strawcore/examples/emit_json.rs

// Differential-verification JSON emitter (Phase-1 harness, NOT shipped).
//
// Additive example only — adds no production code. Emits strawcore's
// extraction for one video id as the harness's normalized JSON schema on
// stdout, so it can be diffed against the upstream NewPipe-Extractor (Java)
// oracle. See /root/build/npe-diff/ for the driver + differ.
//
//   cargo run --example emit_json -- <VIDEO_ID> [visionos-on]
//
// Default matches the shipped app path (ExtractOptions::default(), visionOS
// OFF). Pass `visionos-on` to fetch the visionOS client (client 101) — that
// is what upstream NPE does UNCONDITIONALLY, so `visionos-on` is the
// apples-to-apples mode for an NPE fidelity diff.
use std::sync::Arc;

use serde_json::{json, Value};
use strawcore_core::stream::{AudioStream, DeliveryMethod, StreamInfo, VideoStream};
use strawcore_core::youtube::stream_extractor::{stream_info_with, ExtractOptions};
use strawcore_core::NewPipe;

/// Parse the innertube client tag (`&c=WEB` / `&c=ANDROID` / ...) off a
/// googlevideo URL — this is the "which client sourced this stream" signal
/// that made the visionOS outage visible.
fn client_source(url: &str) -> Value {
    url.split(['&', '?'])
        .find_map(|kv| kv.strip_prefix("c="))
        .map(|c| Value::String(c.to_string()))
        .unwrap_or(Value::Null)
}

fn delivery_str(d: DeliveryMethod) -> &'static str {
    match d {
        DeliveryMethod::Progressive => "progressive",
        DeliveryMethod::Dash => "dash",
        DeliveryMethod::Hls => "hls",
        DeliveryMethod::Torrent => "torrent",
    }
}

fn video_json(s: &VideoStream, kind: &str) -> Value {
    json!({
        "kind": kind,
        "itag": s.itag,
        "delivery": delivery_str(s.delivery),
        "format": s.format.extension(),
        "mime": s.format.mime(),
        "codec": s.codec,
        "resolution": if s.resolution.is_empty() { Value::Null } else { json!(s.resolution) },
        "fps": s.fps,
        "bitrate": s.bandwidth,
        "has_url": !s.url.is_empty(),
        "client_source": client_source(&s.url),
    })
}

fn audio_json(s: &AudioStream) -> Value {
    json!({
        "kind": "audio",
        "itag": s.itag,
        "delivery": delivery_str(s.delivery),
        "format": s.format.extension(),
        "mime": s.format.mime(),
        "codec": s.codec,
        "resolution": Value::Null,
        "fps": 0,
        "bitrate": s.average_bitrate_kbps,
        "has_url": !s.url.is_empty(),
        "client_source": client_source(&s.url),
    })
}

fn emit(info: &StreamInfo) -> Value {
    let mut streams: Vec<Value> = Vec::new();
    for s in &info.video_streams {
        streams.push(video_json(s, "muxed"));
    }
    for s in &info.video_only_streams {
        streams.push(video_json(s, "video_only"));
    }
    for s in &info.audio_streams {
        streams.push(audio_json(s));
    }
    // Deterministic order so diffs are stable: (kind, itag, delivery).
    streams.sort_by(|a, b| {
        (
            a["kind"].as_str().unwrap_or(""),
            a["itag"].as_u64().unwrap_or(0),
            a["delivery"].as_str().unwrap_or(""),
        )
            .cmp(&(
                b["kind"].as_str().unwrap_or(""),
                b["itag"].as_u64().unwrap_or(0),
                b["delivery"].as_str().unwrap_or(""),
            ))
    });

    json!({
        "source": "strawcore",
        "video_id": info.video_id,
        "meta": {
            "name": info.name,
            "duration_seconds": info.duration_seconds,
            "stream_type": format!("{:?}", info.stream_type),
            "uploader_id": info.uploader_id,
            "uploader_name": info.uploader_name,
            "view_count": info.view_count,
        },
        "dash_present": info.dash_manifest_url.is_some(),
        "hls_present": info.hls_manifest_url.is_some(),
        "counts": {
            "muxed": info.video_streams.len(),
            "video_only": info.video_only_streams.len(),
            "audio": info.audio_streams.len(),
        },
        "streams": streams,
    })
}

fn main() {
    let mut args = std::env::args().skip(1);
    let vid = args.next().unwrap_or_else(|| "dQw4w9WgXcQ".into());
    let visionos = args.next().as_deref() == Some("visionos-on");

    let dl =
        Arc::new(strawcore_core::downloader::ReqwestDownloader::new().expect("build downloader"));
    NewPipe::init(dl);

    let opts = ExtractOptions {
        fetch_visionos_client: visionos,
        ..Default::default()
    };
    match stream_info_with(&vid, opts) {
        Ok(info) => {
            println!("{}", serde_json::to_string_pretty(&emit(&info)).unwrap());
        }
        Err(e) => {
            eprintln!("strawcore extraction FAILED for {vid}: {e:?}");
            std::process::exit(2);
        }
    }
}