driftwood/src/core/github_enrichment.rs

use super::database::Database;

// --- API response structs ---

#[derive(Debug, serde::Deserialize)]
pub struct GitHubRepoInfo {
    pub stargazers_count: i64,
    pub pushed_at: Option<String>,
    pub description: Option<String>,
}

#[derive(Debug, serde::Deserialize)]
pub struct GitHubReleaseInfo {
    pub tag_name: String,
    pub published_at: Option<String>,
    pub assets: Vec<GitHubReleaseAsset>,
}

#[derive(Debug, serde::Deserialize)]
pub struct GitHubReleaseAsset {
    pub name: String,
    pub browser_download_url: String,
    pub download_count: i64,
    pub size: i64,
}

// --- URL parsing ---

/// Extract (owner, repo) from a GitHub URL.
/// Tries download_url first (most reliable for GitHub releases), then homepage.
pub fn extract_github_repo(homepage: Option<&str>, download_url: &str) -> Option<(String, String)> {
    // Try download URL first - most AppImageHub entries point to GitHub releases
    if let Some(pair) = parse_github_url(download_url) {
        return Some(pair);
    }
    // Fallback to homepage
    if let Some(hp) = homepage {
        if let Some(pair) = parse_github_url(hp) {
            return Some(pair);
        }
    }
    None
}

/// Parse `github.com/{owner}/{repo}` from a URL, stripping .git suffix if present.
fn parse_github_url(url: &str) -> Option<(String, String)> {
    let stripped = url.trim_start_matches("https://")
        .trim_start_matches("http://");

    if !stripped.starts_with("github.com/") {
        return None;
    }

    let path = stripped.strip_prefix("github.com/")?;
    let parts: Vec<&str> = path.splitn(3, '/').collect();
    if parts.len() < 2 {
        return None;
    }

    let owner = parts[0];
    let repo = parts[1]
        .trim_end_matches(".git")
        .split('?').next().unwrap_or(parts[1]);

    if owner.is_empty() || repo.is_empty() {
        return None;
    }

    Some((owner.to_string(), repo.to_string()))
}

// --- API calls ---

fn github_get(url: &str, token: &str) -> Result<(String, u32), String> {
    let mut req = ureq::get(url)
        .header("Accept", "application/vnd.github+json")
        .header("User-Agent", "Driftwood-AppImage-Manager");
    if !token.is_empty() {
        req = req.header("Authorization", &format!("Bearer {}", token));
    }
    let mut response = req.call()
        .map_err(|e| format!("GitHub API error: {}", e))?;

    // Parse rate limit header
    let remaining: u32 = response.headers()
        .get("x-ratelimit-remaining")
        .and_then(|v| v.to_str().ok())
        .and_then(|v| v.parse().ok())
        .unwrap_or(0);

    let body = response.body_mut().read_to_string()
        .map_err(|e| format!("Read error: {}", e))?;

    Ok((body, remaining))
}

pub fn fetch_repo_info(owner: &str, repo: &str, token: &str) -> Result<(GitHubRepoInfo, u32), String> {
    let url = format!("https://api.github.com/repos/{}/{}", owner, repo);
    let (body, remaining) = github_get(&url, token)?;
    let info: GitHubRepoInfo = serde_json::from_str(&body)
        .map_err(|e| format!("Parse error: {}", e))?;
    Ok((info, remaining))
}

pub fn fetch_release_info(owner: &str, repo: &str, token: &str) -> Result<(GitHubReleaseInfo, u32), String> {
    let url = format!("https://api.github.com/repos/{}/{}/releases/latest", owner, repo);
    let (body, remaining) = github_get(&url, token)?;
    let info: GitHubReleaseInfo = serde_json::from_str(&body)
        .map_err(|e| format!("Parse error: {}", e))?;
    Ok((info, remaining))
}

#[derive(Debug, serde::Deserialize)]
struct GitHubReadmeResponse {
    content: String,
    #[serde(default)]
    encoding: String,
}

/// Fetch the README content for a repo (decoded from base64).
pub fn fetch_readme(owner: &str, repo: &str, token: &str) -> Result<(String, u32), String> {
    let url = format!("https://api.github.com/repos/{}/{}/readme", owner, repo);
    let (body, remaining) = github_get(&url, token)?;
    let resp: GitHubReadmeResponse = serde_json::from_str(&body)
        .map_err(|e| format!("Parse error: {}", e))?;

    if resp.encoding != "base64" {
        return Err(format!("Unexpected encoding: {}", resp.encoding));
    }

    // GitHub returns base64 with newlines; strip them before decoding
    let clean = resp.content.replace('\n', "");
    let decoded = base64_decode(&clean)
        .map_err(|e| format!("Base64 decode error: {}", e))?;
    let text = String::from_utf8(decoded)
        .map_err(|e| format!("UTF-8 error: {}", e))?;
    Ok((text, remaining))
}

/// Simple base64 decoder (standard alphabet, no padding required).
fn base64_decode(input: &str) -> Result<Vec<u8>, String> {
    const TABLE: &[u8; 64] = b"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
    let mut output = Vec::with_capacity(input.len() * 3 / 4);
    let mut buf = 0u32;
    let mut bits = 0u32;
    for &b in input.as_bytes() {
        if b == b'=' { break; }
        let val = TABLE.iter().position(|&c| c == b)
            .ok_or_else(|| format!("Invalid base64 char: {}", b as char))? as u32;
        buf = (buf << 6) | val;
        bits += 6;
        if bits >= 8 {
            bits -= 8;
            output.push((buf >> bits) as u8);
            buf &= (1 << bits) - 1;
        }
    }
    Ok(output)
}

// --- AppImage asset filtering ---

/// A simplified release asset for storage (JSON-serializable).
#[derive(Debug, Clone, serde::Serialize, serde::Deserialize)]
pub struct AppImageAsset {
    pub name: String,
    pub url: String,
    pub size: i64,
}

/// Filter release assets to only AppImage files.
pub fn filter_appimage_assets(assets: &[GitHubReleaseAsset]) -> Vec<AppImageAsset> {
    assets.iter()
        .filter(|a| {
            let lower = a.name.to_lowercase();
            lower.ends_with(".appimage") || lower.ends_with(".appimage.zsync")
        })
        .filter(|a| !a.name.to_lowercase().ends_with(".zsync"))
        .map(|a| AppImageAsset {
            name: a.name.clone(),
            url: a.browser_download_url.clone(),
            size: a.size,
        })
        .collect()
}

/// Detect the current system architecture string as used in AppImage filenames.
pub fn detect_arch() -> &'static str {
    #[cfg(target_arch = "x86_64")]
    { "x86_64" }
    #[cfg(target_arch = "aarch64")]
    { "aarch64" }
    #[cfg(not(any(target_arch = "x86_64", target_arch = "aarch64")))]
    { std::env::consts::ARCH }
}

/// Pick the best AppImage asset for the current architecture.
/// Returns the matching asset, or the first one if no arch match.
pub fn pick_best_asset(assets: &[AppImageAsset]) -> Option<&AppImageAsset> {
    if assets.is_empty() {
        return None;
    }
    let arch = detect_arch();
    // Prefer exact arch match in filename
    let arch_match = assets.iter().find(|a| {
        let lower = a.name.to_lowercase();
        lower.contains(&arch.to_lowercase())
    });
    arch_match.or(assets.first())
}

// --- Enrichment logic ---

/// Enrich a catalog app with repo-level info (stars, pushed_at, description).
pub fn enrich_app_repo_info(
    db: &Database,
    app_id: i64,
    owner: &str,
    repo: &str,
    token: &str,
) -> Result<u32, String> {
    let (info, remaining) = fetch_repo_info(owner, repo, token)?;
    db.update_catalog_app_github_metadata(
        app_id, info.stargazers_count, info.pushed_at.as_deref(), info.description.as_deref(),
    ).map_err(|e| format!("DB error: {}", e))?;
    Ok(remaining)
}

/// Enrich a catalog app with release info (version, date, downloads, assets).
pub fn enrich_app_release_info(
    db: &Database,
    app_id: i64,
    owner: &str,
    repo: &str,
    token: &str,
) -> Result<u32, String> {
    let (info, remaining) = fetch_release_info(owner, repo, token)?;

    // Clean version string (strip leading "v")
    let version = info.tag_name.strip_prefix('v')
        .unwrap_or(&info.tag_name)
        .to_string();

    // Sum download counts across all assets
    let total_downloads: i64 = info.assets.iter().map(|a| a.download_count).sum();

    // Extract AppImage assets and pick the best download URL
    let appimage_assets = filter_appimage_assets(&info.assets);
    let best_url = pick_best_asset(&appimage_assets).map(|a| a.url.as_str());
    let assets_json = if appimage_assets.is_empty() {
        None
    } else {
        serde_json::to_string(&appimage_assets).ok()
    };

    db.update_catalog_app_release_info(
        app_id,
        Some(&version),
        info.published_at.as_deref(),
        if total_downloads > 0 { Some(total_downloads) } else { None },
        best_url,
        assets_json.as_deref(),
    ).map_err(|e| format!("DB error: {}", e))?;

    Ok(remaining)
}

/// A GitHub release with body text for changelog display.
#[derive(Debug, serde::Deserialize)]
struct GitHubRelease {
    tag_name: String,
    published_at: Option<String>,
    body: Option<String>,
}

/// Fetch up to 10 recent releases for a repo.
fn fetch_recent_releases(owner: &str, repo: &str, token: &str) -> Result<(Vec<GitHubRelease>, u32), String> {
    let url = format!("https://api.github.com/repos/{}/{}/releases?per_page=10", owner, repo);
    let (body, remaining) = github_get(&url, token)?;
    let releases: Vec<GitHubRelease> = serde_json::from_str(&body)
        .map_err(|e| format!("Parse error: {}", e))?;
    Ok((releases, remaining))
}

/// Enrich a catalog app with release history (version, date, description for last 10 releases).
/// Only populates if the existing release_history is empty.
pub fn enrich_app_release_history(
    db: &Database,
    app_id: i64,
    owner: &str,
    repo: &str,
    token: &str,
) -> Result<u32, String> {
    // Check if release_history is already populated (from AppStream or prior enrichment)
    if let Ok(Some(app)) = db.get_catalog_app(app_id) {
        if app.release_history.as_ref().is_some_and(|h| !h.is_empty()) {
            return Ok(u32::MAX); // already has data, skip
        }
    }

    let (releases, remaining) = fetch_recent_releases(owner, repo, token)?;
    if releases.is_empty() {
        return Ok(remaining);
    }

    // Convert to the same JSON format used by AppStream: [{version, date, description}]
    let history: Vec<serde_json::Value> = releases.iter().map(|r| {
        let version = r.tag_name.strip_prefix('v')
            .unwrap_or(&r.tag_name)
            .to_string();
        let date = r.published_at.as_deref()
            .and_then(|d| d.split('T').next())
            .unwrap_or("");
        let mut obj = serde_json::json!({
            "version": version,
            "date": date,
        });
        if let Some(ref body) = r.body {
            if !body.is_empty() {
                obj["description"] = serde_json::Value::String(body.clone());
            }
        }
        obj
    }).collect();

    let json = serde_json::to_string(&history)
        .map_err(|e| format!("JSON error: {}", e))?;
    db.update_catalog_app_release_history(app_id, &json)
        .map_err(|e| format!("DB error: {}", e))?;

    Ok(remaining)
}

/// Fetch and store the README for a catalog app.
pub fn enrich_app_readme(
    db: &Database,
    app_id: i64,
    owner: &str,
    repo: &str,
    token: &str,
) -> Result<u32, String> {
    let (readme, remaining) = fetch_readme(owner, repo, token)?;
    db.update_catalog_app_readme(app_id, &readme)
        .map_err(|e| format!("DB error: {}", e))?;
    Ok(remaining)
}

/// Background enrichment: process a batch of unenriched apps.
/// Returns (count_enriched, should_continue).
pub fn background_enrich_batch(
    db: &Database,
    token: &str,
    batch_size: i32,
    on_progress: &dyn Fn(i64, i64),
) -> Result<(u32, bool), String> {
    let apps = db.get_unenriched_catalog_apps(batch_size)
        .map_err(|e| format!("DB error: {}", e))?;

    if apps.is_empty() {
        return Ok((0, false));
    }

    let mut enriched = 0u32;

    for app in &apps {
        let owner = match app.github_owner.as_deref() {
            Some(o) => o,
            None => continue,
        };
        let repo = match app.github_repo.as_deref() {
            Some(r) => r,
            None => continue,
        };

        match enrich_app_repo_info(db, app.id, owner, repo, token) {
            Ok(remaining) => {
                enriched += 1;

                // Also fetch release history (changelog data)
                let _ = enrich_app_release_history(db, app.id, owner, repo, token);

                // Report progress
                if let Ok((done, total)) = db.catalog_enrichment_progress() {
                    on_progress(done, total);
                }

                // Stop if rate limit is getting low
                if remaining < 5 {
                    log::info!("GitHub rate limit low ({}), pausing enrichment", remaining);
                    return Ok((enriched, false));
                }
            }
            Err(e) => {
                log::warn!("Failed to enrich {}/{}: {}", owner, repo, e);
                // Mark as enriched anyway so we don't retry forever
                db.update_catalog_app_github_metadata(app.id, 0, None, None).ok();
            }
        }

        // Sleep between calls to be respectful
        std::thread::sleep(std::time::Duration::from_secs(1));
    }

    Ok((enriched, enriched > 0))
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_extract_github_repo_from_download() {
        let result = extract_github_repo(
            None,
            "https://github.com/nickvdp/deno-spreadsheets/releases/download/v0.3.0/app.AppImage",
        );
        assert_eq!(result, Some(("nickvdp".to_string(), "deno-spreadsheets".to_string())));
    }

    #[test]
    fn test_extract_github_repo_from_homepage() {
        let result = extract_github_repo(
            Some("https://github.com/nickvdp/deno-spreadsheets"),
            "https://example.com/download.AppImage",
        );
        assert_eq!(result, Some(("nickvdp".to_string(), "deno-spreadsheets".to_string())));
    }

    #[test]
    fn test_extract_github_repo_with_git_suffix() {
        let result = extract_github_repo(
            Some("https://github.com/user/repo.git"),
            "https://example.com/download.AppImage",
        );
        assert_eq!(result, Some(("user".to_string(), "repo".to_string())));
    }

    #[test]
    fn test_extract_github_repo_non_github() {
        let result = extract_github_repo(
            Some("https://gitlab.com/user/repo"),
            "https://sourceforge.net/download.AppImage",
        );
        assert_eq!(result, None);
    }

    #[test]
    fn test_parse_github_url_empty() {
        assert_eq!(parse_github_url(""), None);
        assert_eq!(parse_github_url("https://github.com/"), None);
        assert_eq!(parse_github_url("https://github.com/user"), None);
    }

    #[test]
    fn test_filter_appimage_assets() {
        let assets = vec![
            GitHubReleaseAsset {
                name: "app-x86_64.AppImage".to_string(),
                browser_download_url: "https://github.com/u/r/releases/download/v1/app-x86_64.AppImage".to_string(),
                download_count: 100,
                size: 50_000_000,
            },
            GitHubReleaseAsset {
                name: "app-aarch64.AppImage".to_string(),
                browser_download_url: "https://github.com/u/r/releases/download/v1/app-aarch64.AppImage".to_string(),
                download_count: 20,
                size: 48_000_000,
            },
            GitHubReleaseAsset {
                name: "app-x86_64.AppImage.zsync".to_string(),
                browser_download_url: "https://github.com/u/r/releases/download/v1/app-x86_64.AppImage.zsync".to_string(),
                download_count: 5,
                size: 1000,
            },
            GitHubReleaseAsset {
                name: "source.tar.gz".to_string(),
                browser_download_url: "https://github.com/u/r/releases/download/v1/source.tar.gz".to_string(),
                download_count: 10,
                size: 2_000_000,
            },
        ];
        let filtered = filter_appimage_assets(&assets);
        assert_eq!(filtered.len(), 2);
        assert_eq!(filtered[0].name, "app-x86_64.AppImage");
        assert_eq!(filtered[1].name, "app-aarch64.AppImage");
    }

    #[test]
    fn test_pick_best_asset_prefers_arch() {
        let assets = vec![
            AppImageAsset {
                name: "app-aarch64.AppImage".to_string(),
                url: "https://example.com/aarch64".to_string(),
                size: 48_000_000,
            },
            AppImageAsset {
                name: "app-x86_64.AppImage".to_string(),
                url: "https://example.com/x86_64".to_string(),
                size: 50_000_000,
            },
        ];
        let best = pick_best_asset(&assets).unwrap();
        // On x86_64 systems this should pick x86_64, on aarch64 it picks aarch64
        let arch = detect_arch();
        assert!(best.name.contains(arch));
    }

    #[test]
    fn test_pick_best_asset_empty() {
        let assets: Vec<AppImageAsset> = vec![];
        assert!(pick_best_asset(&assets).is_none());
    }

    #[test]
    fn test_pick_best_asset_single() {
        let assets = vec![
            AppImageAsset {
                name: "app.AppImage".to_string(),
                url: "https://example.com/app".to_string(),
                size: 50_000_000,
            },
        ];
        let best = pick_best_asset(&assets).unwrap();
        assert_eq!(best.name, "app.AppImage");
    }
}