nab 0.8.2

//! Shared internal utility helpers for CLI and MCP surfaces.

use crate::{AcceleratedClient, CookieSource};

/// Extract the host/domain from a URL, returning an empty string on failure.
#[must_use]
pub fn extract_domain(url: &str) -> String {
    url::Url::parse(url)
        .ok()
        .and_then(|parsed| parsed.host_str().map(str::to_owned))
        .unwrap_or_default()
}

/// Build a `Referer` header value from a URL (scheme + host + `/`).
///
/// Returns `None` if the URL cannot be parsed.
#[must_use]
pub fn build_referer(url: &str) -> Option<String> {
    url::Url::parse(url)
        .ok()
        .map(|parsed| format!("{}://{}/", parsed.scheme(), parsed.host_str().unwrap_or("")))
}

/// Resolve a cookie header for a known domain from an optional browser name.
#[must_use]
pub fn resolve_cookie_header_for_domain(domain: &str, browser: Option<&str>) -> String {
    let Some(browser) = browser else {
        return String::new();
    };

    CookieSource::from_browser_name(browser)
        .get_cookie_header(domain)
        .unwrap_or_default()
}

/// Resolve a cookie header for a URL from an optional browser name.
#[must_use]
pub fn resolve_cookie_header_for_url(url: &str, browser: Option<&str>) -> String {
    let domain = extract_domain(url);
    resolve_cookie_header_for_domain(&domain, browser)
}

/// Attempt to recover article content from Next.js content chunks.
///
/// Called when the initial extraction yields thin content on a Next.js page
/// with `__NEXT_DATA__` containing only metadata. Makes up to 3 secondary HTTP
/// requests: webpack runtime, page component, and content chunk.
///
/// Returns `Some(markdown)` on success, `None` if recovery fails.
pub async fn recover_nextjs_chunks(
    client: &AcceleratedClient,
    html: &str,
    page_url: &str,
) -> Option<String> {
    use crate::content::spa_extract;

    if !spa_extract::is_nextjs_metadata_only(html) {
        return None;
    }

    let script_urls = spa_extract::discover_nextjs_content_chunks(html, page_url);
    if script_urls.len() < 2 {
        return None;
    }

    tracing::debug!("Attempting Next.js content chunk recovery");

    let (webpack_resp, page_resp) =
        tokio::join!(client.fetch(&script_urls[0]), client.fetch(&script_urls[1]),);

    let webpack_js = webpack_resp.ok()?.text().await.ok()?;
    let page_js = page_resp.ok()?.text().await.ok()?;

    let origin = url::Url::parse(page_url)
        .ok()
        .map(|parsed| parsed.origin().unicode_serialization())?;

    let slug = url::Url::parse(page_url).ok().and_then(|parsed| {
        parsed
            .path_segments()
            .and_then(|mut segments| segments.next_back().map(String::from))
    });
    let chunk_urls = spa_extract::resolve_content_chunk_urls_for_slug(
        &webpack_js,
        &page_js,
        &origin,
        slug.as_deref(),
    );
    if chunk_urls.is_empty() {
        tracing::debug!("No content chunk URLs resolved from webpack runtime");
        return None;
    }

    for chunk_url in &chunk_urls {
        tracing::debug!("Fetching content chunk: {chunk_url}");
        if let Ok(resp) = client.fetch(chunk_url).await
            && let Ok(chunk_js) = resp.text().await
            && let Some(content) = spa_extract::extract_jsx_text_content(&chunk_js)
        {
            tracing::info!(
                "Recovered {} chars from Next.js content chunk",
                content.len()
            );
            return Some(content);
        }
    }

    None
}

#[cfg(test)]
mod tests {
    use super::{
        build_referer, extract_domain, resolve_cookie_header_for_domain,
        resolve_cookie_header_for_url,
    };

    #[test]
    fn extract_domain_returns_host_or_empty_string() {
        assert_eq!(extract_domain("https://example.com/path"), "example.com");
        assert_eq!(extract_domain("not a url"), "");
    }

    #[test]
    fn build_referer_preserves_scheme_and_host() {
        assert_eq!(
            build_referer("https://example.com/path?x=1").as_deref(),
            Some("https://example.com/"),
        );
        assert_eq!(build_referer("not a url"), None);
    }

    #[test]
    fn cookie_helpers_return_empty_string_when_browser_is_absent() {
        assert_eq!(resolve_cookie_header_for_domain("example.com", None), "");
        assert_eq!(
            resolve_cookie_header_for_url("https://example.com/path", None),
            "",
        );
    }
}