CosmickVisions commited on
Commit
0b83924
·
verified ·
1 Parent(s): 5290c0d

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +471 -100
app.py CHANGED
@@ -15,6 +15,13 @@ import requests
15
  import json
16
  import re
17
  from datetime import datetime, timedelta
 
 
 
 
 
 
 
18
 
19
  # Load environment variables
20
  load_dotenv()
@@ -29,6 +36,25 @@ if not os.path.exists(FAISS_INDEX_DIR):
29
  # Dictionary to store user-specific vectorstores
30
  user_vectorstores = {}
31
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
32
  # Custom CSS for Tech theme
33
  custom_css = """
34
  :root {
@@ -70,6 +96,102 @@ body { background-color: var(--light-background); font-family: 'Google Sans', 'R
70
  .qa-body { color: var(--dark-text); font-size: 0.95rem; margin-bottom: 10px; }
71
  .qa-meta { display: flex; justify-content: space-between; color: #5F6368; font-size: 0.85rem; }
72
  .tag { background-color: #E8F0FE; color: var(--primary-color); padding: 4px 8px; border-radius: 4px; font-size: 0.8rem; margin-right: 5px; display: inline-block; }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
73
  """
74
 
75
  # Function to process PDF files
@@ -109,8 +231,207 @@ def process_pdf(pdf_file):
109
  os.unlink(pdf_path)
110
  return None, f"Error processing PDF: {str(e)}", {"page_images": [], "total_pages": 0, "total_words": 0}
111
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
112
  # Function to generate chatbot responses with Tech theme
113
- def generate_response(message, session_id, model_name, history):
114
  if not message:
115
  return history
116
  try:
@@ -121,8 +442,8 @@ def generate_response(message, session_id, model_name, history):
121
  if docs:
122
  context = "\n\nRelevant information from uploaded PDF:\n" + "\n".join(f"- {doc.page_content}" for doc in docs)
123
 
124
- # Check if it's a GitHub repo search
125
- if re.match(r'^/github\s+.+', message, re.IGNORECASE):
126
  query = re.sub(r'^/github\s+', '', message, flags=re.IGNORECASE)
127
  repo_results = search_github_repos(query)
128
  if repo_results:
@@ -139,8 +460,8 @@ def generate_response(message, session_id, model_name, history):
139
  history.append((message, "No GitHub repositories found for your query."))
140
  return history
141
 
142
- # Check if it's a Stack Overflow search
143
- if re.match(r'^/stack\s+.+', message, re.IGNORECASE):
144
  query = re.sub(r'^/stack\s+', '', message, flags=re.IGNORECASE)
145
  qa_results = search_stackoverflow(query)
146
  if qa_results:
@@ -437,106 +758,144 @@ def perform_stack_search(query, tag, sort_by):
437
  with gr.Blocks(css=custom_css, theme=gr.themes.Soft()) as demo:
438
  current_session_id = gr.State(None)
439
  pdf_state = gr.State({"page_images": [], "total_pages": 0, "total_words": 0})
 
 
 
 
440
  gr.HTML("""
441
  <div class="header">
442
- <div class="header-title">Tech-Vision</div>
443
- <div class="header-subtitle">Analyze technical documents with Groq's LLM API.</div>
444
  </div>
445
  """)
446
  with gr.Row(elem_classes="container"):
447
  with gr.Column(scale=1, min_width=300):
448
- pdf_file = gr.File(label="Upload PDF Document", file_types=[".pdf"], type="binary")
449
- upload_button = gr.Button("Process PDF", variant="primary")
450
- pdf_status = gr.Markdown("No PDF uploaded yet")
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
451
  model_dropdown = gr.Dropdown(
452
  choices=["llama3-70b-8192", "llama3-8b-8192", "mixtral-8x7b-32768", "gemma-7b-it"],
453
  value="llama3-70b-8192",
454
  label="Select Groq Model"
455
  )
456
-
457
- # Tech Tools Section
458
- gr.Markdown("### Developer Tools", elem_classes="tool-title")
459
- with gr.Group(elem_classes="tool-container"):
460
- with gr.Tabs():
461
- with gr.TabItem("GitHub Search"):
462
- repo_query = gr.Textbox(label="Search Query", placeholder="Enter keywords to search for repositories")
463
- with gr.Row():
464
- language = gr.Dropdown(
465
- choices=["any", "JavaScript", "Python", "Java", "C++", "TypeScript", "Go", "Rust", "PHP", "C#"],
466
- value="any",
467
- label="Language"
468
- )
469
- min_stars = gr.Dropdown(
470
- choices=["0", "10", "50", "100", "1000", "10000"],
471
- value="0",
472
- label="Min Stars"
473
- )
474
- sort_by = gr.Dropdown(
475
- choices=["stars", "forks", "updated"],
476
- value="stars",
477
- label="Sort By"
478
- )
479
- repo_search_btn = gr.Button("Search Repositories")
480
-
481
- with gr.TabItem("Stack Overflow"):
482
- stack_query = gr.Textbox(label="Search Query", placeholder="Enter your technical question")
483
- with gr.Row():
484
- tag = gr.Dropdown(
485
- choices=["any", "python", "javascript", "java", "c++", "react", "node.js", "android", "ios", "sql"],
486
- value="any",
487
- label="Tag"
488
- )
489
- so_sort_by = gr.Dropdown(
490
- choices=["votes", "newest", "activity"],
491
- value="votes",
492
- label="Sort By"
493
- )
494
- so_search_btn = gr.Button("Search Stack Overflow")
495
-
496
- with gr.TabItem("Code Explainer"):
497
- code_input = gr.Textbox(
498
- label="Code to Explain",
499
- placeholder="Paste your code here...",
500
- lines=10
501
- )
502
- explain_btn = gr.Button("Explain Code")
503
-
504
  with gr.Column(scale=2, min_width=600):
505
  with gr.Tabs():
506
  with gr.TabItem("PDF Viewer"):
507
  with gr.Column(elem_classes="pdf-viewer-container"):
508
  page_slider = gr.Slider(minimum=1, maximum=1, step=1, label="Page Number", value=1)
509
  pdf_image = gr.Image(label="PDF Page", type="pil", elem_classes="pdf-viewer-image")
510
- stats_display = gr.Markdown("No PDF uploaded yet", elem_classes="stats-box")
511
 
512
- with gr.TabItem("GitHub Results"):
513
- repo_results = gr.Markdown("Search for repositories to see results here")
 
514
 
515
- with gr.TabItem("Stack Overflow Results"):
516
- stack_results = gr.Markdown("Search for questions to see results here")
517
-
518
- with gr.TabItem("Code Explanation"):
519
- code_explanation = gr.Markdown("Paste your code and click 'Explain Code' to see an explanation here")
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
520
 
 
521
  with gr.Row(elem_classes="container"):
522
  with gr.Column(scale=2, min_width=600):
523
- chatbot = gr.Chatbot(height=500, bubble_full_width=False, show_copy_button=True, elem_classes="chat-container")
524
  with gr.Row():
525
- msg = gr.Textbox(show_label=False, placeholder="Ask about your document, type /github to search repos, or /stack to search Stack Overflow...", scale=5)
 
 
 
 
 
526
  send_btn = gr.Button("Send", scale=1)
527
- clear_btn = gr.Button("Clear Conversation")
 
 
 
 
528
 
529
- # Event Handlers
530
- upload_button.click(
 
 
 
 
531
  process_pdf,
532
  inputs=[pdf_file],
533
- outputs=[current_session_id, pdf_status, pdf_state]
534
  ).then(
535
  update_pdf_viewer,
536
  inputs=[pdf_state],
537
- outputs=[page_slider, pdf_image, stats_display]
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
538
  )
539
 
 
540
  msg.submit(
541
  generate_response,
542
  inputs=[msg, current_session_id, model_dropdown, chatbot],
@@ -549,43 +908,55 @@ with gr.Blocks(css=custom_css, theme=gr.themes.Soft()) as demo:
549
  outputs=[chatbot]
550
  ).then(lambda: "", None, [msg])
551
 
552
- clear_btn.click(
553
- lambda: ([], None, "No PDF uploaded yet", {"page_images": [], "total_pages": 0, "total_words": 0}, 0, None, "No PDF uploaded yet"),
554
- None,
555
- [chatbot, current_session_id, pdf_status, pdf_state, page_slider, pdf_image, stats_display]
 
 
 
 
 
 
 
 
 
 
 
 
556
  )
557
 
 
558
  page_slider.change(
559
  update_image,
560
  inputs=[page_slider, pdf_state],
561
  outputs=[pdf_image]
562
  )
563
 
564
- # Tech tool handlers
565
- repo_search_btn.click(
566
- perform_repo_search,
567
- inputs=[repo_query, language, sort_by, min_stars],
568
- outputs=[repo_results]
569
- )
570
-
571
- so_search_btn.click(
572
- perform_stack_search,
573
- inputs=[stack_query, tag, so_sort_by],
574
- outputs=[stack_results]
575
- )
576
-
577
- explain_btn.click(
578
- explain_code,
579
- inputs=[code_input],
580
- outputs=[code_explanation]
581
  )
582
 
583
- # Add footer with attribution
584
- gr.HTML("""
585
- <div style="text-align: center; margin-top: 20px; padding: 10px; color: #666; font-size: 0.8rem; border-top: 1px solid #eee;">
586
- Created by Calvin Allen Crawford
587
- </div>
588
- """)
589
 
590
  # Launch the app
591
  if __name__ == "__main__":
 
15
  import json
16
  import re
17
  from datetime import datetime, timedelta
18
+ import speech_recognition as sr
19
+ import pyttsx3
20
+ import torch
21
+ from transformers import AutoProcessor, AutoModelForVision2Seq
22
+ import numpy as np
23
+ import pandas as pd
24
+ import openpyxl
25
 
26
  # Load environment variables
27
  load_dotenv()
 
36
  # Dictionary to store user-specific vectorstores
37
  user_vectorstores = {}
38
 
39
+ # Load SmolDocling model for image analysis
40
+ def load_docling_model():
41
+ try:
42
+ processor = AutoProcessor.from_pretrained("ds4sd/SmolDocling-256M-preview")
43
+ model = AutoModelForVision2Seq.from_pretrained("ds4sd/SmolDocling-256M-preview")
44
+ return processor, model
45
+ except Exception as e:
46
+ print(f"Error loading SmolDocling model: {e}")
47
+ return None, None
48
+
49
+ # Initialize SmolDocling model
50
+ docling_processor, docling_model = load_docling_model()
51
+
52
+ # Initialize text-to-speech engine
53
+ tts_engine = pyttsx3.init()
54
+ # Set properties for better speech
55
+ tts_engine.setProperty('rate', 150) # Speed of speech
56
+ tts_engine.setProperty('volume', 0.9) # Volume (0.0 to 1.0)
57
+
58
  # Custom CSS for Tech theme
59
  custom_css = """
60
  :root {
 
96
  .qa-body { color: var(--dark-text); font-size: 0.95rem; margin-bottom: 10px; }
97
  .qa-meta { display: flex; justify-content: space-between; color: #5F6368; font-size: 0.85rem; }
98
  .tag { background-color: #E8F0FE; color: var(--primary-color); padding: 4px 8px; border-radius: 4px; font-size: 0.8rem; margin-right: 5px; display: inline-block; }
99
+ .toggle-container { display: flex; align-items: center; margin-bottom: 15px; }
100
+ .toggle-label { margin-right: 10px; font-weight: 500; }
101
+ .search-toggle { margin-left: 5px; }
102
+ .voice-btn { background-color: var(--primary-color) !important; border-radius: 50% !important; width: 44px !important; height: 44px !important; display: flex !important; align-items: center !important; justify-content: center !important; color: var(--white) !important; box-shadow: 0 2px 5px rgba(0,0,0,0.2) !important; }
103
+ .speak-btn { background-color: var(--secondary-color) !important; border-radius: 24px !important; color: var(--white) !important; padding: 8px 16px !important; font-weight: 500 !important; margin-left: 10px !important; }
104
+ .audio-controls { display: flex; align-items: center; margin-top: 10px; }
105
+
106
+ /* Audio Visualization Elements */
107
+ .audio-visualization {
108
+ display: flex;
109
+ align-items: center;
110
+ justify-content: center;
111
+ gap: 4px;
112
+ height: 40px;
113
+ padding: 10px;
114
+ background-color: rgba(0,0,0,0.05);
115
+ border-radius: 12px;
116
+ margin: 10px 0;
117
+ }
118
+
119
+ .audio-bar {
120
+ width: 3px;
121
+ background-color: var(--accent-color);
122
+ border-radius: 2px;
123
+ height: 5px;
124
+ transition: height 0.1s ease;
125
+ }
126
+
127
+ .audio-status {
128
+ font-size: 0.85rem;
129
+ color: var(--secondary-color);
130
+ text-align: center;
131
+ margin-top: 5px;
132
+ font-style: italic;
133
+ }
134
+
135
+ .recording-indicator {
136
+ width: 12px;
137
+ height: 12px;
138
+ border-radius: 50%;
139
+ background-color: #ff4b4b;
140
+ margin-right: 8px;
141
+ animation: blink 1s infinite;
142
+ }
143
+
144
+ .playing-indicator {
145
+ width: 12px;
146
+ height: 12px;
147
+ border-radius: 50%;
148
+ background-color: #4bff4b;
149
+ margin-right: 8px;
150
+ animation: pulse 1s infinite;
151
+ }
152
+
153
+ @keyframes blink {
154
+ 0% { opacity: 1; }
155
+ 50% { opacity: 0.4; }
156
+ 100% { opacity: 1; }
157
+ }
158
+
159
+ @keyframes pulse {
160
+ 0% { transform: scale(1); }
161
+ 50% { transform: scale(1.2); }
162
+ 100% { transform: scale(1); }
163
+ }
164
+
165
+ .file-upload-enhancement .file-preview {
166
+ max-height: 200px;
167
+ overflow: auto;
168
+ border: 1px solid var(--border-color);
169
+ border-radius: 8px;
170
+ padding: 10px;
171
+ margin-top: 10px;
172
+ background-color: rgba(0,0,0,0.02);
173
+ }
174
+
175
+ .excel-preview-table {
176
+ width: 100%;
177
+ border-collapse: collapse;
178
+ font-size: 0.85rem;
179
+ }
180
+
181
+ .excel-preview-table th, .excel-preview-table td {
182
+ border: 1px solid #ddd;
183
+ padding: 4px 8px;
184
+ text-align: left;
185
+ }
186
+
187
+ .excel-preview-table th {
188
+ background-color: var(--secondary-color);
189
+ color: white;
190
+ }
191
+
192
+ .excel-preview-table tr:nth-child(even) {
193
+ background-color: rgba(0,0,0,0.03);
194
+ }
195
  """
196
 
197
  # Function to process PDF files
 
231
  os.unlink(pdf_path)
232
  return None, f"Error processing PDF: {str(e)}", {"page_images": [], "total_pages": 0, "total_words": 0}
233
 
234
+ # New function to process Excel files
235
+ def process_excel(excel_file):
236
+ if excel_file is None:
237
+ return None, "No file uploaded", {"data_preview": "", "total_sheets": 0, "total_rows": 0}
238
+
239
+ try:
240
+ session_id = str(uuid.uuid4())
241
+ with tempfile.NamedTemporaryFile(suffix=".xlsx", delete=False) as temp_file:
242
+ temp_file.write(excel_file)
243
+ excel_path = temp_file.name
244
+
245
+ # Read Excel file with pandas
246
+ excel_data = pd.ExcelFile(excel_path)
247
+ sheet_names = excel_data.sheet_names
248
+ all_texts = []
249
+ total_rows = 0
250
+
251
+ # Process each sheet
252
+ for sheet in sheet_names:
253
+ df = pd.read_excel(excel_path, sheet_name=sheet)
254
+ total_rows += len(df)
255
+
256
+ # Convert dataframe to text for vectorization
257
+ sheet_text = f"Sheet: {sheet}\n"
258
+ sheet_text += df.to_string(index=False)
259
+ all_texts.append(sheet_text)
260
+
261
+ # Generate HTML preview of first sheet
262
+ first_df = pd.read_excel(excel_path, sheet_name=0)
263
+ preview_rows = min(10, len(first_df))
264
+ data_preview = first_df.head(preview_rows).to_html(classes="excel-preview-table", index=False)
265
+
266
+ # Process for vectorstore
267
+ text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
268
+ chunks = text_splitter.create_documents(all_texts)
269
+ vectorstore = FAISS.from_documents(chunks, embeddings)
270
+ index_path = os.path.join(FAISS_INDEX_DIR, session_id)
271
+ vectorstore.save_local(index_path)
272
+ user_vectorstores[session_id] = vectorstore
273
+
274
+ os.unlink(excel_path)
275
+ excel_state = {"data_preview": data_preview, "total_sheets": len(sheet_names), "total_rows": total_rows}
276
+ return session_id, f"✅ Successfully processed {len(chunks)} text chunks from Excel file", excel_state
277
+ except Exception as e:
278
+ if "excel_path" in locals() and os.path.exists(excel_path):
279
+ os.unlink(excel_path)
280
+ return None, f"Error processing Excel file: {str(e)}", {"data_preview": "", "total_sheets": 0, "total_rows": 0}
281
+
282
+ # Function to analyze image using SmolDocling
283
+ def analyze_image(image_file):
284
+ if image_file is None:
285
+ return "No image uploaded. Please upload an image to analyze."
286
+
287
+ if docling_processor is None or docling_model is None:
288
+ return "SmolDocling model not loaded. Please check your installation."
289
+
290
+ try:
291
+ # Process the image - image_file is a filepath string from Gradio
292
+ image = Image.open(image_file)
293
+
294
+ # Use the SmolDocling model
295
+ inputs = docling_processor(images=image, return_tensors="pt")
296
+ with torch.no_grad():
297
+ outputs = docling_model.generate(
298
+ **inputs,
299
+ max_new_tokens=512,
300
+ temperature=0.1,
301
+ do_sample=False
302
+ )
303
+
304
+ # Decode the output
305
+ result = docling_processor.batch_decode(outputs, skip_special_tokens=True)[0]
306
+
307
+ # Format the result for display with technical emphasis
308
+ analysis = f"## Technical Document Analysis Results\n\n{result}\n\n"
309
+ analysis += "### Technical Insights\n\n"
310
+ analysis += "* The analysis provides technical information extracted from the document image.\n"
311
+ analysis += "* Consider this information as a starting point for further technical investigation.\n"
312
+ analysis += "* For code snippets or technical specifications, verify accuracy before implementation.\n"
313
+
314
+ return analysis
315
+ except Exception as e:
316
+ return f"Error analyzing image: {str(e)}"
317
+
318
+ # Improved function for speech-to-text conversion with status updates
319
+ def speech_to_text(audio_status):
320
+ try:
321
+ # Update status to show we're listening
322
+ audio_status = "Listening... Speak now"
323
+ yield audio_status, gr.update(visible=True), None
324
+
325
+ r = sr.Recognizer()
326
+ with sr.Microphone() as source:
327
+ r.adjust_for_ambient_noise(source)
328
+ audio = r.listen(source, timeout=5, phrase_time_limit=15)
329
+
330
+ # Update status to show processing
331
+ audio_status = "Processing speech..."
332
+ yield audio_status, gr.update(visible=True), None
333
+
334
+ text = r.recognize_google(audio)
335
+ audio_status = "Speech recognized!"
336
+ return audio_status, gr.update(visible=False), text
337
+ except sr.UnknownValueError:
338
+ audio_status = "Could not understand audio. Please try again."
339
+ return audio_status, gr.update(visible=False), None
340
+ except sr.RequestError as e:
341
+ audio_status = f"Error with speech recognition service: {e}"
342
+ return audio_status, gr.update(visible=False), None
343
+ except Exception as e:
344
+ audio_status = f"Error: {str(e)}"
345
+ return audio_status, gr.update(visible=False), None
346
+
347
+ # Improved function for text-to-speech conversion with pyttsx3
348
+ def text_to_speech(audio_status, history):
349
+ if not history:
350
+ return "No text to speak", gr.update(visible=False), None
351
+
352
+ try:
353
+ # Get the last bot response
354
+ last_response = history[-1][1]
355
+
356
+ # Clean up the text (remove markdown and other formatting)
357
+ clean_text = re.sub(r'\*\*|__', '', last_response) # Remove bold/underline
358
+ clean_text = re.sub(r'\[([^\]]+)\]\([^)]+\)', r'\1', clean_text) # Remove links
359
+ clean_text = re.sub(r'#+ ', '', clean_text) # Remove headers
360
+ clean_text = re.sub(r'```[^`]*```', ' Code block removed for speech. ', clean_text) # Remove code blocks
361
+
362
+ # Update status
363
+ audio_status = "Generating speech..."
364
+ yield audio_status, gr.update(visible=True), None
365
+
366
+ # Save to a temporary file
367
+ temp_file = tempfile.NamedTemporaryFile(delete=False, suffix=".mp3")
368
+
369
+ # Use pyttsx3 to generate speech
370
+ tts_engine.save_to_file(clean_text, temp_file.name)
371
+ tts_engine.runAndWait()
372
+
373
+ audio_status = "Speech ready!"
374
+ return audio_status, gr.update(visible=False), temp_file.name
375
+ except Exception as e:
376
+ audio_status = f"Error in text-to-speech: {str(e)}"
377
+ return audio_status, gr.update(visible=False), None
378
+
379
+ # Function to handle different file types
380
+ def process_file(file_data, file_type):
381
+ if file_data is None:
382
+ return None, "No file uploaded", None
383
+
384
+ if file_type == "pdf":
385
+ return process_pdf(file_data)
386
+ elif file_type == "excel":
387
+ return process_excel(file_data)
388
+ elif file_type == "image":
389
+ # For image files, we'll just use them directly for analysis
390
+ # But we'll return a session ID to maintain consistency
391
+ session_id = str(uuid.uuid4())
392
+ return session_id, "✅ Image file ready for analysis", None
393
+ else:
394
+ return None, "Unsupported file type", None
395
+
396
+ # Function for speech-to-text conversion
397
+ def speech_to_text():
398
+ try:
399
+ r = sr.Recognizer()
400
+ with sr.Microphone() as source:
401
+ r.adjust_for_ambient_noise(source)
402
+ audio = r.listen(source)
403
+ text = r.recognize_google(audio)
404
+ return text
405
+ except sr.UnknownValueError:
406
+ return "Could not understand audio. Please try again."
407
+ except sr.RequestError as e:
408
+ return f"Error with speech recognition service: {e}"
409
+ except Exception as e:
410
+ return f"Error converting speech to text: {str(e)}"
411
+
412
+ # Function for text-to-speech conversion
413
+ def text_to_speech(text, history):
414
+ if not text or not history:
415
+ return None
416
+
417
+ try:
418
+ # Get the last bot response
419
+ last_response = history[-1][1]
420
+
421
+ # Convert text to speech
422
+ tts = pyttsx3.init()
423
+ tts.setProperty('rate', 150)
424
+ tts.setProperty('volume', 0.9)
425
+ tts.save_to_file(last_response, "temp_output.mp3")
426
+ tts.runAndWait()
427
+
428
+ return "temp_output.mp3"
429
+ except Exception as e:
430
+ print(f"Error in text-to-speech: {e}")
431
+ return None
432
+
433
  # Function to generate chatbot responses with Tech theme
434
+ def generate_response(message, session_id, model_name, history, web_search_enabled=True):
435
  if not message:
436
  return history
437
  try:
 
442
  if docs:
443
  context = "\n\nRelevant information from uploaded PDF:\n" + "\n".join(f"- {doc.page_content}" for doc in docs)
444
 
445
+ # Check if it's a GitHub repo search and web search is enabled
446
+ if web_search_enabled and re.match(r'^/github\s+.+', message, re.IGNORECASE):
447
  query = re.sub(r'^/github\s+', '', message, flags=re.IGNORECASE)
448
  repo_results = search_github_repos(query)
449
  if repo_results:
 
460
  history.append((message, "No GitHub repositories found for your query."))
461
  return history
462
 
463
+ # Check if it's a Stack Overflow search and web search is enabled
464
+ if web_search_enabled and re.match(r'^/stack\s+.+', message, re.IGNORECASE):
465
  query = re.sub(r'^/stack\s+', '', message, flags=re.IGNORECASE)
466
  qa_results = search_stackoverflow(query)
467
  if qa_results:
 
758
  with gr.Blocks(css=custom_css, theme=gr.themes.Soft()) as demo:
759
  current_session_id = gr.State(None)
760
  pdf_state = gr.State({"page_images": [], "total_pages": 0, "total_words": 0})
761
+ excel_state = gr.State({"data_preview": "", "total_sheets": 0, "total_rows": 0})
762
+ file_type = gr.State("none")
763
+ audio_status = gr.State("Ready")
764
+
765
  gr.HTML("""
766
  <div class="header">
767
+ <div class="header-title">Tech-Vision Enhanced</div>
768
+ <div class="header-subtitle">Analyze technical documents, spreadsheets, and images with AI</div>
769
  </div>
770
  """)
771
  with gr.Row(elem_classes="container"):
772
  with gr.Column(scale=1, min_width=300):
773
+ with gr.Tabs():
774
+ with gr.TabItem("PDF"):
775
+ pdf_file = gr.File(label="Upload PDF Document", file_types=[".pdf"], type="binary")
776
+ pdf_upload_button = gr.Button("Process PDF", variant="primary")
777
+
778
+ with gr.TabItem("Excel"):
779
+ excel_file = gr.File(label="Upload Excel File", file_types=[".xlsx", ".xls"], type="binary")
780
+ excel_upload_button = gr.Button("Process Excel", variant="primary")
781
+
782
+ with gr.TabItem("Image"):
783
+ image_input = gr.File(
784
+ label="Upload Image",
785
+ file_types=["image"],
786
+ type="filepath"
787
+ )
788
+ analyze_btn = gr.Button("Analyze Image")
789
+
790
+ file_status = gr.Markdown("No file uploaded yet")
791
+
792
+ # Model selector
793
  model_dropdown = gr.Dropdown(
794
  choices=["llama3-70b-8192", "llama3-8b-8192", "mixtral-8x7b-32768", "gemma-7b-it"],
795
  value="llama3-70b-8192",
796
  label="Select Groq Model"
797
  )
798
+
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
799
  with gr.Column(scale=2, min_width=600):
800
  with gr.Tabs():
801
  with gr.TabItem("PDF Viewer"):
802
  with gr.Column(elem_classes="pdf-viewer-container"):
803
  page_slider = gr.Slider(minimum=1, maximum=1, step=1, label="Page Number", value=1)
804
  pdf_image = gr.Image(label="PDF Page", type="pil", elem_classes="pdf-viewer-image")
805
+ pdf_stats = gr.Markdown("No PDF uploaded yet", elem_classes="stats-box")
806
 
807
+ with gr.TabItem("Excel Viewer"):
808
+ excel_preview = gr.HTML(label="Excel Preview", elem_classes="file-preview")
809
+ excel_stats = gr.Markdown("No Excel file uploaded yet", elem_classes="stats-box")
810
 
811
+ with gr.TabItem("Image Analysis"):
812
+ image_preview = gr.Image(label="Image Preview", type="pil")
813
+ image_analysis_results = gr.Markdown("Upload an image and click 'Analyze Image' to see analysis results")
814
+
815
+ # Audio visualization elements
816
+ with gr.Row(elem_classes="container"):
817
+ with gr.Column():
818
+ audio_vis = gr.HTML("""
819
+ <div class="audio-visualization">
820
+ <div class="audio-bar" style="height: 5px;"></div>
821
+ <div class="audio-bar" style="height: 12px;"></div>
822
+ <div class="audio-bar" style="height: 18px;"></div>
823
+ <div class="audio-bar" style="height: 15px;"></div>
824
+ <div class="audio-bar" style="height: 10px;"></div>
825
+ <div class="audio-bar" style="height: 20px;"></div>
826
+ <div class="audio-bar" style="height: 14px;"></div>
827
+ <div class="audio-bar" style="height: 8px;"></div>
828
+ </div>
829
+ """, visible=False)
830
+ audio_status_display = gr.Markdown("", elem_classes="audio-status")
831
 
832
+ # Chat interface
833
  with gr.Row(elem_classes="container"):
834
  with gr.Column(scale=2, min_width=600):
835
+ chatbot = gr.Chatbot(height=400, bubble_full_width=False, show_copy_button=True, elem_classes="chat-container")
836
  with gr.Row():
837
+ msg = gr.Textbox(
838
+ show_label=False,
839
+ placeholder="Ask about your document or click the microphone to speak...",
840
+ scale=5
841
+ )
842
+ voice_btn = gr.Button("🎤", elem_classes="voice-btn")
843
  send_btn = gr.Button("Send", scale=1)
844
+
845
+ with gr.Row(elem_classes="audio-controls"):
846
+ clear_btn = gr.Button("Clear Conversation")
847
+ speak_btn = gr.Button("🔊 Speak Response", elem_classes="speak-btn")
848
+ audio_player = gr.Audio(label="Response Audio", type="filepath", visible=False)
849
 
850
+ # Event Handlers for PDF processing
851
+ pdf_upload_button.click(
852
+ lambda x: ("pdf", x),
853
+ inputs=[pdf_file],
854
+ outputs=[file_type, file_status]
855
+ ).then(
856
  process_pdf,
857
  inputs=[pdf_file],
858
+ outputs=[current_session_id, file_status, pdf_state]
859
  ).then(
860
  update_pdf_viewer,
861
  inputs=[pdf_state],
862
+ outputs=[page_slider, pdf_image, pdf_stats]
863
+ )
864
+
865
+ # Event Handlers for Excel processing
866
+ excel_upload_button.click(
867
+ lambda x: ("excel", x),
868
+ inputs=[excel_file],
869
+ outputs=[file_type, file_status]
870
+ ).then(
871
+ process_excel,
872
+ inputs=[excel_file],
873
+ outputs=[current_session_id, file_status, excel_state]
874
+ ).then(
875
+ lambda preview, sheets, rows: (
876
+ preview,
877
+ f"**Excel Statistics:**\nSheets: {sheets}\nTotal Rows: {rows}"
878
+ ),
879
+ inputs=[excel_state["data_preview"], excel_state["total_sheets"], excel_state["total_rows"]],
880
+ outputs=[excel_preview, excel_stats]
881
+ )
882
+
883
+ # Event Handlers for Image Analysis
884
+ analyze_btn.click(
885
+ lambda x: ("image", x),
886
+ inputs=[image_input],
887
+ outputs=[file_type, file_status]
888
+ ).then(
889
+ analyze_image,
890
+ inputs=[image_input],
891
+ outputs=[image_analysis_results]
892
+ ).then(
893
+ lambda x: Image.open(x) if x else None,
894
+ inputs=[image_input],
895
+ outputs=[image_preview]
896
  )
897
 
898
+ # Chat message handling
899
  msg.submit(
900
  generate_response,
901
  inputs=[msg, current_session_id, model_dropdown, chatbot],
 
908
  outputs=[chatbot]
909
  ).then(lambda: "", None, [msg])
910
 
911
+ # Improved speech-to-text with visual feedback
912
+ voice_btn.click(
913
+ speech_to_text,
914
+ inputs=[audio_status],
915
+ outputs=[audio_status_display, audio_vis, msg]
916
+ )
917
+
918
+ # Improved text-to-speech with visual feedback
919
+ speak_btn.click(
920
+ text_to_speech,
921
+ inputs=[audio_status, chatbot],
922
+ outputs=[audio_status_display, audio_vis, audio_player]
923
+ ).then(
924
+ lambda x: gr.update(visible=True) if x else gr.update(visible=False),
925
+ inputs=[audio_player],
926
+ outputs=[audio_player]
927
  )
928
 
929
+ # Page navigation for PDF
930
  page_slider.change(
931
  update_image,
932
  inputs=[page_slider, pdf_state],
933
  outputs=[pdf_image]
934
  )
935
 
936
+ # Clear conversation and reset UI
937
+ clear_btn.click(
938
+ lambda: (
939
+ [], None, "No file uploaded yet",
940
+ {"page_images": [], "total_pages": 0, "total_words": 0},
941
+ {"data_preview": "", "total_sheets": 0, "total_rows": 0},
942
+ "none", 0, None, "No PDF uploaded yet", "",
943
+ "No Excel file uploaded yet", None,
944
+ "Upload an image and click 'Analyze Image' to see results", None,
945
+ gr.update(visible=False), "Ready"
946
+ ),
947
+ None,
948
+ [chatbot, current_session_id, file_status, pdf_state, excel_state,
949
+ file_type, page_slider, pdf_image, pdf_stats, excel_preview,
950
+ excel_stats, image_preview, image_analysis_results, audio_player,
951
+ audio_vis, audio_status_display]
 
952
  )
953
 
954
+ # Add footer with creator attribution
955
+ gr.HTML("""
956
+ <div style="text-align: center; margin-top: 20px; padding: 10px; color: #666; font-size: 0.8rem; border-top: 1px solid #eee;">
957
+ Created by Calvin Allen Crawford
958
+ </div>
959
+ """)
960
 
961
  # Launch the app
962
  if __name__ == "__main__":