feat: enable content generation without source for slide (#351)

Author: jiankaiii

backend/main.py

@@ -1,7 +1,6 @@
 # Copyright (C) 2025 Intel Corporation
 # SPDX-License-Identifier: Apache-2.0
-
-from fastapi import FastAPI, HTTPException, File, UploadFile
+from fastapi import BackgroundTasks, FastAPI, HTTPException, File, UploadFile
 from pydantic import BaseModel
 import fitz  # PyMuPDF
 from pathlib import Path
@@ -12,11 +11,15 @@
 from generate_image_embedding import generate_image_embedding
 from fastapi.responses import FileResponse, JSONResponse
 from generate_pptx import create_pptx
+from generate_pptx import create_pptx
 from starlette.background import BackgroundTask
 import tempfile
 import imagehash
 from PIL import Image
 import io
+import uuid
+from typing import Dict
+import json
 
 app = FastAPI()
 
@@ -26,22 +29,10 @@
 OUTPUT_DIR.mkdir(parents=True, exist_ok=True)
 
 
-@app.post("/parse")
-async def parse_pdf(file: UploadFile = File(...)):
-    """
-    Endpoint to parse a PDF file uploaded via multipart/form-data.
-    Extracts images, generates captions and embeddings, and returns the data.
-    """
-    temp_file_path = None
+def process_pdf_to_file(job_id: str, pdf_path: str, filename: str):
     try:
-        # Create temp file with delete=False to avoid Windows file locking issues
-        with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as temp_file:
-            temp_file.write(await file.read())
-            temp_file_path = temp_file.name
-
-        print(f"DEBUG : Temporary PDF file created at: {temp_file_path}")
-        # Open the PDF file using PyMuPDF (now works on Windows since file is closed)
-        pdf_file = fitz.open(str(temp_file_path))
+        print(f"Processing job {job_id}")
+        pdf_file = fitz.open(str(pdf_path))
         image_data = []
         image_order = 1
         seen_hashes = set()
@@ -88,29 +79,62 @@ async def parse_pdf(file: UploadFile = File(...)):
 
         # Prepare the response data
         response_data = {
-            "name": file.filename,
+            "name": filename,
             "details": f"Extracted {len(image_data)} images from the PDF.",
             "images": image_data,
             "text": extracted_text,
         }
 
-        return JSONResponse(content=response_data)
+        temp_dir = tempfile.gettempdir()
+        result_path = os.path.join(temp_dir, f"{job_id}.json")
+        with open(result_path, "w") as f:
+            json.dump(response_data, f)
 
     except Exception as e:
-        print(f"Error processing PDF: {e}")
-        raise HTTPException(
-            status_code=500, detail=f"An error occurred while processing the PDF: {e}"
-        )
+        print(f"Error in processing pdf job_id: {job_id}: {e}")
+
     finally:
-        # Clean up temporary file on Windows
-        if temp_file_path and os.path.exists(temp_file_path):
-            try:
-                os.unlink(temp_file_path)
-                print(f"DEBUG: Cleaned up temporary file: {temp_file_path}")
-            except Exception as cleanup_error:
-                print(
-                    f"Warning: Failed to clean up temporary file {temp_file_path}: {cleanup_error}"
-                )
+        try:
+            if os.path.exists(pdf_path):
+                os.remove(pdf_path)
+        except Exception as cleanup_err:
+            print(f"Warning: Failed to remove temporary PDF {pdf_path}: {cleanup_err}")
+
+
+@app.post("/upload")
+async def upload_file(
+    file: UploadFile = File(...), background_tasks: BackgroundTasks = None
+):
+    try:
+        # Generate job ID
+        job_id = str(uuid.uuid4())
+        tmp_dir = tempfile.gettempdir()
+        tmp_path = os.path.join(tmp_dir, f"{job_id}_{file.filename}")
+
+        # Save uploaded file to /tmp
+        with open(tmp_path, "wb") as buffer:
+            shutil.copyfileobj(file.file, buffer)
+
+        # Schedule background PDF processing
+        background_tasks.add_task(process_pdf_to_file, job_id, tmp_path, file.filename)
+
+        return {"jobID": job_id}
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=f"Error uploading file: {e}")
+
+
+@app.get("/result/{job_id}")
+def get_result(job_id: str):
+    temp_dir = tempfile.gettempdir()
+    result_path = os.path.join(temp_dir, f"{job_id}.json")
+    if not os.path.exists(result_path):
+        return JSONResponse(
+            status_code=202, content={"message": "PDF processing not complete yet."}
+        )
+
+    with open(result_path, "r") as f:
+        result = json.load(f)
+        return result
 
 
 class PPTXRequest(BaseModel):

frontend/src/app/api/slide/content-generator.ts

@@ -4,6 +4,7 @@
 import { createOllama } from 'ollama-ai-provider'
 import { type CoreMessage, generateText } from 'ai'
 import type { ClientSource } from '@/lib/types/client-source'
+import type { CourseInfo } from '@/lib/types/course-info-types'
 import type {
   LectureContent,
   AssessmentQuestion,
@@ -42,6 +43,7 @@ export async function generateCourseContent(
   sessionLength: number,
   difficultyLevel: string,
   topicName: string,
+  courseInfo?: CourseInfo,
 ): Promise<LectureContent> {
   try {
     // Check for required environment variables
@@ -55,8 +57,11 @@ export async function generateCourseContent(
 
     // Prepare source content
     console.log('Preparing source content...')
-    const { content: assistantContent, metadata: sourceMetadata } =
-      await prepareSourceContent(selectedSources)
+    const { content: assistantContent, metadata: sourceMetadata } = await prepareSourceContent(
+      selectedSources,
+      topicName,
+      courseInfo,
+    )
 
     // Ensure assistant content fits within context window
     const assistantMessage: CoreMessage = {
@@ -112,7 +117,9 @@ export async function generateCourseContent(
 
     const metadataUserMessage: CoreMessage = {
       role: 'user',
-      content: `Generate the title, learning outcomes, and at least 5-10 key terms for a ${difficultyLevel} level ${contentType} on "${topicName}" based STRICTLY on the provided source materials above.`,
+      content: sourceMetadata.usingCourseContext
+        ? `Generate the title, learning outcomes, and at least 5-10 key terms for a ${difficultyLevel} level ${contentType} on "${topicName}" based on standard academic knowledge and best practices for this subject area.`
+        : `Generate the title, learning outcomes, and at least 5-10 key terms for a ${difficultyLevel} level ${contentType} on "${topicName}" based STRICTLY on the provided source materials above.`,
     }
 
     const metadataMessages = [metadataSystemMessage, assistantMessage, metadataUserMessage]
@@ -137,10 +144,17 @@ export async function generateCourseContent(
     const introSystemPrompt = `You are an expert educational content developer. Continue creating a ${difficultyLevel} level ${contentType} on "${topicName}" designed for a ${sessionLength}-minute session.
 
 IMPORTANT INSTRUCTIONS:
-1. You MUST base your content ENTIRELY on the source materials provided.
+${
+  sourceMetadata.usingCourseContext
+    ? `1. Since no specific source materials were provided, base your content on standard academic knowledge for the topic.
+2. Draw from established educational practices and common curriculum content for this subject area.
+3. Create content appropriate for the specified difficulty level and session length.
+4. Ensure the introduction provides context and importance of the topic based on general knowledge.`
+    : `1. You MUST base your content ENTIRELY on the source materials provided.
 2. Extract key concepts, terminology, examples, and explanations directly from the source materials.
 3. Do not introduce concepts or information that is not present in the source materials.
-4. Create an engaging introduction that provides context and importance of the topic.
+4. Create an engaging introduction that provides context and importance of the topic.`
+}
 
 RESPONSE FORMAT:
 Your response MUST be a valid JSON object with EXACTLY these fields:
@@ -157,7 +171,9 @@ CRITICAL: Your response MUST be valid JSON only. Do not include any text, markdo
 
     const introUserMessage: CoreMessage = {
       role: 'user',
-      content: `Generate an engaging introduction for a ${difficultyLevel} level ${contentType} on "${topicName}" with title "${metadataResponse.title}" based STRICTLY on the provided source materials above.`,
+      content: sourceMetadata.usingCourseContext
+        ? `Generate an engaging introduction for a ${difficultyLevel} level ${contentType} on "${topicName}" with title "${metadataResponse.title}" based on standard academic knowledge and best practices for this subject area.`
+        : `Generate an engaging introduction for a ${difficultyLevel} level ${contentType} on "${topicName}" with title "${metadataResponse.title}" based STRICTLY on the provided source materials above.`,
     }
 
     const introMessages = [introSystemMessage, assistantMessage, introUserMessage]
@@ -179,9 +195,15 @@ CRITICAL: Your response MUST be valid JSON only. Do not include any text, markdo
     const specialSlidesSystemPrompt = `You are an expert educational content developer. Continue creating a ${difficultyLevel} level ${contentType} on "${topicName}" designed for a ${sessionLength}-minute session.
 
 IMPORTANT INSTRUCTIONS:
-1. You MUST base your content ENTIRELY on the source materials provided.
+${
+  sourceMetadata.usingCourseContext
+    ? `1. Since no specific source materials were provided, base your content on standard academic knowledge for the topic.
+2. Draw from established educational practices and common curriculum content for this subject area.
+3. Create content appropriate for the specified difficulty level and session length.`
+    : `1. You MUST base your content ENTIRELY on the source materials provided.
 2. Extract key concepts, terminology, examples, and explanations directly from the source materials.
-3. Do not introduce concepts or information that is not present in the source materials.
+3. Do not introduce concepts or information that is not present in the source materials.`
+}
 4. Create ONLY the following special slides:
  - Introduction slide (first slide that introduces the topic)
  - Agenda/Overview slide (outlines what will be covered)
@@ -228,7 +250,9 @@ CRITICAL: Your response MUST be valid JSON only. Do not include any text, markdo
 
     const specialSlidesUserMessage: CoreMessage = {
       role: 'user',
-      content: `Generate the introduction, agenda, assessment, and conclusion slides for a ${difficultyLevel} level ${contentType} on "${topicName}" with title "${metadataResponse.title}" based STRICTLY on the provided source materials above.`,
+      content: sourceMetadata.usingCourseContext
+        ? `Generate the introduction, agenda, assessment, and conclusion slides for a ${difficultyLevel} level ${contentType} on "${topicName}" with title "${metadataResponse.title}" based on standard academic knowledge and best practices for this subject area.`
+        : `Generate the introduction, agenda, assessment, and conclusion slides for a ${difficultyLevel} level ${contentType} on "${topicName}" with title "${metadataResponse.title}" based STRICTLY on the provided source materials above.`,
     }
 
     const specialSlidesMessages = [
@@ -291,9 +315,15 @@ CRITICAL: Your response MUST be valid JSON only. Do not include any text, markdo
       const contentSlidesSystemPrompt = `You are generating content slides ${startSlideNum} through ${endSlideNum} of a total of ${totalContentSlidesNeeded} content slides. Ensure all slides are unique.
 
 IMPORTANT INSTRUCTIONS:
-1. You MUST base your content ENTIRELY on the source materials provided.
+${
+  sourceMetadata.usingCourseContext
+    ? `1. Since no specific source materials were provided, base your content on standard academic knowledge for the topic.
+2. Draw from established educational practices and common curriculum content for this subject area.
+3. Create content appropriate for the specified difficulty level and session length.`
+    : `1. You MUST base your content ENTIRELY on the source materials provided.
 2. Extract key concepts, terminology, examples, and explanations directly from the source materials.
-3. Do not introduce concepts or information that is not present in the source materials.
+3. Do not introduce concepts or information that is not present in the source materials.`
+}
 4. Create detailed teaching slides with substantial content on each slide.
 5. Focus ONLY on core teaching content slides.
 6. Each slide should have comprehensive speaker notes with additional details and examples.
@@ -324,7 +354,11 @@ CRITICAL: Your response MUST be valid JSON only. Do not include any text, markdo
 
       const contentSlidesUserMessage: CoreMessage = {
         role: 'user',
-        content: `Generate content slides ${startSlideNum} through ${endSlideNum} for a ${difficultyLevel} level ${contentType} on "${topicName}" with title "${metadataResponse.title}" based STRICTLY on the provided source materials above. 
+        content: sourceMetadata.usingCourseContext
+          ? `Generate content slides ${startSlideNum} through ${endSlideNum} for a ${difficultyLevel} level ${contentType} on "${topicName}" with title "${metadataResponse.title}" based on standard academic knowledge and best practices for this subject area. 
+        
+DO NOT create introduction, agenda, assessment, or conclusion slides. Focus ONLY on core teaching content slides.`
+          : `Generate content slides ${startSlideNum} through ${endSlideNum} for a ${difficultyLevel} level ${contentType} on "${topicName}" with title "${metadataResponse.title}" based STRICTLY on the provided source materials above. 
         
 DO NOT create introduction, agenda, assessment, or conclusion slides. Focus ONLY on core teaching content slides.`,
       }
@@ -381,7 +415,9 @@ DO NOT create introduction, agenda, assessment, or conclusion slides. Focus ONLY
 
     const activitiesUserMessage: CoreMessage = {
       role: 'user',
-      content: `Generate the activities for a ${difficultyLevel} level ${contentType} on "${topicName}" with title "${metadataResponse.title}" based STRICTLY on the provided source materials above.`,
+      content: sourceMetadata.usingCourseContext
+        ? `Generate the activities for a ${difficultyLevel} level ${contentType} on "${topicName}" with title "${metadataResponse.title}" based on standard academic knowledge and best practices for this subject area.`
+        : `Generate the activities for a ${difficultyLevel} level ${contentType} on "${topicName}" with title "${metadataResponse.title}" based STRICTLY on the provided source materials above.`,
     }
 
     const activitiesMessages = [activitiesSystemMessage, assistantMessage, activitiesUserMessage]
@@ -417,7 +453,9 @@ DO NOT create introduction, agenda, assessment, or conclusion slides. Focus ONLY
 
     const assessmentUserMessage: CoreMessage = {
       role: 'user',
-      content: `Generate assessment ideas (without example questions) for a ${difficultyLevel} level ${contentType} on "${topicName}" with title "${metadataResponse.title}" based STRICTLY on the provided source materials above.`,
+      content: sourceMetadata.usingCourseContext
+        ? `Generate assessment ideas (without example questions) for a ${difficultyLevel} level ${contentType} on "${topicName}" with title "${metadataResponse.title}" based on standard academic knowledge and best practices for this subject area.`
+        : `Generate assessment ideas (without example questions) for a ${difficultyLevel} level ${contentType} on "${topicName}" with title "${metadataResponse.title}" based STRICTLY on the provided source materials above.`,
     }
 
     const assessmentMessages = [assessmentSystemMessage, assistantMessage, assessmentUserMessage]
@@ -479,7 +517,9 @@ DO NOT create introduction, agenda, assessment, or conclusion slides. Focus ONLY
 
     const readingsUserMessage: CoreMessage = {
       role: 'user',
-      content: `Generate further reading suggestions for a ${difficultyLevel} level ${contentType} on "${topicName}" with title "${metadataResponse.title}" based STRICTLY on the provided source materials above.`,
+      content: sourceMetadata.usingCourseContext
+        ? `Generate further reading suggestions for a ${difficultyLevel} level ${contentType} on "${topicName}" with title "${metadataResponse.title}" based on standard academic knowledge and best practices for this subject area.`
+        : `Generate further reading suggestions for a ${difficultyLevel} level ${contentType} on "${topicName}" with title "${metadataResponse.title}" based STRICTLY on the provided source materials above.`,
     }
 
     const readingsMessages = [readingsSystemMessage, assistantMessage, readingsUserMessage]

frontend/src/app/api/slide/route.ts

@@ -139,6 +139,7 @@ export async function POST(req: Request) {
       sessionLength,
       difficultyLevel,
       topicName,
+      courseInfo,
     } = requestData
 
     console.log('Data from request:', {
@@ -149,6 +150,7 @@ export async function POST(req: Request) {
       sessionLength,
       difficultyLevel,
       topicName,
+      courseInfo,
     })
 
     // Generate course content
@@ -160,6 +162,7 @@ export async function POST(req: Request) {
       sessionLength,
       difficultyLevel,
       topicName,
+      courseInfo,
     )
 
     return NextResponse.json(generatedContent)

frontend/src/app/api/slide/types.ts

@@ -4,6 +4,7 @@
 // Type definitions for course content generation
 
 import { ClientSource } from '@/lib/types/client-source'
+import type { CourseInfo } from '@/lib/types/course-info-types'
 
 export interface LectureSlide {
   title: string
@@ -120,6 +121,7 @@ export interface CourseContentRequest {
   sessionLength: number
   difficultyLevel: string
   topicName: string
+  courseInfo?: CourseInfo
   action?: string
   content?: LectureContent
 }