debug_vertical_output.py

#!/usr/bin/env python3
"""
Debug script to identify and fix vertical output issues

Common causes of vertical output:
1. Iterating over string instead of string.split()
2. Incorrect streaming/chunking
3. Character encoding issues
4. JSON parsing issues
5. Improper list iteration
"""

import re

def test_common_vertical_output_bugs():
    """Test common patterns that cause vertical output"""
    
    print("=== Testing Common Vertical Output Patterns ===\n")
    
    # Example 1: Wrong iteration (causes vertical output)
    text = "Analyzed page structure and JavaScript interaction logic"
    
    print("❌ WRONG - Iterating over string directly:")
    for i, char in enumerate(text):
        if i < 10:  # Only show first 10 to avoid spam
            print(f"- {char}")
        elif i == 10:
            print("- ... (continuing vertically)")
            break
    
    print("\n✅ CORRECT - Process string as whole:")
    print(f"Normal output: {text}")
    
    # Example 2: Wrong list handling
    print("\n❌ WRONG - Processing response character by character:")
    response_text = "Normal response"
    # This is what causes vertical output:
    # for char in response_text:
    #     yield char  # Each character yielded separately
    
    print("✅ CORRECT - Process response as complete text:")
    print(f"Complete response: {response_text}")
    
    # Example 3: JSON parsing issues
    print("\n❌ WRONG - Treating JSON string as iterable:")
    json_like = '{"type": "text", "content": "Hello"}'
    print("If we iterate over JSON string:")
    for i, char in enumerate(json_like):
        if i < 15:
            print(f"- {char}")
        elif i == 15:
            print("- ... (vertical output)")
            break
    
    print("\n✅ CORRECT - Parse JSON first:")
    import json
    try:
        parsed = json.loads(json_like)
        print(f"Parsed content: {parsed.get('content', '')}")
    except:
        print("JSON parsing failed")

def analyze_anthropic_streaming_issue():
    """Analyze potential Anthropic streaming response issues"""
    print("\n=== Analyzing Anthropic API Response Patterns ===\n")
    
    # Simulate problematic streaming patterns
    print("Potential issue 1: Character-level streaming")
    response_content = "Analyzed page structure"
    
    print("❌ If processing character by character:")
    print("Output would be:")
    example_chars = ["A", "n", "a", "l", "y", "z", "e", "d", " ", "p", "a", "g", "e"]
    for char in example_chars:
        print(f"- {char}")
    
    print("\n✅ Correct streaming should accumulate:")
    accumulated = ""
    for char in example_chars:
        accumulated += char
    print(f"Final result: {accumulated}")
    
    print("\nPotential issue 2: Incorrect content block processing")
    # Simulate what might happen in create_message
    mock_response = {
        "content": [
            {"type": "text", "text": "Analyzed page structure and JavaScript interaction logic"}
        ]
    }
    
    print("❌ Wrong way - iterating over text characters:")
    text_content = mock_response["content"][0]["text"]
    print(f"If we do: for char in text_content")
    print("Result: Vertical output")
    
    print("\n✅ Correct way - use the text directly:")
    print(f"Direct use: {text_content}")

def check_potential_fixes():
    """Show potential fixes for vertical output"""
    print("\n=== Potential Fixes ===\n")
    
    print("1. Fix streaming in anthropic_provider.py:")
    print("""
    # ❌ Wrong:
    for char in response.content:
        yield char
    
    # ✅ Correct:
    if response.content:
        yield response.content
    """)
    
    print("2. Fix content processing:")
    print("""
    # ❌ Wrong:
    for content_block in response.content:
        if content_block.type == "text":
            for char in content_block.text:  # This causes vertical output
                yield char
    
    # ✅ Correct:
    for content_block in response.content:
        if content_block.type == "text":
            yield content_block.text  # Yield complete text
    """)
    
    print("3. Fix message handling:")
    print("""
    # ❌ Wrong:
    assistant_content = ""
    for item in response:
        assistant_content += item  # If item is individual characters
    
    # ✅ Correct:
    assistant_content = response.get("content", "")  # Get complete content
    """)

def main():
    """Run all debug tests"""
    test_common_vertical_output_bugs()
    analyze_anthropic_streaming_issue()
    check_potential_fixes()
    
    print("\n" + "="*60)
    print("DIAGNOSIS:")
    print("The vertical output (each character on a separate line)")
    print("is typically caused by:")
    print("1. Iterating over a string character by character")
    print("2. Yielding individual characters in streaming responses")
    print("3. Processing API response content incorrectly")
    print("\nTo fix this, check:")
    print("- anthropic_provider.py: create_message method")
    print("- agent.py: response processing loop")
    print("- Any place where string iteration happens")

if __name__ == "__main__":
    main()