File size: 14,379 Bytes
3d9654b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
# Full app setup in one script (modularized)
# Required Libraries
import streamlit as st
from PyPDF2 import PdfReader
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.vectorstores import FAISS
from langchain.text_splitter import CharacterTextSplitter
from langchain.chains import RetrievalQA, LLMChain
from langchain.prompts import PromptTemplate
from langchain_google_genai import GoogleGenerativeAI
import os
import pandas as pd
import plotly.express as px
import uuid
import base64
import tempfile
import fitz  # PyMuPDF
from docx import Document
import google.generativeai as genai
from google.api_core.exceptions import InvalidArgument
from dotenv import load_dotenv

load_dotenv()

st.set_page_config(layout="wide")
st.title("πŸ“š PDF QA App")

# Initialize session state for uploaded files
if "uploaded_files" not in st.session_state:
    st.session_state.uploaded_files = []

# Initialize Gemini model
@st.cache_resource
def load_gemini_model():
    # You'll need to get an API key from Google AI Studio
    api_key = os.getenv("GOOGLE_API_KEY")
    
    if not api_key:
        api_key = st.text_input("Enter your Google API Key", type="password")
        if not api_key:
            st.warning("Please enter a Google API key to continue")
            st.stop()
    
    # Configure the Gemini model
    try:
        # Configure the genai module
        genai.configure(api_key=api_key)
        
        # Verify available models
        models = genai.list_models()
        available_models = [m.name for m in models]
        
        # Check which model is available and select the appropriate one
        gemini_model_name = None
        for model_option in ["gemini-1.5-pro", "gemini-pro", "gemini-1.0-pro"]:
            if any(model_option in model for model in available_models):
                gemini_model_name = model_option
                break
                
        if not gemini_model_name:
            st.error(f"No Gemini model found. Available models: {available_models}")
            st.stop()
            
        st.success(f"Using Gemini model: {gemini_model_name}")
        
        # Initialize the LangChain wrapper for Gemini
        llm = GoogleGenerativeAI(
            model=gemini_model_name,
            google_api_key=api_key,
            temperature=0.3,
            max_output_tokens=512
        )
        return llm
    except Exception as e:
        st.error(f"Error initializing Gemini model: {str(e)}")
        st.stop()

# Session state for chat history
if "chat_history" not in st.session_state:
    st.session_state.chat_history = []
if "analytics" not in st.session_state:
    st.session_state.analytics = []

# File uploader
pdf_files = st.file_uploader("Upload one or more PDFs", type="pdf", accept_multiple_files=True)

# Store uploaded files in session state for later use
if pdf_files:
    st.session_state.uploaded_files = pdf_files

# Interactive PDF Viewer
with st.expander("πŸ“‘ PDF Viewer", expanded=False):
    try:
        if st.session_state.uploaded_files:
            # Display the uploaded files in a selection box
            pdf_file_names = [uploaded_file.name for uploaded_file in st.session_state.uploaded_files]
            pdf_file_names.insert(0, "Select PDF File")
            selected_pdf = st.selectbox("Select a PDF to view", pdf_file_names)
            
            # Retrieve the selected PDF file
            selected_file = None
            for uploaded_file in st.session_state.uploaded_files:
                if uploaded_file.name == selected_pdf:
                    selected_file = uploaded_file
                    break
            
            # Display the selected PDF
            if selected_file and selected_pdf != "Select PDF File":
                st.subheader(f"Viewing PDF: {selected_pdf}")
                
                # Read PDF file
                selected_file.seek(0)  # Reset file pointer to start
                pdf_bytes = selected_file.read()
                selected_file.seek(0)  # Reset file pointer after reading
                
                # Encode the PDF file in base64 for displaying in iframe
                pdf_base64 = base64.b64encode(pdf_bytes).decode('utf-8')
                
                # Display the PDF file in an iframe using an HTML embed
                pdf_display = f'<iframe src="data:application/pdf;base64,{pdf_base64}" width="100%" height="600" type="application/pdf"></iframe>'
                st.markdown(pdf_display, unsafe_allow_html=True)
    except Exception as e:
        st.error(f"Error displaying PDF: {str(e)}")

question = st.text_input("Ask a question across PDFs")

# Helper: Save files to temp and chunk
def load_and_chunk(file):
    # Save file pointer position
    file_pos = file.tell()
    
    # Reset file pointer to start
    file.seek(0)
    
    try:
        reader = PdfReader(file)
        all_text, page_map = "", {}
        for i, page in enumerate(reader.pages):
            text = page.extract_text()
            page_map[i] = text
            all_text += f"\n[Page {i + 1}]\n{text}"
        
        splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
        chunks = splitter.split_text(all_text)
        
        # Reset file pointer to original position
        file.seek(file_pos)
        
        return chunks, page_map
    except Exception as e:
        st.error(f"Error processing PDF {file.name}: {str(e)}")
        file.seek(file_pos)  # Reset file pointer even if there's an error
        return [], {}

# Helper: Create FAISS store
def embed_documents(chunks):
    # Use HuggingFace embeddings instead of OpenAI
    try:
        embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-mpnet-base-v2")
        return FAISS.from_texts(chunks, embeddings)
    except Exception as e:
        st.error(f"Error creating embeddings: {str(e)}")
        return None

# Helper: Display PDF Page (both methods available)
def show_pdf_page(file, page_num, use_iframe=False):
    # Save current position
    file_pos = file.tell()
    
    # Reset file pointer
    file.seek(0)
    
    try:
        if use_iframe:
            # Read the entire PDF
            pdf_bytes = file.read()
            # Encode the PDF file in base64 for displaying in iframe
            pdf_base64 = base64.b64encode(pdf_bytes).decode('utf-8')
            # Display the PDF file in an iframe with page number parameter
            pdf_display = f'<iframe src="data:application/pdf;base64,{pdf_base64}#page={page_num}" width="100%" height="500" type="application/pdf"></iframe>'
            st.markdown(pdf_display, unsafe_allow_html=True)
        else:
            # Original method using PyMuPDF to render as image
            with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as tmp:
                tmp.write(file.read())
                tmp_path = tmp.name
                
            # Open the saved PDF
            doc = fitz.open(tmp_path)
            
            # Validate page number
            if page_num < 1 or page_num > len(doc):
                st.error(f"Invalid page number: {page_num}. Document has {len(doc)} pages.")
                return
                
            page = doc.load_page(page_num - 1)
            pix = page.get_pixmap()
            img_path = tmp_path.replace(".pdf", f"_page{page_num}.png")
            pix.save(img_path)
            st.image(img_path, caption=f"Page {page_num}")
            
            # Clean up
            doc.close()
            try:
                os.unlink(img_path)
                os.unlink(tmp_path)
            except Exception as e:
                pass  # Silently handle cleanup errors
    except Exception as e:
        st.error(f"Error displaying PDF page: {str(e)}")
    finally:
        # Reset file pointer to original position
        file.seek(file_pos)

# Helper: Summarize
@st.cache_data
def summarize_doc(chunks, _llm):
    summary_prompt = PromptTemplate(
        input_variables=["context"],
        template="Summarize this document:\n{context}"
    )
    chain = LLMChain(llm=_llm, prompt=summary_prompt)
    
    # Join only a subset of chunks to avoid token limits
    full_text = " ".join(chunks[:5])  # Limiting to first 5 chunks
    
    try:
        return chain.run({"context": full_text})
    except Exception as e:
        st.error(f"Error during summarization: {str(e)}")
        return "Error: Document too large to summarize or API error. Try with fewer pages."


# Initialize model and DBs
try:
    llm = load_gemini_model()
    file_chunks, vector_dbs, page_maps = {}, {}, {}
    
    if pdf_files:
        with st.spinner("Processing PDF files..."):
            for file in pdf_files:
                chunks, page_map = load_and_chunk(file)
                if chunks:  # Only create db if chunks were successfully extracted
                    db = embed_documents(chunks)
                    if db:  # Only store if db was successfully created
                        file_chunks[file.name] = chunks
                        page_maps[file.name] = page_map
                        vector_dbs[file.name] = db
except Exception as e:
    st.error(f"Error loading model or processing files: {str(e)}")

# Document Summarization UI
if pdf_files and file_chunks:
    with st.expander("πŸ“„ Document Summarization"):
        summarize_option = st.selectbox("Select a document to summarize", 
                                      ["All"] + [f.name for f in pdf_files if f.name in file_chunks])
        if st.button("Summarize"):
            with st.spinner("Summarizing..."):
                try:
                    if summarize_option == "All":
                        for file in pdf_files:
                            if file.name in file_chunks:
                                summary = summarize_doc(file_chunks[file.name], llm)
                                st.subheader(file.name)
                                st.write(summary)
                    else:
                        f = next(f for f in pdf_files if f.name == summarize_option)
                        summary = summarize_doc(file_chunks[f.name], llm)
                        st.subheader(f.name)
                        st.write(summary)
                except Exception as e:
                    st.error(f"Error during summarization: {str(e)}")

# Question Answering UI
results = []
if question and vector_dbs:
    try:
        for fname, db in vector_dbs.items():
            qa = RetrievalQA.from_chain_type(llm=llm, retriever=db.as_retriever())
            
            try:
                result = qa({"query": question})
                answer = result['result']
                
                context_docs = db.similarity_search(question, k=1)
                if context_docs:
                    context = context_docs[0].page_content
                    
                    # Extract page number safely
                    page_num = "Unknown"
                    try:
                        page_num_match = context.split("[Page ")
                        if len(page_num_match) > 1:
                            page_num = page_num_match[1].split("]")[0]
                    except:
                        pass
                    
                    st.markdown(f"### πŸ“˜ {fname} (Page {page_num})")
                    #st.write(highlight_text(context, answer))
                    st.write(answer)
                    
                    
                    
                    st.session_state.chat_history.append({
                        "file": fname, 
                        "page": page_num, 
                        "question": question, 
                        "answer": answer
                    })
                    
                    st.session_state.analytics.append({
                        "file": fname,
                        "page": int(page_num) if page_num.isdigit() else 0,
                        "confidence": 0.9,
                        "question": question
                    })
                    
                    results.append((fname, page_num, question, answer))
            except Exception as e:
                st.error(f"Error processing question for {fname}: {str(e)}")
    except Exception as e:
        st.error(f"Error during question answering: {str(e)}")

# Chat History Panel
if st.session_state.chat_history:
    with st.expander("πŸ’¬ Chat History"):
        for entry in st.session_state.chat_history[::-1]:
            st.markdown(f"**{entry['file']}** | Page {entry['page']}\n> {entry['question']}\n→ {entry['answer']}")

# Downloadable Report
if results:
    with st.expander("πŸ“₯ Download Q&A Report"):
        docx = Document()
        docx.add_heading("PDF QA Report", 0)
        for fname, page, q, a in results:
            docx.add_paragraph(f"File: {fname} | Page: {page}", style="List Bullet")
            docx.add_paragraph(f"Q: {q}")
            docx.add_paragraph(f"A: {a}\n")
        
        try:
            docx_path = os.path.join(tempfile.gettempdir(), f"report_{uuid.uuid4()}.docx")
            docx.save(docx_path)
            with open(docx_path, "rb") as f:
                b64 = base64.b64encode(f.read()).decode()
                st.markdown(f"[Download DOCX Report](data:application/octet-stream;base64,{b64})", unsafe_allow_html=True)
            # Clean up temporary files
            try:
                os.unlink(docx_path)
            except:
                pass
        except Exception as e:
            st.error(f"Error creating downloadable report: {str(e)}")

# Analytics Dashboard
if st.session_state.analytics:
    with st.expander("πŸ“Š Analytics Dashboard"):
        df = pd.DataFrame(st.session_state.analytics)
        col1, col2 = st.columns(2)
        with col1:
            st.dataframe(df)
        with col2:
            try:
                fig = px.histogram(df, x="file", color="page", title="Answer Distribution by File")
                st.plotly_chart(fig, use_container_width=True)
            except Exception as e:
                st.error(f"Error generating analytics chart: {str(e)}")
        
        st.markdown("Use filters below to explore:")
        file_filter = st.selectbox("Filter by file", ["All"] + list(df["file"].unique()))
        if file_filter != "All":
            st.dataframe(df[df["file"] == file_filter])