file citation mayheming

Author: Veikkosuhonen

chunkingTest.js

@@ -1,39 +1,85 @@
 import { Box, Paper, Typography } from '@mui/material'
-import { FileCitation } from '../../../shared/types'
+import { CourseAssistant, FileCitation } from '../../../shared/types'
 import { Message } from '../../types'
+import { useQuery } from '@tanstack/react-query'
+import { useCourseAssistant } from '../../hooks/useCourseAssistant'
+import { useParams } from 'react-router-dom'
+
+const useFileCitationText = (citation: FileCitation, vectorStoreId: string) => {
+  const { data, ...rest } = useQuery({
+    queryKey: ['file-citation-content', vectorStoreId, citation.file_id],
+    queryFn: async () => {
+      const response = await fetch(`/api/files/${vectorStoreId}/${citation.file_id}`)
+      if (!response.ok) {
+        throw new Error('Failed to fetch file text')
+      }
+      return response.text()
+    },
+    retry: false,
+  })
+
+  return {
+    text: data,
+    ...rest,
+  }
+}
+
+const Citation = ({ citation, courseAssistant }: { citation: FileCitation, courseAssistant: CourseAssistant }) => {
+  const { text, isLoading, error } = useFileCitationText(citation, courseAssistant.vector_store_id)
 
-const Citation = ({ citation }: { citation: FileCitation }) => {
   return (
     <Box>
       <Typography variant="body2" color="textSecondary">
         {citation.filename} (Index: {citation.index})
       </Typography>
+      {isLoading ? (
+        <Typography variant="body2" color="textSecondary">
+          Loading citation text...
+        </Typography>
+      ) : error ? (
+        <Typography variant="body2" color="error">
+          Error loading citation text: {error.message}
+        </Typography>
+      ) : (
+        <Typography variant="body2">{text}</Typography>
+      )}
     </Box>
   )
 }
 
-const MessageCitations = ({ citations }: { citations: FileCitation[] }) => {
+const MessageCitations = ({ citations, courseAssistant }: { citations: FileCitation[]; courseAssistant: CourseAssistant }) => {
   return (
     <Box sx={{ mt: 1, fontSize: '0.875rem', color: 'gray' }}>
       Citations:
       {citations.map((citation, index) => (
-        <Citation key={index} citation={citation} />
+        <Citation key={index} citation={citation} courseAssistant={courseAssistant} />
       ))}
     </Box>
   )
 }
 
 export const CitationsBox = ({ messages, citations }: { messages: Message[]; citations: FileCitation[] }) => {
+  const { courseId } = useParams()
+  const { courseAssistant, isLoading } = useCourseAssistant(courseId)
+
   const messageCitations = [...messages.map((message) => (Array.isArray(message.citations) ? message.citations : [])), citations]
 
   console.log('CitationsBox messageCitations', messageCitations)
 
   return (
     <Paper>
       <Box sx={{ width: 300, padding: 2 }}>
-        {messageCitations.map((c, index) => (
-          <MessageCitations key={index} citations={c} />
-        ))}
+        {courseAssistant ? (
+          messageCitations.map((c, index) => <MessageCitations key={index} citations={c} courseAssistant={courseAssistant} />)
+        ) : isLoading ? (
+          <Typography variant="body2" color="textSecondary">
+            Loading course assistant...
+          </Typography>
+        ) : (
+          <Typography variant="body2" color="error">
+            No course assistant found for this course.
+          </Typography>
+        )}
       </Box>
     </Paper>
   )

package-lock.json

@@ -0,0 +1,24 @@
+import { useQuery } from '@tanstack/react-query'
+import { CourseAssistant } from '../../shared/types'
+
+export const useCourseAssistant = (courseId: string | null) => {
+  const queryKey = ['courseAssistant', courseId]
+
+  const { data: courseAssistant, ...rest } = useQuery<CourseAssistant>({
+    queryKey,
+    queryFn: async () => {
+      const response = await fetch(`/api/assistants/for-course/${courseId}`)
+      if (!response.ok) {
+        throw new Error('Failed to fetch course assistant')
+      }
+      return response.json()
+    },
+    enabled: !!courseId,
+    retry: false,
+  })
+
+  return {
+    courseAssistant,
+    ...rest,
+  }
+}

src/client/components/ChatV2/CitationsBox.tsx

@@ -0,0 +1,19 @@
+import { Router } from 'express'
+import { courseAssistants } from '../util/azure/courseAssistants'
+
+const router = Router()
+
+router.get('/for-course/:courseId', async (req, res) => {
+  const { courseId } = req.params
+
+  const courseAssistant = courseAssistants.find((assistant) => assistant.course_id === courseId)
+
+  if (!courseAssistant) {
+    res.status(404).send('Course assistant not found')
+    return
+  }
+
+  res.send(courseAssistant)
+})
+
+export default router

src/client/hooks/useCourseAssistant.ts

@@ -0,0 +1,36 @@
+import { Router } from 'express'
+import { getAzureOpenAIClient } from '../util/azure/client'
+import { z } from 'zod'
+
+const router = Router()
+
+const GetFileParamsSchema = z.object({
+  vectorStoreId: z.string(),
+  fileId: z.string(),
+})
+
+router.get('/:vectorStoreId/:fileId', async (req, res) => {
+  const { vectorStoreId, fileId } = GetFileParamsSchema.parse(req.params)
+  const client = getAzureOpenAIClient('curredev4omini')
+
+  const asd = await client.vectorStores.files.retrieve(vectorStoreId, fileId)
+  console.log('asd', asd)
+  if (!asd) {
+    res.status(404).send('File not found')
+    return
+  }
+
+  const fileRes = await client.vectorStores.files.content(vectorStoreId, fileId)
+  console.log('fileRes', fileRes)
+
+  const content = fileRes.data?.[0]?.text
+
+  if (!content) {
+    res.status(404).send('File not found or empty')
+    return
+  }
+
+  res.send(content)
+})
+
+export default router

src/server/routes/assistants.ts

@@ -17,6 +17,8 @@ import emailRouter from './email'
 import adminRouter from './admin'
 import facultyRouter from './faculty'
 import infoTextRouter from './infotext'
+import filesRouter from './files'
+import assistantRouter from './assistants'
 
 const router = express()
 
@@ -36,6 +38,8 @@ router.get('/ping', (_, res) => {
 
 router.use('/ai', openaiRouter)
 router.use('/rag', ragRouter)
+router.use('/files', filesRouter)
+router.use('/assistants', assistantRouter)
 router.use('/users', userRouter)
 router.use('/chatinstances', chatInstancesRouter)
 router.use('/courses', courseRouter)

src/server/routes/files.ts

@@ -13,10 +13,10 @@ import { AzureOpenAI } from 'openai'
 import { Stream } from 'openai/streaming'
 import { FileSearchTool, FunctionTool, ResponseInput, ResponseInputItem, ResponseStreamEvent, ResponseTextAnnotationDeltaEvent } from 'openai/resources/responses/responses'
 
-import { courseAssistants, type CourseAssistant } from './courseAssistants'
+import { courseAssistants } from './courseAssistants'
 import { createFileSearchTool } from './util'
 
-import type { FileCitation, ResponseStreamEventData } from '../../../shared/types'
+import type { CourseAssistant, FileCitation, ResponseStreamEventData } from '../../../shared/types'
 
 const endpoint = `https://${AZURE_RESOURCE}.openai.azure.com/`
 
@@ -28,38 +28,37 @@ export const getAzureOpenAIClient = (deployment: string) =>
     endpoint,
   })
 
-const client = getAzureOpenAIClient(process.env.GPT_4O)
+const client = getAzureOpenAIClient(process.env.GPT_4O_MINI)
 
 export class ResponsesClient {
   model: string
   instructions: string
   tools: FileSearchTool[]
+  courseAssistant: CourseAssistant
 
   constructor({ model, courseId }: { model: string; courseId?: string }) {
     const deploymentId = validModels.find((m) => m.name === model)?.deployment
 
     if (!deploymentId) throw new Error(`Invalid model: ${model}, not one of ${validModels.map((m) => m.name).join(', ')}`)
 
-    let courseAssistant: CourseAssistant
-
     if (courseId) {
-      courseAssistant = courseAssistants.find((assistant) => assistant.course_id === courseId)
+      this.courseAssistant = courseAssistants.find((assistant) => assistant.course_id === courseId)
 
-      if (!courseAssistant) throw new Error(`No course assistant found for course ID: ${courseId}`)
+      if (!this.courseAssistant) throw new Error(`No course assistant found for course ID: ${courseId}`)
     } else {
-      courseAssistant = courseAssistants.find((assistant) => assistant.name === 'default')
+      this.courseAssistant = courseAssistants.find((assistant) => assistant.name === 'default')
     }
 
     const fileSearchTool = courseId
       ? [
           createFileSearchTool({
-            vectorStoreId: courseAssistant.vector_store_id,
+            vectorStoreId: this.courseAssistant.vector_store_id,
           }),
         ]
       : [] // needs to retrun empty array for null
 
     this.model = deploymentId
-    this.instructions = courseAssistant.assistant_instruction
+    this.instructions = this.courseAssistant.assistant_instruction
     this.tools = fileSearchTool
   }
 

src/server/routes/index.ts

@@ -1,9 +1,4 @@
-export interface CourseAssistant {
-  course_id: string | null
-  name: string
-  assistant_instruction: string
-  vector_store_id: string | null
-}
+import { CourseAssistant } from "../../../shared/types";
 
 export const courseAssistants = [
   {