@chnxthr/knowledge-mcp 0.0.0-stage → 0.3.1
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/LICENSE +21 -0
- package/README.md +37 -2
- package/dist/bootstrap.d.ts +1 -0
- package/dist/bootstrap.js +18 -0
- package/dist/bootstrap.js.map +1 -0
- package/dist/cli.d.ts +2 -0
- package/dist/cli.js +5 -0
- package/dist/cli.js.map +1 -0
- package/dist/core/auth.d.ts +38 -0
- package/dist/core/auth.js +93 -0
- package/dist/core/auth.js.map +1 -0
- package/dist/core/config.d.ts +259 -0
- package/dist/core/config.js +172 -0
- package/dist/core/config.js.map +1 -0
- package/dist/core/context.d.ts +25 -0
- package/dist/core/context.js +37 -0
- package/dist/core/context.js.map +1 -0
- package/dist/core/env.d.ts +2 -0
- package/dist/core/env.js +5 -0
- package/dist/core/env.js.map +1 -0
- package/dist/core/log_mask.d.ts +10 -0
- package/dist/core/log_mask.js +88 -0
- package/dist/core/log_mask.js.map +1 -0
- package/dist/core/oauth.d.ts +46 -0
- package/dist/core/oauth.js +507 -0
- package/dist/core/oauth.js.map +1 -0
- package/dist/core/paths.d.ts +5 -0
- package/dist/core/paths.js +47 -0
- package/dist/core/paths.js.map +1 -0
- package/dist/core/scope.d.ts +28 -0
- package/dist/core/scope.js +54 -0
- package/dist/core/scope.js.map +1 -0
- package/dist/core/upstream.d.ts +30 -0
- package/dist/core/upstream.js +164 -0
- package/dist/core/upstream.js.map +1 -0
- package/dist/index.d.ts +23 -0
- package/dist/index.js +21 -0
- package/dist/index.js.map +1 -0
- package/dist/plugin/define.d.ts +144 -0
- package/dist/plugin/define.js +44 -0
- package/dist/plugin/define.js.map +1 -0
- package/dist/plugin/load.d.ts +2 -0
- package/dist/plugin/load.js +41 -0
- package/dist/plugin/load.js.map +1 -0
- package/dist/plugin/ping.d.ts +9 -0
- package/dist/plugin/ping.js +29 -0
- package/dist/plugin/ping.js.map +1 -0
- package/dist/plugin/registry.d.ts +62 -0
- package/dist/plugin/registry.js +248 -0
- package/dist/plugin/registry.js.map +1 -0
- package/dist/plugin/result.d.ts +20 -0
- package/dist/plugin/result.js +59 -0
- package/dist/plugin/result.js.map +1 -0
- package/dist/plugins/drive/chunker.d.ts +28 -0
- package/dist/plugins/drive/chunker.js +356 -0
- package/dist/plugins/drive/chunker.js.map +1 -0
- package/dist/plugins/drive/clean_text.d.ts +13 -0
- package/dist/plugins/drive/clean_text.js +31 -0
- package/dist/plugins/drive/clean_text.js.map +1 -0
- package/dist/plugins/drive/client.d.ts +94 -0
- package/dist/plugins/drive/client.js +220 -0
- package/dist/plugins/drive/client.js.map +1 -0
- package/dist/plugins/drive/embeddings.d.ts +15 -0
- package/dist/plugins/drive/embeddings.js +118 -0
- package/dist/plugins/drive/embeddings.js.map +1 -0
- package/dist/plugins/drive/exclude.d.ts +11 -0
- package/dist/plugins/drive/exclude.js +36 -0
- package/dist/plugins/drive/exclude.js.map +1 -0
- package/dist/plugins/drive/extract_text.d.ts +26 -0
- package/dist/plugins/drive/extract_text.js +95 -0
- package/dist/plugins/drive/extract_text.js.map +1 -0
- package/dist/plugins/drive/get_token.d.ts +5 -0
- package/dist/plugins/drive/get_token.js +128 -0
- package/dist/plugins/drive/get_token.js.map +1 -0
- package/dist/plugins/drive/index.d.ts +5 -0
- package/dist/plugins/drive/index.js +72 -0
- package/dist/plugins/drive/index.js.map +1 -0
- package/dist/plugins/drive/indexer.d.ts +69 -0
- package/dist/plugins/drive/indexer.js +204 -0
- package/dist/plugins/drive/indexer.js.map +1 -0
- package/dist/plugins/drive/mathpix.d.ts +4 -0
- package/dist/plugins/drive/mathpix.js +102 -0
- package/dist/plugins/drive/mathpix.js.map +1 -0
- package/dist/plugins/drive/ooxml.d.ts +17 -0
- package/dist/plugins/drive/ooxml.js +152 -0
- package/dist/plugins/drive/ooxml.js.map +1 -0
- package/dist/plugins/drive/path.d.ts +40 -0
- package/dist/plugins/drive/path.js +87 -0
- package/dist/plugins/drive/path.js.map +1 -0
- package/dist/plugins/drive/qdrant.d.ts +123 -0
- package/dist/plugins/drive/qdrant.js +148 -0
- package/dist/plugins/drive/qdrant.js.map +1 -0
- package/dist/plugins/drive/reindex.d.ts +2 -0
- package/dist/plugins/drive/reindex.js +263 -0
- package/dist/plugins/drive/reindex.js.map +1 -0
- package/dist/plugins/drive/settings.d.ts +11 -0
- package/dist/plugins/drive/settings.js +13 -0
- package/dist/plugins/drive/settings.js.map +1 -0
- package/dist/plugins/drive/sync.d.ts +16 -0
- package/dist/plugins/drive/sync.js +139 -0
- package/dist/plugins/drive/sync.js.map +1 -0
- package/dist/plugins/drive/sync_state.d.ts +3 -0
- package/dist/plugins/drive/sync_state.js +26 -0
- package/dist/plugins/drive/sync_state.js.map +1 -0
- package/dist/plugins/drive/tools/list_drive_folder.d.ts +104 -0
- package/dist/plugins/drive/tools/list_drive_folder.js +221 -0
- package/dist/plugins/drive/tools/list_drive_folder.js.map +1 -0
- package/dist/plugins/drive/tools/read_drive_file.d.ts +66 -0
- package/dist/plugins/drive/tools/read_drive_file.js +206 -0
- package/dist/plugins/drive/tools/read_drive_file.js.map +1 -0
- package/dist/plugins/drive/tools/search_drive.d.ts +52 -0
- package/dist/plugins/drive/tools/search_drive.js +78 -0
- package/dist/plugins/drive/tools/search_drive.js.map +1 -0
- package/dist/plugins/drive/tools/semantic_search_drive.d.ts +109 -0
- package/dist/plugins/drive/tools/semantic_search_drive.js +228 -0
- package/dist/plugins/drive/tools/semantic_search_drive.js.map +1 -0
- package/dist/plugins/github/index.d.ts +1 -0
- package/dist/plugins/github/index.js +27 -0
- package/dist/plugins/github/index.js.map +1 -0
- package/dist/plugins/github/repo_tree.d.ts +57 -0
- package/dist/plugins/github/repo_tree.js +123 -0
- package/dist/plugins/github/repo_tree.js.map +1 -0
- package/dist/plugins/index.d.ts +11 -0
- package/dist/plugins/index.js +7 -0
- package/dist/plugins/index.js.map +1 -0
- package/dist/plugins/railway/index.d.ts +1 -0
- package/dist/plugins/railway/index.js +21 -0
- package/dist/plugins/railway/index.js.map +1 -0
- package/dist/plugins/railway/tools.d.ts +114 -0
- package/dist/plugins/railway/tools.js +165 -0
- package/dist/plugins/railway/tools.js.map +1 -0
- package/dist/plugins/slack/api.d.ts +63 -0
- package/dist/plugins/slack/api.js +280 -0
- package/dist/plugins/slack/api.js.map +1 -0
- package/dist/plugins/slack/index.d.ts +1 -0
- package/dist/plugins/slack/index.js +80 -0
- package/dist/plugins/slack/index.js.map +1 -0
- package/dist/plugins/slack/tools/slack_files.d.ts +87 -0
- package/dist/plugins/slack/tools/slack_files.js +204 -0
- package/dist/plugins/slack/tools/slack_files.js.map +1 -0
- package/dist/plugins/slack/tools/slack_list_channels.d.ts +46 -0
- package/dist/plugins/slack/tools/slack_list_channels.js +197 -0
- package/dist/plugins/slack/tools/slack_list_channels.js.map +1 -0
- package/dist/plugins/slack/tools/slack_read_channel.d.ts +56 -0
- package/dist/plugins/slack/tools/slack_read_channel.js +150 -0
- package/dist/plugins/slack/tools/slack_read_channel.js.map +1 -0
- package/dist/plugins/slack/tools/slack_search.d.ts +53 -0
- package/dist/plugins/slack/tools/slack_search.js +132 -0
- package/dist/plugins/slack/tools/slack_search.js.map +1 -0
- package/dist/run.d.ts +8 -0
- package/dist/run.js +270 -0
- package/dist/run.js.map +1 -0
- package/dist/server/http.d.ts +5 -0
- package/dist/server/http.js +406 -0
- package/dist/server/http.js.map +1 -0
- package/dist/server/identity.d.ts +17 -0
- package/dist/server/identity.js +47 -0
- package/dist/server/identity.js.map +1 -0
- package/dist/server/mcp.d.ts +7 -0
- package/dist/server/mcp.js +55 -0
- package/dist/server/mcp.js.map +1 -0
- package/dist/server/stdio.d.ts +2 -0
- package/dist/server/stdio.js +40 -0
- package/dist/server/stdio.js.map +1 -0
- package/dist/store/index.d.ts +11 -0
- package/dist/store/index.js +31 -0
- package/dist/store/index.js.map +1 -0
- package/dist/store/keys.d.ts +9 -0
- package/dist/store/keys.js +32 -0
- package/dist/store/keys.js.map +1 -0
- package/dist/store/sqlite.d.ts +38 -0
- package/dist/store/sqlite.js +316 -0
- package/dist/store/sqlite.js.map +1 -0
- package/dist/store/supabase.d.ts +35 -0
- package/dist/store/supabase.js +269 -0
- package/dist/store/supabase.js.map +1 -0
- package/dist/store/supabase_client.d.ts +10 -0
- package/dist/store/supabase_client.js +31 -0
- package/dist/store/supabase_client.js.map +1 -0
- package/dist/store/tool_log.d.ts +12 -0
- package/dist/store/tool_log.js +8 -0
- package/dist/store/tool_log.js.map +1 -0
- package/dist/store/types.d.ts +96 -0
- package/dist/store/types.js +2 -0
- package/dist/store/types.js.map +1 -0
- package/dist/version.d.ts +1 -0
- package/dist/version.js +11 -0
- package/dist/version.js.map +1 -0
- package/package.json +80 -4
- package/sql/supabase.sql +185 -0
|
@@ -0,0 +1,263 @@
|
|
|
1
|
+
import { getDriveClient } from './client.js';
|
|
2
|
+
import { chunkMarkdown } from './chunker.js';
|
|
3
|
+
import { embedTexts } from './embeddings.js';
|
|
4
|
+
import { extractText } from './extract_text.js';
|
|
5
|
+
import { isExcludedName } from './exclude.js';
|
|
6
|
+
import { ensureCollection, getQdrantClient, DRIVE_COLLECTION, } from './qdrant.js';
|
|
7
|
+
// 색인 코어 로직 SSOT — drive_sync(증분)와 공유해 drift 방지(특히 makeChunkId 일치 = orphan 로직 정합).
|
|
8
|
+
import { makeChunkId, classifyFile, contentHash, buildChunkPayload, deriveBizMeta, embedInputForChunk, SUPPORTED_MIMES, DRIVE_ROOTS, MAX_FILE_BYTES, } from './indexer.js';
|
|
9
|
+
import { existsSync, mkdirSync, readFileSync, writeFileSync, unlinkSync } from 'fs';
|
|
10
|
+
import { join } from 'path';
|
|
11
|
+
import { dataDir } from '../../core/paths.js';
|
|
12
|
+
// 체크포인트 — 처리 완료 file_id 기록 → 재시작 시 skip (429 중단 후 재개·재실행 비용 절감).
|
|
13
|
+
// 데이터 폴더에 둔다(패키지 폴더는 npm 설치본이라 쓰기 불가일 수 있다).
|
|
14
|
+
const CHECKPOINT_PATH = join(dataDir(), 'drive-reindex-checkpoint.json');
|
|
15
|
+
function loadCheckpoint() {
|
|
16
|
+
if (!existsSync(CHECKPOINT_PATH))
|
|
17
|
+
return new Set();
|
|
18
|
+
try {
|
|
19
|
+
return new Set(JSON.parse(readFileSync(CHECKPOINT_PATH, 'utf8')));
|
|
20
|
+
}
|
|
21
|
+
catch {
|
|
22
|
+
return new Set();
|
|
23
|
+
}
|
|
24
|
+
}
|
|
25
|
+
function saveCheckpoint(done) {
|
|
26
|
+
mkdirSync(dataDir(), { recursive: true, mode: 0o700 });
|
|
27
|
+
writeFileSync(CHECKPOINT_PATH, JSON.stringify([...done]));
|
|
28
|
+
}
|
|
29
|
+
function clearCheckpoint() {
|
|
30
|
+
if (existsSync(CHECKPOINT_PATH))
|
|
31
|
+
unlinkSync(CHECKPOINT_PATH);
|
|
32
|
+
}
|
|
33
|
+
function parseArgs(args) {
|
|
34
|
+
let roots = DRIVE_ROOTS;
|
|
35
|
+
let dryRun = false;
|
|
36
|
+
let recreate = false;
|
|
37
|
+
let partial = false;
|
|
38
|
+
for (let i = 0; i < args.length; i++) {
|
|
39
|
+
if (args[i] === '--root' && args[i + 1]) {
|
|
40
|
+
// override default with single root (source=dev 가정)
|
|
41
|
+
roots = [{ id: args[++i], name: 'custom', source: 'dev' }];
|
|
42
|
+
partial = true;
|
|
43
|
+
}
|
|
44
|
+
else if (args[i] === '--biz-only') {
|
|
45
|
+
roots = DRIVE_ROOTS.filter((r) => r.source === 'biz');
|
|
46
|
+
partial = true;
|
|
47
|
+
}
|
|
48
|
+
else if (args[i] === '--dev-only') {
|
|
49
|
+
roots = DRIVE_ROOTS.filter((r) => r.source === 'dev');
|
|
50
|
+
partial = true;
|
|
51
|
+
}
|
|
52
|
+
else if (args[i] === '--dry-run') {
|
|
53
|
+
dryRun = true;
|
|
54
|
+
}
|
|
55
|
+
else if (args[i] === '--recreate') {
|
|
56
|
+
// collection 비우고 새로 (orphan chunk 제거 — 정제·chunk 수 변동 반영)
|
|
57
|
+
recreate = true;
|
|
58
|
+
}
|
|
59
|
+
}
|
|
60
|
+
return { roots, dryRun, recreate, partial };
|
|
61
|
+
}
|
|
62
|
+
/**
|
|
63
|
+
* 폴더 ID 의 지원 형식 파일 recursive list (deprecated 제외).
|
|
64
|
+
*/
|
|
65
|
+
async function listAllMarkdown(drive, folderId, pathPrefix) {
|
|
66
|
+
const out = [];
|
|
67
|
+
const subfolderTasks = []; // 하위 폴더 recursion 을 병렬로
|
|
68
|
+
let pageToken;
|
|
69
|
+
do {
|
|
70
|
+
const resp = await drive.files.list({
|
|
71
|
+
q: `'${folderId}' in parents and trashed=false`,
|
|
72
|
+
fields: 'nextPageToken, files(id,name,mimeType,modifiedTime,size)',
|
|
73
|
+
pageSize: 100,
|
|
74
|
+
pageToken,
|
|
75
|
+
includeItemsFromAllDrives: true,
|
|
76
|
+
supportsAllDrives: true,
|
|
77
|
+
});
|
|
78
|
+
for (const f of resp.data.files ?? []) {
|
|
79
|
+
if (!f.id || !f.name)
|
|
80
|
+
continue;
|
|
81
|
+
const currentPath = pathPrefix ? `${pathPrefix}/${f.name}` : f.name;
|
|
82
|
+
if (f.mimeType === 'application/vnd.google-apps.folder') {
|
|
83
|
+
// 제외 폴더 (민감 PII/계약/HR + 개발 아티팩트. SSOT: exclude.ts) skip — recursion 자체 안 함
|
|
84
|
+
if (isExcludedName(f.name))
|
|
85
|
+
continue;
|
|
86
|
+
// gather as parallel tasks instead of sequential await (hundreds of folders: minutes -> seconds)
|
|
87
|
+
subfolderTasks.push(listAllMarkdown(drive, f.id, currentPath));
|
|
88
|
+
}
|
|
89
|
+
else if (f.mimeType && SUPPORTED_MIMES.has(f.mimeType)) {
|
|
90
|
+
out.push({
|
|
91
|
+
id: f.id,
|
|
92
|
+
name: f.name,
|
|
93
|
+
path: currentPath,
|
|
94
|
+
modifiedTime: f.modifiedTime ?? '',
|
|
95
|
+
mimeType: f.mimeType,
|
|
96
|
+
size: f.size ? Number(f.size) : undefined,
|
|
97
|
+
});
|
|
98
|
+
}
|
|
99
|
+
}
|
|
100
|
+
pageToken = resp.data.nextPageToken ?? undefined;
|
|
101
|
+
} while (pageToken);
|
|
102
|
+
// 하위 폴더 병렬 recursion 결과 합치기
|
|
103
|
+
const childResults = await Promise.all(subfolderTasks);
|
|
104
|
+
for (const child of childResults)
|
|
105
|
+
out.push(...child);
|
|
106
|
+
return out;
|
|
107
|
+
}
|
|
108
|
+
async function downloadMarkdown(drive, fileId) {
|
|
109
|
+
const resp = await drive.files.get({ fileId, alt: 'media', supportsAllDrives: true }, { responseType: 'text' });
|
|
110
|
+
return resp.data;
|
|
111
|
+
}
|
|
112
|
+
/**
|
|
113
|
+
* 이번 목록(keep)에 없는 파일의 벡터를 지운다. 반환 = 지운 파일 수.
|
|
114
|
+
* 목록 단계가 실패하면 main 이 그 전에 FATAL 로 끝나므로, 여기 도달했다면 목록은 완전하다.
|
|
115
|
+
*/
|
|
116
|
+
async function pruneStaleVectors(keep) {
|
|
117
|
+
const qdrant = await getQdrantClient();
|
|
118
|
+
const stale = new Set();
|
|
119
|
+
let offset = undefined;
|
|
120
|
+
do {
|
|
121
|
+
const page = await qdrant.scroll(DRIVE_COLLECTION, {
|
|
122
|
+
limit: 1000,
|
|
123
|
+
offset: offset ?? undefined,
|
|
124
|
+
with_payload: ['file_id'],
|
|
125
|
+
with_vector: false,
|
|
126
|
+
});
|
|
127
|
+
for (const pt of page.points) {
|
|
128
|
+
const fid = pt.payload?.file_id;
|
|
129
|
+
if (typeof fid === 'string' && !keep.has(fid))
|
|
130
|
+
stale.add(fid);
|
|
131
|
+
}
|
|
132
|
+
offset = page.next_page_offset ?? null;
|
|
133
|
+
} while (offset !== null);
|
|
134
|
+
for (const fid of stale) {
|
|
135
|
+
await qdrant.delete(DRIVE_COLLECTION, {
|
|
136
|
+
wait: true,
|
|
137
|
+
filter: { must: [{ key: 'file_id', match: { value: fid } }] },
|
|
138
|
+
});
|
|
139
|
+
}
|
|
140
|
+
return stale.size;
|
|
141
|
+
}
|
|
142
|
+
/** `knowledge-mcp drive reindex [--root ID | --dev-only | --biz-only] [--dry-run] [--recreate]` */
|
|
143
|
+
export async function runReindex(argv) {
|
|
144
|
+
const { roots, dryRun, recreate, partial } = parseArgs(argv);
|
|
145
|
+
const startTs = Date.now();
|
|
146
|
+
console.log(`[index] start — roots=${roots.map((r) => r.name).join(',')} dryRun=${dryRun} recreate=${recreate}`);
|
|
147
|
+
// 1. Qdrant collection 보장 (--recreate 시 기존 비우고 새로)
|
|
148
|
+
if (!dryRun) {
|
|
149
|
+
if (recreate) {
|
|
150
|
+
try {
|
|
151
|
+
await (await getQdrantClient()).deleteCollection(DRIVE_COLLECTION);
|
|
152
|
+
console.log(`[index] collection '${DRIVE_COLLECTION}' 삭제 (recreate)`);
|
|
153
|
+
}
|
|
154
|
+
catch {
|
|
155
|
+
/* 없으면 무시 */
|
|
156
|
+
}
|
|
157
|
+
}
|
|
158
|
+
await ensureCollection();
|
|
159
|
+
console.log(`[index] collection '${DRIVE_COLLECTION}' ready`);
|
|
160
|
+
}
|
|
161
|
+
// 2. Drive 모든 markdown list (root 별)
|
|
162
|
+
const drive = await getDriveClient();
|
|
163
|
+
const allFiles = [];
|
|
164
|
+
for (const root of roots) {
|
|
165
|
+
const files = await listAllMarkdown(drive, root.id, '');
|
|
166
|
+
console.log(`[index] '${root.name}' (${root.source}) — ${files.length} files`);
|
|
167
|
+
for (const f of files)
|
|
168
|
+
allFiles.push({ ...f, source: root.source });
|
|
169
|
+
}
|
|
170
|
+
console.log(`[index] total ${allFiles.length} files`);
|
|
171
|
+
// 3. 파일별 download + chunk + embed + upsert
|
|
172
|
+
// 체크포인트 — recreate 면 초기화, 아니면 이전 진행분 skip (429 중단 후 재개)
|
|
173
|
+
if (recreate)
|
|
174
|
+
clearCheckpoint();
|
|
175
|
+
const done = dryRun ? new Set() : loadCheckpoint();
|
|
176
|
+
if (done.size > 0)
|
|
177
|
+
console.log(`[index] 체크포인트 — 이미 처리 ${done.size}개 skip`);
|
|
178
|
+
let totalChunks = 0;
|
|
179
|
+
let totalTokens = 0;
|
|
180
|
+
let processed = 0;
|
|
181
|
+
for (const file of allFiles) {
|
|
182
|
+
processed++;
|
|
183
|
+
if (done.has(file.id))
|
|
184
|
+
continue; // 체크포인트 — 이미 처리됨
|
|
185
|
+
if (file.size && file.size > MAX_FILE_BYTES) {
|
|
186
|
+
console.log(`[index] skip (too large: ${file.size}B > ${MAX_FILE_BYTES}B): [${file.source}] ${file.path}`);
|
|
187
|
+
done.add(file.id);
|
|
188
|
+
if (!dryRun)
|
|
189
|
+
saveCheckpoint(done);
|
|
190
|
+
continue;
|
|
191
|
+
}
|
|
192
|
+
try {
|
|
193
|
+
const { text, needsOcr } = await extractText(drive, file.id, file.mimeType);
|
|
194
|
+
if (needsOcr || text.trim().length === 0) {
|
|
195
|
+
console.log(`[index] skip (ocr/empty): [${file.source}] ${file.path}`);
|
|
196
|
+
done.add(file.id);
|
|
197
|
+
if (!dryRun)
|
|
198
|
+
saveCheckpoint(done);
|
|
199
|
+
continue;
|
|
200
|
+
}
|
|
201
|
+
const chunks = chunkMarkdown(text);
|
|
202
|
+
if (chunks.length === 0) {
|
|
203
|
+
console.log(`[index] skip empty: [${file.source}] ${file.path}`);
|
|
204
|
+
// 처리 완료로 기록 — 빠뜨리면 체크포인트가 영구히 남아 다음 재색인이 바뀐 파일을 건너뛴다.
|
|
205
|
+
done.add(file.id);
|
|
206
|
+
if (!dryRun)
|
|
207
|
+
saveCheckpoint(done);
|
|
208
|
+
continue;
|
|
209
|
+
}
|
|
210
|
+
const classification = classifyFile(file.path, file.source);
|
|
211
|
+
const bizMeta = deriveBizMeta(file.path, file.source);
|
|
212
|
+
const hash = contentHash(text); // 무변경 skip(EFF-1) 판정용 — payload 동봉 (SSOT: drive_indexer)
|
|
213
|
+
const fileTokens = chunks.reduce((s, c) => s + c.tokens, 0);
|
|
214
|
+
totalChunks += chunks.length;
|
|
215
|
+
totalTokens += fileTokens;
|
|
216
|
+
if (dryRun) {
|
|
217
|
+
console.log(`[index] [${processed}/${allFiles.length}] [${file.source}] ${file.path} — ${chunks.length} chunks, ${fileTokens} tok`);
|
|
218
|
+
continue;
|
|
219
|
+
}
|
|
220
|
+
// Embed — heading breadcrumb 포함(embedInputForChunk). 증분 경로(drive_indexer.indexFile)와 동일 입력.
|
|
221
|
+
const vectors = await embedTexts(chunks.map(embedInputForChunk));
|
|
222
|
+
// Qdrant upsert — payload 구성은 buildChunkPayload SSOT (drive_indexer) 공유
|
|
223
|
+
const points = chunks.map((c, idx) => ({
|
|
224
|
+
id: makeChunkId(file.id, idx),
|
|
225
|
+
vector: vectors[idx],
|
|
226
|
+
payload: buildChunkPayload(c, file, idx, chunks.length, classification, bizMeta, hash),
|
|
227
|
+
}));
|
|
228
|
+
const qdrant = await getQdrantClient();
|
|
229
|
+
// B-1: 이 파일의 기존 청크를 전부 제거한 뒤 새로 upsert (줄어든 경우 유령 청크 잔류 방지).
|
|
230
|
+
// 검증된 match 필터만 사용(range 필터는 비인덱싱 payload 에서 400). --recreate 면 매칭 0건(no-op).
|
|
231
|
+
// 풀 reindex 는 배치 작업이라 delete~upsert 사이의 짧은 공백 허용.
|
|
232
|
+
await qdrant.delete(DRIVE_COLLECTION, {
|
|
233
|
+
wait: false,
|
|
234
|
+
filter: { must: [{ key: 'file_id', match: { value: file.id } }] },
|
|
235
|
+
});
|
|
236
|
+
await qdrant.upsert(DRIVE_COLLECTION, {
|
|
237
|
+
wait: false,
|
|
238
|
+
points,
|
|
239
|
+
});
|
|
240
|
+
done.add(file.id); // 체크포인트 기록 (upsert 성공)
|
|
241
|
+
saveCheckpoint(done);
|
|
242
|
+
console.log(`[index] [${processed}/${allFiles.length}] [${file.source}] ${file.path} — ${chunks.length} chunks upserted (${fileTokens} tok)`);
|
|
243
|
+
}
|
|
244
|
+
catch (err) {
|
|
245
|
+
// 실패 파일은 체크포인트에 기록 안 함 → 재실행 시 재시도
|
|
246
|
+
console.error(`[index] ERROR [${file.source}] ${file.path}:`, err instanceof Error ? err.message : err);
|
|
247
|
+
}
|
|
248
|
+
}
|
|
249
|
+
// 4. 이번 목록에 없는 파일의 벡터 정리 (전체 root 실행일 때만)
|
|
250
|
+
if (!dryRun && !partial) {
|
|
251
|
+
const pruned = await pruneStaleVectors(new Set(allFiles.map((f) => f.id)));
|
|
252
|
+
console.log(`[index] 정리 — 목록에 없는 파일 ${pruned}개의 벡터 삭제`);
|
|
253
|
+
}
|
|
254
|
+
const elapsedSec = ((Date.now() - startTs) / 1000).toFixed(1);
|
|
255
|
+
// OpenAI text-embedding-3-small: $0.02/1M tokens.
|
|
256
|
+
const costUsd = ((totalTokens / 1_000_000) * 0.02).toFixed(4);
|
|
257
|
+
console.log(`\n[index] DONE — ${totalChunks} chunks, ${totalTokens} tokens, ` +
|
|
258
|
+
`~$${costUsd} (OpenAI text-embedding-3-small), ${elapsedSec}s`);
|
|
259
|
+
// 전부 성공적으로 끝났으면 체크포인트 정리 (실패가 남아있으면 보존)
|
|
260
|
+
if (!dryRun && done.size >= allFiles.length)
|
|
261
|
+
clearCheckpoint();
|
|
262
|
+
}
|
|
263
|
+
//# sourceMappingURL=reindex.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"reindex.js","sourceRoot":"","sources":["../../../src/plugins/drive/reindex.ts"],"names":[],"mappings":"AAoBA,OAAO,EAAE,cAAc,EAAE,MAAM,aAAa,CAAC;AAC7C,OAAO,EAAE,aAAa,EAAoB,MAAM,cAAc,CAAC;AAC/D,OAAO,EAAE,UAAU,EAAE,MAAM,iBAAiB,CAAC;AAC7C,OAAO,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAChD,OAAO,EAAE,cAAc,EAAE,MAAM,cAAc,CAAC;AAC9C,OAAO,EACL,gBAAgB,EAChB,eAAe,EACf,gBAAgB,GAEjB,MAAM,aAAa,CAAC;AACrB,kFAAkF;AAClF,OAAO,EACL,WAAW,EACX,YAAY,EACZ,WAAW,EACX,iBAAiB,EACjB,aAAa,EACb,kBAAkB,EAClB,eAAe,EACf,WAAW,EACX,cAAc,GACf,MAAM,cAAc,CAAC;AACtB,OAAO,EAAE,UAAU,EAAE,SAAS,EAAE,YAAY,EAAE,aAAa,EAAE,UAAU,EAAE,MAAM,IAAI,CAAC;AACpF,OAAO,EAAE,IAAI,EAAE,MAAM,MAAM,CAAC;AAC5B,OAAO,EAAE,OAAO,EAAE,MAAM,qBAAqB,CAAC;AAE9C,iEAAiE;AACjE,6CAA6C;AAC7C,MAAM,eAAe,GAAG,IAAI,CAAC,OAAO,EAAE,EAAE,+BAA+B,CAAC,CAAC;AACzE,SAAS,cAAc;IACrB,IAAI,CAAC,UAAU,CAAC,eAAe,CAAC;QAAE,OAAO,IAAI,GAAG,EAAE,CAAC;IACnD,IAAI,CAAC;QACH,OAAO,IAAI,GAAG,CAAC,IAAI,CAAC,KAAK,CAAC,YAAY,CAAC,eAAe,EAAE,MAAM,CAAC,CAAa,CAAC,CAAC;IAChF,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,IAAI,GAAG,EAAE,CAAC;IACnB,CAAC;AACH,CAAC;AACD,SAAS,cAAc,CAAC,IAAiB;IACvC,SAAS,CAAC,OAAO,EAAE,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC;IACvD,aAAa,CAAC,eAAe,EAAE,IAAI,CAAC,SAAS,CAAC,CAAC,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC;AAC5D,CAAC;AACD,SAAS,eAAe;IACtB,IAAI,UAAU,CAAC,eAAe,CAAC;QAAE,UAAU,CAAC,eAAe,CAAC,CAAC;AAC/D,CAAC;AAgBD,SAAS,SAAS,CAAC,IAAc;IAC/B,IAAI,KAAK,GAAgB,WAAW,CAAC;IACrC,IAAI,MAAM,GAAG,KAAK,CAAC;IACnB,IAAI,QAAQ,GAAG,KAAK,CAAC;IACrB,IAAI,OAAO,GAAG,KAAK,CAAC;IACpB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACrC,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,QAAQ,IAAI,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,EAAE,CAAC;YACxC,oDAAoD;YACpD,KAAK,GAAG,CAAC,EAAE,EAAE,EAAE,IAAI,CAAC,EAAE,CAAC,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAC;YAC3D,OAAO,GAAG,IAAI,CAAC;QACjB,CAAC;aAAM,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,YAAY,EAAE,CAAC;YACpC,KAAK,GAAG,WAAW,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,KAAK,CAAC,CAAC;YACtD,OAAO,GAAG,IAAI,CAAC;QACjB,CAAC;aAAM,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,YAAY,EAAE,CAAC;YACpC,KAAK,GAAG,WAAW,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,KAAK,CAAC,CAAC;YACtD,OAAO,GAAG,IAAI,CAAC;QACjB,CAAC;aAAM,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,WAAW,EAAE,CAAC;YACnC,MAAM,GAAG,IAAI,CAAC;QAChB,CAAC;aAAM,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,YAAY,EAAE,CAAC;YACpC,yDAAyD;YACzD,QAAQ,GAAG,IAAI,CAAC;QAClB,CAAC;IACH,CAAC;IACD,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,QAAQ,EAAE,OAAO,EAAE,CAAC;AAC9C,CAAC;AAWD;;GAEG;AACH,KAAK,UAAU,eAAe,CAC5B,KAAkB,EAClB,QAAgB,EAChB,UAAkB;IAElB,MAAM,GAAG,GAAiB,EAAE,CAAC;IAC7B,MAAM,cAAc,GAA4B,EAAE,CAAC,CAAC,wBAAwB;IAC5E,IAAI,SAA6B,CAAC;IAElC,GAAG,CAAC;QACF,MAAM,IAAI,GAAG,MAAM,KAAK,CAAC,KAAK,CAAC,IAAI,CAAC;YAClC,CAAC,EAAE,IAAI,QAAQ,gCAAgC;YAC/C,MAAM,EAAE,0DAA0D;YAClE,QAAQ,EAAE,GAAG;YACb,SAAS;YACT,yBAAyB,EAAE,IAAI;YAC/B,iBAAiB,EAAE,IAAI;SACxB,CAAC,CAAC;QAEH,KAAK,MAAM,CAAC,IAAI,IAAI,CAAC,IAAI,CAAC,KAAK,IAAI,EAAE,EAAE,CAAC;YACtC,IAAI,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,IAAI;gBAAE,SAAS;YAC/B,MAAM,WAAW,GAAG,UAAU,CAAC,CAAC,CAAC,GAAG,UAAU,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;YACpE,IAAI,CAAC,CAAC,QAAQ,KAAK,oCAAoC,EAAE,CAAC;gBACxD,2EAA2E;gBAC3E,IAAI,cAAc,CAAC,CAAC,CAAC,IAAI,CAAC;oBAAE,SAAS;gBACrC,iGAAiG;gBACjG,cAAc,CAAC,IAAI,CAAC,eAAe,CAAC,KAAK,EAAE,CAAC,CAAC,EAAE,EAAE,WAAW,CAAC,CAAC,CAAC;YACjE,CAAC;iBAAM,IAAI,CAAC,CAAC,QAAQ,IAAI,eAAe,CAAC,GAAG,CAAC,CAAC,CAAC,QAAQ,CAAC,EAAE,CAAC;gBACzD,GAAG,CAAC,IAAI,CAAC;oBACP,EAAE,EAAE,CAAC,CAAC,EAAE;oBACR,IAAI,EAAE,CAAC,CAAC,IAAI;oBACZ,IAAI,EAAE,WAAW;oBACjB,YAAY,EAAE,CAAC,CAAC,YAAY,IAAI,EAAE;oBAClC,QAAQ,EAAE,CAAC,CAAC,QAAQ;oBACpB,IAAI,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,SAAS;iBAC1C,CAAC,CAAC;YACL,CAAC;QACH,CAAC;QAED,SAAS,GAAG,IAAI,CAAC,IAAI,CAAC,aAAa,IAAI,SAAS,CAAC;IACnD,CAAC,QAAQ,SAAS,EAAE;IAEpB,4BAA4B;IAC5B,MAAM,YAAY,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC,cAAc,CAAC,CAAC;IACvD,KAAK,MAAM,KAAK,IAAI,YAAY;QAAE,GAAG,CAAC,IAAI,CAAC,GAAG,KAAK,CAAC,CAAC;IAErD,OAAO,GAAG,CAAC;AACb,CAAC;AAED,KAAK,UAAU,gBAAgB,CAAC,KAAkB,EAAE,MAAc;IAChE,MAAM,IAAI,GAAG,MAAM,KAAK,CAAC,KAAK,CAAC,GAAG,CAChC,EAAE,MAAM,EAAE,GAAG,EAAE,OAAO,EAAE,iBAAiB,EAAE,IAAI,EAAE,EACjD,EAAE,YAAY,EAAE,MAAM,EAAE,CACzB,CAAC;IACF,OAAO,IAAI,CAAC,IAAyB,CAAC;AACxC,CAAC;AAED;;;GAGG;AACH,KAAK,UAAU,iBAAiB,CAAC,IAAiB;IAChD,MAAM,MAAM,GAAG,MAAM,eAAe,EAAE,CAAC;IACvC,MAAM,KAAK,GAAG,IAAI,GAAG,EAAU,CAAC;IAChC,IAAI,MAAM,GAAuC,SAAS,CAAC;IAC3D,GAAG,CAAC;QACF,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,MAAM,CAAC,gBAAgB,EAAE;YACjD,KAAK,EAAE,IAAI;YACX,MAAM,EAAE,MAAM,IAAI,SAAS;YAC3B,YAAY,EAAE,CAAC,SAAS,CAAC;YACzB,WAAW,EAAE,KAAK;SACnB,CAAC,CAAC;QACH,KAAK,MAAM,EAAE,IAAI,IAAI,CAAC,MAAM,EAAE,CAAC;YAC7B,MAAM,GAAG,GAAI,EAAE,CAAC,OAAoD,EAAE,OAAO,CAAC;YAC9E,IAAI,OAAO,GAAG,KAAK,QAAQ,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC;gBAAE,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;QAChE,CAAC;QACD,MAAM,GAAI,IAAI,CAAC,gBAAuD,IAAI,IAAI,CAAC;IACjF,CAAC,QAAQ,MAAM,KAAK,IAAI,EAAE;IAC1B,KAAK,MAAM,GAAG,IAAI,KAAK,EAAE,CAAC;QACxB,MAAM,MAAM,CAAC,MAAM,CAAC,gBAAgB,EAAE;YACpC,IAAI,EAAE,IAAI;YACV,MAAM,EAAE,EAAE,IAAI,EAAE,CAAC,EAAE,GAAG,EAAE,SAAS,EAAE,KAAK,EAAE,EAAE,KAAK,EAAE,GAAG,EAAE,EAAE,CAAC,EAAE;SAC9D,CAAC,CAAC;IACL,CAAC;IACD,OAAO,KAAK,CAAC,IAAI,CAAC;AACpB,CAAC;AAED,mGAAmG;AACnG,MAAM,CAAC,KAAK,UAAU,UAAU,CAAC,IAAc;IAC7C,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,QAAQ,EAAE,OAAO,EAAE,GAAG,SAAS,CAAC,IAAI,CAAC,CAAC;IAC7D,MAAM,OAAO,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;IAE3B,OAAO,CAAC,GAAG,CACT,yBAAyB,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,WAAW,MAAM,aAAa,QAAQ,EAAE,CACpG,CAAC;IAEF,mDAAmD;IACnD,IAAI,CAAC,MAAM,EAAE,CAAC;QACZ,IAAI,QAAQ,EAAE,CAAC;YACb,IAAI,CAAC;gBACH,MAAM,CAAC,MAAM,eAAe,EAAE,CAAC,CAAC,gBAAgB,CAAC,gBAAgB,CAAC,CAAC;gBACnE,OAAO,CAAC,GAAG,CAAC,uBAAuB,gBAAgB,iBAAiB,CAAC,CAAC;YACxE,CAAC;YAAC,MAAM,CAAC;gBACP,YAAY;YACd,CAAC;QACH,CAAC;QACD,MAAM,gBAAgB,EAAE,CAAC;QACzB,OAAO,CAAC,GAAG,CAAC,uBAAuB,gBAAgB,SAAS,CAAC,CAAC;IAChE,CAAC;IAED,qCAAqC;IACrC,MAAM,KAAK,GAAG,MAAM,cAAc,EAAE,CAAC;IACrC,MAAM,QAAQ,GAAkD,EAAE,CAAC;IAEnE,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,MAAM,KAAK,GAAG,MAAM,eAAe,CAAC,KAAK,EAAE,IAAI,CAAC,EAAE,EAAE,EAAE,CAAC,CAAC;QACxD,OAAO,CAAC,GAAG,CAAC,YAAY,IAAI,CAAC,IAAI,MAAM,IAAI,CAAC,MAAM,OAAO,KAAK,CAAC,MAAM,QAAQ,CAAC,CAAC;QAC/E,KAAK,MAAM,CAAC,IAAI,KAAK;YAAE,QAAQ,CAAC,IAAI,CAAC,EAAE,GAAG,CAAC,EAAE,MAAM,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC,CAAC;IACtE,CAAC;IACD,OAAO,CAAC,GAAG,CAAC,iBAAiB,QAAQ,CAAC,MAAM,QAAQ,CAAC,CAAC;IAEtD,2CAA2C;IAC3C,2DAA2D;IAC3D,IAAI,QAAQ;QAAE,eAAe,EAAE,CAAC;IAChC,MAAM,IAAI,GAAG,MAAM,CAAC,CAAC,CAAC,IAAI,GAAG,EAAU,CAAC,CAAC,CAAC,cAAc,EAAE,CAAC;IAC3D,IAAI,IAAI,CAAC,IAAI,GAAG,CAAC;QAAE,OAAO,CAAC,GAAG,CAAC,yBAAyB,IAAI,CAAC,IAAI,QAAQ,CAAC,CAAC;IAE3E,IAAI,WAAW,GAAG,CAAC,CAAC;IACpB,IAAI,WAAW,GAAG,CAAC,CAAC;IACpB,IAAI,SAAS,GAAG,CAAC,CAAC;IAElB,KAAK,MAAM,IAAI,IAAI,QAAQ,EAAE,CAAC;QAC5B,SAAS,EAAE,CAAC;QACZ,IAAI,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;YAAE,SAAS,CAAC,iBAAiB;QAClD,IAAI,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,IAAI,GAAG,cAAc,EAAE,CAAC;YAC5C,OAAO,CAAC,GAAG,CAAC,4BAA4B,IAAI,CAAC,IAAI,OAAO,cAAc,QAAQ,IAAI,CAAC,MAAM,KAAK,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC;YAC3G,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;YAClB,IAAI,CAAC,MAAM;gBAAE,cAAc,CAAC,IAAI,CAAC,CAAC;YAClC,SAAS;QACX,CAAC;QACD,IAAI,CAAC;YACH,MAAM,EAAE,IAAI,EAAE,QAAQ,EAAE,GAAG,MAAM,WAAW,CAAC,KAAK,EAAE,IAAI,CAAC,EAAE,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAC;YAC5E,IAAI,QAAQ,IAAI,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;gBACzC,OAAO,CAAC,GAAG,CAAC,8BAA8B,IAAI,CAAC,MAAM,KAAK,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC;gBACvE,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;gBAClB,IAAI,CAAC,MAAM;oBAAE,cAAc,CAAC,IAAI,CAAC,CAAC;gBAClC,SAAS;YACX,CAAC;YACD,MAAM,MAAM,GAAG,aAAa,CAAC,IAAI,CAAC,CAAC;YACnC,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;gBACxB,OAAO,CAAC,GAAG,CAAC,wBAAwB,IAAI,CAAC,MAAM,KAAK,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC;gBACjE,sDAAsD;gBACtD,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;gBAClB,IAAI,CAAC,MAAM;oBAAE,cAAc,CAAC,IAAI,CAAC,CAAC;gBAClC,SAAS;YACX,CAAC;YAED,MAAM,cAAc,GAAG,YAAY,CAAC,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;YAC5D,MAAM,OAAO,GAAG,aAAa,CAAC,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;YACtD,MAAM,IAAI,GAAG,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC,yDAAyD;YACzF,MAAM,UAAU,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC;YAC5D,WAAW,IAAI,MAAM,CAAC,MAAM,CAAC;YAC7B,WAAW,IAAI,UAAU,CAAC;YAE1B,IAAI,MAAM,EAAE,CAAC;gBACX,OAAO,CAAC,GAAG,CACT,YAAY,SAAS,IAAI,QAAQ,CAAC,MAAM,MAAM,IAAI,CAAC,MAAM,KAAK,IAAI,CAAC,IAAI,MAAM,MAAM,CAAC,MAAM,YAAY,UAAU,MAAM,CACvH,CAAC;gBACF,SAAS;YACX,CAAC;YAED,4FAA4F;YAC5F,MAAM,OAAO,GAAG,MAAM,UAAU,CAAC,MAAM,CAAC,GAAG,CAAC,kBAAkB,CAAC,CAAC,CAAC;YAEjE,wEAAwE;YACxE,MAAM,MAAM,GAAG,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,GAAG,EAAE,EAAE,CAAC,CAAC;gBACrC,EAAE,EAAE,WAAW,CAAC,IAAI,CAAC,EAAE,EAAE,GAAG,CAAC;gBAC7B,MAAM,EAAE,OAAO,CAAC,GAAG,CAAC;gBACpB,OAAO,EAAE,iBAAiB,CAAC,CAAC,EAAE,IAAI,EAAE,GAAG,EAAE,MAAM,CAAC,MAAM,EAAE,cAAc,EAAE,OAAO,EAAE,IAAI,CAAC;aACvF,CAAC,CAAC,CAAC;YAEJ,MAAM,MAAM,GAAG,MAAM,eAAe,EAAE,CAAC;YACvC,6DAA6D;YAC7D,8EAA8E;YAC9E,kDAAkD;YAClD,MAAM,MAAM,CAAC,MAAM,CAAC,gBAAgB,EAAE;gBACpC,IAAI,EAAE,KAAK;gBACX,MAAM,EAAE,EAAE,IAAI,EAAE,CAAC,EAAE,GAAG,EAAE,SAAS,EAAE,KAAK,EAAE,EAAE,KAAK,EAAE,IAAI,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE;aAClE,CAAC,CAAC;YACH,MAAM,MAAM,CAAC,MAAM,CAAC,gBAAgB,EAAE;gBACpC,IAAI,EAAE,KAAK;gBACX,MAAM;aACP,CAAC,CAAC;YAEH,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC,uBAAuB;YAC1C,cAAc,CAAC,IAAI,CAAC,CAAC;YAErB,OAAO,CAAC,GAAG,CACT,YAAY,SAAS,IAAI,QAAQ,CAAC,MAAM,MAAM,IAAI,CAAC,MAAM,KAAK,IAAI,CAAC,IAAI,MAAM,MAAM,CAAC,MAAM,qBAAqB,UAAU,OAAO,CACjI,CAAC;QACJ,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACb,mCAAmC;YACnC,OAAO,CAAC,KAAK,CACX,kBAAkB,IAAI,CAAC,MAAM,KAAK,IAAI,CAAC,IAAI,GAAG,EAC9C,GAAG,YAAY,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,CACzC,CAAC;QACJ,CAAC;IACH,CAAC;IAED,0CAA0C;IAC1C,IAAI,CAAC,MAAM,IAAI,CAAC,OAAO,EAAE,CAAC;QACxB,MAAM,MAAM,GAAG,MAAM,iBAAiB,CAAC,IAAI,GAAG,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;QAC3E,OAAO,CAAC,GAAG,CAAC,0BAA0B,MAAM,UAAU,CAAC,CAAC;IAC1D,CAAC;IAED,MAAM,UAAU,GAAG,CAAC,CAAC,IAAI,CAAC,GAAG,EAAE,GAAG,OAAO,CAAC,GAAG,IAAI,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;IAC9D,kDAAkD;IAClD,MAAM,OAAO,GAAG,CAAC,CAAC,WAAW,GAAG,SAAS,CAAC,GAAG,IAAI,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;IAC9D,OAAO,CAAC,GAAG,CACT,oBAAoB,WAAW,YAAY,WAAW,WAAW;QAC/D,KAAK,OAAO,qCAAqC,UAAU,GAAG,CACjE,CAAC;IACF,wCAAwC;IACxC,IAAI,CAAC,MAAM,IAAI,IAAI,CAAC,IAAI,IAAI,QAAQ,CAAC,MAAM;QAAE,eAAe,EAAE,CAAC;AACjE,CAAC"}
|
|
@@ -0,0 +1,11 @@
|
|
|
1
|
+
/** Drive index roots (SSOT for full reindex + incremental sync). Configured via config.drive.roots. */
|
|
2
|
+
export declare const DRIVE_ROOTS: {
|
|
3
|
+
id: string;
|
|
4
|
+
name: string;
|
|
5
|
+
source: "dev" | "biz";
|
|
6
|
+
}[];
|
|
7
|
+
/**
|
|
8
|
+
* 색인·읽기 파일 크기 상한(바이트). 파일을 통째로 메모리에 올리므로, 거대한 파일 하나가 서버를 죽이지 않게 막는다.
|
|
9
|
+
* Google 문서·시트·슬라이드는 Drive 메타에 size 가 없어(export) 이 검사를 받지 않는다.
|
|
10
|
+
*/
|
|
11
|
+
export declare const MAX_FILE_BYTES: number;
|
|
@@ -0,0 +1,13 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* Drive 색인 범위와 한도 — 가벼운 상수만 둔다. 도구들이 이 값 때문에 색인기(indexer.ts, 토크나이저·Qdrant)를
|
|
3
|
+
* 통째로 불러오지 않게 따로 뺐다.
|
|
4
|
+
*/
|
|
5
|
+
import { getConfig } from '../../core/config.js';
|
|
6
|
+
/** Drive index roots (SSOT for full reindex + incremental sync). Configured via config.drive.roots. */
|
|
7
|
+
export const DRIVE_ROOTS = getConfig().drive.roots;
|
|
8
|
+
/**
|
|
9
|
+
* 색인·읽기 파일 크기 상한(바이트). 파일을 통째로 메모리에 올리므로, 거대한 파일 하나가 서버를 죽이지 않게 막는다.
|
|
10
|
+
* Google 문서·시트·슬라이드는 Drive 메타에 size 가 없어(export) 이 검사를 받지 않는다.
|
|
11
|
+
*/
|
|
12
|
+
export const MAX_FILE_BYTES = Number(process.env.DRIVE_MAX_FILE_BYTES ?? 50_000_000);
|
|
13
|
+
//# sourceMappingURL=settings.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"settings.js","sourceRoot":"","sources":["../../../src/plugins/drive/settings.ts"],"names":[],"mappings":"AAAA;;;GAGG;AACH,OAAO,EAAE,SAAS,EAAE,MAAM,sBAAsB,CAAC;AAEjD,uGAAuG;AACvG,MAAM,CAAC,MAAM,WAAW,GAAG,SAAS,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC;AAEnD;;;GAGG;AACH,MAAM,CAAC,MAAM,cAAc,GAAG,MAAM,CAAC,OAAO,CAAC,GAAG,CAAC,oBAAoB,IAAI,UAAU,CAAC,CAAC"}
|
|
@@ -0,0 +1,16 @@
|
|
|
1
|
+
export interface SyncResult {
|
|
2
|
+
status: string;
|
|
3
|
+
changed: number;
|
|
4
|
+
deleted: number;
|
|
5
|
+
skipped: number;
|
|
6
|
+
errored?: number;
|
|
7
|
+
}
|
|
8
|
+
export declare function runIncrementalSync(): Promise<SyncResult>;
|
|
9
|
+
/**
|
|
10
|
+
* startup 호출 — 1회 실행(첫 실행은 token init) + 주기 polling.
|
|
11
|
+
*
|
|
12
|
+
* setInterval 대신 **자기재귀 setTimeout** — 다음 tick 을 이전 sync 가 끝난 뒤에만 예약해
|
|
13
|
+
* 한 주기가 간격을 초과해도 sync 가 겹치지 않게 한다(중복 임베딩·토큰 되감김 방지).
|
|
14
|
+
* `_running` 재진입 가드는 이중 안전장치.
|
|
15
|
+
*/
|
|
16
|
+
export declare function startDriveSyncScheduler(intervalMin?: number): void;
|
|
@@ -0,0 +1,139 @@
|
|
|
1
|
+
import { getDriveClient } from './client.js';
|
|
2
|
+
import { indexFile, deleteFileVectors, SUPPORTED_MIMES, DRIVE_ROOTS } from './indexer.js';
|
|
3
|
+
import { getSyncToken, setSyncToken } from './sync_state.js';
|
|
4
|
+
import { isExcludedPath } from './exclude.js';
|
|
5
|
+
import { getFolderMeta } from './path.js';
|
|
6
|
+
/**
|
|
7
|
+
* fileId 의 parent chain 을 root 까지 추적해 path + source 복원. dev/biz 외면 null.
|
|
8
|
+
* 폴더 hop 은 drive_path.getFolderMeta(프로세스 캐시·10분 TTL)로 조회 — 변경 파일들이 조상 폴더를
|
|
9
|
+
* 공유하므로 warmup 후 캐시 hit 로 Drive API 왕복이 급감(sync 지연·쿼터 절감). 파일 자신의 parents 만
|
|
10
|
+
* 매번 조회(변경 파일이라 캐시 무의미). 폴더 조회 실패 시 getFolderMeta 가 빈 값 → walk 종료(graceful).
|
|
11
|
+
*/
|
|
12
|
+
async function resolvePathAndSource(drive, fileId, fileName) {
|
|
13
|
+
const folders = [];
|
|
14
|
+
let parents = (await drive.files.get({ fileId, fields: 'parents', supportsAllDrives: true })).data.parents ?? undefined;
|
|
15
|
+
for (let hop = 0; hop < 12 && parents && parents.length > 0; hop++) {
|
|
16
|
+
const pid = parents[0];
|
|
17
|
+
const root = DRIVE_ROOTS.find((r) => r.id === pid);
|
|
18
|
+
if (root)
|
|
19
|
+
return { path: [...folders, fileName].join('/'), source: root.source };
|
|
20
|
+
const meta = await getFolderMeta(drive, pid);
|
|
21
|
+
if (meta.name)
|
|
22
|
+
folders.unshift(meta.name);
|
|
23
|
+
parents = meta.parent ? [meta.parent] : undefined;
|
|
24
|
+
}
|
|
25
|
+
return null; // dev/biz root 하위 아님 → skip
|
|
26
|
+
}
|
|
27
|
+
export async function runIncrementalSync() {
|
|
28
|
+
const drive = await getDriveClient();
|
|
29
|
+
const token = await getSyncToken();
|
|
30
|
+
// 첫 실행 — startPageToken 저장만 (이후 변경부터 추적). 전체 색인은 knowledge-mcp drive reindex.
|
|
31
|
+
if (!token) {
|
|
32
|
+
const { data } = await drive.changes.getStartPageToken({ supportsAllDrives: true });
|
|
33
|
+
if (data.startPageToken)
|
|
34
|
+
await setSyncToken(data.startPageToken);
|
|
35
|
+
return { status: 'initialized', changed: 0, deleted: 0, skipped: 0 };
|
|
36
|
+
}
|
|
37
|
+
let changed = 0;
|
|
38
|
+
let deleted = 0;
|
|
39
|
+
let skipped = 0;
|
|
40
|
+
let errored = 0;
|
|
41
|
+
let cursor = token;
|
|
42
|
+
while (cursor) {
|
|
43
|
+
const data = (await drive.changes.list({
|
|
44
|
+
pageToken: cursor,
|
|
45
|
+
fields: 'nextPageToken, newStartPageToken, changes(removed, fileId, file(id,name,mimeType,modifiedTime,trashed,size))',
|
|
46
|
+
pageSize: 100,
|
|
47
|
+
includeRemoved: true,
|
|
48
|
+
includeItemsFromAllDrives: true,
|
|
49
|
+
supportsAllDrives: true,
|
|
50
|
+
})).data;
|
|
51
|
+
for (const ch of data.changes ?? []) {
|
|
52
|
+
// 파일 1건 실패(Mathpix timeout·OpenAI 5xx 소진·깨진 docx 등)가 throw 하면 루프를 탈출해
|
|
53
|
+
// runIncrementalSync 가 전체 throw → 그 tick 의 setSyncToken 이 호출 안 돼 pageToken 이
|
|
54
|
+
// 전진 못 하고, 다음 tick 에 같은 깨진 파일에서 또 멈춘다(sync 영구 정체). per-file 격리로
|
|
55
|
+
// 실패 파일만 errored 처리하고 나머지·토큰 전진은 계속한다.
|
|
56
|
+
try {
|
|
57
|
+
const f = ch.file;
|
|
58
|
+
if (ch.removed || f?.trashed) {
|
|
59
|
+
if (ch.fileId) {
|
|
60
|
+
await deleteFileVectors(ch.fileId);
|
|
61
|
+
deleted++;
|
|
62
|
+
}
|
|
63
|
+
continue;
|
|
64
|
+
}
|
|
65
|
+
if (!f?.id || !f.name || !f.mimeType || !SUPPORTED_MIMES.has(f.mimeType)) {
|
|
66
|
+
skipped++;
|
|
67
|
+
continue;
|
|
68
|
+
}
|
|
69
|
+
const resolved = await resolvePathAndSource(drive, f.id, f.name);
|
|
70
|
+
// 제외: dev/biz 외 폴더 + 민감(PII/계약/HR/deprecated) + 개발 아티팩트(node_modules 등). SSOT: exclude.ts
|
|
71
|
+
// 색인 대상 밖으로 옮겨졌거나 제외 폴더로 들어간 파일은 남아 있던 벡터를 지운다
|
|
72
|
+
// (건너뛰기만 하면 옛 청크 본문이 semantic_search_drive 에 계속 나온다).
|
|
73
|
+
if (!resolved || isExcludedPath(resolved.path)) {
|
|
74
|
+
await deleteFileVectors(f.id);
|
|
75
|
+
skipped++;
|
|
76
|
+
continue;
|
|
77
|
+
}
|
|
78
|
+
await indexFile(drive, {
|
|
79
|
+
id: f.id,
|
|
80
|
+
name: f.name,
|
|
81
|
+
path: resolved.path,
|
|
82
|
+
modifiedTime: f.modifiedTime ?? '',
|
|
83
|
+
source: resolved.source,
|
|
84
|
+
mimeType: f.mimeType,
|
|
85
|
+
size: f.size ? Number(f.size) : undefined,
|
|
86
|
+
});
|
|
87
|
+
changed++;
|
|
88
|
+
}
|
|
89
|
+
catch (e) {
|
|
90
|
+
errored++;
|
|
91
|
+
console.error(`[drive-sync] 파일 색인 실패 (skip, 토큰은 전진): fileId=${ch.fileId ?? '?'} — ${e instanceof Error ? e.message : e}`);
|
|
92
|
+
}
|
|
93
|
+
}
|
|
94
|
+
if (data.newStartPageToken) {
|
|
95
|
+
await setSyncToken(data.newStartPageToken);
|
|
96
|
+
break;
|
|
97
|
+
}
|
|
98
|
+
cursor = data.nextPageToken ?? undefined;
|
|
99
|
+
}
|
|
100
|
+
return { status: 'synced', changed, deleted, skipped, errored };
|
|
101
|
+
}
|
|
102
|
+
let _started = false;
|
|
103
|
+
let _running = false;
|
|
104
|
+
let _timer = null;
|
|
105
|
+
/**
|
|
106
|
+
* startup 호출 — 1회 실행(첫 실행은 token init) + 주기 polling.
|
|
107
|
+
*
|
|
108
|
+
* setInterval 대신 **자기재귀 setTimeout** — 다음 tick 을 이전 sync 가 끝난 뒤에만 예약해
|
|
109
|
+
* 한 주기가 간격을 초과해도 sync 가 겹치지 않게 한다(중복 임베딩·토큰 되감김 방지).
|
|
110
|
+
* `_running` 재진입 가드는 이중 안전장치.
|
|
111
|
+
*/
|
|
112
|
+
export function startDriveSyncScheduler(intervalMin = 30) {
|
|
113
|
+
if (_started)
|
|
114
|
+
return;
|
|
115
|
+
_started = true;
|
|
116
|
+
const run = async () => {
|
|
117
|
+
if (_running) {
|
|
118
|
+
console.error('[drive-sync] 이전 주기 진행 중 — 이번 tick 건너뜀');
|
|
119
|
+
return;
|
|
120
|
+
}
|
|
121
|
+
_running = true;
|
|
122
|
+
try {
|
|
123
|
+
const r = await runIncrementalSync();
|
|
124
|
+
console.error(`[drive-sync] ${r.status} · changed=${r.changed} deleted=${r.deleted} skipped=${r.skipped} errored=${r.errored ?? 0}`);
|
|
125
|
+
}
|
|
126
|
+
catch (e) {
|
|
127
|
+
console.error('[drive-sync] error:', e instanceof Error ? e.message : e);
|
|
128
|
+
}
|
|
129
|
+
finally {
|
|
130
|
+
_running = false;
|
|
131
|
+
}
|
|
132
|
+
};
|
|
133
|
+
const scheduleNext = () => {
|
|
134
|
+
_timer = setTimeout(() => void run().finally(scheduleNext), intervalMin * 60_000);
|
|
135
|
+
};
|
|
136
|
+
void run().finally(scheduleNext); // startup 1회 후 주기 시작
|
|
137
|
+
console.error(`[drive-sync] scheduler 시작 — ${intervalMin}분 주기 (자기재귀 setTimeout · 재진입 가드)`);
|
|
138
|
+
}
|
|
139
|
+
//# sourceMappingURL=sync.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"sync.js","sourceRoot":"","sources":["../../../src/plugins/drive/sync.ts"],"names":[],"mappings":"AAYA,OAAO,EAAE,cAAc,EAAE,MAAM,aAAa,CAAC;AAC7C,OAAO,EAAE,SAAS,EAAE,iBAAiB,EAAE,eAAe,EAAE,WAAW,EAAE,MAAM,cAAc,CAAC;AAC1F,OAAO,EAAE,YAAY,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AAC7D,OAAO,EAAE,cAAc,EAAE,MAAM,cAAc,CAAC;AAC9C,OAAO,EAAE,aAAa,EAAE,MAAM,WAAW,CAAC;AAE1C;;;;;GAKG;AACH,KAAK,UAAU,oBAAoB,CACjC,KAAkB,EAClB,MAAc,EACd,QAAgB;IAEhB,MAAM,OAAO,GAAa,EAAE,CAAC;IAC7B,IAAI,OAAO,GAAyB,CAClC,MAAM,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,MAAM,EAAE,MAAM,EAAE,SAAS,EAAE,iBAAiB,EAAE,IAAI,EAAE,CAAC,CAC9E,CAAC,IAAI,CAAC,OAAO,IAAI,SAAS,CAAC;IAE5B,KAAK,IAAI,GAAG,GAAG,CAAC,EAAE,GAAG,GAAG,EAAE,IAAI,OAAO,IAAI,OAAO,CAAC,MAAM,GAAG,CAAC,EAAE,GAAG,EAAE,EAAE,CAAC;QACnE,MAAM,GAAG,GAAG,OAAO,CAAC,CAAC,CAAC,CAAC;QACvB,MAAM,IAAI,GAAG,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,KAAK,GAAG,CAAC,CAAC;QACnD,IAAI,IAAI;YAAE,OAAO,EAAE,IAAI,EAAE,CAAC,GAAG,OAAO,EAAE,QAAQ,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,MAAM,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC;QACjF,MAAM,IAAI,GAAG,MAAM,aAAa,CAAC,KAAK,EAAE,GAAG,CAAC,CAAC;QAC7C,IAAI,IAAI,CAAC,IAAI;YAAE,OAAO,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAC1C,OAAO,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;IACpD,CAAC;IACD,OAAO,IAAI,CAAC,CAAC,4BAA4B;AAC3C,CAAC;AAUD,MAAM,CAAC,KAAK,UAAU,kBAAkB;IACtC,MAAM,KAAK,GAAG,MAAM,cAAc,EAAE,CAAC;IACrC,MAAM,KAAK,GAAG,MAAM,YAAY,EAAE,CAAC;IAEnC,8EAA8E;IAC9E,IAAI,CAAC,KAAK,EAAE,CAAC;QACX,MAAM,EAAE,IAAI,EAAE,GAAG,MAAM,KAAK,CAAC,OAAO,CAAC,iBAAiB,CAAC,EAAE,iBAAiB,EAAE,IAAI,EAAE,CAAC,CAAC;QACpF,IAAI,IAAI,CAAC,cAAc;YAAE,MAAM,YAAY,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC;QACjE,OAAO,EAAE,MAAM,EAAE,aAAa,EAAE,OAAO,EAAE,CAAC,EAAE,OAAO,EAAE,CAAC,EAAE,OAAO,EAAE,CAAC,EAAE,CAAC;IACvE,CAAC;IAED,IAAI,OAAO,GAAG,CAAC,CAAC;IAChB,IAAI,OAAO,GAAG,CAAC,CAAC;IAChB,IAAI,OAAO,GAAG,CAAC,CAAC;IAChB,IAAI,OAAO,GAAG,CAAC,CAAC;IAChB,IAAI,MAAM,GAAuB,KAAK,CAAC;IAEvC,OAAO,MAAM,EAAE,CAAC;QACd,MAAM,IAAI,GAAoB,CAC5B,MAAM,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC;YACvB,SAAS,EAAE,MAAM;YACjB,MAAM,EACJ,8GAA8G;YAChH,QAAQ,EAAE,GAAG;YACb,cAAc,EAAE,IAAI;YACpB,yBAAyB,EAAE,IAAI;YAC/B,iBAAiB,EAAE,IAAI;SACxB,CAAC,CACH,CAAC,IAAI,CAAC;QAEP,KAAK,MAAM,EAAE,IAAI,IAAI,CAAC,OAAO,IAAI,EAAE,EAAE,CAAC;YACpC,sEAAsE;YACtE,6EAA6E;YAC7E,gEAAgE;YAChE,uCAAuC;YACvC,IAAI,CAAC;gBACH,MAAM,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC;gBAClB,IAAI,EAAE,CAAC,OAAO,IAAI,CAAC,EAAE,OAAO,EAAE,CAAC;oBAC7B,IAAI,EAAE,CAAC,MAAM,EAAE,CAAC;wBACd,MAAM,iBAAiB,CAAC,EAAE,CAAC,MAAM,CAAC,CAAC;wBACnC,OAAO,EAAE,CAAC;oBACZ,CAAC;oBACD,SAAS;gBACX,CAAC;gBACD,IAAI,CAAC,CAAC,EAAE,EAAE,IAAI,CAAC,CAAC,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC,QAAQ,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC,CAAC,CAAC,QAAQ,CAAC,EAAE,CAAC;oBACzE,OAAO,EAAE,CAAC;oBACV,SAAS;gBACX,CAAC;gBACD,MAAM,QAAQ,GAAG,MAAM,oBAAoB,CAAC,KAAK,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC;gBACjE,0FAA0F;gBAC1F,+CAA+C;gBAC/C,sDAAsD;gBACtD,IAAI,CAAC,QAAQ,IAAI,cAAc,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC;oBAC/C,MAAM,iBAAiB,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;oBAC9B,OAAO,EAAE,CAAC;oBACV,SAAS;gBACX,CAAC;gBACD,MAAM,SAAS,CAAC,KAAK,EAAE;oBACrB,EAAE,EAAE,CAAC,CAAC,EAAE;oBACR,IAAI,EAAE,CAAC,CAAC,IAAI;oBACZ,IAAI,EAAE,QAAQ,CAAC,IAAI;oBACnB,YAAY,EAAE,CAAC,CAAC,YAAY,IAAI,EAAE;oBAClC,MAAM,EAAE,QAAQ,CAAC,MAAM;oBACvB,QAAQ,EAAE,CAAC,CAAC,QAAQ;oBACpB,IAAI,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,SAAS;iBAC1C,CAAC,CAAC;gBACH,OAAO,EAAE,CAAC;YACZ,CAAC;YAAC,OAAO,CAAC,EAAE,CAAC;gBACX,OAAO,EAAE,CAAC;gBACV,OAAO,CAAC,KAAK,CACX,gDAAgD,EAAE,CAAC,MAAM,IAAI,GAAG,MAAM,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,EAAE,CAC3G,CAAC;YACJ,CAAC;QACH,CAAC;QAED,IAAI,IAAI,CAAC,iBAAiB,EAAE,CAAC;YAC3B,MAAM,YAAY,CAAC,IAAI,CAAC,iBAAiB,CAAC,CAAC;YAC3C,MAAM;QACR,CAAC;QACD,MAAM,GAAG,IAAI,CAAC,aAAa,IAAI,SAAS,CAAC;IAC3C,CAAC;IAED,OAAO,EAAE,MAAM,EAAE,QAAQ,EAAE,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,CAAC;AAClE,CAAC;AAED,IAAI,QAAQ,GAAG,KAAK,CAAC;AACrB,IAAI,QAAQ,GAAG,KAAK,CAAC;AACrB,IAAI,MAAM,GAA0B,IAAI,CAAC;AAEzC;;;;;;GAMG;AACH,MAAM,UAAU,uBAAuB,CAAC,WAAW,GAAG,EAAE;IACtD,IAAI,QAAQ;QAAE,OAAO;IACrB,QAAQ,GAAG,IAAI,CAAC;IAEhB,MAAM,GAAG,GAAG,KAAK,IAAI,EAAE;QACrB,IAAI,QAAQ,EAAE,CAAC;YACb,OAAO,CAAC,KAAK,CAAC,uCAAuC,CAAC,CAAC;YACvD,OAAO;QACT,CAAC;QACD,QAAQ,GAAG,IAAI,CAAC;QAChB,IAAI,CAAC;YACH,MAAM,CAAC,GAAG,MAAM,kBAAkB,EAAE,CAAC;YACrC,OAAO,CAAC,KAAK,CACX,gBAAgB,CAAC,CAAC,MAAM,cAAc,CAAC,CAAC,OAAO,YAAY,CAAC,CAAC,OAAO,YAAY,CAAC,CAAC,OAAO,YAAY,CAAC,CAAC,OAAO,IAAI,CAAC,EAAE,CACtH,CAAC;QACJ,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACX,OAAO,CAAC,KAAK,CAAC,qBAAqB,EAAE,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAC3E,CAAC;gBAAS,CAAC;YACT,QAAQ,GAAG,KAAK,CAAC;QACnB,CAAC;IACH,CAAC,CAAC;IAEF,MAAM,YAAY,GAAG,GAAG,EAAE;QACxB,MAAM,GAAG,UAAU,CAAC,GAAG,EAAE,CAAC,KAAK,GAAG,EAAE,CAAC,OAAO,CAAC,YAAY,CAAC,EAAE,WAAW,GAAG,MAAM,CAAC,CAAC;IACpF,CAAC,CAAC;IAEF,KAAK,GAAG,EAAE,CAAC,OAAO,CAAC,YAAY,CAAC,CAAC,CAAC,qBAAqB;IACvD,OAAO,CAAC,KAAK,CAAC,+BAA+B,WAAW,iCAAiC,CAAC,CAAC;AAC7F,CAAC"}
|
|
@@ -0,0 +1,26 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* Drive 증분 sync 의 pageToken 영속화 — 저장소 plugin state (namespace "drive", key "pageToken").
|
|
3
|
+
*
|
|
4
|
+
* 로컬 파일은 컨테이너 재배포마다 사라져(ephemeral) 재배포~다음 전체 reindex 사이의 Drive 변경분이
|
|
5
|
+
* 인덱스에 영구 누락됐다. 그래서 저장소(SQLite 파일은 볼륨, 또는 Supabase)에 둔다.
|
|
6
|
+
*
|
|
7
|
+
* 읽기 실패는 throw 한다. 예전엔 null(= 첫 실행)로 취급해 "지금부터" 토큰으로 덮어써,
|
|
8
|
+
* 일시 장애 한 번에 그 사이 변경분을 영구히 놓쳤다. 이제 그 주기만 건너뛰고 다음 주기에 다시 읽는다.
|
|
9
|
+
*/
|
|
10
|
+
import { getStore } from '../../store/index.js';
|
|
11
|
+
const NS = 'drive';
|
|
12
|
+
const KEY = 'pageToken';
|
|
13
|
+
/** 저장된 토큰. 없으면 null(첫 실행). 저장소 오류는 throw. */
|
|
14
|
+
export async function getSyncToken() {
|
|
15
|
+
const v = await (await getStore()).getState(NS, KEY);
|
|
16
|
+
return typeof v === 'string' && v ? v : null;
|
|
17
|
+
}
|
|
18
|
+
export async function setSyncToken(token) {
|
|
19
|
+
try {
|
|
20
|
+
await (await getStore()).setState(NS, KEY, token);
|
|
21
|
+
}
|
|
22
|
+
catch (e) {
|
|
23
|
+
console.error('[sync_state] setSyncToken failed:', e instanceof Error ? e.message : e);
|
|
24
|
+
}
|
|
25
|
+
}
|
|
26
|
+
//# sourceMappingURL=sync_state.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"sync_state.js","sourceRoot":"","sources":["../../../src/plugins/drive/sync_state.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AACH,OAAO,EAAE,QAAQ,EAAE,MAAM,sBAAsB,CAAC;AAEhD,MAAM,EAAE,GAAG,OAAO,CAAC;AACnB,MAAM,GAAG,GAAG,WAAW,CAAC;AAExB,6CAA6C;AAC7C,MAAM,CAAC,KAAK,UAAU,YAAY;IAChC,MAAM,CAAC,GAAG,MAAM,CAAC,MAAM,QAAQ,EAAE,CAAC,CAAC,QAAQ,CAAC,EAAE,EAAE,GAAG,CAAC,CAAC;IACrD,OAAO,OAAO,CAAC,KAAK,QAAQ,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;AAC/C,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,YAAY,CAAC,KAAa;IAC9C,IAAI,CAAC;QACH,MAAM,CAAC,MAAM,QAAQ,EAAE,CAAC,CAAC,QAAQ,CAAC,EAAE,EAAE,GAAG,EAAE,KAAK,CAAC,CAAC;IACpD,CAAC;IAAC,OAAO,CAAC,EAAE,CAAC;QACX,OAAO,CAAC,KAAK,CAAC,mCAAmC,EAAE,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IACzF,CAAC;AACH,CAAC"}
|
|
@@ -0,0 +1,104 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* list_drive_folder — Drive 폴더 자식 list (배치 + 재귀 depth).
|
|
3
|
+
*
|
|
4
|
+
* 배치/재귀: 폴더 N개·하위 트리를 1회 호출로. 에이전트가 폴더마다 순차 호출(= per-call
|
|
5
|
+
* 네트워크 floor + 그 사이 LLM 추론 턴 ×N)하던 것을 1회로 접어 체감 지연을 N→1 로 줄인다.
|
|
6
|
+
* read_drive_file 의 fileIds 배치와 같은 패턴(서버가 Drive I/O 를 자기↔구글 빠른 망에서 병렬).
|
|
7
|
+
*
|
|
8
|
+
* - folderNames[] / folderIds[] : 서로 다른 위치의 폴더 여러 개를 병렬 list.
|
|
9
|
+
* - depth (1~3) : 한 폴더의 하위 트리를 N단계까지 한 번에.
|
|
10
|
+
* 예) 고객사 10곳 + 각 폴더 문서 → list_drive_folder({ folderName: "01_고객사", depth: 2 }).
|
|
11
|
+
*
|
|
12
|
+
* 보안: 시작 폴더는 색인 root 하위이고 제외 폴더(PII·계약·HR 등)를 거치지 않을 때만 연다(read_drive_file 과 동일 가드).
|
|
13
|
+
* 이름으로 찾을 때도 root 하위 후보만 고른다. 재귀 walk 는 제외 폴더로 확산하지 않는다
|
|
14
|
+
* (해당 폴더는 이름만 표기하고 자식 미전개 — excluded:true).
|
|
15
|
+
*
|
|
16
|
+
* 안전장치: 총 폴더 방문(=Drive list 호출) 상한 MAX_FOLDER_VISITS — 깊고 넓은 트리의 호출 폭발 방지.
|
|
17
|
+
*/
|
|
18
|
+
import { z } from 'zod';
|
|
19
|
+
declare const inputSchema: z.ZodEffects<z.ZodObject<{
|
|
20
|
+
folderId: z.ZodOptional<z.ZodString>;
|
|
21
|
+
folderName: z.ZodOptional<z.ZodString>;
|
|
22
|
+
folderIds: z.ZodOptional<z.ZodArray<z.ZodString, "many">>;
|
|
23
|
+
folderNames: z.ZodOptional<z.ZodArray<z.ZodString, "many">>;
|
|
24
|
+
depth: z.ZodDefault<z.ZodNumber>;
|
|
25
|
+
pageSize: z.ZodDefault<z.ZodNumber>;
|
|
26
|
+
}, "strip", z.ZodTypeAny, {
|
|
27
|
+
depth: number;
|
|
28
|
+
pageSize: number;
|
|
29
|
+
folderId?: string | undefined;
|
|
30
|
+
folderName?: string | undefined;
|
|
31
|
+
folderIds?: string[] | undefined;
|
|
32
|
+
folderNames?: string[] | undefined;
|
|
33
|
+
}, {
|
|
34
|
+
depth?: number | undefined;
|
|
35
|
+
pageSize?: number | undefined;
|
|
36
|
+
folderId?: string | undefined;
|
|
37
|
+
folderName?: string | undefined;
|
|
38
|
+
folderIds?: string[] | undefined;
|
|
39
|
+
folderNames?: string[] | undefined;
|
|
40
|
+
}>, {
|
|
41
|
+
depth: number;
|
|
42
|
+
pageSize: number;
|
|
43
|
+
folderId?: string | undefined;
|
|
44
|
+
folderName?: string | undefined;
|
|
45
|
+
folderIds?: string[] | undefined;
|
|
46
|
+
folderNames?: string[] | undefined;
|
|
47
|
+
}, {
|
|
48
|
+
depth?: number | undefined;
|
|
49
|
+
pageSize?: number | undefined;
|
|
50
|
+
folderId?: string | undefined;
|
|
51
|
+
folderName?: string | undefined;
|
|
52
|
+
folderIds?: string[] | undefined;
|
|
53
|
+
folderNames?: string[] | undefined;
|
|
54
|
+
}>;
|
|
55
|
+
export declare const listDriveFolderTool: import("../../../plugin/define.js").Tool<z.ZodEffects<z.ZodObject<{
|
|
56
|
+
folderId: z.ZodOptional<z.ZodString>;
|
|
57
|
+
folderName: z.ZodOptional<z.ZodString>;
|
|
58
|
+
folderIds: z.ZodOptional<z.ZodArray<z.ZodString, "many">>;
|
|
59
|
+
folderNames: z.ZodOptional<z.ZodArray<z.ZodString, "many">>;
|
|
60
|
+
depth: z.ZodDefault<z.ZodNumber>;
|
|
61
|
+
pageSize: z.ZodDefault<z.ZodNumber>;
|
|
62
|
+
}, "strip", z.ZodTypeAny, {
|
|
63
|
+
depth: number;
|
|
64
|
+
pageSize: number;
|
|
65
|
+
folderId?: string | undefined;
|
|
66
|
+
folderName?: string | undefined;
|
|
67
|
+
folderIds?: string[] | undefined;
|
|
68
|
+
folderNames?: string[] | undefined;
|
|
69
|
+
}, {
|
|
70
|
+
depth?: number | undefined;
|
|
71
|
+
pageSize?: number | undefined;
|
|
72
|
+
folderId?: string | undefined;
|
|
73
|
+
folderName?: string | undefined;
|
|
74
|
+
folderIds?: string[] | undefined;
|
|
75
|
+
folderNames?: string[] | undefined;
|
|
76
|
+
}>, {
|
|
77
|
+
depth: number;
|
|
78
|
+
pageSize: number;
|
|
79
|
+
folderId?: string | undefined;
|
|
80
|
+
folderName?: string | undefined;
|
|
81
|
+
folderIds?: string[] | undefined;
|
|
82
|
+
folderNames?: string[] | undefined;
|
|
83
|
+
}, {
|
|
84
|
+
depth?: number | undefined;
|
|
85
|
+
pageSize?: number | undefined;
|
|
86
|
+
folderId?: string | undefined;
|
|
87
|
+
folderName?: string | undefined;
|
|
88
|
+
folderIds?: string[] | undefined;
|
|
89
|
+
folderNames?: string[] | undefined;
|
|
90
|
+
}>>;
|
|
91
|
+
export declare function handleListDriveFolder(args: z.infer<typeof inputSchema>): Promise<{
|
|
92
|
+
content: {
|
|
93
|
+
type: "text";
|
|
94
|
+
text: string;
|
|
95
|
+
}[];
|
|
96
|
+
isError: boolean;
|
|
97
|
+
} | {
|
|
98
|
+
content: {
|
|
99
|
+
type: "text";
|
|
100
|
+
text: string;
|
|
101
|
+
}[];
|
|
102
|
+
isError?: undefined;
|
|
103
|
+
}>;
|
|
104
|
+
export {};
|