knowledge recognition from attached picture filess

core_logic.py

@@ -79,12 +79,23 @@ def chat_function(message, history):
     user_text = message.get("text", "")
     files = message.get("files", [])
     
-    # 1. Process Files with character limits
+    # Context Aggregator Buffer for all multi-format assets
     context_from_files = ""
-    for f in files:
-        path = f["path"] if isinstance(f, dict) else f
-        file_content = parse_file(path)
-        context_from_files += file_content
+    
+    # 1. Process Multimodal and Extended Multi-format Files via Perception Agent
+    if files:
+        from perception_agent import read_document_file
+        yield "◌ _Perception Agent initialized: Ingesting uploaded file assets..._"
+        
+        for f in files:
+            # Gradio 6 handles file entries either as dictionaries with a 'path' key or flat strings
+            path = f["path"] if isinstance(f, dict) else f
+            if path and os.path.exists(path):
+                file_content = read_document_file(path)
+                context_from_files += file_content
+                
+        yield "◌ _Perception processing complete. Transmitting compiled structures to the Brain..._"
+
     
     # TRUNCATE FILE CONTEXT: Max ~3000 tokens (approx 12,000 chars)
     if len(context_from_files) > 12000:
@@ -112,7 +123,7 @@ def chat_function(message, history):
             model=model,
             messages=messages,
             stream=True,
-            temperature=0.0,
+            temperature=0.2,
             #max_tokens=1024 # Limit response size to prevent mid-stream cuts
         )
 

perception_agent.py

@@ -0,0 +1,109 @@
+
+# ./perception_agent.py
+
+import os
+import pandas as pd
+from docx import Document
+from pypdf import PdfReader  # Cleanly leverages your requirements.txt package
+from groq import Groq
+from agent_logging import log_agent_action
+
+client = Groq(api_key=os.getenv("GROQ_API_KEY"))
+
+def read_image_file(file_path):
+    """Uses Groq Vision capability to interpret images (.png, .jpg, .bmp)"""
+    import base64
+    try:
+        log_agent_action("PERCEPTION", f"Encoding image for Vision API: {file_path}")
+        with open(file_path, "rb") as image_file:
+            encoded_string = base64.b64encode(image_file.read()).decode('utf-8')
+        
+        ext = os.path.splitext(file_path)[1].lower().replace(".", "")
+        mime_type = f"image/{ext}" if ext != "jpg" else "image/jpeg"
+
+        response = client.chat.completions.create(
+            model="llama-3.2-11b-vision-preview",
+            messages=[
+                {
+                    "role": "user",
+                    "content": [
+                        {"type": "text", "text": "Analyze this technical image. Extract all code, data tables, structural diagrams, or text precisely."},
+                        {
+                            "type": "image_url",
+                            "image_url": {
+                                "url": f"data:{mime_type};base64,{encoded_string}"
+                            }
+                        }
+                    ]
+                }
+            ],
+            temperature=0.0
+        )
+        log_agent_action("PERCEPTION_SUCCESS", f"Vision extraction complete for {file_path}")
+        return f"\n--- Visual Content Extraction from {os.path.basename(file_path)} ---\n{response.choices[0].message.content}\n"
+    except Exception as e:
+        log_agent_action("PERCEPTION_ERROR", f"Vision interpretation failed: {str(e)}")
+        return f"\n[Vision Error processing image {os.path.basename(file_path)}: {str(e)}]\n"
+
+def read_document_file(file_path):
+    """Universal router parsing text, code, spreadsheets, PDFs, and document assets"""
+    ext = os.path.splitext(file_path)[1].lower()
+    filename = os.path.basename(file_path)
+    
+    try:
+        # 1. Plain Text and Markdown Layouts
+        if ext in ['.txt', '.md', '.py', '.json', '.yaml', '.toml', '.css', '.html']:
+            log_agent_action("PERCEPTION", f"Reading plaintext structure: {filename}")
+            with open(file_path, "r", encoding="utf-8", errors="ignore") as f:
+                return f"\n--- Content of File: {filename} ---\n{f.read()}\n"
+        
+        # 2. Excel Data Configurations
+        elif ext in ['.xlsx', '.xls']:
+            log_agent_action("PERCEPTION", f"Parsing Data Spreadsheet: {filename}")
+            excel_data = pd.read_excel(file_path, sheet_name=None)
+            combined_text = f"\n--- Spreadsheet Matrix Extraction: {filename} ---\n"
+            for sheet_name, df in excel_data.items():
+                combined_text += f"\nSheet: {sheet_name}\n"
+                combined_text += df.to_markdown(index=False) + "\n"
+            return combined_text
+        
+        # 3. Microsoft Word Processing
+        elif ext == '.docx':
+            log_agent_action("PERCEPTION", f"Extracting structural Word paragraphs: {filename}")
+            doc = Document(file_path)
+            paragraphs = [p.text for p in doc.paragraphs]
+            return f"\n--- Document Text Extraction: {filename} ---\n" + "\n".join(paragraphs) + "\n"
+        
+        # 4. NEW ADDITION: Portable Document Format (.pdf) Ingestion
+        elif ext == '.pdf':
+            log_agent_action("PERCEPTION", f"Initializing pypdf reader pipeline: {filename}")
+            reader = PdfReader(file_path)
+            pdf_text_buffer = []
+            
+            for index, page in enumerate(reader.pages):
+                extracted_page_text = page.extract_text()
+                if extracted_page_text:
+                    pdf_text_buffer.append(f"--- Page {index + 1} ---\n{extracted_page_text}")
+            
+            if not pdf_text_buffer:
+                log_agent_action("PERCEPTION_WARN", f"PDF contained no raw text layers (possible raw scan): {filename}")
+                return f"\n[System Warning: '{filename}' appears to be an un-OCRed scanned image PDF. Please extract its pages as raw images for CoderG's Vision layer.]\n"
+                
+            log_agent_action("PERCEPTION_SUCCESS", f"Successfully parsed {len(pdf_text_buffer)} pages from {filename}")
+            return f"\n--- PDF Document Content Ingestion: {filename} ---\n" + "\n".join(pdf_text_buffer) + "\n"
+        
+        # 5. Image Vector/Raster Formats
+        elif ext in ['.png', '.jpg', '.jpeg', '.bmp']:
+            return read_image_file(file_path)
+            
+        elif ext == '.doc':
+            log_agent_action("PERCEPTION_WARN", f"Legacy format encountered: {filename}")
+            return f"\n[System Error: Legacy format '{ext}' detected. Please convert '{filename}' to '.docx' for automated ingestion.]\n"
+            
+        else:
+            log_agent_action("PERCEPTION_WARN", f"Unknown asset extension skipped: {filename}")
+            return f"\n[System Warning: Unsupported file format '{ext}' for file '{filename}'. Skipping content ingestion.]\n"
+            
+    except Exception as e:
+        log_agent_action("PERCEPTION_ERROR", f"Failed parsing {filename}: {str(e)}")
+        return f"\n[Error processing document asset {filename}: {str(e)}]\n"