#!/usr/bin/env python3
"""
Test script for LocalAI Content Engine blog generation.
Uses phi4:14b on Gaming PC via Tailscale.
"""

import httpx
import json
import sys
from datetime import datetime

# Gaming PC Ollama endpoint
OLLAMA_URL = "http://matt-pc.tail864e81.ts.net:11434/api/generate"
MODEL = "phi4:14b"

# Sovereign Operator voice — banned words
BANNED_WORDS = [
    "delve", "tapestry", "moreover", "leveraging", "holistic", 
    "paradigm", "synergy", "unlock", "potential", "seamless",
    "robust", "streamline", "utilize", "embark", "journey",
    "transformative", "cutting-edge", "innovative", "groundbreaking"
]

SYSTEM_PROMPT = """You are a technical writer with a sovereign, pragmatic voice.
Your style is direct, specific, and human. You write about home infrastructure, 
AI systems, and the messy reality of building things.

Rules:
- Start with the human moment, not the solution
- Admit wrong turns before revealing fixes
- Use specific versions, error messages, timestamps
- End with actionable takeaways
- NEVER use these words: {banned}
- Write like a competent engineer talking to another engineer
""".format(banned=", ".join(BANNED_WORDS))


def generate_content(prompt: str, max_tokens: int = 2000) -> str:
    """Generate content via Ollama on Gaming PC."""
    
    full_prompt = f"{SYSTEM_PROMPT}\n\n{prompt}\n\nWrite 800-1200 words. Be specific. No fluff."
    
    response = httpx.post(
        OLLAMA_URL,
        json={
            "model": MODEL,
            "prompt": full_prompt,
            "stream": False,
            "options": {
                "temperature": 0.7,
                "num_predict": max_tokens,
                "top_p": 0.9,
            }
        },
        timeout=300.0
    )
    response.raise_for_status()
    return response.json()["response"]


def generate_titles(topic: str) -> list:
    """Generate 3-5 title options for a topic."""
    
    prompt = f"""Topic: {topic}

Generate 5 title options for a blog post. Mix of styles:
- Direct/how-to
- Story-based
- Specific technical angle

Return ONLY a JSON array of strings. No markdown, no explanation.
"""
    
    response = generate_content(prompt, max_tokens=500)
    
    # Extract JSON from response
    try:
        # Find JSON array in response
        start = response.find("[")
        end = response.rfind("]") + 1
        if start >= 0 and end > start:
            return json.loads(response[start:end])
    except:
        pass
    
    # Fallback: parse line by line
    lines = [l.strip("- •\"' ") for l in response.split("\n") if l.strip()]
    return [l for l in lines if l and not l.startswith("{")][:5]


def generate_draft(title: str, outline: list) -> str:
    """Generate a full blog post draft from title and outline."""
    
    outline_text = "\n".join(f"- {item}" for item in outline)
    
    prompt = f"""Title: {title}

Outline:
{outline_text}

Write the full blog post following the outline. Match the voice rules.
Include specific details, error messages, timestamps where relevant.
"""
    
    return generate_content(prompt, max_tokens=4000)


def generate_seo(content: str) -> dict:
    """Generate SEO metadata from content."""
    
    prompt = f"""Given this blog post content, generate:
1. A 1-2 sentence excerpt/summary
2. 5-7 relevant tags (comma-separated)
3. A meta description (under 160 characters)

Content excerpt (first 500 chars):
{content[:500]}

Return ONLY JSON in this format:
{{"excerpt": "...", "tags": "tag1, tag2, tag3", "meta_description": "..."}}
"""
    
    response = generate_content(prompt, max_tokens=500)
    
    try:
        start = response.find("{")
        end = response.rfind("}") + 1
        if start >= 0 and end > start:
            return json.loads(response[start:end])
    except:
        pass
    
    return {
        "excerpt": "Blog post about " + content[:50],
        "tags": "home-lab, ai, infrastructure",
        "meta_description": content[:160]
    }


def test_pipeline():
    """Run a full content generation test."""
    
    print("=" * 60)
    print("LOCALAI CONTENT ENGINE TEST")
    print(f"Model: {MODEL}")
    print(f"Time: {datetime.now().isoformat()}")
    print("=" * 60)
    
    # Test 1: Title Generation
    print("\n📝 TEST 1: Title Generation")
    print("-" * 40)
    topic = "How I migrated from Google Calendar to a self-hosted CalDAV server"
    print(f"Topic: {topic}")
    
    try:
        titles = generate_titles(topic)
        print(f"\nGenerated {len(titles)} titles:")
        for i, title in enumerate(titles, 1):
            print(f"  {i}. {title}")
    except Exception as e:
        print(f"❌ Title generation failed: {e}")
        titles = []
    
    # Test 2: Draft Generation
    print("\n📝 TEST 2: Draft Generation")
    print("-" * 40)
    
    if titles:
        title = titles[0]
    else:
        title = "The Night I Broke DNS: A CalDAV Migration Story"
    
    outline = [
        "The trigger: Aundrea asks why the family calendar is empty",
        "The Google Calendar dependency audit",
        "Choosing Radicale over NextCloud (simplicity wins)",
        "The migration: icalendar + caldav Python scripts",
        "Setting up Cloudflare Email Worker for webhook parsing",
        "The moment it worked: seeing events sync to phones",
        "What I'd do differently next time"
    ]
    
    print(f"Title: {title}")
    print(f"Outline: {len(outline)} points")
    
    try:
        start = datetime.now()
        draft = generate_draft(title, outline)
        elapsed = (datetime.now() - start).total_seconds()
        
        print(f"\n✅ Draft generated in {elapsed:.1f}s")
        print(f"Length: {len(draft)} chars, ~{len(draft.split())} words")
        print(f"\n--- DRAFT PREVIEW (first 800 chars) ---")
        print(draft[:800])
        print("...")
        print(f"\n--- END PREVIEW ---")
        
        # Test 3: SEO Generation
        print("\n📝 TEST 3: SEO Metadata Generation")
        print("-" * 40)
        
        try:
            seo = generate_seo(draft)
            print(f"Excerpt: {seo.get('excerpt', 'N/A')}")
            print(f"Tags: {seo.get('tags', 'N/A')}")
            print(f"Meta: {seo.get('meta_description', 'N/A')}")
        except Exception as e:
            print(f"❌ SEO generation failed: {e}")
        
        # Save output
        output = {
            "title": title,
            "titles": titles,
            "outline": outline,
            "draft": draft,
            "seo": seo if 'seo' in dir() else {},
            "metadata": {
                "model": MODEL,
                "generated_at": datetime.now().isoformat(),
                "elapsed_seconds": elapsed,
                "draft_length": len(draft),
                "word_count": len(draft.split())
            }
        }
        
        output_file = f"/home/hoffmann_admin/.openclaw/workspace-socrates/test_output_{datetime.now().strftime('%Y%m%d_%H%M%S')}.json"
        with open(output_file, 'w') as f:
            json.dump(output, f, indent=2)
        
        print(f"\n💾 Full output saved to: {output_file}")
        
    except Exception as e:
        print(f"❌ Draft generation failed: {e}")
        import traceback
        traceback.print_exc()
    
    print("\n" + "=" * 60)
    print("TEST COMPLETE")
    print("=" * 60)


if __name__ == "__main__":
    test_pipeline()