Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -1,10 +1,16 @@
|
|
1 |
|
2 |
|
|
|
|
|
|
|
3 |
|
4 |
import os
|
|
|
|
|
5 |
import streamlit as st
|
6 |
import numpy as np
|
7 |
import fitz # PyMuPDF
|
|
|
8 |
from ultralytics import YOLO
|
9 |
from sklearn.cluster import KMeans
|
10 |
from sklearn.metrics.pairwise import cosine_similarity
|
@@ -13,27 +19,22 @@ from langchain_community.document_loaders import PyMuPDFLoader
|
|
13 |
from langchain_openai import OpenAIEmbeddings
|
14 |
from langchain_text_splitters import RecursiveCharacterTextSplitter
|
15 |
from langchain_core.prompts import ChatPromptTemplate
|
16 |
-
from sklearn.decomposition import PCA
|
17 |
from langchain_openai import ChatOpenAI
|
18 |
-
import string
|
19 |
import re
|
20 |
-
|
|
|
21 |
|
22 |
# Load the trained model
|
23 |
-
model = YOLO("best.pt")
|
24 |
-
openai_api_key =
|
25 |
|
26 |
# Define the class indices for figures, tables, and text
|
27 |
-
figure_class_index = 4
|
28 |
-
table_class_index = 3
|
29 |
-
|
30 |
-
# Global variables to store embeddings and contents
|
31 |
-
global_embeddings = None
|
32 |
-
global_split_contents = None
|
33 |
|
|
|
34 |
def clean_text(text):
|
35 |
-
|
36 |
-
return text
|
37 |
|
38 |
def remove_references(text):
|
39 |
reference_patterns = [
|
@@ -47,9 +48,10 @@ def remove_references(text):
|
|
47 |
return text
|
48 |
|
49 |
def save_uploaded_file(uploaded_file):
|
50 |
-
|
51 |
-
|
52 |
-
|
|
|
53 |
|
54 |
def summarize_pdf(pdf_file_path, num_clusters=10):
|
55 |
embeddings_model = OpenAIEmbeddings(model="text-embedding-3-small", api_key=openai_api_key)
|
@@ -70,66 +72,21 @@ def summarize_pdf(pdf_file_path, num_clusters=10):
|
|
70 |
loader = PyMuPDFLoader(pdf_file_path)
|
71 |
docs = loader.load()
|
72 |
full_text = "\n".join(doc.page_content for doc in docs)
|
73 |
-
cleaned_full_text = remove_references(full_text)
|
74 |
-
cleaned_full_text = clean_text(cleaned_full_text)
|
75 |
|
76 |
-
text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=0,separators=["\n\n", "\n",".", " "])
|
77 |
split_contents = text_splitter.split_text(cleaned_full_text)
|
78 |
embeddings = embeddings_model.embed_documents(split_contents)
|
79 |
|
80 |
-
X = np.array(embeddings)
|
81 |
kmeans = KMeans(n_clusters=num_clusters, init='k-means++', random_state=0).fit(embeddings)
|
82 |
-
|
83 |
-
|
84 |
-
closest_point_indices = []
|
85 |
-
for center in cluster_centers:
|
86 |
-
distances = np.linalg.norm(embeddings - center, axis=1)
|
87 |
-
closest_point_indices.append(np.argmin(distances))
|
88 |
-
|
89 |
extracted_contents = [split_contents[idx] for idx in closest_point_indices]
|
90 |
-
results = chain.invoke({"topic": ' '.join(extracted_contents)})
|
91 |
-
|
92 |
-
summary_sentences = re.split(r'(?<!\w\.\w.)(?<![A-Z][a-z]\.)(?<=\.|\?)\s', results)
|
93 |
-
summary_embeddings = embeddings_model.embed_documents(summary_sentences)
|
94 |
-
extracted_embeddings = embeddings_model.embed_documents(extracted_contents)
|
95 |
-
similarity_matrix = cosine_similarity(summary_embeddings, extracted_embeddings)
|
96 |
|
97 |
-
|
98 |
-
relevant_sources = []
|
99 |
-
source_mapping = {}
|
100 |
-
sentence_to_source = {}
|
101 |
-
similarity_threshold = 0.6
|
102 |
|
103 |
-
|
104 |
-
if sentence in sentence_to_source:
|
105 |
-
continue
|
106 |
-
max_similarity = max(similarity_matrix[i])
|
107 |
-
if max_similarity >= similarity_threshold:
|
108 |
-
most_similar_idx = np.argmax(similarity_matrix[i])
|
109 |
-
if most_similar_idx not in source_mapping:
|
110 |
-
source_mapping[most_similar_idx] = len(relevant_sources) + 1
|
111 |
-
relevant_sources.append((most_similar_idx, extracted_contents[most_similar_idx]))
|
112 |
-
citation_idx = source_mapping[most_similar_idx]
|
113 |
-
citation = f"([Source {citation_idx}](#source-{citation_idx}))"
|
114 |
-
cited_sentence = re.sub(r'([.!?])$', f" {citation}\\1", sentence)
|
115 |
-
sentence_to_source[sentence] = citation_idx
|
116 |
-
cited_results = cited_results.replace(sentence, cited_sentence)
|
117 |
-
|
118 |
-
sources_list = "\n\n## Sources:\n"
|
119 |
-
for idx, (original_idx, content) in enumerate(relevant_sources):
|
120 |
-
sources_list += f"""
|
121 |
-
<details style="margin: 10px 0; padding: 10px; border: 1px solid #ccc; border-radius: 5px; background-color: #f9f9f9;">
|
122 |
-
<summary style="font-weight: bold; cursor: pointer;">Source {idx + 1}</summary>
|
123 |
-
<pre style="white-space: pre-wrap; word-wrap: break-word; margin-top: 10px;">{content}</pre>
|
124 |
-
</details>
|
125 |
-
"""
|
126 |
-
cited_results += sources_list
|
127 |
-
return cited_results
|
128 |
|
129 |
def qa_pdf(pdf_file_path, query, num_clusters=5, similarity_threshold=0.6):
|
130 |
-
global global_embeddings, global_split_contents
|
131 |
-
|
132 |
-
# Initialize models and embeddings
|
133 |
embeddings_model = OpenAIEmbeddings(model="text-embedding-3-small", api_key=openai_api_key)
|
134 |
llm = ChatOpenAI(model="gpt-3.5-turbo", api_key=openai_api_key, temperature=0.3)
|
135 |
prompt = ChatPromptTemplate.from_template(
|
@@ -142,40 +99,37 @@ def qa_pdf(pdf_file_path, query, num_clusters=5, similarity_threshold=0.6):
|
|
142 |
output_parser = StrOutputParser()
|
143 |
chain = prompt | llm | output_parser
|
144 |
|
145 |
-
|
146 |
-
|
147 |
-
|
148 |
-
|
149 |
-
full_text = "\n".join(doc.page_content for doc in docs)
|
150 |
-
cleaned_full_text = remove_references(full_text)
|
151 |
-
cleaned_full_text = clean_text(cleaned_full_text)
|
152 |
|
153 |
-
|
154 |
-
|
155 |
-
|
156 |
|
157 |
-
# Embed the query and find the most relevant contexts
|
158 |
query_embedding = embeddings_model.embed_query(query)
|
159 |
-
similarity_scores = cosine_similarity([query_embedding],
|
160 |
top_indices = np.argsort(similarity_scores)[-num_clusters:]
|
161 |
-
relevant_contents = [
|
162 |
|
163 |
-
# Generate the answer using the LLM chain
|
164 |
results = chain.invoke({"question": query, "contexts": ' '.join(relevant_contents)})
|
165 |
|
166 |
-
|
167 |
-
answer_sentences = re.split(r'(?<!\w\.\w.)(?<![A-Z][a-z]\.)(?<=\.|\?)\s', results)
|
168 |
-
answer_embeddings = embeddings_model.embed_documents(answer_sentences)
|
169 |
-
relevant_embeddings = embeddings_model.embed_documents(relevant_contents)
|
170 |
-
similarity_matrix = cosine_similarity(answer_embeddings, relevant_embeddings)
|
171 |
|
172 |
-
|
173 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
174 |
relevant_sources = []
|
175 |
source_mapping = {}
|
176 |
sentence_to_source = {}
|
177 |
|
178 |
-
for i, sentence in enumerate(
|
179 |
if sentence in sentence_to_source:
|
180 |
continue
|
181 |
max_similarity = max(similarity_matrix[i])
|
@@ -183,34 +137,42 @@ def qa_pdf(pdf_file_path, query, num_clusters=5, similarity_threshold=0.6):
|
|
183 |
most_similar_idx = np.argmax(similarity_matrix[i])
|
184 |
if most_similar_idx not in source_mapping:
|
185 |
source_mapping[most_similar_idx] = len(relevant_sources) + 1
|
186 |
-
relevant_sources.append((most_similar_idx,
|
187 |
citation_idx = source_mapping[most_similar_idx]
|
188 |
-
citation = f"
|
189 |
cited_sentence = re.sub(r'([.!?])$', f" {citation}\\1", sentence)
|
190 |
sentence_to_source[sentence] = citation_idx
|
191 |
-
|
192 |
|
193 |
-
# Format the sources for markdown rendering
|
194 |
sources_list = "\n\n## Sources:\n"
|
195 |
for idx, (original_idx, content) in enumerate(relevant_sources):
|
196 |
-
sources_list +=
|
197 |
-
<details style="margin:
|
198 |
-
<summary style="font-weight: bold; cursor: pointer;">Source {idx + 1}</summary>
|
199 |
-
<pre style="white-space: pre-wrap; word-wrap: break-word; margin
|
200 |
</details>
|
201 |
"""
|
202 |
-
cited_results += sources_list
|
203 |
-
return cited_results
|
204 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
205 |
|
206 |
-
|
|
|
|
|
|
|
|
|
|
|
207 |
results = model.predict(image)
|
208 |
-
|
209 |
(int(box.xyxy[0][0]), int(box.xyxy[0][1]), int(box.xyxy[0][2]), int(box.xyxy[0][3]), int(box.cls[0]))
|
210 |
for result in results for box in result.boxes
|
211 |
if int(box.cls[0]) in {figure_class_index, table_class_index} and box.conf[0] > confidence_threshold
|
212 |
]
|
213 |
-
return boxes
|
214 |
|
215 |
def crop_images_from_boxes(image, boxes, scale_factor):
|
216 |
figures = []
|
@@ -223,7 +185,6 @@ def crop_images_from_boxes(image, boxes, scale_factor):
|
|
223 |
tables.append(cropped_img)
|
224 |
return figures, tables
|
225 |
|
226 |
-
|
227 |
def process_pdf(pdf_file_path):
|
228 |
doc = fitz.open(pdf_file_path)
|
229 |
all_figures = []
|
@@ -246,213 +207,173 @@ def process_pdf(pdf_file_path):
|
|
246 |
|
247 |
return all_figures, all_tables
|
248 |
|
249 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
250 |
|
251 |
-
#
|
252 |
-
|
253 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
254 |
|
255 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
256 |
st.markdown("""
|
257 |
<style>
|
258 |
-
|
259 |
-
|
260 |
-
|
261 |
-
padding:
|
262 |
-
|
|
|
|
|
|
|
|
|
263 |
}
|
264 |
-
|
265 |
-
|
266 |
-
.section-header {
|
267 |
-
font-size: 2rem;
|
268 |
-
font-weight: bold;
|
269 |
-
color: #343a40;
|
270 |
-
margin-top: 2rem;
|
271 |
-
margin-bottom: 1rem;
|
272 |
-
text-align: center;
|
273 |
-
font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif;
|
274 |
}
|
275 |
-
|
276 |
-
|
277 |
-
|
278 |
-
|
279 |
-
|
280 |
-
|
281 |
-
margin-bottom: 2rem;
|
282 |
-
box-shadow: 0 4px 12px rgba(0, 0, 0, 0.1);
|
283 |
}
|
284 |
-
|
285 |
-
|
286 |
-
.stButton>button {
|
287 |
-
background-color: #007bff;
|
288 |
-
color: white;
|
289 |
-
padding: 0.6rem 1.2rem;
|
290 |
-
border-radius: 5px;
|
291 |
-
border: none;
|
292 |
-
cursor: pointer;
|
293 |
-
font-size: 1rem;
|
294 |
-
transition: background-color 0.3s ease, transform 0.3s ease;
|
295 |
}
|
296 |
-
.
|
297 |
-
background-color: #
|
298 |
-
transform: translateY(-2px);
|
299 |
}
|
300 |
-
|
301 |
-
|
302 |
-
.chat-message {
|
303 |
-
padding: 1rem;
|
304 |
-
border-radius: 10px;
|
305 |
-
margin-bottom: 1rem;
|
306 |
-
font-size: 1rem;
|
307 |
-
transition: all 0.3s ease;
|
308 |
-
box-shadow: 0 2px 8px rgba(0, 0, 0, 0.1);
|
309 |
-
}
|
310 |
-
.chat-message.user {
|
311 |
-
background-color: #e6f7ff;
|
312 |
-
border-left: 5px solid #007bff;
|
313 |
-
text-align: left;
|
314 |
}
|
315 |
-
.
|
316 |
-
background-color: #
|
317 |
-
border-left: 5px solid #dc3545;
|
318 |
-
text-align: left;
|
319 |
-
}
|
320 |
-
|
321 |
-
/* Input area */
|
322 |
-
.input-container {
|
323 |
-
display: flex;
|
324 |
-
align-items: center;
|
325 |
-
gap: 10px;
|
326 |
-
margin-top: 1rem;
|
327 |
}
|
328 |
-
|
329 |
-
border: 2px solid #ccc;
|
330 |
-
border-radius: 10px;
|
331 |
-
padding: 10px;
|
332 |
width: 100%;
|
333 |
-
|
334 |
-
|
335 |
-
|
336 |
-
|
|
|
337 |
}
|
338 |
-
|
339 |
-
border-color: #
|
340 |
-
|
341 |
}
|
342 |
-
.
|
343 |
-
|
|
|
344 |
color: white;
|
345 |
-
padding: 0.6rem 1.2rem;
|
346 |
-
border-radius: 5px;
|
347 |
-
border: none;
|
348 |
-
cursor: pointer;
|
349 |
-
font-size: 1rem;
|
350 |
-
transition: background-color 0.3s ease, transform 0.3s ease;
|
351 |
-
}
|
352 |
-
.input-container button:hover {
|
353 |
-
background-color: #0056b3;
|
354 |
-
transform: translateY(-2px);
|
355 |
-
}
|
356 |
-
|
357 |
-
/* Expander */
|
358 |
-
.st-expander {
|
359 |
border: none;
|
360 |
-
|
361 |
-
|
|
|
|
|
|
|
362 |
}
|
363 |
-
|
364 |
-
|
365 |
-
|
366 |
-
font-size: 1rem;
|
367 |
-
color: #343a40;
|
368 |
-
line-height: 1.6;
|
369 |
-
}
|
370 |
-
|
371 |
-
/* Titles and subtitles */
|
372 |
-
.stTitle {
|
373 |
-
color: #343a40;
|
374 |
-
text-align: center;
|
375 |
-
margin-bottom: 1rem;
|
376 |
-
font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif;
|
377 |
-
}
|
378 |
-
.stSubtitle {
|
379 |
-
color: #6c757d;
|
380 |
-
text-align: center;
|
381 |
-
margin-bottom: 1rem;
|
382 |
-
font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif;
|
383 |
}
|
384 |
</style>
|
|
|
|
|
|
|
|
|
385 |
""", unsafe_allow_html=True)
|
386 |
-
|
387 |
-
# Streamlit interface
|
388 |
-
# Streamlit interface
|
389 |
-
st.title("📄 PDF Reading Assistant")
|
390 |
-
st.markdown("### Extract tables, figures, summaries, and answers from your PDF files easily.")
|
391 |
-
|
392 |
-
uploaded_file = st.file_uploader("Upload a PDF", type="pdf")
|
393 |
-
if uploaded_file:
|
394 |
-
file_path = save_uploaded_file(uploaded_file)
|
395 |
-
|
396 |
-
if 'figures' not in st.session_state:
|
397 |
-
st.session_state['figures'] = None
|
398 |
-
if 'tables' not in st.session_state:
|
399 |
-
st.session_state['tables'] = None
|
400 |
-
if 'summary' not in st.session_state:
|
401 |
-
st.session_state['summary'] = None
|
402 |
-
|
403 |
-
with st.container():
|
404 |
-
st.markdown("<div class='section-header'>Extract Tables and Figures</div>", unsafe_allow_html=True)
|
405 |
-
with st.expander("Click to Extract Tables and Figures", expanded=True):
|
406 |
-
with st.container():
|
407 |
-
extract_button = st.button("Extract")
|
408 |
-
if extract_button:
|
409 |
-
figures, tables = process_pdf(file_path)
|
410 |
-
st.session_state['figures'] = figures
|
411 |
-
st.session_state['tables'] = tables
|
412 |
-
|
413 |
-
if st.session_state['figures']:
|
414 |
-
col1, col2 = st.columns(2)
|
415 |
-
with col1:
|
416 |
-
st.write("### Figures")
|
417 |
-
for figure in st.session_state['figures']:
|
418 |
-
st.image(figure, use_column_width=True)
|
419 |
-
with col2:
|
420 |
-
st.write("### Tables")
|
421 |
-
for table in st.session_state['tables']:
|
422 |
-
st.image(table, use_column_width=True)
|
423 |
-
else:
|
424 |
-
st.write("No figures or tables found.")
|
425 |
-
|
426 |
-
with st.container():
|
427 |
-
st.markdown("<div class='section-header'>Get Summary</div>", unsafe_allow_html=True)
|
428 |
-
with st.expander("Click to Generate Summary", expanded=True):
|
429 |
-
with st.container():
|
430 |
-
summary_button = st.button("Generate Summary")
|
431 |
-
if summary_button:
|
432 |
-
summary = summarize_pdf(file_path)
|
433 |
-
st.session_state['summary'] = summary
|
434 |
-
|
435 |
-
if st.session_state['summary']:
|
436 |
-
st.markdown(st.session_state['summary'], unsafe_allow_html=True)
|
437 |
-
|
438 |
-
with st.container():
|
439 |
-
st.markdown("<div class='section-header'>Chat with your PDF</div>", unsafe_allow_html=True)
|
440 |
-
st.write("### Chat with your PDF")
|
441 |
-
if 'chat_history' not in st.session_state:
|
442 |
-
st.session_state['chat_history'] = []
|
443 |
-
|
444 |
-
for chat in st.session_state['chat_history']:
|
445 |
-
chat_user_class = "user" if chat["user"] else ""
|
446 |
-
chat_bot_class = "bot" if chat["bot"] else ""
|
447 |
-
st.markdown(f"<div class='chat-message {chat_user_class}'>{chat['user']}</div>", unsafe_allow_html=True)
|
448 |
-
st.markdown(f"<div class='chat-message {chat_bot_class}'>{chat['bot']}</div>", unsafe_allow_html=True)
|
449 |
-
|
450 |
-
with st.form(key="chat_form", clear_on_submit=True):
|
451 |
-
user_input = st.text_area("Ask a question about the PDF:", key="user_input")
|
452 |
-
submit_button = st.form_submit_button(label="Send")
|
453 |
-
|
454 |
-
if submit_button and user_input:
|
455 |
-
st.session_state['chat_history'].append({"user": user_input, "bot": None})
|
456 |
-
answer = qa_pdf(file_path, user_input)
|
457 |
-
st.session_state['chat_history'][-1]["bot"] = answer
|
458 |
-
st.rerun()
|
|
|
1 |
|
2 |
|
3 |
+
# Load the trained model
|
4 |
+
model = YOLO("best.pt")
|
5 |
+
openai_api_key = os.environ.get("openai_api_key")
|
6 |
|
7 |
import os
|
8 |
+
import io
|
9 |
+
import base64
|
10 |
import streamlit as st
|
11 |
import numpy as np
|
12 |
import fitz # PyMuPDF
|
13 |
+
import tempfile
|
14 |
from ultralytics import YOLO
|
15 |
from sklearn.cluster import KMeans
|
16 |
from sklearn.metrics.pairwise import cosine_similarity
|
|
|
19 |
from langchain_openai import OpenAIEmbeddings
|
20 |
from langchain_text_splitters import RecursiveCharacterTextSplitter
|
21 |
from langchain_core.prompts import ChatPromptTemplate
|
|
|
22 |
from langchain_openai import ChatOpenAI
|
|
|
23 |
import re
|
24 |
+
from PIL import Image
|
25 |
+
from streamlit_chat import message
|
26 |
|
27 |
# Load the trained model
|
28 |
+
model = YOLO("runs\\detect\\train7\\weights\\best.pt")
|
29 |
+
openai_api_key = "sk-proj-J7kj0kbG1m0eIMPWMdjoT3BlbkFJqwZNqQeOYJ9UH6I0efPi"
|
30 |
|
31 |
# Define the class indices for figures, tables, and text
|
32 |
+
figure_class_index = 4
|
33 |
+
table_class_index = 3
|
|
|
|
|
|
|
|
|
34 |
|
35 |
+
# Utility functions
|
36 |
def clean_text(text):
|
37 |
+
return re.sub(r'\s+', ' ', text).strip()
|
|
|
38 |
|
39 |
def remove_references(text):
|
40 |
reference_patterns = [
|
|
|
48 |
return text
|
49 |
|
50 |
def save_uploaded_file(uploaded_file):
|
51 |
+
temp_file = tempfile.NamedTemporaryFile(delete=False)
|
52 |
+
temp_file.write(uploaded_file.getbuffer())
|
53 |
+
temp_file.close()
|
54 |
+
return temp_file.name
|
55 |
|
56 |
def summarize_pdf(pdf_file_path, num_clusters=10):
|
57 |
embeddings_model = OpenAIEmbeddings(model="text-embedding-3-small", api_key=openai_api_key)
|
|
|
72 |
loader = PyMuPDFLoader(pdf_file_path)
|
73 |
docs = loader.load()
|
74 |
full_text = "\n".join(doc.page_content for doc in docs)
|
75 |
+
cleaned_full_text = clean_text(remove_references(full_text))
|
|
|
76 |
|
77 |
+
text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=0, separators=["\n\n", "\n", ".", " "])
|
78 |
split_contents = text_splitter.split_text(cleaned_full_text)
|
79 |
embeddings = embeddings_model.embed_documents(split_contents)
|
80 |
|
|
|
81 |
kmeans = KMeans(n_clusters=num_clusters, init='k-means++', random_state=0).fit(embeddings)
|
82 |
+
closest_point_indices = [np.argmin(np.linalg.norm(embeddings - center, axis=1)) for center in kmeans.cluster_centers_]
|
|
|
|
|
|
|
|
|
|
|
|
|
83 |
extracted_contents = [split_contents[idx] for idx in closest_point_indices]
|
|
|
|
|
|
|
|
|
|
|
|
|
84 |
|
85 |
+
results = chain.invoke({"topic": ' '.join(extracted_contents)})
|
|
|
|
|
|
|
|
|
86 |
|
87 |
+
return generate_citations(results, extracted_contents)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
88 |
|
89 |
def qa_pdf(pdf_file_path, query, num_clusters=5, similarity_threshold=0.6):
|
|
|
|
|
|
|
90 |
embeddings_model = OpenAIEmbeddings(model="text-embedding-3-small", api_key=openai_api_key)
|
91 |
llm = ChatOpenAI(model="gpt-3.5-turbo", api_key=openai_api_key, temperature=0.3)
|
92 |
prompt = ChatPromptTemplate.from_template(
|
|
|
99 |
output_parser = StrOutputParser()
|
100 |
chain = prompt | llm | output_parser
|
101 |
|
102 |
+
loader = PyMuPDFLoader(pdf_file_path)
|
103 |
+
docs = loader.load()
|
104 |
+
full_text = "\n".join(doc.page_content for doc in docs)
|
105 |
+
cleaned_full_text = clean_text(remove_references(full_text))
|
|
|
|
|
|
|
106 |
|
107 |
+
text_splitter = RecursiveCharacterTextSplitter(chunk_size=800, chunk_overlap=0, separators=["\n\n", "\n", ".", " "])
|
108 |
+
split_contents = text_splitter.split_text(cleaned_full_text)
|
109 |
+
embeddings = embeddings_model.embed_documents(split_contents)
|
110 |
|
|
|
111 |
query_embedding = embeddings_model.embed_query(query)
|
112 |
+
similarity_scores = cosine_similarity([query_embedding], embeddings)[0]
|
113 |
top_indices = np.argsort(similarity_scores)[-num_clusters:]
|
114 |
+
relevant_contents = [split_contents[i] for i in top_indices]
|
115 |
|
|
|
116 |
results = chain.invoke({"question": query, "contexts": ' '.join(relevant_contents)})
|
117 |
|
118 |
+
return generate_citations(results, relevant_contents, similarity_threshold)
|
|
|
|
|
|
|
|
|
119 |
|
120 |
+
def generate_citations(text, contents, similarity_threshold=0.6):
|
121 |
+
embeddings_model = OpenAIEmbeddings(model="text-embedding-3-small", api_key=openai_api_key)
|
122 |
+
text_sentences = re.split(r'(?<!\w\.\w.)(?<![A-Z][a-z]\.)(?<=\.|\?)\s', text)
|
123 |
+
text_embeddings = embeddings_model.embed_documents(text_sentences)
|
124 |
+
content_embeddings = embeddings_model.embed_documents(contents)
|
125 |
+
similarity_matrix = cosine_similarity(text_embeddings, content_embeddings)
|
126 |
+
|
127 |
+
cited_text = text
|
128 |
relevant_sources = []
|
129 |
source_mapping = {}
|
130 |
sentence_to_source = {}
|
131 |
|
132 |
+
for i, sentence in enumerate(text_sentences):
|
133 |
if sentence in sentence_to_source:
|
134 |
continue
|
135 |
max_similarity = max(similarity_matrix[i])
|
|
|
137 |
most_similar_idx = np.argmax(similarity_matrix[i])
|
138 |
if most_similar_idx not in source_mapping:
|
139 |
source_mapping[most_similar_idx] = len(relevant_sources) + 1
|
140 |
+
relevant_sources.append((most_similar_idx, contents[most_similar_idx]))
|
141 |
citation_idx = source_mapping[most_similar_idx]
|
142 |
+
citation = f"([Source {citation_idx}](#source-{citation_idx}))"
|
143 |
cited_sentence = re.sub(r'([.!?])$', f" {citation}\\1", sentence)
|
144 |
sentence_to_source[sentence] = citation_idx
|
145 |
+
cited_text = cited_text.replace(sentence, cited_sentence)
|
146 |
|
|
|
147 |
sources_list = "\n\n## Sources:\n"
|
148 |
for idx, (original_idx, content) in enumerate(relevant_sources):
|
149 |
+
sources_list += f"""
|
150 |
+
<details style="margin: 1px 0; padding: 5px; border: 1px solid #ccc; border-radius: 8px; background-color: #f9f9f9; transition: all 0.3s ease;">
|
151 |
+
<summary style="font-weight: bold; cursor: pointer; outline: none; padding: 5px 0; transition: color 0.3s ease;">Source {idx + 1}</summary>
|
152 |
+
<pre style="white-space: pre-wrap; word-wrap: break-word; margin: 1px 0; padding: 10px; background-color: #fff; border-radius: 5px; border: 1px solid #ddd; box-shadow: 0 2px 5px rgba(0, 0, 0, 0.1);">{content}</pre>
|
153 |
</details>
|
154 |
"""
|
|
|
|
|
155 |
|
156 |
+
# Add dummy blanks after the last source
|
157 |
+
dummy_blanks = """
|
158 |
+
<div style="margin: 20px 0;"></div>
|
159 |
+
<div style="margin: 20px 0;"></div>
|
160 |
+
<div style="margin: 20px 0;"></div>
|
161 |
+
<div style="margin: 20px 0;"></div>
|
162 |
+
<div style="margin: 20px 0;"></div>
|
163 |
|
164 |
+
"""
|
165 |
+
|
166 |
+
cited_text += sources_list + dummy_blanks
|
167 |
+
return cited_text
|
168 |
+
|
169 |
+
def infer_image_and_get_boxes(image, confidence_threshold=0.8):
|
170 |
results = model.predict(image)
|
171 |
+
return [
|
172 |
(int(box.xyxy[0][0]), int(box.xyxy[0][1]), int(box.xyxy[0][2]), int(box.xyxy[0][3]), int(box.cls[0]))
|
173 |
for result in results for box in result.boxes
|
174 |
if int(box.cls[0]) in {figure_class_index, table_class_index} and box.conf[0] > confidence_threshold
|
175 |
]
|
|
|
176 |
|
177 |
def crop_images_from_boxes(image, boxes, scale_factor):
|
178 |
figures = []
|
|
|
185 |
tables.append(cropped_img)
|
186 |
return figures, tables
|
187 |
|
|
|
188 |
def process_pdf(pdf_file_path):
|
189 |
doc = fitz.open(pdf_file_path)
|
190 |
all_figures = []
|
|
|
207 |
|
208 |
return all_figures, all_tables
|
209 |
|
210 |
+
def image_to_base64(img):
|
211 |
+
buffered = io.BytesIO()
|
212 |
+
img = Image.fromarray(img)
|
213 |
+
img.save(buffered, format="PNG")
|
214 |
+
return base64.b64encode(buffered.getvalue()).decode()
|
215 |
+
|
216 |
+
def on_btn_click():
|
217 |
+
del st.session_state.chat_history[:]
|
218 |
|
219 |
+
# Streamlit interface
|
220 |
+
|
221 |
+
# Custom CSS for the file uploader
|
222 |
+
uploadercss='''
|
223 |
+
<style>
|
224 |
+
[data-testid='stFileUploader'] {
|
225 |
+
width: max-content;
|
226 |
+
}
|
227 |
+
[data-testid='stFileUploader'] section {
|
228 |
+
padding: 0;
|
229 |
+
float: left;
|
230 |
+
}
|
231 |
+
[data-testid='stFileUploader'] section > input + div {
|
232 |
+
display: none;
|
233 |
+
}
|
234 |
+
[data-testid='stFileUploader'] section + div {
|
235 |
+
float: right;
|
236 |
+
padding-top: 0;
|
237 |
+
}
|
238 |
+
|
239 |
+
</style>
|
240 |
+
'''
|
241 |
+
|
242 |
+
st.set_page_config(page_title="PDF Reading Assistant", page_icon="📄")
|
243 |
+
|
244 |
+
# Initialize chat history in session state if not already present
|
245 |
+
if 'chat_history' not in st.session_state:
|
246 |
+
st.session_state.chat_history = []
|
247 |
+
|
248 |
+
st.title("📄 PDF Reading Assistant")
|
249 |
+
st.markdown("### Extract tables, figures, summaries, and answers from your PDF files easily.")
|
250 |
+
chat_placeholder = st.empty()
|
251 |
+
|
252 |
+
# File uploader for PDF
|
253 |
+
uploaded_file = st.file_uploader("Upload a PDF", type="pdf")
|
254 |
+
st.markdown(uploadercss, unsafe_allow_html=True)
|
255 |
+
if uploaded_file:
|
256 |
+
file_path = save_uploaded_file(uploaded_file)
|
257 |
|
258 |
+
# Chat container where all messages will be displayed
|
259 |
+
chat_container = st.container()
|
260 |
+
user_input = st.chat_input("Ask a question about the pdf......", key="user_input")
|
261 |
+
with chat_container:
|
262 |
+
# Scrollable chat messages
|
263 |
+
for idx, chat in enumerate(st.session_state.chat_history):
|
264 |
+
if chat.get("user"):
|
265 |
+
message(chat["user"], is_user=True, allow_html=True, key=f"user_{idx}", avatar_style="initials", seed="user")
|
266 |
+
if chat.get("bot"):
|
267 |
+
message(chat["bot"], is_user=False, allow_html=True, key=f"bot_{idx}",seed="bot")
|
268 |
+
|
269 |
+
# Input area and buttons for user interaction
|
270 |
+
with st.form(key="chat_form", clear_on_submit=True,border=False):
|
271 |
+
|
272 |
+
col1, col2, col3 = st.columns([1, 1, 1])
|
273 |
+
with col1:
|
274 |
+
summary_button = st.form_submit_button("Generate Summary")
|
275 |
+
with col2:
|
276 |
+
extract_button = st.form_submit_button("Extract Tables and Figures")
|
277 |
+
with col3:
|
278 |
+
st.form_submit_button("Clear message", on_click=on_btn_click)
|
279 |
+
|
280 |
+
# Handle responses based on user input and button presses
|
281 |
+
if summary_button:
|
282 |
+
with st.spinner("Generating summary..."):
|
283 |
+
summary = summarize_pdf(file_path)
|
284 |
+
st.session_state.chat_history.append({"user": "Generate Summary", "bot": summary})
|
285 |
+
st.rerun()
|
286 |
+
|
287 |
+
if extract_button:
|
288 |
+
with st.spinner("Extracting tables and figures..."):
|
289 |
+
figures, tables = process_pdf(file_path)
|
290 |
+
if figures:
|
291 |
+
st.session_state.chat_history.append({"user": "Figures"})
|
292 |
+
|
293 |
+
for idx, figure in enumerate(figures):
|
294 |
+
figure_base64 = image_to_base64(figure)
|
295 |
+
result_html = f'<img src="data:image/png;base64,{figure_base64}" style="width:100%; display:block;" alt="Figure {idx+1}"/>'
|
296 |
+
st.session_state.chat_history.append({"bot": f"Figure {idx+1} {result_html}"})
|
297 |
+
if tables:
|
298 |
+
st.session_state.chat_history.append({"user": "Tables"})
|
299 |
+
for idx, table in enumerate(tables):
|
300 |
+
table_base64 = image_to_base64(table)
|
301 |
+
result_html = f'<img src="data:image/png;base64,{table_base64}" style="width:100%; display:block;" alt="Table {idx+1}"/>'
|
302 |
+
st.session_state.chat_history.append({"bot": f"Table {idx+1} {result_html}"})
|
303 |
+
st.rerun()
|
304 |
+
|
305 |
+
if user_input:
|
306 |
+
st.session_state.chat_history.append({"user": user_input, "bot": None})
|
307 |
+
with st.spinner("Processing..."):
|
308 |
+
answer = qa_pdf(file_path, user_input)
|
309 |
+
st.session_state.chat_history[-1]["bot"] = answer
|
310 |
+
st.rerun()
|
311 |
+
|
312 |
+
# Additional CSS and JavaScript to ensure the chat container is scrollable and scrolls to the bottom
|
313 |
st.markdown("""
|
314 |
<style>
|
315 |
+
#chat-container {
|
316 |
+
max-height: 500px;
|
317 |
+
overflow-y: auto;
|
318 |
+
padding: 1rem;
|
319 |
+
border: 1px solid #ddd;
|
320 |
+
border-radius: 8px;
|
321 |
+
background-color: #fefefe;
|
322 |
+
box-shadow: 0 2px 4px rgba(0, 0, 0, 0.1);
|
323 |
+
transition: background-color 0.3s ease;
|
324 |
}
|
325 |
+
#chat-container:hover {
|
326 |
+
background-color: #f9f9f9;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
327 |
}
|
328 |
+
.stChatMessage {
|
329 |
+
padding: 0.75rem;
|
330 |
+
margin: 0.75rem 0;
|
331 |
+
border-radius: 8px;
|
332 |
+
box-shadow: 0 1px 3px rgba(0, 0, 0, 0.1);
|
333 |
+
transition: background-color 0.3s ease;
|
|
|
|
|
334 |
}
|
335 |
+
.stChatMessage--user {
|
336 |
+
background-color: #E3F2FD;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
337 |
}
|
338 |
+
.stChatMessage--user:hover {
|
339 |
+
background-color: #BBDEFB;
|
|
|
340 |
}
|
341 |
+
.stChatMessage--bot {
|
342 |
+
background-color: #EDE7F6;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
343 |
}
|
344 |
+
.stChatMessage--bot:hover {
|
345 |
+
background-color: #D1C4E9;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
346 |
}
|
347 |
+
textarea {
|
|
|
|
|
|
|
348 |
width: 100%;
|
349 |
+
padding: 1rem;
|
350 |
+
border: 1px solid #ddd;
|
351 |
+
border-radius: 8px;
|
352 |
+
box-shadow: inset 0 1px 3px rgba(0, 0, 0, 0.1);
|
353 |
+
transition: border-color 0.3s ease, box-shadow 0.3s ease;
|
354 |
}
|
355 |
+
textarea:focus {
|
356 |
+
border-color: #4CAF50;
|
357 |
+
box-shadow: 0 0 5px rgba(76, 175, 80, 0.5);
|
358 |
}
|
359 |
+
.stButton > button {
|
360 |
+
width: 100%;
|
361 |
+
background-color: #4CAF50;
|
362 |
color: white;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
363 |
border: none;
|
364 |
+
border-radius: 8px;
|
365 |
+
padding: 0.75rem;
|
366 |
+
font-size: 16px;
|
367 |
+
box-shadow: 0 2px 4px rgba(0, 0, 0, 0.1);
|
368 |
+
transition: background-color 0.3s ease, box-shadow 0.3s ease;
|
369 |
}
|
370 |
+
.stButton > button:hover {
|
371 |
+
background-color: #45A049;
|
372 |
+
box-shadow: 0 4px 8px rgba(0, 0, 0, 0.1);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
373 |
}
|
374 |
</style>
|
375 |
+
<script>
|
376 |
+
const chatContainer = document.getElementById('chat-container');
|
377 |
+
chatContainer.scrollTop = chatContainer.scrollHeight;
|
378 |
+
</script>
|
379 |
""", unsafe_allow_html=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|