WordpressGitSync/includes/api-client.php

<?php
/**
 * Oribi Sync — Git provider API client.
 *
 * Supports GitHub, GitLab, Bitbucket Cloud, Gitea / Forgejo,
 * and Azure DevOps REST APIs.
 * Fetches repository tree and file contents using PAT authentication.
 */

if ( ! defined( 'ABSPATH' ) ) exit;

// ─── Supported providers ──────────────────────────────────────────────────────

/**
 * Return the list of supported providers for the admin dropdown.
 */
function oribi_sync_providers(): array {
    return [
        'github'    => 'GitHub',
        'gitlab'    => 'GitLab',
        'bitbucket' => 'Bitbucket Cloud',
        'gitea'     => 'Gitea / Forgejo',
        'azure'     => 'Azure DevOps',
    ];
}

/**
 * Get the configured provider (stored in options).
 */
function oribi_sync_get_provider(): string {
    $provider = get_option( 'oribi_sync_provider', '' );
    if ( ! empty( $provider ) && array_key_exists( $provider, oribi_sync_providers() ) ) {
        return $provider;
    }
    // Auto-detect fallback for existing installs without the option
    $repo = get_option( 'oribi_sync_repo', '' );
    return oribi_sync_detect_provider( $repo );
}

/**
 * Auto-detect the Git provider from the repo URL.
 * Used as fallback when no explicit provider is set.
 *
 * @return string Provider key.
 */
function oribi_sync_detect_provider( string $repo_url ): string {
    if ( stripos( $repo_url, 'github.com' ) !== false )    return 'github';
    if ( stripos( $repo_url, 'gitlab.com' ) !== false )    return 'gitlab';
    if ( stripos( $repo_url, 'gitlab' ) !== false )        return 'gitlab';
    if ( stripos( $repo_url, 'bitbucket.org' ) !== false ) return 'bitbucket';
    if ( stripos( $repo_url, 'dev.azure.com' ) !== false ) return 'azure';
    if ( stripos( $repo_url, 'visualstudio.com' ) !== false ) return 'azure';
    // Assume Gitea for unknown self-hosted (most compatible generic API)
    return 'gitea';
}

// ─── URL parsing ──────────────────────────────────────────────────────────────

/**
 * Parse owner/repo (and optionally project) from a Git URL.
 *
 * Accepts HTTPS and SSH styles for all providers.
 *
 * @return array{owner: string, repo: string, host?: string, project?: string}|WP_Error
 */
function oribi_sync_parse_repo_url( string $url ) {
    // Azure DevOps:  https://dev.azure.com/{org}/{project}/_git/{repo}
    if ( preg_match( '#https?://dev\.azure\.com/([^/]+)/([^/]+)/_git/([^/\s]+?)(?:\.git)?$#i', $url, $m ) ) {
        return [ 'owner' => $m[1], 'project' => $m[2], 'repo' => $m[3], 'host' => 'dev.azure.com' ];
    }
    // Azure DevOps legacy:  https://{org}.visualstudio.com/{project}/_git/{repo}
    if ( preg_match( '#https?://([^.]+)\.visualstudio\.com/([^/]+)/_git/([^/\s]+?)(?:\.git)?$#i', $url, $m ) ) {
        return [ 'owner' => $m[1], 'project' => $m[2], 'repo' => $m[3], 'host' => $m[1] . '.visualstudio.com' ];
    }
    // Generic HTTPS:  https://host/owner/repo
    if ( preg_match( '#https?://([^/]+)/([^/]+)/([^/\s.]+?)(?:\.git)?(?:/)?$#i', $url, $m ) ) {
        return [ 'owner' => $m[2], 'repo' => $m[3], 'host' => $m[1] ];
    }
    // SSH:  git@host:owner/repo.git
    if ( preg_match( '#[^@]+@([^:]+):([^/]+)/([^/\s.]+?)(?:\.git)?$#i', $url, $m ) ) {
        return [ 'owner' => $m[2], 'repo' => $m[3], 'host' => $m[1] ];
    }
    return new WP_Error( 'oribi_sync_bad_url', 'Could not parse owner/repo from the repository URL.' );
}

// ─── API base URLs ────────────────────────────────────────────────────────────

/**
 * Build the base API URL for the configured provider.
 */
function oribi_sync_api_base( string $provider, array $parsed ): string {
    $owner = $parsed['owner'];
    $repo  = $parsed['repo'];
    $host  = $parsed['host'] ?? '';

    switch ( $provider ) {
        case 'gitlab':
            $api_host = ( $host && $host !== 'gitlab.com' ) ? $host : 'gitlab.com';
            $encoded  = rawurlencode( $owner . '/' . $repo );
            return "https://{$api_host}/api/v4/projects/{$encoded}";

        case 'bitbucket':
            return "https://api.bitbucket.org/2.0/repositories/{$owner}/{$repo}";

        case 'azure':
            $project = $parsed['project'] ?? $repo;
            $org     = $owner;
            return "https://dev.azure.com/{$org}/{$project}/_apis/git/repositories/{$repo}";

        case 'gitea':
            // Works for Gitea, Forgejo, and most Gitea-compatible hosts
            $api_host = $host ?: 'localhost:3000';
            return "https://{$api_host}/api/v1/repos/{$owner}/{$repo}";

        case 'github':
        default:
            $api_host = ( $host && $host !== 'github.com' ) ? $host . '/api/v3' : 'api.github.com';
            return "https://{$api_host}/repos/{$owner}/{$repo}";
    }
}

// ─── Auth headers ─────────────────────────────────────────────────────────────

/**
 * Build authorization headers for the provider.
 *
 * @return array  Headers array to merge into request.
 */
function oribi_sync_auth_headers( string $provider, string $pat ): array {
    switch ( $provider ) {
        case 'bitbucket':
            // Bitbucket Cloud app passwords use Basic auth (username:app_password)
            // If PAT contains ':', treat as username:password; otherwise Bearer
            if ( strpos( $pat, ':' ) !== false ) {
                return [ 'Authorization' => 'Basic ' . base64_encode( $pat ) ];
            }
            return [ 'Authorization' => 'Bearer ' . $pat ];

        case 'azure':
            // Azure DevOps PATs use Basic auth with empty username
            return [ 'Authorization' => 'Basic ' . base64_encode( ':' . $pat ) ];

        case 'gitlab':
            return [ 'PRIVATE-TOKEN' => $pat ];

        case 'gitea':
            return [ 'Authorization' => 'token ' . $pat ];

        case 'github':
        default:
            return [ 'Authorization' => 'Bearer ' . $pat ];
    }
}

// ─── API request ──────────────────────────────────────────────────────────────

/**
 * Perform an authenticated GET request to the Git API.
 *
 * @return array|WP_Error  Decoded JSON body or WP_Error.
 */
function oribi_sync_api_get( string $url, string $provider, string $pat ) {
    $headers = array_merge(
        oribi_sync_auth_headers( $provider, $pat ),
        [
            'Accept'     => 'application/json',
            'User-Agent' => 'Oribi-Sync-WP/' . ORIBI_SYNC_VERSION,
        ]
    );

    // Provider-specific Accept overrides
    if ( $provider === 'github' ) {
        $headers['Accept'] = 'application/vnd.github+json';
    } elseif ( $provider === 'azure' ) {
        $headers['Accept'] = 'application/json';
    }

    $response = wp_remote_get( $url, [
        'timeout' => 30,
        'headers' => $headers,
    ] );

    if ( is_wp_error( $response ) ) {
        return $response;
    }

    $code = wp_remote_retrieve_response_code( $response );
    $body = wp_remote_retrieve_body( $response );

    if ( $code < 200 || $code >= 300 ) {
        return new WP_Error(
            'oribi_sync_api_error',
            sprintf( 'Git API returned HTTP %d: %s', $code, wp_trim_words( $body, 30, '…' ) )
        );
    }

    $decoded = json_decode( $body, true );
    if ( json_last_error() !== JSON_ERROR_NONE ) {
        return new WP_Error( 'oribi_sync_json_error', 'Failed to parse API JSON response.' );
    }

    return $decoded;
}

// ─── Tree fetching ────────────────────────────────────────────────────────────

/**
 * Fetch the repository tree (recursive) for the given branch.
 *
 * Returns a flat list of file entries with 'path' and 'type'.
 *
 * @return array|WP_Error
 */
function oribi_sync_fetch_tree( string $api_base, string $branch, string $provider, string $pat ) {
    switch ( $provider ) {

        // ── GitLab ──────────────────────────────────────────────────────
        case 'gitlab':
            $entries = [];
            $page    = 1;
            do {
                $url = $api_base . '/repository/tree?' . http_build_query( [
                    'ref'       => $branch,
                    'recursive' => 'true',
                    'per_page'  => 100,
                    'page'      => $page,
                ] );
                $result = oribi_sync_api_get( $url, $provider, $pat );
                if ( is_wp_error( $result ) ) return $result;
                if ( empty( $result ) ) break;

                foreach ( $result as $item ) {
                    $entries[] = [
                        'path' => $item['path'],
                        'type' => $item['type'] === 'blob' ? 'blob' : $item['type'],
                        'sha'  => $item['id'] ?? '',  // GitLab uses 'id' for blob SHA
                    ];
                }
                $page++;
            } while ( count( $result ) === 100 );
            return $entries;

        // ── Bitbucket Cloud ─────────────────────────────────────────────
        case 'bitbucket':
            $entries = [];
            $url     = $api_base . '/src/' . rawurlencode( $branch ) . '/?pagelen=100&max_depth=10';
            // Bitbucket returns directory listings; we recurse via 'next' links
            while ( $url ) {
                $result = oribi_sync_api_get( $url, $provider, $pat );
                if ( is_wp_error( $result ) ) return $result;

                foreach ( $result['values'] ?? [] as $item ) {
                    if ( ( $item['type'] ?? '' ) === 'commit_file' ) {
                        $entries[] = [ 'path' => $item['path'], 'type' => 'blob' ];
                    }
                }
                $url = $result['next'] ?? null;
            }
            return $entries;

        // ── Azure DevOps ───────────────────────────────────────────────
        case 'azure':
            $url = $api_base . '/items?' . http_build_query( [
                'recursionLevel'   => 'full',
                'versionDescriptor.version'     => $branch,
                'versionDescriptor.versionType' => 'branch',
                'api-version'      => '7.0',
            ] );
            $result = oribi_sync_api_get( $url, $provider, $pat );
            if ( is_wp_error( $result ) ) return $result;

            $entries = [];
            foreach ( $result['value'] ?? [] as $item ) {
                if ( ! $item['isFolder'] ) {
                    // Azure paths start with '/' — strip leading slash
                    $path = ltrim( $item['path'], '/' );
                    $entries[] = [ 'path' => $path, 'type' => 'blob' ];
                }
            }
            return $entries;

        // ── Gitea / Forgejo ─────────────────────────────────────────────
        case 'gitea':
            $url = $api_base . '/git/trees/' . rawurlencode( $branch ) . '?recursive=true';
            $result = oribi_sync_api_get( $url, $provider, $pat );
            if ( is_wp_error( $result ) ) return $result;

            if ( ! isset( $result['tree'] ) ) {
                return new WP_Error( 'oribi_sync_tree_error', 'Unexpected tree response from Gitea.' );
            }
            return array_map( function ( $item ) {
                return [ 'path' => $item['path'], 'type' => $item['type'], 'sha' => $item['sha'] ?? '' ];
            }, $result['tree'] );

        // ── GitHub (default) ───────────────────────────────────────────
        case 'github':
        default:
            $url = $api_base . '/git/trees/' . rawurlencode( $branch ) . '?recursive=1';
            $result = oribi_sync_api_get( $url, $provider, $pat );
            if ( is_wp_error( $result ) ) return $result;

            if ( ! isset( $result['tree'] ) ) {
                return new WP_Error( 'oribi_sync_tree_error', 'Unexpected tree response structure.' );
            }
            return array_map( function ( $item ) {
                return [ 'path' => $item['path'], 'type' => $item['type'], 'sha' => $item['sha'] ?? '' ];
            }, $result['tree'] );
    }
}

// ─── File fetching ────────────────────────────────────────────────────────────

/**
 * Fetch raw file content from the repository.
 *
 * @return string|WP_Error  Raw file content.
 */
function oribi_sync_fetch_file( string $api_base, string $branch, string $file_path, string $provider, string $pat ) {
    $encoded_path = implode( '/', array_map( 'rawurlencode', explode( '/', $file_path ) ) );

    switch ( $provider ) {
        case 'gitlab':
            $url    = $api_base . '/repository/files/' . rawurlencode( $file_path ) . '/raw?' . http_build_query( [ 'ref' => $branch ] );
            $accept = 'text/plain';
            break;

        case 'bitbucket':
            $url    = $api_base . '/src/' . rawurlencode( $branch ) . '/' . $encoded_path;
            $accept = 'text/plain';
            break;

        case 'azure':
            $url = $api_base . '/items?' . http_build_query( [
                'path'                          => '/' . $file_path,
                'versionDescriptor.version'     => $branch,
                'versionDescriptor.versionType' => 'branch',
                'api-version'                   => '7.0',
                '\$format'                       => 'octetStream',
            ] );
            $accept = 'application/octet-stream';
            break;

        case 'gitea':
            $url    = $api_base . '/raw/' . $encoded_path . '?ref=' . rawurlencode( $branch );
            $accept = 'text/plain';
            break;

        case 'github':
        default:
            $url    = $api_base . '/contents/' . $encoded_path . '?ref=' . rawurlencode( $branch );
            $accept = 'application/vnd.github.raw+json';
            break;
    }

    $headers = array_merge(
        oribi_sync_auth_headers( $provider, $pat ),
        [
            'Accept'     => $accept,
            'User-Agent' => 'Oribi-Sync-WP/' . ORIBI_SYNC_VERSION,
        ]
    );

    $response = wp_remote_get( $url, [
        'timeout' => 30,
        'headers' => $headers,
    ] );

    if ( is_wp_error( $response ) ) {
        return $response;
    }

    $code = wp_remote_retrieve_response_code( $response );
    if ( $code < 200 || $code >= 300 ) {
        return new WP_Error(
            'oribi_sync_file_error',
            sprintf( 'Failed to fetch %s (HTTP %d)', $file_path, $code )
        );
    }

    return wp_remote_retrieve_body( $response );
}

// ─── Tree filtering ───────────────────────────────────────────────────────────

/**
 * Filter tree entries to only those under a given directory prefix.
 *
 * @param array  $tree      Tree from oribi_sync_fetch_tree().
 * @param string $prefix    Directory prefix (e.g. 'pages/').
 * @param bool   $recursive Whether to include files in subdirectories (default: false).
 *
 * @return array  Filtered entries (blobs only).
 */
function oribi_sync_filter_tree( array $tree, string $prefix, bool $recursive = false ): array {
    $prefix = rtrim( $prefix, '/' ) . '/';
    $out    = [];

    foreach ( $tree as $entry ) {
        if ( $entry['type'] !== 'blob' ) continue;
        if ( strpos( $entry['path'], $prefix ) !== 0 ) continue;
        $relative = substr( $entry['path'], strlen( $prefix ) );
        // Skip sub-directory files unless recursive is enabled
        if ( ! $recursive && strpos( $relative, '/' ) !== false ) continue;
        $out[] = $entry;
    }

    return $out;
}