Meziantou.Framework

Converting to PDF/A

edit

PdfDocument.ConvertToPdfA changes an existing document, such as a file a user sends, so that it can be saved as PDF/A-2b, PDF/A-2u, PDF/A-3b or PDF/A-3u. It removes what PDF/A does not allow and completes what it needs, without moving anything on the pages, then checks the document as the save does:

var document = PdfDocument.Open(file);
var result = document.ConvertToPdfA(PdfAConformance.PdfA2B);

// When the conversion succeeds, the save writes a PDF/A-2b document; otherwise, the violations tell what it could
// not fix, such as a font that is not embedded and has no substitute
var pdf = result.Succeeded ? document.ToByteArray() : null;
var reasons = string.Join(Environment.NewLine, result.Violations);

When Succeeded is true, PdfAConformance is set to the level, and the save writes a PDF/A document. Changes lists what the conversion did, and Violations what it could not fix. The fixes change the document even when some requirements cannot be met: open the file again to start over.

What the conversion fixes

  • Encryption: removed. Open the document with its password first.
  • Version: a document of PDF 2.0 is set to PDF 1.7, the version of PDF/A-2 and PDF/A-3.
  • Actions: the JavaScript of the document, the actions PDF/A does not allow (Launch, JavaScript, Sound, Movie, ResetForm, ImportData, Hide, ...), the actions of the widgets and every additional action (/AA) are removed. The links to pages and to web sites are kept.
  • Forms: the XFA form is removed; when the form asks the viewer to generate the appearances of the fields, they are generated and the flag is removed.
  • Annotations: the 3D, Sound, Screen, Movie and RichMedia annotations are removed, the appearances other than the normal one are removed, and the missing appearances are generated for the shapes, the lines, the ink, the text markups and the fields.
  • Fonts: a standard font that is not embedded (Helvetica, Times, Courier and their styles, and the names readers accept for them, such as Arial) gets an embedded substitute: Liberation Sans, Serif or Mono, which have the widths of the standard fonts, so the text does not move. It is embedded as a subset of the glyphs of its encoding, with a ToUnicode map.
  • Streams: the streams encoded with LZWDecode, which PDF/A-2 does not allow, are decoded; the save compresses them with FlateDecode.
  • Images and graphics states: the alternate images, the OPI entries, the PostScript XObjects and the transfer functions are removed, and the interpolation of the images is turned off.
  • Embedded files: PDF/A-2 embeds PDF/A documents only, so the other files are removed (use PDF/A-3 to keep them). PDF/A-3 gets their missing names, MIME type (application/octet-stream) and relationship (Unspecified), and associates them with the document.
  • Colors: a document without output intent gets the sRGB one, or the profile of OutputIntentProfile. When the output intent is not RGB, an sRGB DefaultRGB color space gives DeviceRGB its meaning.
  • Metadata: the XMP metadata is removed, and the save writes it again from the document information, whose characters that XML does not allow are removed. Set KeepXmpMetadata to keep it when it has no property of another schema.
  • Optional content: each configuration gets a name, and none keeps automatic states (AS).

What it cannot fix

The conversion reports these, and does not set the level:

  • A font that is not embedded and is not a Latin standard font (such as Garamond, Symbol, ZapfDingbats or a CID font): the library has no substitute for it. Embed it with the tool that made the document.
  • DeviceCMYK without a CMYK output intent: pass the profile of the printing condition in OutputIntentProfile.
  • An annotation without appearance that the library cannot draw, such as a note or a stamp: remove it.
  • For the level u, text whose font gives no Unicode value: convert to the level b.
  • A stream whose data is in an external file, or encoded with LZW data that cannot be decoded.

The conversion is not a validator: it fixes and checks what the save checks (see PDF/A). Validate the converted documents with veraPDF.