Notes from building Confileo — a free, Arabic-first toolkit of 79 PDF, image and document tools. I write about what broke along the way: right-to-left text in PDF extraction, Unicode's missing letters, font embedding, headless PDF rendering, and Angular SSR. Debugging stories, with the actual fix at the end.
Got a photo of a document, a screenshot, or a scanned PDF and need the text as editable, copyable text? That's what OCR (Optical Character Recognition) does — and you can do it free in your browser, w
A PDF can leak in four different ways, and each one has a different fix. A password stops a stranger from opening the file. It does nothing about a bank account number that should never have been in t
You paste 1756771200 into a spreadsheet and want to see a date. Or you export a sheet and your API receives a column of numbers like 46236.5 instead of dates. Both directions look trivial and both go
I spent months writing bidi reconstruction logic for reversed Arabic text in PDF extraction. The actual fix was deleting all of it and upgrading two dependencies. Here is how to tell which one is lying to you.