can't be bothered writing a commit message

breadarrd/src/importer/ffprobe.rs

@@ -0,0 +1,379 @@
+use std::path::Path;
+use std::process::Command;
+
+use anyhow::{bail, Context, Result};
+
+#[derive(Debug, Clone, PartialEq)]
+pub struct AudioStream {
+    pub codec: Option<String>,
+    pub language: Option<String>,
+    pub is_default: bool,
+    pub channels: Option<i64>,
+}
+
+#[derive(Debug, Clone, PartialEq)]
+pub struct SubtitleStream {
+    pub language: Option<String>,
+}
+
+#[derive(Debug, Clone, Default, PartialEq)]
+pub struct MediaProbe {
+    pub duration_secs: Option<f64>,
+    pub container_bitrate: Option<i64>,
+    /// ffprobe's own long-form container name (e.g. "Matroska / WebM") —
+    /// distinct from the file extension, which can lie or just be absent.
+    pub container_format: Option<String>,
+    pub video_codec: Option<String>,
+    pub width: Option<i64>,
+    pub height: Option<i64>,
+    pub video_bitrate: Option<i64>,
+    pub frame_rate: Option<f64>,
+    /// The primary video stream's `color_transfer` tag as ffprobe reports
+    /// it (e.g. "smpte2084", "arib-std-b67", "bt709") — kept as the raw
+    /// string rather than pre-reduced to a bool so a future caller isn't
+    /// stuck with just today's idea of what counts as "HDR" (Dolby Vision
+    /// profiles, for instance, don't all surface the same way here).
+    pub color_transfer: Option<String>,
+    pub audio: Vec<AudioStream>,
+    pub subtitles: Vec<SubtitleStream>,
+    /// The complete ffprobe JSON response, kept verbatim. Every field
+    /// above is a deliberately-narrow, purpose-built read of this same
+    /// payload; this is the escape hatch — future functionality that needs
+    /// something not already surfaced as its own column (chapters, extra
+    /// stream tags, encoder info, etc.) can mine it here without requiring
+    /// a re-probe of the whole library.
+    pub raw_json: String,
+}
+
+impl MediaProbe {
+    /// Best-effort HDR classification from `color_transfer` — PQ
+    /// (smpte2084, the common HDR10/HDR10+/Dolby-Vision-base-layer
+    /// transfer function) or HLG (arib-std-b67). Not exhaustive by design;
+    /// see `color_transfer`'s own doc comment for why the raw value is kept
+    /// around too.
+    pub fn is_hdr(&self) -> bool {
+        matches!(
+            self.color_transfer.as_deref(),
+            Some("smpte2084") | Some("arib-std-b67")
+        )
+    }
+}
+
+fn is_english(lang: Option<&str>) -> bool {
+    matches!(lang, Some("eng") | Some("en"))
+}
+
+impl MediaProbe {
+    pub fn has_english_audio(&self) -> bool {
+        self.audio.iter().any(|a| is_english(a.language.as_deref()))
+    }
+
+    /// True when the default-disposition audio track (if any) is not
+    /// English — mirrors `mkv::default_track_is_english_or_unset`'s "unset
+    /// default isn't a problem" behavior, since the codebase's post-download
+    /// remux fix already treats those two cases identically.
+    pub fn default_audio_is_non_english(&self) -> bool {
+        match self.audio.iter().find(|a| a.is_default) {
+            Some(a) => !is_english(a.language.as_deref()),
+            None => false,
+        }
+    }
+}
+
+/// One `ffprobe` call gets container + every stream's codec/resolution/
+/// bitrate/language/default-disposition in a single JSON payload — no need
+/// for separate audio/video/subtitle passes.
+pub fn probe(path: &Path) -> Result<MediaProbe> {
+    let output = Command::new("ffprobe")
+        .args([
+            "-v",
+            "quiet",
+            "-print_format",
+            "json",
+            "-show_format",
+            "-show_streams",
+        ])
+        .arg(path)
+        .output()
+        .context("failed to run ffprobe")?;
+
+    if !output.status.success() {
+        bail!(
+            "ffprobe failed for {}: {}",
+            path.display(),
+            String::from_utf8_lossy(&output.stderr)
+        );
+    }
+
+    let raw_json = String::from_utf8_lossy(&output.stdout).into_owned();
+    let json: serde_json::Value =
+        serde_json::from_slice(&output.stdout).context("ffprobe output was not valid JSON")?;
+
+    let format = &json["format"];
+    let duration_secs = format["duration"]
+        .as_str()
+        .and_then(|s| s.parse::<f64>().ok());
+    let container_bitrate = format["bit_rate"]
+        .as_str()
+        .and_then(|s| s.parse::<i64>().ok());
+    let container_format = format["format_long_name"].as_str().map(str::to_string);
+
+    let streams = json["streams"].as_array().cloned().unwrap_or_default();
+
+    let mut probe = MediaProbe {
+        duration_secs,
+        container_bitrate,
+        container_format,
+        raw_json,
+        ..Default::default()
+    };
+
+    for stream in &streams {
+        let codec_type = stream["codec_type"].as_str().unwrap_or("");
+        let language = stream["tags"]["language"]
+            .as_str()
+            .or_else(|| stream["tags"]["LANGUAGE"].as_str())
+            .map(str::to_string);
+        match codec_type {
+            "video" if probe.video_codec.is_none() => {
+                // First video stream only — a second "video" stream in a
+                // real-world file is almost always an embedded cover-art
+                // thumbnail, not a second picture track.
+                probe.video_codec = stream["codec_name"].as_str().map(str::to_string);
+                probe.width = stream["width"].as_i64();
+                probe.height = stream["height"].as_i64();
+                probe.video_bitrate = stream["bit_rate"]
+                    .as_str()
+                    .and_then(|s| s.parse::<i64>().ok());
+                probe.frame_rate = stream["avg_frame_rate"]
+                    .as_str()
+                    .and_then(parse_frame_rate_fraction)
+                    .or_else(|| {
+                        stream["r_frame_rate"]
+                            .as_str()
+                            .and_then(parse_frame_rate_fraction)
+                    });
+                probe.color_transfer = stream["color_transfer"].as_str().map(str::to_string);
+            }
+            "audio" => {
+                probe.audio.push(AudioStream {
+                    codec: stream["codec_name"].as_str().map(str::to_string),
+                    language,
+                    is_default: stream["disposition"]["default"].as_i64() == Some(1),
+                    channels: stream["channels"].as_i64(),
+                });
+            }
+            "subtitle" => {
+                probe.subtitles.push(SubtitleStream { language });
+            }
+            _ => {}
+        }
+    }
+
+    Ok(probe)
+}
+
+/// ffprobe reports frame rate as a "num/den" fraction string (e.g.
+/// "24000/1001" for 23.976fps) rather than a plain number. `0/0` means
+/// "unknown" (common for e.g. still-image or data streams that don't
+/// really have one), not a real rate.
+fn parse_frame_rate_fraction(s: &str) -> Option<f64> {
+    let (num, den) = s.split_once('/')?;
+    let num: f64 = num.parse().ok()?;
+    let den: f64 = den.parse().ok()?;
+    if den == 0.0 {
+        return None;
+    }
+    Some(num / den)
+}
+
+/// A cheap header-level probe succeeding only proves the container is
+/// parseable — it doesn't catch a truncated file or mid-stream bit-rot,
+/// which requires actually decoding every frame. `ffmpeg -xerror` does
+/// that: any decode error either exits non-zero or writes to stderr. This
+/// is CPU-bound and can take minutes per file (it reads and decodes the
+/// whole thing), so it's deliberately a separate, opt-in pass rather than
+/// part of every routine scan — see `verify-library` in `main.rs`.
+pub enum DecodeCheck {
+    Ok,
+    Corrupt(String),
+}
+
+pub fn verify_decodable(path: &Path) -> Result<DecodeCheck> {
+    let output = Command::new("ffmpeg")
+        .args(["-v", "error", "-xerror", "-i"])
+        .arg(path)
+        .args(["-f", "null", "-"])
+        .output()
+        .context("failed to run ffmpeg for decode verification")?;
+
+    if output.status.success() && output.stderr.is_empty() {
+        Ok(DecodeCheck::Ok)
+    } else {
+        Ok(DecodeCheck::Corrupt(
+            String::from_utf8_lossy(&output.stderr).into_owned(),
+        ))
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn has_english_audio_true_when_any_track_is_english() {
+        let probe = MediaProbe {
+            audio: vec![
+                AudioStream {
+                    codec: None,
+                    language: Some("jpn".to_string()),
+                    is_default: true,
+                    channels: None,
+                },
+                AudioStream {
+                    codec: None,
+                    language: Some("eng".to_string()),
+                    is_default: false,
+                    channels: None,
+                },
+            ],
+            ..Default::default()
+        };
+        assert!(probe.has_english_audio());
+    }
+
+    #[test]
+    fn has_english_audio_false_with_no_tracks() {
+        assert!(!MediaProbe::default().has_english_audio());
+    }
+
+    #[test]
+    fn default_audio_is_non_english_true_when_default_track_is_not_english() {
+        let probe = MediaProbe {
+            audio: vec![AudioStream {
+                codec: None,
+                language: Some("ita".to_string()),
+                is_default: true,
+                channels: None,
+            }],
+            ..Default::default()
+        };
+        assert!(probe.default_audio_is_non_english());
+    }
+
+    #[test]
+    fn default_audio_is_non_english_false_when_no_default_is_set() {
+        // Mirrors mkv::default_track_is_english_or_unset: an unset default
+        // isn't treated as a problem.
+        let probe = MediaProbe {
+            audio: vec![AudioStream {
+                codec: None,
+                language: Some("ita".to_string()),
+                is_default: false,
+                channels: None,
+            }],
+            ..Default::default()
+        };
+        assert!(!probe.default_audio_is_non_english());
+    }
+
+    #[test]
+    fn default_audio_is_non_english_false_when_default_is_english() {
+        let probe = MediaProbe {
+            audio: vec![AudioStream {
+                codec: None,
+                language: Some("eng".to_string()),
+                is_default: true,
+                channels: None,
+            }],
+            ..Default::default()
+        };
+        assert!(!probe.default_audio_is_non_english());
+    }
+
+    #[test]
+    fn probe_fails_gracefully_for_a_nonexistent_file() {
+        let result = probe(Path::new("/nonexistent/path/to/nothing.mkv"));
+        assert!(result.is_err());
+    }
+
+    /// Generates a tiny real video via ffmpeg's `lavfi` synthetic source —
+    /// validates the actual JSON field extraction (width/height/codec/
+    /// duration/audio language+default) against genuine ffprobe output,
+    /// not just hand-built `MediaProbe` fixtures.
+    fn generate_test_clip(dir: &Path, width: u32, height: u32) -> std::path::PathBuf {
+        let path = dir.join("clip.mkv");
+        let status = Command::new("ffmpeg")
+            .args(["-y", "-f", "lavfi", "-i"])
+            .arg(format!("testsrc=size={width}x{height}:duration=1:rate=1"))
+            .args(["-f", "lavfi", "-i", "sine=frequency=1000:duration=1"])
+            .args(["-metadata:s:a:0", "language=eng"])
+            .args(["-c:v", "libx264", "-c:a", "aac"])
+            .arg(&path)
+            .output()
+            .expect("failed to run ffmpeg to generate a test clip");
+        assert!(
+            status.status.success(),
+            "ffmpeg failed to generate test clip: {}",
+            String::from_utf8_lossy(&status.stderr)
+        );
+        path
+    }
+
+    #[test]
+    fn probe_extracts_real_resolution_and_audio_language_from_a_generated_clip() {
+        let dir =
+            std::env::temp_dir().join(format!("breadarr-ffprobe-test-{}", std::process::id()));
+        std::fs::create_dir_all(&dir).unwrap();
+        let clip = generate_test_clip(&dir, 640, 360);
+
+        let probe = probe(&clip).unwrap();
+        assert_eq!(probe.width, Some(640));
+        assert_eq!(probe.height, Some(360));
+        assert!(probe.video_codec.is_some());
+        assert!(probe.duration_secs.unwrap_or(0.0) > 0.0);
+        assert!(probe.has_english_audio());
+
+        std::fs::remove_dir_all(&dir).unwrap();
+    }
+
+    #[test]
+    fn probe_extracts_extra_metadata_from_a_generated_clip() {
+        let dir = std::env::temp_dir().join(format!(
+            "breadarr-ffprobe-extra-test-{}",
+            std::process::id()
+        ));
+        std::fs::create_dir_all(&dir).unwrap();
+        // Generated at rate=1 (see generate_test_clip), so frame_rate
+        // should come back at (or very near) 1.0.
+        let clip = generate_test_clip(&dir, 640, 360);
+
+        let probe = probe(&clip).unwrap();
+        assert!(
+            probe.frame_rate.is_some_and(|r| (r - 1.0).abs() < 0.1),
+            "frame_rate was {:?}",
+            probe.frame_rate
+        );
+        assert!(
+            probe
+                .container_format
+                .as_deref()
+                .is_some_and(|f| !f.is_empty()),
+            "container_format should be populated"
+        );
+        assert!(
+            probe.audio.first().is_some_and(|a| a.channels.is_some()),
+            "audio channel count should be populated"
+        );
+        // A real generated clip's default color_transfer is unlikely to be
+        // an HDR transfer function — this is really asserting `is_hdr()`
+        // doesn't spuriously fire on ordinary SDR content.
+        assert!(!probe.is_hdr());
+        // The raw payload is kept verbatim and should parse as JSON on its
+        // own — a caller mining it later needs that to actually be true.
+        assert!(!probe.raw_json.is_empty());
+        assert!(serde_json::from_str::<serde_json::Value>(&probe.raw_json).is_ok());
+
+        std::fs::remove_dir_all(&dir).unwrap();
+    }
+}