Word to PDF Converter

This tool reads a .docx file and rebuilds it as a PDF. A Word document is really a ZIP archive of XML, and the converter opens that archive in your browser, walks the paragraphs, runs, lists and tables inside it, and draws each one onto a PDF page with pdf-lib. Page size and margins come from the document's own section settings, so an A4 file stays A4 and a Letter file stays Letter.

Formatting carries over: bold, italic, underline and strikethrough, font sizes and colors, alignment and indentation, heading styles, bulleted and numbered lists including nested levels, inline images, tables with their column widths and cell shading, and hyperlinks that stay clickable in the PDF.

Three pieces do the work and all three are JavaScript running on your machine. JSZip unzips the .docx, the browser's own DOMParser reads document.xml, styles.xml, numbering.xml and theme1.xml out of it, and pdf-lib writes the resulting page objects back out as PDF bytes. There is no round trip to a converter service anywhere in that chain, which you can confirm in the network panel of your developer tools while a file is converting.

That has practical consequences beyond privacy. The page keeps working after you go offline, a 15MB document starts converting instantly instead of after an upload, and conversion speed depends on your own CPU rather than on how busy a shared server happens to be.

Word documents authored in Word rarely name a font outright. They point at a theme -- w:asciiTheme="minorHAnsi" -- and the actual name, usually Calibri or Cambria, lives in theme1.xml. The converter resolves that reference, then maps the family onto one of the fourteen fonts every PDF reader already has: sans-serif faces go to Helvetica, serif faces to Times, monospaced faces to Courier, each with its bold and italic variants.

Those fourteen built-in fonts cover Latin characters only, so an English document embeds nothing and the output stays a few kilobytes. As soon as the text includes Korean, Japanese or Chinese, a Noto Sans CJK face is fetched, cut down to just the glyphs the document actually uses, and embedded. Exact font matching is out of scope -- Malgun Gothic and MS Mincho both come out as Noto Sans CJK -- and any character with no glyph even there is listed for you rather than silently dropped.

Tables are supported, including horizontal merges, cell backgrounds and borders, and rows that break across pages. Vertically merged cells are drawn with the continuation cells left blank. Not supported: text boxes, shapes, WordArt and SmartArt, charts, equations, footnotes and endnotes, comments, multi-column layouts, tables nested inside table cells, and EMF or WMF images. When the converter meets one of these it skips that element, finishes the rest of the document, and tells you afterwards exactly what was left out and how many times -- so a thinner PDF is never a surprise.

Old binary .doc files cannot be read at all; open one in Word and save it as .docx first. Password-protected documents must be unlocked before conversion, since the contents are encrypted.

A converted document usually needs one more step. Photo-heavy reports come out larger than the DOCX did, and the PDF Compressor re-encodes their images to bring the size back down. A cover letter and a résumé converted separately become one submission with the PDF Merger. A contract converted from a Word template can be signed straight away with Sign PDF, which stamps a drawn signature onto the page so nothing has to be printed and scanned back.