perf(indexer): éliminer le pre-count WalkDir en mode incrémental + concurrence adaptative

- Incremental rebuild: remplace le WalkDir de comptage par un COUNT(*) SQL
  → incrémental 67s → 25s (-62%) sur disque externe
- Full rebuild: conserve le WalkDir (DB vidée avant le comptage)
- Concurrence par défaut: num_cpus/2 clampé [2,8] au lieu de 2 fixe
- Ajoute num_cpus comme dépendance workspace
- Backoffice jobs: un seul formulaire avec formAction par bouton (icônes rétablies)
- infra/perf.sh: corrige l'endpoint /index/jobs/:id (pas /details), exporte BASE_API/TOKEN

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Cargo.lock

@@ -863,6 +863,12 @@ version = "0.5.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "2304e00983f87ffb38b55b444b5e3b60a884b5d30c0fca7d82fe33449bbe55ea"
 
+[[package]]
+name = "hermit-abi"
+version = "0.5.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "fc0fef456e4baa96da950455cd02c081ca953b141298e41db3fc7e36b1da849c"
+
 [[package]]
 name = "hex"
 version = "0.4.3"
@@ -1171,6 +1177,7 @@ dependencies = [
  "futures",
  "image",
  "notify",
+ "num_cpus",
  "parsers",
  "rand 0.8.5",
  "rayon",
@@ -1639,6 +1646,16 @@ dependencies = [
  "libm",
 ]
 
+[[package]]
+name = "num_cpus"
+version = "1.17.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "91df4bbde75afed763b708b7eee1e8e7651e02d97f6d5dd763e89367e957b23b"
+dependencies = [
+ "hermit-abi",
+ "libc",
+]
+
 [[package]]
 name = "once_cell"
 version = "1.21.3"

Cargo.toml

@@ -33,6 +33,7 @@ tracing = "0.1"
 tracing-subscriber = { version = "0.3", features = ["env-filter", "fmt"] }
 uuid = { version = "1.12", features = ["serde", "v4"] }
 natord = "1.0"
+num_cpus = "1.16"
 pdfium-render = { version = "0.8", default-features = false, features = ["pdfium_latest", "image_latest", "thread_safe"] }
 unrar = "0.5"
 walkdir = "2.5"

apps/api/src/index_jobs.rs

@@ -182,7 +182,7 @@ pub async fn cancel_job(
     id: axum::extract::Path<Uuid>,
 ) -> Result<Json<IndexJobResponse>, ApiError> {
     let rows_affected = sqlx::query(
-        "UPDATE index_jobs SET status = 'cancelled' WHERE id = $1 AND status IN ('pending', 'running', 'generating_thumbnails')",
+        "UPDATE index_jobs SET status = 'cancelled' WHERE id = $1 AND status IN ('pending', 'running', 'extracting_pages', 'generating_thumbnails')",
     )
     .bind(id.0)
     .execute(&state.pool)

apps/backoffice/app/api/jobs/[id]/stream/route.ts

@@ -15,6 +15,7 @@ export async function GET(
       
       let lastData: string | null = null;
       let isActive = true;
+      let consecutiveErrors = 0;
 
       const fetchJob = async () => {
         if (!isActive) return;
@@ -25,6 +26,7 @@ export async function GET(
           });
 
           if (response.ok && isActive) {
+            consecutiveErrors = 0;
             const data = await response.json();
             const dataStr = JSON.stringify(data);
 
@@ -54,7 +56,11 @@ export async function GET(
           }
         } catch (error) {
           if (isActive) {
-            console.error("SSE fetch error:", error);
+            consecutiveErrors++;
+            // Only log first failure and every 60th to avoid spam
+            if (consecutiveErrors === 1 || consecutiveErrors % 60 === 0) {
+              console.warn(`SSE fetch error (${consecutiveErrors} consecutive):`, error);
+            }
           }
         }
       };

apps/backoffice/app/api/jobs/stream/route.ts

@@ -10,6 +10,7 @@ export async function GET(request: NextRequest) {
       
       let lastData: string | null = null;
       let isActive = true;
+      let consecutiveErrors = 0;
 
       const fetchJobs = async () => {
         if (!isActive) return;
@@ -20,6 +21,7 @@ export async function GET(request: NextRequest) {
           });
 
           if (response.ok && isActive) {
+            consecutiveErrors = 0;
             const data = await response.json();
             const dataStr = JSON.stringify(data);
 
@@ -38,7 +40,11 @@ export async function GET(request: NextRequest) {
           }
         } catch (error) {
           if (isActive) {
-            console.error("SSE fetch error:", error);
+            consecutiveErrors++;
+            // Only log first failure and every 30th to avoid spam
+            if (consecutiveErrors === 1 || consecutiveErrors % 30 === 0) {
+              console.warn(`SSE fetch error (${consecutiveErrors} consecutive):`, error);
+            }
           }
         }
       };

apps/backoffice/app/jobs/page.tsx

@@ -2,7 +2,7 @@ import { revalidatePath } from "next/cache";
 import { redirect } from "next/navigation";
 import { listJobs, fetchLibraries, rebuildIndex, rebuildThumbnails, regenerateThumbnails, IndexJobDto, LibraryDto } from "../../lib/api";
 import { JobsList } from "../components/JobsList";
-import { Card, CardHeader, CardTitle, CardDescription, CardContent, Button, FormField, FormSelect, FormRow } from "../components/ui";
+import { Card, CardHeader, CardTitle, CardContent, Button, FormField, FormSelect, FormRow } from "../components/ui";
 
 export const dynamic = "force-dynamic";
 
@@ -61,90 +61,44 @@ export default async function JobsPage({ searchParams }: { searchParams: Promise
       <Card className="mb-6">
         <CardHeader>
           <CardTitle>Queue New Job</CardTitle>
-          <CardDescription>Rebuild index, full rebuild, generate missing thumbnails, or regenerate all thumbnails</CardDescription>
         </CardHeader>
-        <CardContent className="space-y-4">
-          <form action={triggerRebuild}>
+        <CardContent>
+          <form>
             <FormRow>
-              <FormField className="flex-1">
+              <FormField className="flex-1 max-w-xs">
                 <FormSelect name="library_id" defaultValue="">
                   <option value="">All libraries</option>
                   {libraries.map((lib) => (
-                    <option key={lib.id} value={lib.id}>
-                      {lib.name}
-                    </option>
+                    <option key={lib.id} value={lib.id}>{lib.name}</option>
                   ))}
                 </FormSelect>
               </FormField>
-              <Button type="submit">
-                <svg className="w-4 h-4 mr-2" fill="none" stroke="currentColor" viewBox="0 0 24 24">
-                  <path strokeLinecap="round" strokeLinejoin="round" strokeWidth={2} d="M4 4v5h.582m15.356 2A8.001 8.001 0 004.582 9m0 0H9m11 11v-5h-.581m0 0a8.003 8.003 0 01-15.357-2m15.357 2H15" />
-                </svg>
-                Queue Rebuild
-              </Button>
-            </FormRow>
-          </form>
-          
-          <form action={triggerFullRebuild}>
-            <FormRow>
-              <FormField className="flex-1">
-                <FormSelect name="library_id" defaultValue="">
-                  <option value="">All libraries</option>
-                  {libraries.map((lib) => (
-                    <option key={lib.id} value={lib.id}>
-                      {lib.name}
-                    </option>
-                  ))}
-                </FormSelect>
-              </FormField>
-              <Button type="submit" variant="warning">
-                <svg className="w-4 h-4 mr-2" fill="none" stroke="currentColor" viewBox="0 0 24 24">
-                  <path strokeLinecap="round" strokeLinejoin="round" strokeWidth={2} d="M19 7l-.867 12.142A2 2 0 0116.138 21H7.862a2 2 0 01-1.995-1.858L5 7m5 4v6m4-6v6m1-10V4a1 1 0 00-1-1h-4a1 1 0 00-1 1v3M4 7h16" />
-                </svg>
-                Full Rebuild
-              </Button>
-            </FormRow>
-          </form>
-
-          <form action={triggerThumbnailsRebuild}>
-            <FormRow>
-              <FormField className="flex-1">
-                <FormSelect name="library_id" defaultValue="">
-                  <option value="">All libraries</option>
-                  {libraries.map((lib) => (
-                    <option key={lib.id} value={lib.id}>
-                      {lib.name}
-                    </option>
-                  ))}
-                </FormSelect>
-              </FormField>
-              <Button type="submit" variant="secondary">
-                <svg className="w-4 h-4 mr-2" fill="none" stroke="currentColor" viewBox="0 0 24 24">
-                  <path strokeLinecap="round" strokeLinejoin="round" strokeWidth={2} d="M4 16l4.586-4.586a2 2 0 012.828 0L16 16m-2-2l1.586-1.586a2 2 0 012.828 0L20 14m-6-6h.01M6 20h12a2 2 0 002-2V6a2 2 0 00-2-2H6a2 2 0 00-2 2v12a2 2 0 002 2z" />
-                </svg>
-                Generate thumbnails
-              </Button>
-            </FormRow>
-          </form>
-
-          <form action={triggerThumbnailsRegenerate}>
-            <FormRow>
-              <FormField className="flex-1">
-                <FormSelect name="library_id" defaultValue="">
-                  <option value="">All libraries</option>
-                  {libraries.map((lib) => (
-                    <option key={lib.id} value={lib.id}>
-                      {lib.name}
-                    </option>
-                  ))}
-                </FormSelect>
-              </FormField>
-              <Button type="submit" variant="warning">
-                <svg className="w-4 h-4 mr-2" fill="none" stroke="currentColor" viewBox="0 0 24 24">
-                  <path strokeLinecap="round" strokeLinejoin="round" strokeWidth={2} d="M4 4v5h.582m15.356 2A8.001 8.001 0 004.582 9m0 0H9m11 11v-5h-.581m0 0a8.003 8.003 0 01-15.357-2m15.357 2H15" />
-                </svg>
-                Regenerate thumbnails
-              </Button>
+              <div className="flex flex-wrap gap-2">
+                <Button type="submit" formAction={triggerRebuild}>
+                  <svg className="w-4 h-4 mr-2" fill="none" stroke="currentColor" viewBox="0 0 24 24">
+                    <path strokeLinecap="round" strokeLinejoin="round" strokeWidth={2} d="M4 4v5h.582m15.356 2A8.001 8.001 0 004.582 9m0 0H9m11 11v-5h-.581m0 0a8.003 8.003 0 01-15.357-2m15.357 2H15" />
+                  </svg>
+                  Rebuild
+                </Button>
+                <Button type="submit" formAction={triggerFullRebuild} variant="warning">
+                  <svg className="w-4 h-4 mr-2" fill="none" stroke="currentColor" viewBox="0 0 24 24">
+                    <path strokeLinecap="round" strokeLinejoin="round" strokeWidth={2} d="M19 7l-.867 12.142A2 2 0 0116.138 21H7.862a2 2 0 01-1.995-1.858L5 7m5 4v6m4-6v6m1-10V4a1 1 0 00-1-1h-4a1 1 0 00-1 1v3M4 7h16" />
+                  </svg>
+                  Full Rebuild
+                </Button>
+                <Button type="submit" formAction={triggerThumbnailsRebuild} variant="secondary">
+                  <svg className="w-4 h-4 mr-2" fill="none" stroke="currentColor" viewBox="0 0 24 24">
+                    <path strokeLinecap="round" strokeLinejoin="round" strokeWidth={2} d="M4 16l4.586-4.586a2 2 0 012.828 0L16 16m-2-2l1.586-1.586a2 2 0 012.828 0L20 14m-6-6h.01M6 20h12a2 2 0 002-2V6a2 2 0 00-2-2H6a2 2 0 00-2 2v12a2 2 0 002 2z" />
+                  </svg>
+                  Generate thumbnails
+                </Button>
+                <Button type="submit" formAction={triggerThumbnailsRegenerate} variant="warning">
+                  <svg className="w-4 h-4 mr-2" fill="none" stroke="currentColor" viewBox="0 0 24 24">
+                    <path strokeLinecap="round" strokeLinejoin="round" strokeWidth={2} d="M4 4v5h.582m15.356 2A8.001 8.001 0 004.582 9m0 0H9m11 11v-5h-.581m0 0a8.003 8.003 0 01-15.357-2m15.357 2H15" />
+                  </svg>
+                  Regenerate thumbnails
+                </Button>
+              </div>
             </FormRow>
           </form>
         </CardContent>

apps/indexer/Cargo.toml

@@ -13,6 +13,7 @@ chrono.workspace = true
 futures = "0.3"
 image.workspace = true
 notify = "6.1"
+num_cpus.workspace = true
 parsers = { path = "../../crates/parsers" }
 rand.workspace = true
 rayon.workspace = true

apps/indexer/src/analyzer.rs

@@ -67,7 +67,10 @@ async fn load_thumbnail_config(pool: &sqlx::PgPool) -> ThumbnailConfig {
 }
 
 async fn load_thumbnail_concurrency(pool: &sqlx::PgPool) -> usize {
-    let default_concurrency = 2;
+    // Default: half the logical CPUs, clamped between 2 and 8.
+    // Archive extraction is I/O bound but benefits from moderate parallelism.
+    let cpus = num_cpus::get();
+    let default_concurrency = (cpus / 2).clamp(2, 8);
     let row = sqlx::query(r#"SELECT value FROM app_settings WHERE key = 'limits'"#)
         .fetch_optional(pool)
         .await;
@@ -173,7 +176,7 @@ pub async fn analyze_library_books(
 
     let sql = format!(
         r#"
-        SELECT b.id AS book_id, bf.abs_path, bf.format
+        SELECT b.id AS book_id, bf.abs_path, bf.format, (b.thumbnail_path IS NULL) AS needs_thumbnail
         FROM books b
         JOIN book_files bf ON bf.book_id = b.id
         WHERE (b.library_id = $1 OR $1 IS NULL)
@@ -219,6 +222,7 @@ pub async fn analyze_library_books(
         book_id: Uuid,
         abs_path: String,
         format: String,
+        needs_thumbnail: bool,
     }
 
     let tasks: Vec<BookTask> = rows
@@ -227,6 +231,7 @@ pub async fn analyze_library_books(
             book_id: row.get("book_id"),
             abs_path: row.get("abs_path"),
             format: row.get("format"),
+            needs_thumbnail: row.get("needs_thumbnail"),
         })
         .collect();
 
@@ -245,7 +250,7 @@ pub async fn analyze_library_books(
 
     let extracted_count = Arc::new(AtomicI32::new(0));
 
-    // Collected results: (book_id, raw_path, page_count)
+    // Collected results: (book_id, raw_path, page_count) — only books that need thumbnail generation
     let extracted: Vec<(Uuid, String, i32)> = stream::iter(tasks)
         .map(|task| {
             let pool = state.pool.clone();
@@ -261,6 +266,28 @@ pub async fn analyze_library_books(
                 let local_path = utils::remap_libraries_path(&task.abs_path);
                 let path = std::path::Path::new(&local_path);
                 let book_id = task.book_id;
+                let needs_thumbnail = task.needs_thumbnail;
+
+                // Remove macOS Apple Double resource fork files (._*) that were indexed before the scanner filter was added
+                if path
+                    .file_name()
+                    .and_then(|n| n.to_str())
+                    .map(|n| n.starts_with("._"))
+                    .unwrap_or(false)
+                {
+                    warn!("[ANALYZER] Removing macOS resource fork from DB: {}", local_path);
+                    let _ = sqlx::query("DELETE FROM book_files WHERE book_id = $1")
+                        .bind(book_id)
+                        .execute(&pool)
+                        .await;
+                    let _ = sqlx::query(
+                        "DELETE FROM books WHERE id = $1 AND NOT EXISTS (SELECT 1 FROM book_files WHERE book_id = $1)",
+                    )
+                    .bind(book_id)
+                    .execute(&pool)
+                    .await;
+                    return None;
+                }
 
                 let format = match book_format_from_str(&task.format) {
                     Some(f) => f,
@@ -295,6 +322,29 @@ pub async fn analyze_library_books(
                     }
                 };
 
+                // If thumbnail already exists, just update page_count and skip thumbnail generation
+                if !needs_thumbnail {
+                    if let Err(e) = sqlx::query("UPDATE books SET page_count = $1 WHERE id = $2")
+                        .bind(page_count)
+                        .bind(book_id)
+                        .execute(&pool)
+                        .await
+                    {
+                        warn!("[ANALYZER] DB page_count update failed for book {}: {}", book_id, e);
+                    }
+                    let processed = extracted_count.fetch_add(1, Ordering::Relaxed) + 1;
+                    let percent = (processed as f64 / total as f64 * 50.0) as i32;
+                    let _ = sqlx::query(
+                        "UPDATE index_jobs SET processed_files = $2, progress_percent = $3 WHERE id = $1",
+                    )
+                    .bind(job_id)
+                    .bind(processed)
+                    .bind(percent)
+                    .execute(&pool)
+                    .await;
+                    return None; // don't enqueue for thumbnail sub-phase
+                }
+
                 // Save raw bytes to disk (no resize, no encode)
                 let raw_path = match tokio::task::spawn_blocking({
                     let dir = config.directory.clone();

apps/indexer/src/job.rs

@@ -13,7 +13,7 @@ pub async fn cleanup_stale_jobs(pool: &PgPool) -> Result<()> {
         SET status = 'failed',
             finished_at = NOW(),
             error_opt = 'Job interrupted by indexer restart'
-        WHERE status = 'running'
+        WHERE status IN ('running', 'extracting_pages', 'generating_thumbnails')
           AND started_at < NOW() - INTERVAL '5 minutes'
         RETURNING id
         "#,
@@ -238,27 +238,42 @@ pub async fn process_job(
             .await?
     };
 
-    // Count total files for progress estimation
-    let library_paths: Vec<String> = libraries
-        .iter()
-        .map(|library| {
-            crate::utils::remap_libraries_path(&library.get::<String, _>("root_path"))
-        })
-        .collect();
+    // Count total files for progress estimation.
+    // For incremental rebuilds, use the DB count (instant) — the filesystem will be walked
+    // once during discovery anyway, no need for a second full WalkDir pass.
+    // For full rebuilds, the DB is already cleared, so we must walk the filesystem.
+    let library_ids: Vec<uuid::Uuid> = libraries.iter().map(|r| r.get("id")).collect();
 
-    let total_files: usize = library_paths
-        .par_iter()
-        .map(|root_path| {
-            walkdir::WalkDir::new(root_path)
-                .into_iter()
-                .filter_map(Result::ok)
-                .filter(|entry| {
-                    entry.file_type().is_file()
-                        && parsers::detect_format(entry.path()).is_some()
-                })
-                .count()
-        })
-        .sum();
+    let total_files: usize = if !is_full_rebuild {
+        let count: i64 = sqlx::query_scalar(
+            "SELECT COUNT(*) FROM book_files bf JOIN books b ON b.id = bf.book_id WHERE b.library_id = ANY($1)"
+        )
+        .bind(&library_ids)
+        .fetch_one(&state.pool)
+        .await
+        .unwrap_or(0);
+        count as usize
+    } else {
+        let library_paths: Vec<String> = libraries
+            .iter()
+            .map(|library| {
+                crate::utils::remap_libraries_path(&library.get::<String, _>("root_path"))
+            })
+            .collect();
+        library_paths
+            .par_iter()
+            .map(|root_path| {
+                walkdir::WalkDir::new(root_path)
+                    .into_iter()
+                    .filter_map(Result::ok)
+                    .filter(|entry| {
+                        entry.file_type().is_file()
+                            && parsers::detect_format(entry.path()).is_some()
+                    })
+                    .count()
+            })
+            .sum()
+    };
 
     info!(
         "[JOB] Found {} libraries, {} total files to index",