Merge pull request 'responsiveness' (#26) from responsiveness into main

Reviewed-on: http://gitea.lan/gpatti/MediaLore/pulls/26

.gitignore

@@ -10,3 +10,4 @@ medialore.db-wal
 tsconfig.tsbuildinfo
 .session_secret
 .vscode/
+*.traineddata

src/app/api/ai-settings/ocr/route.ts

@@ -0,0 +1,11 @@
+import { NextRequest, NextResponse } from 'next/server'
+import { requireAuth } from '@/lib/auth'
+import { getAiConfig } from '@/lib/app-settings'
+
+export async function GET(request: NextRequest) {
+  const auth = await requireAuth(request)
+  if (auth instanceof NextResponse) return auth
+
+  const { ocrMode, ocrLanguages } = getAiConfig()
+  return NextResponse.json({ ocrMode, ocrLanguages })
+}

src/app/api/ai-tagging/extract-text/route.ts

@@ -3,14 +3,14 @@ import { requireLibraryAccess } from '@/lib/auth'
 import { enqueueJob } from '@/lib/ai-jobs'
 
 export async function POST(request: NextRequest) {
-  let body: { itemKey?: string }
+  let body: { itemKey?: string; ocrLanguages?: string }
   try {
     body = await request.json()
   } catch {
     return NextResponse.json({ error: 'Invalid JSON body' }, { status: 400 })
   }
 
-  const { itemKey } = body
+  const { itemKey, ocrLanguages } = body
   if (!itemKey || typeof itemKey !== 'string') {
     return NextResponse.json({ error: 'itemKey is required' }, { status: 400 })
   }
@@ -19,6 +19,12 @@ export async function POST(request: NextRequest) {
   const auth = await requireLibraryAccess(request, libraryId)
   if (auth instanceof NextResponse) return auth
 
-  const jobId = enqueueJob(itemKey, 'extract', libraryId)
+  const jobId = enqueueJob(
+    itemKey,
+    'extract',
+    libraryId,
+    undefined,
+    ocrLanguages ? { ocrLanguages } : undefined,
+  )
   return NextResponse.json({ jobId }, { status: 202 })
 }

src/app/api/browse/route.ts

@@ -1,4 +1,5 @@
 import fs from 'fs'
+import path from 'path'
 import { NextRequest, NextResponse } from 'next/server'
 import { getLibrary, resolveLibraryRoot, resolveAndJail } from '@/lib/libraries'
 import { scanDirectory, scanDirectoryRecursive } from '@/lib/files'
@@ -31,6 +32,21 @@ export async function GET(request: NextRequest) {
   const listing = recursive
     ? scanDirectoryRecursive(root, libraryId, subpath)
     : scanDirectory(root, libraryId, subpath)
+
+  // Annotate image entries with whether they have extracted text
+  const db = getDb()
+  const rows = db
+    .prepare('SELECT item_key FROM media_items WHERE library_id = ? AND extracted_text IS NOT NULL')
+    .all(libraryId) as { item_key: string }[]
+  const withText = new Set(rows.map((r) => r.item_key))
+
+  listing.entries = listing.entries.map((e) => {
+    if (e.type !== 'file' || e.mediaType !== 'image') return e
+    const relPath = subpath ? path.join(subpath, e.name) : e.name
+    const itemKey = `${libraryId}:mixed_file:${encodeURIComponent(relPath)}`
+    return { ...e, hasExtractedText: withText.has(itemKey) }
+  })
+
   return NextResponse.json(listing)
 }
 

src/components/DoomScrollView.tsx

@@ -48,8 +48,10 @@ export default function DoomScrollView({ items, videoContext = 'mixed', onClose,
   const [showOriginal, setShowOriginal] = useState(false)
   const [extracting, setExtracting] = useState(false)
   const [extractError, setExtractError] = useState<string | null>(null)
+  const [extractPending, setExtractPending] = useState(false)
 
   const videoRef = useRef<HTMLVideoElement>(null)
+  const extractPollRef = useRef<ReturnType<typeof setInterval> | null>(null)
   const cooldownRef = useRef(false)
   const touchStartY = useRef<number | null>(null)
 
@@ -126,14 +128,19 @@ export default function DoomScrollView({ items, videoContext = 'mixed', onClose,
     return () => clearTimeout(id)
   }, [autoPlayEnabled, isPaused, autoPlaySeconds, current?.url, goNext])
 
-  // Fetch extracted text for current item
+  // Fetch extracted text for current item; clear any in-flight poll on item change
   useEffect(() => {
+    if (extractPollRef.current) {
+      clearInterval(extractPollRef.current)
+      extractPollRef.current = null
+    }
     setExtractedText(null)
     setTranslatedText(null)
     setShowTextOverlay(false)
     setShowOriginal(false)
     setExtracting(false)
     setExtractError(null)
+    setExtractPending(false)
     if (!current?.itemKey) return
     fetch(`/api/ai-tagging/fields?itemKey=${encodeURIComponent(current.itemKey)}`)
       .then((r) => r.json())
@@ -144,6 +151,13 @@ export default function DoomScrollView({ items, videoContext = 'mixed', onClose,
       .catch(() => {})
   }, [current?.itemKey])
 
+  // Clean up poll on unmount
+  useEffect(() => {
+    return () => {
+      if (extractPollRef.current) clearInterval(extractPollRef.current)
+    }
+  }, [])
+
   useEffect(() => {
     const handleKey = (e: KeyboardEvent) => {
       if (e.key === 'Escape') { onClose(); return }
@@ -184,23 +198,44 @@ export default function DoomScrollView({ items, videoContext = 'mixed', onClose,
 
   const handleExtractText = async () => {
     if (!current?.itemKey) return
+    const itemKey = current.itemKey
     setExtracting(true)
     setExtractError(null)
+    setExtractPending(false)
     try {
       const res = await fetch('/api/ai-tagging/extract-text', {
         method: 'POST',
         headers: { 'Content-Type': 'application/json' },
-        body: JSON.stringify({ itemKey: current.itemKey }),
+        body: JSON.stringify({ itemKey }),
       })
+      if (res.status === 202) {
+        // Job queued — poll until it completes (up to 5 min)
+        setExtractPending(true)
+        const deadline = Date.now() + 5 * 60 * 1000
+        extractPollRef.current = setInterval(async () => {
+          if (Date.now() > deadline) {
+            if (extractPollRef.current) clearInterval(extractPollRef.current)
+            setExtractPending(false)
+            return
+          }
+          try {
+            const r = await fetch(`/api/ai-tagging/fields?itemKey=${encodeURIComponent(itemKey)}`)
+            const data: { extractedText: string | null; extractedTextTranslated: string | null } = await r.json()
+            if (data.extractedText) {
+              if (extractPollRef.current) clearInterval(extractPollRef.current)
+              setExtractPending(false)
+              setExtractedText(data.extractedText)
+              setTranslatedText(data.extractedTextTranslated)
+              setShowTextOverlay(true)
+            }
+          } catch { /* ignore */ }
+        }, 2000)
+        return
+      }
       if (!res.ok) {
         const data = await res.json().catch(() => ({}))
         throw new Error((data as { error?: string }).error ?? 'Extraction failed')
       }
-      if (res.status === 202) {
-        setExtractError('Queued — check AI Integrations for progress')
-        setTimeout(() => setExtractError(null), 4000)
-        return
-      }
       const result = await res.json()
       setExtractedText(result.extractedText || null)
       setTranslatedText(result.translatedText || null)
@@ -301,7 +336,7 @@ export default function DoomScrollView({ items, videoContext = 'mixed', onClose,
       {/* Text overlay */}
       {showTextOverlay && displayText && (
         <div
-          className="absolute bottom-16 left-4 right-4 z-20 rounded-xl p-4"
+          className="absolute bottom-4 left-4 right-4 z-20 rounded-xl p-4 max-w-fit"
           style={{ backgroundColor: 'rgba(0,0,0,0.75)' }}
           onClick={(e) => e.stopPropagation()}
         >
@@ -371,15 +406,20 @@ export default function DoomScrollView({ items, videoContext = 'mixed', onClose,
           ) : current?.itemKey && current?.mediaType === 'image' ? (
             <button
               onClick={handleExtractText}
-              disabled={extracting}
+              disabled={extracting || extractPending}
               className="w-9 h-9 rounded-full flex items-center justify-center transition-opacity hover:opacity-100 opacity-70 disabled:opacity-40"
               style={{
-                backgroundColor: extractError ? 'rgba(127,29,29,0.8)' : 'rgba(0,0,0,0.5)',
+                backgroundColor: extractPending
+                  ? 'var(--accent)'
+                  : extractError
+                    ? 'rgba(127,29,29,0.8)'
+                    : 'rgba(0,0,0,0.5)',
                 color: extractError ? '#fca5a5' : '#fff',
               }}
-              aria-label="Extract text"
+              aria-label={extractPending ? 'Extracting text…' : 'Extract text'}
+              title={extractPending ? 'Queued — extracting text…' : extractError ?? 'Extract text'}
             >
-              {extracting ? (
+              {extracting || extractPending ? (
                 <span className="animate-spin" style={{ display: 'inline-block', fontSize: '0.75rem' }}>⟳</span>
               ) : (
                 <svg width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" strokeWidth="2" strokeLinecap="round" strokeLinejoin="round">

src/components/mixed/ImageLightbox.tsx

@@ -1,6 +1,6 @@
 'use client'
 
-import { useEffect, useRef, useState } from 'react'
+import { useEffect, useRef, useState, useCallback } from 'react'
 import TagSelector from '@/components/tags/TagSelector'
 
 interface Props {
@@ -25,16 +25,32 @@ export default function ImageLightbox({ url, name, onClose, onPrev, onNext, item
   const [extractedText, setExtractedText] = useState<string | null>(null)
   const [translatedText, setTranslatedText] = useState<string | null>(null)
   const [extracting, setExtracting] = useState(false)
+  const [extractPending, setExtractPending] = useState(false)
   const [extractError, setExtractError] = useState<string | null>(null)
   const [retranslating, setRetranslating] = useState(false)
+  const [translatePending, setTranslatePending] = useState(false)
   const [editedExtractedText, setEditedExtractedText] = useState<string>('')
   const [savingText, setSavingText] = useState(false)
   const [sourceLanguage, setSourceLanguage] = useState('')
 
+  // Description state (moved from TagSelector)
+  const [aiDescription, setAiDescription] = useState<string | null>(null)
+  const [generatingDesc, setGeneratingDesc] = useState(false)
+  const [descPending, setDescPending] = useState(false)
+  const [descError, setDescError] = useState<string | null>(null)
+
+  // OCR settings
+  const [ocrMode, setOcrMode] = useState<string | null>(null)
+  const [defaultOcrLanguages, setDefaultOcrLanguages] = useState('eng')
+  const [ocrLanguageInput, setOcrLanguageInput] = useState('')
+
   // Text overlay state
   const [showTextOverlay, setShowTextOverlay] = useState(false)
   const [showOriginal, setShowOriginal] = useState(false)
 
+  // Polling ref
+  const pollRef = useRef<ReturnType<typeof setInterval> | null>(null)
+
   // Determine if this is an image file (for text extraction controls)
   const isImage = /\.(jpe?g|png|gif|webp|bmp|tiff?)$/i.test(name)
 
@@ -42,18 +58,68 @@ export default function ImageLightbox({ url, name, onClose, onPrev, onNext, item
   const displayText = (translatedText && !showOriginal) ? translatedText : extractedText
 
   // Fetch existing AI fields on mount / item change
-  useEffect(() => {
-    if (!itemKey) return
-    fetch(`/api/ai-tagging/fields?itemKey=${encodeURIComponent(itemKey)}`)
+  const fetchAiFields = useCallback(() => {
+    if (!itemKey) return Promise.resolve()
+    return fetch(`/api/ai-tagging/fields?itemKey=${encodeURIComponent(itemKey)}`)
       .then((r) => r.json())
-      .then((data: { extractedText: string | null; extractedTextTranslated: string | null }) => {
+      .then((data: { extractedText: string | null; extractedTextTranslated: string | null; aiDescription: string | null }) => {
         setExtractedText(data.extractedText)
         setEditedExtractedText(data.extractedText ?? '')
         setTranslatedText(data.extractedTextTranslated)
+        setAiDescription(data.aiDescription)
       })
       .catch(() => {})
   }, [itemKey])
 
+  useEffect(() => {
+    fetchAiFields()
+    fetch('/api/ai-settings/ocr')
+      .then((r) => r.json())
+      .then((d: { ocrMode: string; ocrLanguages: string }) => {
+        setOcrMode(d.ocrMode)
+        setDefaultOcrLanguages(d.ocrLanguages)
+      })
+      .catch(() => {})
+    return () => {
+      if (pollRef.current) clearInterval(pollRef.current)
+    }
+  }, [fetchAiFields])
+
+  // Start polling fields every 2s until data changes or 5-min timeout
+  const startPolling = useCallback((snapshotText: string | null, snapshotTranslated: string | null, snapshotDesc: string | null) => {
+    if (!itemKey) return
+    if (pollRef.current) clearInterval(pollRef.current)
+    const deadline = Date.now() + 5 * 60 * 1000
+    pollRef.current = setInterval(async () => {
+      if (Date.now() > deadline) {
+        clearInterval(pollRef.current!)
+        pollRef.current = null
+        setExtractPending(false)
+        setTranslatePending(false)
+        setDescPending(false)
+        return
+      }
+      try {
+        const r = await fetch(`/api/ai-tagging/fields?itemKey=${encodeURIComponent(itemKey)}`)
+        const data: { extractedText: string | null; extractedTextTranslated: string | null; aiDescription: string | null } = await r.json()
+        const textChanged = data.extractedText !== snapshotText
+        const translationChanged = data.extractedTextTranslated !== snapshotTranslated
+        const descChanged = data.aiDescription !== snapshotDesc
+        if (textChanged || translationChanged || descChanged) {
+          clearInterval(pollRef.current!)
+          pollRef.current = null
+          setExtractedText(data.extractedText)
+          setEditedExtractedText(data.extractedText ?? '')
+          setTranslatedText(data.extractedTextTranslated)
+          setAiDescription(data.aiDescription)
+          setExtractPending(false)
+          setTranslatePending(false)
+          setDescPending(false)
+        }
+      } catch { /* ignore */ }
+    }, 2000)
+  }, [itemKey])
+
   useEffect(() => {
     const handleKey = (e: KeyboardEvent) => {
       if (e.key === 'Escape') onClose()
@@ -72,6 +138,38 @@ export default function ImageLightbox({ url, name, onClose, onPrev, onNext, item
     if (e.target === overlayRef.current) onClose()
   }
 
+  const handleGenerateDescription = async () => {
+    if (!itemKey) return
+    setGeneratingDesc(true)
+    setDescError(null)
+    setDescPending(false)
+    try {
+      const res = await fetch('/api/ai-tagging/describe', {
+        method: 'POST',
+        headers: { 'Content-Type': 'application/json' },
+        body: JSON.stringify({ itemKey }),
+      })
+      if (res.status === 202) {
+        setDescPending(true)
+        startPolling(extractedText, translatedText, aiDescription)
+        return
+      }
+      if (!res.ok) {
+        const data = await res.json().catch(() => ({}))
+        throw new Error((data as { error?: string }).error ?? 'Failed to generate description')
+      }
+      const { description } = await res.json()
+      setAiDescription(description)
+    } catch (err) {
+      setDescError(err instanceof Error ? err.message : 'Failed to generate description')
+      setTimeout(() => setDescError(null), 4000)
+    } finally {
+      setGeneratingDesc(false)
+    }
+  }
+
+  const smallBtn = 'w-7 h-7 rounded-full flex items-center justify-center transition-colors flex-shrink-0'
+
   return (
     <div
       ref={overlayRef}
@@ -79,13 +177,13 @@ export default function ImageLightbox({ url, name, onClose, onPrev, onNext, item
       style={{ backgroundColor: 'rgba(0,0,0,0.9)', height: '100vh', maxHeight: '100vh' }}
       onClick={handleOverlayClick}
     >
-      {/* Toolbar */}
+      {/* Toolbar — collapses to just filename + text overlay when tag panel is open */}
       <div className={`flex items-center justify-between w-full flex-shrink-0 ${showTags ? '' : 'max-w-4xl'}`}>
         <span className="text-sm truncate max-w-[80%]" style={{ color: 'var(--text-secondary)' }}>
           {name}
         </span>
         <div className="flex items-center gap-2 flex-shrink-0">
-          {/* Text overlay button — only shown when extracted text exists */}
+          {/* Text overlay button — always shown when text exists */}
           {extractedText && (
             <button
               onClick={(e) => { e.stopPropagation(); setShowTextOverlay((v) => !v) }}
@@ -110,22 +208,17 @@ export default function ImageLightbox({ url, name, onClose, onPrev, onNext, item
               </svg>
             </button>
           )}
+          {/* These buttons only show in the toolbar when the tag panel is closed */}
+          {!showTags && (
+            <>
               {itemKey && (
                 <button
-              onClick={(e) => { e.stopPropagation(); setShowTags((v) => !v) }}
+                  onClick={(e) => { e.stopPropagation(); setShowTags(true) }}
                   className="w-12 h-12 rounded-full flex items-center justify-center text-sm transition-colors"
-              style={{
-                backgroundColor: showTags ? 'var(--accent)' : 'var(--surface)',
-                color: showTags ? '#fff' : 'var(--text-primary)',
-                fontSize: '1.5rem',
-              }}
-              onMouseEnter={(e) => {
-                if (!showTags) (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--surface-hover)'
-              }}
-              onMouseLeave={(e) => {
-                if (!showTags) (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--surface)'
-              }}
-              aria-label={showTags ? 'Hide tags' : 'Show tags'}
+                  style={{ backgroundColor: 'var(--surface)', color: 'var(--text-primary)', fontSize: '1.5rem' }}
+                  onMouseEnter={(e) => (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--surface-hover)'}
+                  onMouseLeave={(e) => (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--surface)'}
+                  aria-label="Show tags"
                   title="Tags"
                 >
                   🏷
@@ -173,12 +266,14 @@ export default function ImageLightbox({ url, name, onClose, onPrev, onNext, item
                 onClick={onClose}
                 className="w-12 h-12 rounded-full flex items-center justify-center text-sm transition-colors"
                 style={{ backgroundColor: 'var(--surface)', color: 'var(--text-primary)', fontSize: '1.5rem' }}
-            onMouseEnter={(e) => ((e.currentTarget as HTMLElement).style.backgroundColor = 'var(--surface-hover)')}
-            onMouseLeave={(e) => ((e.currentTarget as HTMLElement).style.backgroundColor = 'var(--surface)')}
+                onMouseEnter={(e) => (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--surface-hover)'}
+                onMouseLeave={(e) => (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--surface)'}
                 aria-label="Close"
               >
                 ✕
               </button>
+            </>
+          )}
         </div>
       </div>
 
@@ -237,43 +332,149 @@ export default function ImageLightbox({ url, name, onClose, onPrev, onNext, item
               </div>
             )}
           </div>
+
           {/* Tag panel */}
           <div
-            className="w-80 h-full max-h-full flex-shrink-0 rounded-xl overflow-y-auto p-4"
+            className="w-80 h-full max-h-full flex-shrink-0 rounded-xl overflow-y-auto p-4 flex flex-col gap-4"
             style={{ backgroundColor: 'var(--surface)', border: '1px solid var(--border)' }}
             onClick={(e) => e.stopPropagation()}
           >
-            <p className="text-xs font-semibold uppercase tracking-wider mb-3" style={{ color: 'var(--text-secondary)' }}>
-              Tags
+            {/* Panel header — hide panel + AI tagger + close lightbox */}
+            <div className="flex items-center justify-between flex-shrink-0">
+              <button
+                onClick={() => setShowTags(false)}
+                className={smallBtn}
+                style={{ backgroundColor: 'var(--border)', color: 'var(--text-secondary)', fontSize: '0.85rem' }}
+                onMouseEnter={(e) => (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--text-secondary)'}
+                onMouseLeave={(e) => (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--border)'}
+                aria-label="Hide panel"
+                title="Hide panel"
+              >
+                ›
+              </button>
+              <div className="flex items-center gap-1.5">
+                {onAiTag && (
+                  <button
+                    onClick={async (e) => {
+                      e.stopPropagation()
+                      setAiTagging(true)
+                      setAiTagError(null)
+                      try {
+                        await onAiTag()
+                        setTagRefreshKey((k) => k + 1)
+                        onTagsChanged?.()
+                      } catch (err) {
+                        setAiTagError(err instanceof Error ? err.message : 'AI tagging failed')
+                        setTimeout(() => setAiTagError(null), 4000)
+                      } finally {
+                        setAiTagging(false)
+                      }
+                    }}
+                    disabled={aiTagging}
+                    className={`${smallBtn} disabled:opacity-50`}
+                    style={{
+                      backgroundColor: aiTagError ? '#7f1d1d' : 'var(--border)',
+                      color: aiTagError ? '#fca5a5' : 'var(--text-secondary)',
+                      fontSize: '1rem',
+                    }}
+                    onMouseEnter={(e) => {
+                      if (!aiTagging && !aiTagError) (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--text-secondary)'
+                    }}
+                    onMouseLeave={(e) => {
+                      if (!aiTagError) (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--border)'
+                    }}
+                    aria-label="AI Tag this image"
+                    title={aiTagError ?? (aiTagging ? 'Tagging…' : 'AI Tag')}
+                  >
+                    {aiTagging ? <span className="animate-spin" style={{ display: 'inline-block' }}>⟳</span> : '✨'}
+                  </button>
+                )}
+                <button
+                  onClick={onClose}
+                  className={smallBtn}
+                  style={{ backgroundColor: 'var(--border)', color: 'var(--text-secondary)', fontSize: '0.85rem' }}
+                  onMouseEnter={(e) => (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--text-secondary)'}
+                  onMouseLeave={(e) => (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--border)'}
+                  aria-label="Close"
+                  title="Close"
+                >
+                  ✕
+                </button>
+              </div>
+            </div>
+
+            {/* Description section */}
+            <div className="flex flex-col gap-1" style={{ borderTop: '1px solid var(--border)', paddingTop: '1rem' }}>
+              <p className="text-xs font-semibold uppercase tracking-wider mb-2" style={{ color: 'var(--text-secondary)' }}>
+                Description
               </p>
-            <TagSelector itemKey={itemKey!} onTagsChanged={onTagsChanged} refreshKey={tagRefreshKey} />
+              {aiDescription && (
+                <p className="text-xs italic mb-2" style={{ color: 'var(--text-secondary)' }}>
+                  {aiDescription}
+                </p>
+              )}
+              <div className="flex items-center gap-1.5">
+                <button
+                  onClick={handleGenerateDescription}
+                  disabled={generatingDesc || descPending}
+                  className="text-xs px-2 py-0.5 rounded-full transition-colors disabled:opacity-50"
+                  style={{
+                    backgroundColor: descPending ? 'var(--accent)' : 'var(--border)',
+                    color: descPending ? '#fff' : 'var(--text-secondary)',
+                  }}
+                  onMouseEnter={(e) => {
+                    if (!generatingDesc && !descPending) {
+                      ;(e.currentTarget as HTMLElement).style.backgroundColor = 'var(--text-secondary)'
+                      ;(e.currentTarget as HTMLElement).style.color = 'var(--background)'
+                    }
+                  }}
+                  onMouseLeave={(e) => {
+                    if (!descPending) {
+                      ;(e.currentTarget as HTMLElement).style.backgroundColor = 'var(--border)'
+                      ;(e.currentTarget as HTMLElement).style.color = 'var(--text-secondary)'
+                    }
+                  }}
+                  title={aiDescription ? 'Regenerate AI description' : 'Generate AI description'}
+                >
+                  {generatingDesc ? '⟳ Generating…' : descPending ? '⟳ Queued…' : aiDescription ? '✦ Regenerate Description' : '✦ Generate Description'}
+                </button>
+                {descError && (
+                  <span className="text-xs" style={{ color: '#f87171' }}>{descError}</span>
+                )}
+              </div>
+            </div>
 
             {/* Text extraction section — only for images */}
             {isImage && (
-              <div className="mt-4 pt-4" style={{ borderTop: '1px solid var(--border)' }}>
-                <p className="text-xs font-semibold uppercase tracking-wider mb-3" style={{ color: 'var(--text-secondary)' }}>
+              <div className="flex flex-col gap-2" style={{ borderTop: '1px solid var(--border)', paddingTop: '1rem' }}>
+                <p className="text-xs font-semibold uppercase tracking-wider" style={{ color: 'var(--text-secondary)' }}>
                   Text Extraction
                 </p>
 
+                <div className="flex items-center gap-2 flex-wrap">
                   <button
                     onClick={async () => {
                       setExtracting(true)
                       setExtractError(null)
+                      setExtractPending(false)
                       try {
                         const res = await fetch('/api/ai-tagging/extract-text', {
                           method: 'POST',
                           headers: { 'Content-Type': 'application/json' },
-                        body: JSON.stringify({ itemKey }),
+                          body: JSON.stringify({
+                            itemKey,
+                            ...(ocrLanguageInput.trim() && { ocrLanguages: ocrLanguageInput.trim() }),
+                          }),
                         })
+                        if (res.status === 202) {
+                          setExtractPending(true)
+                          startPolling(extractedText, translatedText, aiDescription)
+                          return
+                        }
                         if (!res.ok) {
                           const data = await res.json().catch(() => ({}))
                           throw new Error((data as { error?: string }).error ?? 'Failed to extract text')
                         }
-                      if (res.status === 202) {
-                        setExtractError('Queued — check AI Integrations for progress')
-                        setTimeout(() => setExtractError(null), 4000)
-                        return
-                      }
                         const result = await res.json()
                         setExtractedText(result.extractedText || null)
                         setEditedExtractedText(result.extractedText || '')
@@ -285,25 +486,47 @@ export default function ImageLightbox({ url, name, onClose, onPrev, onNext, item
                         setExtracting(false)
                       }
                     }}
-                  disabled={extracting}
-                  className="text-xs px-2 py-1 rounded-lg transition-colors disabled:opacity-50 mb-2"
-                  style={{ backgroundColor: 'var(--border)', color: 'var(--text-secondary)' }}
+                    disabled={extracting || extractPending}
+                    className="text-xs px-2 py-1 rounded-lg transition-colors disabled:opacity-50 self-start flex-shrink-0"
+                    style={{
+                      backgroundColor: extractPending ? 'var(--accent)' : 'var(--border)',
+                      color: extractPending ? '#fff' : 'var(--text-secondary)',
+                    }}
                     onMouseEnter={(e) => {
-                    if (!extracting) {
+                      if (!extracting && !extractPending) {
                         ;(e.currentTarget as HTMLElement).style.backgroundColor = 'var(--text-secondary)'
                         ;(e.currentTarget as HTMLElement).style.color = 'var(--background)'
                       }
                     }}
                     onMouseLeave={(e) => {
+                      if (!extractPending) {
                         ;(e.currentTarget as HTMLElement).style.backgroundColor = 'var(--border)'
                         ;(e.currentTarget as HTMLElement).style.color = 'var(--text-secondary)'
+                      }
                     }}
                   >
-                  {extracting ? '⟳ Extracting…' : extractedText ? '🔍 Re-extract Text' : '🔍 Extract Text'}
+                    {extracting ? '⟳ Extracting…' : extractPending ? '⟳ Queued…' : extractedText ? '🔍 Re-extract Text' : '🔍 Extract Text'}
                   </button>
+                  {ocrMode && ocrMode !== 'llm' && (
+                    <input
+                      type="text"
+                      value={ocrLanguageInput}
+                      onChange={(e) => setOcrLanguageInput(e.target.value)}
+                      placeholder={defaultOcrLanguages}
+                      className="text-xs px-2 py-0.5 rounded-full outline-none"
+                      style={{
+                        backgroundColor: 'var(--background)',
+                        border: '1px solid var(--border)',
+                        color: 'var(--text-primary)',
+                        width: 120,
+                      }}
+                      title="Tesseract language(s) for this extraction (e.g. jpn+jpn_vert). Leave blank to use the configured default."
+                    />
+                  )}
+                </div>
 
                 {extractError && (
-                  <p className="text-xs mb-2" style={{ color: '#f87171' }}>{extractError}</p>
+                  <p className="text-xs" style={{ color: '#f87171' }}>{extractError}</p>
                 )}
 
                 {extractedText && (
@@ -380,47 +603,64 @@ export default function ImageLightbox({ url, name, onClose, onPrev, onNext, item
                       <button
                         onClick={async () => {
                           setRetranslating(true)
+                          setTranslatePending(false)
                           try {
                             const res = await fetch('/api/ai-tagging/translate', {
                               method: 'POST',
                               headers: { 'Content-Type': 'application/json' },
                               body: JSON.stringify({ itemKey, ...(sourceLanguage.trim() && { sourceLanguage: sourceLanguage.trim() }) }),
                             })
+                            if (res.status === 202) {
+                              setTranslatePending(true)
+                              startPolling(extractedText, translatedText, aiDescription)
+                              return
+                            }
                             if (!res.ok) {
                               const data = await res.json().catch(() => ({}))
                               throw new Error((data as { error?: string }).error ?? 'Failed to translate')
                             }
-                            if (res.status !== 202) {
                             const result = await res.json()
                             setTranslatedText(result.translatedText || null)
-                            }
                           } catch {
                             // ignore
                           } finally {
                             setRetranslating(false)
                           }
                         }}
-                        disabled={retranslating}
+                        disabled={retranslating || translatePending}
                         className="text-xs px-2 py-0.5 rounded-full transition-colors disabled:opacity-50"
-                        style={{ backgroundColor: 'var(--border)', color: 'var(--text-secondary)' }}
+                        style={{
+                          backgroundColor: translatePending ? 'var(--accent)' : 'var(--border)',
+                          color: translatePending ? '#fff' : 'var(--text-secondary)',
+                        }}
                         onMouseEnter={(e) => {
-                          if (!retranslating) {
+                          if (!retranslating && !translatePending) {
                             ;(e.currentTarget as HTMLElement).style.backgroundColor = 'var(--text-secondary)'
                             ;(e.currentTarget as HTMLElement).style.color = 'var(--background)'
                           }
                         }}
                         onMouseLeave={(e) => {
+                          if (!translatePending) {
                             ;(e.currentTarget as HTMLElement).style.backgroundColor = 'var(--border)'
                             ;(e.currentTarget as HTMLElement).style.color = 'var(--text-secondary)'
+                          }
                         }}
                       >
-                        {retranslating ? '⟳ Translating…' : translatedText ? '🌐 Re-translate' : '🌐 Translate'}
+                        {retranslating ? '⟳ Translating…' : translatePending ? '⟳ Queued…' : translatedText ? '🌐 Re-translate' : '🌐 Translate'}
                       </button>
                     </div>
                   </div>
                 )}
               </div>
             )}
+
+            {/* Tags section */}
+            <div style={{ borderTop: '1px solid var(--border)', paddingTop: '1rem' }}>
+              <p className="text-xs font-semibold uppercase tracking-wider mb-3" style={{ color: 'var(--text-secondary)' }}>
+                Tags
+              </p>
+              <TagSelector itemKey={itemKey!} onTagsChanged={onTagsChanged} refreshKey={tagRefreshKey} hideDescription />
+            </div>
           </div>
         </div>
       ) : (

src/components/mixed/MixedView.tsx

@@ -83,6 +83,9 @@ export default function MixedView({ libraryId, initialPath }: Props) {
     setDoomScrollLoading(false)
   }, [currentPath])
 
+  const [ocrMode, setOcrMode] = useState<string | null>(null)
+  const [defaultOcrLanguages, setDefaultOcrLanguages] = useState('eng')
+
   const fetchAssignments = useCallback(() => {
     fetch(`/api/tags/library-assignments?libraryId=${encodeURIComponent(libraryId)}`)
       .then((r) => r.json())
@@ -92,6 +95,16 @@ export default function MixedView({ libraryId, initialPath }: Props) {
 
   useEffect(() => { fetchAssignments() }, [fetchAssignments])
 
+  useEffect(() => {
+    fetch('/api/ai-settings/ocr')
+      .then((r) => r.json())
+      .then((d: { ocrMode: string; ocrLanguages: string }) => {
+        setOcrMode(d.ocrMode)
+        setDefaultOcrLanguages(d.ocrLanguages)
+      })
+      .catch(() => {})
+  }, [])
+
   const filtersActive = search !== '' || selectedTagIds.size > 0
 
   const fetchRecursive = useCallback(() => {
@@ -387,6 +400,8 @@ export default function MixedView({ libraryId, initialPath }: Props) {
                   entry={entry}
                   onOpen={handleEntry}
                   onTag={handleTagEntry}
+                  ocrMode={ocrMode}
+                  defaultOcrLanguages={defaultOcrLanguages}
                   onAiTag={async (e) => {
                     const itemKey = itemKeyFor(e)
                     const res = await fetch('/api/ai-tagging', {
@@ -401,7 +416,7 @@ export default function MixedView({ libraryId, initialPath }: Props) {
                     fetchAssignments()
                     setFilterRefreshKey((k) => k + 1)
                   }}
-                  onExtractText={async (e) => {
+                  onExtractText={async (e, ocrLanguages) => {
                     if (e.type === 'directory') {
                       // Bulk extract for directory
                       const dirRel = filtersActive ? e.name : (currentPath ? `${currentPath}/${e.name}` : e.name)
@@ -420,7 +435,7 @@ export default function MixedView({ libraryId, initialPath }: Props) {
                       const res = await fetch('/api/ai-tagging/extract-text', {
                         method: 'POST',
                         headers: { 'Content-Type': 'application/json' },
-                        body: JSON.stringify({ itemKey }),
+                        body: JSON.stringify({ itemKey, ...(ocrLanguages && { ocrLanguages }) }),
                       })
                       if (!res.ok) {
                         const data = await res.json().catch(() => ({}))
@@ -594,7 +609,7 @@ export default function MixedView({ libraryId, initialPath }: Props) {
   )
 }
 
-function EntryTile({ entry, onOpen, onTag, onDelete, onRename, onAiTag, onExtractText, onDescribe, onTranslate }: { entry: FileEntry; onOpen: (e: FileEntry) => void; onTag: (e: FileEntry) => void; onDelete?: (e: FileEntry) => void; onRename?: (e: FileEntry, newName: string) => Promise<boolean>; onAiTag?: (e: FileEntry) => Promise<void>; onExtractText?: (e: FileEntry) => Promise<void>; onDescribe?: (e: FileEntry) => Promise<void>; onTranslate?: (e: FileEntry) => Promise<void> }) {
+function EntryTile({ entry, onOpen, onTag, onDelete, onRename, onAiTag, onExtractText, onDescribe, onTranslate, ocrMode, defaultOcrLanguages }: { entry: FileEntry; onOpen: (e: FileEntry) => void; onTag: (e: FileEntry) => void; onDelete?: (e: FileEntry) => void; onRename?: (e: FileEntry, newName: string) => Promise<boolean>; onAiTag?: (e: FileEntry) => Promise<void>; onExtractText?: (e: FileEntry, ocrLanguages?: string) => Promise<void>; onDescribe?: (e: FileEntry) => Promise<void>; onTranslate?: (e: FileEntry) => Promise<void>; ocrMode?: string | null; defaultOcrLanguages?: string }) {
   type ImgState = 'loading' | 'loaded' | 'error'
   const [imgState, setImgState] = useState<ImgState>(
     entry.thumbnailUrl ? 'loading' : 'error'
@@ -615,6 +630,8 @@ function EntryTile({ entry, onOpen, onTag, onDelete, onRename, onAiTag, onExtrac
   const [describeError, setDescribeError] = useState<string | null>(null)
   const [translating, setTranslating] = useState(false)
   const [translateError, setTranslateError] = useState<string | null>(null)
+  const [showOcrPrompt, setShowOcrPrompt] = useState(false)
+  const [ocrLanguageInput, setOcrLanguageInput] = useState('')
 
   useEffect(() => {
     if (!menuOpen) return
@@ -804,16 +821,21 @@ function EntryTile({ entry, onOpen, onTag, onDelete, onRename, onAiTag, onExtrac
                   📝 Describe Folder
                 </button>
               )}
-              {onExtractText && entry.mediaType === 'image' && (
+              {onExtractText && entry.mediaType === 'image' && !showOcrPrompt && (
                 <button
                   onClick={(e) => {
                     e.stopPropagation()
+                    if (ocrMode && ocrMode !== 'llm') {
+                      setOcrLanguageInput('')
+                      setShowOcrPrompt(true)
+                    } else {
                       setMenuOpen(false)
                       setTextExtracting(true)
                       setTextExtractError(null)
                       onExtractText(entry)
                         .catch((err) => setTextExtractError(err instanceof Error ? err.message : 'Text extraction failed'))
                         .finally(() => setTextExtracting(false))
+                    }
                   }}
                   disabled={textExtracting}
                   className="flex items-center gap-2 w-full px-4 py-2 text-sm text-left transition-colors disabled:opacity-50"
@@ -824,6 +846,57 @@ function EntryTile({ entry, onOpen, onTag, onDelete, onRename, onAiTag, onExtrac
                   🔍 Extract Text
                 </button>
               )}
+              {onExtractText && entry.mediaType === 'image' && showOcrPrompt && (
+                <div className="px-4 py-2 flex flex-col gap-2" onClick={(e) => e.stopPropagation()}>
+                  <p className="text-xs" style={{ color: 'var(--text-secondary)' }}>OCR language</p>
+                  <input
+                    autoFocus
+                    type="text"
+                    value={ocrLanguageInput}
+                    onChange={(e) => setOcrLanguageInput(e.target.value)}
+                    onKeyDown={(e) => {
+                      if (e.key === 'Escape') { setShowOcrPrompt(false) }
+                      if (e.key === 'Enter') {
+                        setShowOcrPrompt(false)
+                        setMenuOpen(false)
+                        setTextExtracting(true)
+                        setTextExtractError(null)
+                        onExtractText(entry, ocrLanguageInput.trim() || undefined)
+                          .catch((err) => setTextExtractError(err instanceof Error ? err.message : 'Text extraction failed'))
+                          .finally(() => setTextExtracting(false))
+                      }
+                    }}
+                    placeholder={defaultOcrLanguages ?? 'eng'}
+                    className="text-xs px-2 py-1 rounded-lg outline-none w-full"
+                    style={{ backgroundColor: 'var(--background)', border: '1px solid var(--border)', color: 'var(--text-primary)' }}
+                    title="Tesseract language(s) for this extraction (e.g. jpn+jpn_vert). Leave blank to use the configured default."
+                  />
+                  <div className="flex gap-2">
+                    <button
+                      onClick={() => {
+                        setShowOcrPrompt(false)
+                        setMenuOpen(false)
+                        setTextExtracting(true)
+                        setTextExtractError(null)
+                        onExtractText(entry, ocrLanguageInput.trim() || undefined)
+                          .catch((err) => setTextExtractError(err instanceof Error ? err.message : 'Text extraction failed'))
+                          .finally(() => setTextExtracting(false))
+                      }}
+                      className="text-xs px-2 py-1 rounded-lg"
+                      style={{ backgroundColor: 'var(--accent)', color: '#fff' }}
+                    >
+                      Extract
+                    </button>
+                    <button
+                      onClick={() => setShowOcrPrompt(false)}
+                      className="text-xs px-2 py-1"
+                      style={{ color: 'var(--text-secondary)' }}
+                    >
+                      Cancel
+                    </button>
+                  </div>
+                </div>
+              )}
               {onExtractText && entry.type === 'directory' && (
                 <button
                   onClick={(e) => {
@@ -844,7 +917,7 @@ function EntryTile({ entry, onOpen, onTag, onDelete, onRename, onAiTag, onExtrac
                   🔍 Extract Text for Folder
                 </button>
               )}
-              {onTranslate && entry.mediaType === 'image' && (
+              {onTranslate && entry.mediaType === 'image' && entry.hasExtractedText && (
                 <button
                   onClick={(e) => {
                     e.stopPropagation()

src/components/mixed/VideoPlayerModal.tsx

@@ -45,6 +45,8 @@ export default function VideoPlayerModal({ url, name, onClose, onPrev, onNext, i
     if (e.target === overlayRef.current) onClose()
   }
 
+  const smallBtn = 'w-7 h-7 rounded-full flex items-center justify-center transition-colors flex-shrink-0'
+
   return (
     <div
       ref={overlayRef}
@@ -52,27 +54,21 @@ export default function VideoPlayerModal({ url, name, onClose, onPrev, onNext, i
       style={{ backgroundColor: 'rgba(0,0,0,0.9)', height: '100vh', maxHeight: '100vh' }}
       onClick={handleOverlayClick}
     >
-      {/* Toolbar */}
+      {/* Toolbar — collapses to just filename when tag panel is open */}
       <div className={`flex items-center justify-between w-full flex-shrink-0 ${showTags ? '' : 'max-w-4xl'}`}>
         <span className="text-sm truncate max-w-[80%]" style={{ color: 'var(--text-secondary)' }}>
           {name}
         </span>
+        {!showTags && (
           <div className="flex items-center gap-2 flex-shrink-0">
             {itemKey && (
               <button
-              onClick={(e) => { e.stopPropagation(); setShowTags((v) => !v) }}
+                onClick={(e) => { e.stopPropagation(); setShowTags(true) }}
                 className="w-8 h-8 rounded-full flex items-center justify-center text-sm transition-colors"
-              style={{
-                backgroundColor: showTags ? 'var(--accent)' : 'var(--surface)',
-                color: showTags ? '#fff' : 'var(--text-primary)',
-              }}
-              onMouseEnter={(e) => {
-                if (!showTags) (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--surface-hover)'
-              }}
-              onMouseLeave={(e) => {
-                if (!showTags) (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--surface)'
-              }}
-              aria-label={showTags ? 'Hide tags' : 'Show tags'}
+                style={{ backgroundColor: 'var(--surface)', color: 'var(--text-primary)' }}
+                onMouseEnter={(e) => (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--surface-hover)'}
+                onMouseLeave={(e) => (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--surface)'}
+                aria-label="Show tags"
                 title="Tags"
               >
                 🏷
@@ -119,13 +115,14 @@ export default function VideoPlayerModal({ url, name, onClose, onPrev, onNext, i
               onClick={onClose}
               className="w-8 h-8 rounded-full flex items-center justify-center text-sm flex-shrink-0 transition-colors"
               style={{ backgroundColor: 'var(--surface)', color: 'var(--text-primary)' }}
-            onMouseEnter={(e) => ((e.currentTarget as HTMLElement).style.backgroundColor = 'var(--surface-hover)')}
-            onMouseLeave={(e) => ((e.currentTarget as HTMLElement).style.backgroundColor = 'var(--surface)')}
+              onMouseEnter={(e) => (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--surface-hover)'}
+              onMouseLeave={(e) => (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--surface)'}
               aria-label="Close"
             >
               ✕
             </button>
           </div>
+        )}
       </div>
 
       {showTags ? (
@@ -164,18 +161,86 @@ export default function VideoPlayerModal({ url, name, onClose, onPrev, onNext, i
               </button>
             )}
           </div>
+
           {/* Tag panel */}
           <div
-            className="w-80 h-full max-h-full flex-shrink-0 rounded-xl overflow-y-auto p-4"
+            className="w-80 h-full max-h-full flex-shrink-0 rounded-xl overflow-y-auto p-4 flex flex-col gap-4"
             style={{ backgroundColor: 'var(--surface)', border: '1px solid var(--border)' }}
             onClick={(e) => e.stopPropagation()}
           >
+            {/* Panel header — hide panel + AI tagger + close */}
+            <div className="flex items-center justify-between flex-shrink-0">
+              <button
+                onClick={() => setShowTags(false)}
+                className={smallBtn}
+                style={{ backgroundColor: 'var(--border)', color: 'var(--text-secondary)', fontSize: '0.85rem' }}
+                onMouseEnter={(e) => (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--text-secondary)'}
+                onMouseLeave={(e) => (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--border)'}
+                aria-label="Hide panel"
+                title="Hide panel"
+              >
+                ›
+              </button>
+              <div className="flex items-center gap-1.5">
+                {onAiTag && (
+                  <button
+                    onClick={async (e) => {
+                      e.stopPropagation()
+                      setAiTagging(true)
+                      setAiTagError(null)
+                      try {
+                        await onAiTag()
+                        setTagRefreshKey((k) => k + 1)
+                        onTagsChanged?.()
+                      } catch (err) {
+                        setAiTagError(err instanceof Error ? err.message : 'AI tagging failed')
+                        setTimeout(() => setAiTagError(null), 4000)
+                      } finally {
+                        setAiTagging(false)
+                      }
+                    }}
+                    disabled={aiTagging}
+                    className={`${smallBtn} disabled:opacity-50`}
+                    style={{
+                      backgroundColor: aiTagError ? '#7f1d1d' : 'var(--border)',
+                      color: aiTagError ? '#fca5a5' : 'var(--text-secondary)',
+                      fontSize: '1rem',
+                    }}
+                    onMouseEnter={(e) => {
+                      if (!aiTagging && !aiTagError) (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--text-secondary)'
+                    }}
+                    onMouseLeave={(e) => {
+                      if (!aiTagError) (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--border)'
+                    }}
+                    aria-label="AI Tag this video"
+                    title={aiTagError ?? (aiTagging ? 'Tagging…' : 'AI Tag')}
+                  >
+                    {aiTagging ? <span className="animate-spin" style={{ display: 'inline-block' }}>⟳</span> : '✨'}
+                  </button>
+                )}
+                <button
+                  onClick={onClose}
+                  className={smallBtn}
+                  style={{ backgroundColor: 'var(--border)', color: 'var(--text-secondary)', fontSize: '0.85rem' }}
+                  onMouseEnter={(e) => (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--text-secondary)'}
+                  onMouseLeave={(e) => (e.currentTarget as HTMLElement).style.backgroundColor = 'var(--border)'}
+                  aria-label="Close"
+                  title="Close"
+                >
+                  ✕
+                </button>
+              </div>
+            </div>
+
+            {/* Tags section */}
+            <div style={{ borderTop: '1px solid var(--border)', paddingTop: '1rem' }}>
               <p className="text-xs font-semibold uppercase tracking-wider mb-3" style={{ color: 'var(--text-secondary)' }}>
                 Tags
               </p>
               <TagSelector itemKey={itemKey!} onTagsChanged={onTagsChanged} refreshKey={tagRefreshKey} />
             </div>
           </div>
+        </div>
       ) : (
         <div className="w-full flex-1 min-h-0 flex items-center justify-center overflow-hidden max-h-full relative">
           <video

src/components/tags/TagSelector.tsx

@@ -8,6 +8,7 @@ interface Props {
   itemKey: string
   onTagsChanged?: () => void
   refreshKey?: number
+  hideDescription?: boolean
 }
 
 interface AllTags {
@@ -15,7 +16,7 @@ interface AllTags {
   tags: Tag[]
 }
 
-export default function TagSelector({ itemKey, onTagsChanged, refreshKey }: Props) {
+export default function TagSelector({ itemKey, onTagsChanged, refreshKey, hideDescription }: Props) {
   const [assigned, setAssigned] = useState<{ tags: Tag[]; categories: TagCategory[] }>({
     tags: [],
     categories: [],
@@ -210,6 +211,7 @@ export default function TagSelector({ itemKey, onTagsChanged, refreshKey }: Prop
   return (
     <div className="flex flex-col gap-3">
       {/* AI description */}
+      {!hideDescription && (
         <div className="flex flex-col gap-1">
           {aiDescription && (
             <p className="text-xs italic" style={{ color: 'var(--text-secondary)' }}>
@@ -241,6 +243,7 @@ export default function TagSelector({ itemKey, onTagsChanged, refreshKey }: Prop
             )}
           </div>
         </div>
+      )}
       {/* Assigned tags grouped by category */}
       {assigned.tags.length > 0 && (
         <div className="flex flex-wrap gap-1.5">

src/lib/ai-jobs.ts

@@ -34,6 +34,7 @@ interface AiJobRow {
   started_at: number | null
   completed_at: number | null
   item_title: string | null
+  payload: string | null
 }
 
 function rowToJob(row: AiJobRow): AiJob {
@@ -75,6 +76,7 @@ export function enqueueJob(
   jobType: AiJobType,
   libraryId: string,
   sourceLanguage?: string,
+  payload?: Record<string, string>,
 ): string {
   const db = getDb()
 
@@ -96,9 +98,9 @@ export function enqueueJob(
   const metadata = jobType === 'translate' && sourceLanguage ? sourceLanguage : null
 
   db.prepare(
-    `INSERT INTO ai_jobs (id, item_key, library_id, job_type, status, error, attempt, max_retries, created_at, item_title)
-     VALUES (?, ?, ?, ?, 'queued', ?, 0, ?, ?, ?)`
-  ).run(id, itemKey, libraryId, jobType, metadata, maxRetries, Date.now(), title)
+    `INSERT INTO ai_jobs (id, item_key, library_id, job_type, status, error, attempt, max_retries, created_at, item_title, payload)
+     VALUES (?, ?, ?, ?, 'queued', ?, 0, ?, ?, ?, ?)`
+  ).run(id, itemKey, libraryId, jobType, metadata, maxRetries, Date.now(), title, payload ? JSON.stringify(payload) : null)
 
   // Wake the processor
   wakeProcessor()
@@ -251,6 +253,8 @@ async function processNextJob(): Promise<boolean> {
 
   // Extract sourceLanguage for translate jobs (stored in error field at enqueue)
   const sourceLanguage = row.job_type === 'translate' ? row.error : null
+  // Parse job payload (carries per-call overrides, e.g. ocrLanguages for extract jobs)
+  const jobPayload = row.payload ? (JSON.parse(row.payload) as Record<string, string>) : null
 
   db.prepare(
     "UPDATE ai_jobs SET status = 'running', started_at = ?, error = NULL WHERE id = ?"
@@ -265,7 +269,7 @@ async function processNextJob(): Promise<boolean> {
         await generateItemDescription(row.item_key)
         break
       case 'extract':
-        await extractItemText(row.item_key)
+        await extractItemText(row.item_key, jobPayload?.ocrLanguages)
         break
       case 'translate':
         await translateItemText(row.item_key, sourceLanguage || undefined)

src/lib/ai-tagger.ts

@@ -538,7 +538,7 @@ async function extractWithTesseract(
  * Translation is not performed automatically — call translateItemText() separately.
  * Returns { extractedText, translatedText } where translatedText is always null.
  */
-export async function extractItemText(itemKey: string): Promise<{ extractedText: string; translatedText: string | null }> {
+export async function extractItemText(itemKey: string, ocrLanguagesOverride?: string): Promise<{ extractedText: string; translatedText: string | null }> {
   const libraryId = itemKey.split(':')[0]
   const config = getEffectiveAiConfig(libraryId)
 
@@ -567,7 +567,8 @@ export async function extractItemText(itemKey: string): Promise<{ extractedText:
     throw Object.assign(new Error('Text extraction is only available for images'), { code: 'NO_IMAGE' })
   }
 
-  const { ocrMode, ocrLanguages, ocrConfidenceThreshold } = config
+  const { ocrMode, ocrLanguages: configOcrLanguages, ocrConfidenceThreshold } = config
+  const ocrLanguages = ocrLanguagesOverride?.trim() || configOcrLanguages
 
   // ── Tesseract path ────────────────────────────────────────────────────────
   if (ocrMode === 'tesseract' || ocrMode === 'hybrid') {

src/lib/db.ts

@@ -338,4 +338,12 @@ function migrateAiJobs(db: Database.Database): void {
     CREATE INDEX IF NOT EXISTS ai_jobs_status ON ai_jobs(status);
     CREATE INDEX IF NOT EXISTS ai_jobs_created_at ON ai_jobs(created_at);
   `)
+
+  // Add payload column if not present
+  const aiJobsRow = db
+    .prepare("SELECT sql FROM sqlite_master WHERE type='table' AND name='ai_jobs'")
+    .get() as { sql: string } | undefined
+  if (aiJobsRow && !aiJobsRow.sql.includes('payload')) {
+    db.exec('ALTER TABLE ai_jobs ADD COLUMN payload TEXT')
+  }
 }

src/types/index.ts

@@ -44,6 +44,7 @@ export interface FileEntry {
   mediaType: MediaType | null
   url: string | null
   thumbnailUrl: string | null
+  hasExtractedText?: boolean
 }
 
 export interface Movie {