torttube/sidecar/crates/torttube-sidecar/src/resolve.rs

// resolve.rs — Tier 1 (rustypipe) → Tier 2 (yt-dlp -j fallback)
// SPDX-License-Identifier: GPL-3.0-or-later

use serde_json::Value;

use crate::{run_yt_dlp, HandlerError};

/// Playback-ready single-URL resolve. Asks yt-dlp for `best[ext=mp4]/best` —
/// a combined audio+video format that Kodi can play as a plain HTTP URL.
/// Slower than `resolve()` (~3-5s) but guarantees a working stream.
pub(crate) async fn resolve_play(id: &str) -> Result<Value, HandlerError> {
    let url = format!("https://www.youtube.com/watch?v={id}");
    // -f best[ext=mp4]/best — prefer mp4 progressive, else any best combined.
    // We use -j to get the full info dump; the selected format's URL appears
    // as the top-level "url" field.
    let stdout = run_yt_dlp(&[
        "-j", "--no-warnings", "--no-playlist",
        "-f", "best[ext=mp4]/best",
        &url,
    ])
    .await
    .map_err(|e| classify_yt_dlp_error(&e))?;

    let dump: Value = serde_json::from_slice(&stdout)
        .map_err(|e| HandlerError::Extractor(format!("yt-dlp json parse: {e}")))?;

    let stream_url = dump
        .get("url")
        .and_then(Value::as_str)
        .ok_or_else(|| HandlerError::Extractor("yt-dlp: no top-level url".into()))?
        .to_string();

    tracing::info!(id, "resolve_play ok via yt-dlp combined");

    Ok(serde_json::json!({
        "source": "yt-dlp",
        "stream_url": stream_url,
        "title": dump.get("title"),
        "duration_s": dump.get("duration"),
        "channel_name": dump.get("channel"),
        "channel_id": dump.get("channel_id"),
        "thumbnail": dump.get("thumbnail"),
        "format_id": dump.get("format_id"),
        "ext": dump.get("ext"),
    }))
}

/// Metadata-rich resolve. Tries Tier 1 (rustypipe), falls back to Tier 2 (yt-dlp -j).
/// Returns the full extractor response including separate audio + video streams.
/// Use `resolve_play` instead for direct playback (returns a single combined URL).
pub(crate) async fn resolve(id: &str) -> Result<Value, HandlerError> {
    match tier1_rustypipe(id).await {
        Ok(v) => {
            tracing::info!(id, source = "rustypipe", "resolve ok");
            Ok(v)
        }
        Err(e) => {
            tracing::warn!(id, error = %e, "rustypipe failed; falling back to yt-dlp");
            // Typed errors that mean "video can't be played by anyone" — don't retry yt-dlp,
            // it'll just hit the same wall.
            if matches!(
                e,
                HandlerError::AgeRestricted
                    | HandlerError::PrivateVideo
                    | HandlerError::NotFound
            ) {
                return Err(e);
            }
            tier2_yt_dlp(id).await
        }
    }
}

/// Tier 1 — native rustypipe. Serializes the whole player.details + selected streams as
/// opaque pass-through JSON. The Python addon parses the fields it needs; this keeps us
/// resilient to rustypipe shape evolution and unblocks tier-2 normalization later.
async fn tier1_rustypipe(id: &str) -> Result<Value, HandlerError> {
    use rustypipe::client::RustyPipe;
    use rustypipe::param::StreamFilter;

    let rp = RustyPipe::new();
    let player = rp
        .query()
        .player(id)
        .await
        .map_err(|e| classify_rustypipe_error(&e))?;

    let (video, audio) = player.select_video_audio_stream(&StreamFilter::default());

    let details_json = serde_json::to_value(&player.details)
        .map_err(|e| HandlerError::Internal(format!("serialize details: {e}")))?;
    let video_json = video
        .map(|v| serde_json::to_value(v))
        .transpose()
        .map_err(|e| HandlerError::Internal(format!("serialize video: {e}")))?
        .unwrap_or(Value::Null);
    let audio_json = audio
        .map(|a| serde_json::to_value(a))
        .transpose()
        .map_err(|e| HandlerError::Internal(format!("serialize audio: {e}")))?
        .unwrap_or(Value::Null);

    Ok(serde_json::json!({
        "source": "rustypipe",
        "details": details_json,
        "video_stream": video_json,
        "audio_stream": audio_json,
    }))
}

/// Classify a yt-dlp shell-out error into one of our typed handler errors.
/// yt-dlp's stderr is freeform English; we match on substrings, case-insensitive
/// via a lowercase copy, but preserve the original message in the returned error.
fn classify_yt_dlp_error(e: &anyhow::Error) -> HandlerError {
    let original = e.to_string();
    let lower = original.to_lowercase();
    if lower.contains("age") {
        HandlerError::AgeRestricted
    } else if lower.contains("private") {
        HandlerError::PrivateVideo
    } else if lower.contains("not available") || lower.contains("does not exist") {
        HandlerError::NotFound
    } else if lower.contains("geo") || lower.contains("region") {
        HandlerError::RegionBlocked
    } else {
        HandlerError::Extractor(original)
    }
}

/// Classify a rustypipe error into one of our typed handler errors.
/// rustypipe's error enum varies by version; we match on the Display string for resilience.
fn classify_rustypipe_error(e: &dyn std::fmt::Display) -> HandlerError {
    let msg = e.to_string().to_lowercase();
    if msg.contains("age") && msg.contains("restrict") {
        HandlerError::AgeRestricted
    } else if msg.contains("region") || msg.contains("country") || msg.contains("geo") {
        HandlerError::RegionBlocked
    } else if msg.contains("private") {
        HandlerError::PrivateVideo
    } else if msg.contains("not found") || msg.contains("unavailable") {
        HandlerError::NotFound
    } else if msg.contains("network") || msg.contains("timeout") || msg.contains("connect") {
        HandlerError::Network(msg)
    } else {
        HandlerError::Extractor(msg)
    }
}

/// Tier 2 — shell out to yt-dlp -j.
async fn tier2_yt_dlp(id: &str) -> Result<Value, HandlerError> {
    let url = format!("https://www.youtube.com/watch?v={id}");
    let stdout = run_yt_dlp(&["-j", "--no-warnings", "--no-playlist", &url])
        .await
        .map_err(|e| classify_yt_dlp_error(&e))?;

    let dump: Value = serde_json::from_slice(&stdout)
        .map_err(|e| HandlerError::Extractor(format!("yt-dlp json parse: {e}")))?;

    // yt-dlp's JSON has a `formats` array. We pass it through largely as-is — the addon
    // can pick what inputstream.adaptive wants. Shape it to match our protocol.
    let streams: Vec<Value> = dump
        .get("formats")
        .and_then(Value::as_array)
        .cloned()
        .unwrap_or_default()
        .into_iter()
        .filter_map(|f| {
            let url = f.get("url")?.as_str()?.to_string();
            let vcodec = f.get("vcodec").and_then(Value::as_str).unwrap_or("none");
            let acodec = f.get("acodec").and_then(Value::as_str).unwrap_or("none");
            let is_audio_only = vcodec == "none" && acodec != "none";
            let is_video_only = vcodec != "none" && acodec == "none";
            Some(serde_json::json!({
                "url": url,
                "itag": f.get("format_id").and_then(|v| v.as_str()).and_then(|s| s.parse::<u32>().ok()),
                "mime": f.get("ext"),
                "width": f.get("width"),
                "height": f.get("height"),
                "bitrate": f.get("tbr"),
                "is_audio_only": is_audio_only,
                "is_video_only": is_video_only,
            }))
        })
        .collect();

    Ok(serde_json::json!({
        "source": "yt-dlp",
        "title": dump.get("title"),
        "duration_s": dump.get("duration"),
        "channel_name": dump.get("channel"),
        "channel_id": dump.get("channel_id"),
        "streams": streams,
    }))
}