quickwit

backend/src/workers/ocr.rs

@@ -1,6 +1,5 @@
 use std::{
-    fmt,
-    fs,
+    fmt, fs,
     io::{ErrorKind, Write},
     process::Command,
     sync::Arc,
@@ -19,7 +18,7 @@ use tracing::{error, info, warn};
 use uuid::Uuid;
 
 use crate::{
-    jobs::JOB_GENERATE_OCR_TEXT,
+    jobs::{enqueue_job, JOB_GENERATE_OCR_TEXT, JOB_INDEX_DOCUMENT_TEXT},
     models::{Document, DocumentAsset, DocumentVersion, NewDocumentAsset},
     schema::{document_assets, document_versions, documents},
     state::AppState,
@@ -30,7 +29,7 @@ use super::{JobExecution, JobHandler};
 pub const OCR_TEXT_ASSET_TYPE: &str = "ocr-text";
 const MIN_TEXT_LENGTH: usize = 50;
 
-#[derive(Debug, Deserialize)]
+#[derive(Clone, Debug, Deserialize)]
 struct OcrPayload {
     document_id: Uuid,
     document_version_id: Uuid,
@@ -63,8 +62,10 @@ impl JobHandler for GenerateOcrTextJob {
         };
 
         let state_clone = state.clone();
+        let payload_clone = payload.clone();
         let context =
-            match task::spawn_blocking(move || load_ocr_context(state_clone, &payload)).await {
+            match task::spawn_blocking(move || load_ocr_context(state_clone, &payload_clone)).await
+            {
                 Ok(Ok(ctx)) => ctx,
                 Ok(Err(err)) => {
                     warn!(job_id = %job.id, error = %err, "ocr job will retry");
@@ -155,7 +156,15 @@ impl JobHandler for GenerateOcrTextJob {
         })
         .await
         {
-            Ok(Ok(())) => JobExecution::Success,
+            Ok(Ok(())) => {
+                if state.config.quickwit_endpoint.is_some() && state.config.quickwit_index.is_some()
+                {
+                    if let Err(err) = enqueue_index_job(&state, &payload) {
+                        warn!(job_id = %job.id, error = %err, "failed to enqueue index job");
+                    }
+                }
+                JobExecution::Success
+            }
             Ok(Err(err)) => {
                 warn!(job_id = %job.id, error = %err, "failed to persist ocr metadata");
                 JobExecution::Retry {
@@ -334,7 +343,8 @@ fn run_ocr(bytes: &[u8]) -> Result<Option<String>, OcrError> {
                 )));
             }
 
-            let text = fs::read_to_string(sidecar.path()).map_err(|err| OcrError::Failed(err.to_string()))?;
+            let text = fs::read_to_string(sidecar.path())
+                .map_err(|err| OcrError::Failed(err.to_string()))?;
             if text.trim().chars().count() >= MIN_TEXT_LENGTH {
                 Ok(Some(text))
             } else {
@@ -392,6 +402,21 @@ fn persist_ocr_metadata(
     Ok(())
 }
 
+fn enqueue_index_job(state: &AppState, payload: &OcrPayload) -> Result<(), String> {
+    let mut conn = state.db().map_err(|err| format!("{err:?}"))?;
+    enqueue_job(
+        &mut conn,
+        JOB_INDEX_DOCUMENT_TEXT,
+        json!({
+            "document_id": payload.document_id,
+            "document_version_id": payload.document_version_id,
+        }),
+        None,
+    )
+    .map(|_| ())
+    .map_err(|err| err.to_string())
+}
+
 pub fn document_is_pdf(document: &Document) -> bool {
     document_meta_is_pdf(&PdfDocumentMeta {
         content_type: document.content_type.clone(),