A PDF can look correct while its extracted text reads in the wrong order. A Markdown document can contain valid Unicode and still export with missing characters. For multilingual work, preserving the words also means preserving how people read them.
Use this guide to convert a multilingual document locally and check its reading order, spacing, and fonts before you share it.
Convert and review a multilingual document
- Preview the source. Identify its language, text direction, and whether a PDF contains actual text or only scanned page images.
- Choose the conversion in FileHop. Use PDF to Markdown for an editable text copy or Markdown to PDF for a document you can share.
- Check a representative passage. Include names, accents, script-specific marks, mixed English and non-English text, and any numbers or punctuation that matter.
- Review the output. Check reading order and fonts as well as appearance. Keep the original alongside the converted file.
Start with the text a person should read
A PDF describes where things are drawn on a page. Those drawing instructions do not necessarily describe a paragraph in reading order. Reading every visible mark from left to right can scramble Arabic or Hebrew, especially when the same line contains English words and numbers.
FileHop’s PDF reading path reconstructs text flow and handles right-to-left runs when producing readable text. It also uses a PDF’s replacement text, when supplied, so a visually drawn glyph can yield the text the document intended it to represent.
The recent changes improve script and spacing handling and retain blank pages in PDF-to-Markdown conversion. They address the document’s content and structure rather than treating a PDF as one long sequence of characters.
Compare the Markdown with the source PDF, especially around columns, tables, and mixed-direction paragraphs. The conversion prioritizes readable text, so the layout can differ.
Spaces are part of the language
Inserting a space between every extracted text fragment can damage Chinese or Japanese text. Removing spaces indiscriminately can damage a language that needs word spacing, including Korean.
FileHop’s PDF text handling distinguishes Han characters, Japanese kana, and related punctuation when joining fragments. The same approach is not applied indiscriminately to Hangul. That distinction helps keep extracted text useful for reading and subsequent editing.
A font must contain the characters you use
When you export Markdown to PDF, the main document typeface may not cover every script in the document. Missing coverage can turn perfectly good text into empty boxes.
FileHop’s Markdown-to-PDF exporter looks for appropriate installed font families for scripts its main faces do not cover. The mappings include Arabic and Hebrew; Devanagari, Bengali, Gurmukhi, Gujarati, Tamil, Telugu, Kannada, and Malayalam; Thai; and Japanese, Korean, and Chinese text.
That list describes document-font handling. It is separate from the languages offered in the app’s interface. For example, a document can contain Malayalam text even though Malayalam is not currently an interface-language choice.
Font availability still matters. An installed family is used only if it is present, and coverage can differ between computers. If an exported document shows missing characters, check the fonts available for that script and review another export before sharing it. Support in this exporter does not imply identical font behavior in every other FileHop tool.
Keep Arabic and Hebrew paragraphs pointed the right way
In Markdown-to-PDF export, FileHop considers a paragraph or heading’s first directional letter when its direction differs from the document’s. This helps an Arabic paragraph inside an English document, or an English paragraph inside a right-to-left document, keep an appropriate reading direction.
For a document whose primary language is Arabic, Markdown front matter can declare the language:
---
lang: ar
---
# تقرير المشروع
هذا ملخص المشروع.
## English summary
The project is ready for review. Language metadata guides document text behavior; it does not translate the content. Inspect mixed-language headings, punctuation, lists, and tables in the exported PDF.
Read older HTML using its declared encoding
Local HTML files are not always UTF-8. A file may declare another character encoding, and decoding it incorrectly can replace accented letters or whole passages with unreadable text.
FileHop’s HTML text-reading path now considers the file’s byte-order mark and declared encoding when decoding the content. This is relevant when extracting text from saved pages and older HTML documents. It is not a claim that every malformed file can be repaired automatically.
Text extraction does not recognize words from scanned images by itself. Scans need an OCR step; that step’s language coverage and privacy depend on the OCR workflow you choose. FileHop also does not translate a document merely because it supports the script.
The local conversion paths described here do not require uploading your documents. If you connect an AI agent to help with translation or editing, that agent’s data handling is a separate part of the workflow.
For agent workflows and format constraints, see the PDF and document reference and conversion options. To choose your interface language, read FileHop speaks your language.