|
|
4109
|
+
const implementationPathBoost = preferSourcePaths && isLikelyImplementationPath(candidate.metadata.filePath) ? 0.08 : 0;
|
|
4110
|
+
const isReadmePath = candidate.metadata.filePath.toLowerCase().includes("readme");
|
|
4111
|
+
const testDocPenalty = preferSourcePaths && likelyTestOrDoc ? 0.12 : 0;
|
|
4112
|
+
const readmeDocBoost = !preferSourcePaths && isReadmePath ? 0.08 : 0;
|
|
4113
|
+
const identifierBoost = hasIdentifierMatch ? 0.12 : 0;
|
|
4114
|
+
const tokenCoverage = queryTokenList.length > 0 ? (exactOrPrefixNameHits + pathOverlap + chunkTypeHits) / queryTokenList.length : 0;
|
|
4115
|
+
const coverageBoost = Math.min(0.12, tokenCoverage * 0.06);
|
|
4116
|
+
const deterministicBoost = exactOrPrefixNameHits * 0.08 + pathOverlap * 0.03 + chunkTypeHits * 0.02 + coverageBoost + identifierBoost + implementationPathBoost - testDocPenalty + readmeDocBoost + chunkTypeBoost(candidate.metadata.chunkType);
|
|
4117
|
+
return {
|
|
4118
|
+
candidate,
|
|
4119
|
+
boostedScore: candidate.score + deterministicBoost,
|
|
4120
|
+
originalIndex: idx,
|
|
4121
|
+
hasIdentifierMatch,
|
|
4122
|
+
implementationChunk: isImplementationChunkType(candidate.metadata.chunkType),
|
|
4123
|
+
isLikelyImplementationPath: isLikelyImplementationPath(candidate.metadata.filePath),
|
|
4124
|
+
isTestOrDocPath: likelyTestOrDoc,
|
|
4125
|
+
isReadmePath
|
|
4126
|
+
};
|
|
4127
|
+
});
|
|
4128
|
+
head.sort((a, b) => {
|
|
4129
|
+
if (b.boostedScore !== a.boostedScore) return b.boostedScore - a.boostedScore;
|
|
4130
|
+
if (b.candidate.score !== a.candidate.score) return b.candidate.score - a.candidate.score;
|
|
4131
|
+
if (a.originalIndex !== b.originalIndex) return a.originalIndex - b.originalIndex;
|
|
4132
|
+
return a.candidate.id.localeCompare(b.candidate.id);
|
|
4133
|
+
});
|
|
4134
|
+
if (preferSourcePaths) {
|
|
4135
|
+
head.sort((a, b) => {
|
|
4136
|
+
const aId = a.hasIdentifierMatch ? 1 : 0;
|
|
4137
|
+
const bId = b.hasIdentifierMatch ? 1 : 0;
|
|
4138
|
+
if (aId !== bId) return bId - aId;
|
|
4139
|
+
const aImpl = a.implementationChunk ? 1 : 0;
|
|
4140
|
+
const bImpl = b.implementationChunk ? 1 : 0;
|
|
4141
|
+
if (aImpl !== bImpl) return bImpl - aImpl;
|
|
4142
|
+
const aImplementationPath = a.isLikelyImplementationPath ? 1 : 0;
|
|
4143
|
+
const bImplementationPath = b.isLikelyImplementationPath ? 1 : 0;
|
|
4144
|
+
if (aImplementationPath !== bImplementationPath) return bImplementationPath - aImplementationPath;
|
|
4145
|
+
const aTestDoc = a.isTestOrDocPath ? 1 : 0;
|
|
4146
|
+
const bTestDoc = b.isTestOrDocPath ? 1 : 0;
|
|
4147
|
+
if (aTestDoc !== bTestDoc) return aTestDoc - bTestDoc;
|
|
4148
|
+
return 0;
|
|
4149
|
+
});
|
|
4150
|
+
} else if (docIntent === "docs") {
|
|
4151
|
+
head.sort((a, b) => {
|
|
4152
|
+
const aReadme = a.isReadmePath ? 1 : 0;
|
|
4153
|
+
const bReadme = b.isReadmePath ? 1 : 0;
|
|
4154
|
+
if (aReadme !== bReadme) return bReadme - aReadme;
|
|
4155
|
+
return 0;
|
|
4156
|
+
});
|
|
4157
|
+
}
|
|
4158
|
+
const tail = candidates.slice(topN);
|
|
4159
|
+
return [...head.map((entry) => entry.candidate), ...tail];
|
|
4160
|
+
}
|
|
4161
|
+
function rankHybridResults(query, semanticResults, keywordResults, options) {
|
|
4162
|
+
const overfetchLimit = Math.max(options.limit * 4, options.limit);
|
|
4163
|
+
const fused = options.fusionStrategy === "rrf" ? fuseResultsRrf(semanticResults, keywordResults, options.rrfK, overfetchLimit) : fuseResultsWeighted(semanticResults, keywordResults, options.hybridWeight, overfetchLimit);
|
|
4164
|
+
const rerankPoolLimit = Math.max(overfetchLimit, options.rerankTopN * 3, options.limit * 6);
|
|
4165
|
+
const rerankPool = fused.slice(0, rerankPoolLimit);
|
|
4166
|
+
const intent = classifyQueryIntentRaw(query);
|
|
4167
|
+
return rerankResults(query, rerankPool, options.rerankTopN, {
|
|
4168
|
+
prioritizeSourcePaths: intent === "source"
|
|
4169
|
+
});
|
|
4170
|
+
}
|
|
4171
|
+
function rankSemanticOnlyResults(query, semanticResults, options) {
|
|
4172
|
+
const overfetchLimit = Math.max(options.limit * 4, options.limit);
|
|
4173
|
+
const bounded = semanticResults.slice(0, overfetchLimit);
|
|
4174
|
+
return rerankResults(query, bounded, options.rerankTopN, {
|
|
4175
|
+
prioritizeSourcePaths: options.prioritizeSourcePaths ?? false
|
|
4176
|
+
});
|
|
4177
|
+
}
|
|
4178
|
+
function promoteIdentifierMatches(query, combined, semanticCandidates, keywordCandidates, database, branchChunkIds) {
|
|
4179
|
+
if (combined.length === 0) {
|
|
4180
|
+
return combined;
|
|
4181
|
+
}
|
|
4182
|
+
if (classifyQueryIntentRaw(query) !== "source") {
|
|
4183
|
+
return combined;
|
|
4184
|
+
}
|
|
4185
|
+
const identifierHints = extractIdentifierHints(query);
|
|
4186
|
+
if (identifierHints.length === 0) {
|
|
4187
|
+
return combined;
|
|
4188
|
+
}
|
|
4189
|
+
const combinedById = new Map(combined.map((candidate) => [candidate.id, candidate]));
|
|
4190
|
+
const candidateUnion = /* @__PURE__ */ new Map();
|
|
4191
|
+
for (const candidate of semanticCandidates) {
|
|
4192
|
+
candidateUnion.set(candidate.id, candidate);
|
|
4193
|
+
}
|
|
4194
|
+
for (const candidate of keywordCandidates) {
|
|
4195
|
+
if (!candidateUnion.has(candidate.id)) {
|
|
4196
|
+
candidateUnion.set(candidate.id, candidate);
|
|
4197
|
+
}
|
|
4198
|
+
}
|
|
4199
|
+
if (database) {
|
|
4200
|
+
for (const identifier of identifierHints) {
|
|
4201
|
+
const symbols = database.getSymbolsByName(identifier);
|
|
4202
|
+
for (const symbol of symbols) {
|
|
4203
|
+
const chunks = database.getChunksByFile(symbol.filePath);
|
|
4204
|
+
for (const chunk of chunks) {
|
|
4205
|
+
if (branchChunkIds && !branchChunkIds.has(chunk.chunkId)) {
|
|
4206
|
+
continue;
|
|
4207
|
+
}
|
|
4208
|
+
const chunkType = chunk.nodeType ?? "other";
|
|
4209
|
+
if (!isImplementationChunkType(chunkType)) {
|
|
4210
|
+
continue;
|
|
4211
|
+
}
|
|
4212
|
+
if (!isLikelyImplementationPath(chunk.filePath)) {
|
|
4213
|
+
continue;
|
|
4214
|
+
}
|
|
4215
|
+
if (chunk.startLine > symbol.startLine || chunk.endLine < symbol.endLine) {
|
|
4216
|
+
continue;
|
|
4217
|
+
}
|
|
4218
|
+
const existing = combinedById.get(chunk.chunkId) ?? candidateUnion.get(chunk.chunkId);
|
|
4219
|
+
const metadata = existing?.metadata ?? {
|
|
4220
|
+
filePath: chunk.filePath,
|
|
4221
|
+
startLine: chunk.startLine,
|
|
4222
|
+
endLine: chunk.endLine,
|
|
4223
|
+
chunkType,
|
|
4224
|
+
name: chunk.name ?? void 0,
|
|
4225
|
+
language: chunk.language,
|
|
4226
|
+
hash: chunk.contentHash
|
|
4227
|
+
};
|
|
4228
|
+
const baselineScore = existing?.score ?? 0.5;
|
|
4229
|
+
candidateUnion.set(chunk.chunkId, {
|
|
4230
|
+
id: chunk.chunkId,
|
|
4231
|
+
score: Math.min(1, baselineScore + 0.5),
|
|
4232
|
+
metadata
|
|
4233
|
+
});
|
|
4234
|
+
}
|
|
4235
|
+
}
|
|
4236
|
+
}
|
|
4237
|
+
}
|
|
4238
|
+
const promoted = [];
|
|
4239
|
+
for (const candidate of candidateUnion.values()) {
|
|
4240
|
+
const filePathLower = candidate.metadata.filePath.toLowerCase();
|
|
4241
|
+
const nameLower = (candidate.metadata.name ?? "").toLowerCase();
|
|
4242
|
+
const exactIdentifierMatch = identifierHints.some((hint) => nameLower === hint);
|
|
4243
|
+
const hasIdentifierMatch = exactIdentifierMatch || identifierHints.some(
|
|
4244
|
+
(hint) => nameLower.includes(hint) || filePathLower.includes(hint)
|
|
4245
|
+
);
|
|
4246
|
+
if (!hasIdentifierMatch) {
|
|
4247
|
+
continue;
|
|
4248
|
+
}
|
|
4249
|
+
if (!isImplementationChunkType(candidate.metadata.chunkType)) {
|
|
4250
|
+
continue;
|
|
4251
|
+
}
|
|
4252
|
+
if (!isLikelyImplementationPath(candidate.metadata.filePath)) {
|
|
4253
|
+
continue;
|
|
4254
|
+
}
|
|
4255
|
+
const existing = combinedById.get(candidate.id) ?? candidate;
|
|
4256
|
+
const rescueBoost = exactIdentifierMatch ? 0.45 : 0.25;
|
|
4257
|
+
const boostedScore = Math.min(1, Math.max(existing.score, candidate.score) + rescueBoost);
|
|
4258
|
+
promoted.push({
|
|
4259
|
+
id: existing.id,
|
|
4260
|
+
score: boostedScore,
|
|
4261
|
+
metadata: existing.metadata
|
|
4262
|
+
});
|
|
4263
|
+
}
|
|
4264
|
+
if (promoted.length === 0) {
|
|
4265
|
+
return combined;
|
|
4266
|
+
}
|
|
4267
|
+
promoted.sort((a, b) => b.score - a.score || a.id.localeCompare(b.id));
|
|
4268
|
+
const promotedIds = new Set(promoted.map((candidate) => candidate.id));
|
|
4269
|
+
const remainder = combined.filter((candidate) => !promotedIds.has(candidate.id));
|
|
4270
|
+
return [...promoted, ...remainder];
|
|
4271
|
+
}
|
|
4272
|
+
function buildSymbolDefinitionLane(query, database, branchChunkIds, limit, fallbackCandidates) {
|
|
4273
|
+
if (classifyQueryIntentRaw(query) !== "source") {
|
|
4274
|
+
return [];
|
|
4275
|
+
}
|
|
4276
|
+
const identifierHints = extractIdentifierHints(query);
|
|
4277
|
+
const codeTermHints = extractCodeTermHints(query);
|
|
4278
|
+
if (identifierHints.length === 0 && codeTermHints.length === 0) {
|
|
4279
|
+
return [];
|
|
4280
|
+
}
|
|
4281
|
+
const symbolCandidates = /* @__PURE__ */ new Map();
|
|
4282
|
+
const filePathHint = extractFilePathHint(query);
|
|
4283
|
+
const primaryHint = extractPrimaryIdentifierQueryHint(query);
|
|
4284
|
+
const upsertChunkCandidate = (chunk, identifier, normalizedIdentifier, baseScore) => {
|
|
4285
|
+
if (branchChunkIds && !branchChunkIds.has(chunk.chunkId)) {
|
|
4286
|
+
return;
|
|
4287
|
+
}
|
|
4288
|
+
const chunkType = chunk.nodeType ?? "other";
|
|
4289
|
+
if (!isImplementationChunkType(chunkType)) {
|
|
4290
|
+
return;
|
|
4291
|
+
}
|
|
4292
|
+
if (!isLikelyImplementationPath(chunk.filePath)) {
|
|
4293
|
+
return;
|
|
4294
|
+
}
|
|
4295
|
+
const nameLower = (chunk.name ?? "").toLowerCase();
|
|
4296
|
+
const exactName = nameLower === identifier || nameLower.replace(/_/g, "") === normalizedIdentifier;
|
|
4297
|
+
const base = baseScore ?? (exactName ? 0.99 : 0.88);
|
|
4298
|
+
const existing = symbolCandidates.get(chunk.chunkId);
|
|
4299
|
+
if (!existing || base > existing.score) {
|
|
4300
|
+
symbolCandidates.set(chunk.chunkId, {
|
|
4301
|
+
id: chunk.chunkId,
|
|
4302
|
+
score: base,
|
|
4303
|
+
metadata: {
|
|
4304
|
+
filePath: chunk.filePath,
|
|
4305
|
+
startLine: chunk.startLine,
|
|
4306
|
+
endLine: chunk.endLine,
|
|
4307
|
+
chunkType,
|
|
4308
|
+
name: chunk.name ?? void 0,
|
|
4309
|
+
language: chunk.language,
|
|
4310
|
+
hash: chunk.contentHash
|
|
4311
|
+
}
|
|
4312
|
+
});
|
|
4313
|
+
}
|
|
4314
|
+
};
|
|
4315
|
+
const normalizedHints = identifierHints.flatMap((hint) => [
|
|
4316
|
+
hint,
|
|
4317
|
+
hint.replace(/_/g, ""),
|
|
4318
|
+
hint.replace(/_/g, "-")
|
|
4319
|
+
]).filter((hint, idx, arr) => hint.length >= 3 && arr.indexOf(hint) === idx).slice(0, 6);
|
|
4320
|
+
for (const identifier of normalizedHints) {
|
|
4321
|
+
const symbols = [
|
|
4322
|
+
...database.getSymbolsByName(identifier),
|
|
4323
|
+
...database.getSymbolsByNameCi(identifier)
|
|
4324
|
+
];
|
|
4325
|
+
const chunksByName = [
|
|
4326
|
+
...database.getChunksByName(identifier),
|
|
4327
|
+
...database.getChunksByNameCi(identifier)
|
|
4328
|
+
];
|
|
4329
|
+
const normalizedIdentifier = identifier.replace(/_/g, "");
|
|
4330
|
+
const dedupSymbols = /* @__PURE__ */ new Map();
|
|
4331
|
+
for (const symbol of symbols) {
|
|
4332
|
+
dedupSymbols.set(symbol.id, symbol);
|
|
4333
|
+
}
|
|
4334
|
+
for (const symbol of dedupSymbols.values()) {
|
|
4335
|
+
const chunks = database.getChunksByFile(symbol.filePath);
|
|
4336
|
+
for (const chunk of chunks) {
|
|
4337
|
+
if (chunk.startLine > symbol.startLine || chunk.endLine < symbol.endLine) {
|
|
4338
|
+
continue;
|
|
4339
|
+
}
|
|
4340
|
+
upsertChunkCandidate(chunk, identifier, normalizedIdentifier);
|
|
4341
|
+
}
|
|
4342
|
+
}
|
|
4343
|
+
const dedupChunksByName = /* @__PURE__ */ new Map();
|
|
4344
|
+
for (const chunk of chunksByName) {
|
|
4345
|
+
dedupChunksByName.set(chunk.chunkId, chunk);
|
|
4346
|
+
}
|
|
4347
|
+
for (const chunk of dedupChunksByName.values()) {
|
|
4348
|
+
upsertChunkCandidate(chunk, identifier, normalizedIdentifier);
|
|
4349
|
+
}
|
|
4350
|
+
}
|
|
4351
|
+
if (filePathHint && primaryHint) {
|
|
4352
|
+
const primaryChunks = [
|
|
4353
|
+
...database.getChunksByName(primaryHint),
|
|
4354
|
+
...database.getChunksByNameCi(primaryHint)
|
|
4355
|
+
];
|
|
4356
|
+
const dedupPrimaryChunks = /* @__PURE__ */ new Map();
|
|
4357
|
+
for (const chunk of primaryChunks) {
|
|
4358
|
+
dedupPrimaryChunks.set(chunk.chunkId, chunk);
|
|
4359
|
+
}
|
|
4360
|
+
for (const chunk of dedupPrimaryChunks.values()) {
|
|
4361
|
+
if (!pathMatchesHint(chunk.filePath, filePathHint)) {
|
|
4362
|
+
continue;
|
|
4363
|
+
}
|
|
4364
|
+
const normalizedPrimary = primaryHint.replace(/_/g, "");
|
|
4365
|
+
upsertChunkCandidate(chunk, primaryHint, normalizedPrimary, 1);
|
|
4366
|
+
}
|
|
4367
|
+
}
|
|
4368
|
+
const ranked = Array.from(symbolCandidates.values()).sort((a, b) => b.score - a.score || a.id.localeCompare(b.id));
|
|
4369
|
+
if (ranked.length === 0) {
|
|
4370
|
+
const implementationFallback = fallbackCandidates.filter(
|
|
4371
|
+
(candidate) => isImplementationChunkType(candidate.metadata.chunkType) && isLikelyImplementationPath(candidate.metadata.filePath)
|
|
4372
|
+
);
|
|
4373
|
+
for (const candidate of implementationFallback) {
|
|
4374
|
+
const nameLower = (candidate.metadata.name ?? "").toLowerCase();
|
|
4375
|
+
const pathLower = candidate.metadata.filePath.toLowerCase();
|
|
4376
|
+
const exactHintMatch = normalizedHints.some((hint) => nameLower === hint || nameLower.replace(/_/g, "") === hint.replace(/_/g, ""));
|
|
4377
|
+
const tokenizedName = tokenizeTextForRanking(nameLower);
|
|
4378
|
+
const tokenHits = codeTermHints.filter((term) => tokenizedName.has(term) || pathLower.includes(term)).length;
|
|
4379
|
+
if (!exactHintMatch && tokenHits === 0) {
|
|
4380
|
+
continue;
|
|
4381
|
+
}
|
|
4382
|
+
const laneScore = exactHintMatch ? Math.min(1, Math.max(candidate.score, 0.97)) : Math.min(0.95, Math.max(candidate.score, 0.82 + tokenHits * 0.03));
|
|
4383
|
+
symbolCandidates.set(candidate.id, {
|
|
4384
|
+
id: candidate.id,
|
|
4385
|
+
score: laneScore,
|
|
4386
|
+
metadata: candidate.metadata
|
|
4387
|
+
});
|
|
4388
|
+
}
|
|
4389
|
+
if (symbolCandidates.size === 0) {
|
|
4390
|
+
const queryTokenSet = tokenizeTextForRanking(query);
|
|
4391
|
+
const rankedFallback = implementationFallback.map((candidate) => {
|
|
4392
|
+
const nameTokens = tokenizeTextForRanking(candidate.metadata.name ?? "");
|
|
4393
|
+
const pathTokens = splitPathTokens(candidate.metadata.filePath);
|
|
4394
|
+
let overlap = 0;
|
|
4395
|
+
for (const token of queryTokenSet) {
|
|
4396
|
+
if (nameTokens.has(token) || pathTokens.has(token)) {
|
|
4397
|
+
overlap += 1;
|
|
4398
|
+
}
|
|
4399
|
+
}
|
|
4400
|
+
const overlapScore = queryTokenSet.size > 0 ? overlap / queryTokenSet.size : 0;
|
|
4401
|
+
return {
|
|
4402
|
+
candidate,
|
|
4403
|
+
overlapScore
|
|
4404
|
+
};
|
|
4405
|
+
}).filter((entry) => entry.overlapScore > 0).sort((a, b) => b.overlapScore - a.overlapScore || b.candidate.score - a.candidate.score).slice(0, Math.max(limit, 3));
|
|
4406
|
+
for (const entry of rankedFallback) {
|
|
4407
|
+
symbolCandidates.set(entry.candidate.id, {
|
|
4408
|
+
id: entry.candidate.id,
|
|
4409
|
+
score: Math.min(0.94, Math.max(entry.candidate.score, 0.8 + entry.overlapScore * 0.1)),
|
|
4410
|
+
metadata: entry.candidate.metadata
|
|
4411
|
+
});
|
|
4412
|
+
}
|
|
4413
|
+
}
|
|
4414
|
+
}
|
|
4415
|
+
const withFallback = Array.from(symbolCandidates.values()).sort((a, b) => b.score - a.score || a.id.localeCompare(b.id));
|
|
4416
|
+
return withFallback.slice(0, Math.max(limit * 2, limit));
|
|
4417
|
+
}
|
|
4418
|
+
function buildIdentifierDefinitionLane(query, candidates, limit) {
|
|
4419
|
+
if (classifyQueryIntentRaw(query) !== "source") {
|
|
4420
|
+
return [];
|
|
4421
|
+
}
|
|
4422
|
+
const primaryHint = extractPrimaryIdentifierQueryHint(query);
|
|
4423
|
+
if (!primaryHint) {
|
|
4424
|
+
return [];
|
|
4425
|
+
}
|
|
4426
|
+
const hints = [primaryHint, ...extractIdentifierHints(query), ...extractCodeTermHints(query)].slice(0, 8);
|
|
4427
|
+
const scored = candidates.filter(
|
|
4428
|
+
(candidate) => isLikelyImplementationPath(candidate.metadata.filePath) && isImplementationChunkType(candidate.metadata.chunkType)
|
|
4429
|
+
).map((candidate) => {
|
|
4430
|
+
const matchScore = scoreIdentifierMatch(candidate.metadata.name, candidate.metadata.filePath, hints);
|
|
4431
|
+
return {
|
|
4432
|
+
candidate,
|
|
4433
|
+
matchScore
|
|
4434
|
+
};
|
|
4435
|
+
}).filter((entry) => entry.matchScore > 0).sort((a, b) => {
|
|
4436
|
+
if (b.matchScore !== a.matchScore) return b.matchScore - a.matchScore;
|
|
4437
|
+
if (b.candidate.score !== a.candidate.score) return b.candidate.score - a.candidate.score;
|
|
4438
|
+
return a.candidate.id.localeCompare(b.candidate.id);
|
|
4439
|
+
}).slice(0, Math.max(limit * 2, 10));
|
|
4440
|
+
return scored.map((entry) => ({
|
|
4441
|
+
id: entry.candidate.id,
|
|
4442
|
+
score: Math.min(1, 0.9 + entry.matchScore * 0.09),
|
|
4443
|
+
metadata: entry.candidate.metadata
|
|
4444
|
+
}));
|
|
4445
|
+
}
|
|
4446
|
+
function mergeTieredResults(symbolLane, hybridLane, limit) {
|
|
4447
|
+
if (symbolLane.length === 0) {
|
|
4448
|
+
return hybridLane.slice(0, limit);
|
|
4449
|
+
}
|
|
4450
|
+
const out = [];
|
|
4451
|
+
const seen = /* @__PURE__ */ new Set();
|
|
4452
|
+
for (const candidate of symbolLane) {
|
|
4453
|
+
if (seen.has(candidate.id)) continue;
|
|
4454
|
+
out.push(candidate);
|
|
4455
|
+
seen.add(candidate.id);
|
|
4456
|
+
if (out.length >= limit) return out;
|
|
4457
|
+
}
|
|
4458
|
+
for (const candidate of hybridLane) {
|
|
4459
|
+
if (seen.has(candidate.id)) continue;
|
|
4460
|
+
out.push(candidate);
|
|
4461
|
+
seen.add(candidate.id);
|
|
4462
|
+
if (out.length >= limit) return out;
|
|
4463
|
+
}
|
|
4464
|
+
return out;
|
|
4465
|
+
}
|
|
4466
|
+
function unionCandidates(semanticCandidates, keywordCandidates) {
|
|
4467
|
+
const byId = /* @__PURE__ */ new Map();
|
|
4468
|
+
for (const candidate of semanticCandidates) {
|
|
4469
|
+
byId.set(candidate.id, candidate);
|
|
4470
|
+
}
|
|
4471
|
+
for (const candidate of keywordCandidates) {
|
|
4472
|
+
const existing = byId.get(candidate.id);
|
|
4473
|
+
if (!existing || candidate.score > existing.score) {
|
|
4474
|
+
byId.set(candidate.id, candidate);
|
|
4475
|
+
}
|
|
4476
|
+
}
|
|
4477
|
+
return Array.from(byId.values());
|
|
4478
|
+
}
|
|
3485
4479
|
var Indexer = class {
|
|
3486
4480
|
config;
|
|
3487
4481
|
projectRoot;
|
|
@@ -3984,6 +4978,79 @@ var Indexer = class {
|
|
|
3984
4978
|
if (chunkDataBatch.length > 0) {
|
|
3985
4979
|
database.upsertChunksBatch(chunkDataBatch);
|
|
3986
4980
|
}
|
|
4981
|
+
const allSymbolIds = /* @__PURE__ */ new Set();
|
|
4982
|
+
const symbolsByFile = /* @__PURE__ */ new Map();
|
|
4983
|
+
for (let i = 0; i < parsedFiles.length; i++) {
|
|
4984
|
+
const parsed = parsedFiles[i];
|
|
4985
|
+
const changedFile = changedFiles[i];
|
|
4986
|
+
database.deleteCallEdgesByFile(parsed.path);
|
|
4987
|
+
database.deleteSymbolsByFile(parsed.path);
|
|
4988
|
+
const fileSymbols = [];
|
|
4989
|
+
for (const chunk of parsed.chunks) {
|
|
4990
|
+
if (!chunk.name || !CALL_GRAPH_SYMBOL_CHUNK_TYPES.has(chunk.chunkType)) continue;
|
|
4991
|
+
const symbolId = `sym_${hashContent(parsed.path + ":" + chunk.name + ":" + chunk.chunkType + ":" + chunk.startLine).slice(0, 16)}`;
|
|
4992
|
+
const symbol = {
|
|
4993
|
+
id: symbolId,
|
|
4994
|
+
filePath: parsed.path,
|
|
4995
|
+
name: chunk.name,
|
|
4996
|
+
kind: chunk.chunkType,
|
|
4997
|
+
startLine: chunk.startLine,
|
|
4998
|
+
startCol: 0,
|
|
4999
|
+
endLine: chunk.endLine,
|
|
5000
|
+
endCol: 0,
|
|
5001
|
+
language: chunk.language
|
|
5002
|
+
};
|
|
5003
|
+
fileSymbols.push(symbol);
|
|
5004
|
+
allSymbolIds.add(symbolId);
|
|
5005
|
+
}
|
|
5006
|
+
const symbolsByName = /* @__PURE__ */ new Map();
|
|
5007
|
+
for (const symbol of fileSymbols) {
|
|
5008
|
+
const existing = symbolsByName.get(symbol.name) ?? [];
|
|
5009
|
+
existing.push(symbol);
|
|
5010
|
+
symbolsByName.set(symbol.name, existing);
|
|
5011
|
+
}
|
|
5012
|
+
if (fileSymbols.length > 0) {
|
|
5013
|
+
database.upsertSymbolsBatch(fileSymbols);
|
|
5014
|
+
symbolsByFile.set(parsed.path, fileSymbols);
|
|
5015
|
+
}
|
|
5016
|
+
const fileLanguage = parsed.chunks[0]?.language;
|
|
5017
|
+
if (!fileLanguage || !CALL_GRAPH_LANGUAGES.has(fileLanguage)) continue;
|
|
5018
|
+
const callSites = extractCalls(changedFile.content, fileLanguage);
|
|
5019
|
+
if (callSites.length === 0) continue;
|
|
5020
|
+
const edges = [];
|
|
5021
|
+
for (const site of callSites) {
|
|
5022
|
+
const enclosingSymbol = fileSymbols.find(
|
|
5023
|
+
(sym) => site.line >= sym.startLine && site.line <= sym.endLine
|
|
5024
|
+
);
|
|
5025
|
+
if (!enclosingSymbol) continue;
|
|
5026
|
+
const edgeId = `edge_${hashContent(enclosingSymbol.id + ":" + site.calleeName + ":" + site.line + ":" + site.column).slice(0, 16)}`;
|
|
5027
|
+
edges.push({
|
|
5028
|
+
id: edgeId,
|
|
5029
|
+
fromSymbolId: enclosingSymbol.id,
|
|
5030
|
+
targetName: site.calleeName,
|
|
5031
|
+
toSymbolId: void 0,
|
|
5032
|
+
callType: site.callType,
|
|
5033
|
+
line: site.line,
|
|
5034
|
+
col: site.column,
|
|
5035
|
+
isResolved: false
|
|
5036
|
+
});
|
|
5037
|
+
}
|
|
5038
|
+
if (edges.length > 0) {
|
|
5039
|
+
database.upsertCallEdgesBatch(edges);
|
|
5040
|
+
for (const edge of edges) {
|
|
5041
|
+
const candidates = symbolsByName.get(edge.targetName);
|
|
5042
|
+
if (candidates && candidates.length === 1) {
|
|
5043
|
+
database.resolveCallEdge(edge.id, candidates[0].id);
|
|
5044
|
+
}
|
|
5045
|
+
}
|
|
5046
|
+
}
|
|
5047
|
+
}
|
|
5048
|
+
for (const filePath of unchangedFilePaths) {
|
|
5049
|
+
const existingSymbols = database.getSymbolsByFile(filePath);
|
|
5050
|
+
for (const sym of existingSymbols) {
|
|
5051
|
+
allSymbolIds.add(sym.id);
|
|
5052
|
+
}
|
|
5053
|
+
}
|
|
3987
5054
|
let removedCount = 0;
|
|
3988
5055
|
for (const [chunkId] of existingChunks) {
|
|
3989
5056
|
if (!currentChunkIds.has(chunkId)) {
|
|
@@ -4005,6 +5072,8 @@ var Indexer = class {
|
|
|
4005
5072
|
if (pendingChunks.length === 0 && removedCount === 0) {
|
|
4006
5073
|
database.clearBranch(this.currentBranch);
|
|
4007
5074
|
database.addChunksToBranchBatch(this.currentBranch, Array.from(currentChunkIds));
|
|
5075
|
+
database.clearBranchSymbols(this.currentBranch);
|
|
5076
|
+
database.addSymbolsToBranchBatch(this.currentBranch, Array.from(allSymbolIds));
|
|
4008
5077
|
this.fileHashCache = currentFileHashes;
|
|
4009
5078
|
this.saveFileHashCache();
|
|
4010
5079
|
stats.durationMs = Date.now() - startTime;
|
|
@@ -4021,6 +5090,8 @@ var Indexer = class {
|
|
|
4021
5090
|
if (pendingChunks.length === 0) {
|
|
4022
5091
|
database.clearBranch(this.currentBranch);
|
|
4023
5092
|
database.addChunksToBranchBatch(this.currentBranch, Array.from(currentChunkIds));
|
|
5093
|
+
database.clearBranchSymbols(this.currentBranch);
|
|
5094
|
+
database.addSymbolsToBranchBatch(this.currentBranch, Array.from(allSymbolIds));
|
|
4024
5095
|
store.save();
|
|
4025
5096
|
invertedIndex.save();
|
|
4026
5097
|
this.fileHashCache = currentFileHashes;
|
|
@@ -4161,6 +5232,8 @@ var Indexer = class {
|
|
|
4161
5232
|
});
|
|
4162
5233
|
database.clearBranch(this.currentBranch);
|
|
4163
5234
|
database.addChunksToBranchBatch(this.currentBranch, Array.from(currentChunkIds));
|
|
5235
|
+
database.clearBranchSymbols(this.currentBranch);
|
|
5236
|
+
database.addSymbolsToBranchBatch(this.currentBranch, Array.from(allSymbolIds));
|
|
4164
5237
|
store.save();
|
|
4165
5238
|
invertedIndex.save();
|
|
4166
5239
|
this.fileHashCache = currentFileHashes;
|
|
@@ -4271,15 +5344,22 @@ var Indexer = class {
|
|
|
4271
5344
|
}
|
|
4272
5345
|
const maxResults = limit ?? this.config.search.maxResults;
|
|
4273
5346
|
const hybridWeight = options?.hybridWeight ?? this.config.search.hybridWeight;
|
|
5347
|
+
const fusionStrategy = this.config.search.fusionStrategy;
|
|
5348
|
+
const rrfK = this.config.search.rrfK;
|
|
5349
|
+
const rerankTopN = this.config.search.rerankTopN;
|
|
4274
5350
|
const filterByBranch = options?.filterByBranch ?? true;
|
|
4275
5351
|
this.logger.search("debug", "Starting search", {
|
|
4276
5352
|
query,
|
|
4277
5353
|
maxResults,
|
|
4278
5354
|
hybridWeight,
|
|
5355
|
+
fusionStrategy,
|
|
5356
|
+
rrfK,
|
|
5357
|
+
rerankTopN,
|
|
4279
5358
|
filterByBranch
|
|
4280
5359
|
});
|
|
4281
5360
|
const embeddingStartTime = import_perf_hooks.performance.now();
|
|
4282
|
-
const
|
|
5361
|
+
const embeddingQuery = stripFilePathHint(query);
|
|
5362
|
+
const embedding = await this.getQueryEmbedding(embeddingQuery, provider);
|
|
4283
5363
|
const embeddingMs = import_perf_hooks.performance.now() - embeddingStartTime;
|
|
4284
5364
|
const vectorStartTime = import_perf_hooks.performance.now();
|
|
4285
5365
|
const semanticResults = store.search(embedding, maxResults * 4);
|
|
@@ -4287,16 +5367,74 @@ var Indexer = class {
|
|
|
4287
5367
|
const keywordStartTime = import_perf_hooks.performance.now();
|
|
4288
5368
|
const keywordResults = await this.keywordSearch(query, maxResults * 4);
|
|
4289
5369
|
const keywordMs = import_perf_hooks.performance.now() - keywordStartTime;
|
|
4290
|
-
const fusionStartTime = import_perf_hooks.performance.now();
|
|
4291
|
-
const combined = this.fuseResults(semanticResults, keywordResults, hybridWeight, maxResults * 4);
|
|
4292
|
-
const fusionMs = import_perf_hooks.performance.now() - fusionStartTime;
|
|
4293
5370
|
let branchChunkIds = null;
|
|
4294
5371
|
if (filterByBranch && this.currentBranch !== "default") {
|
|
4295
5372
|
branchChunkIds = new Set(database.getBranchChunkIds(this.currentBranch));
|
|
4296
5373
|
}
|
|
4297
|
-
const
|
|
5374
|
+
const prefilterStartTime = import_perf_hooks.performance.now();
|
|
5375
|
+
const shouldPrefilterByBranch = branchChunkIds !== null && branchChunkIds.size > 0;
|
|
5376
|
+
const prefilteredSemantic = shouldPrefilterByBranch && branchChunkIds ? semanticResults.filter((r) => branchChunkIds.has(r.id)) : semanticResults;
|
|
5377
|
+
const prefilteredKeyword = shouldPrefilterByBranch && branchChunkIds ? keywordResults.filter((r) => branchChunkIds.has(r.id)) : keywordResults;
|
|
5378
|
+
const semanticCandidates = shouldPrefilterByBranch && semanticResults.length > 0 && prefilteredSemantic.length === 0 ? semanticResults : prefilteredSemantic;
|
|
5379
|
+
const keywordCandidates = shouldPrefilterByBranch && keywordResults.length > 0 && prefilteredKeyword.length === 0 ? keywordResults : prefilteredKeyword;
|
|
5380
|
+
const prefilterMs = import_perf_hooks.performance.now() - prefilterStartTime;
|
|
5381
|
+
if (branchChunkIds && branchChunkIds.size === 0) {
|
|
5382
|
+
this.logger.search("warn", "Branch prefilter skipped because branch catalog is empty", {
|
|
5383
|
+
branch: this.currentBranch
|
|
5384
|
+
});
|
|
5385
|
+
}
|
|
5386
|
+
if (shouldPrefilterByBranch && semanticResults.length > 0 && prefilteredSemantic.length === 0) {
|
|
5387
|
+
this.logger.search("warn", "Branch prefilter produced no semantic overlap, using unfiltered semantic candidates", {
|
|
5388
|
+
branch: this.currentBranch
|
|
5389
|
+
});
|
|
5390
|
+
}
|
|
5391
|
+
if (shouldPrefilterByBranch && keywordResults.length > 0 && prefilteredKeyword.length === 0) {
|
|
5392
|
+
this.logger.search("warn", "Branch prefilter produced no keyword overlap, using unfiltered keyword candidates", {
|
|
5393
|
+
branch: this.currentBranch
|
|
5394
|
+
});
|
|
5395
|
+
}
|
|
5396
|
+
const fusionStartTime = import_perf_hooks.performance.now();
|
|
5397
|
+
const combined = rankHybridResults(query, semanticCandidates, keywordCandidates, {
|
|
5398
|
+
fusionStrategy,
|
|
5399
|
+
rrfK,
|
|
5400
|
+
rerankTopN,
|
|
5401
|
+
limit: maxResults,
|
|
5402
|
+
hybridWeight
|
|
5403
|
+
});
|
|
5404
|
+
const fusionMs = import_perf_hooks.performance.now() - fusionStartTime;
|
|
5405
|
+
const rescued = promoteIdentifierMatches(
|
|
5406
|
+
query,
|
|
5407
|
+
combined,
|
|
5408
|
+
semanticCandidates,
|
|
5409
|
+
keywordCandidates,
|
|
5410
|
+
database,
|
|
5411
|
+
branchChunkIds
|
|
5412
|
+
);
|
|
5413
|
+
const union = unionCandidates(semanticCandidates, keywordCandidates);
|
|
5414
|
+
const deterministicIdentifierLane = buildDeterministicIdentifierPass(
|
|
5415
|
+
query,
|
|
5416
|
+
union,
|
|
5417
|
+
maxResults
|
|
5418
|
+
);
|
|
5419
|
+
const identifierLane = buildIdentifierDefinitionLane(
|
|
5420
|
+
query,
|
|
5421
|
+
union,
|
|
5422
|
+
maxResults
|
|
5423
|
+
);
|
|
5424
|
+
const symbolLane = buildSymbolDefinitionLane(
|
|
5425
|
+
query,
|
|
5426
|
+
database,
|
|
5427
|
+
branchChunkIds,
|
|
5428
|
+
maxResults,
|
|
5429
|
+
union
|
|
5430
|
+
);
|
|
5431
|
+
const prePrimaryLane = mergeTieredResults(deterministicIdentifierLane, identifierLane, maxResults * 4);
|
|
5432
|
+
const primaryLane = mergeTieredResults(prePrimaryLane, symbolLane, maxResults * 4);
|
|
5433
|
+
const tiered = mergeTieredResults(primaryLane, rescued, maxResults * 4);
|
|
5434
|
+
const sourceIntent = classifyQueryIntentRaw(query) === "source";
|
|
5435
|
+
const hasCodeHints = extractCodeTermHints(query).length > 0 || extractIdentifierHints(query).length > 0;
|
|
5436
|
+
const baseFiltered = tiered.filter((r) => {
|
|
4298
5437
|
if (r.score < this.config.search.minScore) return false;
|
|
4299
|
-
if (branchChunkIds && !branchChunkIds.has(r.id)) return false;
|
|
4300
5438
|
if (options?.fileType) {
|
|
4301
5439
|
const ext = r.metadata.filePath.split(".").pop()?.toLowerCase();
|
|
4302
5440
|
if (ext !== options.fileType.toLowerCase().replace(/^\./, "")) return false;
|
|
@@ -4309,7 +5447,11 @@ var Indexer = class {
|
|
|
4309
5447
|
if (r.metadata.chunkType !== options.chunkType) return false;
|
|
4310
5448
|
}
|
|
4311
5449
|
return true;
|
|
4312
|
-
})
|
|
5450
|
+
});
|
|
5451
|
+
const implementationOnly = baseFiltered.filter(
|
|
5452
|
+
(r) => isLikelyImplementationPath(r.metadata.filePath) && isImplementationChunkType(r.metadata.chunkType)
|
|
5453
|
+
);
|
|
5454
|
+
const filtered = (sourceIntent && hasCodeHints && implementationOnly.length > 0 ? implementationOnly : baseFiltered).slice(0, maxResults);
|
|
4313
5455
|
const totalSearchMs = import_perf_hooks.performance.now() - searchStartTime;
|
|
4314
5456
|
this.logger.recordSearch(totalSearchMs, {
|
|
4315
5457
|
embeddingMs,
|
|
@@ -4324,6 +5466,7 @@ var Indexer = class {
|
|
|
4324
5466
|
embeddingMs: Math.round(embeddingMs * 100) / 100,
|
|
4325
5467
|
vectorMs: Math.round(vectorMs * 100) / 100,
|
|
4326
5468
|
keywordMs: Math.round(keywordMs * 100) / 100,
|
|
5469
|
+
prefilterMs: Math.round(prefilterMs * 100) / 100,
|
|
4327
5470
|
fusionMs: Math.round(fusionMs * 100) / 100
|
|
4328
5471
|
});
|
|
4329
5472
|
const metadataOnly = options?.metadataOnly ?? false;
|
|
@@ -4377,34 +5520,6 @@ var Indexer = class {
|
|
|
4377
5520
|
results.sort((a, b) => b.score - a.score);
|
|
4378
5521
|
return results.slice(0, limit);
|
|
4379
5522
|
}
|
|
4380
|
-
fuseResults(semanticResults, keywordResults, keywordWeight, limit) {
|
|
4381
|
-
const semanticWeight = 1 - keywordWeight;
|
|
4382
|
-
const fusedScores = /* @__PURE__ */ new Map();
|
|
4383
|
-
for (const r of semanticResults) {
|
|
4384
|
-
fusedScores.set(r.id, {
|
|
4385
|
-
score: r.score * semanticWeight,
|
|
4386
|
-
metadata: r.metadata
|
|
4387
|
-
});
|
|
4388
|
-
}
|
|
4389
|
-
for (const r of keywordResults) {
|
|
4390
|
-
const existing = fusedScores.get(r.id);
|
|
4391
|
-
if (existing) {
|
|
4392
|
-
existing.score += r.score * keywordWeight;
|
|
4393
|
-
} else {
|
|
4394
|
-
fusedScores.set(r.id, {
|
|
4395
|
-
score: r.score * keywordWeight,
|
|
4396
|
-
metadata: r.metadata
|
|
4397
|
-
});
|
|
4398
|
-
}
|
|
4399
|
-
}
|
|
4400
|
-
const results = Array.from(fusedScores.entries()).map(([id, data]) => ({
|
|
4401
|
-
id,
|
|
4402
|
-
score: data.score,
|
|
4403
|
-
metadata: data.metadata
|
|
4404
|
-
}));
|
|
4405
|
-
results.sort((a, b) => b.score - a.score);
|
|
4406
|
-
return results.slice(0, limit);
|
|
4407
|
-
}
|
|
4408
5523
|
async getStatus() {
|
|
4409
5524
|
const { store, configuredProviderInfo } = await this.ensureInitialized();
|
|
4410
5525
|
return {
|
|
@@ -4427,6 +5542,7 @@ var Indexer = class {
|
|
|
4427
5542
|
this.fileHashCache.clear();
|
|
4428
5543
|
this.saveFileHashCache();
|
|
4429
5544
|
database.clearBranch(this.currentBranch);
|
|
5545
|
+
database.clearBranchSymbols(this.currentBranch);
|
|
4430
5546
|
database.deleteMetadata("index.version");
|
|
4431
5547
|
database.deleteMetadata("index.embeddingProvider");
|
|
4432
5548
|
database.deleteMetadata("index.embeddingModel");
|
|
@@ -4455,6 +5571,8 @@ var Indexer = class {
|
|
|
4455
5571
|
removedCount++;
|
|
4456
5572
|
}
|
|
4457
5573
|
database.deleteChunksByFile(filePath);
|
|
5574
|
+
database.deleteCallEdgesByFile(filePath);
|
|
5575
|
+
database.deleteSymbolsByFile(filePath);
|
|
4458
5576
|
removedFilePaths.push(filePath);
|
|
4459
5577
|
}
|
|
4460
5578
|
}
|
|
@@ -4464,6 +5582,8 @@ var Indexer = class {
|
|
|
4464
5582
|
}
|
|
4465
5583
|
const gcOrphanEmbeddings = database.gcOrphanEmbeddings();
|
|
4466
5584
|
const gcOrphanChunks = database.gcOrphanChunks();
|
|
5585
|
+
const gcOrphanSymbols = database.gcOrphanSymbols();
|
|
5586
|
+
const gcOrphanCallEdges = database.gcOrphanCallEdges();
|
|
4467
5587
|
this.logger.recordGc(removedCount, gcOrphanChunks, gcOrphanEmbeddings);
|
|
4468
5588
|
this.logger.gc("info", "Health check complete", {
|
|
4469
5589
|
removedStale: removedCount,
|
|
@@ -4471,7 +5591,7 @@ var Indexer = class {
|
|
|
4471
5591
|
orphanChunks: gcOrphanChunks,
|
|
4472
5592
|
removedFiles: removedFilePaths.length
|
|
4473
5593
|
});
|
|
4474
|
-
return { removed: removedCount, filePaths: removedFilePaths, gcOrphanEmbeddings, gcOrphanChunks };
|
|
5594
|
+
return { removed: removedCount, filePaths: removedFilePaths, gcOrphanEmbeddings, gcOrphanChunks, gcOrphanSymbols, gcOrphanCallEdges };
|
|
4475
5595
|
}
|
|
4476
5596
|
async retryFailedBatches() {
|
|
4477
5597
|
const { store, provider, invertedIndex } = await this.ensureInitialized();
|
|
@@ -4577,9 +5697,29 @@ var Indexer = class {
|
|
|
4577
5697
|
if (filterByBranch && this.currentBranch !== "default") {
|
|
4578
5698
|
branchChunkIds = new Set(database.getBranchChunkIds(this.currentBranch));
|
|
4579
5699
|
}
|
|
4580
|
-
const
|
|
5700
|
+
const prefilterStartTime = import_perf_hooks.performance.now();
|
|
5701
|
+
const shouldPrefilterByBranch = branchChunkIds !== null && branchChunkIds.size > 0;
|
|
5702
|
+
const prefilteredSemantic = shouldPrefilterByBranch && branchChunkIds ? semanticResults.filter((r) => branchChunkIds.has(r.id)) : semanticResults;
|
|
5703
|
+
const semanticCandidates = shouldPrefilterByBranch && semanticResults.length > 0 && prefilteredSemantic.length === 0 ? semanticResults : prefilteredSemantic;
|
|
5704
|
+
const prefilterMs = import_perf_hooks.performance.now() - prefilterStartTime;
|
|
5705
|
+
if (branchChunkIds && branchChunkIds.size === 0) {
|
|
5706
|
+
this.logger.search("warn", "Branch prefilter skipped because branch catalog is empty", {
|
|
5707
|
+
branch: this.currentBranch
|
|
5708
|
+
});
|
|
5709
|
+
}
|
|
5710
|
+
if (shouldPrefilterByBranch && semanticResults.length > 0 && prefilteredSemantic.length === 0) {
|
|
5711
|
+
this.logger.search("warn", "Branch prefilter produced no semantic overlap, using unfiltered semantic candidates", {
|
|
5712
|
+
branch: this.currentBranch
|
|
5713
|
+
});
|
|
5714
|
+
}
|
|
5715
|
+
const rerankTopN = this.config.search.rerankTopN;
|
|
5716
|
+
const ranked = rankSemanticOnlyResults(code, semanticCandidates, {
|
|
5717
|
+
rerankTopN,
|
|
5718
|
+
limit,
|
|
5719
|
+
prioritizeSourcePaths: false
|
|
5720
|
+
});
|
|
5721
|
+
const filtered = ranked.filter((r) => {
|
|
4581
5722
|
if (r.score < this.config.search.minScore) return false;
|
|
4582
|
-
if (branchChunkIds && !branchChunkIds.has(r.id)) return false;
|
|
4583
5723
|
if (options?.excludeFile) {
|
|
4584
5724
|
if (r.metadata.filePath === options.excludeFile) return false;
|
|
4585
5725
|
}
|
|
@@ -4608,7 +5748,8 @@ var Indexer = class {
|
|
|
4608
5748
|
results: filtered.length,
|
|
4609
5749
|
totalMs: Math.round(totalSearchMs * 100) / 100,
|
|
4610
5750
|
embeddingMs: Math.round(embeddingMs * 100) / 100,
|
|
4611
|
-
vectorMs: Math.round(vectorMs * 100) / 100
|
|
5751
|
+
vectorMs: Math.round(vectorMs * 100) / 100,
|
|
5752
|
+
prefilterMs: Math.round(prefilterMs * 100) / 100
|
|
4612
5753
|
});
|
|
4613
5754
|
return Promise.all(
|
|
4614
5755
|
filtered.map(async (r) => {
|
|
@@ -4637,6 +5778,14 @@ var Indexer = class {
|
|
|
4637
5778
|
})
|
|
4638
5779
|
);
|
|
4639
5780
|
}
|
|
5781
|
+
async getCallers(targetName) {
|
|
5782
|
+
const { database } = await this.ensureInitialized();
|
|
5783
|
+
return database.getCallersWithContext(targetName, this.currentBranch);
|
|
5784
|
+
}
|
|
5785
|
+
async getCallees(symbolId) {
|
|
5786
|
+
const { database } = await this.ensureInitialized();
|
|
5787
|
+
return database.getCallees(symbolId, this.currentBranch);
|
|
5788
|
+
}
|
|
4640
5789
|
};
|
|
4641
5790
|
|
|
4642
5791
|
// node_modules/chokidar/index.js
|
|
@@ -6695,7 +7844,7 @@ ${formatted.join("\n")}
|
|
|
6695
7844
|
Use Read tool to examine specific files.`;
|
|
6696
7845
|
}
|
|
6697
7846
|
function formatHealthCheck(result) {
|
|
6698
|
-
if (result.removed === 0 && result.gcOrphanEmbeddings === 0 && result.gcOrphanChunks === 0) {
|
|
7847
|
+
if (result.removed === 0 && result.gcOrphanEmbeddings === 0 && result.gcOrphanChunks === 0 && result.gcOrphanSymbols === 0 && result.gcOrphanCallEdges === 0) {
|
|
6699
7848
|
return "Index is healthy. No stale entries found.";
|
|
6700
7849
|
}
|
|
6701
7850
|
const lines = [`Health check complete:`];
|
|
@@ -6708,6 +7857,12 @@ function formatHealthCheck(result) {
|
|
|
6708
7857
|
if (result.gcOrphanChunks > 0) {
|
|
6709
7858
|
lines.push(` Garbage collected orphan chunks: ${result.gcOrphanChunks}`);
|
|
6710
7859
|
}
|
|
7860
|
+
if (result.gcOrphanSymbols > 0) {
|
|
7861
|
+
lines.push(` Garbage collected orphan symbols: ${result.gcOrphanSymbols}`);
|
|
7862
|
+
}
|
|
7863
|
+
if (result.gcOrphanCallEdges > 0) {
|
|
7864
|
+
lines.push(` Garbage collected orphan call edges: ${result.gcOrphanCallEdges}`);
|
|
7865
|
+
}
|
|
6711
7866
|
if (result.filePaths.length > 0) {
|
|
6712
7867
|
lines.push(` Cleaned paths: ${result.filePaths.join(", ")}`);
|
|
6713
7868
|
}
|
|
@@ -6907,6 +8062,42 @@ var codebase_search = (0, import_plugin.tool)({
|
|
|
6907
8062
|
${formatSearchResults(results, "score")}`;
|
|
6908
8063
|
}
|
|
6909
8064
|
});
|
|
8065
|
+
var call_graph = (0, import_plugin.tool)({
|
|
8066
|
+
description: "Query the call graph to find callers or callees of a function/method. Use to understand code flow and dependencies between functions.",
|
|
8067
|
+
args: {
|
|
8068
|
+
name: z.string().describe("Function or method name to query"),
|
|
8069
|
+
direction: z.enum(["callers", "callees"]).default("callers").describe("Direction: 'callers' finds who calls this function, 'callees' finds what this function calls"),
|
|
8070
|
+
symbolId: z.string().optional().describe("Symbol ID (required for 'callees' direction, returned by previous call_graph queries)")
|
|
8071
|
+
},
|
|
8072
|
+
async execute(args) {
|
|
8073
|
+
const indexer = getIndexer();
|
|
8074
|
+
if (args.direction === "callees") {
|
|
8075
|
+
if (!args.symbolId) {
|
|
8076
|
+
return "Error: 'symbolId' is required when direction is 'callees'. First use direction='callers' to find the symbol ID.";
|
|
8077
|
+
}
|
|
8078
|
+
const callees = await indexer.getCallees(args.symbolId);
|
|
8079
|
+
if (callees.length === 0) {
|
|
8080
|
+
return `No callees found for symbol ${args.symbolId}. The function may not call any other tracked functions.`;
|
|
8081
|
+
}
|
|
8082
|
+
const formatted2 = callees.map(
|
|
8083
|
+
(e, i) => `[${i + 1}] \u2192 ${e.targetName} (${e.callType}) at line ${e.line}${e.isResolved ? ` [resolved: ${e.toSymbolId}]` : " [unresolved]"}`
|
|
8084
|
+
);
|
|
8085
|
+
return `${args.name} calls ${callees.length} function(s):
|
|
8086
|
+
|
|
8087
|
+
${formatted2.join("\n")}`;
|
|
8088
|
+
}
|
|
8089
|
+
const callers = await indexer.getCallers(args.name);
|
|
8090
|
+
if (callers.length === 0) {
|
|
8091
|
+
return `No callers found for "${args.name}". It may not be called by any tracked function, or the index needs updating.`;
|
|
8092
|
+
}
|
|
8093
|
+
const formatted = callers.map(
|
|
8094
|
+
(e, i) => `[${i + 1}] \u2190 from ${e.fromSymbolName ?? "<unknown>"} in ${e.fromSymbolFilePath ?? "<unknown file>"} [${e.fromSymbolId}] (${e.callType}) at line ${e.line}${e.isResolved ? " [resolved]" : " [unresolved]"}`
|
|
8095
|
+
);
|
|
8096
|
+
return `"${args.name}" is called by ${callers.length} function(s):
|
|
8097
|
+
|
|
8098
|
+
${formatted.join("\n")}`;
|
|
8099
|
+
}
|
|
8100
|
+
});
|
|
6910
8101
|
|
|
6911
8102
|
// src/commands/loader.ts
|
|
6912
8103
|
var import_fs5 = require("fs");
|
|
@@ -7011,7 +8202,8 @@ var plugin = async ({ directory }) => {
|
|
|
7011
8202
|
index_health_check,
|
|
7012
8203
|
index_metrics,
|
|
7013
8204
|
index_logs,
|
|
7014
|
-
find_similar
|
|
8205
|
+
find_similar,
|
|
8206
|
+
call_graph
|
|
7015
8207
|
},
|
|
7016
8208
|
async config(cfg) {
|
|
7017
8209
|
cfg.command = cfg.command ?? {};
|