import json
import re
import sys

ALLOWED_VERTICALS = {
    "Software Engineering", "Information Technology", "Finance",
    "Corporate Governance", "Marketing", "Operations", "Administration",
    "Executive", "Geopolitics", "Legal", "Public Relations", "Design",
    "Sales", "Human Resources", "Customer Support",
    # Approved 2026-08-16, seeded from existing IT security roles.
    "Cybersecurity",
}

def validate_matrix(file_path):
    try:
        with open(file_path, 'r', encoding='utf-8') as f:
            data = json.load(f)
    except Exception as e:
        print(f"FAILED: Invalid JSON syntax in {file_path} - {e}")
        return False

    if not isinstance(data, list):
        print("FAILED: Top-level JSON structure must be an array.")
        return False

    errors_found = 0

    for idx, agent in enumerate(data):
        title = agent.get('title', f'Record #{idx}')
        agent_id = agent.get('id', '')
        vertical = agent.get('vertical', '')
        keywords = agent.get('keywords', [])
        gem = agent.get('gem', '')

        # 1. ID Slug Verification
        expected_slug = re.sub(r'[^a-z0-9]+', '-', title.lower()).strip('-')
        if agent_id != expected_slug:
            print(f"[{title}] ID Error: Expected '{expected_slug}', got '{agent_id}'")
            errors_found += 1

        # 1b. Vertical Verification - an unknown vertical silently forks a
        # category and shifts every accent colour after it in app.js.
        if vertical not in ALLOWED_VERTICALS:
            print(f"[{title}] Vertical Error: '{vertical}' is not an approved vertical.")
            errors_found += 1

        # 2. Keywords Verification
        if not isinstance(keywords, list) or not (5 <= len(keywords) <= 8):
            print(f"[{title}] Keywords Error: Must contain 5 to 8 strings. Found {len(keywords)}.")
            errors_found += 1
        elif not all(isinstance(k, str) and k.islower() for k in keywords):
            print(f"[{title}] Keywords Error: All keywords must be lowercase strings.")
            errors_found += 1

        # 3. Prohibited Characters Check
        for bad_char in ["#", "|", "$$", "```"]:
            if bad_char in gem:
                print(f"[{title}] Gem Error: Contains forbidden character sequence '{bad_char}'.")
                errors_found += 1

        # 4. Word Count Check (120 - 200 words)
        words = gem.split()
        if not (120 <= len(words) <= 200):
            print(f"[{title}] Word Count Error: Gem contains {len(words)} words (Must be 120-200).")
            errors_found += 1

        # 5. Exact 5-Part Structural Check
        lines = [line.strip() for line in gem.strip().split('\n') if line.strip()]
        
        if len(lines) != 8:
            print(f"[{title}] Structure Error: Must have exactly 8 lines. Found {len(lines)}.")
            errors_found += 1
            continue

        if not lines[0].startswith(f"You are {title},"):
            print(f"[{title}] Line 1 Error: Must start with 'You are {title},'")
            errors_found += 1

        if not lines[1].startswith("Task:"):
            print(f"[{title}] Line 2 Error: Must start with 'Task:'")
            errors_found += 1

        if lines[2] != "Rules:":
            print(f"[{title}] Line 3 Error: Must be exactly 'Rules:'")
            errors_found += 1

        if not lines[3].startswith("- "):
            print(f"[{title}] Line 4 Error (Scope Rule): Must start with '- '")
            errors_found += 1

        if not lines[4].startswith("- "):
            print(f"[{title}] Line 5 Error (Method Rule): Must start with '- '")
            errors_found += 1

        req_missing_info = "- If key details are missing, ask exactly one clarifying question, then proceed with stated assumptions."
        if lines[5] != req_missing_info:
            print(f"[{title}] Line 6 Error: Missing info rule text does not match required string.")
            errors_found += 1

        if not lines[6].startswith("- If asked something outside") or "say it's out of scope and name the right kind of expert instead." not in lines[6]:
            print(f"[{title}] Line 7 Error: Out-of-scope rule text formatting mismatch.")
            errors_found += 1

        if not lines[7].startswith("Output:"):
            print(f"[{title}] Line 8 Error: Must start with 'Output:'")
            errors_found += 1

    if errors_found == 0:
        print(f"SUCCESS: All {len(data)} agents in '{file_path}' passed validator gates.")
        return True
    else:
        print(f"\nFAILED: Found {errors_found} total validation issues.")
        return False

if __name__ == "__main__":
    target = sys.argv[1] if len(sys.argv) > 1 else "agents_matrix.json"
    # Exit code is the gate: 0 = SUCCESS, 1 = FAILED. Scripts may rely on it.
    sys.exit(0 if validate_matrix(target) else 1)