@hsingjui/contextweaver 0.0.5 → 0.0.6
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/dist/{SearchService-SRT2KP6A.js → SearchService-ZI7QP3NE.js} +3 -3
- package/dist/{chunk-XHJNV3MK.js → chunk-2SIQBQ2M.js} +5 -5
- package/dist/{chunk-T337P433.js → chunk-AA3ILFHL.js} +162 -73
- package/dist/{chunk-C7XDGBT5.js → chunk-JVKVSTQ3.js} +1 -3
- package/dist/{chunk-VW5RACJC.js → chunk-OXPWRE3G.js} +1 -1
- package/dist/{chunk-5MZUQNOD.js → chunk-TJHS7BN7.js} +2 -2
- package/dist/{codebaseRetrieval-7SCFUVKL.js → codebaseRetrieval-VZ2E5JYE.js} +3 -3
- package/dist/index.js +25 -22
- package/dist/{lock-PX2BX2YN.js → lock-RC33CJZA.js} +1 -1
- package/dist/scanner-NM7WPJJE.js +10 -0
- package/dist/{server-GSXFZX6I.js → server-SOOY6RCA.js} +3 -3
- package/package.json +12 -12
- package/dist/scanner-7AZ4CHAR.js +0 -10
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
import {
|
|
2
2
|
getIndexer,
|
|
3
3
|
getVectorStore
|
|
4
|
-
} from "./chunk-
|
|
4
|
+
} from "./chunk-TJHS7BN7.js";
|
|
5
5
|
import {
|
|
6
6
|
initDb,
|
|
7
7
|
isChunksFtsInitialized,
|
|
@@ -9,11 +9,11 @@ import {
|
|
|
9
9
|
searchChunksFts,
|
|
10
10
|
searchFilesFts,
|
|
11
11
|
segmentQuery
|
|
12
|
-
} from "./chunk-
|
|
12
|
+
} from "./chunk-OXPWRE3G.js";
|
|
13
13
|
import {
|
|
14
14
|
isDebugEnabled,
|
|
15
15
|
logger
|
|
16
|
-
} from "./chunk-
|
|
16
|
+
} from "./chunk-JVKVSTQ3.js";
|
|
17
17
|
import {
|
|
18
18
|
getEmbeddingConfig,
|
|
19
19
|
getRerankerConfig
|
|
@@ -1,9 +1,9 @@
|
|
|
1
1
|
import {
|
|
2
2
|
generateProjectId
|
|
3
|
-
} from "./chunk-
|
|
3
|
+
} from "./chunk-OXPWRE3G.js";
|
|
4
4
|
import {
|
|
5
5
|
logger
|
|
6
|
-
} from "./chunk-
|
|
6
|
+
} from "./chunk-JVKVSTQ3.js";
|
|
7
7
|
|
|
8
8
|
// src/mcp/tools/codebaseRetrieval.ts
|
|
9
9
|
import fs from "fs";
|
|
@@ -68,8 +68,8 @@ function isProjectIndexed(projectId) {
|
|
|
68
68
|
return fs.existsSync(dbPath);
|
|
69
69
|
}
|
|
70
70
|
async function ensureIndexed(repoPath, projectId, onProgress) {
|
|
71
|
-
const { withLock } = await import("./lock-
|
|
72
|
-
const { scan } = await import("./scanner-
|
|
71
|
+
const { withLock } = await import("./lock-RC33CJZA.js");
|
|
72
|
+
const { scan } = await import("./scanner-NM7WPJJE.js");
|
|
73
73
|
await withLock(projectId, "index", async () => {
|
|
74
74
|
const wasIndexed = isProjectIndexed(projectId);
|
|
75
75
|
if (!wasIndexed) {
|
|
@@ -129,7 +129,7 @@ async function handleCodebaseRetrieval(args, configOverride = ZEN_CONFIG_OVERRID
|
|
|
129
129
|
},
|
|
130
130
|
"MCP \u67E5\u8BE2\u6784\u5EFA"
|
|
131
131
|
);
|
|
132
|
-
const { SearchService } = await import("./SearchService-
|
|
132
|
+
const { SearchService } = await import("./SearchService-ZI7QP3NE.js");
|
|
133
133
|
const service = new SearchService(projectId, repo_path, configOverride);
|
|
134
134
|
await service.init();
|
|
135
135
|
logger.debug("SearchService \u521D\u59CB\u5316\u5B8C\u6210");
|
|
@@ -2,7 +2,7 @@ import {
|
|
|
2
2
|
closeAllIndexers,
|
|
3
3
|
closeAllVectorStores,
|
|
4
4
|
getIndexer
|
|
5
|
-
} from "./chunk-
|
|
5
|
+
} from "./chunk-TJHS7BN7.js";
|
|
6
6
|
import {
|
|
7
7
|
batchDelete,
|
|
8
8
|
batchUpdateMtime,
|
|
@@ -16,10 +16,10 @@ import {
|
|
|
16
16
|
getStoredEmbeddingDimensions,
|
|
17
17
|
initDb,
|
|
18
18
|
setStoredEmbeddingDimensions
|
|
19
|
-
} from "./chunk-
|
|
19
|
+
} from "./chunk-OXPWRE3G.js";
|
|
20
20
|
import {
|
|
21
21
|
logger
|
|
22
|
-
} from "./chunk-
|
|
22
|
+
} from "./chunk-JVKVSTQ3.js";
|
|
23
23
|
import {
|
|
24
24
|
getEmbeddingConfig,
|
|
25
25
|
getExcludePatterns
|
|
@@ -165,7 +165,7 @@ import path2 from "path";
|
|
|
165
165
|
import pLimit from "p-limit";
|
|
166
166
|
|
|
167
167
|
// src/chunking/ParserPool.ts
|
|
168
|
-
import Parser from "tree-sitter";
|
|
168
|
+
import Parser from "@keqingmoe/tree-sitter";
|
|
169
169
|
var GRAMMAR_MODULES = {
|
|
170
170
|
typescript: "tree-sitter-typescript",
|
|
171
171
|
javascript: "tree-sitter-javascript",
|
|
@@ -227,20 +227,28 @@ function isLanguageSupported(language) {
|
|
|
227
227
|
var LANGUAGE_SPECS = {
|
|
228
228
|
typescript: {
|
|
229
229
|
hierarchy: /* @__PURE__ */ new Set([
|
|
230
|
-
// 类和接口
|
|
231
230
|
"class_declaration",
|
|
232
231
|
"abstract_class_declaration",
|
|
233
232
|
"interface_declaration",
|
|
234
|
-
// 函数
|
|
235
233
|
"function_declaration",
|
|
236
234
|
"generator_function_declaration",
|
|
237
235
|
"method_definition",
|
|
238
236
|
"arrow_function",
|
|
239
|
-
// 模块
|
|
240
237
|
"export_statement",
|
|
241
238
|
"import_statement"
|
|
242
239
|
]),
|
|
243
|
-
nameFields: ["name", "id"]
|
|
240
|
+
nameFields: ["name", "id"],
|
|
241
|
+
nameNodeTypes: /* @__PURE__ */ new Set(["identifier", "type_identifier", "property_identifier"]),
|
|
242
|
+
prefixMap: {
|
|
243
|
+
class_declaration: "class ",
|
|
244
|
+
abstract_class_declaration: "abstract class ",
|
|
245
|
+
interface_declaration: "interface ",
|
|
246
|
+
function_declaration: "fn ",
|
|
247
|
+
generator_function_declaration: "fn* ",
|
|
248
|
+
method_definition: "",
|
|
249
|
+
arrow_function: ""
|
|
250
|
+
},
|
|
251
|
+
commentTypes: /* @__PURE__ */ new Set(["comment"])
|
|
244
252
|
},
|
|
245
253
|
javascript: {
|
|
246
254
|
hierarchy: /* @__PURE__ */ new Set([
|
|
@@ -250,57 +258,94 @@ var LANGUAGE_SPECS = {
|
|
|
250
258
|
"method_definition",
|
|
251
259
|
"arrow_function"
|
|
252
260
|
]),
|
|
253
|
-
nameFields: ["name", "id"]
|
|
261
|
+
nameFields: ["name", "id"],
|
|
262
|
+
nameNodeTypes: /* @__PURE__ */ new Set(["identifier", "property_identifier"]),
|
|
263
|
+
prefixMap: {
|
|
264
|
+
class_declaration: "class ",
|
|
265
|
+
function_declaration: "fn ",
|
|
266
|
+
generator_function_declaration: "fn* ",
|
|
267
|
+
method_definition: "",
|
|
268
|
+
arrow_function: ""
|
|
269
|
+
},
|
|
270
|
+
commentTypes: /* @__PURE__ */ new Set(["comment"])
|
|
254
271
|
},
|
|
255
272
|
python: {
|
|
256
273
|
hierarchy: /* @__PURE__ */ new Set(["class_definition", "function_definition", "decorated_definition"]),
|
|
257
|
-
nameFields: ["name"]
|
|
274
|
+
nameFields: ["name"],
|
|
275
|
+
nameNodeTypes: /* @__PURE__ */ new Set(["identifier"]),
|
|
276
|
+
prefixMap: {
|
|
277
|
+
class_definition: "class ",
|
|
278
|
+
function_definition: "def ",
|
|
279
|
+
decorated_definition: ""
|
|
280
|
+
},
|
|
281
|
+
commentTypes: /* @__PURE__ */ new Set(["comment"])
|
|
258
282
|
},
|
|
259
283
|
go: {
|
|
260
284
|
hierarchy: /* @__PURE__ */ new Set([
|
|
261
|
-
// 函数和方法
|
|
262
285
|
"function_declaration",
|
|
263
286
|
"method_declaration",
|
|
264
|
-
// 类型定义
|
|
265
287
|
"type_spec",
|
|
266
288
|
"type_declaration",
|
|
267
|
-
// 结构体和接口
|
|
268
289
|
"struct_type",
|
|
269
290
|
"interface_type"
|
|
270
291
|
]),
|
|
271
|
-
nameFields: ["name"]
|
|
292
|
+
nameFields: ["name"],
|
|
293
|
+
nameNodeTypes: /* @__PURE__ */ new Set(["identifier", "type_identifier", "field_identifier"]),
|
|
294
|
+
prefixMap: {
|
|
295
|
+
function_declaration: "func ",
|
|
296
|
+
method_declaration: "func ",
|
|
297
|
+
type_spec: "type ",
|
|
298
|
+
type_declaration: "type ",
|
|
299
|
+
struct_type: "struct ",
|
|
300
|
+
interface_type: "interface "
|
|
301
|
+
},
|
|
302
|
+
commentTypes: /* @__PURE__ */ new Set(["comment"])
|
|
272
303
|
},
|
|
273
304
|
rust: {
|
|
274
305
|
hierarchy: /* @__PURE__ */ new Set([
|
|
275
|
-
// 函数
|
|
276
306
|
"function_item",
|
|
277
|
-
// 结构体、枚举、trait
|
|
278
307
|
"struct_item",
|
|
279
308
|
"enum_item",
|
|
280
309
|
"trait_item",
|
|
281
|
-
// impl 块
|
|
282
310
|
"impl_item",
|
|
283
|
-
// 模块
|
|
284
311
|
"mod_item",
|
|
285
|
-
// 类型别名
|
|
286
312
|
"type_item"
|
|
287
313
|
]),
|
|
288
|
-
nameFields: ["name"]
|
|
314
|
+
nameFields: ["name"],
|
|
315
|
+
nameNodeTypes: /* @__PURE__ */ new Set(["identifier", "type_identifier"]),
|
|
316
|
+
prefixMap: {
|
|
317
|
+
function_item: "fn ",
|
|
318
|
+
struct_item: "struct ",
|
|
319
|
+
enum_item: "enum ",
|
|
320
|
+
trait_item: "trait ",
|
|
321
|
+
impl_item: "impl ",
|
|
322
|
+
mod_item: "mod ",
|
|
323
|
+
type_item: "type "
|
|
324
|
+
},
|
|
325
|
+
commentTypes: /* @__PURE__ */ new Set(["line_comment", "block_comment"])
|
|
289
326
|
},
|
|
290
327
|
java: {
|
|
291
328
|
hierarchy: /* @__PURE__ */ new Set([
|
|
292
|
-
// 类和接口
|
|
293
329
|
"class_declaration",
|
|
294
330
|
"interface_declaration",
|
|
295
331
|
"enum_declaration",
|
|
296
332
|
"annotation_type_declaration",
|
|
297
|
-
// 方法和构造函数
|
|
298
333
|
"method_declaration",
|
|
299
334
|
"constructor_declaration",
|
|
300
|
-
// 记录类型 (Java 14+)
|
|
301
335
|
"record_declaration"
|
|
302
336
|
]),
|
|
303
|
-
nameFields: ["name", "identifier"]
|
|
337
|
+
nameFields: ["name", "identifier"],
|
|
338
|
+
nameNodeTypes: /* @__PURE__ */ new Set(["identifier"]),
|
|
339
|
+
prefixMap: {
|
|
340
|
+
class_declaration: "class ",
|
|
341
|
+
interface_declaration: "interface ",
|
|
342
|
+
enum_declaration: "enum ",
|
|
343
|
+
annotation_type_declaration: "@interface ",
|
|
344
|
+
method_declaration: "",
|
|
345
|
+
constructor_declaration: "",
|
|
346
|
+
record_declaration: "record "
|
|
347
|
+
},
|
|
348
|
+
commentTypes: /* @__PURE__ */ new Set(["line_comment", "block_comment"])
|
|
304
349
|
}
|
|
305
350
|
};
|
|
306
351
|
function getLanguageSpec(language) {
|
|
@@ -466,11 +511,11 @@ var SemanticSplitter = class {
|
|
|
466
511
|
code;
|
|
467
512
|
language;
|
|
468
513
|
constructor(config = {}) {
|
|
469
|
-
const maxChunkSize = config.maxChunkSize ??
|
|
514
|
+
const maxChunkSize = config.maxChunkSize ?? 2500;
|
|
470
515
|
this.config = {
|
|
471
516
|
maxChunkSize,
|
|
472
|
-
minChunkSize: config.minChunkSize ??
|
|
473
|
-
chunkOverlap: config.chunkOverlap ??
|
|
517
|
+
minChunkSize: config.minChunkSize ?? 100,
|
|
518
|
+
chunkOverlap: config.chunkOverlap ?? 200,
|
|
474
519
|
// 物理字符硬上限:默认为 maxChunkSize * 4(假设 1 token ≈ 4 chars)
|
|
475
520
|
maxRawChars: config.maxRawChars ?? maxChunkSize * 4
|
|
476
521
|
};
|
|
@@ -619,22 +664,10 @@ ${displayCode}`,
|
|
|
619
664
|
let nextContext = context;
|
|
620
665
|
const spec = getLanguageSpec(this.language);
|
|
621
666
|
if (spec?.hierarchy.has(node.type)) {
|
|
622
|
-
|
|
623
|
-
for (const child of node.namedChildren) {
|
|
624
|
-
if (child.type === "identifier" || child.type === "type_identifier" || child.type === "name") {
|
|
625
|
-
name = child.text;
|
|
626
|
-
break;
|
|
627
|
-
}
|
|
628
|
-
}
|
|
629
|
-
if (!name && node.firstNamedChild) {
|
|
630
|
-
const firstChild = node.firstNamedChild;
|
|
631
|
-
if (firstChild.text.length <= 100 && !firstChild.text.includes("\n")) {
|
|
632
|
-
name = firstChild.text;
|
|
633
|
-
}
|
|
634
|
-
}
|
|
667
|
+
const name = this.extractNodeName(node, spec);
|
|
635
668
|
if (name) {
|
|
636
|
-
const
|
|
637
|
-
nextContext = [...context, `${
|
|
669
|
+
const prefix = spec.prefixMap[node.type] ?? "";
|
|
670
|
+
nextContext = [...context, `${prefix}${name}`];
|
|
638
671
|
}
|
|
639
672
|
}
|
|
640
673
|
if (nodeSize <= this.config.maxChunkSize) {
|
|
@@ -651,21 +684,33 @@ ${displayCode}`,
|
|
|
651
684
|
return this.mergeAdjacentWindows(childWindows);
|
|
652
685
|
}
|
|
653
686
|
/**
|
|
654
|
-
*
|
|
687
|
+
* 从节点中提取名称(数据驱动)
|
|
655
688
|
*/
|
|
656
|
-
|
|
657
|
-
|
|
658
|
-
|
|
659
|
-
|
|
660
|
-
|
|
661
|
-
|
|
689
|
+
extractNodeName(node, spec) {
|
|
690
|
+
for (const child of node.namedChildren) {
|
|
691
|
+
if (spec.nameNodeTypes.has(child.type)) {
|
|
692
|
+
return child.text;
|
|
693
|
+
}
|
|
694
|
+
}
|
|
695
|
+
if (node.firstNamedChild) {
|
|
696
|
+
const firstChild = node.firstNamedChild;
|
|
697
|
+
if (firstChild.text.length <= 100 && !firstChild.text.includes("\n")) {
|
|
698
|
+
return firstChild.text;
|
|
699
|
+
}
|
|
700
|
+
}
|
|
701
|
+
return null;
|
|
662
702
|
}
|
|
663
703
|
/**
|
|
664
704
|
* Gap-Aware 相邻窗口合并
|
|
665
705
|
*
|
|
666
|
-
*
|
|
706
|
+
* 使用三重预算策略:
|
|
667
707
|
* - NWS 预算:控制有效代码量
|
|
668
708
|
* - Raw 预算:控制物理字符数,防止大量注释撑爆 Token
|
|
709
|
+
* - 语义边界惩罚:不同 contextPath 的窗口合并门槛更高
|
|
710
|
+
*
|
|
711
|
+
* 前向吸附策略:
|
|
712
|
+
* - 如果当前窗口以 comment 结尾,将 comment 推到下一个窗口
|
|
713
|
+
* - 保证 JSDoc/注释与其描述的代码在同一个 chunk
|
|
669
714
|
*/
|
|
670
715
|
mergeAdjacentWindows(windows) {
|
|
671
716
|
if (windows.length === 0) return [];
|
|
@@ -673,6 +718,11 @@ ${displayCode}`,
|
|
|
673
718
|
let current = windows[0];
|
|
674
719
|
for (let i = 1; i < windows.length; i++) {
|
|
675
720
|
const next = windows[i];
|
|
721
|
+
this.forwardAbsorbComments(current, next);
|
|
722
|
+
if (current.nodes.length === 0) {
|
|
723
|
+
current = next;
|
|
724
|
+
continue;
|
|
725
|
+
}
|
|
676
726
|
const currentStart = current.nodes[0].startIndex;
|
|
677
727
|
const currentEnd = current.nodes[current.nodes.length - 1].endIndex;
|
|
678
728
|
const nextStart = next.nodes[0].startIndex;
|
|
@@ -680,13 +730,18 @@ ${displayCode}`,
|
|
|
680
730
|
const gapNws = this.adapter.nws(currentEnd, nextStart);
|
|
681
731
|
const combinedNws = current.size + gapNws + next.size;
|
|
682
732
|
const combinedRawLen = nextEnd - currentStart;
|
|
733
|
+
const sameContext = this.isSameContext(current.contextPath, next.contextPath);
|
|
734
|
+
const boundaryPenalty = sameContext ? 1 : 0.7;
|
|
683
735
|
const isTiny = current.size < this.config.minChunkSize;
|
|
684
|
-
const
|
|
685
|
-
const
|
|
736
|
+
const effectiveBudget = this.config.maxChunkSize * boundaryPenalty;
|
|
737
|
+
const fitsNwsBudget = combinedNws <= effectiveBudget || isTiny && combinedNws < effectiveBudget * 1.5;
|
|
738
|
+
const fitsRawBudget = combinedRawLen <= this.config.maxRawChars * boundaryPenalty;
|
|
686
739
|
if (fitsNwsBudget && fitsRawBudget) {
|
|
687
740
|
current.nodes.push(...next.nodes);
|
|
688
741
|
current.size = combinedNws;
|
|
689
|
-
|
|
742
|
+
if (next.contextPath.length > current.contextPath.length) {
|
|
743
|
+
current.contextPath = next.contextPath;
|
|
744
|
+
}
|
|
690
745
|
} else {
|
|
691
746
|
merged.push(current);
|
|
692
747
|
current = next;
|
|
@@ -695,6 +750,60 @@ ${displayCode}`,
|
|
|
695
750
|
merged.push(current);
|
|
696
751
|
return merged;
|
|
697
752
|
}
|
|
753
|
+
/**
|
|
754
|
+
* 前向吸附:将 current 尾部的 comment 节点推到 next 头部
|
|
755
|
+
*
|
|
756
|
+
* 这确保 JSDoc/docstring/注释与其描述的函数/方法在同一个 chunk 中,
|
|
757
|
+
* 而不是被切到前一个 chunk 的末尾。
|
|
758
|
+
*
|
|
759
|
+
* 注意:此方法会直接修改 current 和 next
|
|
760
|
+
*/
|
|
761
|
+
forwardAbsorbComments(current, next) {
|
|
762
|
+
const spec = getLanguageSpec(this.language);
|
|
763
|
+
const commentTypes = spec?.commentTypes ?? /* @__PURE__ */ new Set(["comment"]);
|
|
764
|
+
const absorbedNodes = [];
|
|
765
|
+
let absorbedNws = 0;
|
|
766
|
+
while (current.nodes.length > 0) {
|
|
767
|
+
const lastNode = current.nodes[current.nodes.length - 1];
|
|
768
|
+
if (commentTypes.has(lastNode.type)) {
|
|
769
|
+
current.nodes.pop();
|
|
770
|
+
const nodeNws = this.adapter.nws(lastNode.startIndex, lastNode.endIndex);
|
|
771
|
+
absorbedNodes.unshift(lastNode);
|
|
772
|
+
absorbedNws += nodeNws;
|
|
773
|
+
current.size -= nodeNws;
|
|
774
|
+
} else {
|
|
775
|
+
break;
|
|
776
|
+
}
|
|
777
|
+
}
|
|
778
|
+
if (absorbedNodes.length > 0) {
|
|
779
|
+
const gapNws = next.nodes.length > 0 ? this.adapter.nws(
|
|
780
|
+
absorbedNodes[absorbedNodes.length - 1].endIndex,
|
|
781
|
+
next.nodes[0].startIndex
|
|
782
|
+
) : 0;
|
|
783
|
+
next.nodes.unshift(...absorbedNodes);
|
|
784
|
+
next.size += absorbedNws + gapNws;
|
|
785
|
+
}
|
|
786
|
+
}
|
|
787
|
+
/**
|
|
788
|
+
* 检查两个 contextPath 是否属于同一语义单元
|
|
789
|
+
*
|
|
790
|
+
* 规则:如果两者的公共前缀长度 >= 较短路径长度,认为是同一单元
|
|
791
|
+
* 例如:
|
|
792
|
+
* - ["file", "class A", "method foo"] 和 ["file", "class A", "method bar"] -> false(不同方法)
|
|
793
|
+
* - ["file", "class A"] 和 ["file", "class A", "method foo"] -> true(父子关系)
|
|
794
|
+
*/
|
|
795
|
+
isSameContext(a, b) {
|
|
796
|
+
const minLen = Math.min(a.length, b.length);
|
|
797
|
+
let commonLen = 0;
|
|
798
|
+
for (let i = 0; i < minLen; i++) {
|
|
799
|
+
if (a[i] === b[i]) {
|
|
800
|
+
commonLen++;
|
|
801
|
+
} else {
|
|
802
|
+
break;
|
|
803
|
+
}
|
|
804
|
+
}
|
|
805
|
+
return commonLen >= minLen;
|
|
806
|
+
}
|
|
698
807
|
/**
|
|
699
808
|
* 将窗口转换为最终的 ProcessedChunk
|
|
700
809
|
*
|
|
@@ -771,25 +880,6 @@ ${displayCode}`,
|
|
|
771
880
|
}
|
|
772
881
|
return Math.max(0, result);
|
|
773
882
|
}
|
|
774
|
-
/**
|
|
775
|
-
* 计算两个路径数组的最长公共前缀(LCA)
|
|
776
|
-
*
|
|
777
|
-
* 用于合并窗口时更新 contextPath,避免 breadcrumb 误标
|
|
778
|
-
* 例如:["file", "class A", "method foo"] 和 ["file", "class A", "method bar"]
|
|
779
|
-
* => ["file", "class A"]
|
|
780
|
-
*/
|
|
781
|
-
commonPrefix(a, b) {
|
|
782
|
-
const result = [];
|
|
783
|
-
const len = Math.min(a.length, b.length);
|
|
784
|
-
for (let i = 0; i < len; i++) {
|
|
785
|
-
if (a[i] === b[i]) {
|
|
786
|
-
result.push(a[i]);
|
|
787
|
-
} else {
|
|
788
|
-
break;
|
|
789
|
-
}
|
|
790
|
-
}
|
|
791
|
-
return result;
|
|
792
|
-
}
|
|
793
883
|
};
|
|
794
884
|
function generateVectorText(code, contextPath) {
|
|
795
885
|
const breadcrumb = contextPath.join(" > ");
|
|
@@ -1052,7 +1142,6 @@ async function scan(rootPath, options = {}) {
|
|
|
1052
1142
|
setStoredEmbeddingDimensions(db, currentDimensions);
|
|
1053
1143
|
}
|
|
1054
1144
|
if (forceReindex) {
|
|
1055
|
-
logger.info("\u5F3A\u5236\u91CD\u65B0\u7D22\u5F15...");
|
|
1056
1145
|
clear(db);
|
|
1057
1146
|
if (options.vectorIndex !== false) {
|
|
1058
1147
|
const embeddingConfig = getEmbeddingConfig();
|
|
@@ -75,9 +75,7 @@ function createFormattedStream(filePath) {
|
|
|
75
75
|
const { level: _l, time: _t, pid: _p, hostname: _h, name: _n, msg: _m, ...extra } = log;
|
|
76
76
|
let line = `${time} [${level}] ${msg}`;
|
|
77
77
|
if (Object.keys(extra).length > 0) {
|
|
78
|
-
|
|
79
|
-
line += `
|
|
80
|
-
${extraLines}`;
|
|
78
|
+
line += ` ${JSON.stringify(extra)}`;
|
|
81
79
|
}
|
|
82
80
|
writeStream.write(`${line}
|
|
83
81
|
`, callback);
|
|
@@ -4,10 +4,10 @@ import {
|
|
|
4
4
|
batchUpsertChunkFts,
|
|
5
5
|
clearVectorIndexHash,
|
|
6
6
|
isChunksFtsInitialized
|
|
7
|
-
} from "./chunk-
|
|
7
|
+
} from "./chunk-OXPWRE3G.js";
|
|
8
8
|
import {
|
|
9
9
|
logger
|
|
10
|
-
} from "./chunk-
|
|
10
|
+
} from "./chunk-JVKVSTQ3.js";
|
|
11
11
|
import {
|
|
12
12
|
getEmbeddingConfig
|
|
13
13
|
} from "./chunk-SKBAE26T.js";
|
|
@@ -1,9 +1,9 @@
|
|
|
1
1
|
import {
|
|
2
2
|
codebaseRetrievalSchema,
|
|
3
3
|
handleCodebaseRetrieval
|
|
4
|
-
} from "./chunk-
|
|
5
|
-
import "./chunk-
|
|
6
|
-
import "./chunk-
|
|
4
|
+
} from "./chunk-2SIQBQ2M.js";
|
|
5
|
+
import "./chunk-OXPWRE3G.js";
|
|
6
|
+
import "./chunk-JVKVSTQ3.js";
|
|
7
7
|
import "./chunk-SKBAE26T.js";
|
|
8
8
|
export {
|
|
9
9
|
codebaseRetrievalSchema,
|
package/dist/index.js
CHANGED
|
@@ -1,22 +1,30 @@
|
|
|
1
1
|
#!/usr/bin/env node
|
|
2
2
|
import {
|
|
3
3
|
scan
|
|
4
|
-
} from "./chunk-
|
|
5
|
-
import "./chunk-
|
|
4
|
+
} from "./chunk-AA3ILFHL.js";
|
|
5
|
+
import "./chunk-TJHS7BN7.js";
|
|
6
6
|
import {
|
|
7
7
|
generateProjectId
|
|
8
|
-
} from "./chunk-
|
|
8
|
+
} from "./chunk-OXPWRE3G.js";
|
|
9
9
|
import {
|
|
10
10
|
logger
|
|
11
|
-
} from "./chunk-
|
|
11
|
+
} from "./chunk-JVKVSTQ3.js";
|
|
12
12
|
import "./chunk-SKBAE26T.js";
|
|
13
13
|
|
|
14
14
|
// src/index.ts
|
|
15
15
|
import { promises as fs } from "fs";
|
|
16
16
|
import os from "os";
|
|
17
17
|
import path from "path";
|
|
18
|
+
import { fileURLToPath } from "url";
|
|
18
19
|
import cac from "cac";
|
|
20
|
+
var __dirname = path.dirname(fileURLToPath(import.meta.url));
|
|
21
|
+
var pkgPath = path.resolve(__dirname, "../package.json");
|
|
22
|
+
var pkg = JSON.parse(await fs.readFile(pkgPath, "utf-8"));
|
|
19
23
|
var cli = cac("contextweaver");
|
|
24
|
+
if (process.argv.includes("-v") || process.argv.includes("--version")) {
|
|
25
|
+
console.log(pkg.version);
|
|
26
|
+
process.exit(0);
|
|
27
|
+
}
|
|
20
28
|
cli.command("init", "\u521D\u59CB\u5316 ContextWeaver \u914D\u7F6E").action(async () => {
|
|
21
29
|
const configDir = path.join(os.homedir(), ".contextweaver");
|
|
22
30
|
const envFile = path.join(configDir, ".env");
|
|
@@ -80,38 +88,33 @@ cli.command("index [path]", "\u626B\u63CF\u4EE3\u7801\u5E93\u5E76\u5EFA\u7ACB\u7
|
|
|
80
88
|
}
|
|
81
89
|
const startTime = Date.now();
|
|
82
90
|
try {
|
|
91
|
+
let lastLoggedPercent = 0;
|
|
83
92
|
const stats = await scan(rootPath, {
|
|
84
93
|
force: options.force,
|
|
85
94
|
onProgress: (current, total, message) => {
|
|
86
95
|
if (total !== void 0) {
|
|
87
|
-
const percent = (current / total * 100)
|
|
88
|
-
|
|
89
|
-
|
|
90
|
-
|
|
91
|
-
|
|
92
|
-
logger.info({ current }, "\u626B\u63CF\u8FDB\u5EA6");
|
|
96
|
+
const percent = Math.floor(current / total * 100);
|
|
97
|
+
if (percent >= lastLoggedPercent + 30 && percent < 100) {
|
|
98
|
+
logger.info(`\u7D22\u5F15\u8FDB\u5EA6: ${percent}% - ${message || ""}`);
|
|
99
|
+
lastLoggedPercent = Math.floor(percent / 30) * 30;
|
|
100
|
+
}
|
|
93
101
|
}
|
|
94
102
|
}
|
|
95
103
|
});
|
|
96
104
|
process.stdout.write("\n");
|
|
97
105
|
const duration = ((Date.now() - startTime) / 1e3).toFixed(2);
|
|
98
|
-
logger.info(
|
|
99
|
-
logger.info(
|
|
100
|
-
|
|
101
|
-
|
|
102
|
-
logger.info(`\u4FEE\u6539: ${stats.modified}`);
|
|
103
|
-
logger.info(`\u672A\u53D8: ${stats.unchanged}`);
|
|
104
|
-
logger.info(`\u5220\u9664: ${stats.deleted}`);
|
|
105
|
-
logger.info(`\u8DF3\u8FC7: ${stats.skipped}`);
|
|
106
|
-
logger.info(`\u9519\u8BEF: ${stats.errors}`);
|
|
106
|
+
logger.info(`\u7D22\u5F15\u5B8C\u6210 (${duration}s)`);
|
|
107
|
+
logger.info(
|
|
108
|
+
`\u603B\u6570:${stats.totalFiles} \u65B0\u589E:${stats.added} \u4FEE\u6539:${stats.modified} \u672A\u53D8:${stats.unchanged} \u5220\u9664:${stats.deleted} \u8DF3\u8FC7:${stats.skipped} \u9519\u8BEF:${stats.errors}`
|
|
109
|
+
);
|
|
107
110
|
} catch (err) {
|
|
108
111
|
const error = err;
|
|
109
|
-
logger.error({ err, stack: error.stack }, `\
|
|
112
|
+
logger.error({ err, stack: error.stack }, `\u7D22\u5F15\u5931\u8D25: ${error.message}`);
|
|
110
113
|
process.exit(1);
|
|
111
114
|
}
|
|
112
115
|
});
|
|
113
116
|
cli.command("mcp", "\u542F\u52A8 MCP \u670D\u52A1\u5668").action(async () => {
|
|
114
|
-
const { startMcpServer } = await import("./server-
|
|
117
|
+
const { startMcpServer } = await import("./server-SOOY6RCA.js");
|
|
115
118
|
try {
|
|
116
119
|
await startMcpServer();
|
|
117
120
|
} catch (err) {
|
|
@@ -133,7 +136,7 @@ cli.command("search", "\u672C\u5730\u68C0\u7D22\uFF08\u53C2\u6570\u5BF9\u9F50 MC
|
|
|
133
136
|
}
|
|
134
137
|
const technicalTerms = (options.technicalTerms || "").split(",").map((t) => t.trim()).filter(Boolean);
|
|
135
138
|
const useZen = options.zen !== false;
|
|
136
|
-
const { handleCodebaseRetrieval } = await import("./codebaseRetrieval-
|
|
139
|
+
const { handleCodebaseRetrieval } = await import("./codebaseRetrieval-VZ2E5JYE.js");
|
|
137
140
|
const response = await handleCodebaseRetrieval(
|
|
138
141
|
{
|
|
139
142
|
repo_path: repoPath,
|
|
@@ -1,11 +1,11 @@
|
|
|
1
1
|
import {
|
|
2
2
|
codebaseRetrievalSchema,
|
|
3
3
|
handleCodebaseRetrieval
|
|
4
|
-
} from "./chunk-
|
|
5
|
-
import "./chunk-
|
|
4
|
+
} from "./chunk-2SIQBQ2M.js";
|
|
5
|
+
import "./chunk-OXPWRE3G.js";
|
|
6
6
|
import {
|
|
7
7
|
logger
|
|
8
|
-
} from "./chunk-
|
|
8
|
+
} from "./chunk-JVKVSTQ3.js";
|
|
9
9
|
import "./chunk-SKBAE26T.js";
|
|
10
10
|
|
|
11
11
|
// src/mcp/server.ts
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@hsingjui/contextweaver",
|
|
3
|
-
"version": "0.0.
|
|
3
|
+
"version": "0.0.6",
|
|
4
4
|
"description": "A context weaving tool for LLMs",
|
|
5
5
|
"license": "MIT",
|
|
6
6
|
"author": "hsingjui",
|
|
@@ -29,9 +29,10 @@
|
|
|
29
29
|
"fmt": "biome check --write ./src"
|
|
30
30
|
},
|
|
31
31
|
"dependencies": {
|
|
32
|
-
"@
|
|
32
|
+
"@keqingmoe/tree-sitter": "^0.26.2",
|
|
33
|
+
"@lancedb/lancedb": "^0.22.0",
|
|
33
34
|
"@modelcontextprotocol/sdk": "^1.25.1",
|
|
34
|
-
"better-sqlite3": "^
|
|
35
|
+
"better-sqlite3": "^12.2.0",
|
|
35
36
|
"cac": "^6.7.14",
|
|
36
37
|
"chardet": "^2.1.1",
|
|
37
38
|
"dotenv": "^17.2.3",
|
|
@@ -40,28 +41,27 @@
|
|
|
40
41
|
"ignore": "^7.0.5",
|
|
41
42
|
"p-limit": "^7.2.0",
|
|
42
43
|
"pino": "^10.1.0",
|
|
43
|
-
"tree-sitter": "0.
|
|
44
|
-
"tree-sitter-
|
|
45
|
-
"tree-sitter-
|
|
46
|
-
"tree-sitter-
|
|
47
|
-
"tree-sitter-
|
|
48
|
-
"tree-sitter-
|
|
49
|
-
"tree-sitter-typescript": "0.20.5",
|
|
44
|
+
"tree-sitter-go": "0.23.4",
|
|
45
|
+
"tree-sitter-java": "0.23.5",
|
|
46
|
+
"tree-sitter-javascript": "0.23.1",
|
|
47
|
+
"tree-sitter-python": "0.25.0",
|
|
48
|
+
"tree-sitter-rust": "0.21.0",
|
|
49
|
+
"tree-sitter-typescript": "0.23.2",
|
|
50
50
|
"zod": "^4.2.1"
|
|
51
51
|
},
|
|
52
52
|
"devDependencies": {
|
|
53
53
|
"@biomejs/biome": "2.3.10",
|
|
54
54
|
"@types/better-sqlite3": "^7.6.13",
|
|
55
|
-
"@types/node": "^
|
|
55
|
+
"@types/node": "^24.0.0",
|
|
56
56
|
"knip": "^5.78.0",
|
|
57
57
|
"tsup": "^8.5.1",
|
|
58
58
|
"typescript": "^5.9.3"
|
|
59
59
|
},
|
|
60
60
|
"pnpm": {
|
|
61
61
|
"onlyBuiltDependencies": [
|
|
62
|
+
"@keqingmoe/tree-sitter",
|
|
62
63
|
"better-sqlite3",
|
|
63
64
|
"esbuild",
|
|
64
|
-
"tree-sitter",
|
|
65
65
|
"tree-sitter-go",
|
|
66
66
|
"tree-sitter-java",
|
|
67
67
|
"tree-sitter-javascript",
|