support python code execution agent

app.py

@@ -13,6 +13,7 @@ import requests
 from requests.exceptions import RequestException
 from youtube_transcript_api import YouTubeTranscriptApi
 from urllib.parse import urlparse, parse_qs
+import json
 
 from dotenv import load_dotenv
 
@@ -452,30 +453,98 @@ def get_youtube_transcript(video_id: str, language: str = "en") -> str:
     except Exception as e:
         return f"Error retrieving transcript: {str(e)}"
 
-# Tool to analyze the transcript and answer a question
-# @tool
-# def analyze_transcript(transcript: str, question: str) -> str:
-#     """
-#     Analyze the provided video transcript to answer a specific question.
-    
-#     Args:
-#         transcript: The full transcript text of the video
-#         question: The specific question to answer based on the transcript
+
+@tool
+def load_text_file(file_path: str, detect_format: bool = True) -> str:
+    """
+    Loads a text file and optionally detects and processes its format (plain text, code, JSON, etc.).
     
-#     Returns:
-#         The answer to the question based on the transcript content
-#     """
-#     # This tool leverages the LLM's understanding capability
-#     # The implementation is simple because the LLM will do the analysis
-#     if not transcript or transcript.startswith("Error"):
-#         return f"Unable to analyze transcript: {transcript}"
+    Args:
+        file_path: Path to the text file to load
+        detect_format: Whether to automatically detect and process the format (default: True)
         
-#     # Check if transcript has enough content to analyze
-#     if len(transcript.split()) < 10:
-#         return "The transcript is too short or incomplete to properly analyze."
-    
-#     return "TRANSCRIPT_ANALYSIS_PLACEHOLDER"  # This will be replaced by LLM reasoning
+    Returns:
+        String containing the file content, possibly formatted based on detected type
+    """
 
+    if not os.path.exists(file_path):
+        return f"Error: File not found at {file_path}"
+
+    try:
+        # Read the file content
+        with open(file_path, 'r', encoding='utf-8') as file:
+            content = file.read()
+
+        if not detect_format:
+            return f"File content ({os.path.basename(file_path)}):\n\n{content}"
+
+        # Get file extension
+        _, ext = os.path.splitext(file_path)
+        ext = ext.lower()
+
+        # Handle based on file extension or content detection
+        if ext in ['.json', '.geojson']:
+            # Process JSON
+            try:
+                parsed_json = json.loads(content)
+                formatted_json = json.dumps(parsed_json, indent=2)
+                return f"JSON content ({os.path.basename(file_path)}):\n\n{formatted_json}"
+            except json.JSONDecodeError:
+                return f"Warning: File has JSON extension but content is not valid JSON.\n\n{content}"
+
+        elif ext in ['.py', '.js', '.ts', '.java', '.c', '.cpp', '.cs', '.php', '.rb', '.go', '.rs', '.swift']:
+            # It's a code file, return with appropriate formatting
+            return f"Code file ({os.path.basename(file_path)}, {ext[1:]} language):\n\n{content}"
+
+        elif ext in ['.csv', '.tsv']:
+            # Handle CSV/TSV files with preview
+            lines = content.strip().split('\n')
+            preview_lines = lines[:min(10, len(lines))]
+            preview = '\n'.join(preview_lines)
+
+            if len(lines) > 10:
+                preview += f"\n\n[...and {len(lines) - 10} more lines]"
+
+            return f"Tabular data file ({os.path.basename(file_path)}):\n\n{preview}"
+
+        else:
+            # Try to detect JSON content regardless of extension
+            if content.strip().startswith('{') and content.strip().endswith('}'):
+                try:
+                    parsed_json = json.loads(content)
+                    formatted_json = json.dumps(parsed_json, indent=2)
+                    return f"Detected JSON content ({os.path.basename(file_path)}):\n\n{formatted_json}"
+                except json.JSONDecodeError:
+                    pass  # Not valid JSON, continue with other detection
+
+            # Try to detect if it might be code
+            code_indicators = [
+                'def ', 'class ', 'function ', 'import ', 'from ', 'var ', 'let ', 'const ',
+                '#include', 'package ', 'using ', 'public class', 'int main'
+            ]
+
+            if any(indicator in content for indicator in code_indicators):
+                language = "unknown programming language"
+                return f"Detected code file ({os.path.basename(file_path)}, {language}):\n\n{content}"
+
+            # Check if it's XML/HTML-like
+            if re.search(r'<\w+>.*?</\w+>', content, re.DOTALL) or content.strip().startswith('<?xml'):
+                return f"Markup language file ({os.path.basename(file_path)}):\n\n{content}"
+
+            # Default to plain text
+            return f"Plain text file ({os.path.basename(file_path)}):\n\n{content}"
+
+    except UnicodeDecodeError:
+        # Try different encoding if UTF-8 fails
+        try:
+            with open(file_path, 'r', encoding='latin-1') as file:
+                content = file.read()
+            return f"File content ({os.path.basename(file_path)}, non-UTF-8 encoding):\n\n{content}"
+        except Exception as e:
+            return f"Error reading file (encoding issues): {str(e)}"
+    except Exception as e:
+        return f"Error reading file: {str(e)}"
+    
 
 
 class BasicAgent:
@@ -628,6 +697,23 @@ class BasicAgent:
         """)
         )
 
+        self.python_code_executer = CodeAgent(
+            model=reasoning_model,
+            tools=[load_text_file,
+                   FinalAnswerTool()],
+            name="python_code_executer",
+            description=textwrap.dedent("""\
+                You are an expert assistant that can execute Python code.
+
+                Execute the Python code and return the final answer using the `final_answer` tool.
+            """),
+            additional_authorized_imports=["json", "re", "pandas", "numpy", "math", "collections", "itertools", "stat", "statistics", "queue", "unicodedata", "time", "random", "datetime"],
+            verbosity_level=1,
+            max_steps=5,
+            planning_interval=1,
+            #executor_type="e2b",
+            #use_e2b_executor=True
+        )
 
         # Define the Manager Agent
         # This agent manages tasks and delegates to other agents
@@ -638,11 +724,13 @@ class BasicAgent:
                    load_xlsx_file_as_dataframe,
                    load_xlsx_file_as_markdown,
                    query_wikipedia,
+                   load_text_file,
                    FinalAnswerTool()],
             # Specify the agents this manager oversees
             managed_agents=[self.web_search_specialist_agent,
                             self.reasoning_agent,
-                            self.youtube_qa_agent],
+                            self.youtube_qa_agent,
+                            self.python_code_executer],
             name="manager_agent",  # Give the manager agent a name
             description="Manages tasks by delegating to other agents.",  # Describe its role
             additional_authorized_imports=[