Free, open-source file converter. Convert 263 formats offline by drag & drop — images, PDF, Word, Excel, PowerPoint, HWP/HWPX, video, audio, e-books and archives. No upload, no Office needed.
C# audio-converteravaloniabatch-conversionconvertercsharpdocument-converterdocx-to-pdfdotnetdrag-and-dropfile-conversionfile-converterheic-converterhwphwpximage-converterlinuxmacospdf-convertervideo-converterwindows
feat(ocr): add Apple Vision native and Tesseract OCR engine support
smjrifle216282a748cd
diff --git a/docs/ENGINES.md b/docs/ENGINES.mdindex a7c4ce25..38907d75 100644--- a/docs/ENGINES.md+++ b/docs/ENGINES.md@@ -34,6 +34,8 @@ Filee chooses engines automatically. Settings → *Engines* shows their status,| Ghostscript | **optional download** (~20 MB, 31 MB on disk) | EPS/PS → PDF, PDF → EPS/PS, PostScript-based AI → PDF (images through PDF and PDFium); smaller PDFs when a preset asks for it || FFmpeg | **optional download** (~100 MB, 272 MB on disk) | **all video and audio**: video ↔ video (H.264, HEVC, AV1), video → animated GIF / APNG, a still frame or every frame, audio and subtitle extraction, audio ↔ audio (also APE, WavPack, TTA, DSD, DTS, ALAC, …), GIF, APNG and photos → MP4/WEBM/MOV, Live / Motion Photos → their video, a clip of a video → an iPhone Live Photo or Android Motion Photo (experimental), images → HEIC || Calibre | **optional download** (~216 MB, 660 MB on disk) | rare e-book formats: LIT, LRF, CHM, PDB, PML, RB, SNB, TCR, OEB → EPUB; EPUB → MOBI, AZW3, LIT, LRF, PDB, PML, RB, SNB, TCR; e-books → PDF (ranked above the HWPX writer, see below) |+| Apple Vision | part of macOS (macOS only) | **scanned PDFs and images → editable text / DOCX / Markdown / HWPX** via native neural OCR (Latin, CJK, Cyrillic); zero download size |+| Tesseract | **optional, install yourself** (~30 MB + the trained data of each language) | OCR for scripts Apple Vision doesn't know: Devanagari (Nepali, Hindi, Sanskrit, Marathi), Cyrillic beyond Russian/Ukrainian, Thai, Greek, Arabic script and more; the only OCR on Windows and Linux (Linux has no built-in option, so this is required there) |DOCX, XLSX, XLS, ODS and PPTX → PDF need neither Microsoft Office nor LibreOffice: they are read in-process, written as HWPXand rendered by rhwp. Anything the built-in readers understand (including PDF) is written as DOCX by the DOCX writer.diff --git a/src/Filee.App/Assets/i18n/ar.json b/src/Filee.App/Assets/i18n/ar.jsonindex 7ee866f4..c27034f0 100644--- a/src/Filee.App/Assets/i18n/ar.json+++ b/src/Filee.App/Assets/i18n/ar.json@@ -140,6 +140,7 @@"engine.ghostscript.description": "تحويل EPS وPostScript وملفات Illustrator القديمة إلى PDF وصور؛ وتحويل PDF إلى EPS وPS.","engine.calibre.description": "صيغ الكتب الإلكترونية النادرة (LIT وLRF وCHM وPDB وRB…) وإخراج Kindle (MOBI وAZW3).\nعند وضعه فوق محرك كتابة HWPX، يُنشئ أيضًا PDF من الكتب الإلكترونية (بسرعة أكبر بكثير للكتب الطويلة).","engine.libreoffice.description": "قراءة الصيغ القديمة والأقل شيوعًا: DOC وPPT وPages وKeynote وNumbers وWordPerfect وClarisWorks وPublisher وVisio…\nوالحفظ بصيغ DOC وXLS وPPT وODT وODP وقوالب OpenDocument وEMF.",+ "engine.image-ocr.description": "استخراج النص من ملفات PDF الممسوحة ضوئيًا والصور إلى مستندات قابلة للتحرير (Word والنص وHWPX).","profile.images": "الصور","profile.pdf": "PDF","profile.office": "المستندات",@@ -519,6 +520,14 @@"presets.leave_out_pictures": "استبعاد الصور (النص فقط)","presets.leave_out_pictures_hint": "عند التعطيل: تُحفَظ الصور في مجلد name_files بجانب ملف Markdown، وتُدرَج روابطها فيه.\nعند التفعيل: النص فقط، دون مجلد.","presets.target_folder": "مجلد (استخراج الأرشيفات)",+ "presets.group.ocr": "التعرف البصري على الأحرف (OCR)",+ "presets.ocr_enabled": "التعرف على النص في المستندات الممسوحة ضوئيًا والصور",+ "presets.ocr_hint": "استخراج نص قابل للتحرير من الصفحات الممسوحة ضوئيًا والصور إلى Word والنص والمستندات.",+ "presets.ocr_language": "لغة التعرف",+ "presets.ocr_language_hint": "اللغة المستخدمة للتعرف بدقة على الأحرف والكلمات.",+ "presets.ocr_language.auto": "تلقائي / الافتراضي للنظام",+ "presets.ocr_engine_required": "يتطلب Tesseract. قم بتثبيته من الإعدادات ← المحركات",+ "presets.ocr_disabled_linux": "لا يحتوي Linux على OCR مدمج. قم بتثبيت Tesseract من الإعدادات ← المحركات للتمكين.","presets.group.media": "الفيديو والصوت","presets.media_quality": "الجودة","presets.media_quality_hint": "عندما تستطيع الصيغة الجديدة الاحتفاظ بالترميزات كما هي (MOV من iPhone إلى MP4)، تتغير الحاوية فقط.\nدون فقدان جودة، مع الاحتفاظ بـ HDR، خلال ثوانٍ.\nخيار «صغير» يعيد الترميز دائمًا لتقليل الحجم.",@@ -652,6 +661,11 @@"engines.guidance.linux.libreoffice": "ثبّته باستخدام مدير الحزم (على Ubuntu/Debian: sudo apt install libreoffice)، ثم اختر «التحقق مجددًا».\nيتطلب فتح HWP أيضًا Java وإضافة H2Orestart.","engines.guidance.linux.ghostscript": "ثبّته باستخدام مدير الحزم (على Ubuntu/Debian: sudo apt install ghostscript)، ثم اختر «التحقق مجددًا».","engines.guidance.windows": "لا يوفر Filee تنزيلًا لـ {0} لهذا الحاسوب.\nاختر «استخدام نسختي…» وحدّد نسخة مثبّتة.",+ "engines.package.tesseract.name": "Tesseract (التعرف على النصوص متعدد اللغات)",+ "engines.package.tesseract.description": "يتعرف على النص بأكثر من 100 لغة، بما في ذلك الديفاناغارية (النيبالية/الهندية) والعربية والخطوط النادرة.\nعلى macOS، يعمل Apple Vision أصلاً للغات الإنجليزية ولغات CJK واللغات الأوروبية بدونه.",+ "engines.guidance.macos.tesseract": "قم بتثبيت Tesseract باستخدام Homebrew (brew install tesseract tesseract-lang)، ثم اختر “تحقق مرة أخرى”.\nأو اختر “استخدام نسختي…” وحدد مجلد bin الخاص به.",+ "engines.guidance.linux.tesseract": "قم بتثبيت Tesseract باستخدام مدير الحزم (Ubuntu/Debian: sudo apt install tesseract-ocr tesseract-ocr-all)، ثم اختر “تحقق مرة أخرى”.\nأو اختر “استخدام نسختي…” وحدد مجلد البرنامج الخاص به.",+ "engines.guidance.windows.tesseract": "قم بتثبيت Tesseract من UB-Mannheim (github.com/UB-Mannheim/tesseract)، ثم اختر “تحقق مرة أخرى” أو حدد مجلد التثبيت الخاص به.","engines.use_own_app_folder": "اختر المجلد الذي يحتوي على {0}.app","engines.progress_detail": "{0} من {1} · {2}/ث · {3}","engines.eta.estimating": "جارٍ تقدير الوقت المتبقي…",diff --git a/src/Filee.App/Assets/i18n/en.json b/src/Filee.App/Assets/i18n/en.jsonindex d6024c30..31353d40 100644--- a/src/Filee.App/Assets/i18n/en.json+++ b/src/Filee.App/Assets/i18n/en.json@@ -147,6 +147,7 @@"engine.ghostscript.description": "EPS, PostScript and older Illustrator files to PDF and images; PDF to EPS and PS.","engine.calibre.description": "Rare e-book formats (LIT, LRF, CHM, PDB, RB…) and Kindle output (MOBI, AZW3).\nAbove the HWPX writer, it also makes the PDFs of e-books (much faster for long books).","engine.libreoffice.description": "Reads older and less common formats: DOC, PPT, Pages, Keynote, Numbers, WordPerfect, ClarisWorks, Publisher, Visio…\nAlso saves as DOC, XLS, PPT, ODT, ODP, OpenDocument templates and EMF.",+ "engine.image-ocr.description": "Extracts text from scanned PDFs and pictures into editable documents (Word, text, HWPX).","profile.images": "Images","profile.pdf": "PDF",@@ -534,6 +535,14 @@"presets.leave_out_pictures": "Leave out pictures (text only)","presets.leave_out_pictures_hint": "Off: the pictures are saved in a \"name_files\" folder next to the Markdown file and linked from it.\nOn: only the text, no folder.","presets.target_folder": "Folder (extract archives)",+ "presets.group.ocr": "Optical Character Recognition (OCR)",+ "presets.ocr_enabled": "Recognize text in scans and images",+ "presets.ocr_hint": "Extracts editable text from scanned pages and images into Word, text, and documents.",+ "presets.ocr_language": "Recognition language",+ "presets.ocr_language_hint": "Language used to recognize characters and words accurately.",+ "presets.ocr_language.auto": "Automatic / System Default",+ "presets.ocr_engine_required": "Needs Tesseract. Install it in Settings → Engines",+ "presets.ocr_disabled_linux": "Linux has no built-in OCR. Install Tesseract in Settings → Engines to enable.","presets.group.media": "Video & audio","presets.media_quality": "Quality","presets.media_quality_hint": "When the new format can hold the codecs as they are (iPhone MOV → MP4), the video is only rewrapped.\nNo quality lost, HDR kept, done in seconds.\nSmall always re-encodes to make the file smaller.",@@ -669,6 +678,11 @@"engines.guidance.linux.libreoffice": "Install it with your package manager (Ubuntu/Debian: sudo apt install libreoffice), then choose “Check again”.\nOpening HWP also needs Java and the H2Orestart extension.","engines.guidance.linux.ghostscript": "Install it with your package manager (Ubuntu/Debian: sudo apt install ghostscript), then choose “Check again”.","engines.guidance.windows": "Filee has no download of {0} for this PC.\nChoose “Use my copy…” and select an installed copy.",+ "engines.package.tesseract.name": "Tesseract (multilingual text recognition)",+ "engines.package.tesseract.description": "Recognizes text in over 100 languages, including Devanagari (Nepali/Hindi), Arabic, and rare scripts.\nOn macOS, Apple Vision works natively for English, CJK, and European languages without it.",+ "engines.guidance.macos.tesseract": "Install Tesseract with Homebrew (brew install tesseract tesseract-lang), then choose “Check again”.\nOr choose “Use my copy…” and select its bin folder.",+ "engines.guidance.linux.tesseract": "Install Tesseract with your package manager (Ubuntu/Debian: sudo apt install tesseract-ocr tesseract-ocr-all), then choose “Check again”.\nOr choose “Use my copy…” and select its program folder.",+ "engines.guidance.windows.tesseract": "Install Tesseract from UB-Mannheim (github.com/UB-Mannheim/tesseract), then choose “Check again” or select its install folder.","engines.use_own_app_folder": "Choose the folder containing {0}.app","engines.progress_detail": "{0} of {1} · {2}/s · {3}","engines.eta.estimating": "estimating time left…",diff --git a/src/Filee.App/Assets/i18n/es.json b/src/Filee.App/Assets/i18n/es.jsonindex 89007a3e..b3da5d10 100644--- a/src/Filee.App/Assets/i18n/es.json+++ b/src/Filee.App/Assets/i18n/es.json@@ -140,6 +140,7 @@"engine.ghostscript.description": "EPS, PostScript y archivos antiguos de Illustrator a PDF e imágenes; PDF a EPS y PS.","engine.calibre.description": "Formatos raros de libros electrónicos (LIT, LRF, CHM, PDB, RB…) y salida Kindle (MOBI, AZW3).\nPor encima del escritor HWPX, también crea los PDF de libros electrónicos (mucho más rápido para libros largos).","engine.libreoffice.description": "Lee formatos antiguos y poco habituales: DOC, PPT, Pages, Keynote, Numbers, WordPerfect, ClarisWorks, Publisher, Visio…\nTambién guarda como DOC, XLS, PPT, ODT, ODP, plantillas OpenDocument y EMF.",+ "engine.image-ocr.description": "Extrae texto de PDF escaneados e imágenes a documentos editables (Word, texto, HWPX).","profile.images": "Imágenes","profile.pdf": "PDF","profile.office": "Documentos",@@ -519,6 +520,14 @@"presets.leave_out_pictures": "Omitir imágenes (solo texto)","presets.leave_out_pictures_hint": "Desactivado: las imágenes se guardan en una carpeta \"nombre_files\" junto al archivo Markdown y se enlazan desde él.\nActivado: solo el texto, sin carpeta.","presets.target_folder": "Carpeta (extraer archivos comprimidos)",+ "presets.group.ocr": "Reconocimiento óptico de caracteres (OCR)",+ "presets.ocr_enabled": "Reconocer texto en escaneos e imágenes",+ "presets.ocr_hint": "Extrae texto editable de páginas escaneadas e imágenes a Word, texto y documentos.",+ "presets.ocr_language": "Idioma de reconocimiento",+ "presets.ocr_language_hint": "Idioma utilizado para reconocer caracteres y palabras con precisión.",+ "presets.ocr_language.auto": "Automático / Predeterminado del sistema",+ "presets.ocr_engine_required": "Requiere Tesseract. Instálalo en Configuración → Motores",+ "presets.ocr_disabled_linux": "Linux no tiene OCR integrado. Instala Tesseract en Configuración → Motores para activarlo.","presets.group.media": "Vídeo y audio","presets.media_quality": "Calidad","presets.media_quality_hint": "Cuando el formato nuevo puede conservar los códecs tal como están (MOV de iPhone → MP4), el vídeo solo se cambia de contenedor sin recodificar.\nSin pérdida de calidad, HDR conservado, listo en segundos.\nPequeño siempre recodifica para que el archivo pese menos.",@@ -652,6 +661,11 @@"engines.guidance.linux.libreoffice": "Instálalo con tu gestor de paquetes (Ubuntu/Debian: sudo apt install libreoffice) y luego elige «Comprobar otra vez».\nAbrir HWP necesita además Java y la extensión H2Orestart.","engines.guidance.linux.ghostscript": "Instálalo con tu gestor de paquetes (Ubuntu/Debian: sudo apt install ghostscript) y luego elige «Comprobar otra vez».","engines.guidance.windows": "Filee no tiene una descarga de {0} para este PC.\nElige «Usar mi copia…» y selecciona una copia instalada.",+ "engines.package.tesseract.name": "Tesseract (reconocimiento de texto multilingüe)",+ "engines.package.tesseract.description": "Reconoce texto en más de 100 idiomas, incluidos devanagari (nepalí/hindi), árabe y alfabetos poco comunes.\nEn macOS, Apple Vision funciona de forma nativa para inglés, CJK e idiomas europeos sin él.",+ "engines.guidance.macos.tesseract": "Instala Tesseract con Homebrew (brew install tesseract tesseract-lang), luego elige “Comprobar de nuevo”.\nO elige “Usar mi copia…” y selecciona su carpeta bin.",+ "engines.guidance.linux.tesseract": "Instala Tesseract con el gestor de paquetes (Ubuntu/Debian: sudo apt install tesseract-ocr tesseract-ocr-all), luego elige “Comprobar de nuevo”.\nO elige “Usar mi copia…” y selecciona su carpeta de programas.",+ "engines.guidance.windows.tesseract": "Instala Tesseract desde UB-Mannheim (github.com/UB-Mannheim/tesseract), luego elige “Comprobar de nuevo” o selecciona su carpeta de instalación.","engines.use_own_app_folder": "Elige la carpeta que contiene {0}.app","engines.progress_detail": "{0} de {1} · {2}/s · {3}","engines.eta.estimating": "calculando el tiempo restante…",diff --git a/src/Filee.App/Assets/i18n/hi.json b/src/Filee.App/Assets/i18n/hi.jsonindex dde78607..deaca383 100644--- a/src/Filee.App/Assets/i18n/hi.json+++ b/src/Filee.App/Assets/i18n/hi.json@@ -140,6 +140,7 @@"engine.ghostscript.description": "EPS, PostScript और पुरानी Illustrator फ़ाइलें PDF और इमेज में; PDF से EPS और PS।","engine.calibre.description": "दुर्लभ ई-बुक फ़ॉर्मैट (LIT, LRF, CHM, PDB, RB…) और Kindle आउटपुट (MOBI, AZW3)।\nHWPX writer के ऊपर होने पर यह ई-बुक के PDF भी बनाता है (लंबी किताबों के लिए बहुत तेज़)।","engine.libreoffice.description": "पुराने और कम प्रचलित फ़ॉर्मैट पढ़ता है: DOC, PPT, Pages, Keynote, Numbers, WordPerfect, ClarisWorks, Publisher, Visio…\nDOC, XLS, PPT, ODT, ODP, OpenDocument टेम्पलेट और EMF के रूप में सहेजता भी है।",+ "engine.image-ocr.description": "स्कैन किए गए PDF और चित्रों से टेक्स्ट निकाल कर संपादन योग्य दस्तावेज़ों (Word, टेक्स्ट, HWPX) में बदलता है।","profile.images": "इमेज","profile.pdf": "PDF","profile.office": "दस्तावेज़",@@ -519,6 +520,14 @@"presets.leave_out_pictures": "तस्वीरें छोड़ दें (केवल टेक्स्ट)","presets.leave_out_pictures_hint": "बंद: तस्वीरें Markdown फ़ाइल के बगल में \"name_files\" फ़ोल्डर में सहेजी जाती हैं और उसी से लिंक होती हैं।\nचालू: केवल टेक्स्ट, कोई फ़ोल्डर नहीं।","presets.target_folder": "फ़ोल्डर (आर्काइव निकालने के लिए)",+ "presets.group.ocr": "ऑप्टिकल कैरेक्टर रिकग्निशन (OCR)",+ "presets.ocr_enabled": "स्कैन और छवियों में टेक्स्ट पहचानें",+ "presets.ocr_hint": "स्कैन किए गए पेजों और छवियों से संपादन योग्य टेक्स्ट निकाल कर Word, टेक्स्ट और दस्तावेज़ों में बदलें।",+ "presets.ocr_language": "पहचान भाषा",+ "presets.ocr_language_hint": "वर्णों और शब्दों को सटीक रूप से पहचानने के लिए उपयोग की जाने वाली भाषा।",+ "presets.ocr_language.auto": "स्वचालित / सिस्टम डिफ़ॉल्ट",+ "presets.ocr_engine_required": "Tesseract की आवश्यकता है। इसे सेटिंग्स → इंजन में इंस्टॉल करें",+ "presets.ocr_disabled_linux": "Linux में कोई अंतर्निहित OCR नहीं है। सक्षम करने के लिए सेटिंग्स → इंजन में Tesseract इंस्टॉल करें।","presets.group.media": "वीडियो और ऑडियो","presets.media_quality": "क्वालिटी","presets.media_quality_hint": "जब नया फ़ॉर्मैट कोडेक को वैसे ही रख सकता है (iPhone MOV → MP4), तो वीडियो केवल दोबारा लपेटा जाता है।\nकोई क्वालिटी नहीं घटती, HDR बना रहता है, और कुछ सेकंड में हो जाता है।\nछोटा विकल्प फ़ाइल घटाने के लिए हमेशा दोबारा एन्कोड करता है।",@@ -652,6 +661,11 @@"engines.guidance.linux.libreoffice": "इसे अपने पैकेज मैनेजर से इंस्टॉल करें (Ubuntu/Debian: sudo apt install libreoffice), फिर “फिर से जाँचें” चुनें।\nHWP खोलने के लिए Java और H2Orestart एक्सटेंशन भी चाहिए।","engines.guidance.linux.ghostscript": "इसे अपने पैकेज मैनेजर से इंस्टॉल करें (Ubuntu/Debian: sudo apt install ghostscript), फिर “फिर से जाँचें” चुनें।","engines.guidance.windows": "इस PC के लिए Filee के पास {0} का डाउनलोड नहीं है।\n“मेरी कॉपी इस्तेमाल करें…” चुनें और इंस्टॉल की गई कॉपी चुनें।",+ "engines.package.tesseract.name": "Tesseract (बहुभाषी टेक्स्ट पहचान)",+ "engines.package.tesseract.description": "देवनागरी (नेपाली/हिंदी), अरबी और दुर्लभ लिपियों सहित 100 से अधिक भाषाओं में टेक्स्ट पहचानता है।\nmacOS पर Apple Vision अंग्रेजी, CJK और यूरोपीय भाषाओं के लिए बिना इसके भी काम करता है।",+ "engines.guidance.macos.tesseract": "Homebrew से Tesseract इंस्टॉल करें (brew install tesseract tesseract-lang), फिर “दोबारा जांचें” चुनें।\nया “मेरी कॉपी इस्तेमाल करें…” चुनें और इसके bin फ़ोल्डर का चयन करें।",+ "engines.guidance.linux.tesseract": "अपने पैकेज मैनेजर से Tesseract इंस्टॉल करें (Ubuntu/Debian: sudo apt install tesseract-ocr tesseract-ocr-all), फिर “दोबारा जांचें” चुनें।\nया “मेरी कॉपी इस्तेमाल करें…” चुनें और इसका प्रोग्राम फ़ोल्डर चुनें।",+ "engines.guidance.windows.tesseract": "UB-Mannheim (github.com/UB-Mannheim/tesseract) से Tesseract इंस्टॉल करें, फिर “दोबारा जांचें” चुनें या इसके इंस्टाल फ़ोल्डर का चयन करें।","engines.use_own_app_folder": "वह फ़ोल्डर चुनें जिसमें {0}.app है","engines.progress_detail": "{1} में से {0} · {2}/s · {3}","engines.eta.estimating": "बचा हुआ समय आँका जा रहा है…",diff --git a/src/Filee.App/Assets/i18n/ja.json b/src/Filee.App/Assets/i18n/ja.jsonindex afc5f5d3..bc537324 100644--- a/src/Filee.App/Assets/i18n/ja.json+++ b/src/Filee.App/Assets/i18n/ja.json@@ -140,6 +140,7 @@"engine.ghostscript.description": "EPS、PostScript、古いIllustratorファイルをPDFや画像に変換し、PDFをEPSやPSに変換します。","engine.calibre.description": "一般的でない電子書籍形式(LIT、LRF、CHM、PDB、RBなど)を変換し、Kindle用のMOBI、AZW3形式で保存できます。\n「HWPX writer」より上に置くと、電子書籍のPDF作成もCalibreが担当します(長い本では大幅に速くなります)。","engine.libreoffice.description": "古い形式や一般的でない形式(DOC、PPT、Pages、Keynote、Numbers、WordPerfect、ClarisWorks、Publisher、Visioなど)を読み込みます。\nDOC、XLS、PPT、ODT、ODP、OpenDocumentテンプレート、EMF形式での保存にも対応します。",+ "engine.image-ocr.description": "スキャンしたPDFや画像からテキストを抽出し、編集可能な文書(Word、テキスト、HWPX)に変換します。","profile.images": "画像","profile.pdf": "PDF","profile.office": "文書",@@ -519,6 +520,14 @@"presets.leave_out_pictures": "画像を含めない(テキストのみ)","presets.leave_out_pictures_hint": "オフ:画像をMarkdownファイルと同じ場所の「name_files」フォルダーに保存し、Markdownファイルからリンクします。\nオン:テキストだけを保存し、フォルダーは作成しません。","presets.target_folder": "フォルダー(アーカイブを展開)",+ "presets.group.ocr": "光学式文字認識 (OCR)",+ "presets.ocr_enabled": "スキャンや画像内のテキストを認識",+ "presets.ocr_hint": "スキャンしたページや画像から編集可能なテキストを抽出し、Word、テキスト、ドキュメントに変換します。",+ "presets.ocr_language": "認識言語",+ "presets.ocr_language_hint": "文字や単語を正確に認識するために使用する言語。",+ "presets.ocr_language.auto": "自動 / システムのデフォルト",+ "presets.ocr_engine_required": "Tesseractが必要です。設定 → エンジンからインストールしてください",+ "presets.ocr_disabled_linux": "Linuxには組み込みOCRがありません。有効にするには設定 → エンジンでTesseractをインストールしてください。","presets.group.media": "動画・音声","presets.media_quality": "品質","presets.media_quality_hint": "変換先が元のコーデックに対応している場合(iPhoneのMOV → MP4など)、動画を再エンコードせずに形式だけを変えます。\n画質を保ち、HDRも維持したまま数秒で完了します。\n「小さめ」はファイルサイズを減らすため、常に再エンコードします。",@@ -652,6 +661,11 @@"engines.guidance.linux.libreoffice": "パッケージマネージャーでインストール(Ubuntu/Debian:sudo apt install libreoffice)してから、「再確認」を選んでください。\nHWPを開くには、JavaとH2Orestart拡張機能も必要です。","engines.guidance.linux.ghostscript": "パッケージマネージャーでインストール(Ubuntu/Debian:sudo apt install ghostscript)してから、「再確認」を選んでください。","engines.guidance.windows": "このPC向けの{0}はFileeからダウンロードできません。\n「手元のプログラムを使う…」を選び、インストール済みのものを指定してください。",+ "engines.package.tesseract.name": "Tesseract(多言語テキスト認識)",+ "engines.package.tesseract.description": "デーヴァナーガリー(ネパール語/ヒンディー語)、アラビア語、希少言語を含む100以上の言語のテキストを認識します。\nmacOSではApple Visionが英語、CJK、ヨーロッパ言語をネイティブにサポートしています。",+ "engines.guidance.macos.tesseract": "HomebrewでTesseractをインストール(brew install tesseract tesseract-lang)してから「再確認」を選択してください。\nまたは「手持ちのコピーを使う…」を選択してbinフォルダーを指定してください。",+ "engines.guidance.linux.tesseract": "パッケージマネージャーでTesseractをインストール(Ubuntu/Debian: sudo apt install tesseract-ocr tesseract-ocr-all)してから「再確認」を選択してください。\nまたは「手持ちのコピーを使う…」を選択してプログラムフォルダーを指定してください。",+ "engines.guidance.windows.tesseract": "UB-Mannheim(github.com/UB-Mannheim/tesseract)からTesseractをインストールし、「再確認」を選択するかインストールフォルダーを指定してください。","engines.use_own_app_folder": "{0}.appを含むフォルダーを選択してください","engines.progress_detail": "{0}/{1} · {2}/秒 · {3}","engines.eta.estimating": "残り時間を計算中…",diff --git a/src/Filee.App/Assets/i18n/ko.json b/src/Filee.App/Assets/i18n/ko.jsonindex 115bfc7e..379ada5e 100644--- a/src/Filee.App/Assets/i18n/ko.json+++ b/src/Filee.App/Assets/i18n/ko.json@@ -147,6 +147,7 @@"engine.ghostscript.description": "EPS, 포스트스크립트, 예전 일러스트레이터 파일을 PDF와 이미지로, PDF를 EPS·PS로.","engine.calibre.description": "드문 전자책 형식(LIT, LRF, CHM, PDB, RB…)과 킨들용 출력(MOBI, AZW3).\nHWPX writer보다 위에 두면 전자책 PDF도 만들어요(긴 책은 훨씬 빨라요).","engine.libreoffice.description": "DOC, PPT, Pages, Keynote, Numbers, 워드퍼펙트, ClarisWorks, Publisher, Visio 같은 오래되었거나 드문 형식을 읽어요.\nDOC·XLS·PPT·ODT·ODP, 오픈도큐먼트 서식 파일, EMF로 저장도 해요.",+ "engine.image-ocr.description": "스캔된 PDF 및 이미지에서 텍스트를 추출하여 편집 가능한 문서(Word, 텍스트, HWPX)로 변환합니다.","profile.images": "이미지","profile.pdf": "PDF",@@ -534,6 +535,14 @@"presets.leave_out_pictures": "그림 빼기 (텍스트만)","presets.leave_out_pictures_hint": "끄면: 그림을 마크다운 파일 옆 \"이름_files\" 폴더에 저장하고 파일에서 연결해요.\n켜면: 폴더 없이 텍스트만 남겨요.","presets.target_folder": "폴더 (압축 풀기)",+ "presets.group.ocr": "광학 문자 인식 (OCR)",+ "presets.ocr_enabled": "스캔 및 이미지에서 텍스트 인식",+ "presets.ocr_hint": "스캔된 페이지와 이미지에서 편집 가능한 텍스트를 추출하여 Word, 텍스트 및 문서로 변환합니다.",+ "presets.ocr_language": "인식 언어",+ "presets.ocr_language_hint": "문자와 단어를 정확하게 인식하는 데 사용되는 언어입니다.",+ "presets.ocr_language.auto": "자동 / 시스템 기본값",+ "presets.ocr_engine_required": "Tesseract가 필요합니다. 설정 → 엔진에서 설치해 주세요",+ "presets.ocr_disabled_linux": "Linux에는 내장 OCR이 없습니다. 활성화하려면 설정 → 엔진에서 Tesseract를 설치하세요.","presets.group.media": "동영상·오디오","presets.media_quality": "품질","presets.media_quality_hint": "새 형식에 코덱을 그대로 담을 수 있으면(아이폰 MOV → MP4 등) 다시 인코딩하지 않아요.\n화질과 HDR이 그대로고 몇 초면 끝나요.\n'작게'는 용량을 줄이려고 항상 다시 인코딩해요.",@@ -669,6 +678,11 @@"engines.guidance.linux.libreoffice": "패키지 관리자로 설치(Ubuntu/Debian: sudo apt install libreoffice)한 뒤 “다시 확인”을 누르세요.\nHWP를 열려면 Java와 H2Orestart 확장도 필요해요.","engines.guidance.linux.ghostscript": "패키지 관리자로 설치(Ubuntu/Debian: sudo apt install ghostscript)한 뒤 “다시 확인”을 누르세요.","engines.guidance.windows": "이 PC용 {0} 다운로드는 없어요.\n“내 프로그램 쓰기…”로 설치된 프로그램을 선택해 주세요.",+ "engines.package.tesseract.name": "Tesseract (다국어 텍스트 인식)",+ "engines.package.tesseract.description": "데바나가리(네팔어/힌디어), 아랍어 및 희귀 문자를 포함한 100개 이상의 언어로 텍스트를 인식합니다.\nmacOS에서는 Apple Vision이 내장되어 영어, CJK 및 유럽어는 추가 설치 없이 동작합니다.",+ "engines.guidance.macos.tesseract": "Homebrew로 Tesseract를 설치(brew install tesseract tesseract-lang)한 뒤 “다시 확인”을 선택하세요.\n또는 “내 사본 사용…”을 선택하고 bin 폴더를 지정하세요.",+ "engines.guidance.linux.tesseract": "패키지 관리자로 Tesseract를 설치(Ubuntu/Debian: sudo apt install tesseract-ocr tesseract-ocr-all)한 뒤 “다시 확인”을 선택하세요.\n또는 “내 사본 사용…”을 선택하고 프로그램 폴더를 지정하세요.",+ "engines.guidance.windows.tesseract": "UB-Mannheim(github.com/UB-Mannheim/tesseract)에서 Tesseract를 설치한 뒤 “다시 확인”을 선택하거나 설치 폴더를 지정하세요.","engines.use_own_app_folder": "{0}.app이 들어 있는 폴더 선택","engines.progress_detail": "{0} / {1} · {2}/s · {3}","engines.eta.estimating": "남은 시간 계산 중…",diff --git a/src/Filee.App/Assets/i18n/pl.json b/src/Filee.App/Assets/i18n/pl.jsonindex 2e2da36b..15e82d6b 100644--- a/src/Filee.App/Assets/i18n/pl.json+++ b/src/Filee.App/Assets/i18n/pl.json@@ -147,6 +147,7 @@"engine.ghostscript.description": "EPS, PostScript i starsze pliki Illustratora do PDF i obrazów; PDF do EPS i PS.","engine.calibre.description": "Rzadkie formaty e-booków (LIT, LRF, CHM, PDB, RB…) i zapis dla Kindle (MOBI, AZW3).\nUmieszczony nad HWPX writer tworzy też PDF-y e-booków (znacznie szybciej przy długich książkach).","engine.libreoffice.description": "Odczytuje starsze i rzadsze formaty: DOC, PPT, Pages, Keynote, Numbers, WordPerfect, ClarisWorks, Publisher, Visio…\nZapisuje też jako DOC, XLS, PPT, ODT, ODP, szablony OpenDocument i EMF.",+ "engine.image-ocr.description": "Wyodrębnia tekst ze skanowanych plików PDF i obrazów do formatu edytowalnych dokumentów (Word, tekst, HWPX).","profile.images": "Obrazy","profile.pdf": "PDF",@@ -534,6 +535,14 @@"presets.leave_out_pictures": "Pomiń obrazy (tylko tekst)","presets.leave_out_pictures_hint": "Wyłączone: obrazy trafiają do folderu „nazwa_files” obok pliku Markdown, który do nich linkuje.\nWłączone: tylko tekst, bez folderu.","presets.target_folder": "Folder (rozpakuj archiwa)",+ "presets.group.ocr": "Optyczne rozpoznawanie znaków (OCR)",+ "presets.ocr_enabled": "Rozpoznaj tekst w skanach i obrazach",+ "presets.ocr_hint": "Wyodrębnia edytowalny tekst ze skanowanych stron i obrazów do formatu Word, tekstu i dokumentów.",+ "presets.ocr_language": "Język rozpoznawania",+ "presets.ocr_language_hint": "Język używany do dokładnego rozpoznawania znaków i słów.",+ "presets.ocr_language.auto": "Automatycznie / Domyślny systemu",+ "presets.ocr_engine_required": "Wymaga Tesseract. Zainstaluj w Ustawienia → Silniki",+ "presets.ocr_disabled_linux": "System Linux nie ma wbudowanego OCR. Zainstaluj Tesseract w Ustawienia → Silniki, aby włączyć.","presets.group.media": "Wideo i audio","presets.media_quality": "Jakość","presets.media_quality_hint": "Gdy nowy format może przyjąć kodeki bez zmian (MOV z iPhone’a → MP4), wideo jest tylko przepakowywane.\nBez utraty jakości, z zachowaniem HDR, w kilka sekund.\nMała zawsze koduje ponownie, by zmniejszyć plik.",@@ -669,6 +678,11 @@"engines.guidance.linux.libreoffice": "Zainstaluj menedżerem pakietów (Ubuntu/Debian: sudo apt install libreoffice), potem wybierz „Sprawdź ponownie”.\nOtwieranie HWP wymaga też Javy i rozszerzenia H2Orestart.","engines.guidance.linux.ghostscript": "Zainstaluj menedżerem pakietów (Ubuntu/Debian: sudo apt install ghostscript), potem wybierz „Sprawdź ponownie”.","engines.guidance.windows": "Filee nie ma do pobrania {0} dla tego komputera.\nWybierz „Użyj mojej kopii…” i wskaż zainstalowaną kopię.",+ "engines.package.tesseract.name": "Tesseract (wielojęzyczne rozpoznawanie tekstu)",+ "engines.package.tesseract.description": "Rozpoznaje tekst w ponad 100 językach, w tym dewanagari (nepalski/hindi), arabskim i rzadkich pismach.\nNa systemie macOS Apple Vision działa natywnie dla języka angielskiego, CJK i języków europejskich bez tego pakietu.",+ "engines.guidance.macos.tesseract": "Zainstaluj Tesseract za pomocą Homebrew (brew install tesseract tesseract-lang), a następnie wybierz „Sprawdź ponownie”.\nLub wybierz „Użyj mojej kopii…” i wskaż folder bin.",+ "engines.guidance.linux.tesseract": "Zainstaluj Tesseract za pomocą menedżera pakietów (Ubuntu/Debian: sudo apt install tesseract-ocr tesseract-ocr-all), a następnie wybierz „Sprawdź ponownie”.\nLub wybierz „Użyj mojej kopii…” i wskaż folder programu.",+ "engines.guidance.windows.tesseract": "Zainstaluj Tesseract z UB-Mannheim (github.com/UB-Mannheim/tesseract), a następnie wybierz „Sprawdź ponownie” lub wskaż folder instalacji.","engines.use_own_app_folder": "Wybierz folder zawierający {0}.app","engines.progress_detail": "{0} z {1} · {2}/s · {3}","engines.eta.estimating": "szacowanie pozostałego czasu…",diff --git a/src/Filee.App/Assets/i18n/pt-BR.json b/src/Filee.App/Assets/i18n/pt-BR.jsonindex 0b7cd7cd..ad32a285 100644--- a/src/Filee.App/Assets/i18n/pt-BR.json+++ b/src/Filee.App/Assets/i18n/pt-BR.json@@ -147,6 +147,7 @@"engine.ghostscript.description": "EPS, PostScript e arquivos antigos do Illustrator para PDF e imagens; PDF para EPS e PS.","engine.calibre.description": "Formatos raros de e-book (LIT, LRF, CHM, PDB, RB…) e saída Kindle (MOBI, AZW3).\nAcima do gravador de HWPX, também gera os PDFs de e-books (bem mais rápido em livros longos).","engine.libreoffice.description": "Lê formatos antigos e menos comuns: DOC, PPT, Pages, Keynote, Numbers, WordPerfect, ClarisWorks, Publisher, Visio…\nTambém salva como DOC, XLS, PPT, ODT, ODP, modelos OpenDocument e EMF.",+ "engine.image-ocr.description": "Extrai texto de PDFs digitalizados e imagens para documentos editáveis (Word, texto, HWPX).","profile.images": "Imagens","profile.pdf": "PDF",@@ -534,6 +535,14 @@"presets.leave_out_pictures": "Deixar as imagens de fora (só texto)","presets.leave_out_pictures_hint": "Desligado: as imagens são salvas numa pasta “nome_files” ao lado do arquivo Markdown e referenciadas nele.\nLigado: só o texto, sem pasta.","presets.target_folder": "Pasta (extrair arquivos compactados)",+ "presets.group.ocr": "Reconhecimento Óptico de Caracteres (OCR)",+ "presets.ocr_enabled": "Reconhecer texto em digitalizações e imagens",+ "presets.ocr_hint": "Extrai texto editável de páginas digitalizadas e imagens para Word, texto e documentos.",+ "presets.ocr_language": "Idioma de reconhecimento",+ "presets.ocr_language_hint": "Idioma usado para reconhecer caracteres e palavras com precisão.",+ "presets.ocr_language.auto": "Automático / Padrão do sistema",+ "presets.ocr_engine_required": "Requer Tesseract. Instale em Configurações → Motores",+ "presets.ocr_disabled_linux": "O Linux não possui OCR integrado. Instale o Tesseract em Configurações → Motores para ativar.","presets.group.media": "Vídeo e áudio","presets.media_quality": "Qualidade","presets.media_quality_hint": "Quando o novo formato pode guardar os codecs como estão (MOV do iPhone → MP4), o vídeo só é reempacotado.\nSem perda de qualidade, HDR mantido, pronto em segundos.\nPequeno sempre recodifica para deixar o arquivo menor.",@@ -669,6 +678,11 @@"engines.guidance.linux.libreoffice": "Instale com o gerenciador de pacotes (Ubuntu/Debian: sudo apt install libreoffice) e escolha “Verificar de novo”.\nAbrir HWP também precisa de Java e da extensão H2Orestart.","engines.guidance.linux.ghostscript": "Instale com o gerenciador de pacotes (Ubuntu/Debian: sudo apt install ghostscript) e escolha “Verificar de novo”.","engines.guidance.windows": "O Filee não tem download de {0} para este PC.\nEscolha “Usar minha cópia…” e selecione uma cópia instalada.",+ "engines.package.tesseract.name": "Tesseract (reconhecimento de texto multilíngue)",+ "engines.package.tesseract.description": "Reconhece texto em mais de 100 idiomas, incluindo devanágari (nepalês/híndi), árabe e alfabetos raros.\nNo macOS, o Apple Vision funciona nativamente para inglês, CJK e idiomas europeus sem ele.",+ "engines.guidance.macos.tesseract": "Instale o Tesseract com o Homebrew (brew install tesseract tesseract-lang) e selecione “Verificar novamente”.\nOu escolha “Usar minha cópia…” e selecione a pasta bin.",+ "engines.guidance.linux.tesseract": "Instale o Tesseract com o gerenciador de pacotes (Ubuntu/Debian: sudo apt install tesseract-ocr tesseract-ocr-all) e selecione “Verificar novamente”.\nOu escolha “Usar minha cópia…” e selecione a pasta do programa.",+ "engines.guidance.windows.tesseract": "Instale o Tesseract do UB-Mannheim (github.com/UB-Mannheim/tesseract) e selecione “Verificar novamente” ou selecione sua pasta de instalação.","engines.use_own_app_folder": "Escolha a pasta que contém {0}.app","engines.progress_detail": "{0} de {1} · {2}/s · {3}","engines.eta.estimating": "estimando o tempo restante…",diff --git a/src/Filee.App/Assets/i18n/ru.json b/src/Filee.App/Assets/i18n/ru.jsonindex 3eb526c7..a12e691e 100644--- a/src/Filee.App/Assets/i18n/ru.json+++ b/src/Filee.App/Assets/i18n/ru.json@@ -147,6 +147,7 @@"engine.ghostscript.description": "EPS, PostScript и старые файлы Illustrator — в PDF и картинки; PDF — в EPS и PS.","engine.calibre.description": "Редкие форматы электронных книг (LIT, LRF, CHM, PDB, RB…) и вывод для Kindle (MOBI, AZW3).\nЕсли стоит выше HWPX writer, он же делает PDF из электронных книг (для длинных книг намного быстрее).","engine.libreoffice.description": "Читает старые и редкие форматы: DOC, PPT, Pages, Keynote, Numbers, WordPerfect, ClarisWorks, Publisher, Visio…\nА ещё сохраняет в DOC, XLS, PPT, ODT, ODP, шаблоны OpenDocument и EMF.",+ "engine.image-ocr.description": "Извлекает текст из отсканированных PDF и изображений в редактируемые документы (Word, текст, HWPX).","profile.images": "Изображения","profile.pdf": "PDF",@@ -534,6 +535,14 @@"presets.leave_out_pictures": "Без картинок (только текст)","presets.leave_out_pictures_hint": "Выключено: картинки сохраняются в папку «имя_files» рядом с файлом Markdown и связаны с ним ссылками.\nВключено: только текст, без папки.","presets.target_folder": "Папка (распаковка архивов)",+ "presets.group.ocr": "Оптическое распознавание символов (OCR)",+ "presets.ocr_enabled": "Распознавать текст в сканах и изображениях",+ "presets.ocr_hint": "Извлекает редактируемый текст из отсканированных страниц и изображений в Word, текст и документы.",+ "presets.ocr_language": "Язык распознавания",+ "presets.ocr_language_hint": "Язык для точного распознавания символов и слов.",+ "presets.ocr_language.auto": "Автоматически / По умолчанию",+ "presets.ocr_engine_required": "Требуется Tesseract. Установите его в Настройки → Движки",+ "presets.ocr_disabled_linux": "В Linux нет встроенного OCR. Установите Tesseract в Настройки → Движки, чтобы включить.","presets.group.media": "Видео и аудио","presets.media_quality": "Качество","presets.media_quality_hint": "Если новый формат может принять кодеки как есть (MOV с iPhone → MP4), видео только переупаковывается.\nБез потери качества, HDR сохраняется, за секунды.\n«Компактное» всегда перекодирует, чтобы уменьшить файл.",@@ -669,6 +678,11 @@"engines.guidance.linux.libreoffice": "Установите менеджером пакетов (Ubuntu/Debian: sudo apt install libreoffice), затем выберите «Проверить снова».\nДля открытия HWP также нужны Java и расширение H2Orestart.","engines.guidance.linux.ghostscript": "Установите менеджером пакетов (Ubuntu/Debian: sudo apt install ghostscript), затем выберите «Проверить снова».","engines.guidance.windows": "Filee не может скачать {0} для этого компьютера.\nВыберите «Использовать свою копию…» и укажите установленную копию.",+ "engines.package.tesseract.name": "Tesseract (многоязычное распознавание текста)",+ "engines.package.tesseract.description": "Распознает текст на более чем 100 языках, включая деванагари (непальский/хинди), арабский и редкие письменности.\nВ macOS Apple Vision работает нативно для английского, CJK и европейских языков без него.",+ "engines.guidance.macos.tesseract": "Установите Tesseract с помощью Homebrew (brew install tesseract tesseract-lang), затем выберите «Проверить снова».\nИли выберите «Использовать мою копию…» и укажите папку bin.",+ "engines.guidance.linux.tesseract": "Установите Tesseract с помощью пакетного менеджера (Ubuntu/Debian: sudo apt install tesseract-ocr tesseract-ocr-all), затем выберите «Проверить снова».\nИли выберите «Использовать мою копию…» и укажите папку программы.",+ "engines.guidance.windows.tesseract": "Установите Tesseract из UB-Mannheim (github.com/UB-Mannheim/tesseract), затем выберите «Проверить снова» или укажите папку установки.","engines.use_own_app_folder": "Выберите папку, в которой лежит {0}.app","engines.progress_detail": "{0} из {1} · {2}/с · {3}","engines.eta.estimating": "оцениваем, сколько осталось…",diff --git a/src/Filee.App/Assets/i18n/zh-CN.json b/src/Filee.App/Assets/i18n/zh-CN.jsonindex 6247a2b0..e3342628 100644--- a/src/Filee.App/Assets/i18n/zh-CN.json+++ b/src/Filee.App/Assets/i18n/zh-CN.json@@ -147,6 +147,7 @@"engine.ghostscript.description": "EPS、PostScript 和旧版 Illustrator 文件转为 PDF 和图片;PDF 转为 EPS 和 PS。","engine.calibre.description": "少见的电子书格式(LIT、LRF、CHM、PDB、RB…)以及 Kindle 输出(MOBI、AZW3)。\n排在 HWPX writer 之上时,也由它把电子书转为 PDF(长书快得多)。","engine.libreoffice.description": "读取较旧或少见的格式,如 DOC、PPT、Pages、Keynote、Numbers、WordPerfect、ClarisWorks、Publisher、Visio。\n也能保存为 DOC、XLS、PPT、ODT、ODP、OpenDocument 模板或 EMF。",+ "engine.image-ocr.description": "从扫描 PDF 和图片中提取文字并转换为可编辑文档(Word、文本、HWPX)。","profile.images": "图片","profile.pdf": "PDF",@@ -534,6 +535,14 @@"presets.leave_out_pictures": "不保留图片(仅文本)","presets.leave_out_pictures_hint": "关闭时,图片保存在 Markdown 文件旁的“名称_files”文件夹中,并从文件中链接。\n开启时,只保留文字,不生成文件夹。","presets.target_folder": "文件夹(解压缩)",+ "presets.group.ocr": "光学字符识别 (OCR)",+ "presets.ocr_enabled": "识别扫描件和图像中的文本",+ "presets.ocr_hint": "从扫描页面和图像中提取可编辑文本并转换为 Word、文本及文档。",+ "presets.ocr_language": "识别语言",+ "presets.ocr_language_hint": "用于准确识别字符和文字的语言。",+ "presets.ocr_language.auto": "自动 / 系统默认",+ "presets.ocr_engine_required": "需要 Tesseract。请在“设置 → 转换引擎”中安装",+ "presets.ocr_disabled_linux": "Linux 没有内置 OCR。请在“设置 → 转换引擎”中安装 Tesseract 以启用。","presets.group.media": "视频和音频","presets.media_quality": "质量","presets.media_quality_hint": "新格式能直接容纳原编码时(如 iPhone MOV → MP4),只重新封装。\n画质无损、保留 HDR,几秒即可完成。\n“小”总会重新编码以减小文件。",@@ -669,6 +678,11 @@"engines.guidance.linux.libreoffice": "用软件包管理器安装(Ubuntu/Debian 下运行 sudo apt install libreoffice)后点“重新检查”。\n打开 HWP 还需要 Java 和 H2Orestart 扩展。","engines.guidance.linux.ghostscript": "用软件包管理器安装(Ubuntu/Debian 下运行 sudo apt install ghostscript)后点“重新检查”。","engines.guidance.windows": "没有适用于这台电脑的 {0} 下载。\n点“使用我的程序…”选择已安装的程序。",+ "engines.package.tesseract.name": "Tesseract(多语言文字识别)",+ "engines.package.tesseract.description": "支持梵文(尼泊尔语/印地语)、阿拉伯语等 100 多种语言的文字识别。\n在 macOS 上,系统内置 Apple Vision 可直接识别英语、中日韩及主要欧洲语言。",+ "engines.guidance.macos.tesseract": "使用 Homebrew 安装 Tesseract(brew install tesseract tesseract-lang),然后选择“重新检查”。\n或选择“使用我的副本…”并指定其 bin 文件夹。",+ "engines.guidance.linux.tesseract": "使用包管理器安装 Tesseract(Ubuntu/Debian: sudo apt install tesseract-ocr tesseract-ocr-all),然后选择“重新检查”。\n或选择“使用我的副本…”并指定其程序文件夹。",+ "engines.guidance.windows.tesseract": "从 UB-Mannheim (github.com/UB-Mannheim/tesseract) 安装 Tesseract,然后选择“重新检查”或指定其安装文件夹。","engines.use_own_app_folder": "选择包含 {0}.app 的文件夹","engines.progress_detail": "{0} / {1} · {2}/s · {3}","engines.eta.estimating": "正在估算剩余时间…",diff --git a/src/Filee.App/Services/AppHost.cs b/src/Filee.App/Services/AppHost.csindex f06ce0dc..de8a4e87 100644--- a/src/Filee.App/Services/AppHost.cs+++ b/src/Filee.App/Services/AppHost.cs@@ -4,6 +4,7 @@ using Filee.App.Services.Triggers;using Filee.App.ViewModels;using Filee.Core.Conversion;using Filee.Core.Localization;+using Filee.Core.Ocr;using Filee.Core.Platform;using Filee.Core.Settings;using Filee.Core.Web;@@ -48,6 +49,8 @@ public static class AppHostservices.AddSingleton<ElevationService>();// Conversion engines+ if (OperatingSystem.IsMacOS())+ OcrEngineRegistry.Register(new MacOsVisionOcrEngine());services.AddSingleton(_ => new EngineEnvironment(dataDir));services.AddSingleton(sp => EngineRegistry.CreateAll(sp.GetRequiredService<EngineEnvironment>(), sp.GetRequiredService<ILoggerFactory>()));services.AddSingleton(sp => new ConverterCatalog(sp.GetRequiredService<IReadOnlyList<IConverter>>()));diff --git a/src/Filee.App/Services/EngineDownloadService.cs b/src/Filee.App/Services/EngineDownloadService.csindex 2bd41056..14cd8cac 100644--- a/src/Filee.App/Services/EngineDownloadService.cs+++ b/src/Filee.App/Services/EngineDownloadService.cs@@ -44,7 +44,9 @@ public sealed partial class EnginePackageState : ObservableObjectpublic string Sizes => EngineDownloads.CanDownload(Package)? _loc.Format("engines.package.size", FormatBytes(EngineDownloads.DownloadSize(Package)), FormatBytes(Package.InstalledSize))- : _loc["engines.package.system_required"];+ : CanSystemInstall(Package)+ ? _loc.Format("engines.package.size", FormatBytes(350_000_000), FormatBytes(Package.InstalledSize))+ : _loc["engines.package.system_required"];/// <summary>/// How to get an engine Filee doesn't download on this system (FFmpeg and Ghostscript from Homebrew on macOS,@@ -54,7 +56,7 @@ public sealed partial class EnginePackageState : ObservableObject{get{- if (EngineDownloads.CanDownload(Package))+ if (EngineDownloads.CanDownload(Package) || CanSystemInstall(Package))return null;var system = OperatingSystem.IsMacOS() ? "macos" : OperatingSystem.IsLinux() ? "linux" : "windows";var key = $"engines.guidance.{system}.{Package.Id}";@@ -114,7 +116,7 @@ public sealed partial class EnginePackageState : ObservableObject/// <summary>The engine is there: Filee's download or the user's own copy.</summary>public bool IsInstalled => Status is EnginePackageStatus.Installed or EnginePackageStatus.OwnCopy;- public bool CanRemove => Status == EnginePackageStatus.Installed;+ public bool CanRemove => Status == EnginePackageStatus.Installed || (Package.Id == "tesseract" && Status == EnginePackageStatus.OwnCopy && FromSystem);/// <summary>The user chose this copy ("Stop using it" applies).</summary>public bool UsesOwnCopy => Status == EnginePackageStatus.OwnCopy && !FromSystem;public bool ShowsCopyFolder => Status == EnginePackageStatus.OwnCopy;@@ -122,8 +124,33 @@ public sealed partial class EnginePackageState : ObservableObject|| (Status == EnginePackageStatus.OwnCopy && FromSystem);public bool IsBusy => Status is EnginePackageStatus.Queued or EnginePackageStatus.Downloading or EnginePackageStatus.Unpacking;/// <summary>Also when a copy on the PC is used: Filee's download is the tested version and wins once installed.</summary>- public bool CanInstall => EngineDownloads.CanDownload(Package) && (Status is EnginePackageStatus.NotInstalled or EnginePackageStatus.Failed- || (Status == EnginePackageStatus.OwnCopy && FromSystem));+ public bool CanInstall => (EngineDownloads.CanDownload(Package) && (Status is EnginePackageStatus.NotInstalled or EnginePackageStatus.Failed || (Status == EnginePackageStatus.OwnCopy && FromSystem)))+ || (CanSystemInstall(Package) && Status is EnginePackageStatus.NotInstalled or EnginePackageStatus.Failed);++ /// <summary>+ /// Whether the app can drive an in-place install of <paramref name="package"/> on this machine: Homebrew on+ /// macOS, winget on Windows. Linux has no handler here (package managers vary too much to drive one+ /// generically), so it keeps the same guidance-text + "Use my copy…" flow as every other engine.+ /// </summary>+ internal static bool CanSystemInstall(EnginePackage package) =>+ package.Id == "tesseract" && (OperatingSystem.IsMacOS() ? LocateBrew() is not null : OperatingSystem.IsWindows());++ /// <summary>Homebrew's own binary, at its two standard prefixes or on PATH; null when it isn't installed.</summary>+ internal static string? LocateBrew()+ {+ if (File.Exists("/opt/homebrew/bin/brew")) return "/opt/homebrew/bin/brew";+ if (File.Exists("/usr/local/bin/brew")) return "/usr/local/bin/brew";+ var path = Environment.GetEnvironmentVariable("PATH");+ if (path is not null)+ {+ foreach (var part in path.Split(':'))+ {+ var candidate = Path.Combine(part, "brew");+ if (File.Exists(candidate)) return candidate;+ }+ }+ return null;+ }public string StatusText => Status switch{@@ -258,32 +285,40 @@ public sealed class EngineDownloadServiceawait _oneAtATime.WaitAsync(cancellation.Token);try{- // Progress<T> reports on the UI thread it was created on.- var clock = Stopwatch.StartNew();- var rate = new TransferRate();- var progress = new Progress<EngineInstallProgress>(p =>+ if (state.Package.Id == "tesseract" && !EngineDownloads.CanDownload(state.Package)){- if (!state.IsBusy)- return;- state.Status = p.Stage == EngineInstallStage.Downloading ? EnginePackageStatus.Downloading : EnginePackageStatus.Unpacking;- state.Progress = p.Fraction * 100;- if (p.Stage == EngineInstallStage.Downloading && p.BytesTotal > 0)- {- rate.Add(clock.Elapsed, p.BytesDone);- state.ShowTransfer(p.BytesDone, p.BytesTotal, rate);- }- else+ await InstallTesseractSystemAsync(state, cancellation.Token);+ }+ else+ {+ // Progress<T> reports on the UI thread it was created on.+ var clock = Stopwatch.StartNew();+ var rate = new TransferRate();+ var progress = new Progress<EngineInstallProgress>(p =>{- state.ProgressDetail = "";- }- });- await Task.Run(() => _installer.InstallAsync(state.Package, progress, cancellation.Token), cancellation.Token);+ if (!state.IsBusy)+ return;+ state.Status = p.Stage == EngineInstallStage.Downloading ? EnginePackageStatus.Downloading : EnginePackageStatus.Unpacking;+ state.Progress = p.Fraction * 100;+ if (p.Stage == EngineInstallStage.Downloading && p.BytesTotal > 0)+ {+ rate.Add(clock.Elapsed, p.BytesDone);+ state.ShowTransfer(p.BytesDone, p.BytesTotal, rate);+ }+ else+ {+ state.ProgressDetail = "";+ }+ });+ await Task.Run(() => _installer.InstallAsync(state.Package, progress, cancellation.Token), cancellation.Token);+ }}finally{_oneAtATime.Release();}- state.Status = EnginePackageStatus.Installed;+ EngineEnvironment.ForgetSystemCopies();+ state.Status = StatusOf(state);_catalog.Refresh();if (state.Package.ConverterIds.Contains("libreoffice"))WarmUpInBackground(TimeSpan.Zero);@@ -305,6 +340,142 @@ public sealed class EngineDownloadService}}+ private static async Task InstallTesseractSystemAsync(EnginePackageState state, CancellationToken cancellationToken)+ {+ try+ {+ if (OperatingSystem.IsMacOS())+ {+ var brew = EnginePackageState.LocateBrew();+ if (brew is null)+ throw new InvalidOperationException("Homebrew was not found. Install Homebrew from https://brew.sh first.");++ state.Status = EnginePackageStatus.Downloading;+ state.Progress = 5;+ state.ProgressDetail = "Starting Homebrew install...";++ var psi = new ProcessStartInfo+ {+ FileName = brew,+ ArgumentList = { "install", "tesseract", "tesseract-lang" },+ RedirectStandardOutput = true,+ RedirectStandardError = true,+ UseShellExecute = false,+ CreateNoWindow = true+ };+ var brewDir = Path.GetDirectoryName(brew);+ var currentPath = Environment.GetEnvironmentVariable("PATH") ?? "";+ if (brewDir is not null && !currentPath.Contains(brewDir))+ psi.Environment["PATH"] = brewDir + ":" + currentPath;+ psi.Environment["HOMEBREW_NO_AUTO_UPDATE"] = "1";+ psi.Environment["HOMEBREW_NO_ENV_HINTS"] = "1";++ using var process = new Process { StartInfo = psi };+ process.OutputDataReceived += (_, e) =>+ {+ if (!string.IsNullOrWhiteSpace(e.Data))+ {+ Dispatcher.UIThread.Post(() =>+ {+ var line = e.Data.Trim();+ state.ProgressDetail = line;+ if (line.StartsWith("==> Downloading", StringComparison.OrdinalIgnoreCase))+ {+ state.Status = EnginePackageStatus.Downloading;+ if (state.Progress < 50) state.Progress += 10;+ }+ else if (line.StartsWith("==> Pouring", StringComparison.OrdinalIgnoreCase) ||+ line.StartsWith("==> Installing", StringComparison.OrdinalIgnoreCase))+ {+ state.Status = EnginePackageStatus.Unpacking;+ if (state.Progress < 90) state.Progress += 10;+ }+ });+ }+ };+ process.ErrorDataReceived += (_, e) =>+ {+ if (!string.IsNullOrWhiteSpace(e.Data))+ {+ Dispatcher.UIThread.Post(() =>+ {+ state.ProgressDetail = e.Data.Trim();+ });+ }+ };++ process.Start();+ process.BeginOutputReadLine();+ process.BeginErrorReadLine();++ using (cancellationToken.Register(() =>+ {+ try { process.Kill(true); } catch { }+ }))+ {+ await process.WaitForExitAsync(cancellationToken);+ }++ if (process.ExitCode != 0)+ throw new InvalidOperationException($"Homebrew install failed with exit code {process.ExitCode}.");++ state.Progress = 100;+ state.ProgressDetail = "";+ }+ else if (OperatingSystem.IsWindows())+ {+ state.Status = EnginePackageStatus.Downloading;+ state.Progress = 10;+ state.ProgressDetail = "Installing Tesseract with winget...";++ var psi = new ProcessStartInfo+ {+ FileName = "winget",+ ArgumentList = { "install", "UB-Mannheim.TesseractOCR", "--accept-source-agreements", "--accept-package-agreements" },+ RedirectStandardOutput = true,+ RedirectStandardError = true,+ UseShellExecute = false,+ CreateNoWindow = true+ };++ using var process = new Process { StartInfo = psi };+ process.OutputDataReceived += (_, e) =>+ {+ if (!string.IsNullOrWhiteSpace(e.Data))+ {+ Dispatcher.UIThread.Post(() =>+ {+ state.ProgressDetail = e.Data.Trim();+ });+ }+ };++ process.Start();+ process.BeginOutputReadLine();++ using (cancellationToken.Register(() =>+ {+ try { process.Kill(true); } catch { }+ }))+ {+ await process.WaitForExitAsync(cancellationToken);+ }++ if (process.ExitCode != 0)+ throw new InvalidOperationException($"winget install failed with exit code {process.ExitCode}.");+ }+ else+ {+ throw new PlatformNotSupportedException("Automatic installation is not supported on this platform. Please install tesseract using your package manager.");+ }+ }+ finally+ {+ state.Cancellation = null;+ state.ProgressDetail = "";+ }+ }+public void Cancel(EnginePackageState state) => state.Cancellation?.Cancel();/// <summary>@@ -346,8 +517,45 @@ public sealed class EngineDownloadServicereturn false;try{+ if (state.Package.Id == "tesseract")+ {+ if (OperatingSystem.IsMacOS())+ {+ var brew = EnginePackageState.LocateBrew();+ if (brew is not null)+ {+ var psi = new ProcessStartInfo+ {+ FileName = brew,+ ArgumentList = { "uninstall", "tesseract", "tesseract-lang" },+ UseShellExecute = false,+ CreateNoWindow = true+ };+ var brewDir = Path.GetDirectoryName(brew);+ var currentPath = Environment.GetEnvironmentVariable("PATH") ?? "";+ if (brewDir is not null && !currentPath.Contains(brewDir))+ psi.Environment["PATH"] = brewDir + ":" + currentPath;+ psi.Environment["HOMEBREW_NO_AUTO_UPDATE"] = "1";+ psi.Environment["HOMEBREW_NO_ENV_HINTS"] = "1";+ using var process = Process.Start(psi);+ process?.WaitForExit();+ }+ }+ else if (OperatingSystem.IsWindows())+ {+ using var process = Process.Start(new ProcessStartInfo+ {+ FileName = "winget",+ ArgumentList = { "uninstall", "UB-Mannheim.TesseractOCR" },+ UseShellExecute = false,+ CreateNoWindow = true+ });+ process?.WaitForExit();+ }+ EngineEnvironment.ForgetSystemCopies();+ }_installer.Uninstall(state.Package);- state.Status = EnginePackageStatus.NotInstalled;+ state.Status = StatusOf(state);_catalog.Refresh();return true;}diff --git a/src/Filee.App/ViewModels/PresetEditorViewModel.cs b/src/Filee.App/ViewModels/PresetEditorViewModel.csindex d76f638b..43bc2b5f 100644--- a/src/Filee.App/ViewModels/PresetEditorViewModel.cs+++ b/src/Filee.App/ViewModels/PresetEditorViewModel.cs@@ -6,6 +6,7 @@using CommunityToolkit.Mvvm.ComponentModel;using Filee.Core.Formats;using Filee.Core.Localization;+using Filee.Core.Ocr;using Filee.Core.Presets;namespace Filee.App.ViewModels;@@ -97,6 +98,94 @@ public sealed partial class PresetEditorViewModel : ObservableObjectAnimationSpeeds = [.. ImageOptions.AnimationSpeeds.Append(_edit.Image.PlaybackSpeed).Distinct().Order().Select(s => new Choice<double>(s, s == 1 ? loc["presets.animation_speed.original"] : SpeedLabel(s)))];+ OcrLanguages =+ [+ new Choice<string?>(null, loc["presets.ocr_language.auto"]),+ // Popular / Primary Languages+ new Choice<string?>("en", "English"),+ new Choice<string?>("ne", "Nepali / नेपाली"),+ new Choice<string?>("hi", "Hindi / हिन्दी"),+ new Choice<string?>("zh-Hans", "Chinese Simplified / 简体中文"),+ new Choice<string?>("zh-Hant", "Chinese Traditional / 繁體中文"),+ new Choice<string?>("ja", "Japanese / 日本語"),+ new Choice<string?>("ko", "Korean / 한국어"),+ new Choice<string?>("es", "Spanish / Español"),+ new Choice<string?>("fr", "French / Français"),+ new Choice<string?>("de", "German / Deutsch"),+ new Choice<string?>("it", "Italian / Italiano"),+ new Choice<string?>("pt", "Portuguese / Português"),+ new Choice<string?>("ru", "Russian / Русский"),+ new Choice<string?>("ar", "Arabic / العربية"),+ new Choice<string?>("th", "Thai / ไทย"),+ new Choice<string?>("el", "Greek / Ελληνικά"),+ new Choice<string?>("uk", "Ukrainian / Українська"),+ new Choice<string?>("tr", "Turkish / Türkçe"),+ new Choice<string?>("vi", "Vietnamese / Tiếng Việt"),+ new Choice<string?>("id", "Indonesian / Bahasa Indonesia"),+ new Choice<string?>("pl", "Polish / Polski"),+ new Choice<string?>("nl", "Dutch / Nederlands"),+ new Choice<string?>("sv", "Swedish / Svenska"),+ new Choice<string?>("da", "Danish / Dansk"),+ new Choice<string?>("fi", "Finnish / Suomi"),+ new Choice<string?>("no", "Norwegian / Norsk"),+ new Choice<string?>("cs", "Czech / Čeština"),+ new Choice<string?>("sk", "Slovak / Slovenčina"),+ new Choice<string?>("hu", "Hungarian / Magyar"),+ new Choice<string?>("ro", "Romanian / Română"),+ new Choice<string?>("bg", "Bulgarian / Български"),+ new Choice<string?>("hr", "Croatian / Hrvatski"),+ new Choice<string?>("sr-Cyrl", "Serbian (Cyrillic) / Српски"),+ new Choice<string?>("sr-Latn", "Serbian (Latin) / Srpski"),+ new Choice<string?>("sl", "Slovenian / Slovenščina"),+ new Choice<string?>("et", "Estonian / Eesti"),+ new Choice<string?>("lv", "Latvian / Latviešu"),+ new Choice<string?>("lt", "Lithuanian / Lietuvių"),+ new Choice<string?>("be", "Belarusian / Беларуская"),+ // South Asian / Indic Languages+ new Choice<string?>("sa", "Sanskrit / संस्कृतम्"),+ new Choice<string?>("mr", "Marathi / मराठी"),+ new Choice<string?>("ta", "Tamil / தமிழ்"),+ new Choice<string?>("te", "Telugu / తెలుగు"),+ new Choice<string?>("ur", "Urdu / اردو"),+ new Choice<string?>("fa", "Persian / فارسی"),+ new Choice<string?>("ps", "Pashto / پښتو"),+ new Choice<string?>("sd", "Sindhi / سنڌي"),+ new Choice<string?>("ug", "Uyghur / ئۇيغۇرچە"),+ // European, Regional & World Languages+ new Choice<string?>("af", "Afrikaans"),+ new Choice<string?>("az", "Azerbaijani / Azərbaycan"),+ new Choice<string?>("bs", "Bosnian / Bosanski"),+ new Choice<string?>("ca", "Catalan / Català"),+ new Choice<string?>("cy", "Welsh / Cymraeg"),+ new Choice<string?>("eu", "Basque / Euskara"),+ new Choice<string?>("ga", "Irish / Gaeilge"),+ new Choice<string?>("gl", "Galician / Galego"),+ new Choice<string?>("is", "Icelandic / Íslenska"),+ new Choice<string?>("ka", "Georgian / ქართული"),+ new Choice<string?>("ku", "Kurdish / Kurdî"),+ new Choice<string?>("la", "Latin / Latina"),+ new Choice<string?>("lb", "Luxembourgish / Lëtzebuergesch"),+ new Choice<string?>("mi", "Maori / Te Reo Māori"),+ new Choice<string?>("mk", "Macedonian / Македонски"),+ new Choice<string?>("mn", "Mongolian / Монгол"),+ new Choice<string?>("ms", "Malay / Bahasa Melayu"),+ new Choice<string?>("mt", "Maltese / Malti"),+ new Choice<string?>("oc", "Occitan / Lenga d'òc"),+ new Choice<string?>("qu", "Quechua / Runasimi"),+ new Choice<string?>("sq", "Albanian / Shqip"),+ new Choice<string?>("sw", "Swahili / Kiswahili"),+ new Choice<string?>("tl", "Tagalog / Filipino"),+ new Choice<string?>("uz", "Uzbek / Oʻzbekcha"),+ new Choice<string?>("kk", "Kazakh / Қазақша"),+ new Choice<string?>("ky", "Kyrgyz / Кыргызча"),+ new Choice<string?>("tg", "Tajik / Тоҷикӣ"),+ new Choice<string?>("tt", "Tatar / Татарча"),+ ];+ // Vision and Tesseract each claim a real, limited set of languages (SupportedLanguages); the old PaddleOCR+ // engine claimed over 100 it could never actually recognize (no bundled model, no install path). Only offer+ // what a registered engine can really attempt, so picking a language never leads straight to "engine missing".+ OcrLanguages = [.. OcrLanguages.Where(c => c.Value is null || OcrEngineRegistry.AnyEngineSupports(c.Value))];+_name = string.IsNullOrWhiteSpace(_edit.Name) ? loc.DisplayName(_edit) : _edit.Name;_target = Targets.FirstOrDefault(t => t.Value == _edit.TargetFormat) ?? Targets[1];_resize = ResizeModes.First(r => r.Value == _edit.Image.Resize);@@ -163,6 +252,13 @@ public sealed partial class PresetEditorViewModel : ObservableObject_pattern = _edit.Output.FileNamePattern;_keepDates = _edit.Output.KeepDates;_zipResult = _edit.Output.ZipResult;+ _ocr = _edit.Document.Ocr;+ _selectedOcrLanguage = OcrLanguages.FirstOrDefault(l =>+ l.Value == _edit.Document.OcrLanguage ||+ (l.Value is not null && _edit.Document.OcrLanguage is not null &&+ (_edit.Document.OcrLanguage.StartsWith(l.Value, StringComparison.OrdinalIgnoreCase) ||+ l.Value.StartsWith(_edit.Document.OcrLanguage, StringComparison.OrdinalIgnoreCase))))+ ?? OcrLanguages[0];}public string PresetId => _original.Id;@@ -193,6 +289,7 @@ public sealed partial class PresetEditorViewModel : ObservableObject/// <summary>Playback speeds of animations, 0.25× to 4× (ImageOptions.AnimationSpeed).</summary>public IReadOnlyList<Choice<double>> AnimationSpeeds { get; }+ public IReadOnlyList<Choice<string?>> OcrLanguages { get; }[ObservableProperty, NotifyPropertyChangedFor(nameof(NameError), nameof(IsValid))]private string _name;@@ -206,7 +303,8 @@ public sealed partial class PresetEditorViewModel : ObservableObjectnameof(AudioCodecs), nameof(AudioCodecChoice), nameof(ShowAudioCodec),nameof(ShowArchive), nameof(ShowArchiveOptions), nameof(ShowExtractNote),nameof(ShowSaveEveryFrame), nameof(ShowAnimationSpeed), nameof(ShowTrace), nameof(ShowExtractMotionPhoto),- nameof(ShowLivePhotoChoice), nameof(LivePhotoKinds), nameof(LivePhotoChoice), nameof(IsLivePhoto), nameof(LivePhotoHint))]+ nameof(ShowLivePhotoChoice), nameof(LivePhotoKinds), nameof(LivePhotoChoice), nameof(IsLivePhoto), nameof(LivePhotoHint),+ nameof(ShowOcr), nameof(ShowOcrLanguage), nameof(ShowOcrMissingWarning), nameof(OcrDisabledOnLinux))]private Choice<string> _target;private LivePhotoKind _livePhotoValue;@@ -393,6 +491,8 @@ public sealed partial class PresetEditorViewModel : ObservableObject/// <summary>Markdown: the text only, no picture folder next to it (DocumentOptions.LeaveOutPictures).</summary>[ObservableProperty] private bool _leaveOutPictures;+ [ObservableProperty, NotifyPropertyChangedFor(nameof(ShowOcrLanguage)), NotifyPropertyChangedFor(nameof(ShowOcrMissingWarning))] private bool _ocr;+ [ObservableProperty, NotifyPropertyChangedFor(nameof(HasOcrEngine)), NotifyPropertyChangedFor(nameof(ShowOcrMissingWarning))] private Choice<string?> _selectedOcrLanguage;[ObservableProperty] private string _subfolderName;/// <summary>The subfolder only for a file that becomes several (OutputRule.SubfolderOnlyForSeveralFiles).</summary>[ObservableProperty] private bool _subfolderOnlyForSeveralFiles;@@ -497,6 +597,17 @@ public sealed partial class PresetEditorViewModel : ObservableObjectpublic bool ShowDocument => TargetId == "pdf";/// <summary>Markdown can't hold pictures: they go into a folder next to it, or are left out.</summary>public bool ShowLeaveOutPictures => TargetId == "md";++ /// <summary>OCR (Optical Character Recognition) for text recognition in scanned documents / images.</summary>+ public bool ShowOcr => !IsTool && (TargetCategory is FormatCategory.Document or FormatCategory.Text or FormatCategory.Ebook+ || TargetId is "docx" or "txt" or "md" or "hwpx" or "pdf" or "rtf" or "odt");+ public bool ShowOcrLanguage => ShowOcr && Ocr;+ public bool HasOcrEngine => OcrEngineRegistry.GetDefault(SelectedOcrLanguage?.Value) is not null;+ public bool HasAnyOcrEngine => OcrEngineRegistry.All.Any(e => e.IsAvailable);+ /// <summary>Linux has no built-in OCR (unlike Apple Vision on macOS), so Tesseract isn't just optional there: with+ /// nothing installed, OCR can't run at all, and the toggle is disabled rather than left on to fail at conversion time.</summary>+ public bool OcrDisabledOnLinux => ShowOcr && OperatingSystem.IsLinux() && !HasAnyOcrEngine;+ public bool ShowOcrMissingWarning => ShowOcr && !OcrDisabledOnLinux && (!HasAnyOcrEngine || (Ocr && !HasOcrEngine));public bool ShowMedia => ShowMediaQuality || ShowMaxHeight || ShowRemoveAudio || ShowAudioChannels || ShowVideoCodec || ShowAudioCodec;/// <summary>Mono / stereo for every audio target but AMR, which is mono by definition.</summary>public bool ShowAudioChannels => TargetIsAudio && TargetId != "amr";@@ -598,6 +709,8 @@ public sealed partial class PresetEditorViewModel : ObservableObject_original.Document.PdfA = PdfA;// Only where the option is shown, as with SaveEveryFrame below._original.Document.LeaveOutPictures = ShowLeaveOutPictures && LeaveOutPictures;+ _original.Document.Ocr = ShowOcr && Ocr;+ _original.Document.OcrLanguage = ShowOcr && Ocr ? SelectedOcrLanguage.Value : null;var media = _original.Media;media.Quality = MediaQuality.Value;diff --git a/src/Filee.App/Views/PresetEditorView.axaml b/src/Filee.App/Views/PresetEditorView.axamlindex 8c81df87..de37f45a 100644--- a/src/Filee.App/Views/PresetEditorView.axaml+++ b/src/Filee.App/Views/PresetEditorView.axaml@@ -296,6 +296,38 @@</StackPanel></Border>+ <!-- Documents: Optical Character Recognition (OCR) for scanned pages and images -->+ <Border x:Name="OcrCard" Classes="card" Padding="14" IsVisible="{Binding ShowOcr}">+ <StackPanel Spacing="10">+ <TextBlock Classes="h2" Text="{DynamicResource presets.group.ocr}" />+ <StackPanel Spacing="2">+ <CheckBox Content="{DynamicResource presets.ocr_enabled}" IsChecked="{Binding Ocr}" IsEnabled="{Binding !OcrDisabledOnLinux}" />+ <TextBlock Classes="caption" TextWrapping="Wrap" Margin="28,0,0,0" Text="{DynamicResource presets.ocr_hint}" />+ </StackPanel>+ <Border Classes="badge" Margin="28,2,0,0" IsVisible="{Binding OcrDisabledOnLinux}" HorizontalAlignment="Left">+ <StackPanel Orientation="Horizontal" Spacing="6">+ <PathIcon Width="12" Height="12" Data="{StaticResource Icon.Alert}" Foreground="{DynamicResource WarningBrush}" />+ <TextBlock Classes="caption" Foreground="{DynamicResource WarningBrush}" Text="{DynamicResource presets.ocr_disabled_linux}" />+ </StackPanel>+ </Border>+ <Border Classes="badge" Margin="28,2,0,0" IsVisible="{Binding ShowOcrMissingWarning}" HorizontalAlignment="Left">+ <StackPanel Orientation="Horizontal" Spacing="6">+ <PathIcon Width="12" Height="12" Data="{StaticResource Icon.Alert}" Foreground="{DynamicResource WarningBrush}" />+ <TextBlock Classes="caption" Foreground="{DynamicResource WarningBrush}" Text="{DynamicResource presets.ocr_engine_required}" />+ </StackPanel>+ </Border>+ <StackPanel Spacing="2" Margin="28,4,0,0" IsVisible="{Binding ShowOcrLanguage}">+ <Grid ColumnDefinitions="*,12,*">+ <StackPanel>+ <TextBlock Classes="label" Text="{DynamicResource presets.ocr_language}" />+ <ComboBox ItemsSource="{Binding OcrLanguages}" SelectedItem="{Binding SelectedOcrLanguage}" HorizontalAlignment="Stretch" />+ </StackPanel>+ </Grid>+ <TextBlock Classes="caption" TextWrapping="Wrap" Margin="0,4,0,0" Text="{DynamicResource presets.ocr_language_hint}" />+ </StackPanel>+ </StackPanel>+ </Border>+<!-- PDF options (render settings also apply when a PDF is converted to images) --><Border Classes="card" Padding="14" IsVisible="{Binding ShowPdf}"><StackPanel Spacing="10">diff --git a/src/Filee.Cli/CliHost.cs b/src/Filee.Cli/CliHost.csindex 14bb8b43..dae34da3 100644--- a/src/Filee.Cli/CliHost.cs+++ b/src/Filee.Cli/CliHost.cs@@ -34,6 +34,8 @@ internal sealed class CliHost : IAsyncDisposableEngineEnvironment.LibreOfficeForDocuments = store.Settings.LibreOfficeForDocuments;EngineEnvironment.LibreOfficeForSheets = store.Settings.LibreOfficeForSheets;var converters = EngineRegistry.CreateAll(new EngineEnvironment(store.Directory));+ if (OperatingSystem.IsMacOS())+ Filee.Core.Ocr.OcrEngineRegistry.Register(new Filee.Platform.MacOS.MacOsVisionOcrEngine());var catalog = new ConverterCatalog(converters) { Priority = store.Settings.EnginePriority };var queue = new JobQueue(catalog, EngineRegistry.FindPdfMerger(converters), combiner: EngineRegistry.FindFileCombiner(converters),tiffMerger: EngineRegistry.FindTiffMerger(converters));diff --git a/src/Filee.Cli/Filee.Cli.csproj b/src/Filee.Cli/Filee.Cli.csprojindex af728be3..1883ea6d 100644--- a/src/Filee.Cli/Filee.Cli.csproj+++ b/src/Filee.Cli/Filee.Cli.csproj@@ -14,6 +14,7 @@<ItemGroup><ProjectReference Include="..\Filee.Core\Filee.Core.csproj" /><ProjectReference Include="..\Filee.Engines\Filee.Engines.csproj" />+ <ProjectReference Include="..\Filee.Platform.MacOS\Filee.Platform.MacOS.csproj" Condition="$([MSBuild]::IsOSPlatform('OSX'))" /></ItemGroup><!-- Error messages and built-in preset names, in English (the app's translation file). -->diff --git a/src/Filee.Core/Ocr/IOcrEngine.cs b/src/Filee.Core/Ocr/IOcrEngine.csnew file mode 100644index 00000000..4f320abd--- /dev/null+++ b/src/Filee.Core/Ocr/IOcrEngine.cs@@ -0,0 +1,30 @@+// The contract every OCR engine implements (Apple Vision, Tesseract): recognize text and word boxes in an image,+// so callers like PdfDocumentReader can place the result back onto the page without knowing which engine ran.++namespace Filee.Core.Ocr;++/// <summary>+/// A service that extracts text and bounding boxes from images.+/// </summary>+public interface IOcrEngine+{+ /// <summary>Unique engine identifier (e.g. "vision", "tesseract").</summary>+ string Id { get; }++ /// <summary>Display name (e.g. "Apple Vision", "Tesseract OCR").</summary>+ string DisplayName { get; }++ /// <summary>Whether this engine is available and ready on the current machine.</summary>+ bool IsAvailable { get; }++ /// <summary>Supported languages (BCP-47 codes, or empty if generic/all).</summary>+ IReadOnlyList<string> SupportedLanguages { get; }++ /// <summary>+ /// Recognizes text in an image file.+ /// </summary>+ /// <param name="imagePath">Path to an image file (PNG, JPG, TIFF, etc.).</param>+ /// <param name="language">Optional language hint or code.</param>+ /// <param name="cancellationToken">Cancellation token.</param>+ Task<OcrPageResult> RecognizeAsync(string imagePath, string? language = null, CancellationToken cancellationToken = default);+}diff --git a/src/Filee.Core/Ocr/OcrEngineRegistry.cs b/src/Filee.Core/Ocr/OcrEngineRegistry.csnew file mode 100644index 00000000..504bdf7d--- /dev/null+++ b/src/Filee.Core/Ocr/OcrEngineRegistry.cs@@ -0,0 +1,135 @@+// Where callers (PdfDocumentReader, ImageOcrConverter, the preset editor) find an OCR engine without depending on+// Filee.Platform.MacOS or Filee.Engines directly: Tesseract registers itself through EngineRegistry.CreateAll on+// every platform, Apple Vision is discovered lazily and by reflection since it lives in a macOS-only assembly.++namespace Filee.Core.Ocr;++/// <summary>+/// Registry of available OCR engines.+/// </summary>+public static class OcrEngineRegistry+{+ private static readonly List<IOcrEngine> RegisteredEngines = [];+ private static readonly Lock Sync = new();++ public static void Register(IOcrEngine engine)+ {+ ArgumentNullException.ThrowIfNull(engine);+ lock (Sync)+ {+ if (!RegisteredEngines.Any(e => e.Id.Equals(engine.Id, StringComparison.OrdinalIgnoreCase)))+ {+ RegisteredEngines.Add(engine);+ }+ }+ }++ public static IReadOnlyList<IOcrEngine> All+ {+ get+ {+ lock (Sync)+ {+ DiscoverNativeEngine();+ return [.. RegisteredEngines];+ }+ }+ }++ /// <summary>+ /// The engine to use. With a specific <paramref name="language"/>, only an engine whose+ /// <see cref="IOcrEngine.SupportedLanguages"/> actually covers it is returned — never an engine that would+ /// silently misread a script it doesn't know (e.g. Apple Vision guessing at Devanagari), even if that engine is+ /// otherwise the platform's native/preferred one. With no language ("Automatic"), the platform's native engine+ /// (Apple Vision on macOS: zero install, GPU/NPU accelerated) is preferred, since the document's script isn't+ /// known ahead of time; picking a non-Latin-only language from the preset editor is what routes to Tesseract.+ /// </summary>+ public static IOcrEngine? GetDefault(string? language = null)+ {+ lock (Sync)+ {+ DiscoverNativeEngine();++ if (!string.IsNullOrWhiteSpace(language))+ return RegisteredEngines.FirstOrDefault(e => e.IsAvailable && Supports(e, language));++ var native = RegisteredEngines.FirstOrDefault(e => e.Id.Equals("vision", StringComparison.OrdinalIgnoreCase) && e.IsAvailable);+ return native ?? RegisteredEngines.FirstOrDefault(e => e.IsAvailable);+ }+ }++ /// <summary>+ /// The exact code to pass to <paramref name="engine"/>'s <see cref="IOcrEngine.RecognizeAsync"/> for Filee's+ /// <paramref name="language"/> code: Apple Vision needs its own full BCP-47 entry ("en" → "en-US"), engines that+ /// map Filee's codes themselves (Tesseract) get the code unchanged.+ /// </summary>+ public static string? ResolveLanguage(IOcrEngine? engine, string? language)+ {+ if (engine is null || string.IsNullOrWhiteSpace(language))+ return language;+ return engine.SupportedLanguages.FirstOrDefault(l =>+ l.Equals(language, StringComparison.OrdinalIgnoreCase)+ || l.StartsWith(language + "-", StringComparison.OrdinalIgnoreCase)) ?? language;+ }++ /// <summary>+ /// Whether any registered engine (installed or not) claims to support <paramref name="language"/> — for the+ /// preset editor's language picker, so it never offers a language no engine could ever attempt. Use+ /// <see cref="GetDefault"/> to find the engine that can actually run it right now.+ /// </summary>+ public static bool AnyEngineSupports(string language)+ {+ lock (Sync)+ {+ DiscoverNativeEngine();+ return RegisteredEngines.Any(e => Supports(e, language));+ }+ }++ private static bool Supports(IOcrEngine engine, string language) =>+ engine.SupportedLanguages.Count == 0+ || engine.SupportedLanguages.Any(l => l.Equals(language, StringComparison.OrdinalIgnoreCase)+ || l.StartsWith(language + "-", StringComparison.OrdinalIgnoreCase));++ /// <summary>Registers the platform's native engine (Apple Vision) the first time it's needed.</summary>+ private static void DiscoverNativeEngine()+ {+ if (!OperatingSystem.IsMacOS() || RegisteredEngines.Any(e => e.Id.Equals("vision", StringComparison.OrdinalIgnoreCase)))+ return;+ try+ {+ var asm = AppDomain.CurrentDomain.GetAssemblies().FirstOrDefault(a => a.GetName().Name == "Filee.Platform.MacOS");+ if (asm is null)+ {+ try+ {+ asm = System.Runtime.Loader.AssemblyLoadContext.Default.LoadFromAssemblyName(new System.Reflection.AssemblyName("Filee.Platform.MacOS"));+ }+ catch+ {+ var baseDir = AppContext.BaseDirectory;+ var path = System.IO.Path.Combine(baseDir, "Filee.Platform.MacOS.dll");+ if (System.IO.File.Exists(path))+ asm = System.Runtime.Loader.AssemblyLoadContext.Default.LoadFromAssemblyPath(path);+ }+ }++ var macEngineType = asm?.GetType("Filee.Platform.MacOS.MacOsVisionOcrEngine")+ ?? Type.GetType("Filee.Platform.MacOS.MacOsVisionOcrEngine, Filee.Platform.MacOS");+ if (macEngineType is not null && Activator.CreateInstance(macEngineType) is IOcrEngine macEngine && macEngine.IsAvailable)+ RegisteredEngines.Add(macEngine);+ }+ catch+ {+ // Ignore discovery failure+ }+ }++ public static IOcrEngine? GetById(string id)+ {+ lock (Sync)+ {+ return RegisteredEngines.FirstOrDefault(e => e.Id.Equals(id, StringComparison.OrdinalIgnoreCase));+ }+ }+}diff --git a/src/Filee.Core/Ocr/OcrResult.cs b/src/Filee.Core/Ocr/OcrResult.csnew file mode 100644index 00000000..2d60ac93--- /dev/null+++ b/src/Filee.Core/Ocr/OcrResult.cs@@ -0,0 +1,39 @@+// The result of an IOcrEngine.RecognizeAsync call: text and boxes in image pixel coordinates, grouped into words+// and lines so a caller can either read the plain text or reconstruct layout (PdfDocumentReader scales these boxes+// back onto the PDF page, one shared baseline per line rather than per word).++namespace Filee.Core.Ocr;++/// <summary>+/// A bounding box in image pixel coordinates (Left/Top to Right/Bottom).+/// </summary>+public readonly record struct OcrRect(double Left, double Top, double Right, double Bottom)+{+ public double Width => Math.Max(0, Right - Left);+ public double Height => Math.Max(0, Bottom - Top);+}++/// <summary>+/// A recognized word with its text, bounding box, and recognition confidence.+/// </summary>+public sealed record OcrWord(string Text, OcrRect Box, float Confidence = 1.0f);++/// <summary>+/// A recognized line composed of words.+/// </summary>+public sealed record OcrLine(string Text, IReadOnlyList<OcrWord> Words, OcrRect Box);++/// <summary>+/// The complete OCR result of an image or page.+/// </summary>+public sealed record OcrPageResult(+ string Text,+ IReadOnlyList<OcrLine> Lines,+ IReadOnlyList<OcrWord> Words,+ int ImageWidth,+ int ImageHeight,+ string? Language = null)+{+ public static OcrPageResult Empty(int width = 0, int height = 0) =>+ new("", [], [], width, height);+}diff --git a/src/Filee.Core/Presets/Preset.cs b/src/Filee.Core/Presets/Preset.csindex ed631510..eada26a5 100644--- a/src/Filee.Core/Presets/Preset.cs+++ b/src/Filee.Core/Presets/Preset.cs@@ -423,6 +423,16 @@ public sealed class DocumentOptions/// </summary>public bool LeaveOutPictures { get; set; }+ /// <summary>+ /// Optical Character Recognition for scanned pages and images.+ /// </summary>+ public bool Ocr { get; set; } = true;++ /// <summary>+ /// Optical Character Recognition language code (e.g. "en-US", "ko-KR"). Null uses the system default.+ /// </summary>+ public string? OcrLanguage { get; set; }+public DocumentOptions Clone() => (DocumentOptions)MemberwiseClone();}diff --git a/src/Filee.Engines/EngineRegistry.cs b/src/Filee.Engines/EngineRegistry.csindex b9c0f010..facebb67 100644--- a/src/Filee.Engines/EngineRegistry.cs+++ b/src/Filee.Engines/EngineRegistry.cs@@ -1,6 +1,7 @@// THE place where conversion engines are registered. Adding an engine = one line in CreateAll.using Filee.Core.Conversion;+using Filee.Core.Ocr;using Filee.Engines.Archives;using Filee.Engines.Cad;using Filee.Engines.Data;@@ -15,6 +16,7 @@ using Filee.Engines.Infrastructure;using Filee.Engines.Magick;using Filee.Engines.Media;using Filee.Engines.Models3D;+using Filee.Engines.Ocr;using Filee.Engines.Office;using Filee.Engines.Pdf;using Filee.Engines.Subtitles;@@ -37,38 +39,44 @@ public static class EngineRegistry/// The app's log; FFmpeg writes which hardware encoders work and when one failed, LibreOffice when a file needed/// its container repacked./// </param>- public static IReadOnlyList<IConverter> CreateAll(EngineEnvironment env, ILoggerFactory? loggers = null) =>- [- new MagickImageConverter(),- new PdfSharpConverter(),- new PdfiumConverter(),- new RhwpConverter(),- new UnhwpConverter(),- new MarkdownConverter(),- new SpreadsheetConverter(),- new OoxmlConverter(),- new HwpxConverter(),- new ArchiveConverter(),- new VectorConverter(),- new IcnsConverter(),- new FontConverter(),- new DocxWriterConverter(),- new PdfTextConverter(),- new PdfSlidesConverter(),- new EmlConverter(),- new DataConverter(),- new SubtitleConverter(),- new WicImageConverter(),- new FfmpegConverter(loggers?.CreateLogger<FfmpegConverter>()),- new CadConverter(),- new ModelConverter(),- new GeoConverter(),- new EbookConverter(),- new PandocConverter(),- new GhostscriptConverter(),- new CalibreConverter(),- new LibreOfficeConverter(env, loggers?.CreateLogger<LibreOfficeConverter>()),- ];+ public static IReadOnlyList<IConverter> CreateAll(EngineEnvironment env, ILoggerFactory? loggers = null)+ {+ OcrEngineRegistry.Register(new TesseractOcrEngine());++ return+ [+ new MagickImageConverter(),+ new PdfSharpConverter(),+ new PdfiumConverter(),+ new RhwpConverter(),+ new UnhwpConverter(),+ new MarkdownConverter(),+ new SpreadsheetConverter(),+ new OoxmlConverter(),+ new HwpxConverter(),+ new ArchiveConverter(),+ new VectorConverter(),+ new IcnsConverter(),+ new FontConverter(),+ new DocxWriterConverter(),+ new ImageOcrConverter(),+ new PdfTextConverter(),+ new PdfSlidesConverter(),+ new EmlConverter(),+ new DataConverter(),+ new SubtitleConverter(),+ new WicImageConverter(),+ new FfmpegConverter(loggers?.CreateLogger<FfmpegConverter>()),+ new CadConverter(),+ new ModelConverter(),+ new GeoConverter(),+ new EbookConverter(),+ new PandocConverter(),+ new GhostscriptConverter(),+ new CalibreConverter(),+ new LibreOfficeConverter(env, loggers?.CreateLogger<LibreOfficeConverter>()),+ ];+ }/// <summary>Finds the PDF merger among the converters.</summary>public static IPdfMerger? FindPdfMerger(IEnumerable<IConverter> converters) =>diff --git a/src/Filee.Engines/Filee.Engines.csproj b/src/Filee.Engines/Filee.Engines.csprojindex ca144acd..1c3d7420 100644--- a/src/Filee.Engines/Filee.Engines.csproj+++ b/src/Filee.Engines/Filee.Engines.csproj@@ -39,6 +39,7 @@<InternalsVisibleTo Include="Filee" /><InternalsVisibleTo Include="Filee.Engines.Tests" /><InternalsVisibleTo Include="Filee.App.Tests" />+ <InternalsVisibleTo Include="Filee.Platform.Tests" /></ItemGroup><ItemGroup>diff --git a/src/Filee.Engines/Hwp/Hwpx/DocumentReaders.cs b/src/Filee.Engines/Hwp/Hwpx/DocumentReaders.csindex b98e78d0..2920b655 100644--- a/src/Filee.Engines/Hwp/Hwpx/DocumentReaders.cs+++ b/src/Filee.Engines/Hwp/Hwpx/DocumentReaders.cs@@ -4,6 +4,7 @@using System.Text;using System.Text.Json.Nodes;+using Filee.Core.Presets;using Filee.Engines.Ebooks;using Filee.Engines.Email;using Filee.Engines.Hwp.Hwpx.Docx;@@ -105,9 +106,15 @@ internal static class DocumentReaders/// <summary>Reads <paramref name="path"/> (of format <paramref name="format"/>) into the document model.</summary>/// <param name="workDirectory">Scratch folder of the job: extracted pictures go into a new folder inside it.</param>- public static async Task<HDocument> ReadAsync(string path, string format, string workDirectory, CancellationToken cancellationToken)+ public static async Task<HDocument> ReadAsync(string path, string format, string workDirectory, CancellationToken cancellationToken, Preset? preset = null){var media = Path.Combine(workDirectory, $"media-{Guid.NewGuid():N}");+ if (format.Equals("pdf", StringComparison.OrdinalIgnoreCase))+ {+ var enableOcr = preset?.Document.Ocr ?? true;+ var ocrLang = preset?.Document.OcrLanguage;+ return await Task.Run(() => PdfDocumentReader.Read(path, media, null, ocrLang, enableOcr), cancellationToken);+ }if (BuiltIn.TryGetValue(format, out var reader))return await Task.Run(() => reader(path, media), cancellationToken);if (ThroughPandoc.TryGetValue(format, out var pandocReader))diff --git a/src/Filee.Engines/Hwp/Hwpx/HwpxConverter.cs b/src/Filee.Engines/Hwp/Hwpx/HwpxConverter.csindex bf6f570e..fdaa1d19 100644--- a/src/Filee.Engines/Hwp/Hwpx/HwpxConverter.cs+++ b/src/Filee.Engines/Hwp/Hwpx/HwpxConverter.cs@@ -45,7 +45,7 @@ public sealed class HwpxConverter : IConverterpublic async Task<IReadOnlyList<string>> ConvertAsync(ConversionStep step, IProgress<double>? progress, CancellationToken cancellationToken){progress?.Report(0.1);- var document = await DocumentReaders.ReadAsync(step.InputPath, step.From, step.WorkDirectory, cancellationToken);+ var document = await DocumentReaders.ReadAsync(step.InputPath, step.From, step.WorkDirectory, cancellationToken, step.Preset);progress?.Report(0.6);var output = step.Output.Allocate("hwpx");diff --git a/src/Filee.Engines/Infrastructure/EngineDownloads.cs b/src/Filee.Engines/Infrastructure/EngineDownloads.csindex 7ae3e302..4a150684 100644--- a/src/Filee.Engines/Infrastructure/EngineDownloads.cs+++ b/src/Filee.Engines/Infrastructure/EngineDownloads.cs@@ -44,12 +44,17 @@ public static class EngineDownloadsnew("ghostscript", OperatingSystem.IsWindows() ? ["ghostscript", "vcruntime"] : ["ghostscript"], 31_000_000, ["ghostscript"]),new("ffmpeg", ["ffmpeg"], 272_000_000, ["ffmpeg"]),new("calibre", ["calibre"], 663_000_000, ["calibre"]),+ // No Filee-hosted download on any platform yet (Tesseract's Linux/macOS builds are dynamically linked+ // against leptonica/libpng/etc., not xcopy-deployable the way FFmpeg's static builds are): install it with+ // the OS's own package manager or installer, then "Use my copy…", same as Ghostscript on macOS and Linux.+ new("tesseract", [], 30_000_000, ["image-ocr"]),];/// <summary>Total download size of a package in bytes.</summary>public static long DownloadSize(EnginePackage package) => package.Components.Sum(c => Components.GetValueOrDefault(c)?.Size ?? 0);- public static bool CanDownload(EnginePackage package) => package.Components.Count > 0 && package.Components.All(Components.ContainsKey);+ public static bool CanDownload(EnginePackage package) =>+ package.Components.Count > 0 && package.Components.All(Components.ContainsKey);public static string? UnavailableReason(EnginePackage package){@@ -61,6 +66,7 @@ public static class EngineDownloads"ffmpeg" => "Install FFmpeg with Homebrew (brew install ffmpeg), then check again or select its bin folder.","ghostscript" => "Install Ghostscript with Homebrew (brew install ghostscript), then check again or select its bin folder.","calibre" => "The calibre Filee downloads needs macOS 14 or later. Install calibre 7.26 from calibre-ebook.com and select its .app folder.",+ "tesseract" => "Install Tesseract with Homebrew (brew install tesseract tesseract-lang for every language), then check again or select its bin folder. Apple Vision already recognizes English, CJK and a few other languages without it.",_ => $"No Filee download is available for {RuntimeIdentifier}. Install {package.Id} for this Mac and select its program or .app folder.",};if (OperatingSystem.IsLinux())@@ -68,8 +74,11 @@ public static class EngineDownloads{"libreoffice" => "Install LibreOffice using your distribution's package manager (Ubuntu/Debian: sudo apt install libreoffice), then check again. HWP import through LibreOffice also needs Java and the H2Orestart extension.","ghostscript" => "Install Ghostscript using your distribution's package manager (Ubuntu/Debian: sudo apt install ghostscript), then check again.",+ "tesseract" => "Install Tesseract using your distribution's package manager (Ubuntu/Debian: sudo apt install tesseract-ocr, plus tesseract-ocr-<lang> for each language, e.g. tesseract-ocr-nep), then check again. Linux has no built-in OCR, so this is required for OCR to work at all.",_ => $"No Filee download is available for {RuntimeIdentifier}. Install {package.Id} using your distribution's package manager and select its program folder.",};+ if (package.Id == "tesseract")+ return "Install Tesseract from https://github.com/UB-Mannheim/tesseract (ticking the languages you need), then check again or select its install folder.";return $"No Filee download is available for {RuntimeIdentifier}. Select a compatible installed copy.";}diff --git a/src/Filee.Engines/Infrastructure/EngineEnvironment.cs b/src/Filee.Engines/Infrastructure/EngineEnvironment.csindex a94979e6..12a9da9e 100644--- a/src/Filee.Engines/Infrastructure/EngineEnvironment.cs+++ b/src/Filee.Engines/Infrastructure/EngineEnvironment.cs@@ -84,6 +84,7 @@ public sealed class EngineEnvironment(string dataDirectory)["calibre"] = [windows ? "ebook-convert.exe" : "ebook-convert"],["rhwp"] = [windows ? "rhwp.exe" : "rhwp"],["7zip"] = [windows ? "7z.exe" : "7zz"],+ ["tesseract"] = [windows ? "tesseract.exe" : "tesseract"],};public static string ProgramName(string name) => OperatingSystem.IsWindows() ? name@@ -217,6 +218,8 @@ public sealed class EngineEnvironment(string dataDirectory)}if (id == "calibre")folders.AddRange(["/Applications/calibre.app", Path.Combine(Environment.GetFolderPath(Environment.SpecialFolder.UserProfile), "Applications", "calibre.app"), "/opt/calibre"]);+ if (id == "tesseract")+ folders.AddRange(["/opt/homebrew/opt/tesseract/bin", "/usr/local/opt/tesseract/bin"]);}foreach (var root in (OperatingSystem.IsWindows()? new[] { Environment.SpecialFolder.ProgramFiles, Environment.SpecialFolder.ProgramFilesX86 }@@ -231,6 +234,10 @@ public sealed class EngineEnvironment(string dataDirectory)case "calibre":folders.Add(Path.Combine(root, "Calibre2"));break;+ case "tesseract":+ // The official UB-Mannheim installer's default location.+ folders.Add(Path.Combine(root, "Tesseract-OCR"));+ break;case "ghostscript":// C:\Program Files\gs\gs10.04.0\bin, newest firsttrydiff --git a/src/Filee.Engines/Infrastructure/EngineInstaller.cs b/src/Filee.Engines/Infrastructure/EngineInstaller.csindex 12fe5fac..ec2cdc70 100644--- a/src/Filee.Engines/Infrastructure/EngineInstaller.cs+++ b/src/Filee.Engines/Infrastructure/EngineInstaller.cs@@ -53,11 +53,11 @@ public sealed class EngineInstaller}/// <summary>True when every component is installed (any version).</summary>- public bool IsInstalled(EnginePackage package) => package.Components.All(c => _components.ContainsKey(c) && Marker(c) is not null);+ public bool IsInstalled(EnginePackage package) => package.Components.Count > 0 && package.Components.All(c => _components.ContainsKey(c) && Marker(c) is not null);/// <summary>True when every component is installed in the version engines.json pins.</summary>public bool IsUpToDate(EnginePackage package) =>- package.Components.All(c => _components.TryGetValue(c, out var component)+ package.Components.Count > 0 && package.Components.All(c => _components.TryGetValue(c, out var component)&& string.Equals(Marker(c), component.Sha256, StringComparison.OrdinalIgnoreCase));/// <summary>Downloads and unpacks the components that are missing or outdated.</summary>diff --git a/src/Filee.Engines/Ocr/ImageOcrConverter.cs b/src/Filee.Engines/Ocr/ImageOcrConverter.csnew file mode 100644index 00000000..64e2d217--- /dev/null+++ b/src/Filee.Engines/Ocr/ImageOcrConverter.cs@@ -0,0 +1,96 @@+// Image → TXT, DOCX and Markdown via whichever OCR engine OcrEngineRegistry resolves (Apple Vision on macOS,+// Tesseract everywhere else). Allows dropping scans, photos or screenshots directly and extracting editable text.++using System.Text;+using Filee.Core.Conversion;+using Filee.Core.Ocr;+using Filee.Engines.Hwp.Hwpx;+using Filee.Engines.Hwp.Hwpx.Docx;+using Filee.Engines.Infrastructure;++namespace Filee.Engines.Ocr;++/// <summary>+/// Converts raster images directly to text documents using Optical Character Recognition.+/// </summary>+public sealed class ImageOcrConverter : IConverter+{+ private static readonly string[] ImageFormats = ["png", "jpg", "jpeg", "webp", "tiff", "bmp"];++ public string Id => "image-ocr";+ public string DisplayName => "Image OCR (built-in)";+ public int MaxParallelism => 0;++ public IReadOnlyList<ConversionEdge> Edges { get; } =+ ImageFormats.SelectMany(from => new ConversionEdge[]+ {+ new(from, "txt", 15, DirectOnly: true),+ new(from, "docx", 15, DirectOnly: true),+ new(from, "md", 15, DirectOnly: true),+ }).ToList();++ public EngineStatus GetStatus()+ {+ var engine = OcrEngineRegistry.GetDefault();+ if (engine is { IsAvailable: true })+ return EngineStatus.Available($"Image → TXT, DOCX, Markdown ({engine.DisplayName})", EngineVersions.BuiltIn);++ return EngineStatus.Unavailable("engine.reason.not_installed", "No supported OCR engine found on this system.");+ }++ public async Task<IReadOnlyList<string>> ConvertAsync(ConversionStep step, IProgress<double>? progress, CancellationToken cancellationToken)+ {+ if (!step.Preset.Document.Ocr)+ throw new InvalidOperationException("OCR is disabled in preset settings.");++ var language = step.Preset.Document.OcrLanguage;+ var engine = OcrEngineRegistry.GetDefault(language);+ if (engine is null || !engine.IsAvailable)+ throw new InvalidOperationException(string.IsNullOrWhiteSpace(language)+ ? "No OCR engine is installed. Install Tesseract in Settings → Engines."+ : $"No installed OCR engine recognizes '{language}'. Install Tesseract with that language's trained data in Settings → Engines.");++ progress?.Report(0.1);+ var result = await engine.RecognizeAsync(step.InputPath, OcrEngineRegistry.ResolveLanguage(engine, language), cancellationToken);+ cancellationToken.ThrowIfCancellationRequested();+ progress?.Report(0.7);++ if (string.IsNullOrWhiteSpace(result.Text))+ throw new InvalidOperationException("No text could be recognized in the image.");++ var output = step.Output.Allocate(step.To);+ if (output is null)+ return [];++ switch (step.To)+ {+ case "txt":+ await File.WriteAllTextAsync(output, result.Text, new UTF8Encoding(encoderShouldEmitUTF8Identifier: true), cancellationToken);+ break;++ case "md":+ var mdContent = string.Join("\n\n", result.Lines.Select(l => l.Text));+ await File.WriteAllTextAsync(output, mdContent, new UTF8Encoding(encoderShouldEmitUTF8Identifier: false), cancellationToken);+ break;++ case "docx":+ var doc = new HDocument();+ var section = new HSection();+ foreach (var line in result.Lines)+ {+ var p = new HParagraph();+ p.Inlines.Add(new HText(line.Text, default));+ section.Blocks.Add(p);+ }+ doc.Sections.Add(section);+ await Task.Run(() => DocxWriter.Write(doc, output), cancellationToken);+ break;++ default:+ throw new NotSupportedException($"Target format {step.To} is not supported by Image OCR.");+ }++ progress?.Report(1.0);+ return [output];+ }+}diff --git a/src/Filee.Engines/Ocr/TesseractOcrEngine.cs b/src/Filee.Engines/Ocr/TesseractOcrEngine.csnew file mode 100644index 00000000..2baea8ce--- /dev/null+++ b/src/Filee.Engines/Ocr/TesseractOcrEngine.cs@@ -0,0 +1,243 @@+// Tesseract OCR engine: a single self-contained binary plus pinned .traineddata files, found the same way as every+// other optional engine (OwnCopies, Filee's own copy, then one already on the PC). The real cross-platform, no-Python+// fallback for platforms without a native OCR API (Windows, Linux) and for scripts Apple Vision doesn't recognize.++using System.Globalization;+using Filee.Core.Ocr;+using Filee.Engines.Infrastructure;++namespace Filee.Engines.Ocr;++/// <summary>+/// Optical Character Recognition engine powered by Tesseract (Apache-2.0), run as a subprocess. Covers every+/// platform and, with the matching .traineddata installed, scripts Apple Vision does not (Devanagari, Cyrillic+/// beyond Russian/Ukrainian, Thai, Greek, Arabic script, ...).+/// </summary>+public sealed class TesseractOcrEngine : IOcrEngine+{+ private static readonly TimeSpan Timeout = TimeSpan.FromMinutes(2);++ public string Id => "tesseract";+ public string DisplayName => "Tesseract OCR";++ public bool IsAvailable => LocateExecutable() is not null;++ /// <summary>+ /// Filee's language codes this engine can be asked to recognize, limited to what Tesseract's own trained-data+ /// set actually ships (tessdata_fast/best): asking for a code outside this list always fails, so the UI and+ /// <see cref="OcrEngineRegistry"/> only offer what can really work, instead of PaddleOCR's old claimed-but-never+ /// bundled 100+ languages.+ /// </summary>+ public IReadOnlyList<string> SupportedLanguages => [.. LanguageMap.Keys];++ /// <summary>Filee language code (the OCR language picker, BCP-47-ish) → Tesseract's own 3-letter code.</summary>+ private static readonly IReadOnlyDictionary<string, string> LanguageMap = new Dictionary<string, string>(StringComparer.OrdinalIgnoreCase)+ {+ ["en"] = "eng",+ ["ne"] = "nep",+ ["hi"] = "hin",+ ["zh-Hans"] = "chi_sim",+ ["zh-Hant"] = "chi_tra",+ ["ja"] = "jpn",+ ["ko"] = "kor",+ ["es"] = "spa",+ ["fr"] = "fra",+ ["de"] = "deu",+ ["it"] = "ita",+ ["pt"] = "por",+ ["ru"] = "rus",+ ["ar"] = "ara",+ ["th"] = "tha",+ ["el"] = "ell",+ ["uk"] = "ukr",+ ["tr"] = "tur",+ ["vi"] = "vie",+ ["id"] = "ind",+ ["pl"] = "pol",+ ["nl"] = "nld",+ ["sv"] = "swe",+ ["da"] = "dan",+ ["fi"] = "fin",+ ["no"] = "nor",+ ["cs"] = "ces",+ ["sk"] = "slk",+ ["hu"] = "hun",+ ["ro"] = "ron",+ ["bg"] = "bul",+ ["hr"] = "hrv",+ ["sr-Cyrl"] = "srp",+ ["sr-Latn"] = "srp_latn",+ ["sl"] = "slv",+ ["et"] = "est",+ ["lv"] = "lav",+ ["lt"] = "lit",+ ["be"] = "bel",+ ["sa"] = "san",+ ["mr"] = "mar",+ ["ta"] = "tam",+ ["te"] = "tel",+ ["ur"] = "urd",+ ["fa"] = "fas",+ ["ps"] = "pus",+ ["sd"] = "snd",+ ["ug"] = "uig",+ ["af"] = "afr",+ ["az"] = "aze",+ ["bs"] = "bos",+ ["ca"] = "cat",+ ["cy"] = "cym",+ ["eu"] = "eus",+ ["ga"] = "gle",+ ["gl"] = "glg",+ ["is"] = "isl",+ ["ka"] = "kat",+ ["ku"] = "kmr",+ ["la"] = "lat",+ ["lb"] = "ltz",+ ["mi"] = "mri",+ ["mk"] = "mkd",+ ["mn"] = "mon",+ ["ms"] = "msa",+ ["mt"] = "mlt",+ ["oc"] = "oci",+ ["qu"] = "que",+ ["sq"] = "sqi",+ ["sw"] = "swa",+ ["tl"] = "fil",+ ["uz"] = "uzb",+ ["kk"] = "kaz",+ ["ky"] = "kir",+ ["tg"] = "tgk",+ ["tt"] = "tat",+ };++ /// <summary>Tesseract's own code for a Filee language code, or null when Tesseract has no matching trained data.</summary>+ public static string? MapLanguage(string? language)+ {+ if (string.IsNullOrWhiteSpace(language))+ return "eng";+ return LanguageMap.TryGetValue(language.Trim(), out var code) ? code : null;+ }++ public static string? LocateExecutable() =>+ EngineEnvironment.OwnProgram("tesseract", "tesseract.exe")+ ?? EngineEnvironment.BundledProgram("tesseract", "tesseract.exe")+ ?? EngineEnvironment.SystemProgram("tesseract", "tesseract.exe");++ /// <summary>+ /// The tessdata folder next to <paramref name="executable"/>: Homebrew, apt and the official Windows installer+ /// all keep trained-data files at one of these relative locations. Null lets tesseract fall back to its own+ /// compiled-in default (TESSDATA_PREFIX or the path it was built with), same as running it from a shell.+ /// </summary>+ public static string? LocateTessdataDir(string executable)+ {+ var bin = Path.GetDirectoryName(executable);+ if (bin is null)+ return null;+ var root = Path.GetFullPath(Path.Combine(bin, ".."));+ string[] candidates =+ [+ Path.Combine(bin, "tessdata"),+ Path.Combine(root, "share", "tessdata"),+ Path.Combine(root, "share", "tesseract-ocr", "5", "tessdata"),+ Path.Combine(root, "share", "tesseract-ocr", "4.00", "tessdata"),+ "/usr/share/tesseract-ocr/5/tessdata",+ "/usr/share/tesseract-ocr/4.00/tessdata",+ "/usr/share/tessdata",+ ];+ return candidates.FirstOrDefault(c => Directory.Exists(c) && Directory.EnumerateFiles(c, "*.traineddata").Any());+ }++ /// <summary>Whether <paramref name="language"/> has its .traineddata installed, not just a known mapping.</summary>+ public bool HasLanguageData(string? language)+ {+ var exe = LocateExecutable();+ var code = MapLanguage(language);+ if (exe is null || code is null)+ return false;+ var dir = LocateTessdataDir(exe);+ if (dir is null)+ return true; // Can't check; let tesseract itself report a missing language if it turns out to be wrong.+ return File.Exists(Path.Combine(dir, $"{code}.traineddata"));+ }++ public async Task<OcrPageResult> RecognizeAsync(string imagePath, string? language = null, CancellationToken cancellationToken = default)+ {+ if (!File.Exists(imagePath))+ throw new FileNotFoundException("Image file not found for OCR.", imagePath);++ var exe = LocateExecutable() ?? throw new InvalidOperationException("Tesseract is not installed.");+ var code = MapLanguage(language)+ ?? throw new InvalidOperationException($"Tesseract has no trained data for language '{language}'.");++ var args = new List<string> { imagePath, "stdout", "-l", code };+ if (LocateTessdataDir(exe) is { } tessdata)+ args.AddRange(["--tessdata-dir", tessdata]);+ args.Add("tsv");++ var result = await ProcessRunner.RunAsync(exe, [.. args], Timeout, cancellationToken);+ if (result.ExitCode != 0)+ {+ throw new InvalidOperationException(result.StandardError.Contains("Failed loading language", StringComparison.OrdinalIgnoreCase)+ ? $"Tesseract has no trained data installed for '{code}'. Install the {language} language pack and try again."+ : $"Tesseract could not read the image. {result.StandardError.Trim()}");+ }++ return ParseTsv(result.StandardOutput, code);+ }++ private static OcrPageResult ParseTsv(string tsv, string language)+ {+ var words = new List<OcrWord>();+ var lineGroups = new List<(string Key, List<OcrWord> Words)>();+ var imgWidth = 0;+ var imgHeight = 0;++ foreach (var row in tsv.Split('\n', StringSplitOptions.RemoveEmptyEntries))+ {+ var cols = row.Split('\t');+ // level, page_num, block_num, par_num, line_num, word_num, left, top, width, height, conf, text+ if (cols.Length < 12)+ continue;+ if (cols[0] == "1")+ {+ // Page-level row: left/top/width/height is the whole image, needed to scale word boxes back+ // to the source page (PdfDocumentReader) even when the recognized text doesn't fill the image.+ int.TryParse(cols[8], out imgWidth);+ int.TryParse(cols[9], out imgHeight);+ continue;+ }+ if (cols[0] != "5")+ continue;+ var text = cols[11].Trim();+ if (text.Length == 0)+ continue;+ if (!double.TryParse(cols[6], NumberStyles.Float, CultureInfo.InvariantCulture, out var left)+ || !double.TryParse(cols[7], NumberStyles.Float, CultureInfo.InvariantCulture, out var top)+ || !double.TryParse(cols[8], NumberStyles.Float, CultureInfo.InvariantCulture, out var width)+ || !double.TryParse(cols[9], NumberStyles.Float, CultureInfo.InvariantCulture, out var height))+ continue;+ float.TryParse(cols[10], NumberStyles.Float, CultureInfo.InvariantCulture, out var conf);++ var word = new OcrWord(text, new OcrRect(left, top, left + width, top + height), Math.Max(0, conf) / 100f);+ words.Add(word);++ var lineKey = string.Join('.', cols[1], cols[2], cols[3], cols[4]);+ var group = lineGroups.Count > 0 && lineGroups[^1].Key == lineKey ? lineGroups[^1] : (lineKey, []);+ if (lineGroups.Count == 0 || lineGroups[^1].Key != lineKey)+ lineGroups.Add(group);+ group.Words.Add(word);+ }++ var lines = lineGroups.Select(g =>+ {+ var text = string.Join(' ', g.Words.Select(w => w.Text));+ var box = new OcrRect(+ g.Words.Min(w => w.Box.Left), g.Words.Min(w => w.Box.Top),+ g.Words.Max(w => w.Box.Right), g.Words.Max(w => w.Box.Bottom));+ return new OcrLine(text, g.Words, box);+ }).ToList();++ return new OcrPageResult(string.Join('\n', lines.Select(l => l.Text)), lines, words, imgWidth, imgHeight, language);+ }+}diff --git a/src/Filee.Engines/Office/DocxWriterConverter.cs b/src/Filee.Engines/Office/DocxWriterConverter.csindex 0b2deb23..9fc095ed 100644--- a/src/Filee.Engines/Office/DocxWriterConverter.cs+++ b/src/Filee.Engines/Office/DocxWriterConverter.cs@@ -32,7 +32,7 @@ public sealed class DocxWriterConverter : IConverterpublic async Task<IReadOnlyList<string>> ConvertAsync(ConversionStep step, IProgress<double>? progress, CancellationToken cancellationToken){progress?.Report(0.1);- var document = await DocumentReaders.ReadAsync(step.InputPath, step.From, step.WorkDirectory, cancellationToken);+ var document = await DocumentReaders.ReadAsync(step.InputPath, step.From, step.WorkDirectory, cancellationToken, step.Preset);progress?.Report(0.6);var output = step.Output.Allocate("docx");diff --git a/src/Filee.Engines/Pdf/PdfDocumentReader.cs b/src/Filee.Engines/Pdf/PdfDocumentReader.csindex cb4a9058..bd67cde4 100644--- a/src/Filee.Engines/Pdf/PdfDocumentReader.cs+++ b/src/Filee.Engines/Pdf/PdfDocumentReader.cs@@ -17,6 +17,7 @@using System.Text;using System.Text.RegularExpressions;+using Filee.Core.Ocr;using Filee.Engines.Hwp.Hwpx;using Filee.Engines.Infrastructure;using PDFtoImage;@@ -102,23 +103,37 @@ internal sealed partial class PdfDocumentReaderprivate readonly string _path;private readonly string _mediaFolder;private readonly bool _textOnly;+ private readonly IOcrEngine? _ocrEngine;+ private readonly string? _ocrLanguage;+ private readonly bool _enableOcr;private byte[]? _pdfBytes;private int _pictures;- private PdfDocumentReader(string path, string mediaFolder, bool textOnly)+ private PdfDocumentReader(string path, string mediaFolder, bool textOnly, IOcrEngine? ocrEngine = null, string? ocrLanguage = null, bool enableOcr = true){_path = path;_mediaFolder = mediaFolder;_textOnly = textOnly;+ _ocrEngine = ocrEngine;+ _ocrLanguage = ocrLanguage;+ _enableOcr = enableOcr;}/// <summary>Reads <paramref name="path"/> with pictures (extracted into <paramref name="mediaFolder"/>) and scans.</summary>- public static HDocument Read(string path, string mediaFolder) => new PdfDocumentReader(path, mediaFolder, textOnly: false).ReadDocument();+ public static HDocument Read(string path, string mediaFolder) =>+ new PdfDocumentReader(path, mediaFolder, textOnly: false).ReadDocument();++ public static HDocument Read(string path, string mediaFolder, IOcrEngine? ocrEngine, string? ocrLanguage = null, bool enableOcr = true) =>+ new PdfDocumentReader(path, mediaFolder, textOnly: false, ocrEngine, ocrLanguage, enableOcr).ReadDocument();/// <summary>/// Reads only the text, including invisible text (the OCR layer of scanned PDFs); pictures and scans are left out./// </summary>- public static HDocument ReadText(string path) => new PdfDocumentReader(path, "", textOnly: true).ReadDocument();+ public static HDocument ReadText(string path) =>+ new PdfDocumentReader(path, "", textOnly: true).ReadDocument();++ public static HDocument ReadText(string path, IOcrEngine? ocrEngine, string? ocrLanguage = null, bool enableOcr = true) =>+ new PdfDocumentReader(path, "", textOnly: true, ocrEngine, ocrLanguage, enableOcr).ReadDocument();// ───────────────────────── Model of a page ─────────────────────────@@ -237,7 +252,7 @@ internal sealed partial class PdfDocumentReadernew("The PDF is password-protected. Open it with its password and save a copy without one, then convert that copy.");/// <summary>- /// The text of each page in <paramref name="pages"/> (zero-based), read like <see cref="ReadText"/> (reading order,+ /// The text of each page in <paramref name="pages"/> (zero-based), read like <see cref="ReadText(string)"/> (reading order,/// invisible OCR text included) but page by page, with running heads and page numbers kept: "" for a page without/// text. Paragraphs are separated by an empty line, lines end with CR LF./// </summary>@@ -308,13 +323,83 @@ internal sealed partial class PdfDocumentReaderlead.StartBaseLine.Y - crop.Bottom, Math.Max(1, size), Format(lead, size), link));}+ if (words.Count == 0)+ {+ // First check if the page already had an invisible OCR layer that was skipped for formatting+ foreach (var word in pageWords)+ {+ if (string.IsNullOrWhiteSpace(word.Text))+ continue;+ var letters = word.Letters.Where(l => !string.IsNullOrWhiteSpace(l.Value)).ToList();+ if (letters.Count == 0)+ continue;+ var lead = string.IsNullOrWhiteSpace(word.Lead.Value) ? letters[0] : word.Lead;+ var box = Shift(word.Box, crop);+ var size = Median(letters.Select(LetterSize));+ var center = new PdfPoint((box.Left + box.Right) / 2, (box.Top + box.Bottom) / 2);+ var link = links.FirstOrDefault(l => Inside(l.Bounds, center)).Uri;+ words.Add(new TextWord(CleanText(word.Text), box.Left, box.Right, box.Top, box.Bottom,+ lead.StartBaseLine.Y - crop.Bottom, Math.Max(1, size), Format(lead, size), link));+ }+ }+if (words.Count == 0 || (rotated && !_textOnly)){- // A scan, a page of drawings, or a turned page: the rendered page keeps everything.- content.Scanned = true;- if (!_textOnly)- content.ScanPath = RenderScan(page.Number - 1);- return content;+ if (_enableOcr)+ {+ var ocr = _ocrEngine ?? OcrEngineRegistry.GetDefault(_ocrLanguage);+ if (ocr is not null && ocr.IsAvailable)+ {+ var scanPath = RenderScan(page.Number - 1);+ try+ {+ var ocrResult = ocr.RecognizeAsync(scanPath, OcrEngineRegistry.ResolveLanguage(ocr, _ocrLanguage)).GetAwaiter().GetResult();+ if (ocrResult.Words.Count > 0)+ {+ var scaleX = content.Width / (double)ocrResult.ImageWidth;+ var scaleY = content.Height / (double)ocrResult.ImageHeight;+ // One baseline/top/bottom per OCR-detected line, not per word: on a skewed or photographed+ // scan each word's own pixel box drifts slightly in Y, which (once scaled to PDF points)+ // is enough for Lines() below to split a single real text line into several interleaved+ // reconstructed lines. Tesseract already groups words into lines; use that box.+ foreach (var line in ocrResult.Lines)+ {+ var pdfTop = content.Height - (line.Box.Top * scaleY);+ var pdfBottom = content.Height - (line.Box.Bottom * scaleY);+ var size = Math.Max(4, pdfTop - pdfBottom);+ var format = new HCharFormat(Size: (int)Math.Round(size * 100));+ foreach (var w in line.Words)+ {+ var pdfLeft = w.Box.Left * scaleX;+ var pdfRight = w.Box.Right * scaleX;+ words.Add(new TextWord(CleanText(w.Text), pdfLeft, pdfRight, pdfTop, pdfBottom, pdfBottom, size, format, null));+ }+ }+ }+ }+ catch+ {+ // Fall back to pure scan if OCR encounters an error+ }++ if (words.Count == 0)+ {+ content.Scanned = true;+ if (!_textOnly)+ content.ScanPath = scanPath;+ return content;+ }+ }+ }++ if (words.Count == 0)+ {+ // A scan, a page of drawings, or a turned page: the rendered page keeps everything.+ content.Scanned = true;+ if (!_textOnly)+ content.ScanPath = RenderScan(page.Number - 1);+ return content;+ }}// Plain text (TXT, page texts for slides) keeps table cells as paragraphs in reading order.@@ -484,8 +569,9 @@ internal sealed partial class PdfDocumentReaderprivate string NewMediaPath(string extension){- Directory.CreateDirectory(_mediaFolder);- return Path.Combine(_mediaFolder, $"pdf-{++_pictures}.{extension}");+ var folder = string.IsNullOrEmpty(_mediaFolder) ? Path.Combine(Path.GetTempPath(), "filee-media") : _mediaFolder;+ Directory.CreateDirectory(folder);+ return Path.Combine(folder, $"pdf-{++_pictures}.{extension}");}private SKBitmap Render(int pageIndex)@@ -966,7 +1052,7 @@ internal sealed partial class PdfDocumentReaderprivate static string? NullIfBlank(string? text) => string.IsNullOrWhiteSpace(text) ? null : text.Trim();- /// <summary>Plain text of a document read by <see cref="ReadText"/>: paragraphs separated by an empty line.</summary>+ /// <summary>Plain text of a document read by <see cref="ReadText(string)"/>: paragraphs separated by an empty line.</summary>public static string PlainText(HDocument document){var sb = new StringBuilder();diff --git a/src/Filee.Engines/Pdf/PdfTextConverter.cs b/src/Filee.Engines/Pdf/PdfTextConverter.csindex 86859ffb..1588eed2 100644--- a/src/Filee.Engines/Pdf/PdfTextConverter.cs+++ b/src/Filee.Engines/Pdf/PdfTextConverter.cs@@ -40,7 +40,7 @@ public sealed class PdfTextConverter : IConverterprivate static IReadOnlyList<string> Text(ConversionStep step, IProgress<double>? progress, CancellationToken cancellationToken){progress?.Report(0.1);- var text = PdfDocumentReader.PlainText(PdfDocumentReader.ReadText(step.InputPath));+ var text = PdfDocumentReader.PlainText(PdfDocumentReader.ReadText(step.InputPath, null, step.Preset.Document.OcrLanguage, step.Preset.Document.Ocr));cancellationToken.ThrowIfCancellationRequested();if (text.Length == 0)throw NoText();@@ -59,8 +59,8 @@ public sealed class PdfTextConverter : IConverterprogress?.Report(0.1);var textOnly = step.Preset.Document.LeaveOutPictures;var document = textOnly- ? PdfDocumentReader.ReadText(step.InputPath)- : PdfDocumentReader.Read(step.InputPath, Path.Combine(step.WorkDirectory, $"media-{Guid.NewGuid():N}"));+ ? PdfDocumentReader.ReadText(step.InputPath, null, step.Preset.Document.OcrLanguage, step.Preset.Document.Ocr)+ : PdfDocumentReader.Read(step.InputPath, Path.Combine(step.WorkDirectory, $"media-{Guid.NewGuid():N}"), null, step.Preset.Document.OcrLanguage, step.Preset.Document.Ocr);cancellationToken.ThrowIfCancellationRequested();if (textOnly && PdfDocumentReader.PlainText(document).Length == 0)throw NoText();diff --git a/src/Filee.Platform.MacOS/MacOsVisionOcrEngine.cs b/src/Filee.Platform.MacOS/MacOsVisionOcrEngine.csnew file mode 100644index 00000000..780f9e86--- /dev/null+++ b/src/Filee.Platform.MacOS/MacOsVisionOcrEngine.cs@@ -0,0 +1,272 @@+// OCR through Apple's Vision framework (VNRecognizeTextRequest), called directly via the Objective-C runtime so+// Filee.Platform.MacOS stays free of an Objective-C bridging library. Zero install size and GPU/NPU accelerated;+// the preferred engine on macOS whenever its SupportedLanguages cover what was asked for.++using System.Runtime.InteropServices;+using System.Runtime.Versioning;+using System.Text;+using Filee.Core.Ocr;++namespace Filee.Platform.MacOS;++/// <summary>Optical Character Recognition engine powered by Apple's Vision framework (macOS only).</summary>+[SupportedOSPlatform("macos")]+public sealed class MacOsVisionOcrEngine : IOcrEngine+{+ private static readonly Lazy<bool> VisionAvailable = new(() =>+ {+ if (!OperatingSystem.IsMacOS())+ return false;+ try+ {+ NativeLibrary.Load("/System/Library/Frameworks/Vision.framework/Vision");+ return objc_getClass("VNRecognizeTextRequest") != 0;+ }+ catch+ {+ return false;+ }+ });++ public string Id => "vision";++ public string DisplayName => "Apple Vision";++ public bool IsAvailable => VisionAvailable.Value;++ public IReadOnlyList<string> SupportedLanguages { get; } =+ [+ "en-US", "fr-FR", "it-IT", "de-DE", "es-ES", "pt-BR", "zh-Hans", "zh-Hant", "ja-JP", "ko-KR", "ru-RU", "uk-UA"+ ];++ public Task<OcrPageResult> RecognizeAsync(string imagePath, string? language = null, CancellationToken cancellationToken = default)+ {+ if (!IsAvailable)+ return Task.FromResult(OcrPageResult.Empty());++ return Task.Run(() =>+ {+ cancellationToken.ThrowIfCancellationRequested();+ using var pool = new AutoreleasePool();++ var (width, height) = GetImageDimensions(imagePath);+ if (width <= 0 || height <= 0)+ (width, height) = (1000, 1000);++ var nsPath = CFStringCreateWithCString(0, imagePath, 0x08000100);+ if (nsPath == 0)+ return OcrPageResult.Empty(width, height);++ try+ {+ var urlClass = objc_getClass("NSURL");+ var fileUrl = Send(urlClass, sel_registerName("fileURLWithPath:"), nsPath);++ var reqClass = objc_getClass("VNRecognizeTextRequest");+ var request = Send(Send(reqClass, sel_registerName("alloc")), sel_registerName("init"));+ SendInteger(request, sel_registerName("setRecognitionLevel:"), 0); // 0 = VNRequestTextRecognitionLevelAccurate+ SendBool(request, sel_registerName("setUsesLanguageCorrection:"), true);++ if (!string.IsNullOrEmpty(language))+ {+ var langString = CFStringCreateWithCString(0, language, 0x08000100);+ if (langString != 0)+ {+ var arrayClass = objc_getClass("NSArray");+ var langsArray = Send(arrayClass, sel_registerName("arrayWithObject:"), langString);+ Send(request, sel_registerName("setRecognitionLanguages:"), langsArray);+ CFRelease(langString);+ }+ }++ var dictClass = objc_getClass("NSDictionary");+ var emptyDict = Send(dictClass, sel_registerName("dictionary"));+ var handlerClass = objc_getClass("VNImageRequestHandler");+ var handler = Send(Send(handlerClass, sel_registerName("alloc")), sel_registerName("initWithURL:options:"), fileUrl, emptyDict);++ var arrayClass2 = objc_getClass("NSArray");+ var requestsArray = Send(arrayClass2, sel_registerName("arrayWithObject:"), request);++ if (!SendPerform(handler, sel_registerName("performRequests:error:"), requestsArray, out _))+ return OcrPageResult.Empty(width, height);++ var results = Send(request, sel_registerName("results"));+ if (results == 0)+ return OcrPageResult.Empty(width, height);++ var count = (int)CFArrayGetCount(results);+ var lines = new List<OcrLine>();+ var allWords = new List<OcrWord>();+ var fullText = new StringBuilder();++ for (var i = 0; i < count; i++)+ {+ cancellationToken.ThrowIfCancellationRequested();+ var obs = CFArrayGetValueAtIndex(results, i);+ var candidates = Send(obs, sel_registerName("topCandidates:"), 1);+ if (candidates == 0 || CFArrayGetCount(candidates) == 0)+ continue;++ var cand = CFArrayGetValueAtIndex(candidates, 0);+ var nsStr = Send(cand, sel_registerName("string"));+ var utf8Ptr = Send(nsStr, sel_registerName("UTF8String"));+ var text = Marshal.PtrToStringUTF8(utf8Ptr);+ if (string.IsNullOrWhiteSpace(text))+ continue;++ var conf = SendFloat(cand, sel_registerName("confidence"));+ var normBox = SendRect(obs, sel_registerName("boundingBox"));++ // Apple Vision returns normalized coordinates with bottom-left origin:+ // normBox.X (0..1 from left), normBox.Y (0..1 from bottom),+ // normBox.Width, normBox.Height.+ // Convert to pixel coordinates with top-left origin:+ var pixelLeft = normBox.X * width;+ var pixelRight = (normBox.X + normBox.Width) * width;+ var pixelTop = (1.0 - normBox.Y - normBox.Height) * height;+ var pixelBottom = (1.0 - normBox.Y) * height;+ var lineBox = new OcrRect(pixelLeft, pixelTop, pixelRight, pixelBottom);++ // Estimate word boxes along the line+ var rawWords = text.Split(' ', StringSplitOptions.RemoveEmptyEntries);+ var wordsInLine = new List<OcrWord>();+ if (rawWords.Length == 1)+ {+ var word = new OcrWord(rawWords[0], lineBox, conf);+ wordsInLine.Add(word);+ allWords.Add(word);+ }+ else if (rawWords.Length > 1)+ {+ var totalChars = rawWords.Sum(w => w.Length);+ var currentX = pixelLeft;+ var usableWidth = Math.Max(0, lineBox.Width);+ for (var w = 0; w < rawWords.Length; w++)+ {+ var wStr = rawWords[w];+ var wWidth = totalChars > 0 ? (usableWidth * wStr.Length / (double)totalChars) : (usableWidth / rawWords.Length);+ var wBox = new OcrRect(currentX, pixelTop, currentX + wWidth, pixelBottom);+ var word = new OcrWord(wStr, wBox, conf);+ wordsInLine.Add(word);+ allWords.Add(word);+ currentX += wWidth;+ }+ }++ lines.Add(new OcrLine(text, wordsInLine, lineBox));+ fullText.AppendLine(text);+ }++ return new OcrPageResult(fullText.ToString().TrimEnd(), lines, allWords, width, height, language);+ }+ finally+ {+ CFRelease(nsPath);+ }+ }, cancellationToken);+ }++ private static (int Width, int Height) GetImageDimensions(string path)+ {+ try+ {+ var nsImageClass = objc_getClass("NSImage");+ var nsPath = CFStringCreateWithCString(0, path, 0x08000100);+ if (nsPath == 0)+ return (0, 0);+ try+ {+ var alloc = Send(nsImageClass, sel_registerName("alloc"));+ var image = Send(alloc, sel_registerName("initWithContentsOfFile:"), nsPath);+ if (image == 0)+ return (0, 0);+ var size = SendSize(image, sel_registerName("size"));+ return ((int)Math.Round(size.Width), (int)Math.Round(size.Height));+ }+ finally+ {+ CFRelease(nsPath);+ }+ }+ catch+ {+ return (0, 0);+ }+ }++ [StructLayout(LayoutKind.Sequential)]+ private struct CGRect+ {+ public double X;+ public double Y;+ public double Width;+ public double Height;+ }++ private readonly struct AutoreleasePool : IDisposable+ {+ private readonly nint _pool = Send(Send(objc_getClass("NSAutoreleasePool"), sel_registerName("alloc")), sel_registerName("init"));+ public AutoreleasePool() { }+ public void Dispose() => SendVoid(_pool, sel_registerName("drain"));+ }++ [StructLayout(LayoutKind.Sequential)]+ private struct CGSize+ {+ public double Width;+ public double Height;+ }++ private const string CoreFoundation = "/System/Library/Frameworks/CoreFoundation.framework/CoreFoundation";+ private const string ObjC = "/usr/lib/libobjc.A.dylib";++ [DllImport(CoreFoundation)]+ private static extern nint CFArrayGetCount(nint array);++ [DllImport(CoreFoundation)]+ private static extern nint CFArrayGetValueAtIndex(nint array, nint index);++ [DllImport(CoreFoundation)]+ private static extern nint CFStringCreateWithCString(nint allocator, [MarshalAs(UnmanagedType.LPUTF8Str)] string value, uint encoding);++ [DllImport(CoreFoundation)]+ private static extern void CFRelease(nint obj);++ [DllImport(ObjC)]+ private static extern nint objc_getClass([MarshalAs(UnmanagedType.LPUTF8Str)] string name);++ [DllImport(ObjC)]+ private static extern nint sel_registerName([MarshalAs(UnmanagedType.LPUTF8Str)] string name);++ [DllImport(ObjC, EntryPoint = "objc_msgSend")]+ private static extern nint Send(nint receiver, nint selector);++ [DllImport(ObjC, EntryPoint = "objc_msgSend")]+ private static extern nint Send(nint receiver, nint selector, nint argument);++ [DllImport(ObjC, EntryPoint = "objc_msgSend")]+ private static extern nint Send(nint receiver, nint selector, nint arg1, nint arg2);++ [DllImport(ObjC, EntryPoint = "objc_msgSend")]+ [return: MarshalAs(UnmanagedType.I1)]+ private static extern bool SendBool(nint receiver, nint selector, [MarshalAs(UnmanagedType.I1)] bool argument);++ [DllImport(ObjC, EntryPoint = "objc_msgSend")]+ private static extern void SendInteger(nint receiver, nint selector, nint value);++ [DllImport(ObjC, EntryPoint = "objc_msgSend")]+ [return: MarshalAs(UnmanagedType.I1)]+ private static extern bool SendPerform(nint receiver, nint selector, nint requests, out nint error);++ [DllImport(ObjC, EntryPoint = "objc_msgSend")]+ private static extern float SendFloat(nint receiver, nint selector);++ [DllImport(ObjC, EntryPoint = "objc_msgSend")]+ private static extern CGRect SendRect(nint receiver, nint selector);++ [DllImport(ObjC, EntryPoint = "objc_msgSend")]+ private static extern CGSize SendSize(nint receiver, nint selector);++ [DllImport(ObjC, EntryPoint = "objc_msgSend")]+ private static extern void SendVoid(nint receiver, nint selector);+}diff --git a/tests/Filee.App.Tests/EnginesPageTests.cs b/tests/Filee.App.Tests/EnginesPageTests.csindex f1c6edec..dcc491a4 100644--- a/tests/Filee.App.Tests/EnginesPageTests.cs+++ b/tests/Filee.App.Tests/EnginesPageTests.cs@@ -111,4 +111,17 @@ public class EnginesPageTestsvm.Dispose();}}++ [AvaloniaFact]+ public void Tesseract_package_has_install_available_when_not_installed()+ {+ TestServices.EnsureInitialized("en");+ var downloads = AppHost.Get<EngineDownloadService>();+ var tesseract = downloads.Packages.FirstOrDefault(p => p.Package.Id == "tesseract");+ Assert.NotNull(tesseract);+ if (EnginePackageState.CanSystemInstall(tesseract.Package))+ {+ Assert.True(tesseract.CanInstall || tesseract.IsInstalled);+ }+ }}diff --git a/tests/Filee.Core.Tests/OcrEngineRegistryTests.cs b/tests/Filee.Core.Tests/OcrEngineRegistryTests.csnew file mode 100644index 00000000..1a6f16e6--- /dev/null+++ b/tests/Filee.Core.Tests/OcrEngineRegistryTests.cs@@ -0,0 +1,88 @@+using Filee.Core.Ocr;+using Filee.Core.Presets;++namespace Filee.Core.Tests;++public class OcrEngineRegistryTests+{+ private sealed class MockOcrEngine(string id, string name, bool available, IReadOnlyList<string> languages) : IOcrEngine+ {+ public string Id => id;+ public string DisplayName => name;+ public bool IsAvailable => available;+ public IReadOnlyList<string> SupportedLanguages => languages;++ public Task<OcrPageResult> RecognizeAsync(string imagePath, string? language = null, CancellationToken cancellationToken = default)+ {+ var word = new OcrWord("test", new OcrRect(0, 0, 10, 10), 0.99f);+ var line = new OcrLine("test", [word], new OcrRect(0, 0, 10, 10));+ return Task.FromResult(new OcrPageResult("test", [line], [word], 100, 100, language));+ }+ }++ [Fact]+ public void OcrRect_computes_width_and_height_correctly()+ {+ var rect = new OcrRect(10, 20, 110, 70);+ Assert.Equal(100, rect.Width);+ Assert.Equal(50, rect.Height);+ }++ [Fact]+ public void OcrPageResult_Empty_has_zero_words_and_lines()+ {+ var empty = OcrPageResult.Empty(200, 300);+ Assert.Empty(empty.Text);+ Assert.Empty(empty.Lines);+ Assert.Empty(empty.Words);+ Assert.Equal(200, empty.ImageWidth);+ Assert.Equal(300, empty.ImageHeight);+ }++ [Fact]+ public void Register_and_retrieve_engine()+ {+ var mock = new MockOcrEngine("mock-ocr", "Mock OCR", true, ["en", "ne"]);+ OcrEngineRegistry.Register(mock);++ Assert.Contains(OcrEngineRegistry.All, e => e.Id == "mock-ocr");+ Assert.NotNull(OcrEngineRegistry.GetById("mock-ocr"));+ Assert.Same(mock, OcrEngineRegistry.GetById("mock-ocr"));+ }++ [Fact]+ public void DocumentOptions_defaults_and_cloning()+ {+ var options = new DocumentOptions();+ Assert.True(options.Ocr);+ Assert.Null(options.OcrLanguage);++ options.Ocr = false;+ options.OcrLanguage = "ne";++ var clone = options.Clone();+ Assert.False(clone.Ocr);+ Assert.Equal("ne", clone.OcrLanguage);+ }++ [Fact]+ public void GetDefault_routes_a_language_to_the_engine_that_actually_supports_it()+ {+ // A language no other test's mock claims, so this isn't order-dependent on the shared registry.+ var tesseract = new MockOcrEngine("tesseract", "Tesseract OCR", true, ["th", "el"]);+ OcrEngineRegistry.Register(tesseract);++ var thaiEngine = OcrEngineRegistry.GetDefault("th");+ Assert.NotNull(thaiEngine);+ Assert.Equal("tesseract", thaiEngine.Id);+ }++ [Fact]+ public void GetDefault_returns_null_when_no_available_engine_supports_the_language()+ {+ var unrelated = new MockOcrEngine("unrelated-mock", "Unrelated", true, ["fr"]);+ OcrEngineRegistry.Register(unrelated);++ Assert.Null(OcrEngineRegistry.GetDefault("a-language-nothing-claims"));+ }+}diff --git a/tests/Filee.Engines.Tests/TesseractOcrTests.cs b/tests/Filee.Engines.Tests/TesseractOcrTests.csnew file mode 100644index 00000000..b5fe537a--- /dev/null+++ b/tests/Filee.Engines.Tests/TesseractOcrTests.cs@@ -0,0 +1,74 @@+using Filee.Core.Ocr;+using Filee.Engines.Ocr;+using SkiaSharp;++namespace Filee.Engines.Tests;++public sealed class TesseractOcrTests+{+ // Pure mapping logic: runs even where Tesseract itself isn't installed (CI).+ [Theory]+ [InlineData("en", "eng")]+ [InlineData("ne", "nep")]+ [InlineData("zh-Hans", "chi_sim")]+ [InlineData("zh-Hant", "chi_tra")]+ [InlineData(null, "eng")]+ public void MapLanguage_maps_known_codes(string? filee, string tesseract) =>+ Assert.Equal(tesseract, TesseractOcrEngine.MapLanguage(filee));++ [Fact]+ public void MapLanguage_returns_null_for_a_code_with_no_trained_data()+ {+ // Unlike the old PaddleOCR engine, which silently fell back to English for any code it didn't+ // recognize, an unmapped code must be reported as unsupported rather than guessed at.+ Assert.Null(TesseractOcrEngine.MapLanguage("xal"));+ }++ [Fact]+ public async Task RecognizesTextInAGeneratedImage()+ {+ var engine = new TesseractOcrEngine();+ Assert.SkipUnless(engine.IsAvailable, "Tesseract is not installed on this machine.");++ var imagePath = Path.Combine(Path.GetTempPath(), $"filee-tesseract-test-{Guid.NewGuid():N}.png");+ try+ {+ using (var bitmap = new SKBitmap(400, 120))+ using (var canvas = new SKCanvas(bitmap))+ using (var font = new SKFont(SKTypeface.Default, 48))+ using (var paint = new SKPaint { Color = SKColors.Black, IsAntialias = true })+ {+ canvas.Clear(SKColors.White);+ canvas.DrawText("Filee OCR Test", 10, 70, font, paint);+ using var image = SKImage.FromBitmap(bitmap);+ using var data = image.Encode(SKEncodedImageFormat.Png, 100);+ using var stream = File.OpenWrite(imagePath);+ data.SaveTo(stream);+ }++ var result = await engine.RecognizeAsync(imagePath, "en", TestContext.Current.CancellationToken);++ Assert.Contains("Filee", result.Text, StringComparison.OrdinalIgnoreCase);+ Assert.NotEmpty(result.Words);+ Assert.True(result.ImageWidth > 0 && result.ImageHeight > 0);+ }+ finally+ {+ File.Delete(imagePath);+ }+ }++ [Fact]+ public async Task RecognizeAsync_throws_a_clear_error_for_an_unmapped_language()+ {+ var engine = new TesseractOcrEngine();+ Assert.SkipUnless(engine.IsAvailable, "Tesseract is not installed on this machine.");++ // Any existing file: the unmapped-language check happens before the image is read.+ var imagePath = Path.Combine(AppContext.BaseDirectory, "Filee.Engines.Tests.dll");++ var ex = await Assert.ThrowsAsync<InvalidOperationException>(() =>+ engine.RecognizeAsync(imagePath, "xal", TestContext.Current.CancellationToken));+ Assert.Contains("xal", ex.Message);+ }+}diff --git a/tests/Filee.Platform.Tests/Filee.Platform.Tests.csproj b/tests/Filee.Platform.Tests/Filee.Platform.Tests.csprojindex a3d0e81a..27f0ce7f 100644--- a/tests/Filee.Platform.Tests/Filee.Platform.Tests.csproj+++ b/tests/Filee.Platform.Tests/Filee.Platform.Tests.csproj@@ -8,6 +8,7 @@<ProjectReference Include="..\..\src\Filee.Platform.Linux\Filee.Platform.Linux.csproj" /><ProjectReference Include="..\..\src\Filee.Platform.MacOS\Filee.Platform.MacOS.csproj" /><ProjectReference Include="..\..\src\Filee.Platform.Windows\Filee.Platform.Windows.csproj" />+ <ProjectReference Include="..\..\src\Filee.Engines\Filee.Engines.csproj" /><PackageReference Include="Microsoft.NET.Test.Sdk" /><PackageReference Include="xunit.v3" /><PackageReference Include="xunit.runner.visualstudio" />diff --git a/tests/Filee.Platform.Tests/VisionOcrTests.cs b/tests/Filee.Platform.Tests/VisionOcrTests.csnew file mode 100644index 00000000..de906b16--- /dev/null+++ b/tests/Filee.Platform.Tests/VisionOcrTests.cs@@ -0,0 +1,198 @@+using Filee.Core.Conversion;+using Filee.Core.Ocr;+using Filee.Core.Presets;+using Filee.Engines.Hwp.Hwpx;+using Filee.Engines.Ocr;+using Filee.Engines.Pdf;+using Filee.Platform.MacOS;++namespace Filee.Platform.Tests;++public sealed class VisionOcrTests+{+ private static string FindRepoRoot()+ {+ var dir = new DirectoryInfo(AppContext.BaseDirectory);+ while (dir is not null && !File.Exists(Path.Combine(dir.FullName, "Filee.slnx")))+ dir = dir.Parent;+ return dir!.FullName;+ }++ [Fact]+ public async Task VisionOcrRecognizesTextInImageOnMacOS()+ {+ if (!OperatingSystem.IsMacOS())+ {+ Assert.Skip("Requires macOS Vision framework.");+ return;+ }++ var engine = new MacOsVisionOcrEngine();+ Assert.True(engine.IsAvailable);+ Assert.Equal("vision", engine.Id);+ Assert.NotEmpty(engine.SupportedLanguages);++ var imagePath = Path.Combine(FindRepoRoot(), "docs", "media", "banner-en.jpg");+ Assert.True(File.Exists(imagePath), $"Test image {imagePath} should exist");++ var result = await engine.RecognizeAsync(imagePath, "en-US", TestContext.Current.CancellationToken);++ Assert.NotEmpty(result.Text);+ Assert.NotEmpty(result.Lines);+ Assert.NotEmpty(result.Words);+ Assert.True(result.ImageWidth > 0);+ Assert.True(result.ImageHeight > 0);++ // The banner image has "Filee"+ Assert.Contains("Filee", result.Text, StringComparison.OrdinalIgnoreCase);+ }++ [Fact]+ public async Task ImageOcrConverterConvertsImageToTextAndDocx()+ {+ if (!OperatingSystem.IsMacOS())+ {+ Assert.Skip("Requires macOS Vision framework.");+ return;+ }++ var engine = new MacOsVisionOcrEngine();+ OcrEngineRegistry.Register(engine);++ var converter = new ImageOcrConverter();+ Assert.True(converter.GetStatus().IsAvailable);++ var imagePath = Path.Combine(FindRepoRoot(), "docs", "media", "banner-en.jpg");+ var workDir = Path.Combine(Path.GetTempPath(), $"filee-test-{Guid.NewGuid():N}");+ Directory.CreateDirectory(workDir);++ try+ {+ var txtOut = Path.Combine(workDir, "banner.txt");+ var txtStep = new ConversionStep(+ InputPath: imagePath,+ From: "jpg",+ To: "txt",+ Preset: new Preset(),+ Output: new FixedPath(txtOut),+ WorkDirectory: workDir);++ var txtOutputs = await converter.ConvertAsync(txtStep, null, TestContext.Current.CancellationToken);+ Assert.Single(txtOutputs);+ Assert.True(File.Exists(txtOutputs[0]));+ var txt = await File.ReadAllTextAsync(txtOutputs[0], TestContext.Current.CancellationToken);+ Assert.Contains("Filee", txt, StringComparison.OrdinalIgnoreCase);++ var docxOut = Path.Combine(workDir, "banner.docx");+ var docxStep = new ConversionStep(+ InputPath: imagePath,+ From: "jpg",+ To: "docx",+ Preset: new Preset(),+ Output: new FixedPath(docxOut),+ WorkDirectory: workDir);++ var docxOutputs = await converter.ConvertAsync(docxStep, null, TestContext.Current.CancellationToken);+ Assert.Single(docxOutputs);+ Assert.True(File.Exists(docxOutputs[0]));+ Assert.True(new FileInfo(docxOutputs[0]).Length > 1000);+ }+ finally+ {+ Directory.Delete(workDir, recursive: true);+ }+ }++ [Fact]+ public void PdfDocumentReaderExtractsEditableTextFromScannedPdf()+ {+ if (!OperatingSystem.IsMacOS())+ {+ Assert.Skip("Requires macOS Vision framework.");+ return;+ }++ var engine = new MacOsVisionOcrEngine();+ OcrEngineRegistry.Register(engine);++ var imagePath = Path.Combine(FindRepoRoot(), "docs", "media", "banner-en.jpg");+ var workDir = Path.Combine(Path.GetTempPath(), $"filee-pdf-test-{Guid.NewGuid():N}");+ Directory.CreateDirectory(workDir);++ try+ {+ var pdfPath = Path.Combine(workDir, "scanned.pdf");++ // Create a pure scanned PDF with only a bitmap image using PdfSharp+ using (var pdfDoc = new PdfSharp.Pdf.PdfDocument())+ {+ var page = pdfDoc.AddPage();+ page.Width = PdfSharp.Drawing.XUnit.FromPoint(600);+ page.Height = PdfSharp.Drawing.XUnit.FromPoint(400);+ using var gfx = PdfSharp.Drawing.XGraphics.FromPdfPage(page);+ using var img = PdfSharp.Drawing.XImage.FromFile(imagePath);+ gfx.DrawImage(img, 0, 0, 600, 400);+ pdfDoc.Save(pdfPath);+ }++ Assert.True(File.Exists(pdfPath));++ // Read the scanned PDF using PdfDocumentReader with OCR enabled+ var mediaFolder = Path.Combine(workDir, "media");+ var document = PdfDocumentReader.Read(pdfPath, mediaFolder, engine);++ Assert.NotEmpty(document.Sections);+ var paragraphs = document.Sections.SelectMany(s => s.Blocks).OfType<HParagraph>().ToList();+ Assert.NotEmpty(paragraphs);++ var allText = string.Join(" ", paragraphs.SelectMany(p => p.Inlines).OfType<HText>().Select(t => t.Text));+ Assert.Contains("Filee", allText, StringComparison.OrdinalIgnoreCase);++ // Verify plain text reader also extracts the OCR text+ var plainDoc = PdfDocumentReader.ReadText(pdfPath, engine);+ var plainText = PdfDocumentReader.PlainText(plainDoc);+ Assert.Contains("Filee", plainText, StringComparison.OrdinalIgnoreCase);+ }+ finally+ {+ Directory.Delete(workDir, recursive: true);+ }+ }++ [Fact]+ public void ScannedPdfWithoutOcrStaysPictureOnly()+ {+ var imagePath = Path.Combine(FindRepoRoot(), "docs", "media", "banner-en.jpg");+ var workDir = Path.Combine(Path.GetTempPath(), $"filee-pdf-noocr-{Guid.NewGuid():N}");+ Directory.CreateDirectory(workDir);++ try+ {+ var pdfPath = Path.Combine(workDir, "scanned.pdf");+ using (var pdfDoc = new PdfSharp.Pdf.PdfDocument())+ {+ var page = pdfDoc.AddPage();+ page.Width = PdfSharp.Drawing.XUnit.FromPoint(600);+ page.Height = PdfSharp.Drawing.XUnit.FromPoint(400);+ using var gfx = PdfSharp.Drawing.XGraphics.FromPdfPage(page);+ using var img = PdfSharp.Drawing.XImage.FromFile(imagePath);+ gfx.DrawImage(img, 0, 0, 600, 400);+ pdfDoc.Save(pdfPath);+ }++ var document = PdfDocumentReader.Read(pdfPath, Path.Combine(workDir, "media"), null, null, enableOcr: false);+ var paragraphs = document.Sections.SelectMany(s => s.Blocks).OfType<HParagraph>()+ .SelectMany(p => p.Inlines).OfType<HText>().Select(t => t.Text).ToList();+ Assert.Empty(paragraphs);+ }+ finally+ {+ Directory.Delete(workDir, recursive: true);+ }+ }++ private sealed class FixedPath(string path) : IOutputAllocator+ {+ public string? Allocate(string extension, string? suffix = null) => path;+ }+}