"""GitHub Repository Analyzer Agent."""

from __future__ import annotations

from researchlink.agents.base import BaseAgent
from researchlink.providers.base import AgentRole
from researchlink.schemas.repository import RepoAnalysis
from researchlink.services.github_client import (
    detect_language_and_frameworks,
    fetch_readme,
    fetch_repo_metadata,
    fetch_tree,
    parse_github_url,
)


class GitHubRepoAnalyzerAgent(BaseAgent):
    name = "GitHubRepoAnalyzerAgent"
    task_role = AgentRole.analysis

    def run(self, github_url: str | None) -> RepoAnalysis:
        analysis = RepoAnalysis(url=github_url)

        if not github_url:
            analysis.analysis_notes.append("No GitHub URL provided.")
            return analysis

        owner, repo_name = parse_github_url(github_url)
        if not owner or not repo_name:
            analysis.analysis_notes.append(f"Could not parse owner/repo from URL: {github_url}")
            return analysis

        analysis.owner = owner
        analysis.repo_name = repo_name
        self.log(f"Analyzing {owner}/{repo_name}...")

        # Fetch metadata
        meta = fetch_repo_metadata(owner, repo_name)
        if "error" in meta:
            analysis.repo_accessible = False
            analysis.access_error = meta["error"]
            analysis.analysis_notes.append(f"GitHub API error: {meta['error']}")
            self.warn(f"GitHub API error: {meta['error']}")
            return analysis

        analysis.repo_accessible = True

        # Fetch README
        readme = fetch_readme(owner, repo_name)
        analysis.readme_content = readme[:3000] if readme else None

        # Fetch tree
        tree = fetch_tree(owner, repo_name)
        analysis.top_level_folders = [
            item["path"] for item in tree if item.get("type") == "tree"
        ]
        analysis.top_level_files = [
            item["path"] for item in tree if item.get("type") == "blob"
        ]

        # Detect language/frameworks
        detected = detect_language_and_frameworks(tree, readme)
        analysis.detected_language = detected["languages"][0] if detected["languages"] else None
        analysis.detected_frameworks = detected["frameworks"]
        analysis.setup_commands = detected["setup_commands"]
        analysis.has_tests = detected["has_tests"]
        analysis.test_paths = detected["test_paths"]

        # Heuristic: is this likely the official implementation?
        readme_lower = (readme or "").lower()
        analysis.is_likely_official_impl = any(
            kw in readme_lower for kw in ["official", "paper", "implementation", "code for"]
        )
        analysis.official_impl_confidence = "medium" if analysis.is_likely_official_impl else "low"

        analysis.analysis_notes.append(
            "Repository analysis based on GitHub API (public access, unauthenticated). "
            "Exact reproducibility has NOT been verified."
        )

        self.log(
            f"Repo analysis done. Language: {analysis.detected_language}, "
            f"frameworks: {analysis.detected_frameworks}"
        )
        return analysis