PDF importer (#7610)

Signed-off-by: Denis Bykhov <[email protected]>

.vscode/launch.json

@@ -584,7 +584,8 @@
         "PASSWORD": "password",
         "AVATAR_PATH": "./assets/avatar.png",
         "AVATAR_CONTENT_TYPE": ".png",
-        "LOVE_ENDPOINT": "http://localhost:8096"
+        "STORAGE_CONFIG": "minio|localhost?accessKey=minioadmin&secretKey=minioadmin",
+        "LOVE_ENDPOINT": "http://localhost:8096",
       },
       "runtimeArgs": ["--nolazy", "-r", "ts-node/register"],
       "sourceMaps": true,

services/ai-bot/pod-ai-bot/package.json

@@ -46,6 +46,7 @@
     "eslint-plugin-n": "^15.4.0",
     "eslint-plugin-node": "^11.1.0",
     "eslint-plugin-promise": "^6.1.1",
+    "@types/uuid": "^8.3.1",
     "jest": "^29.7.0",
     "prettier": "^3.1.0",
     "ts-jest": "^29.1.1",
@@ -56,6 +57,8 @@
     "@hcengineering/account": "^0.6.0",
     "@hcengineering/ai-bot": "^0.6.0",
     "@hcengineering/analytics-collector": "^0.6.0",
+    "@hcengineering/document": "^0.6.0",
+    "@hcengineering/attachment": "^0.6.14",
     "@hcengineering/chunter": "^0.6.20",
     "@hcengineering/client": "^0.6.18",
     "@hcengineering/client-resources": "^0.6.27",
@@ -72,6 +75,8 @@
     "@hcengineering/server-token": "^0.6.11",
     "@hcengineering/setting": "^0.6.17",
     "@hcengineering/text": "^0.6.5",
+    "@hcengineering/rank": "^0.6.4",
+    "@hcengineering/server-storage": "^0.6.0",
     "@hcengineering/workbench": "^0.6.16",
     "@hcengineering/love": "^0.6.0",
     "cors": "^2.8.5",
@@ -80,6 +85,7 @@
     "fast-equals": "^5.0.1",
     "form-data": "^4.0.0",
     "js-tiktoken": "^1.0.14",
+    "uuid": "^8.3.2",
     "mongodb": "^6.12.0",
     "openai": "^4.56.0",
     "ws": "^8.18.0"

services/ai-bot/pod-ai-bot/src/config.ts

@@ -36,6 +36,7 @@ interface Config {
   MaxHistoryRecords: number
   Port: number
   LoveEndpoint: string
+  DataLabApiKey: string
 }
 
 const parseNumber = (str: string | undefined): number | undefined => (str !== undefined ? Number(str) : undefined)
@@ -61,7 +62,8 @@ const config: Config = (() => {
     MaxContentTokens: parseNumber(process.env.MAX_CONTENT_TOKENS) ?? 128 * 100,
     MaxHistoryRecords: parseNumber(process.env.MAX_HISTORY_RECORDS) ?? 500,
     Port: parseNumber(process.env.PORT) ?? 4010,
-    LoveEndpoint: process.env.LOVE_ENDPOINT ?? ''
+    LoveEndpoint: process.env.LOVE_ENDPOINT ?? '',
+    DataLabApiKey: process.env.DATALAB_API_KEY ?? ''
   }
 
   const missingEnv = (Object.keys(params) as Array<keyof Config>).filter((key) => params[key] === undefined)

services/ai-bot/pod-ai-bot/src/controller.ts

@@ -29,22 +29,24 @@ import {
   TranslateRequest,
   TranslateResponse
 } from '@hcengineering/ai-bot'
+import { Markup, MeasureContext, Ref, WorkspaceId } from '@hcengineering/core'
+import { Room } from '@hcengineering/love'
 import { WorkspaceInfoRecord } from '@hcengineering/server-ai-bot'
 import { getTransactorEndpoint } from '@hcengineering/server-client'
 import { generateToken } from '@hcengineering/server-token'
-import OpenAI from 'openai'
-import { encodingForModel } from 'js-tiktoken'
 import { htmlToMarkup, markupToHTML } from '@hcengineering/text'
-import { Markup, MeasureContext, Ref, WorkspaceId } from '@hcengineering/core'
-import { Room } from '@hcengineering/love'
+import { encodingForModel } from 'js-tiktoken'
+import OpenAI from 'openai'
 
-import { WorkspaceClient } from './workspace/workspaceClient'
+import { StorageAdapter } from '@hcengineering/server-core'
+import { buildStorageFromConfig, storageConfigFromEnv } from '@hcengineering/server-storage'
 import config from './config'
 import { DbStorage } from './storage'
-import { SupportWsClient } from './workspace/supportWsClient'
 import { AIReplyTransferData } from './types'
 import { tryAssignToWorkspace } from './utils/account'
 import { translateHtml } from './utils/openai'
+import { SupportWsClient } from './workspace/supportWsClient'
+import { WorkspaceClient } from './workspace/workspaceClient'
 
 const CLOSE_INTERVAL_MS = 10 * 60 * 1000 // 10 minutes
 
@@ -54,6 +56,7 @@ export class AIControl {
   private readonly connectingWorkspaces = new Map<string, Promise<void>>()
 
   readonly aiClient?: OpenAI
+  readonly storageAdapter: StorageAdapter
   readonly encoding = encodingForModel(config.OpenAIModel)
 
   supportClient: SupportWsClient | undefined = undefined
@@ -70,6 +73,7 @@ export class AIControl {
         })
         : undefined
     void this.connectSupportWorkspace()
+    this.storageAdapter = buildStorageFromConfig(storageConfigFromEnv())
   }
 
   async getWorkspaceRecord (workspace: string): Promise<WorkspaceInfoRecord> {
@@ -125,10 +129,26 @@ export class AIControl {
     this.ctx.info('Listen workspace: ', { workspace })
 
     if (workspace === config.SupportWorkspace) {
-      return new SupportWsClient(endpoint, token, workspace, this, this.ctx.newChild(workspace, {}), info)
+      return new SupportWsClient(
+        this.storageAdapter,
+        endpoint,
+        token,
+        workspace,
+        this,
+        this.ctx.newChild(workspace, {}),
+        info
+      )
     }
 
-    return new WorkspaceClient(endpoint, token, workspace, this, this.ctx.newChild(workspace, {}), info)
+    return new WorkspaceClient(
+      this.storageAdapter,
+      endpoint,
+      token,
+      workspace,
+      this,
+      this.ctx.newChild(workspace, {}),
+      info
+    )
   }
 
   async initWorkspaceClient (workspace: string): Promise<void> {

services/ai-bot/pod-ai-bot/src/utils/openai.ts

@@ -13,12 +13,13 @@
 // limitations under the License.
 //
 
-import OpenAI from 'openai'
 import { countTokens } from '@hcengineering/openai'
 import { Tiktoken } from 'js-tiktoken'
-
+import OpenAI from 'openai'
 import config from '../config'
 import { HistoryRecord } from '../types'
+import { WorkspaceClient } from '../workspace/workspaceClient'
+import { getTools } from './tools'
 
 export async function translateHtml (client: OpenAI, html: string, lang: string): Promise<string | undefined> {
   const response = await client.chat.completions.create({
@@ -66,6 +67,58 @@ export async function createChatCompletion (
   return undefined
 }
 
+export async function createChatCompletionWithTools (
+  workspaceClient: WorkspaceClient,
+  client: OpenAI,
+  message: OpenAI.ChatCompletionMessageParam,
+  user?: string,
+  history: OpenAI.ChatCompletionMessageParam[] = [],
+  skipCache = true
+): Promise<
+  | {
+    completion: string | undefined
+    usage: number
+  }
+  | undefined
+  > {
+  const opt: OpenAI.RequestOptions = {}
+  if (skipCache) {
+    opt.headers = { 'cf-skip-cache': 'true' }
+  }
+  try {
+    const res = client.beta.chat.completions
+      .runTools(
+        {
+          messages: [
+            {
+              role: 'system',
+              content: 'Use tools if possible, don`t use previous information after success using tool for user request'
+            },
+            ...history,
+            message
+          ],
+          model: config.OpenAIModel,
+          user,
+          tools: getTools(workspaceClient, user)
+        },
+        opt
+      )
+      .on('message', (message) => {
+        console.log(message)
+      })
+    const str = await res.finalContent()
+    const usage = (await res.totalUsage()).completion_tokens
+    return {
+      completion: str ?? undefined,
+      usage
+    }
+  } catch (e) {
+    console.error(e)
+  }
+
+  return undefined
+}
+
 export async function requestSummary (
   aiClient: OpenAI,
   encoding: Tiktoken,