Fixes for ytmusic search not working without ytmusicapi

scripts/ytmusic_search.py

@@ -0,0 +1,105 @@
+#!/usr/bin/env python3
+"""YouTube Music search helper for shanty-dl.
+
+Requires: pip install ytmusicapi
+
+Usage:
+    ytmusic_search.py search <query>           # general search
+    ytmusic_search.py artist <artist>          # bulk search by artist (up to 100 songs)
+    ytmusic_search.py track <artist> <title>   # search for a specific track
+
+Output: JSON array of results to stdout, each with:
+    - videoId: YouTube video ID
+    - title: track title
+    - artist: primary artist name
+    - album: album name (if available)
+    - duration_seconds: duration in seconds (if available)
+"""
+
+import json
+import sys
+
+from ytmusicapi import YTMusic
+
+
+def search_general(yt: YTMusic, query: str) -> list[dict]:
+    """Search YouTube Music for songs matching a general query."""
+    try:
+        results = yt.search(query, filter="songs", limit=5)
+    except Exception as e:
+        print(f"Search error: {e}", file=sys.stderr)
+        return []
+    return _format_results(results)
+
+
+def search_artist_bulk(yt: YTMusic, artist: str) -> list[dict]:
+    """Fetch up to 100 songs by an artist."""
+    try:
+        results = yt.search(artist, filter="songs", limit=100)
+    except Exception as e:
+        print(f"Search error: {e}", file=sys.stderr)
+        return []
+    return _format_results(results)
+
+
+def search_track(yt: YTMusic, artist: str, title: str) -> list[dict]:
+    """Search for a specific track by artist and title."""
+    try:
+        results = yt.search(f"{artist} {title}", filter="songs", limit=5)
+    except Exception as e:
+        print(f"Search error: {e}", file=sys.stderr)
+        return []
+    return _format_results(results)
+
+
+def _format_results(results: list[dict]) -> list[dict]:
+    """Convert ytmusicapi results to a simple JSON-friendly format."""
+    formatted = []
+    for r in results:
+        if not r.get("videoId"):
+            continue
+        artists = r.get("artists", [])
+        artist_name = artists[0]["name"] if artists else None
+        album = r.get("album")
+        album_name = album["name"] if isinstance(album, dict) else None
+        formatted.append({
+            "videoId": r["videoId"],
+            "title": r.get("title", ""),
+            "artist": artist_name,
+            "album": album_name,
+            "duration_seconds": r.get("duration_seconds"),
+        })
+    return formatted
+
+
+def main():
+    if len(sys.argv) < 3:
+        print("Usage: ytmusic_search.py <search|artist|track> <args...>", file=sys.stderr)
+        sys.exit(1)
+
+    yt = YTMusic()
+    command = sys.argv[1]
+
+    if command == "search":
+        query = " ".join(sys.argv[2:])
+        results = search_general(yt, query)
+    elif command == "artist":
+        artist = " ".join(sys.argv[2:])
+        results = search_artist_bulk(yt, artist)
+    elif command == "track":
+        if len(sys.argv) < 4:
+            print("Usage: ytmusic_search.py track <artist> <title>", file=sys.stderr)
+            sys.exit(1)
+        # Artist is argv[2], title is the rest
+        artist = sys.argv[2]
+        title = " ".join(sys.argv[3:])
+        results = search_track(yt, artist, title)
+    else:
+        print(f"Unknown command: {command}", file=sys.stderr)
+        sys.exit(1)
+
+    json.dump(results, sys.stdout)
+
+
+if __name__ == "__main__":
+    main()

src/ytdlp.rs

@@ -15,22 +15,6 @@ pub enum SearchSource {
     YouTube,
 }
 
-impl SearchSource {
-    fn prefix(&self) -> &str {
-        match self {
-            SearchSource::YouTubeMusic => "ytmusicsearch5",
-            SearchSource::YouTube => "ytsearch5",
-        }
-    }
-
-    fn source_name(&self) -> &str {
-        match self {
-            SearchSource::YouTubeMusic => "youtube_music",
-            SearchSource::YouTube => "youtube",
-        }
-    }
-}
-
 impl std::str::FromStr for SearchSource {
     type Err = String;
 
@@ -111,13 +95,53 @@ impl YtDlpBackend {
         Ok(String::from_utf8_lossy(&output.stdout).to_string())
     }
 
-    /// Search using a specific source prefix.
-    async fn search_with_source(
-        &self,
-        query: &str,
-        source: SearchSource,
-    ) -> DlResult<Vec<SearchResult>> {
-        let search_query = format!("{}:{}", source.prefix(), query);
+    /// Search via the ytmusicapi Python helper script.
+    async fn search_ytmusic(&self, query: &str) -> DlResult<Vec<SearchResult>> {
+        let script = self.find_ytmusic_script()?;
+
+        tracing::debug!(query = query, "searching YouTube Music via ytmusicapi");
+
+        let output = Command::new("python3")
+            .args([script.to_str().unwrap(), "search", query])
+            .stdout(Stdio::piped())
+            .stderr(Stdio::piped())
+            .output()
+            .await
+            .map_err(|e| {
+                if e.kind() == std::io::ErrorKind::NotFound {
+                    DlError::BackendNotFound(
+                        "python3 not found — required for YouTube Music search".into(),
+                    )
+                } else {
+                    DlError::Io(e)
+                }
+            })?;
+
+        if !output.status.success() {
+            let stderr = String::from_utf8_lossy(&output.stderr).to_string();
+            tracing::warn!(stderr = %stderr, "ytmusic search failed");
+            return Err(DlError::BackendError(format!("ytmusic search failed: {stderr}")));
+        }
+
+        let stdout = String::from_utf8_lossy(&output.stdout);
+        let entries: Vec<YtMusicSearchEntry> = serde_json::from_str(stdout.trim())
+            .map_err(|e| DlError::BackendError(format!("failed to parse ytmusic output: {e}")))?;
+
+        Ok(entries
+            .into_iter()
+            .map(|e| SearchResult {
+                url: format!("https://music.youtube.com/watch?v={}", e.video_id),
+                title: e.title,
+                artist: e.artist,
+                duration: e.duration_seconds,
+                source: "youtube_music".into(),
+            })
+            .collect())
+    }
+
+    /// Search using yt-dlp's ytsearch prefix.
+    async fn search_ytsearch(&self, query: &str) -> DlResult<Vec<SearchResult>> {
+        let search_query = format!("ytsearch5:{query}");
         let output = self
             .run_ytdlp(&[
                 "--dump-json",
@@ -139,7 +163,7 @@ impl YtDlpBackend {
                         title: entry.title.unwrap_or_else(|| "Unknown".into()),
                         artist: entry.artist.or(entry.uploader).or(entry.channel),
                         duration: entry.duration,
-                        source: source.source_name().into(),
+                        source: "youtube".into(),
                     });
                 }
                 Err(e) => {
@@ -150,6 +174,39 @@ impl YtDlpBackend {
 
         Ok(results)
     }
+
+    /// Locate the ytmusic_search.py helper script.
+    /// Checks: next to the binary, in the crate's scripts/ dir, or on PATH.
+    fn find_ytmusic_script(&self) -> DlResult<PathBuf> {
+        // Check next to the current executable
+        if let Ok(exe) = std::env::current_exe() {
+            let beside_exe = exe.parent().unwrap_or(std::path::Path::new(".")).join("ytmusic_search.py");
+            if beside_exe.exists() {
+                return Ok(beside_exe);
+            }
+        }
+
+        // Check common install locations
+        for dir in &[
+            "/usr/share/shanty",
+            "/usr/local/share/shanty",
+        ] {
+            let path = PathBuf::from(dir).join("ytmusic_search.py");
+            if path.exists() {
+                return Ok(path);
+            }
+        }
+
+        // Check relative to crate root (for development)
+        let dev_path = PathBuf::from(env!("CARGO_MANIFEST_DIR")).join("scripts/ytmusic_search.py");
+        if dev_path.exists() {
+            return Ok(dev_path);
+        }
+
+        Err(DlError::BackendNotFound(
+            "ytmusic_search.py not found — install it next to the shanty-dl binary or in /usr/share/shanty/".into(),
+        ))
+    }
 }
 
 impl DownloadBackend for YtDlpBackend {
@@ -161,7 +218,10 @@ impl DownloadBackend for YtDlpBackend {
 
     async fn search(&self, query: &str) -> DlResult<Vec<SearchResult>> {
         // Try primary search source
-        let results = self.search_with_source(query, self.search_source).await;
+        let results = match self.search_source {
+            SearchSource::YouTubeMusic => self.search_ytmusic(query).await,
+            SearchSource::YouTube => self.search_ytsearch(query).await,
+        };
 
         match results {
             Ok(ref r) if !r.is_empty() => return results,
@@ -173,19 +233,26 @@ impl DownloadBackend for YtDlpBackend {
             }
         }
 
-        // Fallback to the other source
-        if self.fallback_search && self.search_source == SearchSource::YouTubeMusic {
-            tracing::info!("falling back to YouTube search");
-            let fallback = self
-                .search_with_source(query, SearchSource::YouTube)
-                .await?;
-            if !fallback.is_empty() {
-                return Ok(fallback);
+        // Fallback: if ytmusic failed/empty, try ytsearch; and vice versa
+        if self.fallback_search {
+            let fallback_results = match self.search_source {
+                SearchSource::YouTubeMusic => {
+                    tracing::info!("falling back to YouTube search");
+                    self.search_ytsearch(query).await
+                }
+                SearchSource::YouTube => {
+                    tracing::info!("falling back to YouTube Music search");
+                    self.search_ytmusic(query).await
+                }
+            };
+            match fallback_results {
+                Ok(ref r) if !r.is_empty() => return fallback_results,
+                Ok(_) => tracing::debug!("fallback search also returned no results"),
+                Err(ref e) => tracing::debug!(error = %e, "fallback search also failed"),
             }
         }
 
-        // If primary returned Ok([]) and fallback also empty, return that
-        // If primary returned Err, propagate it
+        // Return the original result (Ok([]) or Err)
         results
     }
 
@@ -255,15 +322,17 @@ impl DownloadBackend for YtDlpBackend {
             DlError::BackendError(format!("failed to parse yt-dlp output: {e}"))
         })?;
 
-        // yt-dlp may change the extension after conversion
+        // --print-json reports the pre-extraction filename (e.g. .webm),
+        // but --extract-audio produces a file with the target format extension.
+        // Use requested_downloads if available, otherwise fix the extension.
         let file_path = if let Some(ref requested) = info.requested_downloads {
             if let Some(first) = requested.first() {
                 PathBuf::from(&first.filepath)
             } else {
-                PathBuf::from(&info.filename)
+                fix_extension(&info.filename, config.format)
             }
         } else {
-            PathBuf::from(&info.filename)
+            fix_extension(&info.filename, config.format)
         };
 
         tracing::info!(
@@ -282,6 +351,34 @@ impl DownloadBackend for YtDlpBackend {
     }
 }
 
+/// Replace the file extension with the target audio format.
+/// yt-dlp's --print-json reports the pre-extraction filename (e.g. .webm),
+/// but --extract-audio produces a file with the converted extension.
+fn fix_extension(filename: &str, format: crate::backend::AudioFormat) -> PathBuf {
+    let mut path = PathBuf::from(filename);
+    let new_ext = match format {
+        crate::backend::AudioFormat::Best => return path, // keep original
+        crate::backend::AudioFormat::Opus => "opus",
+        crate::backend::AudioFormat::Mp3 => "mp3",
+        crate::backend::AudioFormat::Flac => "flac",
+    };
+    path.set_extension(new_ext);
+    path
+}
+
+// --- ytmusicapi search result types ---
+
+#[derive(Debug, Deserialize)]
+struct YtMusicSearchEntry {
+    #[serde(rename = "videoId")]
+    video_id: String,
+    title: String,
+    artist: Option<String>,
+    #[allow(dead_code)]
+    album: Option<String>,
+    duration_seconds: Option<f64>,
+}
+
 // --- yt-dlp JSON output types ---
 
 #[derive(Debug, Deserialize)]
@@ -339,9 +436,13 @@ mod tests {
     }
 
     #[test]
-    fn test_search_source_prefix() {
-        assert_eq!(SearchSource::YouTubeMusic.prefix(), "ytmusicsearch5");
-        assert_eq!(SearchSource::YouTube.prefix(), "ytsearch5");
+    fn test_parse_ytmusic_entry() {
+        let json = r#"[{"videoId": "abc123", "title": "Time", "artist": "Pink Floyd", "album": "DSOTM", "duration_seconds": 413.0}]"#;
+        let entries: Vec<super::YtMusicSearchEntry> = serde_json::from_str(json).unwrap();
+        assert_eq!(entries.len(), 1);
+        assert_eq!(entries[0].video_id, "abc123");
+        assert_eq!(entries[0].title, "Time");
+        assert_eq!(entries[0].artist.as_deref(), Some("Pink Floyd"));
     }
 
     #[test]