chapter extraction via ffprobe during library sync (#5)

ffprobe module in adapter-common: parse JSON output into Vec<Chapter>,
should_probe_chapters gate (movie OR >45min), graceful failure.
Add set_chapters/set_role setters to MediaItem.

Cargo.lock

@@ -25,6 +25,7 @@ dependencies = [
  "serde",
  "serde_json",
  "sqlx",
+ "tokio",
  "tracing",
  "uuid",
 ]
@@ -1795,6 +1796,7 @@ name = "presentation"
 version = "0.1.0"
 dependencies = [
  "adapter-auth",
+ "adapter-common",
  "adapter-event-publisher",
  "adapter-jellyfin",
  "adapter-local-files",
@@ -3372,6 +3374,7 @@ name = "worker"
 version = "0.1.0"
 dependencies = [
  "adapter-auth",
+ "adapter-common",
  "adapter-event-publisher",
  "adapter-jellyfin",
  "adapter-local-files",

crates/adapters/adapter-common/Cargo.toml

@@ -11,3 +11,4 @@ uuid = { workspace = true }
 serde = { workspace = true }
 serde_json = { workspace = true }
 tracing = { workspace = true }
+tokio = { workspace = true }

crates/adapters/adapter-common/src/ffprobe.rs

@@ -0,0 +1,180 @@
+use domain::{Chapter, ContentType, SourceUri};
+use serde::Deserialize;
+
+const CHAPTER_PROBE_MIN_DURATION_SECS: u32 = 2700;
+
+#[derive(Deserialize)]
+struct FfprobeOutput {
+    #[serde(default)]
+    chapters: Vec<FfprobeChapter>,
+}
+
+#[derive(Deserialize)]
+struct FfprobeChapter {
+    #[serde(default)]
+    start_time: String,
+    #[serde(default)]
+    end_time: String,
+    #[serde(default)]
+    tags: Option<FfprobeChapterTags>,
+}
+
+#[derive(Deserialize)]
+struct FfprobeChapterTags {
+    title: Option<String>,
+}
+
+pub fn parse_chapters_json(json: &str) -> Vec<Chapter> {
+    let output: FfprobeOutput = match serde_json::from_str(json) {
+        Ok(o) => o,
+        Err(_) => return Vec::new(),
+    };
+    output
+        .chapters
+        .into_iter()
+        .map(|c| {
+            let title = c.tags.and_then(|t| t.title);
+            let start_secs = c.start_time.parse::<f64>().unwrap_or(0.0);
+            let end_secs = c.end_time.parse::<f64>().unwrap_or(0.0);
+            Chapter::new(title, start_secs, end_secs)
+        })
+        .collect()
+}
+
+pub fn should_probe_chapters(content_type: &ContentType, duration_secs: u32) -> bool {
+    matches!(content_type, ContentType::Movie) || duration_secs > CHAPTER_PROBE_MIN_DURATION_SECS
+}
+
+pub async fn extract_chapters(source_uri: &SourceUri) -> Vec<Chapter> {
+    let path = match source_uri {
+        SourceUri::FilePath { path } => path.clone(),
+        SourceUri::NetworkUrl { url } => url.clone(),
+    };
+
+    let result = tokio::process::Command::new("ffprobe")
+        .args([
+            "-v",
+            "quiet",
+            "-print_format",
+            "json",
+            "-show_chapters",
+            &path,
+        ])
+        .output()
+        .await;
+
+    match result {
+        Ok(output) if output.status.success() => {
+            let json = String::from_utf8_lossy(&output.stdout);
+            parse_chapters_json(&json)
+        }
+        Ok(output) => {
+            tracing::warn!(
+                path = %path,
+                stderr = %String::from_utf8_lossy(&output.stderr),
+                "ffprobe exited with non-zero status"
+            );
+            Vec::new()
+        }
+        Err(e) => {
+            tracing::warn!(error = %e, "ffprobe not available or failed to execute");
+            Vec::new()
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn parse_ffprobe_json_with_chapters() {
+        let json = r#"{
+            "chapters": [
+                {
+                    "id": 0,
+                    "time_base": "1/1000",
+                    "start": 0,
+                    "start_time": "0.000000",
+                    "end": 300000,
+                    "end_time": "300.000000",
+                    "tags": { "title": "Opening" }
+                },
+                {
+                    "id": 1,
+                    "time_base": "1/1000",
+                    "start": 300000,
+                    "start_time": "300.000000",
+                    "end": 1800000,
+                    "end_time": "1800.000000",
+                    "tags": { "title": "Main Feature" }
+                },
+                {
+                    "id": 2,
+                    "time_base": "1/1000",
+                    "start": 1800000,
+                    "start_time": "1800.000000",
+                    "end": 2100000,
+                    "end_time": "2100.000000"
+                }
+            ]
+        }"#;
+
+        let chapters = parse_chapters_json(json);
+        assert_eq!(chapters.len(), 3);
+
+        assert_eq!(chapters[0].title(), Some("Opening"));
+        assert!((chapters[0].start_secs() - 0.0).abs() < f64::EPSILON);
+        assert!((chapters[0].end_secs() - 300.0).abs() < f64::EPSILON);
+
+        assert_eq!(chapters[1].title(), Some("Main Feature"));
+        assert!((chapters[1].start_secs() - 300.0).abs() < f64::EPSILON);
+        assert!((chapters[1].end_secs() - 1800.0).abs() < f64::EPSILON);
+
+        assert_eq!(chapters[2].title(), None);
+        assert!((chapters[2].start_secs() - 1800.0).abs() < f64::EPSILON);
+        assert!((chapters[2].end_secs() - 2100.0).abs() < f64::EPSILON);
+    }
+
+    #[test]
+    fn parse_ffprobe_json_no_chapters() {
+        let json = r#"{ "chapters": [] }"#;
+        let chapters = parse_chapters_json(json);
+        assert!(chapters.is_empty());
+    }
+
+    #[test]
+    fn parse_ffprobe_json_missing_chapters_key() {
+        let json = r#"{}"#;
+        let chapters = parse_chapters_json(json);
+        assert!(chapters.is_empty());
+    }
+
+    #[test]
+    fn parse_ffprobe_json_invalid() {
+        let chapters = parse_chapters_json("not json");
+        assert!(chapters.is_empty());
+    }
+
+    #[test]
+    fn short_items_skip_probe() {
+        assert!(!should_probe_chapters(&ContentType::Episode, 1800));
+        assert!(!should_probe_chapters(&ContentType::Short, 300));
+    }
+
+    #[test]
+    fn movie_always_probed() {
+        assert!(should_probe_chapters(&ContentType::Movie, 600));
+        assert!(should_probe_chapters(&ContentType::Movie, 7200));
+    }
+
+    #[test]
+    fn long_episode_probed() {
+        assert!(should_probe_chapters(&ContentType::Episode, 3600));
+    }
+
+    #[test]
+    fn episode_under_threshold_skipped() {
+        assert!(!should_probe_chapters(&ContentType::Episode, 2700));
+    }
+}

crates/domain/src/models/media.rs

@@ -226,6 +226,14 @@ impl MediaItem {
     pub fn chapters(&self) -> &[Chapter] {
         &self.chapters
     }
+
+    pub fn set_chapters(&mut self, chapters: Vec<Chapter>) {
+        self.chapters = chapters;
+    }
+
+    pub fn set_role(&mut self, role: MediaRole) {
+        self.role = role;
+    }
 }
 
 #[derive(Debug, Clone, Serialize, Deserialize)]