This commit is contained in:
2025-11-10 23:06:53 +01:00
parent 8dddb0b940
commit 8788cd3e25
11 changed files with 1336 additions and 1177 deletions
+282 -370
View File
@@ -8,7 +8,7 @@ use pdfium_render::prelude::*;
use serde::Deserialize;
use serde_json::{json, Map, Value};
use tokio::task;
use tracing::{error, info, warn};
use tracing::{info, warn};
use uuid::Uuid;
use crate::{
@@ -27,16 +27,20 @@ use crate::{
use super::{
analyze::determine_thumbnail_support,
common::{load_document_version, load_version_assets},
fetch_version_object, handle_fetch_error, JobExecution, JobHandler,
job_execution_from_task_error,
taskflow::{
document::DocumentVersionTaskContext, BoxedTask, Task, TaskContext, TaskError,
TaskExecutor, TaskPlanner, TaskResult,
},
JobExecution, JobHandler,
};
const THUMBNAIL_WIDTH: u32 = 512;
const THUMBNAIL_HEIGHT: u32 = 512;
pub const THUMBNAIL_WIDTH: u32 = 512;
pub const THUMBNAIL_HEIGHT: u32 = 512;
const PREVIEW_WIDTH: u32 = THUMBNAIL_WIDTH * 4;
const PREVIEW_HEIGHT: u32 = THUMBNAIL_HEIGHT * 4;
const THUMBNAIL_ASSET_TYPE: &str = "thumbnail";
const PREVIEW_ASSET_TYPE: &str = "preview";
pub const THUMBNAIL_ASSET_TYPE: &str = "thumbnail";
pub const PREVIEW_ASSET_TYPE: &str = "preview";
#[derive(Debug, Deserialize, Clone)]
struct ThumbnailPayload {
@@ -75,268 +79,109 @@ impl JobHandler for GenerateThumbnailsJob {
}
};
let state_clone = state.clone();
let tenant_id = job.tenant_id;
let payload_for_context = payload.clone();
let initial = match task::spawn_blocking(move || {
load_thumbnail_context(state_clone, tenant_id, payload_for_context)
})
.await
{
Ok(Ok(ctx)) => ctx,
Ok(Err(err)) => {
warn!(job_id = %job.id, error = %err, "thumbnail job will retry");
return JobExecution::Retry {
delay: Duration::from_secs(30),
error: err,
};
}
Err(join_err) => {
error!(job_id = %job.id, error = %join_err, "thumbnail task panicked");
return JobExecution::Retry {
delay: Duration::from_secs(60),
error: format!("worker panicked: {join_err}"),
};
}
};
let mut context = DocumentVersionTaskContext::new(
job.id,
JOB_GENERATE_THUMBNAILS,
job.tenant_id,
payload.document_id,
payload.document_version_id,
payload.force,
state.config.worker_max_document_bytes,
state.clone(),
storage,
);
if initial.skip {
info!(job_id = %job.id, "thumbnails already exist; skipping");
return JobExecution::Success;
let planner = ThumbnailPlanner {
force: payload.force,
};
match TaskExecutor::run(&planner, &mut context).await {
Ok(()) => JobExecution::Success,
Err(err) => job_execution_from_task_error(err),
}
}
}
struct ThumbnailPlanner {
force: bool,
}
#[async_trait]
impl TaskPlanner<DocumentVersionTaskContext> for ThumbnailPlanner {
async fn plan(
&self,
_ctx: &mut DocumentVersionTaskContext,
) -> TaskResult<Vec<BoxedTask<DocumentVersionTaskContext>>> {
Ok(vec![Box::new(GenerateThumbnailsTask::new(self.force))])
}
}
pub struct GenerateThumbnailsTask {
force: bool,
}
impl GenerateThumbnailsTask {
pub fn new(force: bool) -> Self {
Self { force }
}
}
#[async_trait]
impl Task<DocumentVersionTaskContext> for GenerateThumbnailsTask {
fn name(&self) -> &'static str {
"generate-thumbnails"
}
async fn execute(&self, ctx: &mut DocumentVersionTaskContext) -> TaskResult<()> {
let context = build_thumbnail_context(ctx, self.force).await?;
if context.skip {
info!(job_id = %ctx.job_id(), "thumbnails already exist; skipping");
return Ok(());
}
let bytes = match fetch_version_object(
&initial.version,
&storage,
&initial.version.s3_key,
state.config.worker_max_document_bytes,
)
.await
{
Ok(bytes) => bytes,
Err(err) => return handle_fetch_error(&job, err, "thumbnail fetch failed; will retry"),
};
let generation = match generate_preview_and_thumbnail(&initial.document, &bytes) {
Ok(result) => result,
Err(err) => {
return JobExecution::Failed { error: err };
}
};
let bytes = ctx.buffered_object().await?;
let generation =
generate_preview_and_thumbnail(&context.document, bytes).map_err(TaskError::fail)?;
if let Some(page_count) = generation.page_count {
let state_clone = state.clone();
let document_id = initial.document.id;
let version_id = initial.version.id;
let tenant_id = initial.tenant_id;
match task::spawn_blocking(move || {
persist_document_page_count(
state_clone,
tenant_id,
document_id,
version_id,
page_count,
let state = ctx.state().clone();
let tenant_id = context.tenant_id;
let document_id = context.document.id;
let version_id = context.version.id;
task::spawn_blocking(move || {
persist_document_page_count(state, tenant_id, document_id, version_id, page_count)
})
.await
.map_err(|err| {
TaskError::retry(
Duration::from_secs(60),
format!("page count task panicked: {err}"),
)
})
.await
{
Ok(Ok(())) => {}
Ok(Err(err)) => {
warn!(
job_id = %job.id,
document_id = %document_id,
version_id = %version_id,
error = %err,
"failed to update document page count metadata; retrying"
);
return JobExecution::Retry {
delay: Duration::from_secs(30),
error: err,
};
}
Err(join_err) => {
error!(
job_id = %job.id,
document_id = %document_id,
version_id = %version_id,
error = %join_err,
"page count metadata task panicked"
);
return JobExecution::Retry {
delay: Duration::from_secs(60),
error: format!("metadata panic: {join_err}"),
};
}
}
})?
.map_err(|err| TaskError::retry(Duration::from_secs(30), err))?;
}
if let Some(existing_preview) = &initial.existing_preview {
for object in &initial.existing_preview_objects {
if let Err(err) = storage.delete_object(&object.s3_key).await {
warn!(
job_id = %job.id,
error = %err,
s3_key = %object.s3_key,
"failed to delete existing preview object"
);
}
}
let tenant_id = initial.document.tenant_id;
let asset_id = existing_preview.id;
let state_clone = state.clone();
match task::spawn_blocking(move || -> AppResult<()> {
let mut conn = state_clone.db_for_tenant(tenant_id)?;
delete_asset(&mut conn, tenant_id, asset_id)
})
.await
{
Ok(Ok(())) => {}
Ok(Err(err)) => {
warn!(
job_id = %job.id,
error = ?err,
asset_id = %asset_id,
"failed to remove preview metadata after deletion"
);
}
Err(join_err) => {
warn!(
job_id = %job.id,
error = %join_err,
asset_id = %asset_id,
"failed to remove preview metadata: task panicked"
);
}
}
}
if let Some(existing_thumbnail) = &initial.existing_thumbnail {
for object in &initial.existing_thumbnail_objects {
if let Err(err) = storage.delete_object(&object.s3_key).await {
warn!(
job_id = %job.id,
error = %err,
s3_key = %object.s3_key,
"failed to delete existing thumbnail object"
);
}
}
let tenant_id = initial.document.tenant_id;
let asset_id = existing_thumbnail.id;
let state_clone = state.clone();
match task::spawn_blocking(move || -> AppResult<()> {
let mut conn = state_clone.db_for_tenant(tenant_id)?;
delete_asset(&mut conn, tenant_id, asset_id)
})
.await
{
Ok(Ok(())) => {}
Ok(Err(err)) => {
warn!(
job_id = %job.id,
error = ?err,
asset_id = %asset_id,
"failed to remove thumbnail metadata after deletion"
);
}
Err(join_err) => {
warn!(
job_id = %job.id,
error = %join_err,
asset_id = %asset_id,
"failed to remove thumbnail metadata: task panicked"
);
}
}
}
remove_existing_assets(ctx, &context).await;
let preview_asset_id = Uuid::new_v4();
let thumbnail_asset_id = Uuid::new_v4();
let mut preview_objects: Vec<AssetObjectPersistence> =
Vec::with_capacity(generation.preview.objects.len());
for (index, image) in generation.preview.objects.iter().enumerate() {
if index + 1 > i32::MAX as usize {
return JobExecution::Failed {
error: "too many preview objects".to_string(),
};
}
let ordinal = (index + 1) as i32;
let s3_key = document_asset_object_key(
initial.document.id,
initial.version.version_number,
PREVIEW_ASSET_TYPE,
preview_asset_id,
ordinal,
);
if let Err(err) = storage
.put_object(
&s3_key,
image.image_bytes.clone(),
Some("image/webp".into()),
None,
)
.await
{
warn!(job_id = %job.id, error = %err, ordinal, "failed to upload preview; retrying");
return JobExecution::Retry {
delay: Duration::from_secs(30),
error: err.to_string(),
};
}
preview_objects.push(AssetObjectPersistence {
ordinal,
s3_key,
width: image.width,
height: image.height,
});
}
let mut thumbnail_objects: Vec<AssetObjectPersistence> =
Vec::with_capacity(generation.thumbnail.objects.len());
for (index, image) in generation.thumbnail.objects.iter().enumerate() {
if index + 1 > i32::MAX as usize {
return JobExecution::Failed {
error: "too many thumbnail objects".to_string(),
};
}
let ordinal = (index + 1) as i32;
let s3_key = document_asset_object_key(
initial.document.id,
initial.version.version_number,
THUMBNAIL_ASSET_TYPE,
thumbnail_asset_id,
ordinal,
);
if let Err(err) = storage
.put_object(
&s3_key,
image.image_bytes.clone(),
Some("image/webp".into()),
None,
)
.await
{
warn!(job_id = %job.id, error = %err, ordinal, "failed to upload thumbnail; retrying");
return JobExecution::Retry {
delay: Duration::from_secs(30),
error: err.to_string(),
};
}
thumbnail_objects.push(AssetObjectPersistence {
ordinal,
s3_key,
width: image.width,
height: image.height,
});
}
let preview_objects = upload_generated_objects(
ctx,
&context,
PREVIEW_ASSET_TYPE,
preview_asset_id,
&generation.preview,
)
.await?;
let thumbnail_objects = upload_generated_objects(
ctx,
&context,
THUMBNAIL_ASSET_TYPE,
thumbnail_asset_id,
&generation.thumbnail,
)
.await?;
let asset_persistences = vec![
AssetPersistence {
@@ -351,33 +196,186 @@ impl JobHandler for GenerateThumbnailsJob {
},
];
let state_clone = state.clone();
match task::spawn_blocking(move || {
persist_assets_metadata(state_clone, &initial, &asset_persistences)
let state = ctx.state().clone();
let tenant_id = context.document.tenant_id;
let version_id = context.version.id;
task::spawn_blocking(move || {
persist_assets_metadata(state, tenant_id, version_id, &asset_persistences)
})
.await
{
Ok(Ok(())) => {}
Ok(Err(err)) => {
warn!(job_id = %job.id, error = %err, "failed to persist thumbnail metadata; retrying");
return JobExecution::Retry {
delay: Duration::from_secs(30),
error: err,
};
}
Err(join_err) => {
error!(job_id = %job.id, error = %join_err, "thumbnail metadata update panicked");
return JobExecution::Retry {
delay: Duration::from_secs(30),
error: format!("metadata update panic: {join_err}"),
};
}
}
.map_err(|err| {
TaskError::retry(
Duration::from_secs(60),
format!("thumbnail metadata task panicked: {err}"),
)
})?
.map_err(|err| TaskError::retry(Duration::from_secs(30), err))?;
JobExecution::Success
ctx.invalidate_asset_cache();
Ok(())
}
}
async fn build_thumbnail_context(
ctx: &mut DocumentVersionTaskContext,
force: bool,
) -> TaskResult<ThumbnailContext> {
let document = ctx.document().await?.clone();
let version = ctx.version().await?.clone();
let tenant_id = ctx.tenant_id();
let (supported, _) = determine_thumbnail_support(&document);
if !supported {
return Ok(ThumbnailContext {
document,
version,
existing_thumbnail: None,
existing_thumbnail_objects: Vec::new(),
existing_preview: None,
existing_preview_objects: Vec::new(),
skip: true,
tenant_id,
});
}
let assets = ctx.assets().await?;
let (existing_thumbnail, existing_thumbnail_objects) = assets
.get(THUMBNAIL_ASSET_TYPE)
.map(|entry| (Some(entry.asset.clone()), entry.objects.clone()))
.unwrap_or((None, Vec::new()));
let (existing_preview, existing_preview_objects) = assets
.get(PREVIEW_ASSET_TYPE)
.map(|entry| (Some(entry.asset.clone()), entry.objects.clone()))
.unwrap_or((None, Vec::new()));
let expected_cardinality = expected_asset_cardinality(&document, &version);
let preview_cardinality = existing_preview
.as_ref()
.and_then(|asset| asset.cardinality)
.unwrap_or_else(|| existing_preview_objects.len() as i32);
let thumbnail_cardinality = existing_thumbnail
.as_ref()
.and_then(|asset| asset.cardinality)
.unwrap_or_else(|| existing_thumbnail_objects.len() as i32);
let needs_regeneration = preview_cardinality < expected_cardinality
|| thumbnail_cardinality < expected_cardinality
|| (existing_preview_objects.len() as i32) < expected_cardinality
|| (existing_thumbnail_objects.len() as i32) < expected_cardinality;
let skip =
existing_thumbnail.is_some() && existing_preview.is_some() && !force && !needs_regeneration;
Ok(ThumbnailContext {
document,
version,
existing_thumbnail,
existing_thumbnail_objects,
existing_preview,
existing_preview_objects,
skip,
tenant_id,
})
}
async fn remove_existing_assets(ctx: &DocumentVersionTaskContext, context: &ThumbnailContext) {
if let Some(existing_preview) = &context.existing_preview {
delete_asset_with_objects(ctx, existing_preview, &context.existing_preview_objects).await;
}
if let Some(existing_thumbnail) = &context.existing_thumbnail {
delete_asset_with_objects(ctx, existing_thumbnail, &context.existing_thumbnail_objects)
.await;
}
}
async fn delete_asset_with_objects(
ctx: &DocumentVersionTaskContext,
asset: &DocumentAsset,
objects: &[DocumentAssetObject],
) {
for object in objects {
if let Err(err) = ctx.storage().delete_object(&object.s3_key).await {
warn!(
job_id = %ctx.job_id(),
error = %err,
s3_key = %object.s3_key,
"failed to delete existing asset object"
);
}
}
let tenant_id = ctx.tenant_id();
let asset_id = asset.id;
let state = ctx.state().clone();
match task::spawn_blocking(move || -> AppResult<()> {
let mut conn = state.db_for_tenant(tenant_id)?;
delete_asset(&mut conn, tenant_id, asset_id)
})
.await
{
Ok(Ok(())) => {}
Ok(Err(err)) => {
warn!(
job_id = %ctx.job_id(),
asset_id = %asset_id,
error = ?err,
"failed to delete asset metadata"
);
}
Err(join_err) => {
warn!(
job_id = %ctx.job_id(),
asset_id = %asset_id,
error = %join_err,
"failed to delete asset metadata task panicked"
);
}
}
}
async fn upload_generated_objects(
ctx: &DocumentVersionTaskContext,
context: &ThumbnailContext,
asset_type: &str,
asset_id: Uuid,
asset: &GeneratedAsset,
) -> TaskResult<Vec<AssetObjectPersistence>> {
let mut objects = Vec::with_capacity(asset.objects.len());
for (index, image) in asset.objects.iter().enumerate() {
if index + 1 > i32::MAX as usize {
return Err(TaskError::fail("too many generated asset objects"));
}
let ordinal = (index + 1) as i32;
let s3_key = document_asset_object_key(
context.document.id,
context.version.version_number,
asset_type,
asset_id,
ordinal,
);
ctx.storage()
.put_object(
&s3_key,
image.image_bytes.clone(),
Some("image/webp".into()),
None,
)
.await
.map_err(|err| TaskError::retry(Duration::from_secs(30), err.to_string()))?;
objects.push(AssetObjectPersistence {
ordinal,
s3_key,
width: image.width,
height: image.height,
});
}
Ok(objects)
}
struct ThumbnailContext {
document: Document,
version: DocumentVersion,
@@ -418,83 +416,11 @@ struct AssetPersistence {
objects: Vec<AssetObjectPersistence>,
}
fn load_thumbnail_context(
state: Arc<AppState>,
tenant_id: Uuid,
payload: ThumbnailPayload,
) -> Result<ThumbnailContext, String> {
let base = load_document_version(
state.as_ref(),
tenant_id,
payload.document_id,
payload.document_version_id,
)?;
let mut conn = state
.db_for_tenant(base.tenant_id)
.map_err(|err| format!("{err:?}"))?;
let mut assets = load_version_assets(
&mut conn,
base.tenant_id,
base.version.id,
&[THUMBNAIL_ASSET_TYPE, PREVIEW_ASSET_TYPE],
)?;
let (existing_thumbnail, existing_thumbnail_objects) = assets
.remove(THUMBNAIL_ASSET_TYPE)
.map(|entry| (Some(entry.asset), entry.objects))
.unwrap_or((None, Vec::new()));
let (existing_preview, existing_preview_objects) = assets
.remove(PREVIEW_ASSET_TYPE)
.map(|entry| (Some(entry.asset), entry.objects))
.unwrap_or((None, Vec::new()));
let (supported, _) = determine_thumbnail_support(&base.document);
if !supported {
return Err("thumbnail generation not supported for this document".into());
}
let expected_cardinality = expected_asset_cardinality(&base.document, &base.version);
let preview_cardinality = existing_preview
.as_ref()
.and_then(|asset| asset.cardinality)
.unwrap_or_else(|| existing_preview_objects.len() as i32);
let thumbnail_cardinality = existing_thumbnail
.as_ref()
.and_then(|asset| asset.cardinality)
.unwrap_or_else(|| existing_thumbnail_objects.len() as i32);
let needs_regeneration = preview_cardinality < expected_cardinality
|| thumbnail_cardinality < expected_cardinality
|| (existing_preview_objects.len() as i32) < expected_cardinality
|| (existing_thumbnail_objects.len() as i32) < expected_cardinality;
let skip = existing_thumbnail.is_some()
&& existing_preview.is_some()
&& !payload.force
&& !needs_regeneration;
Ok(ThumbnailContext {
document: base.document,
version: base.version,
existing_thumbnail,
existing_thumbnail_objects,
existing_preview,
existing_preview_objects,
skip,
tenant_id: base.tenant_id,
})
}
fn generate_preview_and_thumbnail(
document: &Document,
bytes: &[u8],
) -> Result<GeneratedAssets, String> {
let is_pdf = document_is_pdf(document);
if is_pdf {
if document_is_pdf(document) {
let pdf_assets = generate_pdf_assets(bytes)?;
Ok(GeneratedAssets {
preview: pdf_assets.preview,
@@ -623,28 +549,14 @@ fn encode_dynamic_image(image: image::DynamicImage) -> Result<GeneratedImage, St
fn persist_assets_metadata(
state: Arc<AppState>,
context: &ThumbnailContext,
tenant_id: Uuid,
version_id: Uuid,
assets: &[AssetPersistence],
) -> Result<(), String> {
let tenant_id = context.document.tenant_id;
let mut conn = state
.db_for_tenant(tenant_id)
.map_err(|err| format!("{err:?}"))?;
if let Some(existing_preview) = &context.existing_preview {
diesel::delete(document_assets::table.filter(document_assets::id.eq(existing_preview.id)))
.execute(&mut conn)
.map_err(|err| format!("{err:?}"))?;
}
if let Some(existing_thumbnail) = &context.existing_thumbnail {
diesel::delete(
document_assets::table.filter(document_assets::id.eq(existing_thumbnail.id)),
)
.execute(&mut conn)
.map_err(|err| format!("{err:?}"))?;
}
for asset in assets {
if asset.objects.is_empty() {
return Err(format!(
@@ -661,7 +573,7 @@ fn persist_assets_metadata(
let new_asset = NewDocumentAsset {
id: asset.asset_id,
document_version_id: context.version.id,
document_version_id: version_id,
asset_type: asset.asset_type.to_string(),
mime_type: "image/webp".to_string(),
metadata: json!({
@@ -688,7 +600,7 @@ fn persist_assets_metadata(
diesel::delete(
document_asset_objects::table
.filter(document_asset_objects::asset_id.eq(asset.asset_id))
.filter(document_asset_objects::asset_id.eq(new_asset.id))
.filter(document_asset_objects::tenant_id.eq(tenant_id)),
)
.execute(&mut conn)
@@ -707,7 +619,7 @@ fn persist_assets_metadata(
let new_object = NewDocumentAssetObject {
id: Uuid::new_v4(),
asset_id: asset.asset_id,
asset_id: new_asset.id,
ordinal: object.ordinal,
s3_key: object.s3_key.clone(),
metadata: object_metadata,