Add document type selection to DocumentManager and DocumentUploader; implement processDocuments function for improved document processing

Novapool · Novapool · commit fe2d81a0cffb · 2025-03-30T02:58:12.000-04:00
diff --git a/frontend/src/components/DocumentManager.tsx b/frontend/src/components/DocumentManager.tsx
@@ -20,6 +20,7 @@ const DocumentManager: React.FC = () => {
   const [isUploading, setIsUploading] = useState<boolean>(false);
   const [status, setStatus] = useState<string>('');
   const [error, setError] = useState<string>('');
+  const [documentType, setDocumentType] = useState<DocumentType>(DOCUMENT_TYPES.SYLLABUS);
   
   const functions = getFunctions();
 
@@ -51,6 +52,10 @@ const DocumentManager: React.FC = () => {
     }
   };
 
+  const handleTypeChange = (e: React.ChangeEvent<HTMLSelectElement>) => {
+    setDocumentType(e.target.value as DocumentType);
+  };
+
   const handleUpload = async (e: React.FormEvent) => {
     e.preventDefault();
     
@@ -155,6 +160,22 @@ const DocumentManager: React.FC = () => {
       
       <form onSubmit={handleUpload} style={styles.form}>
         <div style={styles.fileInputContainer}>
+          <label htmlFor="documentType" style={styles.fileInputLabel}>
+            Document Type:
+          </label>
+          <select
+            id="documentType"
+            value={documentType}
+            onChange={handleTypeChange}
+            style={styles.select}
+            disabled={isUploading}
+          >
+            <option value={DOCUMENT_TYPES.SYLLABUS}>Syllabus</option>
+            <option value={DOCUMENT_TYPES.TRANSCRIPT}>Transcript</option>
+            <option value={DOCUMENT_TYPES.GRADES}>Grades</option>
+            <option value={DOCUMENT_TYPES.OTHER}>Other</option>
+          </select>
+          
           <label htmlFor="fileInput" style={styles.fileInputLabel}>
             Select Documents
           </label>
@@ -262,6 +283,13 @@ const styles = {
     fontWeight: 'bold',
     color: '#555',
   },
+  select: {
+    padding: '10px',
+    borderRadius: '4px',
+    border: '1px solid #ddd',
+    fontSize: '16px',
+    marginBottom: '15px',
+  },
   fileInput: {
     padding: '10px 0',
   },
diff --git a/frontend/src/components/DocumentProcessingStatus.tsx b/frontend/src/components/DocumentProcessingStatus.tsx
@@ -128,10 +128,8 @@ const DocumentProcessingStatus: React.FC<DocumentProcessingStatusProps> = ({ onP
     setStatus('Processing documents...');
     
     try {
-      const formatDocumentsData = httpsCallable(functions, 'formatDocumentsData');
-      const result = await formatDocumentsData({
-        forceProcess: true // Add this flag to bypass any waiting checks
-      });
+      const processDocuments = httpsCallable(functions, 'processDocuments');
+      const result = await processDocuments({});
       
       if ((result.data as any).success) {
         setStatus('Documents processed successfully');
@@ -183,10 +181,10 @@ const DocumentProcessingStatus: React.FC<DocumentProcessingStatusProps> = ({ onP
     setError(null);
     
     try {
-      // We'll use the formatDocumentsData function to process all documents
-      const formatDocumentsData = httpsCallable(functions, 'formatDocumentsData');
-      const result = await formatDocumentsData({
-        forceProcess: true // Add this flag to bypass any waiting checks
+      // Use the processDocuments function to process documents
+      const processDocuments = httpsCallable(functions, 'processDocuments');
+      const result = await processDocuments({
+        documentId: documentId // Pass the specific document ID to process
       });
       
       if ((result.data as any).success) {
diff --git a/frontend/src/components/DocumentUploader.tsx b/frontend/src/components/DocumentUploader.tsx
@@ -2,6 +2,7 @@ import React, { useState, useEffect } from 'react';
 import { useAuth } from '../contexts/AuthContext';
 import { getFunctions, httpsCallable } from 'firebase/functions';
 import { getFirestore, collection, doc, getDoc, getDocs, onSnapshot, query, orderBy, limit } from 'firebase/firestore';
+import DOCUMENT_TYPES, { DocumentType } from '../constants/documentTypes';
 
 interface Document {
   id: string;
@@ -19,7 +20,7 @@ interface Prediction {
 const DocumentUploader: React.FC = () => {
   const { currentUser } = useAuth();
   const [file, setFile] = useState<File | null>(null);
-  const [documentType, setDocumentType] = useState<string>('syllabus');
+  const [documentType, setDocumentType] = useState<string>(DOCUMENT_TYPES.SYLLABUS);
   const [uploadProgress, setUploadProgress] = useState<number>(0);
   const [isUploading, setIsUploading] = useState<boolean>(false);
   const [documents, setDocuments] = useState<Document[]>([]);
@@ -37,7 +38,7 @@ const DocumentUploader: React.FC = () => {
     // Get user's documents
     const getUserDocuments = async () => {
       try {
-        const getUserDocumentsFunction = httpsCallable(functions, 'get_user_documents');
+        const getUserDocumentsFunction = httpsCallable(functions, 'getUserDocuments');
         const result = await getUserDocumentsFunction({});
         if (result.data && (result.data as any).success) {
           setDocuments((result.data as any).documents || []);
@@ -94,8 +95,8 @@ const DocumentUploader: React.FC = () => {
         const base64String = fileReader.result as string;
         
         try {
-          // Call upload_and_process_document cloud function
-          const uploadAndProcessDocument = httpsCallable(functions, 'upload_and_process_document');
+          // Call uploadDocument cloud function
+          const uploadAndProcessDocument = httpsCallable(functions, 'uploadDocument');
           const result = await uploadAndProcessDocument({
             documentType: documentType,
             documentBase64: base64String
@@ -224,8 +225,8 @@ const DocumentUploader: React.FC = () => {
         credit_hours: syllabusData.credit_hours || '3'
       };
       
-      // Call predict_final_grade cloud function
-      const predictFinalGrade = httpsCallable(functions, 'predict_final_grade');
+      // Call predictFinalGrade cloud function
+      const predictFinalGrade = httpsCallable(functions, 'predictFinalGrade');
       const result = await predictFinalGrade({ courseData });
       
       const data = result.data as any;
@@ -236,7 +237,7 @@ const DocumentUploader: React.FC = () => {
         setIsUploading(false);
         
         // Refresh documents list
-        const getUserDocuments = httpsCallable(functions, 'get_user_documents');
+        const getUserDocuments = httpsCallable(functions, 'getUserDocuments');
         const docsResult = await getUserDocuments({});
         if (docsResult.data && (docsResult.data as any).success) {
           setDocuments((docsResult.data as any).documents || []);
@@ -295,8 +296,10 @@ const DocumentUploader: React.FC = () => {
                 style={styles.select}
                 disabled={isUploading}
               >
-                <option value="syllabus">Syllabus</option>
-                <option value="transcript">Transcript</option>
+                <option value={DOCUMENT_TYPES.SYLLABUS}>Syllabus</option>
+                <option value={DOCUMENT_TYPES.TRANSCRIPT}>Transcript</option>
+                <option value={DOCUMENT_TYPES.GRADES}>Grades</option>
+                <option value={DOCUMENT_TYPES.OTHER}>Other</option>
               </select>
             </div>
             
diff --git a/functions-node/index.js b/functions-node/index.js
@@ -4,17 +4,107 @@ const pdfParse = require('pdf-parse');
 const tmp = require('tmp');
 const fs = require('fs');
 const OpenAI = require('openai');
+const cors = require('cors')({ origin: true });
 const { calculateCurrentGrade } = require('./calculateGrade');
 const { predictFinalGrade } = require('./predictGrade');
 const { formatDocumentsData } = require('./formatDocumentsData');
 const { DOCUMENT_TYPES, normalizeDocumentType } = require('./utils/documentUtils');
+const corsMiddleware = require('./utils/corsMiddleware');
 
 // Initialize Firebase Admin
 admin.initializeApp();
 
 // Export functions
 exports.calculateCurrentGrade = calculateCurrentGrade;
 exports.predictFinalGrade = predictFinalGrade;
+
+/**
+ * Cloud Function to trigger processing of documents
+ * This handles the extraction and formatting of document data
+ * Can process all documents or a specific document if documentId is provided
+ */
+exports.processDocuments = functions.https.onCall(async (data, context) => {
+  try {
+    // Ensure user is authenticated
+    if (!context.auth) {
+      throw new functions.https.HttpsError('unauthenticated', 'User must be authenticated');
+    }
+    
+    const userId = context.auth.uid;
+    const specificDocumentId = data?.documentId;
+    
+    console.log(`Processing documents for user ${userId}${specificDocumentId ? `, document ID: ${specificDocumentId}` : ''}`);
+    
+    // Get documents that need processing
+    const db = admin.firestore();
+    const documentsRef = db.collection('users').doc(userId).collection('documents');
+    
+    let snapshot;
+    if (specificDocumentId) {
+      // If a specific document ID is provided, only process that document
+      const docRef = documentsRef.doc(specificDocumentId);
+      const docSnapshot = await docRef.get();
+      
+      if (!docSnapshot.exists) {
+        throw new functions.https.HttpsError('not-found', 'Document not found');
+      }
+      
+      // Create a QuerySnapshot-like object with a single document
+      snapshot = {
+        empty: false,
+        size: 1,
+        docs: [docSnapshot],
+        forEach: (callback) => callback(docSnapshot)
+      };
+    } else {
+      // Otherwise, get all documents that need processing
+      const query = documentsRef.where('status', 'in', ['uploaded', 'extracted']);
+      snapshot = await query.get();
+    }
+    
+    if (snapshot.empty) {
+      return {
+        success: false,
+        message: 'No documents found for processing'
+      };
+    }
+    
+    console.log(`Found ${snapshot.size} documents to process`);
+    
+    // Process each document
+    const processPromises = [];
+    snapshot.forEach(doc => {
+      const data = doc.data();
+      
+      // If document is uploaded but not extracted, extract text
+      if (data.status === 'uploaded' && data.filePath) {
+        processPromises.push(extractTextFromPdf(userId, doc.id, data.filePath));
+      }
+      // If document is extracted but not processed, update status
+      else if (data.status === 'extracted') {
+        const docRef = db.collection('users').doc(userId).collection('documents').doc(doc.id);
+        processPromises.push(docRef.update({
+          status: 'processing',
+          processingStartedAt: admin.firestore.FieldValue.serverTimestamp()
+        }));
+      }
+    });
+    
+    await Promise.all(processPromises);
+    
+    // Once all documents are processed, format the data
+    const formattedData = await formatDocumentsData(userId, true);
+    
+    return {
+      success: true,
+      message: 'Documents processed successfully',
+      documentsProcessed: snapshot.size
+    };
+  } catch (error) {
+    console.error('Error processing documents:', error);
+    throw new functions.https.HttpsError('internal', `Processing failed: ${error.message}`);
+  }
+});
 exports.formatDocumentsData = functions.https.onCall(async (data, context) => {
   try {
     // Ensure user is authenticated
diff --git a/functions-node/utils/corsMiddleware.js b/functions-node/utils/corsMiddleware.js
@@ -0,0 +1,16 @@
+const cors = require('cors')({ origin: true });
+
+/**
+ * Middleware to handle CORS for Firebase Cloud Functions
+ * @param {Function} handler - The function handler to wrap with CORS support
+ * @returns {Function} A function that handles CORS and then calls the handler
+ */
+const corsMiddleware = (handler) => {
+  return (req, res) => {
+    return cors(req, res, () => {
+      return handler(req, res);
+    });
+  };
+};
+
+module.exports = corsMiddleware;
