@khoinguyen2002/doc-mcp 1.0.0 → 1.0.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/db/vector.d.ts +5 -5
- package/dist/db/vector.d.ts.map +1 -1
- package/dist/db/vector.js +39 -33
- package/dist/mcp-server.js +7 -3
- package/dist/tools/driveTools.d.ts +1 -1
- package/dist/tools/driveTools.d.ts.map +1 -1
- package/dist/tools/driveTools.js +28 -17
- package/dist/tools/knowledgeTools.d.ts.map +1 -1
- package/dist/tools/knowledgeTools.js +28 -7
- package/package.json +1 -1
- package/src/db/vector.ts +39 -34
- package/src/mcp-server.ts +7 -3
- package/src/tools/driveTools.ts +33 -19
- package/src/tools/knowledgeTools.ts +27 -8
- package/src/hooks/driveSync.ts +0 -108
package/dist/db/vector.d.ts
CHANGED
|
@@ -1,8 +1,8 @@
|
|
|
1
1
|
export declare function initVectorDB(): Promise<void>;
|
|
2
2
|
export declare function embedText(text: string): Promise<number[]>;
|
|
3
|
-
export declare function upsertProjectDocument(
|
|
4
|
-
export declare function searchProjectMemory(
|
|
5
|
-
export declare function deleteProjectDocument(
|
|
6
|
-
export declare function checkProjectDocumentExists(
|
|
7
|
-
export declare function getProjectDocumentMetadata(
|
|
3
|
+
export declare function upsertProjectDocument(folderId: string, text: string, metadata?: Record<string, any>): Promise<void>;
|
|
4
|
+
export declare function searchProjectMemory(folderId: string, query: string, topK?: number): Promise<any[]>;
|
|
5
|
+
export declare function deleteProjectDocument(folderId: string, fileId: string): Promise<void>;
|
|
6
|
+
export declare function checkProjectDocumentExists(folderId: string, fileId: string): Promise<boolean>;
|
|
7
|
+
export declare function getProjectDocumentMetadata(folderId: string): Promise<Record<string, string>>;
|
|
8
8
|
//# sourceMappingURL=vector.d.ts.map
|
package/dist/db/vector.d.ts.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"vector.d.ts","sourceRoot":"","sources":["../../src/db/vector.ts"],"names":[],"mappings":"AAOA,wBAAsB,YAAY,kBAqCjC;AAED,wBAAsB,SAAS,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,CAwB/D;AAED,wBAAsB,qBAAqB,CAAC,
|
|
1
|
+
{"version":3,"file":"vector.d.ts","sourceRoot":"","sources":["../../src/db/vector.ts"],"names":[],"mappings":"AAOA,wBAAsB,YAAY,kBAqCjC;AAED,wBAAsB,SAAS,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,CAwB/D;AAED,wBAAsB,qBAAqB,CAAC,QAAQ,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,QAAQ,GAAE,MAAM,CAAC,MAAM,EAAE,GAAG,CAAM,GAAG,OAAO,CAAC,IAAI,CAAC,CA0B7H;AAED,wBAAsB,mBAAmB,CAAC,QAAQ,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,GAAE,MAAU,GAAG,OAAO,CAAC,GAAG,EAAE,CAAC,CAiC3G;AAED,wBAAsB,qBAAqB,CAAC,QAAQ,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC,CAa3F;AAED,wBAAsB,0BAA0B,CAAC,QAAQ,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,GAAG,OAAO,CAAC,OAAO,CAAC,CAanG;AAED,wBAAsB,0BAA0B,CAAC,QAAQ,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAwBlG"}
|
package/dist/db/vector.js
CHANGED
|
@@ -24,7 +24,7 @@ export async function initVectorDB() {
|
|
|
24
24
|
},
|
|
25
25
|
});
|
|
26
26
|
await client.createPayloadIndex(COLLECTION_NAME, {
|
|
27
|
-
field_name: "
|
|
27
|
+
field_name: "folderId",
|
|
28
28
|
field_schema: "keyword",
|
|
29
29
|
});
|
|
30
30
|
await client.createPayloadIndex(COLLECTION_NAME, {
|
|
@@ -61,7 +61,7 @@ export async function embedText(text) {
|
|
|
61
61
|
}
|
|
62
62
|
return json.data[0].embedding;
|
|
63
63
|
}
|
|
64
|
-
export async function upsertProjectDocument(
|
|
64
|
+
export async function upsertProjectDocument(folderId, text, metadata = {}) {
|
|
65
65
|
await initVectorDB();
|
|
66
66
|
if (!client)
|
|
67
67
|
throw new Error("Qdrant not initialized");
|
|
@@ -73,7 +73,7 @@ export async function upsertProjectDocument(projectId, text, metadata = {}) {
|
|
|
73
73
|
id: uuidv4(),
|
|
74
74
|
vector: vector,
|
|
75
75
|
payload: {
|
|
76
|
-
|
|
76
|
+
folderId,
|
|
77
77
|
text,
|
|
78
78
|
source: metadata.source || "user",
|
|
79
79
|
file_id: metadata.file_id || null,
|
|
@@ -84,71 +84,77 @@ export async function upsertProjectDocument(projectId, text, metadata = {}) {
|
|
|
84
84
|
}
|
|
85
85
|
]
|
|
86
86
|
});
|
|
87
|
-
console.error(`Upserted document chunk for
|
|
87
|
+
console.error(`Upserted document chunk for folder ${folderId}`);
|
|
88
88
|
}
|
|
89
|
-
export async function searchProjectMemory(
|
|
89
|
+
export async function searchProjectMemory(folderId, query, topK = 3) {
|
|
90
90
|
await initVectorDB();
|
|
91
91
|
if (!client)
|
|
92
92
|
throw new Error("Qdrant not initialized");
|
|
93
|
-
|
|
94
|
-
|
|
95
|
-
|
|
96
|
-
|
|
97
|
-
|
|
98
|
-
|
|
99
|
-
|
|
100
|
-
|
|
101
|
-
|
|
102
|
-
|
|
103
|
-
|
|
93
|
+
try {
|
|
94
|
+
const queryVector = await embedText(query);
|
|
95
|
+
const results = await client.search(COLLECTION_NAME, {
|
|
96
|
+
vector: queryVector,
|
|
97
|
+
limit: topK,
|
|
98
|
+
with_payload: true,
|
|
99
|
+
filter: {
|
|
100
|
+
must: [
|
|
101
|
+
{
|
|
102
|
+
key: "folderId",
|
|
103
|
+
match: {
|
|
104
|
+
value: folderId
|
|
105
|
+
}
|
|
104
106
|
}
|
|
105
|
-
|
|
106
|
-
|
|
107
|
-
}
|
|
108
|
-
|
|
109
|
-
|
|
110
|
-
|
|
111
|
-
|
|
112
|
-
|
|
113
|
-
|
|
114
|
-
}
|
|
107
|
+
]
|
|
108
|
+
}
|
|
109
|
+
});
|
|
110
|
+
// Map to match LanceDB format expected by other tools
|
|
111
|
+
return results.map(r => ({
|
|
112
|
+
id: r.id,
|
|
113
|
+
vector: r.vector,
|
|
114
|
+
...r.payload
|
|
115
|
+
}));
|
|
116
|
+
}
|
|
117
|
+
catch (err) {
|
|
118
|
+
console.error("Qdrant search error:", err.message);
|
|
119
|
+
return [];
|
|
120
|
+
}
|
|
115
121
|
}
|
|
116
|
-
export async function deleteProjectDocument(
|
|
122
|
+
export async function deleteProjectDocument(folderId, fileId) {
|
|
117
123
|
await initVectorDB();
|
|
118
124
|
if (!client)
|
|
119
125
|
return;
|
|
120
126
|
await client.delete(COLLECTION_NAME, {
|
|
121
127
|
filter: {
|
|
122
128
|
must: [
|
|
123
|
-
{ key: "
|
|
129
|
+
{ key: "folderId", match: { value: folderId } },
|
|
124
130
|
{ key: "file_id", match: { value: fileId } }
|
|
125
131
|
]
|
|
126
132
|
}
|
|
127
133
|
});
|
|
128
|
-
console.error(`Deleted old chunks from Qdrant for ${
|
|
134
|
+
console.error(`Deleted old chunks from Qdrant for ${folderId} / ${fileId}`);
|
|
129
135
|
}
|
|
130
|
-
export async function checkProjectDocumentExists(
|
|
136
|
+
export async function checkProjectDocumentExists(folderId, fileId) {
|
|
131
137
|
await initVectorDB();
|
|
132
138
|
if (!client)
|
|
133
139
|
return false;
|
|
134
140
|
const res = await client.count(COLLECTION_NAME, {
|
|
135
141
|
filter: {
|
|
136
142
|
must: [
|
|
137
|
-
{ key: "
|
|
143
|
+
{ key: "folderId", match: { value: folderId } },
|
|
138
144
|
{ key: "file_id", match: { value: fileId } }
|
|
139
145
|
]
|
|
140
146
|
}
|
|
141
147
|
});
|
|
142
148
|
return res.count > 0;
|
|
143
149
|
}
|
|
144
|
-
export async function getProjectDocumentMetadata(
|
|
150
|
+
export async function getProjectDocumentMetadata(folderId) {
|
|
145
151
|
await initVectorDB();
|
|
146
152
|
if (!client)
|
|
147
153
|
return {};
|
|
148
154
|
const res = await client.scroll(COLLECTION_NAME, {
|
|
149
155
|
filter: {
|
|
150
156
|
must: [
|
|
151
|
-
{ key: "
|
|
157
|
+
{ key: "folderId", match: { value: folderId } },
|
|
152
158
|
{ key: "source", match: { value: "google_drive" } }
|
|
153
159
|
]
|
|
154
160
|
},
|
package/dist/mcp-server.js
CHANGED
|
@@ -16,15 +16,19 @@ const server = new McpServer({
|
|
|
16
16
|
});
|
|
17
17
|
// Register tools
|
|
18
18
|
server.registerTool("list_drive_files", {
|
|
19
|
-
description: "List and search for Google Drive documents in
|
|
19
|
+
description: "List and search for Google Drive documents and subfolders in a specific folder.",
|
|
20
20
|
inputSchema: {
|
|
21
21
|
keyword: z
|
|
22
22
|
.string()
|
|
23
23
|
.optional()
|
|
24
24
|
.describe("Optional keyword to search for in document titles"),
|
|
25
|
+
targetFolderId: z
|
|
26
|
+
.string()
|
|
27
|
+
.optional()
|
|
28
|
+
.describe("Optional Google Drive folder ID to list contents from. Defaults to the root knowledge folder."),
|
|
25
29
|
},
|
|
26
|
-
}, async ({ keyword }) => {
|
|
27
|
-
const res = await listDriveFiles(keyword);
|
|
30
|
+
}, async ({ keyword, targetFolderId }) => {
|
|
31
|
+
const res = await listDriveFiles(keyword, targetFolderId);
|
|
28
32
|
if (!res.success) {
|
|
29
33
|
return {
|
|
30
34
|
content: [{ type: "text", text: `Error: ${res.error}` }],
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"driveTools.d.ts","sourceRoot":"","sources":["../../src/tools/driveTools.ts"],"names":[],"mappings":"AAiCA,wBAAsB,cAAc,CAAC,OAAO,CAAC,EAAE,MAAM;;;;;;;;
|
|
1
|
+
{"version":3,"file":"driveTools.d.ts","sourceRoot":"","sources":["../../src/tools/driveTools.ts"],"names":[],"mappings":"AAiCA,wBAAsB,cAAc,CAAC,OAAO,CAAC,EAAE,MAAM,EAAE,cAAc,CAAC,EAAE,MAAM;;;;;;;;GAoC7E;AAED,wBAAsB,kBAAkB,CAAC,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM;;;;;;;;GA6CxE;AAED,wBAAsB,iBAAiB,CAAC,MAAM,EAAE,MAAM;;;;;;;;GAsCrD;AAED,wBAAsB,eAAe,CAAC,QAAQ,EAAE,MAAM;;;;;;GAqDrD"}
|
package/dist/tools/driveTools.js
CHANGED
|
@@ -19,8 +19,8 @@ function getDriveClient() {
|
|
|
19
19
|
});
|
|
20
20
|
return google.drive({ version: "v3", auth });
|
|
21
21
|
}
|
|
22
|
-
export async function listDriveFiles(keyword) {
|
|
23
|
-
const folderId =
|
|
22
|
+
export async function listDriveFiles(keyword, targetFolderId) {
|
|
23
|
+
const folderId = targetFolderId || config.DOC_MCP_DRIVE_FOLDER_ID;
|
|
24
24
|
if (!folderId) {
|
|
25
25
|
return {
|
|
26
26
|
success: false,
|
|
@@ -29,14 +29,14 @@ export async function listDriveFiles(keyword) {
|
|
|
29
29
|
}
|
|
30
30
|
try {
|
|
31
31
|
const drive = getDriveClient();
|
|
32
|
-
let q = "mimeType = 'application/vnd.google-apps.document'";
|
|
32
|
+
let q = "(mimeType = 'application/vnd.google-apps.document' or mimeType = 'application/vnd.google-apps.folder') and trashed = false";
|
|
33
33
|
q = `'${folderId}' in parents and ${q}`;
|
|
34
34
|
if (keyword) {
|
|
35
35
|
q = `name contains '${keyword}' and ${q}`;
|
|
36
36
|
}
|
|
37
37
|
const res = await drive.files.list({
|
|
38
38
|
q,
|
|
39
|
-
fields: "files(id, name, description)",
|
|
39
|
+
fields: "files(id, name, description, mimeType)",
|
|
40
40
|
spaces: "drive",
|
|
41
41
|
pageSize: 50,
|
|
42
42
|
supportsAllDrives: true,
|
|
@@ -92,7 +92,7 @@ export async function syncSingleDocument(fileId, folderId) {
|
|
|
92
92
|
return { synced: false, driveModifiedTime };
|
|
93
93
|
}
|
|
94
94
|
export async function readDriveDocument(fileId) {
|
|
95
|
-
const folderId =
|
|
95
|
+
const folderId = config.DOC_MCP_DRIVE_FOLDER_ID;
|
|
96
96
|
if (!folderId) {
|
|
97
97
|
return {
|
|
98
98
|
success: false,
|
|
@@ -130,17 +130,28 @@ export async function readDriveDocument(fileId) {
|
|
|
130
130
|
export async function syncFolderState(folderId) {
|
|
131
131
|
try {
|
|
132
132
|
const drive = getDriveClient();
|
|
133
|
-
|
|
134
|
-
|
|
135
|
-
|
|
136
|
-
|
|
137
|
-
|
|
138
|
-
|
|
139
|
-
|
|
140
|
-
|
|
141
|
-
|
|
142
|
-
|
|
143
|
-
|
|
133
|
+
async function getAllDocumentsFlat() {
|
|
134
|
+
let allDocs = [];
|
|
135
|
+
let pageToken = undefined;
|
|
136
|
+
do {
|
|
137
|
+
const docsRes = await drive.files.list({
|
|
138
|
+
// Chú ý: Đéo check parentId nữa, gom sạch sành sanh mọi file .doc mà Service Account nhìn thấy
|
|
139
|
+
q: `mimeType = 'application/vnd.google-apps.document' and trashed = false`,
|
|
140
|
+
fields: "nextPageToken, files(id, name, modifiedTime)",
|
|
141
|
+
spaces: "drive",
|
|
142
|
+
pageSize: 100, // Google API limit mỗi page, tự động nhảy trang nếu nhiều hơn
|
|
143
|
+
pageToken,
|
|
144
|
+
supportsAllDrives: true,
|
|
145
|
+
includeItemsFromAllDrives: true,
|
|
146
|
+
});
|
|
147
|
+
if (docsRes.data.files) {
|
|
148
|
+
allDocs = allDocs.concat(docsRes.data.files);
|
|
149
|
+
}
|
|
150
|
+
pageToken = docsRes.data.nextPageToken || undefined;
|
|
151
|
+
} while (pageToken);
|
|
152
|
+
return allDocs;
|
|
153
|
+
}
|
|
154
|
+
const driveFiles = await getAllDocumentsFlat();
|
|
144
155
|
const dbMetaMap = await getProjectDocumentMetadata(folderId);
|
|
145
156
|
// Sync updated or new files
|
|
146
157
|
for (const file of driveFiles) {
|
|
@@ -153,7 +164,7 @@ export async function syncFolderState(folderId) {
|
|
|
153
164
|
}
|
|
154
165
|
// Delete removed files from DB
|
|
155
166
|
for (const dbFileId of Object.keys(dbMetaMap)) {
|
|
156
|
-
if (!driveFiles.find(f => f.id === dbFileId)) {
|
|
167
|
+
if (!driveFiles.find((f) => f.id === dbFileId)) {
|
|
157
168
|
await deleteProjectDocument(folderId, dbFileId);
|
|
158
169
|
}
|
|
159
170
|
}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"knowledgeTools.d.ts","sourceRoot":"","sources":["../../src/tools/knowledgeTools.ts"],"names":[],"mappings":"
|
|
1
|
+
{"version":3,"file":"knowledgeTools.d.ts","sourceRoot":"","sources":["../../src/tools/knowledgeTools.ts"],"names":[],"mappings":"AAIA,wBAAsB,aAAa,CAAC,OAAO,EAAE,MAAM;;;;;;;;GAmBlD;AAED,wBAAsB,eAAe,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,GAAE,MAAU;;;;;;;;GAoCpE"}
|
|
@@ -1,25 +1,34 @@
|
|
|
1
|
+
import { config } from "../config.js";
|
|
1
2
|
import { upsertProjectDocument, searchProjectMemory } from "../db/vector.js";
|
|
2
3
|
import { syncFolderState } from "./driveTools.js";
|
|
3
4
|
export async function saveAgentNote(content) {
|
|
4
|
-
const folderId =
|
|
5
|
+
const folderId = config.DOC_MCP_DRIVE_FOLDER_ID;
|
|
5
6
|
if (!folderId) {
|
|
6
|
-
return {
|
|
7
|
+
return {
|
|
8
|
+
success: false,
|
|
9
|
+
error: "DOC_MCP_DRIVE_FOLDER_ID is not configured.",
|
|
10
|
+
};
|
|
7
11
|
}
|
|
8
12
|
try {
|
|
9
|
-
// We use folderId as the "projectId" parameter for vector-db namespace
|
|
10
13
|
await upsertProjectDocument(folderId, content, {
|
|
11
14
|
source: "agent",
|
|
12
15
|
});
|
|
13
|
-
return {
|
|
16
|
+
return {
|
|
17
|
+
success: true,
|
|
18
|
+
message: "Successfully stored note in vector memory.",
|
|
19
|
+
};
|
|
14
20
|
}
|
|
15
21
|
catch (err) {
|
|
16
22
|
return { success: false, error: `Failed to store note: ${err.message}` };
|
|
17
23
|
}
|
|
18
24
|
}
|
|
19
25
|
export async function searchKnowledge(query, topK = 3) {
|
|
20
|
-
const folderId =
|
|
26
|
+
const folderId = config.DOC_MCP_DRIVE_FOLDER_ID;
|
|
21
27
|
if (!folderId) {
|
|
22
|
-
return {
|
|
28
|
+
return {
|
|
29
|
+
success: false,
|
|
30
|
+
error: "DOC_MCP_DRIVE_FOLDER_ID is not configured.",
|
|
31
|
+
};
|
|
23
32
|
}
|
|
24
33
|
try {
|
|
25
34
|
// Auto-sync folder state before searching
|
|
@@ -30,10 +39,22 @@ export async function searchKnowledge(query, topK = 3) {
|
|
|
30
39
|
}
|
|
31
40
|
return {
|
|
32
41
|
success: true,
|
|
33
|
-
results: results.map((r) =>
|
|
42
|
+
results: results.map((r) => {
|
|
43
|
+
let title = "Unknown Source";
|
|
44
|
+
if (r.metadata) {
|
|
45
|
+
try {
|
|
46
|
+
const metaObj = JSON.parse(r.metadata);
|
|
47
|
+
if (metaObj.title)
|
|
48
|
+
title = metaObj.title;
|
|
49
|
+
}
|
|
50
|
+
catch (e) { }
|
|
51
|
+
}
|
|
52
|
+
return `[File: ${title} | File ID: ${r.file_id || 'N/A'}]\n${r.text}`;
|
|
53
|
+
}).join("\n\n---\n\n"),
|
|
34
54
|
};
|
|
35
55
|
}
|
|
36
56
|
catch (err) {
|
|
57
|
+
console.error("searchKnowledge outer catch:", err.message);
|
|
37
58
|
return { success: false, error: `Failed to search: ${err.message}` };
|
|
38
59
|
}
|
|
39
60
|
}
|
package/package.json
CHANGED
package/src/db/vector.ts
CHANGED
|
@@ -28,7 +28,7 @@ export async function initVectorDB() {
|
|
|
28
28
|
},
|
|
29
29
|
});
|
|
30
30
|
await client.createPayloadIndex(COLLECTION_NAME, {
|
|
31
|
-
field_name: "
|
|
31
|
+
field_name: "folderId",
|
|
32
32
|
field_schema: "keyword",
|
|
33
33
|
});
|
|
34
34
|
await client.createPayloadIndex(COLLECTION_NAME, {
|
|
@@ -70,7 +70,7 @@ export async function embedText(text: string): Promise<number[]> {
|
|
|
70
70
|
return json.data[0].embedding;
|
|
71
71
|
}
|
|
72
72
|
|
|
73
|
-
export async function upsertProjectDocument(
|
|
73
|
+
export async function upsertProjectDocument(folderId: string, text: string, metadata: Record<string, any> = {}): Promise<void> {
|
|
74
74
|
await initVectorDB();
|
|
75
75
|
if (!client) throw new Error("Qdrant not initialized");
|
|
76
76
|
|
|
@@ -83,7 +83,7 @@ export async function upsertProjectDocument(projectId: string, text: string, met
|
|
|
83
83
|
id: uuidv4(),
|
|
84
84
|
vector: vector,
|
|
85
85
|
payload: {
|
|
86
|
-
|
|
86
|
+
folderId,
|
|
87
87
|
text,
|
|
88
88
|
source: metadata.source || "user",
|
|
89
89
|
file_id: metadata.file_id || null,
|
|
@@ -95,62 +95,67 @@ export async function upsertProjectDocument(projectId: string, text: string, met
|
|
|
95
95
|
]
|
|
96
96
|
});
|
|
97
97
|
|
|
98
|
-
console.error(`Upserted document chunk for
|
|
98
|
+
console.error(`Upserted document chunk for folder ${folderId}`);
|
|
99
99
|
}
|
|
100
100
|
|
|
101
|
-
export async function searchProjectMemory(
|
|
101
|
+
export async function searchProjectMemory(folderId: string, query: string, topK: number = 3): Promise<any[]> {
|
|
102
102
|
await initVectorDB();
|
|
103
103
|
if (!client) throw new Error("Qdrant not initialized");
|
|
104
104
|
|
|
105
|
-
|
|
106
|
-
|
|
107
|
-
|
|
108
|
-
|
|
109
|
-
|
|
110
|
-
|
|
111
|
-
|
|
112
|
-
|
|
113
|
-
|
|
114
|
-
|
|
115
|
-
|
|
116
|
-
|
|
105
|
+
try {
|
|
106
|
+
const queryVector = await embedText(query);
|
|
107
|
+
|
|
108
|
+
const results = await client.search(COLLECTION_NAME, {
|
|
109
|
+
vector: queryVector,
|
|
110
|
+
limit: topK,
|
|
111
|
+
with_payload: true,
|
|
112
|
+
filter: {
|
|
113
|
+
must: [
|
|
114
|
+
{
|
|
115
|
+
key: "folderId",
|
|
116
|
+
match: {
|
|
117
|
+
value: folderId
|
|
118
|
+
}
|
|
117
119
|
}
|
|
118
|
-
|
|
119
|
-
|
|
120
|
-
}
|
|
121
|
-
});
|
|
120
|
+
]
|
|
121
|
+
}
|
|
122
|
+
});
|
|
122
123
|
|
|
123
|
-
|
|
124
|
-
|
|
125
|
-
|
|
126
|
-
|
|
127
|
-
|
|
128
|
-
|
|
124
|
+
// Map to match LanceDB format expected by other tools
|
|
125
|
+
return results.map(r => ({
|
|
126
|
+
id: r.id,
|
|
127
|
+
vector: r.vector,
|
|
128
|
+
...r.payload
|
|
129
|
+
}));
|
|
130
|
+
} catch (err: any) {
|
|
131
|
+
console.error("Qdrant search error:", err.message);
|
|
132
|
+
return [];
|
|
133
|
+
}
|
|
129
134
|
}
|
|
130
135
|
|
|
131
|
-
export async function deleteProjectDocument(
|
|
136
|
+
export async function deleteProjectDocument(folderId: string, fileId: string): Promise<void> {
|
|
132
137
|
await initVectorDB();
|
|
133
138
|
if (!client) return;
|
|
134
139
|
|
|
135
140
|
await client.delete(COLLECTION_NAME, {
|
|
136
141
|
filter: {
|
|
137
142
|
must: [
|
|
138
|
-
{ key: "
|
|
143
|
+
{ key: "folderId", match: { value: folderId } },
|
|
139
144
|
{ key: "file_id", match: { value: fileId } }
|
|
140
145
|
]
|
|
141
146
|
}
|
|
142
147
|
});
|
|
143
|
-
console.error(`Deleted old chunks from Qdrant for ${
|
|
148
|
+
console.error(`Deleted old chunks from Qdrant for ${folderId} / ${fileId}`);
|
|
144
149
|
}
|
|
145
150
|
|
|
146
|
-
export async function checkProjectDocumentExists(
|
|
151
|
+
export async function checkProjectDocumentExists(folderId: string, fileId: string): Promise<boolean> {
|
|
147
152
|
await initVectorDB();
|
|
148
153
|
if (!client) return false;
|
|
149
154
|
|
|
150
155
|
const res = await client.count(COLLECTION_NAME, {
|
|
151
156
|
filter: {
|
|
152
157
|
must: [
|
|
153
|
-
{ key: "
|
|
158
|
+
{ key: "folderId", match: { value: folderId } },
|
|
154
159
|
{ key: "file_id", match: { value: fileId } }
|
|
155
160
|
]
|
|
156
161
|
}
|
|
@@ -158,14 +163,14 @@ export async function checkProjectDocumentExists(projectId: string, fileId: stri
|
|
|
158
163
|
return res.count > 0;
|
|
159
164
|
}
|
|
160
165
|
|
|
161
|
-
export async function getProjectDocumentMetadata(
|
|
166
|
+
export async function getProjectDocumentMetadata(folderId: string): Promise<Record<string, string>> {
|
|
162
167
|
await initVectorDB();
|
|
163
168
|
if (!client) return {};
|
|
164
169
|
|
|
165
170
|
const res = await client.scroll(COLLECTION_NAME, {
|
|
166
171
|
filter: {
|
|
167
172
|
must: [
|
|
168
|
-
{ key: "
|
|
173
|
+
{ key: "folderId", match: { value: folderId } },
|
|
169
174
|
{ key: "source", match: { value: "google_drive" } }
|
|
170
175
|
]
|
|
171
176
|
},
|
package/src/mcp-server.ts
CHANGED
|
@@ -27,16 +27,20 @@ const server = new McpServer({
|
|
|
27
27
|
server.registerTool(
|
|
28
28
|
"list_drive_files",
|
|
29
29
|
{
|
|
30
|
-
description: "List and search for Google Drive documents in
|
|
30
|
+
description: "List and search for Google Drive documents and subfolders in a specific folder.",
|
|
31
31
|
inputSchema: {
|
|
32
32
|
keyword: z
|
|
33
33
|
.string()
|
|
34
34
|
.optional()
|
|
35
35
|
.describe("Optional keyword to search for in document titles"),
|
|
36
|
+
targetFolderId: z
|
|
37
|
+
.string()
|
|
38
|
+
.optional()
|
|
39
|
+
.describe("Optional Google Drive folder ID to list contents from. Defaults to the root knowledge folder."),
|
|
36
40
|
},
|
|
37
41
|
},
|
|
38
|
-
async ({ keyword }) => {
|
|
39
|
-
const res = await listDriveFiles(keyword);
|
|
42
|
+
async ({ keyword, targetFolderId }) => {
|
|
43
|
+
const res = await listDriveFiles(keyword, targetFolderId);
|
|
40
44
|
if (!res.success) {
|
|
41
45
|
return {
|
|
42
46
|
content: [{ type: "text", text: `Error: ${res.error}` }],
|
package/src/tools/driveTools.ts
CHANGED
|
@@ -31,8 +31,8 @@ function getDriveClient() {
|
|
|
31
31
|
return google.drive({ version: "v3", auth });
|
|
32
32
|
}
|
|
33
33
|
|
|
34
|
-
export async function listDriveFiles(keyword?: string) {
|
|
35
|
-
const folderId =
|
|
34
|
+
export async function listDriveFiles(keyword?: string, targetFolderId?: string) {
|
|
35
|
+
const folderId = targetFolderId || config.DOC_MCP_DRIVE_FOLDER_ID;
|
|
36
36
|
if (!folderId) {
|
|
37
37
|
return {
|
|
38
38
|
success: false,
|
|
@@ -42,7 +42,7 @@ export async function listDriveFiles(keyword?: string) {
|
|
|
42
42
|
|
|
43
43
|
try {
|
|
44
44
|
const drive = getDriveClient();
|
|
45
|
-
let q = "mimeType = 'application/vnd.google-apps.document'";
|
|
45
|
+
let q = "(mimeType = 'application/vnd.google-apps.document' or mimeType = 'application/vnd.google-apps.folder') and trashed = false";
|
|
46
46
|
q = `'${folderId}' in parents and ${q}`;
|
|
47
47
|
|
|
48
48
|
if (keyword) {
|
|
@@ -51,7 +51,7 @@ export async function listDriveFiles(keyword?: string) {
|
|
|
51
51
|
|
|
52
52
|
const res = await drive.files.list({
|
|
53
53
|
q,
|
|
54
|
-
fields: "files(id, name, description)",
|
|
54
|
+
fields: "files(id, name, description, mimeType)",
|
|
55
55
|
spaces: "drive",
|
|
56
56
|
pageSize: 50,
|
|
57
57
|
supportsAllDrives: true,
|
|
@@ -112,12 +112,12 @@ export async function syncSingleDocument(fileId: string, folderId: string) {
|
|
|
112
112
|
}
|
|
113
113
|
return { synced: true, content, driveModifiedTime };
|
|
114
114
|
}
|
|
115
|
-
|
|
115
|
+
|
|
116
116
|
return { synced: false, driveModifiedTime };
|
|
117
117
|
}
|
|
118
118
|
|
|
119
119
|
export async function readDriveDocument(fileId: string) {
|
|
120
|
-
const folderId =
|
|
120
|
+
const folderId = config.DOC_MCP_DRIVE_FOLDER_ID;
|
|
121
121
|
if (!folderId) {
|
|
122
122
|
return {
|
|
123
123
|
success: false,
|
|
@@ -127,7 +127,7 @@ export async function readDriveDocument(fileId: string) {
|
|
|
127
127
|
|
|
128
128
|
try {
|
|
129
129
|
const result = await syncSingleDocument(fileId, folderId);
|
|
130
|
-
|
|
130
|
+
|
|
131
131
|
// If not synced just now, we need to fetch content to return to the user
|
|
132
132
|
let content = result.content;
|
|
133
133
|
if (!content) {
|
|
@@ -159,19 +159,33 @@ export async function readDriveDocument(fileId: string) {
|
|
|
159
159
|
export async function syncFolderState(folderId: string) {
|
|
160
160
|
try {
|
|
161
161
|
const drive = getDriveClient();
|
|
162
|
-
let q = "mimeType = 'application/vnd.google-apps.document'";
|
|
163
|
-
q = `'${folderId}' in parents and ${q}`;
|
|
164
162
|
|
|
165
|
-
|
|
166
|
-
|
|
167
|
-
|
|
168
|
-
|
|
169
|
-
|
|
170
|
-
|
|
171
|
-
|
|
172
|
-
|
|
163
|
+
async function getAllDocumentsFlat(): Promise<any[]> {
|
|
164
|
+
let allDocs: any[] = [];
|
|
165
|
+
let pageToken: string | undefined = undefined;
|
|
166
|
+
|
|
167
|
+
do {
|
|
168
|
+
const docsRes: any = await drive.files.list({
|
|
169
|
+
// Chú ý: Đéo check parentId nữa, gom sạch sành sanh mọi file .doc mà Service Account nhìn thấy
|
|
170
|
+
q: `mimeType = 'application/vnd.google-apps.document' and trashed = false`,
|
|
171
|
+
fields: "nextPageToken, files(id, name, modifiedTime)",
|
|
172
|
+
spaces: "drive",
|
|
173
|
+
pageSize: 100, // Google API limit mỗi page, tự động nhảy trang nếu nhiều hơn
|
|
174
|
+
pageToken,
|
|
175
|
+
supportsAllDrives: true,
|
|
176
|
+
includeItemsFromAllDrives: true,
|
|
177
|
+
});
|
|
178
|
+
|
|
179
|
+
if (docsRes.data.files) {
|
|
180
|
+
allDocs = allDocs.concat(docsRes.data.files);
|
|
181
|
+
}
|
|
182
|
+
pageToken = docsRes.data.nextPageToken || undefined;
|
|
183
|
+
} while (pageToken);
|
|
184
|
+
|
|
185
|
+
return allDocs;
|
|
186
|
+
}
|
|
173
187
|
|
|
174
|
-
const driveFiles =
|
|
188
|
+
const driveFiles = await getAllDocumentsFlat();
|
|
175
189
|
const dbMetaMap = await getProjectDocumentMetadata(folderId);
|
|
176
190
|
|
|
177
191
|
// Sync updated or new files
|
|
@@ -185,7 +199,7 @@ export async function syncFolderState(folderId: string) {
|
|
|
185
199
|
|
|
186
200
|
// Delete removed files from DB
|
|
187
201
|
for (const dbFileId of Object.keys(dbMetaMap)) {
|
|
188
|
-
if (!driveFiles.find(f => f.id === dbFileId)) {
|
|
202
|
+
if (!driveFiles.find((f) => f.id === dbFileId)) {
|
|
189
203
|
await deleteProjectDocument(folderId, dbFileId);
|
|
190
204
|
}
|
|
191
205
|
}
|
|
@@ -1,27 +1,35 @@
|
|
|
1
|
+
import { config } from "../config.js";
|
|
1
2
|
import { upsertProjectDocument, searchProjectMemory } from "../db/vector.js";
|
|
2
3
|
import { syncFolderState } from "./driveTools.js";
|
|
3
4
|
|
|
4
5
|
export async function saveAgentNote(content: string) {
|
|
5
|
-
const folderId =
|
|
6
|
+
const folderId = config.DOC_MCP_DRIVE_FOLDER_ID;
|
|
6
7
|
if (!folderId) {
|
|
7
|
-
return {
|
|
8
|
+
return {
|
|
9
|
+
success: false,
|
|
10
|
+
error: "DOC_MCP_DRIVE_FOLDER_ID is not configured.",
|
|
11
|
+
};
|
|
8
12
|
}
|
|
9
|
-
|
|
10
13
|
try {
|
|
11
|
-
// We use folderId as the "projectId" parameter for vector-db namespace
|
|
12
14
|
await upsertProjectDocument(folderId, content, {
|
|
13
15
|
source: "agent",
|
|
14
16
|
});
|
|
15
|
-
return {
|
|
17
|
+
return {
|
|
18
|
+
success: true,
|
|
19
|
+
message: "Successfully stored note in vector memory.",
|
|
20
|
+
};
|
|
16
21
|
} catch (err: any) {
|
|
17
22
|
return { success: false, error: `Failed to store note: ${err.message}` };
|
|
18
23
|
}
|
|
19
24
|
}
|
|
20
25
|
|
|
21
26
|
export async function searchKnowledge(query: string, topK: number = 3) {
|
|
22
|
-
const folderId =
|
|
27
|
+
const folderId = config.DOC_MCP_DRIVE_FOLDER_ID;
|
|
23
28
|
if (!folderId) {
|
|
24
|
-
return {
|
|
29
|
+
return {
|
|
30
|
+
success: false,
|
|
31
|
+
error: "DOC_MCP_DRIVE_FOLDER_ID is not configured.",
|
|
32
|
+
};
|
|
25
33
|
}
|
|
26
34
|
|
|
27
35
|
try {
|
|
@@ -36,7 +44,18 @@ export async function searchKnowledge(query: string, topK: number = 3) {
|
|
|
36
44
|
|
|
37
45
|
return {
|
|
38
46
|
success: true,
|
|
39
|
-
results: results
|
|
47
|
+
results: results
|
|
48
|
+
.map((r: any) => {
|
|
49
|
+
let title = "Unknown Source";
|
|
50
|
+
if (r.metadata) {
|
|
51
|
+
try {
|
|
52
|
+
const metaObj = JSON.parse(r.metadata);
|
|
53
|
+
if (metaObj.title) title = metaObj.title;
|
|
54
|
+
} catch (e) {}
|
|
55
|
+
}
|
|
56
|
+
return `[File: ${title} | File ID: ${r.file_id || "N/A"}]\n${r.text}`;
|
|
57
|
+
})
|
|
58
|
+
.join("\n\n---\n\n"),
|
|
40
59
|
};
|
|
41
60
|
} catch (err: any) {
|
|
42
61
|
return { success: false, error: `Failed to search: ${err.message}` };
|
package/src/hooks/driveSync.ts
DELETED
|
@@ -1,108 +0,0 @@
|
|
|
1
|
-
import {
|
|
2
|
-
getProjectDocumentMetadata,
|
|
3
|
-
deleteProjectDocument,
|
|
4
|
-
upsertProjectDocument,
|
|
5
|
-
} from "../db/vector.js";
|
|
6
|
-
import { RecursiveCharacterTextSplitter } from "@langchain/textsplitters";
|
|
7
|
-
// Remove childLogger dependency to fully decouple from core
|
|
8
|
-
// import { childLogger } from "@workspace/core";
|
|
9
|
-
import { config } from "../config.js";
|
|
10
|
-
|
|
11
|
-
const log = {
|
|
12
|
-
info: (obj: any, msg: string) => console.log(`[driveSync] ${msg}`, obj),
|
|
13
|
-
error: (obj: any, msg: string) => console.error(`[driveSync] ${msg}`, obj)
|
|
14
|
-
};
|
|
15
|
-
|
|
16
|
-
export async function syncProjectDriveFiles(
|
|
17
|
-
projectId: string,
|
|
18
|
-
onSyncMessage?: (msg: string) => void,
|
|
19
|
-
): Promise<void> {
|
|
20
|
-
const dbMeta = await getProjectDocumentMetadata(projectId);
|
|
21
|
-
const fileIds = Object.keys(dbMeta);
|
|
22
|
-
|
|
23
|
-
if (fileIds.length === 0) return;
|
|
24
|
-
|
|
25
|
-
log.info({ projectId, fileCount: fileIds.length }, "Checking Drive files for updates...");
|
|
26
|
-
|
|
27
|
-
const { google } = await import("googleapis");
|
|
28
|
-
|
|
29
|
-
const clientEmail = config.DOC_MCP_GOOGLE_CLIENT_EMAIL;
|
|
30
|
-
let privateKey = config.DOC_MCP_GOOGLE_PRIVATE_KEY;
|
|
31
|
-
if (!clientEmail || !privateKey) return;
|
|
32
|
-
if (privateKey.startsWith('"') && privateKey.endsWith('"')) {
|
|
33
|
-
privateKey = privateKey.slice(1, -1);
|
|
34
|
-
}
|
|
35
|
-
privateKey = privateKey.replace(/\\n/g, "\n");
|
|
36
|
-
|
|
37
|
-
const auth = new google.auth.JWT({
|
|
38
|
-
email: clientEmail,
|
|
39
|
-
key: privateKey,
|
|
40
|
-
scopes: ["https://www.googleapis.com/auth/drive.readonly"],
|
|
41
|
-
});
|
|
42
|
-
|
|
43
|
-
const drive = google.drive({ version: "v3", auth });
|
|
44
|
-
|
|
45
|
-
let updatedCount = 0;
|
|
46
|
-
for (const fileId of fileIds) {
|
|
47
|
-
try {
|
|
48
|
-
const fileInfo = await drive.files.get({
|
|
49
|
-
fileId: fileId,
|
|
50
|
-
fields: "id, name, modifiedTime, trashed",
|
|
51
|
-
supportsAllDrives: true,
|
|
52
|
-
});
|
|
53
|
-
|
|
54
|
-
if (fileInfo.data.trashed) {
|
|
55
|
-
log.info({ fileId, projectId }, "File trashed on Drive, deleting from VectorDB...");
|
|
56
|
-
await deleteProjectDocument(projectId, fileId);
|
|
57
|
-
continue;
|
|
58
|
-
}
|
|
59
|
-
|
|
60
|
-
const driveModifiedTime = fileInfo.data.modifiedTime || "";
|
|
61
|
-
const dbModifiedTime = dbMeta[fileId];
|
|
62
|
-
|
|
63
|
-
if (driveModifiedTime !== dbModifiedTime) {
|
|
64
|
-
if (onSyncMessage) {
|
|
65
|
-
onSyncMessage(`🔄 Syncing updated file: ${fileInfo.data.name}...`);
|
|
66
|
-
}
|
|
67
|
-
|
|
68
|
-
log.info({ fileId, projectId }, "File updated on Drive, syncing...");
|
|
69
|
-
await deleteProjectDocument(projectId, fileId);
|
|
70
|
-
|
|
71
|
-
const res = await drive.files.export({
|
|
72
|
-
fileId: fileId,
|
|
73
|
-
mimeType: "text/plain",
|
|
74
|
-
});
|
|
75
|
-
|
|
76
|
-
const content = res.data;
|
|
77
|
-
if (typeof content === "string" && content.trim() !== "") {
|
|
78
|
-
const splitter = new RecursiveCharacterTextSplitter({
|
|
79
|
-
chunkSize: config.CHUNK_SIZE,
|
|
80
|
-
chunkOverlap: config.CHUNK_OVERLAP,
|
|
81
|
-
});
|
|
82
|
-
const chunks = await splitter.splitText(content);
|
|
83
|
-
|
|
84
|
-
for (const chunk of chunks) {
|
|
85
|
-
await upsertProjectDocument(projectId, chunk, {
|
|
86
|
-
title: fileInfo.data.name,
|
|
87
|
-
file_id: fileId,
|
|
88
|
-
source: "google_drive",
|
|
89
|
-
modified_time: driveModifiedTime,
|
|
90
|
-
});
|
|
91
|
-
}
|
|
92
|
-
updatedCount++;
|
|
93
|
-
}
|
|
94
|
-
}
|
|
95
|
-
} catch (err: any) {
|
|
96
|
-
if (err.code === 404) {
|
|
97
|
-
log.info({ fileId, projectId }, "File not found on Drive, deleting from VectorDB...");
|
|
98
|
-
await deleteProjectDocument(projectId, fileId);
|
|
99
|
-
} else {
|
|
100
|
-
log.error({ fileId, err: err.message }, "Error syncing drive file");
|
|
101
|
-
}
|
|
102
|
-
}
|
|
103
|
-
}
|
|
104
|
-
|
|
105
|
-
if (updatedCount > 0 && onSyncMessage) {
|
|
106
|
-
onSyncMessage(`✅ Synced ${updatedCount} files from Google Drive.`);
|
|
107
|
-
}
|
|
108
|
-
}
|