backend dry
This commit is contained in:
@@ -18,7 +18,13 @@ use crate::{
|
||||
storage::TenantStorage,
|
||||
};
|
||||
|
||||
use super::{ocr::OCR_TEXT_ASSET_TYPE, JobExecution, JobHandler};
|
||||
use super::{
|
||||
fetch_version_object,
|
||||
handle_fetch_error,
|
||||
ocr::OCR_TEXT_ASSET_TYPE,
|
||||
JobExecution,
|
||||
JobHandler,
|
||||
};
|
||||
|
||||
#[derive(Debug, Deserialize)]
|
||||
struct IndexPayload {
|
||||
@@ -114,21 +120,23 @@ impl JobHandler for IndexDocumentTextJob {
|
||||
}
|
||||
|
||||
let s3_key = context.text_s3_key.unwrap();
|
||||
let text = match storage.get_object(&s3_key).await {
|
||||
Ok(bytes) => match String::from_utf8(bytes) {
|
||||
Ok(text) => text,
|
||||
Err(err) => {
|
||||
warn!(job_id = %job.id, error = %err, "ocr text not valid UTF-8");
|
||||
return JobExecution::Failed {
|
||||
error: "ocr text not valid UTF-8".into(),
|
||||
};
|
||||
}
|
||||
},
|
||||
let bytes = match fetch_version_object(
|
||||
&context.version,
|
||||
&storage,
|
||||
&s3_key,
|
||||
state.config.worker_max_document_bytes,
|
||||
)
|
||||
.await
|
||||
{
|
||||
Ok(bytes) => bytes,
|
||||
Err(err) => return handle_fetch_error(job, err, "failed to download ocr text"),
|
||||
};
|
||||
let text = match String::from_utf8(bytes) {
|
||||
Ok(text) => text,
|
||||
Err(err) => {
|
||||
warn!(job_id = %job.id, error = %err, "failed to download ocr text");
|
||||
return JobExecution::Retry {
|
||||
delay: Duration::from_secs(30),
|
||||
error: err.to_string(),
|
||||
warn!(job_id = %job.id, error = %err, "ocr text not valid UTF-8");
|
||||
return JobExecution::Failed {
|
||||
error: "ocr text not valid UTF-8".into(),
|
||||
};
|
||||
}
|
||||
};
|
||||
|
||||
Reference in New Issue
Block a user