推进 OCR UI 与 provider 删除验收

rust/crates/mnote-web/src/routes/local_ocr.rs

@@ -199,8 +199,13 @@ pub(crate) async fn create_job(
         match run_mineru_ocr(&root, &plan, token.unwrap_or_default()).await {
             Ok(markdown) => markdown,
             Err(error) => {
-                let failed_entry =
-                    build_index_entry(&plan, "failed", now, "", Some(redact_error(error.message())));
+                let failed_entry = build_index_entry(
+                    &plan,
+                    "failed",
+                    now,
+                    "",
+                    Some(redact_error(error.message())),
+                );
                 upsert_ocr_index_entry(&root, failed_entry)?;
                 return Err(error.with_context(&context));
             }
@@ -457,7 +462,11 @@ pub(crate) fn strip_ocr_frontmatter(markdown: &str) -> &str {
     trimmed[end + 4..].trim_start_matches('\n')
 }
 
-async fn run_mineru_ocr(root: &Path, plan: &OcrSidecarPlan, token: String) -> Result<String, WebError> {
+async fn run_mineru_ocr(
+    root: &Path,
+    plan: &OcrSidecarPlan,
+    token: String,
+) -> Result<String, WebError> {
     let config = MineruClientConfig {
         api_base_url: mineru_api_base_url(),
         token,
@@ -515,7 +524,10 @@ async fn create_mineru_upload_task(
         .map_err(|error| {
             WebError::bad_gateway_code(
                 "mineru_upload_task_failed",
-                format!("MinerU 上传任务创建失败: {}", redact_error(&error.to_string())),
+                format!(
+                    "MinerU 上传任务创建失败: {}",
+                    redact_error(&error.to_string())
+                ),
             )
         })?;
     let status = response.status();
@@ -527,8 +539,8 @@ async fn create_mineru_upload_task(
         ));
     }
     let value = parse_mineru_json(&body, "mineru_upload_task_json_invalid")?;
-    let batch_id = find_json_string_by_keys(&value, &["batch_id", "batchId", "id"])
-        .ok_or_else(|| {
+    let batch_id =
+        find_json_string_by_keys(&value, &["batch_id", "batchId", "id"]).ok_or_else(|| {
             WebError::bad_gateway_code(
                 "mineru_upload_task_batch_missing",
                 "MinerU 上传任务响应缺少 batch_id",
@@ -559,7 +571,10 @@ async fn upload_mineru_source(
         .map_err(|error| {
             WebError::bad_gateway_code(
                 "mineru_source_upload_failed",
-                format!("MinerU 源文件上传失败: {}", redact_error(&error.to_string())),
+                format!(
+                    "MinerU 源文件上传失败: {}",
+                    redact_error(&error.to_string())
+                ),
             )
         })?;
     if !response.status().is_success() {
@@ -610,7 +625,14 @@ async fn poll_mineru_result_zip_url(
         }
         if let Some(zip_url) = find_json_string_by_keys(
             &value,
-            &["full_zip_url", "fullZipUrl", "zip_url", "zipUrl", "result_url", "resultUrl"],
+            &[
+                "full_zip_url",
+                "fullZipUrl",
+                "zip_url",
+                "zipUrl",
+                "result_url",
+                "resultUrl",
+            ],
         ) {
             return Ok(zip_url);
         }
@@ -629,7 +651,10 @@ async fn download_mineru_result_zip(
     let response = client.get(zip_url).send().await.map_err(|error| {
         WebError::bad_gateway_code(
             "mineru_result_download_failed",
-            format!("MinerU 结果包下载失败: {}", redact_error(&error.to_string())),
+            format!(
+                "MinerU 结果包下载失败: {}",
+                redact_error(&error.to_string())
+            ),
         )
     })?;
     if !response.status().is_success() {
@@ -638,12 +663,19 @@ async fn download_mineru_result_zip(
             format!("MinerU 结果包下载失败: HTTP {}", response.status().as_u16()),
         ));
     }
-    response.bytes().await.map(|bytes| bytes.to_vec()).map_err(|error| {
-        WebError::bad_gateway_code(
-            "mineru_result_download_failed",
-            format!("MinerU 结果包读取失败: {}", redact_error(&error.to_string())),
-        )
-    })
+    response
+        .bytes()
+        .await
+        .map(|bytes| bytes.to_vec())
+        .map_err(|error| {
+            WebError::bad_gateway_code(
+                "mineru_result_download_failed",
+                format!(
+                    "MinerU 结果包读取失败: {}",
+                    redact_error(&error.to_string())
+                ),
+            )
+        })
 }
 
 fn extract_mineru_markdown_from_zip(bytes: &[u8]) -> Result<String, WebError> {
@@ -678,7 +710,8 @@ fn extract_mineru_markdown_from_zip(bytes: &[u8]) -> Result<String, WebError> {
     candidates
         .into_iter()
         .max_by_key(|(name, markdown)| {
-            let preferred = name.ends_with("/full.md") || name == "full.md" || name.ends_with("/result.md");
+            let preferred =
+                name.ends_with("/full.md") || name == "full.md" || name.ends_with("/result.md");
             (preferred, markdown.len())
         })
         .map(|(_, markdown)| markdown)
@@ -1307,9 +1340,7 @@ mod tests {
             writer
                 .start_file("full.md", zip::write::SimpleFileOptions::default())
                 .expect("zip start file");
-            writer
-                .write_all(markdown.as_bytes())
-                .expect("zip markdown");
+            writer.write_all(markdown.as_bytes()).expect("zip markdown");
             writer.finish().expect("zip finish");
         }
         bytes.into_inner()
@@ -1570,7 +1601,9 @@ mod tests {
                 }),
             );
         let mock_handle = tokio::spawn(async move {
-            axum::serve(listener, mock_mineru).await.expect("mock mineru server");
+            axum::serve(listener, mock_mineru)
+                .await
+                .expect("mock mineru server");
         });
 
         let old_token = std::env::var("MNOTE_MINERU_API_TOKEN").ok();
@@ -1643,12 +1676,9 @@ mod tests {
         assert_eq!(payload["job"]["provider"].as_str(), Some("mineru"));
         assert_eq!(upload_count.load(Ordering::SeqCst), 1);
         assert_eq!(poll_count.load(Ordering::SeqCst), 1);
-        let sidecar = fs::read_to_string(
-            root.join("docs")
-                .join("Page.ocr")
-                .join("photo.png.ocr.md"),
-        )
-        .expect("sidecar");
+        let sidecar =
+            fs::read_to_string(root.join("docs").join("Page.ocr").join("photo.png.ocr.md"))
+                .expect("sidecar");
         assert!(sidecar.contains("provider: mineru"));
         assert!(sidecar.contains("识别文本"));
 

rust/crates/mnote-web/src/routes/local_search_index.rs

@@ -185,6 +185,18 @@ pub(crate) fn refresh_local_search_index_for_path(
     index
         .resources
         .retain(|resource| resource.path != relative_path);
+    if local_ocr::is_local_ocr_sidecar_relative_path(&relative_path) {
+        index.built_at = now_ms();
+        write_local_search_index(root_path, &index)?;
+        return Ok(json!({
+            "version": index.version,
+            "rootUri": index.root_uri,
+            "workspaceId": index.workspace_id,
+            "builtAt": index.built_at,
+            "documentCount": index.documents.len(),
+            "resourceCount": index.resources.len()
+        }));
+    }
     let absolute_path = root_path.join(&relative_path);
     if absolute_path.exists() && absolute_path.is_file() && is_markdown_path(&absolute_path) {
         index
@@ -1168,6 +1180,11 @@ mod tests {
             "---\nmnote_ocr_version: 1\nprovider: mock\nmodel_version: vlm\nowner_document: ../Page.md\nsource_path: ./Page.assets/photo.png\nsource_root_relative_path: docs/Page.assets/photo.png\nsource_size: 3\nsource_mtime_ms: 1\nstatus: done\ncreated_at: 1\nupdated_at: 1\n---\n\nOCR-only-token 识别正文\n",
         )
         .expect("ocr markdown");
+        fs::write(
+            root.join("docs").join("Page.ocr").join("photo.png-704905.ocr.md"),
+            "---\nmnote_ocr_version: 1\nprovider: mock\nmodel_version: vlm\nowner_document: ../Page.md\nsource_path: ./Page.assets/photo.png\nsource_root_relative_path: docs/Page.assets/photo.png\nsource_size: 3\nsource_mtime_ms: 2\nstatus: done\ncreated_at: 2\nupdated_at: 2\n---\n\nOCR-hash-token 识别正文\n",
+        )
+        .expect("hashed ocr markdown");
         fs::create_dir_all(root.join(".mnote")).expect("mnote dir");
         fs::write(
             root.join(".mnote").join("ocr-index.json"),
@@ -1208,6 +1225,22 @@ mod tests {
         )
         .expect("without ocr");
         assert_eq!(without_ocr["results"].as_array().map(Vec::len), Some(0));
+        let without_hashed_ocr = query_local_search_index(
+            &root,
+            &root_uri,
+            workspace_id,
+            "OCR-hash-token",
+            None,
+            10,
+            false,
+            false,
+            false,
+        )
+        .expect("without hashed ocr");
+        assert_eq!(
+            without_hashed_ocr["results"].as_array().map(Vec::len),
+            Some(0)
+        );
 
         let with_ocr = query_local_search_index(
             &root,
@@ -1245,6 +1278,24 @@ mod tests {
             .documents
             .iter()
             .any(|document| document.path == "docs/Page.ocr/photo.png.ocr.md"));
+        assert!(!index
+            .documents
+            .iter()
+            .any(|document| document.path == "docs/Page.ocr/photo.png-704905.ocr.md"));
+        refresh_local_search_index_for_path(
+            &root,
+            &root_uri,
+            workspace_id,
+            "docs/Page.ocr/photo.png-704905.ocr.md",
+        )
+        .expect("refresh hashed ocr sidecar");
+        let refreshed_index = read_local_search_index(&root)
+            .expect("read refreshed search index")
+            .expect("refreshed search index");
+        assert!(!refreshed_index
+            .documents
+            .iter()
+            .any(|document| document.path == "docs/Page.ocr/photo.png-704905.ocr.md"));
         let _ = fs::remove_dir_all(&root);
     }