video thumbnails

backend/Cargo.lock

@@ -436,6 +436,17 @@ dependencies = [
  "nom",
 ]
 
+[[package]]
+name = "cfb"
+version = "0.7.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d38f2da7a0a2c4ccf0065be06397cc26a81f4e528be095826eee9d4adbb8c60f"
+dependencies = [
+ "byteorder",
+ "fnv",
+ "uuid",
+]
+
 [[package]]
 name = "cfg-if"
 version = "1.0.4"
@@ -1568,6 +1579,15 @@ dependencies = [
  "serde_core",
 ]
 
+[[package]]
+name = "infer"
+version = "0.19.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a588916bfdfd92e71cacef98a63d9b1f0d74d6599980d11894290e7ddefffcf7"
+dependencies = [
+ "cfb",
+]
+
 [[package]]
 name = "ipnet"
 version = "2.11.0"
@@ -2134,6 +2154,7 @@ dependencies = [
  "http-body-util",
  "hyper",
  "image",
+ "infer",
  "jsonwebtoken",
  "mime_guess",
  "once_cell",

backend/Cargo.toml

@@ -47,6 +47,7 @@ futures-util = "0.3"
 url = "2.5"
 once_cell = "1.19"
 regex = "1.11"
+infer = "0.19"
 utoipa = { version = "4.2", default-features = false, features = ["chrono", "uuid", "preserve_order"] }
 clap = { version = "4.5", features = ["derive"] }
 

backend/Dockerfile

@@ -123,6 +123,7 @@ RUN apt-get update \
         tesseract-ocr \
         ghostscript \
         qpdf \
+        ffmpeg \
     && rm -rf /var/lib/apt/lists/* \
     && mkdir -p /usr/local/lib \
     && useradd --system --create-home --uid 10001 appuser

backend/src/storage.rs

@@ -27,6 +27,8 @@ pub trait ObjectStorage: Send + Sync + 'static {
 
     async fn get_object(&self, key: &str) -> Result<Vec<u8>>;
 
+    async fn get_object_range(&self, key: &str, start: u64, end: Option<u64>) -> Result<Vec<u8>>;
+
     async fn delete_object(&self, key: &str) -> Result<()>;
 }
 
@@ -104,6 +106,15 @@ impl ObjectStorage for S3Storage {
         Ok(data.into_bytes().to_vec())
     }
 
+    async fn get_object_range(&self, key: &str, start: u64, end: Option<u64>) -> Result<Vec<u8>> {
+        let data = self
+            .bucket
+            .get_object_range(key, start, end)
+            .await
+            .context("failed to download ranged object from S3")?;
+        Ok(data.into_bytes().to_vec())
+    }
+
     async fn delete_object(&self, key: &str) -> Result<()> {
         self.bucket
             .delete_object(key)
@@ -167,6 +178,16 @@ impl TenantStorage {
         self.inner.get_object(&qualified).await
     }
 
+    pub async fn get_object_range(
+        &self,
+        key: &str,
+        start: u64,
+        end: Option<u64>,
+    ) -> Result<Vec<u8>> {
+        let qualified = self.qualify(key);
+        self.inner.get_object_range(&qualified, start, end).await
+    }
+
     pub async fn delete_object(&self, key: &str) -> Result<()> {
         let qualified = self.qualify(key);
         self.inner.delete_object(&qualified).await

backend/src/test_support/mod.rs

@@ -112,6 +112,26 @@ impl ObjectStorage for FakeStorage {
             .ok_or_else(|| anyhow!("object {key} missing"))
     }
 
+    async fn get_object_range(
+        &self,
+        key: &str,
+        start: u64,
+        end: Option<u64>,
+    ) -> Result<Vec<u8>> {
+        let guard = self.objects.lock().await;
+        let bytes = guard
+            .get(key)
+            .map(|obj| obj.bytes.clone())
+            .ok_or_else(|| anyhow!("object {key} missing"))?;
+
+        let start_idx = start as usize;
+        let end_idx = end.map(|idx| idx.saturating_add(1) as usize).unwrap_or(bytes.len());
+        if start_idx >= bytes.len() {
+            return Ok(Vec::new());
+        }
+        Ok(bytes[start_idx..end_idx.min(bytes.len())].to_vec())
+    }
+
     async fn delete_object(&self, key: &str) -> Result<()> {
         let mut guard = self.objects.lock().await;
         guard.remove(key);

backend/src/workers/analyze.rs

@@ -1,8 +1,10 @@
-use std::collections::HashSet;
+use std::{collections::HashSet, sync::Arc, time::Duration};
-use std::sync::Arc;
 
 use async_trait::async_trait;
+use diesel::prelude::*;
+use infer;
 use serde::Deserialize;
+use tokio::task;
 use uuid::Uuid;
 
 use crate::{
@@ -16,7 +18,8 @@ use super::{
     job_execution_from_task_error,
     ocr::{GenerateOcrTask, OCR_TEXT_ASSET_TYPE},
     taskflow::{
-        document::DocumentVersionTaskContext, BoxedTask, TaskExecutor, TaskPlanner, TaskResult,
+        document::DocumentVersionTaskContext, BoxedTask, Task, TaskError, TaskExecutor,
+        TaskPlanner, TaskResult,
     },
     thumbnails::GenerateThumbnailsTask,
     JobExecution, JobHandler,
@@ -99,6 +102,8 @@ struct AnalyzePlanner {
     state: Arc<AppState>,
 }
 
+const MIME_SNIFF_BYTES: usize = 8192;
+
 impl AnalyzePlanner {
     fn new(force: bool, state: Arc<AppState>) -> Self {
         Self { force, state }
@@ -114,6 +119,8 @@ impl TaskPlanner<DocumentVersionTaskContext> for AnalyzePlanner {
         let document = ctx.document().await?.clone();
         let mut tasks: Vec<BoxedTask<DocumentVersionTaskContext>> = Vec::new();
 
+        tasks.push(Box::new(EnsureMimeTask));
+
         let (thumbnail_supported, _) = determine_thumbnail_support(&document);
         if thumbnail_supported {
             tasks.push(Box::new(GenerateThumbnailsTask::new(self.force)));
@@ -142,6 +149,85 @@ impl TaskPlanner<DocumentVersionTaskContext> for AnalyzePlanner {
     }
 }
 
+struct EnsureMimeTask;
+
+#[async_trait]
+impl Task<DocumentVersionTaskContext> for EnsureMimeTask {
+    fn name(&self) -> &'static str {
+        "ensure-mime-type"
+    }
+
+    async fn execute(&self, ctx: &mut DocumentVersionTaskContext) -> TaskResult<()> {
+        let document = ctx.document().await?.clone();
+        let current = document.mime_type.clone();
+
+        let guessed = guess_mime_type(ctx, &document).await?;
+        let desired = match guessed {
+            Some(mime) if current.as_deref() != Some(mime.as_str()) => Some(mime),
+            _ => None,
+        };
+
+        if let Some(new_mime) = desired {
+            update_document_mime(ctx, document.id, new_mime).await?;
+        }
+
+        Ok(())
+    }
+}
+
+async fn guess_mime_type(
+    ctx: &mut DocumentVersionTaskContext,
+    document: &Document,
+) -> TaskResult<Option<String>> {
+    let bytes = ctx.object_head(MIME_SNIFF_BYTES).await?;
+    Ok(sniff_mime(&bytes, &document.original_name))
+}
+
+fn sniff_mime(bytes: &[u8], original_name: &str) -> Option<String> {
+    if let Some(kind) = infer::get(bytes) {
+        return Some(kind.mime_type().to_string());
+    }
+
+    mime_guess::from_path(original_name)
+        .first_raw()
+        .map(|value| value.to_string())
+}
+
+async fn update_document_mime(
+    ctx: &mut DocumentVersionTaskContext,
+    document_id: Uuid,
+    mime_type: String,
+) -> TaskResult<()> {
+    let tenant_id = ctx.tenant_id();
+    let state = ctx.state().clone();
+    let mime_type_clone = mime_type.clone();
+
+    task::spawn_blocking(move || -> Result<(), String> {
+        let mut conn = state
+            .db_for_tenant(tenant_id)
+            .map_err(|err| format!("{err:?}"))?;
+
+        diesel::update(
+            crate::schema::documents::table.filter(crate::schema::documents::id.eq(document_id)),
+        )
+        .set(crate::schema::documents::mime_type.eq(Some(mime_type_clone)))
+        .execute(&mut conn)
+        .map_err(|err| format!("{err:?}"))
+        .map(|_| ())
+    })
+    .await
+    .map_err(|err| {
+        TaskError::retry(
+            Duration::from_secs(60),
+            format!("mime update task panicked: {err}"),
+        )
+    })?
+    .map_err(|err| TaskError::retry(Duration::from_secs(30), err))?;
+
+    ctx.set_document_mime(Some(mime_type));
+    Ok(())
+}
+
 pub(crate) fn determine_thumbnail_support(document: &Document) -> (bool, Option<String>) {
     let supported_mimes: HashSet<&'static str> = [
         "image/jpeg",
@@ -151,6 +237,12 @@ pub(crate) fn determine_thumbnail_support(document: &Document) -> (bool, Option<
         "image/bmp",
         "image/webp",
         "application/pdf",
+        "video/mp4",
+        "video/quicktime",
+        "video/webm",
+        "video/x-msvideo",
+        "video/x-ms-wmv",
+        "video/x-matroska",
     ]
     .into_iter()
     .collect();
@@ -168,7 +260,8 @@ pub(crate) fn determine_thumbnail_support(document: &Document) -> (bool, Option<
         .map(|ext| ext.to_ascii_lowercase())
     {
         let supported_exts = [
-            "jpg", "jpeg", "png", "gif", "tif", "tiff", "bmp", "webp", "pdf",
+            "jpg", "jpeg", "png", "gif", "tif", "tiff", "bmp", "webp", "pdf", "mp4", "m4v", "mov",
+            "webm", "mkv", "avi", "wmv",
         ];
         if supported_exts.contains(&ext.as_str()) {
             return (true, None);
@@ -195,3 +288,21 @@ fn document_supports_ocr(document: &Document) -> bool {
                 .unwrap_or(false)
         })
 }
+
+#[cfg(test)]
+mod tests {
+    use super::sniff_mime;
+
+    #[test]
+    fn sniff_mime_prefers_magic_bytes() {
+        const PNG_HEADER: [u8; 8] = [0x89, b'P', b'N', b'G', 0x0D, 0x0A, 0x1A, 0x0A];
+        let mime = sniff_mime(&PNG_HEADER, "file.txt");
+        assert_eq!(mime.as_deref(), Some("image/png"));
+    }
+
+    #[test]
+    fn sniff_mime_falls_back_to_extension() {
+        let mime = sniff_mime(b"not enough to detect", "video.mp4");
+        assert_eq!(mime.as_deref(), Some("video/mp4"));
+    }
+}

backend/src/workers/taskflow/document.rs

@@ -9,7 +9,7 @@ use crate::models::{Document, DocumentVersion};
 use crate::state::AppState;
 use crate::storage::TenantStorage;
 use crate::workers::common::{load_document_version, load_version_assets, LoadedAsset};
-use crate::workers::{fetch_version_object, FetchVersionError};
+use crate::workers::{check_worker_document_limit, fetch_version_object, FetchVersionError};
 
 use super::{TaskContext, TaskError, TaskResult};
 
@@ -86,6 +86,12 @@ impl DocumentVersionTaskContext {
         self.assets = None;
     }
 
+    pub fn set_document_mime(&mut self, mime: Option<String>) {
+        if let Some(document) = self.document.as_mut() {
+            document.mime_type = mime;
+        }
+    }
+
     pub fn state(&self) -> &Arc<AppState> {
         &self.state
     }
@@ -158,6 +164,28 @@ impl DocumentVersionTaskContext {
         Ok(self.object_bytes.as_deref().expect("bytes hydrated"))
     }
 
+    pub async fn object_head(&mut self, max_bytes: usize) -> TaskResult<Vec<u8>> {
+        let version = self.version().await?.clone();
+        check_worker_document_limit(version.size_bytes, self.max_document_bytes).map_err(
+            |(size, limit)| {
+                TaskError::fail(format!(
+                    "document size {size} bytes exceeds worker limit of {limit} bytes"
+                ))
+            },
+        )?;
+
+        let end = max_bytes.saturating_sub(1) as u64;
+        self.storage
+            .get_object_range(&version.s3_key, 0, Some(end))
+            .await
+            .map_err(|err| {
+                TaskError::retry(
+                    DEFAULT_RETRY_DELAY,
+                    format!("failed to fetch ranged object: {err}"),
+                )
+            })
+    }
+
     async fn ensure_document_loaded(&mut self) -> TaskResult<()> {
         if self.document.is_some() && self.version.is_some() {
             return Ok(());

backend/src/workers/thumbnails.rs

@@ -1,12 +1,13 @@
-use std::{convert::TryInto, io::Cursor, panic, sync::Arc, time::Duration};
+use std::{convert::TryInto, io::Cursor, panic, process::Stdio, sync::Arc, time::Duration};
 
 use async_trait::async_trait;
 use chrono::Utc;
 use diesel::{pg::upsert::excluded, prelude::*};
 use image::{GenericImageView, ImageFormat, ImageReader};
 use pdfium_render::prelude::*;
+use serde::Deserialize;
 use serde_json::{Map, Value};
-use tokio::task;
+use tokio::{process::Command, task, time::timeout};
 use tracing::{info, warn};
 use uuid::Uuid;
 
@@ -17,6 +18,7 @@ use crate::{
     schema::{document_assets, document_versions},
     state::AppState,
     utils::storage_paths::document_asset_key,
+    workers::check_worker_document_limit,
 };
 
 use super::{
@@ -28,6 +30,8 @@ pub const THUMBNAIL_WIDTH: u32 = 512;
 pub const THUMBNAIL_HEIGHT: u32 = 512;
 const RENDER_WIDTH: u32 = THUMBNAIL_WIDTH * 4;
 const RENDER_HEIGHT: u32 = THUMBNAIL_HEIGHT * 4;
+const VIDEO_PRESIGN_TTL: Duration = Duration::from_secs(300);
+const FFMPEG_TIMEOUT: Duration = Duration::from_secs(30);
 pub const THUMBNAIL_ASSET_TYPE: &str = "thumbnail";
 
 pub struct GenerateThumbnailsTask {
@@ -54,8 +58,12 @@ impl Task<DocumentVersionTaskContext> for GenerateThumbnailsTask {
             return Ok(());
         }
 
-        let bytes = ctx.buffered_object().await?;
+        let generation = if document_is_video(&context.document) {
-        let generation = generate_thumbnails(&context.document, bytes).map_err(TaskError::fail)?;
+            generate_video_thumbnail(ctx, &context).await?
+        } else {
+            let bytes = ctx.buffered_object().await?;
+            generate_thumbnails(&context.document, bytes).map_err(TaskError::fail)?
+        };
 
         if let Some(page_count) = generation.page_count {
             let state = ctx.state().clone();
@@ -275,6 +283,46 @@ fn generate_thumbnails(document: &Document, bytes: &[u8]) -> Result<GeneratedAss
     }
 }
 
+async fn generate_video_thumbnail(
+    ctx: &DocumentVersionTaskContext,
+    context: &ThumbnailContext,
+) -> TaskResult<GeneratedAssets> {
+    if let Err((size, limit)) =
+        check_worker_document_limit(context.version.size_bytes, ctx.max_document_bytes())
+    {
+        return Err(TaskError::fail(format!(
+            "document size {size} bytes exceeds worker limit of {limit} bytes"
+        )));
+    }
+
+    let presigned_url = ctx
+        .storage()
+        .presign_get_object(&context.version.s3_key, VIDEO_PRESIGN_TTL, None)
+        .await
+        .map_err(|err| {
+            TaskError::retry(
+                Duration::from_secs(30),
+                format!("failed to presign video for thumbnail: {err}"),
+            )
+        })?;
+
+    let probe = probe_video_metadata(&presigned_url)
+        .await
+        .map_err(TaskError::fail)?;
+    let timestamp = pick_thumbnail_timestamp(probe.duration);
+
+    let frame_bytes = extract_video_frame(&presigned_url, timestamp)
+        .await
+        .map_err(TaskError::fail)?;
+
+    let thumbnail = generate_image_assets(&frame_bytes).map_err(TaskError::fail)?;
+
+    Ok(GeneratedAssets {
+        thumbnail,
+        page_count: None,
+    })
+}
+
 fn generate_image_assets(bytes: &[u8]) -> Result<GeneratedAsset, String> {
     let reader = ImageReader::new(Cursor::new(bytes))
         .with_guessed_format()
@@ -352,6 +400,107 @@ fn generate_pdf_assets(bytes: &[u8]) -> Result<PdfGeneratedAssets, String> {
     })
 }
 
+#[derive(Deserialize)]
+struct FfprobeOutput {
+    format: Option<FfprobeFormat>,
+}
+
+#[derive(Deserialize)]
+struct FfprobeFormat {
+    duration: Option<String>,
+}
+
+struct VideoProbe {
+    duration: Option<f64>,
+}
+
+async fn probe_video_metadata(url: &str) -> Result<VideoProbe, String> {
+    let mut cmd = Command::new("ffprobe");
+    cmd.arg("-v")
+        .arg("error")
+        .arg("-show_entries")
+        .arg("format=duration")
+        .arg("-of")
+        .arg("json")
+        .arg(url)
+        .stdout(Stdio::piped());
+
+    let output = timeout(FFMPEG_TIMEOUT, cmd.output())
+        .await
+        .map_err(|_| "ffprobe timed out".to_string())?
+        .map_err(|err| format!("ffprobe failed to start: {err}"))?;
+
+    if !output.status.success() {
+        return Err(format!(
+            "ffprobe exited with status {}: {}",
+            output.status,
+            String::from_utf8_lossy(&output.stderr)
+        ));
+    }
+
+    let parsed: FfprobeOutput = serde_json::from_slice(&output.stdout)
+        .map_err(|err| format!("failed to parse ffprobe output: {err}"))?;
+
+    let duration = parsed
+        .format
+        .and_then(|format| format.duration)
+        .and_then(|dur| dur.parse::<f64>().ok());
+
+    Ok(VideoProbe { duration })
+}
+
+fn pick_thumbnail_timestamp(duration: Option<f64>) -> f64 {
+    if let Some(duration) = duration {
+        if duration.is_finite() && duration > 0.0 {
+            let target = duration * 0.2;
+            let end = (duration - 1.0).max(0.0);
+            return target.max(2.0).min(end).max(0.0);
+        }
+    }
+    2.0
+}
+
+async fn extract_video_frame(url: &str, timestamp_secs: f64) -> Result<Vec<u8>, String> {
+    let timestamp_arg = format!("{timestamp_secs:.3}");
+
+    let mut cmd = Command::new("ffmpeg");
+    cmd.arg("-hide_banner")
+        .arg("-loglevel")
+        .arg("error")
+        .arg("-nostdin")
+        .arg("-ss")
+        .arg(timestamp_arg)
+        .arg("-i")
+        .arg(url)
+        .arg("-frames:v")
+        .arg("1")
+        .arg("-f")
+        .arg("image2pipe")
+        .arg("-vcodec")
+        .arg("png")
+        .arg("-")
+        .stdout(Stdio::piped());
+
+    let output = timeout(FFMPEG_TIMEOUT, cmd.output())
+        .await
+        .map_err(|_| "ffmpeg timed out".to_string())?
+        .map_err(|err| format!("ffmpeg failed to start: {err}"))?;
+
+    if !output.status.success() {
+        return Err(format!(
+            "ffmpeg exited with status {}: {}",
+            output.status,
+            String::from_utf8_lossy(&output.stderr)
+        ));
+    }
+
+    if output.stdout.is_empty() {
+        return Err("ffmpeg produced no frame data".to_string());
+    }
+
+    Ok(output.stdout)
+}
+
 fn encode_dynamic_image(image: image::DynamicImage) -> Result<GeneratedImage, String> {
     let (width, height) = image.dimensions();
     let mut cursor = Cursor::new(Vec::new());
@@ -461,6 +610,27 @@ fn persist_document_page_count(
     Ok(())
 }
 
+fn document_is_video(document: &Document) -> bool {
+    const VIDEO_MIME_TYPES: [&str; 6] = [
+        "video/mp4",
+        "video/quicktime",
+        "video/webm",
+        "video/x-msvideo",
+        "video/x-ms-wmv",
+        "video/x-matroska",
+    ];
+    if let Some(mime) = document.mime_type.as_deref() {
+        if VIDEO_MIME_TYPES
+            .iter()
+            .any(|candidate| mime.eq_ignore_ascii_case(candidate))
+        {
+            return true;
+        }
+    }
+
+    false
+}
+
 fn document_is_pdf(document: &Document) -> bool {
     document
         .mime_type

backend/tests/download_flow.rs

@@ -3,22 +3,22 @@ use axum::http::StatusCode;
 use papercrate::test_support::{acquire_db_lock, body_to_vec, TestApp, TestUserRole};
 use serde::Deserialize;
 
-#[derive(Deserialize)]
+#[derive(Clone, Deserialize)]
 struct DocumentDetail {
     document: DocumentInfo,
 }
 
-#[derive(Deserialize)]
+#[derive(Clone, Deserialize)]
 struct DocumentInfo {
     current_version: Option<DocumentVersion>,
 }
 
-#[derive(Deserialize)]
+#[derive(Clone, Deserialize)]
 struct DocumentVersion {
     download: DownloadLink,
 }
 
-#[derive(Deserialize)]
+#[derive(Clone, Deserialize)]
 struct DownloadLink {
     url: String,
     expires_at: i64,
@@ -47,14 +47,15 @@ async fn document_download_redirects_when_proxy_disabled() -> Result<()> {
     assert_eq!(upload.status(), StatusCode::CREATED);
     let body = body_to_vec(upload.into_body()).await?;
     let detail: DocumentDetail = serde_json::from_slice(&body)?;
-    let download_path = detail
+    let download_link = detail
         .document
         .current_version
         .as_ref()
         .expect("missing version")
         .download
-        .url
         .clone();
+    assert!(download_link.expires_at > 0);
+    let download_path = download_link.url.clone();
 
     let redirect = app.get(&download_path, None).await?;
     assert_eq!(redirect.status(), StatusCode::TEMPORARY_REDIRECT);