opencode-rag-plugin 1.12.5 → 1.12.9
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/cli.js +59 -1
- package/dist/cli.js.map +1 -1
- package/dist/content/doc.d.ts +3 -0
- package/dist/content/doc.js +12 -0
- package/dist/content/doc.js.map +1 -0
- package/dist/content/docx.d.ts +3 -0
- package/dist/content/docx.js +12 -0
- package/dist/content/docx.js.map +1 -0
- package/dist/content/excel.d.ts +3 -0
- package/dist/content/excel.js +12 -0
- package/dist/content/excel.js.map +1 -0
- package/dist/content/image.d.ts +5 -0
- package/dist/content/image.js +24 -0
- package/dist/content/image.js.map +1 -0
- package/dist/content/pdf.d.ts +3 -0
- package/dist/content/pdf.js +12 -0
- package/dist/content/pdf.js.map +1 -0
- package/dist/content/reader.d.ts +18 -0
- package/dist/content/reader.js +104 -0
- package/dist/content/reader.js.map +1 -0
- package/dist/content/types.d.ts +5 -0
- package/dist/content/types.js +2 -0
- package/dist/content/types.js.map +1 -0
- package/dist/core/bootstrap.js +2 -2
- package/dist/core/bootstrap.js.map +1 -1
- package/dist/core/config.d.ts +5 -0
- package/dist/core/config.js +9 -1
- package/dist/core/config.js.map +1 -1
- package/dist/eval/run-token-test.js +2 -2
- package/dist/eval/run-token-test.js.map +1 -1
- package/dist/index.d.ts +2 -0
- package/dist/index.js +2 -0
- package/dist/index.js.map +1 -1
- package/dist/indexer/description-stage.d.ts +14 -0
- package/dist/indexer/description-stage.js +46 -0
- package/dist/indexer/description-stage.js.map +1 -0
- package/dist/indexer/embed-stage.d.ts +8 -0
- package/dist/indexer/embed-stage.js +34 -0
- package/dist/indexer/embed-stage.js.map +1 -0
- package/dist/indexer/metadata.d.ts +1 -0
- package/dist/indexer/metadata.js +99 -0
- package/dist/indexer/metadata.js.map +1 -0
- package/dist/indexer/pipeline.d.ts +22 -0
- package/dist/indexer/pipeline.js +179 -0
- package/dist/indexer/pipeline.js.map +1 -0
- package/dist/indexer/stats.d.ts +29 -0
- package/dist/indexer/stats.js +20 -0
- package/dist/indexer/stats.js.map +1 -0
- package/dist/indexer/watch.d.ts +8 -0
- package/dist/indexer/watch.js +95 -0
- package/dist/indexer/watch.js.map +1 -0
- package/dist/indexer/worker.d.ts +36 -0
- package/dist/indexer/worker.js +151 -0
- package/dist/indexer/worker.js.map +1 -0
- package/dist/indexer.d.ts +7 -54
- package/dist/indexer.js +5 -626
- package/dist/indexer.js.map +1 -1
- package/dist/plugin.d.ts +1 -1
- package/dist/plugin.js +37 -5
- package/dist/plugin.js.map +1 -1
- package/dist/updater.d.ts +19 -0
- package/dist/updater.js +132 -0
- package/dist/updater.js.map +1 -0
- package/dist/vectorstore/factory.d.ts +3 -0
- package/dist/vectorstore/factory.js +13 -0
- package/dist/vectorstore/factory.js.map +1 -0
- package/dist/vectorstore/memory.d.ts +9 -0
- package/dist/vectorstore/memory.js +41 -0
- package/dist/vectorstore/memory.js.map +1 -0
- package/package.json +4 -4
|
@@ -0,0 +1,179 @@
|
|
|
1
|
+
import pLimit from "p-limit";
|
|
2
|
+
import { scanWorkspaceFiles } from "../content/reader.js";
|
|
3
|
+
import { loadManifest, saveManifest } from "../core/manifest.js";
|
|
4
|
+
import { createIndexStats } from "./stats.js";
|
|
5
|
+
import { processFile } from "./worker.js";
|
|
6
|
+
export { createWatchPassScheduler, createWatchIgnore } from "./watch.js";
|
|
7
|
+
function createLogger(logger) {
|
|
8
|
+
return {
|
|
9
|
+
info: logger?.info ?? (() => { }),
|
|
10
|
+
warn: logger?.warn ?? (() => { }),
|
|
11
|
+
};
|
|
12
|
+
}
|
|
13
|
+
export async function runIndexPass(options) {
|
|
14
|
+
const logger = createLogger(options.logger);
|
|
15
|
+
const workspaceFiles = await scanWorkspaceFiles(options.cwd, options.config, logger);
|
|
16
|
+
const loadResult = await loadManifest(options.storePath);
|
|
17
|
+
const manifest = loadResult.manifest;
|
|
18
|
+
let manifestStatus = loadResult.status;
|
|
19
|
+
let rebuildPerformed = false;
|
|
20
|
+
const existingCount = await options.store.count();
|
|
21
|
+
if (options.force || (manifestStatus !== "ok" && existingCount > 0)) {
|
|
22
|
+
await options.store.clear();
|
|
23
|
+
options.keywordIndex?.clear();
|
|
24
|
+
for (const key of Object.keys(manifest.files)) {
|
|
25
|
+
delete manifest.files[key];
|
|
26
|
+
}
|
|
27
|
+
manifest.lastIndexedAt = undefined;
|
|
28
|
+
rebuildPerformed = existingCount > 0 || Boolean(options.force);
|
|
29
|
+
if (manifestStatus !== "ok" && existingCount > 0) {
|
|
30
|
+
logger.warn("Manifest missing or corrupt; rebuilding full index.");
|
|
31
|
+
}
|
|
32
|
+
manifestStatus = options.force ? "missing" : manifestStatus;
|
|
33
|
+
}
|
|
34
|
+
const stats = createIndexStats(workspaceFiles.length, manifestStatus);
|
|
35
|
+
stats.rebuildPerformed = rebuildPerformed;
|
|
36
|
+
for (const file of workspaceFiles) {
|
|
37
|
+
if (file.extractionStatus === "failed" && file.extractionError) {
|
|
38
|
+
stats.extractionFailures++;
|
|
39
|
+
stats.extractionErrors.push({
|
|
40
|
+
filePath: file.filePath,
|
|
41
|
+
error: file.extractionError,
|
|
42
|
+
});
|
|
43
|
+
}
|
|
44
|
+
}
|
|
45
|
+
const currentPaths = new Set(workspaceFiles.map((file) => file.normalizedPath));
|
|
46
|
+
for (const indexedPath of Object.keys(manifest.files)) {
|
|
47
|
+
if (!currentPaths.has(indexedPath)) {
|
|
48
|
+
await options.store.deleteByFilePath(indexedPath);
|
|
49
|
+
options.keywordIndex?.removeByFilePath(indexedPath);
|
|
50
|
+
delete manifest.files[indexedPath];
|
|
51
|
+
stats.deletedFiles++;
|
|
52
|
+
}
|
|
53
|
+
}
|
|
54
|
+
let newCount = 0;
|
|
55
|
+
let modifiedCount = 0;
|
|
56
|
+
let unchangedCount = 0;
|
|
57
|
+
for (const file of workspaceFiles) {
|
|
58
|
+
const previous = manifest.files[file.normalizedPath];
|
|
59
|
+
if (file.isEmpty || file.isTooSmall)
|
|
60
|
+
continue;
|
|
61
|
+
if (previous && previous.hash === file.hash) {
|
|
62
|
+
unchangedCount++;
|
|
63
|
+
}
|
|
64
|
+
else if (previous) {
|
|
65
|
+
modifiedCount++;
|
|
66
|
+
}
|
|
67
|
+
else {
|
|
68
|
+
newCount++;
|
|
69
|
+
}
|
|
70
|
+
}
|
|
71
|
+
logger.info(`Processing ${workspaceFiles.length} files (${newCount} new, ${modifiedCount} modified, ${unchangedCount} unchanged)...`);
|
|
72
|
+
const limit = pLimit(options.config.indexing.concurrency);
|
|
73
|
+
let completed = 0;
|
|
74
|
+
const total = workspaceFiles.length;
|
|
75
|
+
const workerResults = await Promise.all(workspaceFiles.map((file) => limit(async () => {
|
|
76
|
+
const result = await processFile(file, options.cwd, manifest.files[file.normalizedPath], options.config, options.store, options.keywordIndex, options.embedder, options.descriptionProvider, logger);
|
|
77
|
+
completed++;
|
|
78
|
+
if (total > 20 && completed % 50 === 0) {
|
|
79
|
+
logger.info(` Processing file ${completed}/${total}...`);
|
|
80
|
+
}
|
|
81
|
+
return result;
|
|
82
|
+
})));
|
|
83
|
+
aggregateStats(stats, workerResults, manifest);
|
|
84
|
+
manifest.lastIndexedAt = Date.now();
|
|
85
|
+
await saveManifest(options.storePath, manifest);
|
|
86
|
+
await options.keywordIndex?.save(options.storePath);
|
|
87
|
+
stats.finalCount = await options.store.count();
|
|
88
|
+
return stats;
|
|
89
|
+
}
|
|
90
|
+
function aggregateStats(stats, results, manifest) {
|
|
91
|
+
for (const result of results) {
|
|
92
|
+
if (result.isEmpty) {
|
|
93
|
+
stats.skippedEmptyFiles++;
|
|
94
|
+
if (result.isRemoved) {
|
|
95
|
+
delete manifest.files[result.normalizedPath];
|
|
96
|
+
stats.removedFiles++;
|
|
97
|
+
}
|
|
98
|
+
continue;
|
|
99
|
+
}
|
|
100
|
+
if (result.isTooSmall) {
|
|
101
|
+
stats.skippedSmallFiles++;
|
|
102
|
+
if (result.isRemoved) {
|
|
103
|
+
delete manifest.files[result.normalizedPath];
|
|
104
|
+
stats.removedFiles++;
|
|
105
|
+
}
|
|
106
|
+
continue;
|
|
107
|
+
}
|
|
108
|
+
if (result.isUnchanged) {
|
|
109
|
+
stats.unchangedFiles++;
|
|
110
|
+
continue;
|
|
111
|
+
}
|
|
112
|
+
if (result.isRemoved) {
|
|
113
|
+
delete manifest.files[result.normalizedPath];
|
|
114
|
+
stats.removedFiles++;
|
|
115
|
+
continue;
|
|
116
|
+
}
|
|
117
|
+
if (result.isModified) {
|
|
118
|
+
stats.modifiedFiles++;
|
|
119
|
+
}
|
|
120
|
+
else if (result.isNew) {
|
|
121
|
+
stats.newFiles++;
|
|
122
|
+
}
|
|
123
|
+
if (result.chunkCount > 0) {
|
|
124
|
+
manifest.files[result.normalizedPath] = {
|
|
125
|
+
hash: result.hash,
|
|
126
|
+
chunkCount: result.chunkCount,
|
|
127
|
+
indexedAt: Date.now(),
|
|
128
|
+
};
|
|
129
|
+
stats.totalChunks += result.chunkCount;
|
|
130
|
+
stats.batchesFlushed++;
|
|
131
|
+
}
|
|
132
|
+
}
|
|
133
|
+
}
|
|
134
|
+
export async function getIndexStatusSummary(cwd, storePath, config, store) {
|
|
135
|
+
const workspaceFiles = await scanWorkspaceFiles(cwd, config);
|
|
136
|
+
const loadResult = await loadManifest(storePath);
|
|
137
|
+
const storeCount = await store.count();
|
|
138
|
+
if (loadResult.status !== "ok") {
|
|
139
|
+
return {
|
|
140
|
+
manifestStatus: loadResult.status,
|
|
141
|
+
manifestEntries: 0,
|
|
142
|
+
upToDateFiles: 0,
|
|
143
|
+
pendingFiles: workspaceFiles.length,
|
|
144
|
+
rebuildRequired: storeCount > 0,
|
|
145
|
+
};
|
|
146
|
+
}
|
|
147
|
+
const manifest = loadResult.manifest;
|
|
148
|
+
const currentPaths = new Set(workspaceFiles.map((file) => file.normalizedPath));
|
|
149
|
+
let upToDateFiles = 0;
|
|
150
|
+
let pendingFiles = 0;
|
|
151
|
+
for (const file of workspaceFiles) {
|
|
152
|
+
const previous = manifest.files[file.normalizedPath];
|
|
153
|
+
if (file.isEmpty || file.isTooSmall) {
|
|
154
|
+
if (previous)
|
|
155
|
+
pendingFiles++;
|
|
156
|
+
continue;
|
|
157
|
+
}
|
|
158
|
+
if (previous && previous.hash === file.hash) {
|
|
159
|
+
upToDateFiles++;
|
|
160
|
+
}
|
|
161
|
+
else {
|
|
162
|
+
pendingFiles++;
|
|
163
|
+
}
|
|
164
|
+
}
|
|
165
|
+
for (const indexedPath of Object.keys(manifest.files)) {
|
|
166
|
+
if (!currentPaths.has(indexedPath)) {
|
|
167
|
+
pendingFiles++;
|
|
168
|
+
}
|
|
169
|
+
}
|
|
170
|
+
return {
|
|
171
|
+
manifestStatus: loadResult.status,
|
|
172
|
+
manifestEntries: Object.keys(manifest.files).length,
|
|
173
|
+
upToDateFiles,
|
|
174
|
+
pendingFiles,
|
|
175
|
+
lastIndexedAt: manifest.lastIndexedAt,
|
|
176
|
+
rebuildRequired: false,
|
|
177
|
+
};
|
|
178
|
+
}
|
|
179
|
+
//# sourceMappingURL=pipeline.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"pipeline.js","sourceRoot":"","sources":["../../src/indexer/pipeline.ts"],"names":[],"mappings":"AAAA,OAAO,MAAM,MAAM,SAAS,CAAC;AAC7B,OAAO,EAAE,kBAAkB,EAAE,MAAM,sBAAsB,CAAC;AAE1D,OAAO,EAAE,YAAY,EAAE,YAAY,EAAE,MAAM,qBAAqB,CAAC;AAOjE,OAAO,EAAE,gBAAgB,EAA+C,MAAM,YAAY,CAAC;AAC3F,OAAO,EAAE,WAAW,EAAqB,MAAM,aAAa,CAAC;AAG7D,OAAO,EAAE,wBAAwB,EAAE,iBAAiB,EAAE,MAAM,YAAY,CAAC;AAmBzE,SAAS,YAAY,CAAC,MAAwB;IAC5C,OAAO;QACL,IAAI,EAAE,MAAM,EAAE,IAAI,IAAI,CAAC,GAAG,EAAE,GAAE,CAAC,CAAC;QAChC,IAAI,EAAE,MAAM,EAAE,IAAI,IAAI,CAAC,GAAG,EAAE,GAAE,CAAC,CAAC;KACjC,CAAC;AACJ,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,YAAY,CAAC,OAA4B;IAC7D,MAAM,MAAM,GAAG,YAAY,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;IAC5C,MAAM,cAAc,GAAG,MAAM,kBAAkB,CAAC,OAAO,CAAC,GAAG,EAAE,OAAO,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;IACrF,MAAM,UAAU,GAAG,MAAM,YAAY,CAAC,OAAO,CAAC,SAAS,CAAC,CAAC;IACzD,MAAM,QAAQ,GAAG,UAAU,CAAC,QAAQ,CAAC;IACrC,IAAI,cAAc,GAAG,UAAU,CAAC,MAAM,CAAC;IACvC,IAAI,gBAAgB,GAAG,KAAK,CAAC;IAE7B,MAAM,aAAa,GAAG,MAAM,OAAO,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC;IAClD,IAAI,OAAO,CAAC,KAAK,IAAI,CAAC,cAAc,KAAK,IAAI,IAAI,aAAa,GAAG,CAAC,CAAC,EAAE,CAAC;QACpE,MAAM,OAAO,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC;QAC5B,OAAO,CAAC,YAAY,EAAE,KAAK,EAAE,CAAC;QAC9B,KAAK,MAAM,GAAG,IAAI,MAAM,CAAC,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;YAC9C,OAAO,QAAQ,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;QAC7B,CAAC;QACD,QAAQ,CAAC,aAAa,GAAG,SAAS,CAAC;QACnC,gBAAgB,GAAG,aAAa,GAAG,CAAC,IAAI,OAAO,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC;QAC/D,IAAI,cAAc,KAAK,IAAI,IAAI,aAAa,GAAG,CAAC,EAAE,CAAC;YACjD,MAAM,CAAC,IAAI,CAAC,qDAAqD,CAAC,CAAC;QACrE,CAAC;QACD,cAAc,GAAG,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,cAAc,CAAC;IAC9D,CAAC;IAED,MAAM,KAAK,GAAG,gBAAgB,CAAC,cAAc,CAAC,MAAM,EAAE,cAAc,CAAC,CAAC;IACtE,KAAK,CAAC,gBAAgB,GAAG,gBAAgB,CAAC;IAE1C,KAAK,MAAM,IAAI,IAAI,cAAc,EAAE,CAAC;QAClC,IAAI,IAAI,CAAC,gBAAgB,KAAK,QAAQ,IAAI,IAAI,CAAC,eAAe,EAAE,CAAC;YAC/D,KAAK,CAAC,kBAAkB,EAAE,CAAC;YAC3B,KAAK,CAAC,gBAAgB,CAAC,IAAI,CAAC;gBAC1B,QAAQ,EAAE,IAAI,CAAC,QAAQ;gBACvB,KAAK,EAAE,IAAI,CAAC,eAAe;aAC5B,CAAC,CAAC;QACL,CAAC;IACH,CAAC;IAED,MAAM,YAAY,GAAG,IAAI,GAAG,CAAC,cAAc,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC,CAAC;IAChF,KAAK,MAAM,WAAW,IAAI,MAAM,CAAC,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;QACtD,IAAI,CAAC,YAAY,CAAC,GAAG,CAAC,WAAW,CAAC,EAAE,CAAC;YACnC,MAAM,OAAO,CAAC,KAAK,CAAC,gBAAgB,CAAC,WAAW,CAAC,CAAC;YAClD,OAAO,CAAC,YAAY,EAAE,gBAAgB,CAAC,WAAW,CAAC,CAAC;YACpD,OAAO,QAAQ,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC;YACnC,KAAK,CAAC,YAAY,EAAE,CAAC;QACvB,CAAC;IACH,CAAC;IAED,IAAI,QAAQ,GAAG,CAAC,CAAC;IACjB,IAAI,aAAa,GAAG,CAAC,CAAC;IACtB,IAAI,cAAc,GAAG,CAAC,CAAC;IACvB,KAAK,MAAM,IAAI,IAAI,cAAc,EAAE,CAAC;QAClC,MAAM,QAAQ,GAAG,QAAQ,CAAC,KAAK,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC;QACrD,IAAI,IAAI,CAAC,OAAO,IAAI,IAAI,CAAC,UAAU;YAAE,SAAS;QAC9C,IAAI,QAAQ,IAAI,QAAQ,CAAC,IAAI,KAAK,IAAI,CAAC,IAAI,EAAE,CAAC;YAC5C,cAAc,EAAE,CAAC;QACnB,CAAC;aAAM,IAAI,QAAQ,EAAE,CAAC;YACpB,aAAa,EAAE,CAAC;QAClB,CAAC;aAAM,CAAC;YACN,QAAQ,EAAE,CAAC;QACb,CAAC;IACH,CAAC;IACD,MAAM,CAAC,IAAI,CAAC,cAAc,cAAc,CAAC,MAAM,WAAW,QAAQ,SAAS,aAAa,cAAc,cAAc,gBAAgB,CAAC,CAAC;IAEtI,MAAM,KAAK,GAAG,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,QAAQ,CAAC,WAAW,CAAC,CAAC;IAC1D,IAAI,SAAS,GAAG,CAAC,CAAC;IAClB,MAAM,KAAK,GAAG,cAAc,CAAC,MAAM,CAAC;IAEpC,MAAM,aAAa,GAAG,MAAM,OAAO,CAAC,GAAG,CACrC,cAAc,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAC1B,KAAK,CAAC,KAAK,IAAI,EAAE;QACf,MAAM,MAAM,GAAG,MAAM,WAAW,CAC9B,IAAI,EACJ,OAAO,CAAC,GAAG,EACX,QAAQ,CAAC,KAAK,CAAC,IAAI,CAAC,cAAc,CAAC,EACnC,OAAO,CAAC,MAAM,EACd,OAAO,CAAC,KAAK,EACb,OAAO,CAAC,YAAY,EACpB,OAAO,CAAC,QAAQ,EAChB,OAAO,CAAC,mBAAmB,EAC3B,MAAM,CACP,CAAC;QACF,SAAS,EAAE,CAAC;QACZ,IAAI,KAAK,GAAG,EAAE,IAAI,SAAS,GAAG,EAAE,KAAK,CAAC,EAAE,CAAC;YACvC,MAAM,CAAC,IAAI,CAAC,qBAAqB,SAAS,IAAI,KAAK,KAAK,CAAC,CAAC;QAC5D,CAAC;QACD,OAAO,MAAM,CAAC;IAChB,CAAC,CAAC,CACH,CACF,CAAC;IAEF,cAAc,CAAC,KAAK,EAAE,aAAa,EAAE,QAAQ,CAAC,CAAC;IAE/C,QAAQ,CAAC,aAAa,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;IACpC,MAAM,YAAY,CAAC,OAAO,CAAC,SAAS,EAAE,QAAQ,CAAC,CAAC;IAChD,MAAM,OAAO,CAAC,YAAY,EAAE,IAAI,CAAC,OAAO,CAAC,SAAS,CAAC,CAAC;IACpD,KAAK,CAAC,UAAU,GAAG,MAAM,OAAO,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC;IAC/C,OAAO,KAAK,CAAC;AACf,CAAC;AAED,SAAS,cAAc,CACrB,KAAoB,EACpB,OAAuB,EACvB,QAA6F;IAE7F,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;QAC7B,IAAI,MAAM,CAAC,OAAO,EAAE,CAAC;YACnB,KAAK,CAAC,iBAAiB,EAAE,CAAC;YAC1B,IAAI,MAAM,CAAC,SAAS,EAAE,CAAC;gBACrB,OAAO,QAAQ,CAAC,KAAK,CAAC,MAAM,CAAC,cAAc,CAAC,CAAC;gBAC7C,KAAK,CAAC,YAAY,EAAE,CAAC;YACvB,CAAC;YACD,SAAS;QACX,CAAC;QACD,IAAI,MAAM,CAAC,UAAU,EAAE,CAAC;YACtB,KAAK,CAAC,iBAAiB,EAAE,CAAC;YAC1B,IAAI,MAAM,CAAC,SAAS,EAAE,CAAC;gBACrB,OAAO,QAAQ,CAAC,KAAK,CAAC,MAAM,CAAC,cAAc,CAAC,CAAC;gBAC7C,KAAK,CAAC,YAAY,EAAE,CAAC;YACvB,CAAC;YACD,SAAS;QACX,CAAC;QACD,IAAI,MAAM,CAAC,WAAW,EAAE,CAAC;YACvB,KAAK,CAAC,cAAc,EAAE,CAAC;YACvB,SAAS;QACX,CAAC;QACD,IAAI,MAAM,CAAC,SAAS,EAAE,CAAC;YACrB,OAAO,QAAQ,CAAC,KAAK,CAAC,MAAM,CAAC,cAAc,CAAC,CAAC;YAC7C,KAAK,CAAC,YAAY,EAAE,CAAC;YACrB,SAAS;QACX,CAAC;QACD,IAAI,MAAM,CAAC,UAAU,EAAE,CAAC;YACtB,KAAK,CAAC,aAAa,EAAE,CAAC;QACxB,CAAC;aAAM,IAAI,MAAM,CAAC,KAAK,EAAE,CAAC;YACxB,KAAK,CAAC,QAAQ,EAAE,CAAC;QACnB,CAAC;QACD,IAAI,MAAM,CAAC,UAAU,GAAG,CAAC,EAAE,CAAC;YAC1B,QAAQ,CAAC,KAAK,CAAC,MAAM,CAAC,cAAc,CAAC,GAAG;gBACtC,IAAI,EAAE,MAAM,CAAC,IAAI;gBACjB,UAAU,EAAE,MAAM,CAAC,UAAU;gBAC7B,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE;aACtB,CAAC;YACF,KAAK,CAAC,WAAW,IAAI,MAAM,CAAC,UAAU,CAAC;YACvC,KAAK,CAAC,cAAc,EAAE,CAAC;QACzB,CAAC;IACH,CAAC;AACH,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,qBAAqB,CACzC,GAAW,EACX,SAAiB,EACjB,MAAiB,EACjB,KAAkB;IAElB,MAAM,cAAc,GAAG,MAAM,kBAAkB,CAAC,GAAG,EAAE,MAAM,CAAC,CAAC;IAC7D,MAAM,UAAU,GAAG,MAAM,YAAY,CAAC,SAAS,CAAC,CAAC;IACjD,MAAM,UAAU,GAAG,MAAM,KAAK,CAAC,KAAK,EAAE,CAAC;IAEvC,IAAI,UAAU,CAAC,MAAM,KAAK,IAAI,EAAE,CAAC;QAC/B,OAAO;YACL,cAAc,EAAE,UAAU,CAAC,MAAM;YACjC,eAAe,EAAE,CAAC;YAClB,aAAa,EAAE,CAAC;YAChB,YAAY,EAAE,cAAc,CAAC,MAAM;YACnC,eAAe,EAAE,UAAU,GAAG,CAAC;SAChC,CAAC;IACJ,CAAC;IAED,MAAM,QAAQ,GAAG,UAAU,CAAC,QAAQ,CAAC;IACrC,MAAM,YAAY,GAAG,IAAI,GAAG,CAAC,cAAc,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC,CAAC;IAChF,IAAI,aAAa,GAAG,CAAC,CAAC;IACtB,IAAI,YAAY,GAAG,CAAC,CAAC;IAErB,KAAK,MAAM,IAAI,IAAI,cAAc,EAAE,CAAC;QAClC,MAAM,QAAQ,GAAG,QAAQ,CAAC,KAAK,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC;QACrD,IAAI,IAAI,CAAC,OAAO,IAAI,IAAI,CAAC,UAAU,EAAE,CAAC;YACpC,IAAI,QAAQ;gBAAE,YAAY,EAAE,CAAC;YAC7B,SAAS;QACX,CAAC;QAED,IAAI,QAAQ,IAAI,QAAQ,CAAC,IAAI,KAAK,IAAI,CAAC,IAAI,EAAE,CAAC;YAC5C,aAAa,EAAE,CAAC;QAClB,CAAC;aAAM,CAAC;YACN,YAAY,EAAE,CAAC;QACjB,CAAC;IACH,CAAC;IAED,KAAK,MAAM,WAAW,IAAI,MAAM,CAAC,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;QACtD,IAAI,CAAC,YAAY,CAAC,GAAG,CAAC,WAAW,CAAC,EAAE,CAAC;YACnC,YAAY,EAAE,CAAC;QACjB,CAAC;IACH,CAAC;IAED,OAAO;QACL,cAAc,EAAE,UAAU,CAAC,MAAM;QACjC,eAAe,EAAE,MAAM,CAAC,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,MAAM;QACnD,aAAa;QACb,YAAY;QACZ,aAAa,EAAE,QAAQ,CAAC,aAAa;QACrC,eAAe,EAAE,KAAK;KACvB,CAAC;AACJ,CAAC"}
|
|
@@ -0,0 +1,29 @@
|
|
|
1
|
+
export interface IndexRunStats {
|
|
2
|
+
totalFiles: number;
|
|
3
|
+
newFiles: number;
|
|
4
|
+
modifiedFiles: number;
|
|
5
|
+
unchangedFiles: number;
|
|
6
|
+
deletedFiles: number;
|
|
7
|
+
removedFiles: number;
|
|
8
|
+
skippedEmptyFiles: number;
|
|
9
|
+
skippedSmallFiles: number;
|
|
10
|
+
totalChunks: number;
|
|
11
|
+
finalCount: number;
|
|
12
|
+
manifestStatus: "ok" | "missing" | "corrupt";
|
|
13
|
+
rebuildPerformed: boolean;
|
|
14
|
+
batchesFlushed: number;
|
|
15
|
+
extractionFailures: number;
|
|
16
|
+
extractionErrors: Array<{
|
|
17
|
+
filePath: string;
|
|
18
|
+
error: string;
|
|
19
|
+
}>;
|
|
20
|
+
}
|
|
21
|
+
export interface IndexStatusSummary {
|
|
22
|
+
manifestStatus: "ok" | "missing" | "corrupt";
|
|
23
|
+
manifestEntries: number;
|
|
24
|
+
upToDateFiles: number;
|
|
25
|
+
pendingFiles: number;
|
|
26
|
+
lastIndexedAt?: number;
|
|
27
|
+
rebuildRequired: boolean;
|
|
28
|
+
}
|
|
29
|
+
export declare function createIndexStats(totalFiles: number, manifestStatus: IndexRunStats["manifestStatus"]): IndexRunStats;
|
|
@@ -0,0 +1,20 @@
|
|
|
1
|
+
export function createIndexStats(totalFiles, manifestStatus) {
|
|
2
|
+
return {
|
|
3
|
+
totalFiles,
|
|
4
|
+
newFiles: 0,
|
|
5
|
+
modifiedFiles: 0,
|
|
6
|
+
unchangedFiles: 0,
|
|
7
|
+
deletedFiles: 0,
|
|
8
|
+
removedFiles: 0,
|
|
9
|
+
skippedEmptyFiles: 0,
|
|
10
|
+
skippedSmallFiles: 0,
|
|
11
|
+
totalChunks: 0,
|
|
12
|
+
finalCount: 0,
|
|
13
|
+
manifestStatus,
|
|
14
|
+
rebuildPerformed: false,
|
|
15
|
+
batchesFlushed: 0,
|
|
16
|
+
extractionFailures: 0,
|
|
17
|
+
extractionErrors: [],
|
|
18
|
+
};
|
|
19
|
+
}
|
|
20
|
+
//# sourceMappingURL=stats.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"stats.js","sourceRoot":"","sources":["../../src/indexer/stats.ts"],"names":[],"mappings":"AA2BA,MAAM,UAAU,gBAAgB,CAC9B,UAAkB,EAClB,cAA+C;IAE/C,OAAO;QACL,UAAU;QACV,QAAQ,EAAE,CAAC;QACX,aAAa,EAAE,CAAC;QAChB,cAAc,EAAE,CAAC;QACjB,YAAY,EAAE,CAAC;QACf,YAAY,EAAE,CAAC;QACf,iBAAiB,EAAE,CAAC;QACpB,iBAAiB,EAAE,CAAC;QACpB,WAAW,EAAE,CAAC;QACd,UAAU,EAAE,CAAC;QACb,cAAc;QACd,gBAAgB,EAAE,KAAK;QACvB,cAAc,EAAE,CAAC;QACjB,kBAAkB,EAAE,CAAC;QACrB,gBAAgB,EAAE,EAAE;KACrB,CAAC;AACJ,CAAC"}
|
|
@@ -0,0 +1,8 @@
|
|
|
1
|
+
import type { RagConfig } from "../core/config.js";
|
|
2
|
+
export interface WatchPassScheduler {
|
|
3
|
+
notifyChange(): void;
|
|
4
|
+
waitForIdle(): Promise<void>;
|
|
5
|
+
close(): void;
|
|
6
|
+
}
|
|
7
|
+
export declare function createWatchPassScheduler(runPass: () => Promise<void>, onError: (error: unknown) => void, debounceMs?: number): WatchPassScheduler;
|
|
8
|
+
export declare function createWatchIgnore(cwd: string, config: RagConfig, storePath: string): (watchedPath: string) => boolean;
|
|
@@ -0,0 +1,95 @@
|
|
|
1
|
+
import path from "node:path";
|
|
2
|
+
import { manifestPathFor } from "../core/manifest.js";
|
|
3
|
+
export function createWatchPassScheduler(runPass, onError, debounceMs = 300) {
|
|
4
|
+
let timer = null;
|
|
5
|
+
let running = false;
|
|
6
|
+
let rerunRequested = false;
|
|
7
|
+
let closed = false;
|
|
8
|
+
const waiters = [];
|
|
9
|
+
function resolveWaiters() {
|
|
10
|
+
if (running || timer || rerunRequested)
|
|
11
|
+
return;
|
|
12
|
+
while (waiters.length > 0) {
|
|
13
|
+
waiters.shift()?.();
|
|
14
|
+
}
|
|
15
|
+
}
|
|
16
|
+
function schedule() {
|
|
17
|
+
if (closed)
|
|
18
|
+
return;
|
|
19
|
+
if (timer)
|
|
20
|
+
clearTimeout(timer);
|
|
21
|
+
timer = setTimeout(() => {
|
|
22
|
+
timer = null;
|
|
23
|
+
void execute();
|
|
24
|
+
}, debounceMs);
|
|
25
|
+
}
|
|
26
|
+
async function execute() {
|
|
27
|
+
if (closed)
|
|
28
|
+
return;
|
|
29
|
+
if (running) {
|
|
30
|
+
rerunRequested = true;
|
|
31
|
+
return;
|
|
32
|
+
}
|
|
33
|
+
running = true;
|
|
34
|
+
try {
|
|
35
|
+
await runPass();
|
|
36
|
+
}
|
|
37
|
+
catch (error) {
|
|
38
|
+
onError(error);
|
|
39
|
+
}
|
|
40
|
+
finally {
|
|
41
|
+
running = false;
|
|
42
|
+
if (rerunRequested) {
|
|
43
|
+
rerunRequested = false;
|
|
44
|
+
schedule();
|
|
45
|
+
}
|
|
46
|
+
else {
|
|
47
|
+
resolveWaiters();
|
|
48
|
+
}
|
|
49
|
+
}
|
|
50
|
+
}
|
|
51
|
+
return {
|
|
52
|
+
notifyChange() {
|
|
53
|
+
if (closed)
|
|
54
|
+
return;
|
|
55
|
+
if (running) {
|
|
56
|
+
rerunRequested = true;
|
|
57
|
+
return;
|
|
58
|
+
}
|
|
59
|
+
schedule();
|
|
60
|
+
},
|
|
61
|
+
waitForIdle() {
|
|
62
|
+
if (!running && !timer && !rerunRequested) {
|
|
63
|
+
return Promise.resolve();
|
|
64
|
+
}
|
|
65
|
+
return new Promise((resolve) => {
|
|
66
|
+
waiters.push(resolve);
|
|
67
|
+
});
|
|
68
|
+
},
|
|
69
|
+
close() {
|
|
70
|
+
closed = true;
|
|
71
|
+
if (timer) {
|
|
72
|
+
clearTimeout(timer);
|
|
73
|
+
timer = null;
|
|
74
|
+
}
|
|
75
|
+
resolveWaiters();
|
|
76
|
+
},
|
|
77
|
+
};
|
|
78
|
+
}
|
|
79
|
+
export function createWatchIgnore(cwd, config, storePath) {
|
|
80
|
+
const manifestPath = manifestPathFor(storePath);
|
|
81
|
+
const excludeDirs = new Set(config.indexing.excludeDirs);
|
|
82
|
+
return (watchedPath) => {
|
|
83
|
+
const resolved = path.resolve(watchedPath);
|
|
84
|
+
if (resolved.startsWith(storePath))
|
|
85
|
+
return true;
|
|
86
|
+
if (resolved === manifestPath)
|
|
87
|
+
return true;
|
|
88
|
+
const relative = path.relative(cwd, resolved);
|
|
89
|
+
if (!relative || relative.startsWith(".."))
|
|
90
|
+
return false;
|
|
91
|
+
const segments = relative.split(path.sep);
|
|
92
|
+
return segments.some((segment) => excludeDirs.has(segment));
|
|
93
|
+
};
|
|
94
|
+
}
|
|
95
|
+
//# sourceMappingURL=watch.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"watch.js","sourceRoot":"","sources":["../../src/indexer/watch.ts"],"names":[],"mappings":"AAAA,OAAO,IAAI,MAAM,WAAW,CAAC;AAC7B,OAAO,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAStD,MAAM,UAAU,wBAAwB,CACtC,OAA4B,EAC5B,OAAiC,EACjC,aAAqB,GAAG;IAExB,IAAI,KAAK,GAA0B,IAAI,CAAC;IACxC,IAAI,OAAO,GAAG,KAAK,CAAC;IACpB,IAAI,cAAc,GAAG,KAAK,CAAC;IAC3B,IAAI,MAAM,GAAG,KAAK,CAAC;IACnB,MAAM,OAAO,GAAsB,EAAE,CAAC;IAEtC,SAAS,cAAc;QACrB,IAAI,OAAO,IAAI,KAAK,IAAI,cAAc;YAAE,OAAO;QAC/C,OAAO,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAC1B,OAAO,CAAC,KAAK,EAAE,EAAE,EAAE,CAAC;QACtB,CAAC;IACH,CAAC;IAED,SAAS,QAAQ;QACf,IAAI,MAAM;YAAE,OAAO;QACnB,IAAI,KAAK;YAAE,YAAY,CAAC,KAAK,CAAC,CAAC;QAC/B,KAAK,GAAG,UAAU,CAAC,GAAG,EAAE;YACtB,KAAK,GAAG,IAAI,CAAC;YACb,KAAK,OAAO,EAAE,CAAC;QACjB,CAAC,EAAE,UAAU,CAAC,CAAC;IACjB,CAAC;IAED,KAAK,UAAU,OAAO;QACpB,IAAI,MAAM;YAAE,OAAO;QACnB,IAAI,OAAO,EAAE,CAAC;YACZ,cAAc,GAAG,IAAI,CAAC;YACtB,OAAO;QACT,CAAC;QAED,OAAO,GAAG,IAAI,CAAC;QACf,IAAI,CAAC;YACH,MAAM,OAAO,EAAE,CAAC;QAClB,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACf,OAAO,CAAC,KAAK,CAAC,CAAC;QACjB,CAAC;gBAAS,CAAC;YACT,OAAO,GAAG,KAAK,CAAC;YAChB,IAAI,cAAc,EAAE,CAAC;gBACnB,cAAc,GAAG,KAAK,CAAC;gBACvB,QAAQ,EAAE,CAAC;YACb,CAAC;iBAAM,CAAC;gBACN,cAAc,EAAE,CAAC;YACnB,CAAC;QACH,CAAC;IACH,CAAC;IAED,OAAO;QACL,YAAY;YACV,IAAI,MAAM;gBAAE,OAAO;YACnB,IAAI,OAAO,EAAE,CAAC;gBACZ,cAAc,GAAG,IAAI,CAAC;gBACtB,OAAO;YACT,CAAC;YACD,QAAQ,EAAE,CAAC;QACb,CAAC;QACD,WAAW;YACT,IAAI,CAAC,OAAO,IAAI,CAAC,KAAK,IAAI,CAAC,cAAc,EAAE,CAAC;gBAC1C,OAAO,OAAO,CAAC,OAAO,EAAE,CAAC;YAC3B,CAAC;YACD,OAAO,IAAI,OAAO,CAAC,CAAC,OAAO,EAAE,EAAE;gBAC7B,OAAO,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;YACxB,CAAC,CAAC,CAAC;QACL,CAAC;QACD,KAAK;YACH,MAAM,GAAG,IAAI,CAAC;YACd,IAAI,KAAK,EAAE,CAAC;gBACV,YAAY,CAAC,KAAK,CAAC,CAAC;gBACpB,KAAK,GAAG,IAAI,CAAC;YACf,CAAC;YACD,cAAc,EAAE,CAAC;QACnB,CAAC;KACF,CAAC;AACJ,CAAC;AAED,MAAM,UAAU,iBAAiB,CAC/B,GAAW,EACX,MAAiB,EACjB,SAAiB;IAEjB,MAAM,YAAY,GAAG,eAAe,CAAC,SAAS,CAAC,CAAC;IAChD,MAAM,WAAW,GAAG,IAAI,GAAG,CAAC,MAAM,CAAC,QAAQ,CAAC,WAAW,CAAC,CAAC;IAEzD,OAAO,CAAC,WAAmB,EAAW,EAAE;QACtC,MAAM,QAAQ,GAAG,IAAI,CAAC,OAAO,CAAC,WAAW,CAAC,CAAC;QAC3C,IAAI,QAAQ,CAAC,UAAU,CAAC,SAAS,CAAC;YAAE,OAAO,IAAI,CAAC;QAChD,IAAI,QAAQ,KAAK,YAAY;YAAE,OAAO,IAAI,CAAC;QAE3C,MAAM,QAAQ,GAAG,IAAI,CAAC,QAAQ,CAAC,GAAG,EAAE,QAAQ,CAAC,CAAC;QAC9C,IAAI,CAAC,QAAQ,IAAI,QAAQ,CAAC,UAAU,CAAC,IAAI,CAAC;YAAE,OAAO,KAAK,CAAC;QACzD,MAAM,QAAQ,GAAG,QAAQ,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QAC1C,OAAO,QAAQ,CAAC,IAAI,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,WAAW,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC;IAC9D,CAAC,CAAC;AACJ,CAAC"}
|
|
@@ -0,0 +1,36 @@
|
|
|
1
|
+
import type { WorkspaceFile } from "../content/reader.js";
|
|
2
|
+
import type { DescriptionProvider, EmbeddingProvider, KeywordIndex, VectorStore } from "../core/interfaces.js";
|
|
3
|
+
export interface WorkerResult {
|
|
4
|
+
normalizedPath: string;
|
|
5
|
+
hash: string;
|
|
6
|
+
chunkCount: number;
|
|
7
|
+
fileLabel: string;
|
|
8
|
+
isNew: boolean;
|
|
9
|
+
isModified: boolean;
|
|
10
|
+
isUnchanged: boolean;
|
|
11
|
+
isEmpty: boolean;
|
|
12
|
+
isTooSmall: boolean;
|
|
13
|
+
isRemoved: boolean;
|
|
14
|
+
hadChunks: boolean;
|
|
15
|
+
}
|
|
16
|
+
interface Logger {
|
|
17
|
+
info(message: string): void;
|
|
18
|
+
warn(message: string): void;
|
|
19
|
+
}
|
|
20
|
+
interface ManifestFile {
|
|
21
|
+
hash: string;
|
|
22
|
+
chunkCount: number;
|
|
23
|
+
indexedAt?: number;
|
|
24
|
+
}
|
|
25
|
+
export declare function processFile(file: WorkspaceFile, cwd: string, previous: ManifestFile | undefined, config: {
|
|
26
|
+
embedding: {
|
|
27
|
+
documentPrefix?: string;
|
|
28
|
+
};
|
|
29
|
+
indexing: {
|
|
30
|
+
embedBatchSize: number;
|
|
31
|
+
};
|
|
32
|
+
chunking?: {
|
|
33
|
+
nodeTypes?: Record<string, string[]>;
|
|
34
|
+
};
|
|
35
|
+
}, store: VectorStore, keywordIndex: KeywordIndex | undefined, embedder: EmbeddingProvider, descriptionProvider: DescriptionProvider | undefined, logger: Logger): Promise<WorkerResult>;
|
|
36
|
+
export {};
|
|
@@ -0,0 +1,151 @@
|
|
|
1
|
+
import path from "node:path";
|
|
2
|
+
import { chunkFile } from "../chunker/factory.js";
|
|
3
|
+
import { uuid } from "../chunker/uuid.js";
|
|
4
|
+
import { isImageFile } from "../content/image.js";
|
|
5
|
+
import { buildFileMetadataHeader } from "./metadata.js";
|
|
6
|
+
import { generateDescriptions, buildFallbackDescription } from "./description-stage.js";
|
|
7
|
+
export async function processFile(file, cwd, previous, config, store, keywordIndex, embedder, descriptionProvider, logger) {
|
|
8
|
+
const fileLabel = path.relative(cwd, file.filePath);
|
|
9
|
+
if (file.isEmpty) {
|
|
10
|
+
if (previous) {
|
|
11
|
+
await store.deleteByFilePath(file.normalizedPath);
|
|
12
|
+
keywordIndex?.removeByFilePath(file.normalizedPath);
|
|
13
|
+
logger.info(` ${fileLabel} (empty, removed from index)`);
|
|
14
|
+
return { normalizedPath: file.normalizedPath, hash: file.hash, chunkCount: 0, fileLabel, isNew: false, isModified: false, isUnchanged: false, isEmpty: true, isTooSmall: false, isRemoved: true, hadChunks: false };
|
|
15
|
+
}
|
|
16
|
+
logger.info(` ${fileLabel} (empty, skipped)`);
|
|
17
|
+
return { normalizedPath: file.normalizedPath, hash: file.hash, chunkCount: 0, fileLabel, isNew: false, isModified: false, isUnchanged: false, isEmpty: true, isTooSmall: false, isRemoved: false, hadChunks: false };
|
|
18
|
+
}
|
|
19
|
+
if (file.isTooSmall) {
|
|
20
|
+
if (previous) {
|
|
21
|
+
await store.deleteByFilePath(file.normalizedPath);
|
|
22
|
+
keywordIndex?.removeByFilePath(file.normalizedPath);
|
|
23
|
+
logger.info(` ${fileLabel} (too small, removed from index)`);
|
|
24
|
+
return { normalizedPath: file.normalizedPath, hash: file.hash, chunkCount: 0, fileLabel, isNew: false, isModified: false, isUnchanged: false, isEmpty: false, isTooSmall: true, isRemoved: true, hadChunks: false };
|
|
25
|
+
}
|
|
26
|
+
logger.info(` ${fileLabel} (too small, skipped)`);
|
|
27
|
+
return { normalizedPath: file.normalizedPath, hash: file.hash, chunkCount: 0, fileLabel, isNew: false, isModified: false, isUnchanged: false, isEmpty: false, isTooSmall: true, isRemoved: false, hadChunks: false };
|
|
28
|
+
}
|
|
29
|
+
if (previous && previous.hash === file.hash) {
|
|
30
|
+
logger.info(` ${fileLabel} (unchanged)`);
|
|
31
|
+
return { normalizedPath: file.normalizedPath, hash: file.hash, chunkCount: 0, fileLabel, isNew: false, isModified: false, isUnchanged: true, isEmpty: false, isTooSmall: false, isRemoved: false, hadChunks: false };
|
|
32
|
+
}
|
|
33
|
+
let isModified = false;
|
|
34
|
+
if (previous) {
|
|
35
|
+
await store.deleteByFilePath(file.normalizedPath);
|
|
36
|
+
keywordIndex?.removeByFilePath(file.normalizedPath);
|
|
37
|
+
isModified = true;
|
|
38
|
+
}
|
|
39
|
+
let chunks;
|
|
40
|
+
if (isImageFile(file.filePath) && file.content.trim().length > 0) {
|
|
41
|
+
const imgExt = path.extname(file.filePath).toLowerCase();
|
|
42
|
+
const imgRelPath = path.relative(cwd, file.filePath).replace(/\\/g, "/");
|
|
43
|
+
const metaHeader = `[image] [${imgExt.slice(1)}] [${imgRelPath}]`;
|
|
44
|
+
chunks = [{
|
|
45
|
+
id: uuid(),
|
|
46
|
+
content: metaHeader + " " + file.content,
|
|
47
|
+
metadata: {
|
|
48
|
+
filePath: file.filePath,
|
|
49
|
+
startLine: 1,
|
|
50
|
+
endLine: 1,
|
|
51
|
+
language: "image",
|
|
52
|
+
contentType: "image",
|
|
53
|
+
},
|
|
54
|
+
}];
|
|
55
|
+
}
|
|
56
|
+
else {
|
|
57
|
+
chunks = await chunkFile(file.filePath, file.content, config.chunking?.nodeTypes).catch((err) => {
|
|
58
|
+
logger.warn(` ${fileLabel} (chunking failed: ${err.message})`);
|
|
59
|
+
return null;
|
|
60
|
+
});
|
|
61
|
+
}
|
|
62
|
+
if (chunks === null || chunks.length === 0) {
|
|
63
|
+
if (chunks === null) {
|
|
64
|
+
if (previous) {
|
|
65
|
+
return { normalizedPath: file.normalizedPath, hash: file.hash, chunkCount: previous.chunkCount, fileLabel, isNew: false, isModified: false, isUnchanged: true, isEmpty: false, isTooSmall: false, isRemoved: false, hadChunks: true };
|
|
66
|
+
}
|
|
67
|
+
return { normalizedPath: file.normalizedPath, hash: file.hash, chunkCount: 0, fileLabel, isNew: false, isModified: false, isUnchanged: false, isEmpty: false, isTooSmall: false, isRemoved: true, hadChunks: false };
|
|
68
|
+
}
|
|
69
|
+
logger.info(` ${fileLabel} (0 chunks, removed from index)`);
|
|
70
|
+
return { normalizedPath: file.normalizedPath, hash: file.hash, chunkCount: 0, fileLabel, isNew: false, isModified: false, isUnchanged: false, isEmpty: false, isTooSmall: false, isRemoved: true, hadChunks: false };
|
|
71
|
+
}
|
|
72
|
+
keywordIndex?.addChunks(chunks);
|
|
73
|
+
try {
|
|
74
|
+
const docPrefix = config.embedding.documentPrefix ?? "";
|
|
75
|
+
const relPath = path.relative(cwd, file.filePath).replace(/\\/g, "/");
|
|
76
|
+
const isImage = isImageFile(file.filePath);
|
|
77
|
+
const metaHeader = isImage
|
|
78
|
+
? ""
|
|
79
|
+
: buildFileMetadataHeader(file.filePath, cwd, file.content);
|
|
80
|
+
const textToEmbed = [];
|
|
81
|
+
if (descriptionProvider) {
|
|
82
|
+
const { descriptionMap } = await generateDescriptions(chunks, descriptionProvider, logger);
|
|
83
|
+
for (const chunk of chunks) {
|
|
84
|
+
if (isImage) {
|
|
85
|
+
textToEmbed.push(docPrefix + relPath + "\n\n" + chunk.description);
|
|
86
|
+
}
|
|
87
|
+
else {
|
|
88
|
+
const batchDesc = descriptionMap.get(chunk.id);
|
|
89
|
+
if (batchDesc && batchDesc.trim().length > 0) {
|
|
90
|
+
textToEmbed.push(docPrefix + relPath + "\n\n" + metaHeader + "\n\n" + batchDesc + "\n\n" + chunk.content);
|
|
91
|
+
}
|
|
92
|
+
else if (chunk.description) {
|
|
93
|
+
textToEmbed.push(docPrefix + relPath + "\n\n" + metaHeader + "\n\n" + chunk.description + "\n\n" + chunk.content);
|
|
94
|
+
}
|
|
95
|
+
else {
|
|
96
|
+
textToEmbed.push(docPrefix + relPath + "\n\n" + metaHeader + "\n\n" + chunk.content);
|
|
97
|
+
}
|
|
98
|
+
}
|
|
99
|
+
}
|
|
100
|
+
}
|
|
101
|
+
else {
|
|
102
|
+
for (const chunk of chunks) {
|
|
103
|
+
if (isImage) {
|
|
104
|
+
chunk.description = chunk.content;
|
|
105
|
+
textToEmbed.push(docPrefix + relPath + "\n\n" + chunk.description);
|
|
106
|
+
}
|
|
107
|
+
else {
|
|
108
|
+
chunk.description = buildFallbackDescription(chunk);
|
|
109
|
+
textToEmbed.push(docPrefix + relPath + "\n\n" + metaHeader + "\n\n" + chunk.description + "\n\n" + chunk.content);
|
|
110
|
+
}
|
|
111
|
+
}
|
|
112
|
+
}
|
|
113
|
+
const { embedBatch } = await import("../embedder/factory.js");
|
|
114
|
+
const embeddings = await embedBatch(embedder, textToEmbed, config.indexing.embedBatchSize, "document");
|
|
115
|
+
for (let i = 0; i < chunks.length; i++) {
|
|
116
|
+
const emb = embeddings[i];
|
|
117
|
+
if (Array.isArray(emb) && emb.length > 0 && typeof emb[0] === "number") {
|
|
118
|
+
chunks[i].embedding = emb;
|
|
119
|
+
}
|
|
120
|
+
else {
|
|
121
|
+
chunks[i].embedding = undefined;
|
|
122
|
+
}
|
|
123
|
+
}
|
|
124
|
+
const validChunks = chunks.filter((c) => c.embedding && c.embedding.length > 0);
|
|
125
|
+
if (validChunks.length > 0) {
|
|
126
|
+
await store.addChunks(validChunks);
|
|
127
|
+
}
|
|
128
|
+
logger.info(` ${fileLabel} (${chunks.length} chunks${isModified ? ", modified" : ", new"})`);
|
|
129
|
+
return {
|
|
130
|
+
normalizedPath: file.normalizedPath,
|
|
131
|
+
hash: file.hash,
|
|
132
|
+
chunkCount: chunks.length,
|
|
133
|
+
fileLabel,
|
|
134
|
+
isNew: !isModified && !previous,
|
|
135
|
+
isModified,
|
|
136
|
+
isUnchanged: false,
|
|
137
|
+
isEmpty: false,
|
|
138
|
+
isTooSmall: false,
|
|
139
|
+
isRemoved: false,
|
|
140
|
+
hadChunks: chunks.length > 0,
|
|
141
|
+
};
|
|
142
|
+
}
|
|
143
|
+
catch (err) {
|
|
144
|
+
logger.warn(` ${fileLabel} (embed/store failed: ${err.message})`);
|
|
145
|
+
if (previous) {
|
|
146
|
+
return { normalizedPath: file.normalizedPath, hash: file.hash, chunkCount: previous.chunkCount, fileLabel, isNew: false, isModified: false, isUnchanged: true, isEmpty: false, isTooSmall: false, isRemoved: false, hadChunks: true };
|
|
147
|
+
}
|
|
148
|
+
return { normalizedPath: file.normalizedPath, hash: file.hash, chunkCount: 0, fileLabel, isNew: false, isModified: false, isUnchanged: false, isEmpty: false, isTooSmall: false, isRemoved: true, hadChunks: false };
|
|
149
|
+
}
|
|
150
|
+
}
|
|
151
|
+
//# sourceMappingURL=worker.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"worker.js","sourceRoot":"","sources":["../../src/indexer/worker.ts"],"names":[],"mappings":"AAAA,OAAO,IAAI,MAAM,WAAW,CAAC;AAC7B,OAAO,EAAE,SAAS,EAAE,MAAM,uBAAuB,CAAC;AAClD,OAAO,EAAE,IAAI,EAAE,MAAM,oBAAoB,CAAC;AAC1C,OAAO,EAAE,WAAW,EAAE,MAAM,qBAAqB,CAAC;AAQlD,OAAO,EAAE,uBAAuB,EAAE,MAAM,eAAe,CAAC;AACxD,OAAO,EAAE,oBAAoB,EAAE,wBAAwB,EAAE,MAAM,wBAAwB,CAAC;AA2BxF,MAAM,CAAC,KAAK,UAAU,WAAW,CAC/B,IAAmB,EACnB,GAAW,EACX,QAAkC,EAClC,MAIC,EACD,KAAkB,EAClB,YAAsC,EACtC,QAA2B,EAC3B,mBAAoD,EACpD,MAAc;IAEd,MAAM,SAAS,GAAG,IAAI,CAAC,QAAQ,CAAC,GAAG,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAC;IAEpD,IAAI,IAAI,CAAC,OAAO,EAAE,CAAC;QACjB,IAAI,QAAQ,EAAE,CAAC;YACb,MAAM,KAAK,CAAC,gBAAgB,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC;YAClD,YAAY,EAAE,gBAAgB,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC;YACpD,MAAM,CAAC,IAAI,CAAC,KAAK,SAAS,8BAA8B,CAAC,CAAC;YAC1D,OAAO,EAAE,cAAc,EAAE,IAAI,CAAC,cAAc,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,UAAU,EAAE,CAAC,EAAE,SAAS,EAAE,KAAK,EAAE,KAAK,EAAE,UAAU,EAAE,KAAK,EAAE,WAAW,EAAE,KAAK,EAAE,OAAO,EAAE,IAAI,EAAE,UAAU,EAAE,KAAK,EAAE,SAAS,EAAE,IAAI,EAAE,SAAS,EAAE,KAAK,EAAE,CAAC;QACtN,CAAC;QACD,MAAM,CAAC,IAAI,CAAC,KAAK,SAAS,mBAAmB,CAAC,CAAC;QAC/C,OAAO,EAAE,cAAc,EAAE,IAAI,CAAC,cAAc,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,UAAU,EAAE,CAAC,EAAE,SAAS,EAAE,KAAK,EAAE,KAAK,EAAE,UAAU,EAAE,KAAK,EAAE,WAAW,EAAE,KAAK,EAAE,OAAO,EAAE,IAAI,EAAE,UAAU,EAAE,KAAK,EAAE,SAAS,EAAE,KAAK,EAAE,SAAS,EAAE,KAAK,EAAE,CAAC;IACvN,CAAC;IAED,IAAI,IAAI,CAAC,UAAU,EAAE,CAAC;QACpB,IAAI,QAAQ,EAAE,CAAC;YACb,MAAM,KAAK,CAAC,gBAAgB,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC;YAClD,YAAY,EAAE,gBAAgB,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC;YACpD,MAAM,CAAC,IAAI,CAAC,KAAK,SAAS,kCAAkC,CAAC,CAAC;YAC9D,OAAO,EAAE,cAAc,EAAE,IAAI,CAAC,cAAc,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,UAAU,EAAE,CAAC,EAAE,SAAS,EAAE,KAAK,EAAE,KAAK,EAAE,UAAU,EAAE,KAAK,EAAE,WAAW,EAAE,KAAK,EAAE,OAAO,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,SAAS,EAAE,IAAI,EAAE,SAAS,EAAE,KAAK,EAAE,CAAC;QACtN,CAAC;QACD,MAAM,CAAC,IAAI,CAAC,KAAK,SAAS,uBAAuB,CAAC,CAAC;QACnD,OAAO,EAAE,cAAc,EAAE,IAAI,CAAC,cAAc,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,UAAU,EAAE,CAAC,EAAE,SAAS,EAAE,KAAK,EAAE,KAAK,EAAE,UAAU,EAAE,KAAK,EAAE,WAAW,EAAE,KAAK,EAAE,OAAO,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,SAAS,EAAE,KAAK,EAAE,SAAS,EAAE,KAAK,EAAE,CAAC;IACvN,CAAC;IAED,IAAI,QAAQ,IAAI,QAAQ,CAAC,IAAI,KAAK,IAAI,CAAC,IAAI,EAAE,CAAC;QAC5C,MAAM,CAAC,IAAI,CAAC,KAAK,SAAS,cAAc,CAAC,CAAC;QAC1C,OAAO,EAAE,cAAc,EAAE,IAAI,CAAC,cAAc,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,UAAU,EAAE,CAAC,EAAE,SAAS,EAAE,KAAK,EAAE,KAAK,EAAE,UAAU,EAAE,KAAK,EAAE,WAAW,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,UAAU,EAAE,KAAK,EAAE,SAAS,EAAE,KAAK,EAAE,SAAS,EAAE,KAAK,EAAE,CAAC;IACvN,CAAC;IAED,IAAI,UAAU,GAAG,KAAK,CAAC;IACvB,IAAI,QAAQ,EAAE,CAAC;QACb,MAAM,KAAK,CAAC,gBAAgB,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC;QAClD,YAAY,EAAE,gBAAgB,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC;QACpD,UAAU,GAAG,IAAI,CAAC;IACpB,CAAC;IAED,IAAI,MAAM,CAAC;IAEX,IAAI,WAAW,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACjE,MAAM,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,WAAW,EAAE,CAAC;QACzD,MAAM,UAAU,GAAG,IAAI,CAAC,QAAQ,CAAC,GAAG,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAC,OAAO,CAAC,KAAK,EAAE,GAAG,CAAC,CAAC;QACzE,MAAM,UAAU,GAAG,YAAY,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,UAAU,GAAG,CAAC;QAClE,MAAM,GAAG,CAAC;gBACR,EAAE,EAAE,IAAI,EAAE;gBACV,OAAO,EAAE,UAAU,GAAG,GAAG,GAAG,IAAI,CAAC,OAAO;gBACxC,QAAQ,EAAE;oBACR,QAAQ,EAAE,IAAI,CAAC,QAAQ;oBACvB,SAAS,EAAE,CAAC;oBACZ,OAAO,EAAE,CAAC;oBACV,QAAQ,EAAE,OAAO;oBACjB,WAAW,EAAE,OAAO;iBACrB;aACF,CAAC,CAAC;IACL,CAAC;SAAM,CAAC;QACN,MAAM,GAAG,MAAM,SAAS,CAAC,IAAI,CAAC,QAAQ,EAAE,IAAI,CAAC,OAAO,EAAE,MAAM,CAAC,QAAQ,EAAE,SAAS,CAAC,CAAC,KAAK,CAAC,CAAC,GAAG,EAAE,EAAE;YAC9F,MAAM,CAAC,IAAI,CAAC,KAAK,SAAS,sBAAuB,GAAa,CAAC,OAAO,GAAG,CAAC,CAAC;YAC3E,OAAO,IAAI,CAAC;QACd,CAAC,CAAC,CAAC;IACL,CAAC;IAED,IAAI,MAAM,KAAK,IAAI,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC3C,IAAI,MAAM,KAAK,IAAI,EAAE,CAAC;YACpB,IAAI,QAAQ,EAAE,CAAC;gBACb,OAAO,EAAE,cAAc,EAAE,IAAI,CAAC,cAAc,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,UAAU,EAAE,QAAQ,CAAC,UAAU,EAAE,SAAS,EAAE,KAAK,EAAE,KAAK,EAAE,UAAU,EAAE,KAAK,EAAE,WAAW,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,UAAU,EAAE,KAAK,EAAE,SAAS,EAAE,KAAK,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC;YACxO,CAAC;YACD,OAAO,EAAE,cAAc,EAAE,IAAI,CAAC,cAAc,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,UAAU,EAAE,CAAC,EAAE,SAAS,EAAE,KAAK,EAAE,KAAK,EAAE,UAAU,EAAE,KAAK,EAAE,WAAW,EAAE,KAAK,EAAE,OAAO,EAAE,KAAK,EAAE,UAAU,EAAE,KAAK,EAAE,SAAS,EAAE,IAAI,EAAE,SAAS,EAAE,KAAK,EAAE,CAAC;QACvN,CAAC;QACD,MAAM,CAAC,IAAI,CAAC,KAAK,SAAS,iCAAiC,CAAC,CAAC;QAC7D,OAAO,EAAE,cAAc,EAAE,IAAI,CAAC,cAAc,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,UAAU,EAAE,CAAC,EAAE,SAAS,EAAE,KAAK,EAAE,KAAK,EAAE,UAAU,EAAE,KAAK,EAAE,WAAW,EAAE,KAAK,EAAE,OAAO,EAAE,KAAK,EAAE,UAAU,EAAE,KAAK,EAAE,SAAS,EAAE,IAAI,EAAE,SAAS,EAAE,KAAK,EAAE,CAAC;IACvN,CAAC;IAED,YAAY,EAAE,SAAS,CAAC,MAAM,CAAC,CAAC;IAEhC,IAAI,CAAC;QACH,MAAM,SAAS,GAAG,MAAM,CAAC,SAAS,CAAC,cAAc,IAAI,EAAE,CAAC;QACxD,MAAM,OAAO,GAAG,IAAI,CAAC,QAAQ,CAAC,GAAG,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAC,OAAO,CAAC,KAAK,EAAE,GAAG,CAAC,CAAC;QACtE,MAAM,OAAO,GAAG,WAAW,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;QAC3C,MAAM,UAAU,GAAG,OAAO;YACxB,CAAC,CAAC,EAAE;YACJ,CAAC,CAAC,uBAAuB,CAAC,IAAI,CAAC,QAAQ,EAAE,GAAG,EAAE,IAAI,CAAC,OAAO,CAAC,CAAC;QAC9D,MAAM,WAAW,GAAa,EAAE,CAAC;QAEjC,IAAI,mBAAmB,EAAE,CAAC;YACxB,MAAM,EAAE,cAAc,EAAE,GAAG,MAAM,oBAAoB,CAAC,MAAM,EAAE,mBAAmB,EAAE,MAAM,CAAC,CAAC;YAE3F,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;gBAC3B,IAAI,OAAO,EAAE,CAAC;oBACZ,WAAW,CAAC,IAAI,CAAC,SAAS,GAAG,OAAO,GAAG,MAAM,GAAG,KAAK,CAAC,WAAW,CAAC,CAAC;gBACrE,CAAC;qBAAM,CAAC;oBACN,MAAM,SAAS,GAAG,cAAc,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC;oBAC/C,IAAI,SAAS,IAAI,SAAS,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;wBAC7C,WAAW,CAAC,IAAI,CAAC,SAAS,GAAG,OAAO,GAAG,MAAM,GAAG,UAAU,GAAG,MAAM,GAAG,SAAS,GAAG,MAAM,GAAG,KAAK,CAAC,OAAO,CAAC,CAAC;oBAC5G,CAAC;yBAAM,IAAI,KAAK,CAAC,WAAW,EAAE,CAAC;wBAC7B,WAAW,CAAC,IAAI,CAAC,SAAS,GAAG,OAAO,GAAG,MAAM,GAAG,UAAU,GAAG,MAAM,GAAG,KAAK,CAAC,WAAW,GAAG,MAAM,GAAG,KAAK,CAAC,OAAO,CAAC,CAAC;oBACpH,CAAC;yBAAM,CAAC;wBACN,WAAW,CAAC,IAAI,CAAC,SAAS,GAAG,OAAO,GAAG,MAAM,GAAG,UAAU,GAAG,MAAM,GAAG,KAAK,CAAC,OAAO,CAAC,CAAC;oBACvF,CAAC;gBACH,CAAC;YACH,CAAC;QACH,CAAC;aAAM,CAAC;YACN,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;gBAC3B,IAAI,OAAO,EAAE,CAAC;oBACZ,KAAK,CAAC,WAAW,GAAG,KAAK,CAAC,OAAO,CAAC;oBAClC,WAAW,CAAC,IAAI,CAAC,SAAS,GAAG,OAAO,GAAG,MAAM,GAAG,KAAK,CAAC,WAAW,CAAC,CAAC;gBACrE,CAAC;qBAAM,CAAC;oBACN,KAAK,CAAC,WAAW,GAAG,wBAAwB,CAAC,KAAK,CAAC,CAAC;oBACpD,WAAW,CAAC,IAAI,CAAC,SAAS,GAAG,OAAO,GAAG,MAAM,GAAG,UAAU,GAAG,MAAM,GAAG,KAAK,CAAC,WAAW,GAAG,MAAM,GAAG,KAAK,CAAC,OAAO,CAAC,CAAC;gBACpH,CAAC;YACH,CAAC;QACH,CAAC;QAED,MAAM,EAAE,UAAU,EAAE,GAAG,MAAM,MAAM,CAAC,wBAAwB,CAAC,CAAC;QAC9D,MAAM,UAAU,GAAG,MAAM,UAAU,CAAC,QAAQ,EAAE,WAAW,EAAE,MAAM,CAAC,QAAQ,CAAC,cAAc,EAAE,UAAU,CAAC,CAAC;QAEvG,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;YACvC,MAAM,GAAG,GAAG,UAAU,CAAC,CAAC,CAAC,CAAC;YAC1B,IAAI,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,IAAI,GAAG,CAAC,MAAM,GAAG,CAAC,IAAI,OAAO,GAAG,CAAC,CAAC,CAAC,KAAK,QAAQ,EAAE,CAAC;gBACvE,MAAM,CAAC,CAAC,CAAE,CAAC,SAAS,GAAG,GAAe,CAAC;YACzC,CAAC;iBAAM,CAAC;gBACN,MAAM,CAAC,CAAC,CAAE,CAAC,SAAS,GAAG,SAAS,CAAC;YACnC,CAAC;QACH,CAAC;QAED,MAAM,WAAW,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,SAAS,IAAI,CAAC,CAAC,SAAS,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QAChF,IAAI,WAAW,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAC3B,MAAM,KAAK,CAAC,SAAS,CAAC,WAAW,CAAC,CAAC;QACrC,CAAC;QAED,MAAM,CAAC,IAAI,CAAC,KAAK,SAAS,KAAK,MAAM,CAAC,MAAM,UAAU,UAAU,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,OAAO,GAAG,CAAC,CAAC;QAE9F,OAAO;YACL,cAAc,EAAE,IAAI,CAAC,cAAc;YACnC,IAAI,EAAE,IAAI,CAAC,IAAI;YACf,UAAU,EAAE,MAAM,CAAC,MAAM;YACzB,SAAS;YACT,KAAK,EAAE,CAAC,UAAU,IAAI,CAAC,QAAQ;YAC/B,UAAU;YACV,WAAW,EAAE,KAAK;YAClB,OAAO,EAAE,KAAK;YACd,UAAU,EAAE,KAAK;YACjB,SAAS,EAAE,KAAK;YAChB,SAAS,EAAE,MAAM,CAAC,MAAM,GAAG,CAAC;SAC7B,CAAC;IACJ,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,MAAM,CAAC,IAAI,CAAC,KAAK,SAAS,yBAA0B,GAAa,CAAC,OAAO,GAAG,CAAC,CAAC;QAC9E,IAAI,QAAQ,EAAE,CAAC;YACb,OAAO,EAAE,cAAc,EAAE,IAAI,CAAC,cAAc,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,UAAU,EAAE,QAAQ,CAAC,UAAU,EAAE,SAAS,EAAE,KAAK,EAAE,KAAK,EAAE,UAAU,EAAE,KAAK,EAAE,WAAW,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,UAAU,EAAE,KAAK,EAAE,SAAS,EAAE,KAAK,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC;QACxO,CAAC;QACD,OAAO,EAAE,cAAc,EAAE,IAAI,CAAC,cAAc,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,UAAU,EAAE,CAAC,EAAE,SAAS,EAAE,KAAK,EAAE,KAAK,EAAE,UAAU,EAAE,KAAK,EAAE,WAAW,EAAE,KAAK,EAAE,OAAO,EAAE,KAAK,EAAE,UAAU,EAAE,KAAK,EAAE,SAAS,EAAE,IAAI,EAAE,SAAS,EAAE,KAAK,EAAE,CAAC;IACvN,CAAC;AACH,CAAC"}
|
package/dist/indexer.d.ts
CHANGED
|
@@ -1,60 +1,13 @@
|
|
|
1
|
+
export type { IndexRunStats, IndexStatusSummary } from "./indexer/stats.js";
|
|
2
|
+
export { createIndexStats } from "./indexer/stats.js";
|
|
3
|
+
export type { WorkspaceFile } from "./content/reader.js";
|
|
4
|
+
export { scanWorkspaceFiles, walkFiles } from "./content/reader.js";
|
|
5
|
+
export type { RunIndexPassOptions, WatchPassScheduler } from "./indexer/pipeline.js";
|
|
6
|
+
export { runIndexPass, getIndexStatusSummary, createWatchPassScheduler, createWatchIgnore, } from "./indexer/pipeline.js";
|
|
1
7
|
import type { RagConfig } from "./core/config.js";
|
|
2
|
-
import type {
|
|
3
|
-
export interface IndexRunStats {
|
|
4
|
-
totalFiles: number;
|
|
5
|
-
newFiles: number;
|
|
6
|
-
modifiedFiles: number;
|
|
7
|
-
unchangedFiles: number;
|
|
8
|
-
deletedFiles: number;
|
|
9
|
-
removedFiles: number;
|
|
10
|
-
skippedEmptyFiles: number;
|
|
11
|
-
skippedSmallFiles: number;
|
|
12
|
-
totalChunks: number;
|
|
13
|
-
finalCount: number;
|
|
14
|
-
manifestStatus: "ok" | "missing" | "corrupt";
|
|
15
|
-
rebuildPerformed: boolean;
|
|
16
|
-
batchesFlushed: number;
|
|
17
|
-
}
|
|
18
|
-
export interface IndexStatusSummary {
|
|
19
|
-
manifestStatus: "ok" | "missing" | "corrupt";
|
|
20
|
-
manifestEntries: number;
|
|
21
|
-
upToDateFiles: number;
|
|
22
|
-
pendingFiles: number;
|
|
23
|
-
lastIndexedAt?: number;
|
|
24
|
-
rebuildRequired: boolean;
|
|
25
|
-
}
|
|
26
|
-
export interface WorkspaceFile {
|
|
27
|
-
filePath: string;
|
|
28
|
-
normalizedPath: string;
|
|
29
|
-
content: string;
|
|
30
|
-
hash: string;
|
|
31
|
-
isEmpty: boolean;
|
|
32
|
-
isTooSmall: boolean;
|
|
33
|
-
}
|
|
8
|
+
import type { WorkspaceFile } from "./content/reader.js";
|
|
34
9
|
interface Logger {
|
|
35
10
|
info(message: string): void;
|
|
36
11
|
warn(message: string): void;
|
|
37
12
|
}
|
|
38
|
-
export interface RunIndexPassOptions {
|
|
39
|
-
cwd: string;
|
|
40
|
-
storePath: string;
|
|
41
|
-
config: RagConfig;
|
|
42
|
-
store: VectorStore;
|
|
43
|
-
embedder: EmbeddingProvider;
|
|
44
|
-
force?: boolean;
|
|
45
|
-
logger?: Partial<Logger>;
|
|
46
|
-
keywordIndex?: KeywordIndex;
|
|
47
|
-
descriptionProvider?: DescriptionProvider;
|
|
48
|
-
}
|
|
49
|
-
export interface WatchPassScheduler {
|
|
50
|
-
notifyChange(): void;
|
|
51
|
-
waitForIdle(): Promise<void>;
|
|
52
|
-
close(): void;
|
|
53
|
-
}
|
|
54
|
-
export declare function walkFiles(dir: string, extensions: Set<string>, excludeDirs: Set<string>): Promise<string[]>;
|
|
55
13
|
export declare function scanWorkspace(cwd: string, config: RagConfig, logger?: Logger): Promise<WorkspaceFile[]>;
|
|
56
|
-
export declare function runIndexPass(options: RunIndexPassOptions): Promise<IndexRunStats>;
|
|
57
|
-
export declare function getIndexStatusSummary(cwd: string, storePath: string, config: RagConfig, store: VectorStore): Promise<IndexStatusSummary>;
|
|
58
|
-
export declare function createWatchPassScheduler(runPass: () => Promise<void>, onError: (error: unknown) => void, debounceMs?: number): WatchPassScheduler;
|
|
59
|
-
export declare function createWatchIgnore(cwd: string, config: RagConfig, storePath: string): (watchedPath: string) => boolean;
|
|
60
|
-
export {};
|