stability and settings

2025-09-17 12:36:24 +00:00
parent 4d586a46e9
commit 2115238217
9 changed files with 1271 additions and 265 deletions
--- a/settings.yml
+++ b/settings.yml
@@ -0,0 +1,179 @@
+# settings.yml
+
+# General application settings
+app_settings:
+  max_file_size_mb: 2000 # Maximum upload size in Megabytes
+
+# Settings for Optical Character Recognition (OCR) tasks
+ocr_settings:
+  ocrmypdf:
+    deskew: true
+    clean: true
+    optimize: 1
+    force_ocr: true
+
+# Settings for audio transcription tasks
+transcription_settings:
+  whisper:
+    compute_type: "int8"
+    allowed_models:
+      - "tiny"
+      - "base"
+      - "small"
+      - "medium"
+      - "large-v3"
+      - "distil-large-v2"
+
+# --- Conversion Tool Definitions ---
+# Each tool's command is a single string. The backend uses shlex to parse it,
+# so you can use quotes for arguments with spaces.
+# Use placeholders like: {input}, {output}, {output_dir}, {output_ext}, etc.
+conversion_tools:
+  libreoffice:
+    name: "LibreOffice"
+    command_template: 'libreoffice --headless --convert-to {output_ext} --outdir {output_dir} {input}'
+    formats:
+      pdf: "PDF"
+      docx: "Word Document"
+      odt: "OpenDocument Text"
+      html: "HTML"
+      rtf: "Rich Text Format"
+      txt: "Plain Text"
+      xml: "Word 2003 XML"
+      epub: "EPUB"
+      xlsx: "Excel Spreadsheet"
+      ods: "OpenDocument Spreadsheet"
+      csv: "CSV"
+      pptx: "PowerPoint Presentation"
+      odp: "OpenDocument Presentation"
+      svg: "SVG"
+  pandoc:
+    name: "Pandoc"
+    command_template: 'pandoc --standalone {input} -o {output}'
+    formats:
+      docx: "Word Document"
+      odt: "OpenDocument Text"
+      pdf: "PDF"
+      rtf: "Rich Text Format"
+      txt: "Plain Text"
+      tex: "LaTeX"
+      man: "Groff Man Page"
+      epub: "EPUB v3 Book"
+      epub2: "EPUB v2 Book"
+      html: "HTML"
+      html5: "HTML5"
+      pptx: "PowerPoint Presentation"
+      beamer: "Beamer PDF Slides"
+      slidy: "Slidy HTML Slides"
+      md: "Markdown"
+      rst: "reStructuredText"
+      jira: "Jira Wiki Markup"
+      mediawiki: "MediaWiki Markup"
+  ghostscript_pdf:
+    name: "Ghostscript (PDF)"
+    command_template: 'gs -sDEVICE=pdfwrite -dCompatibilityLevel=1.4 -dNOPAUSE -dQUIET -dBATCH {preset} -sOutputFile={output} {input}'
+    formats:
+      screen: "PDF (Optimized for Screen)"
+      ebook: "PDF (Optimized for Ebooks)"
+      printer: "PDF (Optimized for Print)"
+      archive: "PDF/A (for Archiving)"
+  ghostscript_image:
+    name: "Ghostscript (Image)"
+    command_template: 'gs -dNOPAUSE -dBATCH -sDEVICE={device} -r{dpi} -sOutputFile={output} {input}'
+    formats:
+      jpeg_72: "JPEG Image (72 DPI)"
+      jpeg_300: "JPEG Image (300 DPI)"
+      png16m_150: "PNG Image (150 DPI)"
+      png16m_300: "PNG Image (300 DPI)"
+      tiff24nc_300: "TIFF Image (300 DPI)"
+      tiff24nc_600: "TIFF Image (600 DPI)"
+  calibre:
+    name: "Calibre (ebook-convert)"
+    command_template: 'ebook-convert {input} {output}'
+    formats:
+      epub: "EPUB"
+      mobi: "MOBI"
+      azw3: "Amazon Kindle"
+      pdf: "PDF"
+      docx: "Word Document"
+  ffmpeg:
+    name: "FFmpeg"
+    command_template: 'ffmpeg -i {input} -y -preset medium {output}'
+    formats:
+      mp4: "MP4 Video"
+      mkv: "MKV Video"
+      mov: "MOV Video"
+      webm: "WebM Video"
+      mp3: "MP3 Audio"
+      wav: "WAV Audio"
+      flac: "FLAC Audio"
+      gif: "Animated GIF"
+  vips:
+    name: "VIPS"
+    command_template: 'vips copy {input} {output}[Q=90]'
+    formats:
+      jpg: "JPEG Image (Q90)"
+      png: "PNG Image"
+      webp: "WebP Image (Q90)"
+      tiff: "TIFF Image"
+      avif: "AVIF Image"
+  graphicsmagick:
+    name: "GraphicsMagick"
+    command_template: 'gm convert {input} -quality 90 {output}'
+    formats:
+      jpg: "JPEG Image (Q90)"
+      png: "PNG Image"
+      webp: "WebP Image"
+      tiff: "TIFF Image"
+      pdf: "PDF from Images"
+  inkscape:
+    name: "Inkscape"
+    command_template: 'inkscape {input} --export-filename={output}'
+    formats:
+      svg: "SVG (Plain)"
+      png: "PNG Image (96 DPI)"
+      pdf: "PDF Document"
+  libjxl:
+    name: "libjxl (cjxl)"
+    command_template: 'cjxl {input} {output} -q 90'
+    formats:
+      jxl: "JPEG XL (Q90)"
+  resvg:
+    name: "resvg"
+    command_template: 'resvg {input} {output}'
+    formats:
+      png: "PNG from SVG"
+  potrace:
+    name: "Potrace"
+    command_template: 'potrace {input} --svg -o {output}'
+    formats:
+      svg: "SVG from Bitmap"
+  markitdown:
+    name: "Markitdown"
+    command_template: 'markitdown {input} -o {output}'
+    formats:
+      md: "Markdown from Everything!"
+  pngquant:
+    name: "pngquant"
+    command_template: 'pngquant --quality={quality} --speed {speed} --force --output {output} {input}'
+    formats:
+      png_hq: "PNG (High Quality Compression)"
+      png_mq: "PNG (Medium Quality Compression)"
+      png_fast: "PNG (Fast Compression)"
+  sox:
+    name: "SoX Audio Converter"
+    command_template: 'sox {input} -r {samplerate} -b {bitdepth} {output}'
+    formats:
+      wav_48k_24b: "WAV (48kHz, 24-bit)"
+      wav_44k_16b: "WAV (CD, 44.1kHz, 16-bit)"
+      flac_48k_24b: "FLAC (48kHz, 24-bit)"
+      flac_44k_16b: "FLAC (CD, 44.1kHz, 16-bit)"
+      ogg_32k: "Ogg Vorbis (32kHz)"
+      ogg_16k: "Ogg Vorbis (16kHz, Voice)"
+  mozjpeg:
+    name: "MozJPEG"
+    command_template: 'cjpeg -quality {quality} -outfile {output} {input}'
+    formats:
+      jpg_q85: "JPEG (High Quality)"
+      jpg_q75: "JPEG (Web Quality)"
+      jpg_q60: "JPEG (Aggressive Compression)"