"""Phase 0a: Existing Source Environment Discovery Node."""

import logging
import re
from typing import Any, Dict
from app.states.state import GCPArchitectureState
from app.skills.loader import SkillLoader
from app.tools.gcp_scanner import GCPEnvironmentScanner

logger = logging.getLogger(__name__)


def source_discover_node(state: GCPArchitectureState, skill_loader: SkillLoader) -> Dict[str, Any]:
    """Processes Phase 0a Source Environment Discovery: documents existing as-is architecture before target migration."""
    logger.info("Executing source_discover_node")
    skill_prompt = skill_loader.format_skills_for_prompt("source_discover")

    # Extract target GCP project ID from request if specified
    request_summary = state.get("workflow_request", "")
    project_id = state.get("project_id")
    if not project_id and request_summary:
        match = re.search(r"project\s+([a-z0-9-]+)", request_summary, re.IGNORECASE)
        if match:
            project_id = match.group(1)

    # Run live GCP environment scanner
    scanner = GCPEnvironmentScanner(project_id=project_id)
    scan_result = scanner.scan_environment()
    report = scanner.format_scan_report(scan_result)

    source_discovery_doc = report["doc"]
    source_mermaid_diagram = report["mermaid"]

    active_skills = state.get("active_skills", [])
    if "source_discovery" not in active_skills:
        active_skills.append("source_discovery")

    return {
        "source_discovery_doc": source_discovery_doc,
        "source_mermaid_diagram": source_mermaid_diagram,
        "current_phase": "source_discover",
        "active_skills": active_skills,
    }