djoka/bentopdf
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Fix#20 OCR pdf text not selectable

Browse Source
This commit is contained in:
Shirshendu R Tripathi
2025-10-19 22:12:51 +05:30
parent caf23b0ab9
commit ef7b9d0365
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
src/js/logic/ocr-pdf.ts
View File

@@ -144,7 +144,7 @@ async function runOCR() {
binarizeCanvas(context); binarizeCanvas(context);
} }
const result = await worker.recognize(canvas); const result = await worker.recognize(canvas, {}, { text: true, hocr: true });
const data = result.data; const data = result.data;
const newPage = newPdfDoc.addPage([viewport.width, viewport.height]); const newPage = newPdfDoc.addPage([viewport.width, viewport.height]);
const pngImageBytes = await new Promise((resolve) => const pngImageBytes = await new Promise((resolve) =>
@@ -192,7 +192,7 @@ async function runOCR() {
font, font,
size: fontSize, size: fontSize,
color: rgb(0, 0, 0), color: rgb(0, 0, 0),
opacity: 0, opacity: 0.1,
}); });
} catch (error) { } catch (error) {
// If drawing fails despite sanitization, log and skip this word // If drawing fails despite sanitization, log and skip this word