File size: 15,526 Bytes
e7747d3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3d26581
e7747d3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3d26581
 
e7747d3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3d26581
 
 
 
 
e7747d3
 
 
3d26581
4409516
3d26581
e7747d3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3d26581
 
 
e7747d3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3d26581
e7747d3
 
 
 
 
 
 
 
3d26581
 
 
e7747d3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3d26581
 
 
 
 
 
 
 
 
 
e7747d3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3d26581
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e7747d3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3d26581
 
 
 
 
 
 
 
 
 
 
 
 
e7747d3
 
3d26581
 
 
 
 
 
e7747d3
 
 
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
import os
import re
import json
import torch
import numpy as np
import logging
from typing import Dict, List, Tuple, Optional
from tqdm import tqdm
from pydantic import BaseModel
from transformers import (
    AutoTokenizer,
    AutoModelForSeq2SeqLM,
    AutoModelForQuestionAnswering,
    pipeline,
    LogitsProcessor,
    LogitsProcessorList,
    PreTrainedModel,
    PreTrainedTokenizer
)
from sentence_transformers import SentenceTransformer, CrossEncoder
from sklearn.feature_extraction.text import TfidfVectorizer
from rank_bm25 import BM25Okapi
import PyPDF2
from sklearn.cluster import KMeans
import spacy
import subprocess
import gradio as gr

logging.basicConfig(
    level=logging.INFO,
    format="%(asctime)s [%(levelname)s] %(message)s"
)

class ConfidenceCalibrator(LogitsProcessor):
    def __init__(self, calibration_factor: float = 0.9):
        self.calibration_factor = calibration_factor

    def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> torch.FloatTensor:
        return scores / self.calibration_factor

class DocumentResult(BaseModel):
    content: str
    confidence: float
    source_page: int
    supporting_evidence: List[str]

class OptimalModelSelector:
    def __init__(self):
        self.qa_models = {
            "deberta-v3": ("deepset/deberta-v3-large-squad2", 0.87)
        }
        self.summarization_models = {
            "bart": ("facebook/bart-large-cnn", 0.85)
        }
        self.current_models = {}

    def get_best_model(self, task_type: str) -> Tuple[PreTrainedModel, PreTrainedTokenizer, float]:
        model_map = self.qa_models if "qa" in task_type else self.summarization_models
        best_model_name, best_score = max(model_map.items(), key=lambda x: x[1][1])
        if best_model_name not in self.current_models:
            tokenizer = AutoTokenizer.from_pretrained(model_map[best_model_name][0])
            model = (AutoModelForQuestionAnswering if "qa" in task_type
                     else AutoModelForSeq2SeqLM).from_pretrained(model_map[best_model_name][0])
            model = model.eval().half().to('cuda' if torch.cuda.is_available() else 'cpu')
            self.current_models[best_model_name] = (model, tokenizer)
        return *self.current_models[best_model_name], best_score

class PDFAugmentedRetriever:
    def __init__(self, document_texts: List[str]):
        self.documents = [(i, text) for i, text in enumerate(document_texts)]
        self.bm25 = BM25Okapi([text.split() for _, text in self.documents])
        self.encoder = CrossEncoder('cross-encoder/ms-marco-MiniLM-L-6-v2')
        self.tfidf = TfidfVectorizer(stop_words='english').fit([text for _, text in self.documents])

    def retrieve(self, query: str, top_k: int = 8) -> List[Tuple[int, str, float]]:  # Increased from 5 to 8
        bm25_scores = self.bm25.get_scores(query.split())
        semantic_scores = self.encoder.predict([(query, doc) for _, doc in self.documents])
        combined_scores = 0.4 * bm25_scores + 0.6 * np.array(semantic_scores)
        top_indices = np.argsort(combined_scores)[-top_k:][::-1]
        return [(self.documents[i][0], self.documents[i][1], float(combined_scores[i]))
                for i in top_indices]

class DetailedExplainer:
    def __init__(self,
                 explanation_model: str = "google/flan-t5-large",
                 device: int = 0):
        try:
            self.nlp = spacy.load("en_core_web_sm")
        except OSError:
            subprocess.run(["python", "-m", "spacy", "download", "en_core_web_sm"], check=True)
            self.nlp = spacy.load("en_core_web_sm")
        self.explainer = pipeline(
            "text2text-generation",
            model=explanation_model,
            tokenizer=explanation_model,
            device=device,
            max_length=2048,
            max_new_tokens=2000
        )

    def extract_concepts(self, text: str) -> list:
        doc = self.nlp(text)
        concepts = set()
        for chunk in doc.noun_chunks:
            if len(chunk) > 1 and not chunk.root.is_stop:
                concepts.add(chunk.text.strip())
        for ent in doc.ents:
            if ent.label_ in ["PERSON", "ORG", "GPE", "NORP", "EVENT", "WORK_OF_ART"]:
                concepts.add(ent.text.strip())
        return list(concepts)

    def explain_concept(self, concept: str, context: str, min_accuracy: float = 0.50) -> str:
        prompt = (
            f"The following sentence from a PDF is given \n{context}\n\n\n"
            f"Now provide a detailed explanation of the concept '{concept}' mentioned above. "
            f"Include background information, context, examples, and significance. "
            f"Write a comprehensive explanation with at least {int(min_accuracy * 100)}% accuracy. "
            f"Make the explanation thorough and informative, up to 500 words if needed."
        )
        result = self.explainer(
            prompt,
            do_sample=False,
            #max_length=2048,
            max_new_tokens=600
        )
        return result[0]["generated_text"].strip()

    def explain_text(self, text: str, context: str) -> dict:
        concepts = self.extract_concepts(text)
        explanations = {}
        for concept in concepts:
            explanations[concept] = self.explain_concept(concept, context)
        return {"concepts": concepts, "explanations": explanations}

class AdvancedPDFAnalyzer:
    def __init__(self):
        self.logger = logging.getLogger("PDFAnalyzer")
        self.model_selector = OptimalModelSelector()
        self.device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
        self.qa_model, self.qa_tokenizer, _ = self.model_selector.get_best_model("qa")
        self.qa_model = self.qa_model.to(self.device)
        self.summarizer = pipeline(
            "summarization",
            model="facebook/bart-large-cnn",
            device=0 if torch.cuda.is_available() else -1,
            framework="pt",
            max_length=2048,
            min_length=100
        )
        self.logits_processor = LogitsProcessorList([
            ConfidenceCalibrator(calibration_factor=0.85)
        ])
        self.detailed_explainer = DetailedExplainer(device=0 if torch.cuda.is_available() else -1)

    def extract_text_with_metadata(self, file_path: str) -> List[Dict]:
        documents = []
        with open(file_path, 'rb') as f:
            reader = PyPDF2.PdfReader(f)
            for i, page in enumerate(reader.pages):
                text = page.extract_text()
                if not text or not text.strip():
                    continue
                page_number = i + 1
                metadata = {
                    'source': os.path.basename(file_path),
                    'page': page_number,
                    'char_count': len(text),
                    'word_count': len(text.split()),
                }
                documents.append({
                    'content': self._clean_text(text),
                    'metadata': metadata
                })
        if not documents:
            raise ValueError("No extractable content found in PDF")
        return documents

    def _clean_text(self, text: str) -> str:
        text = re.sub(r'[\x00-\x1F\x7F-\x9F]', ' ', text)
        text = re.sub(r'\s+', ' ', text)
        text = re.sub(r'(\w)-\s+(\w)', r'\1\2', text)
        return text.strip()

    def answer_question(self, question: str, documents: List[Dict]) -> Dict:
        retriever = PDFAugmentedRetriever([doc['content'] for doc in documents])
        relevant_contexts = retriever.retrieve(question, top_k=5)  # Increased context retrieval
        answers = []
        
        for page_idx, context, similarity_score in relevant_contexts:
            inputs = self.qa_tokenizer(
                question,
                context,
                add_special_tokens=True,
                return_tensors="pt",
                max_length=1024,  # Increased from 512
                truncation=True,
                padding=True
            )
            inputs = {k: v.to(self.device) for k, v in inputs.items()}
            
            with torch.no_grad():
                outputs = self.qa_model(**inputs)
                start_logits = outputs.start_logits
                end_logits = outputs.end_logits
                
                logits_processor = LogitsProcessorList([ConfidenceCalibrator()])
                start_logits = logits_processor(inputs['input_ids'], start_logits)
                end_logits = logits_processor(inputs['input_ids'], end_logits)
                
                start_prob = torch.nn.functional.softmax(start_logits, dim=-1)
                end_prob = torch.nn.functional.softmax(end_logits, dim=-1)
                
                max_start_score, max_start_idx = torch.max(start_prob, dim=-1)
                max_start_idx_int = max_start_idx.item()
                max_end_score, max_end_idx = torch.max(end_prob[0, max_start_idx_int:], dim=-1)
                max_end_idx_int = max_end_idx.item() + max_start_idx_int
                
                confidence = float((max_start_score * max_end_score) * 0.9 * similarity_score)
                answer_tokens = inputs["input_ids"][0][max_start_idx_int:max_end_idx_int + 1]
                answer = self.qa_tokenizer.decode(answer_tokens, skip_special_tokens=True)
                
                # Enhanced answer extraction for longer responses
                if len(answer.strip()) < 20:  # If answer is too short, try extracting more context
                    # Get more surrounding context
                    extended_start = max(0, max_start_idx_int - 50)
                    extended_end = min(len(inputs["input_ids"][0]), max_end_idx_int + 150)
                    extended_tokens = inputs["input_ids"][0][extended_start:extended_end]
                    extended_answer = self.qa_tokenizer.decode(extended_tokens, skip_special_tokens=True)
                    if len(extended_answer.strip()) > len(answer.strip()):
                        answer = extended_answer
                
                # Only generate explanations if we have a valid answer
                explanations_result = {"concepts": [], "explanations": {}}
                if answer and answer.strip():
                    try:
                        explanations_result = self.detailed_explainer.explain_text(answer, context)
                    except Exception as e:
                        self.logger.warning(f"Failed to generate explanations: {e}")
                
                answers.append({
                    "answer": answer,
                    "confidence": confidence,
                    "context": context,
                    "page_number": documents[page_idx]['metadata']['page'],
                    "explanations": explanations_result
                })
        
        if not answers:
            return {
                "answer": "No confident answer found", 
                "confidence": 0.0, 
                "explanations": {"concepts": [], "explanations": {}},
                "page_number": 0,
                "context": ""
            }
        
        # Get the best answer based on confidence
        best_answer = max(answers, key=lambda x: x['confidence'])
        
        # For comprehensive responses, combine information from multiple high-confidence answers
        if len(answers) > 1:
            high_confidence_answers = [a for a in answers if a['confidence'] > 0.2]
            if len(high_confidence_answers) > 1:
                # Combine explanations from multiple sources
                combined_explanations = {}
                all_concepts = set()
                
                for ans in high_confidence_answers[:3]:  # Use top 3 answers
                    explanations = ans.get("explanations", {}).get("explanations", {})
                    concepts = ans.get("explanations", {}).get("concepts", [])
                    all_concepts.update(concepts)
                    combined_explanations.update(explanations)
                
                best_answer["explanations"]["explanations"] = combined_explanations
                best_answer["explanations"]["concepts"] = list(all_concepts)
        
        # FIXED: Always return the best answer dictionary, just modify the answer text if confidence is low
        if best_answer['confidence'] < 0.3:  # Lowered threshold to be more permissive
            best_answer['answer'] = f"[Low Confidence] {best_answer['answer']}"
        
        return best_answer

# Initialize analyzer (make sure to update the PDF path)
analyzer = AdvancedPDFAnalyzer()

# Global variable to store documents
documents = []

def load_pdf(file_path: str):
    """Load PDF and extract documents"""
    global documents
    try:
        documents = analyzer.extract_text_with_metadata(file_path)
        return f"Successfully loaded PDF with {len(documents)} pages."
    except Exception as e:
        return f"Error loading PDF: {str(e)}"

def ask_question_gradio(question: str):
    if not question.strip():
        return "Please enter a valid question."
    
    if not documents:
        return "❌ No PDF loaded. Please load a PDF first."
    
    try:
        result = analyzer.answer_question(question, documents)
        
        # Ensure we have the expected structure
        answer = result.get('answer', 'No answer found')
        confidence = result.get('confidence', 0.0)
        page_number = result.get('page_number', 0)
        explanations = result.get("explanations", {}).get("explanations", {})
        
        # Format explanations
        explanation_text = ""
        if explanations:
            explanation_text = "\n\n".join(
                f"πŸ”Ή **{concept}**: {desc}"
                for concept, desc in explanations.items()
                if desc and desc.strip()
            )
        
        # Build response
        response_parts = [
            f"πŸ“Œ **Answer**: {answer}",
            f"πŸ”’ **Confidence**: {confidence:.2f}",
            f"πŸ“„ **Page**: {page_number}"
        ]
        
        if explanation_text:
            response_parts.append(f"πŸ“˜ **Explanations**:\n{explanation_text}")
        
        return "\n\n".join(response_parts)
        
    except Exception as e:
        return f"❌ Error: {str(e)}"

# Load your PDF here - update the path to your actual PDF file
pdf_path = "example.pdf"
if os.path.exists(pdf_path):
    load_result = load_pdf(pdf_path)
    print(load_result)
else:
    print(f"PDF file '{pdf_path}' not found. Please update the path.")

demo = gr.Interface(
    fn=ask_question_gradio,
    inputs=gr.Textbox(
        label="Ask a question about the PDF", 
        placeholder="Type your question here...",
        lines=3,
        max_lines=5
    ),
    outputs=gr.Markdown(
        label="Answer",
        value="",
        show_copy_button=True
    ),
    title="Quandans AI - Ask Questions (Up to 2000 words)",
    description="Ask a question based on the document loaded in this system. The system can now provide comprehensive answers up to 2000 words with detailed explanations.",
    examples=[
        "What is the main topic of this document?",
        "Provide a detailed summary of the key points from page 1",
        "What are the conclusions mentioned and explain them in detail?",
        "Give me a comprehensive overview of all the important concepts discussed"
    ],
    theme=gr.themes.Soft(),
    allow_flagging="never"
)

demo.launch()