feat: add option to opt out of embedding full fonts in OCR settings

src/js/logic/ocr-pdf-page.ts

@@ -121,6 +121,9 @@ async function runOCR() {
   );
   const binarize = (document.getElementById('ocr-binarize') as HTMLInputElement)
     .checked;
+  const embedFullFonts = (
+    document.getElementById('ocr-embed-full-fonts') as HTMLInputElement
+  ).checked;
   const whitelist = (
     document.getElementById('ocr-whitelist') as HTMLInputElement
   ).value;
@@ -154,6 +157,7 @@ async function runOCR() {
       resolution: scale,
       binarize,
       whitelist,
+      embedFullFonts,
       onProgress: updateProgress,
     });
 

src/js/utils/ocr.ts

@@ -16,6 +16,7 @@ export interface OcrOptions {
   resolution: number;
   binarize: boolean;
   whitelist: string;
+  embedFullFonts?: boolean;
   onProgress?: (status: string, progress: number) => void;
 }
 
@@ -131,7 +132,14 @@ export async function performOcr(
   pdfBytes: Uint8Array | ArrayBuffer,
   options: OcrOptions
 ): Promise<OcrResult> {
-  const { language, resolution, binarize, whitelist, onProgress } = options;
+  const {
+    language,
+    resolution,
+    binarize,
+    whitelist,
+    embedFullFonts,
+    onProgress,
+  } = options;
   const progress = onProgress || (() => {});
 
   const worker = await createConfiguredTesseractWorker(
@@ -198,14 +206,16 @@ export async function performOcr(
         getFontForLanguage('eng'),
       ]);
       primaryFont = await newPdfDoc.embedFont(scriptFontBytes, {
-        subset: false,
+        subset: !embedFullFonts,
       });
       latinFont = await newPdfDoc.embedFont(latinFontBytes, {
-        subset: false,
+        subset: !embedFullFonts,
       });
     } else {
       const fontBytes = await getFontForLanguage(primaryLang);
-      primaryFont = await newPdfDoc.embedFont(fontBytes, { subset: false });
+      primaryFont = await newPdfDoc.embedFont(fontBytes, {
+        subset: !embedFullFonts,
+      });
       latinFont = primaryFont;
     }
   } catch (e) {

src/pages/ocr-pdf.html

@@ -259,6 +259,16 @@
                 />
                 Binarize Image (Enhance Contrast for Clean Scans)
               </label>
+              <label
+                class="flex items-center gap-2 text-sm text-gray-300 cursor-pointer"
+              >
+                <input
+                  type="checkbox"
+                  id="ocr-embed-full-fonts"
+                  class="w-4 h-4 rounded text-indigo-600 bg-gray-700 border-gray-600"
+                />
+                Embed Full Fonts (Larger file, better compatibility)
+              </label>
               <!-- Whitelist Presets -->
               <div>
                 <label