aitherboard/src/lib/services/content/git-repo-fetcher.ts

/**
 * Service for fetching git repository data from various hosting platforms
 * Supports GitHub, GitLab, Gitea, and other git hosting services
 */

export interface GitRepoInfo {
  name: string;
  description?: string;
  url: string;
  defaultBranch: string;
  branches: GitBranch[];
  commits: GitCommit[];
  files: GitFile[];
  readme?: {
    path: string;
    content: string;
    format: 'markdown' | 'asciidoc';
  };
}

export interface GitBranch {
  name: string;
  commit: {
    sha: string;
    message: string;
    author: string;
    date: string;
  };
}

export interface GitCommit {
  sha: string;
  message: string;
  author: string;
  date: string;
}

export interface GitFile {
  name: string;
  path: string;
  type: 'file' | 'dir';
  size?: number;
}

/**
 * Parse git URL to extract platform, owner, and repo
 */
function parseGitUrl(url: string): { platform: string; owner: string; repo: string; baseUrl: string } | null {
  // GitHub
  const githubMatch = url.match(/github\.com[/:]([^/]+)\/([^/]+?)(?:\.git)?\/?$/);
  if (githubMatch) {
    return {
      platform: 'github',
      owner: githubMatch[1],
      repo: githubMatch[2].replace(/\.git$/, ''),
      baseUrl: 'https://api.github.com'
    };
  }

  // GitLab
  const gitlabMatch = url.match(/gitlab\.com[/:]([^/]+)\/([^/]+?)(?:\.git)?\/?$/);
  if (gitlabMatch) {
    return {
      platform: 'gitlab',
      owner: gitlabMatch[1],
      repo: gitlabMatch[2].replace(/\.git$/, ''),
      baseUrl: 'https://gitlab.com/api/v4'
    };
  }

  // Gitea (generic pattern)
  const giteaMatch = url.match(/(https?:\/\/[^/]+)[/:]([^/]+)\/([^/]+?)(?:\.git)?\/?$/);
  if (giteaMatch) {
    return {
      platform: 'gitea',
      owner: giteaMatch[2],
      repo: giteaMatch[3].replace(/\.git$/, ''),
      baseUrl: `${giteaMatch[1]}/api/v1`
    };
  }

  return null;
}

/**
 * Fetch repository data from GitHub
 */
async function fetchFromGitHub(owner: string, repo: string): Promise<GitRepoInfo | null> {
  try {
    const repoResponse = await fetch(`https://api.github.com/repos/${owner}/${repo}`);
    if (!repoResponse.ok) {
      console.warn(`GitHub API error for repo ${owner}/${repo}: ${repoResponse.status} ${repoResponse.statusText}`);
      return null;
    }
    const repoData = await repoResponse.json();
    
    const defaultBranch = repoData.default_branch || 'main';
    const [branchesResponse, commitsResponse, treeResponse] = await Promise.all([
      fetch(`https://api.github.com/repos/${owner}/${repo}/branches`),
      fetch(`https://api.github.com/repos/${owner}/${repo}/commits?per_page=10`),
      fetch(`https://api.github.com/repos/${owner}/${repo}/git/trees/${defaultBranch}?recursive=1`).catch(() => null)
    ]);

    // Check if responses are OK and parse JSON
    let branchesData: any[] = [];
    let commitsData: any[] = [];
    let treeData: any = null;

    if (branchesResponse && branchesResponse.ok) {
      branchesData = await branchesResponse.json();
      if (!Array.isArray(branchesData)) {
        console.warn('GitHub branches response is not an array:', branchesData);
        branchesData = [];
      }
    } else {
      console.warn(`GitHub API error for branches: ${branchesResponse?.status || 'unknown'}`);
    }

    if (commitsResponse && commitsResponse.ok) {
      commitsData = await commitsResponse.json();
      if (!Array.isArray(commitsData)) {
        console.warn('GitHub commits response is not an array:', commitsData);
        commitsData = [];
      }
    } else {
      console.warn(`GitHub API error for commits: ${commitsResponse?.status || 'unknown'}`);
    }

    if (treeResponse && treeResponse.ok) {
      treeData = await treeResponse.json();
    }

    // Create a map of commit SHAs to commit details for lookup
    const commitMap = new Map<string, { message: string; author: string; date: string }>();
    for (const c of commitsData) {
      if (c.sha) {
        const commitObj = c.commit || {};
        commitMap.set(c.sha, {
          message: commitObj.message ? commitObj.message.split('\n')[0] : '',
          author: commitObj.author?.name || commitObj.committer?.name || 'Unknown',
          date: commitObj.author?.date || commitObj.committer?.date || new Date().toISOString()
        });
      }
    }

    const branches: GitBranch[] = branchesData.map((b: any) => {
      const commitSha = b.commit?.sha || '';
      // Try to get commit details from the commit object first, then fall back to our commit map
      const commitObj = b.commit?.commit || {};
      let commitMessage = commitObj.message ? commitObj.message.split('\n')[0] : '';
      let commitAuthor = commitObj.author?.name || commitObj.committer?.name || '';
      let commitDate = commitObj.author?.date || commitObj.committer?.date || '';
      
      // If commit details are missing, try to find them in our commit map
      if (!commitMessage && commitSha) {
        const mappedCommit = commitMap.get(commitSha);
        if (mappedCommit) {
          commitMessage = mappedCommit.message;
          commitAuthor = mappedCommit.author;
          commitDate = mappedCommit.date;
        }
      }
      
      // Final fallbacks
      if (!commitMessage) commitMessage = 'No commit message';
      if (!commitAuthor) commitAuthor = 'Unknown';
      if (!commitDate) commitDate = new Date().toISOString();
      
      return {
        name: b.name || '',
        commit: {
          sha: commitSha,
          message: commitMessage,
          author: commitAuthor,
          date: commitDate
        }
      };
    });

    const commits: GitCommit[] = commitsData.map((c: any) => {
      const commitObj = c.commit || {};
      const message = commitObj.message ? commitObj.message.split('\n')[0] : '';
      const author = commitObj.author?.name || commitObj.committer?.name || 'Unknown';
      const date = commitObj.author?.date || commitObj.committer?.date || new Date().toISOString();
      
      return {
        sha: c.sha || '',
        message: message,
        author: author,
        date: date
      };
    });

    const files: GitFile[] = treeData?.tree?.filter((item: any) => item.type === 'blob' || item.type === 'tree').map((item: any) => ({
      name: item.path.split('/').pop(),
      path: item.path,
      type: item.type === 'tree' ? 'dir' : 'file',
      size: item.size
    })) || [];

    // Try to fetch README (prioritize .adoc over .md)
    let readme: { path: string; content: string; format: 'markdown' | 'asciidoc' } | undefined;
    const readmeFiles = ['README.adoc', 'README.md', 'README.rst', 'README.txt'];
    for (const readmeFile of readmeFiles) {
      try {
        const readmeData = await fetch(`https://api.github.com/repos/${owner}/${repo}/contents/${readmeFile}`).then(r => {
          if (!r.ok) throw new Error('Not found');
          return r.json();
        });
        if (readmeData.content) {
          const content = atob(readmeData.content.replace(/\s/g, ''));
          readme = {
            path: readmeFile,
            content,
            format: readmeFile.toLowerCase().endsWith('.adoc') ? 'asciidoc' : 'markdown'
          };
          break; // Found a README, stop searching
        }
      } catch {
        continue; // Try next file
      }
    }

    return {
      name: repoData.name,
      description: repoData.description,
      url: repoData.html_url,
      defaultBranch: repoData.default_branch,
      branches,
      commits,
      files,
      readme
    };
  } catch (error) {
    console.error('Error fetching from GitHub:', error);
    return null;
  }
}

/**
 * Fetch repository data from GitLab
 */
async function fetchFromGitLab(owner: string, repo: string, baseUrl: string): Promise<GitRepoInfo | null> {
  try {
    const projectPath = `${owner}/${repo}`;
    const encodedPath = encodeURIComponent(projectPath);
    
    const [repoData, branchesData, commitsData] = await Promise.all([
      fetch(`${baseUrl}/projects/${encodedPath}`).then(r => r.json()),
      fetch(`${baseUrl}/projects/${encodedPath}/repository/branches`).then(r => r.json()),
      fetch(`${baseUrl}/projects/${encodedPath}/repository/commits?per_page=10`).then(r => r.json())
    ]);

    const branches: GitBranch[] = branchesData.map((b: any) => ({
      name: b.name,
      commit: {
        sha: b.commit.id,
        message: b.commit.message.split('\n')[0],
        author: b.commit.author_name,
        date: b.commit.committed_date
      }
    }));

    const commits: GitCommit[] = commitsData.map((c: any) => ({
      sha: c.id,
      message: c.message.split('\n')[0],
      author: c.author_name,
      date: c.committed_date
    }));

    // Fetch file tree
    let files: GitFile[] = [];
    try {
      const treeData = await fetch(`${baseUrl}/projects/${encodedPath}/repository/tree?recursive=true&per_page=100`).then(r => r.json());
      files = treeData.map((item: any) => ({
        name: item.name,
        path: item.path,
        type: item.type === 'tree' ? 'dir' : 'file',
        size: item.size
      }));
    } catch {
      // Tree fetch failed
    }

    // Try to fetch README (prioritize .adoc over .md)
    let readme: { path: string; content: string; format: 'markdown' | 'asciidoc' } | undefined;
    const readmeFiles = ['README.adoc', 'README.md', 'README.rst', 'README.txt'];
    for (const readmeFile of readmeFiles) {
      try {
        const fileData = await fetch(`${baseUrl}/projects/${encodedPath}/repository/files/${encodeURIComponent(readmeFile)}/raw?ref=${repoData.default_branch}`).then(r => {
          if (!r.ok) throw new Error('Not found');
          return r.text();
        });
        readme = {
          path: readmeFile,
          content: fileData,
          format: readmeFile.toLowerCase().endsWith('.adoc') ? 'asciidoc' : 'markdown'
        };
        break; // Found a README, stop searching
      } catch {
        continue; // Try next file
      }
    }

    return {
      name: repoData.name,
      description: repoData.description,
      url: repoData.web_url,
      defaultBranch: repoData.default_branch,
      branches,
      commits,
      files,
      readme
    };
  } catch (error) {
    console.error('Error fetching from GitLab:', error);
    return null;
  }
}

/**
 * Fetch repository data from Gitea
 */
async function fetchFromGitea(owner: string, repo: string, baseUrl: string): Promise<GitRepoInfo | null> {
  try {
    const [repoData, branchesData, commitsData] = await Promise.all([
      fetch(`${baseUrl}/repos/${owner}/${repo}`).then(r => r.json()),
      fetch(`${baseUrl}/repos/${owner}/${repo}/branches`).then(r => r.json()),
      fetch(`${baseUrl}/repos/${owner}/${repo}/commits?limit=10`).then(r => r.json())
    ]);

    const branches: GitBranch[] = branchesData.map((b: any) => ({
      name: b.name,
      commit: {
        sha: b.commit.id,
        message: b.commit.message.split('\n')[0],
        author: b.commit.author.name,
        date: b.commit.timestamp
      }
    }));

    const commits: GitCommit[] = commitsData.map((c: any) => ({
      sha: c.sha,
      message: c.commit.message.split('\n')[0],
      author: c.commit.author.name,
      date: c.commit.timestamp
    }));

    // Fetch file tree
    let files: GitFile[] = [];
    try {
      const treeData = await fetch(`${baseUrl}/repos/${owner}/${repo}/contents?ref=${repoData.default_branch}`).then(r => r.json());
      files = treeData.map((item: any) => ({
        name: item.name,
        path: item.path,
        type: item.type === 'dir' ? 'dir' : 'file',
        size: item.size
      }));
    } catch {
      // Tree fetch failed
    }

    // Try to fetch README (prioritize .adoc over .md)
    let readme: { path: string; content: string; format: 'markdown' | 'asciidoc' } | undefined;
    const readmeFiles = ['README.adoc', 'README.md', 'README.rst', 'README.txt'];
    for (const readmeFile of readmeFiles) {
      try {
        const fileData = await fetch(`${baseUrl}/repos/${owner}/${repo}/contents/${readmeFile}?ref=${repoData.default_branch}`).then(r => {
          if (!r.ok) throw new Error('Not found');
          return r.json();
        });
        if (fileData.content) {
          const content = atob(fileData.content.replace(/\s/g, ''));
          readme = {
            path: readmeFile,
            content,
            format: readmeFile.toLowerCase().endsWith('.adoc') ? 'asciidoc' : 'markdown'
          };
          break; // Found a README, stop searching
        }
      } catch {
        continue; // Try next file
      }
    }

    return {
      name: repoData.name,
      description: repoData.description,
      url: repoData.html_url,
      defaultBranch: repoData.default_branch,
      branches,
      commits,
      files,
      readme
    };
  } catch (error) {
    console.error('Error fetching from Gitea:', error);
    return null;
  }
}

/**
 * Fetch repository data from a git URL
 */
export async function fetchGitRepo(url: string): Promise<GitRepoInfo | null> {
  const parsed = parseGitUrl(url);
  if (!parsed) {
    console.error('Unable to parse git URL:', url);
    return null;
  }

  const { platform, owner, repo, baseUrl } = parsed;

  switch (platform) {
    case 'github':
      return fetchFromGitHub(owner, repo);
    case 'gitlab':
      return fetchFromGitLab(owner, repo, baseUrl);
    case 'gitea':
      return fetchFromGitea(owner, repo, baseUrl);
    default:
      console.error('Unsupported platform:', platform);
      return null;
  }
}

/**
 * Convert SSH git URL to HTTPS format
 */
function convertSshToHttps(url: string): string | null {
  // Handle git@host:user/repo.git format
  const sshMatch = url.match(/git@([^:]+):(.+?)(?:\.git)?$/);
  if (sshMatch) {
    const [, host, path] = sshMatch;
    return `https://${host}/${path}${path.endsWith('.git') ? '' : '.git'}`;
  }
  return null;
}

/**
 * Extract git URLs from a Nostr event
 */
export function extractGitUrls(event: { tags: string[][]; content: string }): string[] {
  const urls: string[] = [];
  
  // Check tags for git URLs (including 'clone' tag which is used in NIP-34)
  for (const tag of event.tags) {
    if (tag[0] === 'r' || tag[0] === 'url' || tag[0] === 'git' || tag[0] === 'clone') {
      const url = tag[1];
      if (!url) continue;
      
      // Convert SSH URLs to HTTPS
      if (url.startsWith('git@')) {
        const httpsUrl = convertSshToHttps(url);
        if (httpsUrl) {
          urls.push(httpsUrl);
          continue;
        }
      }
      
      // Check if it's a git URL
      if (url.includes('github.com') || url.includes('gitlab.com') || url.includes('gitea') || url.includes('.git') || url.startsWith('http')) {
        urls.push(url);
      }
    }
  }
  
  // Check content for git URLs
  const urlRegex = /(https?:\/\/[^\s]+\.git|https?:\/\/(?:github|gitlab|gitea)[^\s]+)/gi;
  const matches = event.content.match(urlRegex);
  if (matches) {
    urls.push(...matches);
  }
  
  return [...new Set(urls)]; // Deduplicate
}