Your first conversion

Converting a document is stream in, stream out. DocWright detects the input format, lays the document out the way Word does, and writes a PDF.

The code

using DocWright;

// One converter serves the whole application: it is thread-safe and caches fonts.
var converter = new DocWrightConverter();

using (FileStream input = File.OpenRead("quarterly-report.docx"))
using (FileStream output = File.Create("quarterly-report.pdf"))
{
    ConversionResult result = converter.Convert(input, output);
    Console.WriteLine($"Wrote {result.PageCount} page(s).");
}

Output

Wrote 3 page(s).

The result

quarterly-report.pdf · page 1 of 3

The first page of the PDF, rendered by DocWright. The table header, borders and the "Page 1 / 3" footer field come straight from the Word document.

What happened

1. The format was detected. Convert sniffs the stream, so the same call accepts .docx, .doc, .rtf, .html and .md. Plain text is the one format that is never guessed. Ask for it with ConvertOptions.SourceFormat = FormatDetection.Text.
2. The document was laid out. Styles were resolved, text measured with the real fonts, and lines, tables and pages broken the way Word breaks them. Fields such as page numbers were filled in.
3. The PDF was written. Fonts are subsetted and embedded, and the output is deterministic. Convert the same file twice and you get the same bytes.

Good practice

• Create one converter and keep it. DocWrightConverter is thread-safe and caches the font engine it builds, which takes about a second the first time. Register it as a singleton in a web application.
• Use any stream. Input can be a FileStream, a MemoryStream or an upload's request body. Output can go to a file, memory or a response.
• Set limits for documents you did not write. Uploads and email attachments can be hostile. Handle errors shows the two limits every server should set.

Next

Load once, render many: parse a document one time, then produce several outputs from it.