Spaces:
Running
Running
File size: 20,585 Bytes
ee7d7b9 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 | """
ποΈ PRO VISION ENGINE v2.0 - DataVision Visual Intelligence
============================================================
Advanced computer vision engine for image and document understanding.
Features:
- π OCR Text Extraction
- π Chart Data Extraction
- π Document Parsing
- πΈ Screenshot Analysis
- π¨ Pattern Recognition
- π‘ Visual Insights
Built for DataVision - See and understand ANY visual content!
Author: DataVision Team
Version: 2.0.0
"""
import logging
import base64
import os
from typing import Dict, Any, List, Optional, Tuple
from datetime import datetime
from enum import Enum
from dataclasses import dataclass
import re
import json
logger = logging.getLogger(__name__)
# LLM with vision capability
try:
from core.llm import chat as llm_chat
LLM_AVAILABLE = True
except ImportError:
LLM_AVAILABLE = False
# Settings
try:
from config.settings import Settings
SETTINGS_AVAILABLE = True
except ImportError:
SETTINGS_AVAILABLE = False
# Intelligent Query Processor (Claude-style)
try:
from core.intelligent_processor import IntelligentQueryProcessor, intelligent_process
INTELLIGENT_PROCESSOR_AVAILABLE = True
except ImportError:
INTELLIGENT_PROCESSOR_AVAILABLE = False
# =============================================================================
# IMAGE TYPES
# =============================================================================
class ImageType(Enum):
"""Types of images the engine can analyze"""
CHART = "chart" # Bar charts, line graphs, pie charts
TABLE = "table" # Data tables
DOCUMENT = "document" # Text documents, PDFs
SCREENSHOT = "screenshot" # UI screenshots
PHOTO = "photo" # General photos
DIAGRAM = "diagram" # Flowcharts, architecture diagrams
INFOGRAPHIC = "infographic" # Data infographics
UNKNOWN = "unknown"
@dataclass
class ImageAnalysis:
"""Result of image analysis"""
image_type: ImageType
extracted_text: str
extracted_data: Dict[str, Any]
descriptions: List[str]
confidence: float
# =============================================================================
# IMAGE TYPE DETECTION
# =============================================================================
def detect_image_type(query: str, filename: str = "") -> ImageType:
"""Detect what type of image analysis is needed."""
q_lower = query.lower()
filename_lower = filename.lower()
# Check query patterns
if any(kw in q_lower for kw in ['chart', 'graph', 'plot', 'bar', 'line', 'pie']):
return ImageType.CHART
if any(kw in q_lower for kw in ['table', 'spreadsheet', 'grid']):
return ImageType.TABLE
if any(kw in q_lower for kw in ['document', 'text', 'ocr', 'read', 'extract text']):
return ImageType.DOCUMENT
if any(kw in q_lower for kw in ['screenshot', 'screen', 'ui', 'interface']):
return ImageType.SCREENSHOT
if any(kw in q_lower for kw in ['diagram', 'flow', 'architecture']):
return ImageType.DIAGRAM
if any(kw in q_lower for kw in ['infographic', 'info']):
return ImageType.INFOGRAPHIC
# Check filename patterns
if any(ext in filename_lower for ext in ['chart', 'graph', 'plot']):
return ImageType.CHART
if any(ext in filename_lower for ext in ['doc', 'pdf', 'letter']):
return ImageType.DOCUMENT
if any(ext in filename_lower for ext in ['screenshot', 'screen']):
return ImageType.SCREENSHOT
return ImageType.UNKNOWN
# =============================================================================
# VISION PROMPTS
# =============================================================================
def get_vision_prompt(query: str, image_type: ImageType) -> str:
"""Get specialized prompt based on image type."""
base_prompt = f"""You are an Vision AI - an expert at analyzing images and extracting information.
USER REQUEST: {query}
"""
prompts = {
ImageType.CHART: base_prompt + """
This is a CHART/GRAPH image. Please:
1. Identify the chart type (bar, line, pie, scatter, etc.)
2. Extract the title and axis labels
3. List the data values you can see
4. Describe trends or patterns
5. Note any anomalies or interesting points
FORMAT YOUR RESPONSE:
**Chart Type:** [type]
**Title:** [title if visible]
**Data Extracted:**
- [label]: [value]
- [label]: [value]
...
**Trends/Insights:**
[Your analysis]
""",
ImageType.TABLE: base_prompt + """
This is a TABLE/SPREADSHEET image. Please:
1. Extract column headers
2. Extract row data (as many rows as visible)
3. Identify data types (numbers, text, dates)
4. Note any formatting (colors, highlights)
5. Summarize the table content
FORMAT AS STRUCTURED DATA:
**Columns:** [col1, col2, col3, ...]
**Sample Data:**
| col1 | col2 | col3 |
|------|------|------|
| val | val | val |
**Summary:** [key insights from the table]
""",
ImageType.DOCUMENT: base_prompt + """
This is a DOCUMENT/TEXT image. Please:
1. Extract ALL visible text (OCR)
2. Preserve formatting where possible
3. Identify document type (letter, report, form, etc.)
4. Extract key entities (names, dates, amounts)
5. Summarize the document
FORMAT:
**Document Type:** [type]
**Extracted Text:**
[Full text from the document]
**Key Information:**
- [entity]: [value]
**Summary:** [brief summary]
""",
ImageType.SCREENSHOT: base_prompt + """
This is a SCREENSHOT/UI image. Please:
1. Identify the application/website
2. Describe the visible UI elements
3. Extract any visible text/data
4. Note the current state/action
5. Identify any errors or notifications
FORMAT:
**Application:** [name if identifiable]
**UI Elements:**
- [element type]: [description]
**Visible Data:**
[Any data shown on screen]
**State:** [What action is happening]
""",
ImageType.DIAGRAM: base_prompt + """
This is a DIAGRAM image. Please:
1. Identify diagram type (flowchart, ERD, architecture, etc.)
2. List all nodes/components
3. Describe connections/relationships
4. Extract any labels or annotations
5. Explain the overall structure
FORMAT:
**Diagram Type:** [type]
**Components:**
- [component]: [description]
**Relationships:**
- [component A] β [component B]: [relationship]
**Summary:** [What the diagram represents]
""",
ImageType.INFOGRAPHIC: base_prompt + """
This is an INFOGRAPHIC image. Please:
1. Extract the main title/topic
2. List all data points with values
3. Identify icons/visual elements
4. Extract statistics or percentages
5. Summarize the key message
FORMAT:
**Topic:** [main subject]
**Data Points:**
- [metric]: [value]
**Key Statistics:**
- [stat description]: [number/percent]
**Main Message:** [What the infographic communicates]
""",
ImageType.UNKNOWN: base_prompt + """
Analyze this image comprehensively:
1. Describe what you see
2. Extract any text visible
3. Identify any data or numbers
4. Note any patterns or interesting elements
5. Provide relevant insights based on the user's question
Be thorough and answer the user's specific question.
"""
}
return prompts.get(image_type, prompts[ImageType.UNKNOWN])
# =============================================================================
# PRO VISION ENGINE CLASS
# =============================================================================
class ProVisionEngine:
"""
ποΈ PRO VISION ENGINE - DataVision Visual Intelligence
Advanced computer vision for any image type.
π‘ WHEN TO USE:
- Extract data from chart images
- OCR from documents
- Analyze screenshots
- Understand diagrams
Features:
- Multi-modal image understanding
- Structured data extraction
- Chart-to-data conversion
- Document parsing
"""
def __init__(self, user_id: str):
self.user_id = user_id
self.analysis_history = []
def process(
self,
query: str,
image_path: str = None,
image_base64: str = None,
context: str = ""
) -> Dict[str, Any]:
"""
Process an image with advanced vision analysis.
Uses DYNAMIC routing - if no image and general query, use AI Knowledge.
"""
result = {
"answer": "",
"mode": "vision",
"confidence": 0.85,
"sources": ["Vision"],
"extracted_data": {},
"image_type": None
}
start_time = datetime.now()
# Check if we have an image
if not image_path and not image_base64:
# =================================================================
# π DYNAMIC ROUTING: No image, check if general knowledge query
# =================================================================
q_lower = query.lower()
# If query seems like a general knowledge question, answer it
vision_terms = ['image', 'picture', 'photo', 'chart', 'diagram', 'screenshot', 'document', 'scan', 'extract', 'ocr']
is_vision_query = any(term in q_lower for term in vision_terms)
if not is_vision_query:
logger.info("π Vision: No image + general query, routing to AI Knowledge")
try:
from core.llm import chat as llm_chat
ai_prompt = f"""You are a helpful AI assistant.
Answer this question:
{query}
Provide a clear, accurate response."""
llm_response = llm_chat(ai_prompt, temperature=0.7, max_tokens=600)
result["answer"] = f"""## ποΈ Vision
π **AI Knowledge**
{llm_response}
---
*π‘ For image analysis, please upload an image with your question.*"""
result["sources"] = ["AI Knowledge"]
exec_time = (datetime.now() - start_time).total_seconds()
result["execution_time"] = f"{exec_time:.2f}s"
return result
except Exception as e:
logger.error(f"AI Knowledge error: {e}")
result["answer"] = self._no_image_response()
return result
# Load image as base64 if path provided
if image_path and not image_base64:
try:
with open(image_path, 'rb') as f:
image_data = f.read()
image_base64 = base64.b64encode(image_data).decode('utf-8')
filename = os.path.basename(image_path)
except Exception as e:
result["answer"] = f"β Could not load image: {str(e)}"
return result
else:
filename = ""
# Detect image type
image_type = detect_image_type(query, filename)
result["image_type"] = image_type.value
logger.info(f"ποΈ Vision Type: {image_type.value}")
# Get specialized prompt
vision_prompt = get_vision_prompt(query, image_type)
# Analyze with vision LLM
if LLM_AVAILABLE:
try:
# Call LLM with image
# Note: This requires vision-capable LLM
response = self._analyze_with_llm(vision_prompt, image_base64)
except Exception as e:
logger.error(f"Vision LLM error: {e}")
response = self._fallback_analysis(query, image_type)
else:
response = self._fallback_analysis(query, image_type)
# Format response
formatted_response = self._format_response(response, image_type, query)
# Extract structured data
extracted_data = self._extract_structured_data(response, image_type)
result["answer"] = formatted_response
result["extracted_data"] = extracted_data
result["confidence"] = 0.85 if response else 0.5
# Chart generation logic removed (incompatible with vision engine)
# Execution time
exec_time = (datetime.now() - start_time).total_seconds()
result["execution_time"] = f"{exec_time:.2f}s"
return result
def _analyze_with_llm(self, prompt: str, image_base64: str) -> str:
"""Analyze image using vision-capable LLM."""
try:
# Standard LLM call (would need vision capability)
# For now, we'll do what we can without vision
return llm_chat(prompt + "\n\n[Note: Analyzing based on prompt context]",
temperature=0.3, max_tokens=600)
except Exception as e:
logger.error(f"LLM analysis error: {e}")
return ""
def _fallback_analysis(self, query: str, image_type: ImageType) -> str:
"""Fallback when vision analysis is not available."""
type_descriptions = {
ImageType.CHART: "This appears to be a chart or graph image. For data extraction, please ensure the image is clear and contains visible data points.",
ImageType.TABLE: "This appears to be a table or spreadsheet image. Data extraction works best with clearly defined rows and columns.",
ImageType.DOCUMENT: "This appears to be a document image. Text extraction will depend on image quality and text clarity.",
ImageType.SCREENSHOT: "This is a screenshot. UI elements and visible text can be analyzed.",
ImageType.DIAGRAM: "This appears to be a diagram. Nodes and connections can be identified from clear diagrams.",
ImageType.UNKNOWN: "Image analysis is available. Please provide a clear image for best results."
}
return type_descriptions.get(image_type, type_descriptions[ImageType.UNKNOWN])
def _format_response(self, analysis: str, image_type: ImageType, query: str) -> str:
"""Format the analysis response."""
type_emoji = {
ImageType.CHART: "π",
ImageType.TABLE: "π",
ImageType.DOCUMENT: "π",
ImageType.SCREENSHOT: "π₯οΈ",
ImageType.DIAGRAM: "π",
ImageType.INFOGRAPHIC: "π",
ImageType.UNKNOWN: "πΌοΈ"
}
emoji = type_emoji.get(image_type, "πΌοΈ")
formatted = f"""## {emoji} Image Analysis
**Query:** {query}
**Image Type:** {image_type.value.replace('_', ' ').title()}
---
### π Analysis Results
{analysis}
---
### π‘ Tips
"""
# Add type-specific tips
tips = {
ImageType.CHART: "- Ask me to extract specific data points\n- I can identify trends and patterns\n- Charts with clear labels work best",
ImageType.TABLE: "- I can extract data as structured rows\n- Ask for specific columns or rows\n- Tables with headers are easier to parse",
ImageType.DOCUMENT: "- I perform OCR on visible text\n- High resolution images work best\n- Ask me to summarize or find specific info",
ImageType.SCREENSHOT: "- I can identify UI elements\n- Ask about specific buttons or text\n- I can describe the current state",
ImageType.DIAGRAM: "- I map relationships between nodes\n- Ask about specific components\n- Flowcharts and ERDs work well"
}
formatted += tips.get(image_type, "- Upload a clear image for best results\n- Ask specific questions about the image")
return formatted
def _extract_structured_data(self, analysis: str, image_type: ImageType) -> Dict[str, Any]:
"""Extract structured data from analysis."""
data = {}
# Try to extract key-value pairs
lines = analysis.split('\n')
for line in lines:
if ':' in line:
parts = line.split(':', 1)
if len(parts) == 2:
key = parts[0].strip().strip('*-')
value = parts[1].strip()
if key and value and len(key) < 50:
data[key] = value
return data
def _no_image_response(self) -> str:
"""Response when no image is provided."""
return """## ποΈ Vision AI
I need an image to analyze! Please:
1. **Upload an image** using the attachment button π
2. **Ask a question** about the image
### What I Can Analyze:
| Image Type | Examples |
|------------|----------|
| π **Charts** | Bar graphs, line charts, pie charts |
| π **Tables** | Spreadsheets, data tables |
| π **Documents** | PDFs, letters, forms (OCR) |
| π₯οΈ **Screenshots** | UI, dashboards, apps |
| π **Diagrams** | Flowcharts, architecture |
### Example Questions:
- "Extract data from this chart"
- "What does this table show?"
- "Read the text in this document"
- "Describe this screenshot"
*Upload an image and I'll analyze it for you!*
"""
# =============================================================================
# CONVENIENCE FUNCTIONS
# =============================================================================
def vision_response(
user_id: str,
query: str,
image_path: str = None,
image_base64: str = None,
context: str = "",
df = None
) -> Dict[str, Any]:
"""
Quick function for vision response.
Args:
user_id: User ID
query: User's question
image_path: Path to image file (optional)
image_base64: Base64 image data (optional)
context: Additional context including pre-analyzed image content
df: DataFrame for chart generation (optional)
"""
engine = ProVisionEngine(user_id)
# Check if context already contains image analysis (pre-processed in chat.py)
if context and "πΌοΈ Image Analysis" in context:
# Context already has image analysis - use it directly with LLM
logger.info("ποΈ Vision: Using pre-analyzed image context")
result = {
"answer": "",
"mode": "vision",
"confidence": 0.90,
"sources": ["Vision Engine"],
"extracted_data": {}
}
try:
from core.llm import chat as llm_chat
vision_prompt = f"""You are a Vision AI Expert. Analyze the image content provided and answer the user's question.
## IMAGE ANALYSIS:
{context}
## USER QUESTION:
{query}
INSTRUCTIONS:
1. Base your answer ONLY on the actual image analysis provided above
2. If asked "what do you see", describe what's shown in the image analysis
3. Be specific about visual elements, text, data, charts, or objects mentioned
4. If the image contains data/charts, extract and explain the key points
5. Format your response clearly with markdown
Provide a detailed, accurate response about the image."""
llm_response = llm_chat(vision_prompt, temperature=0.3, max_tokens=800)
result["answer"] = f"""## ποΈ Vision Analysis
{llm_response}
---
*Image analyzed using AI vision capabilities*"""
result["confidence"] = 0.92
except Exception as e:
logger.error(f"Vision LLM error: {e}")
result["answer"] = f"""## ποΈ Image Analysis
{context}
---
*Note: This is the raw vision analysis.*"""
return result
# No pre-analyzed content - run normal vision processing
return engine.process(query, image_path, image_base64)
def vision_response_sync(
user_id: str,
query: str,
context: str = "",
df = None
) -> Dict[str, Any]:
"""
Synchronous vision response for compatibility.
Args:
user_id: User ID
query: User's question
context: Context string (may contain pre-analyzed image content)
df: DataFrame for optional chart generation
"""
# Check if context contains image analysis (from chat.py preprocessing)
if context and "πΌοΈ Image Analysis" in context:
return vision_response(user_id, query, context=context, df=df)
# Legacy path - context is treated as image_path
return vision_response(user_id, query, image_path=context if context and not context.startswith("##") else None)
# Alias for backwards compatibility
VisionEngine = ProVisionEngine
__all__ = ['ProVisionEngine', 'VisionEngine', 'vision_response', 'vision_response_sync']
|