PDF-Native-raku

This module provides a selection of native implementations of PDF functions.

Just installing this module along with PDF v0.5.7+ provides some increase in performance.

Currently, this module implements a hand-full of functions, mostly related to reading and writing larger PDF files.

So far, just a subset of potential areas are covered:

• the PDF::IO::Filter::Predictor decode and encode functions.
• the widely used PDF::IO::Util pack and unpack functions.
• reading of cross reference tables and PDF 1.5+ cross reference streams.
• writing of strings, numerics, cross-reference tables and streams.

Classes in this Distribution

PDF::Native::Filter::Predictors

These functions implements the predictor stage of TIFF [1] and PNG [2] decoding and encoding.

    use PDF::Native::Filter::Predictors;
    # PNG samples. First bit on each row, is an indicator in the range 0 .. 4
    my $Predictor = PDF::Native::Filter::Predictors::PNG;
    my $Columns = 4;
    my $encoded = blob8.new: [
        2,  0x1, 0x0, 0x10, 0x0,
        2,  0x0, 0x2, 0xcd, 0x0,
        2,  0x0, 0x1, 0x51, 0x0,
        1,  0x0, 0x1, 0x70, 0x0,
        3,  0x0, 0x5, 0x7a, 0x0,
        0,  0x1, 0x2, 0x3,  0x4,
    ];

    my blob8 $decoded = PDF::Native::Filter::Predictors.decode(
                                        $encoded,
                                        :$Columns,
                                        :$Predictor, );

[1] TIFF Predictors [2] PNG Predictors

PDF::Native::Buf

Handles the packing and unpacking of multi-byte quantities as network words. Such as /BitsPerComponent in /Filter /DecodeParms.

Also handles variable byte packing and unpacking. As seen in the /W parameter to XRef streams, and a few other places.

    # pack two 4-byte words into an 8 byte buffer
    use PDF::Native::Buf :pack;
    my blob32 $words .= new(660510, 2634300);
    my blob8 $bytes = pack($words, 24);

    # pack triples as 1 byte, 2 bytes, 1 byte
    my uint32 @in[4;3] = [1, 16, 0], [1, 741, 0], [1, 1030, 0], [1, 1446, 0];
    my @W = packing-widths(@in, 3);    # [1, 2, 0];
    $bytes = pack(@in, @W);

PDF::Native::Reader

Reading of PDF content. Only method so far implemented is read-xref for the fast reading of cross reference indices.

use PDF::Native::Reader;

given PDF::Native::Reader.new {

     enum <free inuse>;

     my Str $xref = (
         'xref',
         '10 4',
         '0000000000 65535 f ',
         '0000000042 00000 n ',
         '0000000069 00000 n ',
         '9000000100 00002 n ',
         ''
     ).join(10.chr);
     my Blob $buf = $xref.encode('latin-1');

     my array $entries = .read-xref($buf);
}

PDF::Native::Writer

Serialization functions have been implemented for a few PDF data-types:

• boolean, real, integers, literal-strings, hex-strings, names and cross reference tables.

use PDF::Native::Writer;

given PDF::Native::Writer {
     say .write-bool(0);    # false
     say .write-bool(1);    # true
     say .write-real(pi);   # 3.14159
     say .write-int(42e3),  # 42000
     say .write-literal("Hi\nthere"); # (Hi\nthere)
     say .write-hex-string("snoopy"); # <736e6f6f7079>
     say .write-name('Hi#there');     # /Hi##there

     # xref entries
     enum <free inuse>;
     my uint64 @xref[4;3] = (
        [0, 65535, free],
        [42, 0, inuse],
        [69, 0, inuse],
        [100, 2, inuse],
     );
     say .write-entries(@xref).lines;
         # 0000000000 65535 f 
         # 0000000042 00000 n 
         # 0000000069 00000 n 
         # 0000000100 00002 n
}

Todo

Some other areas under consideration:

• C equivalents for other PDF::IO::Filter encoding functions, including predictors, ASCII-Hex, ASCII-85 and run-length encoding
• Support for type-1 character transcoding (PDF::Content::Font::Enc::Glyphic)
• Support for PDF::Content::Image, including color-channel separation and (de)multiplexing. GIF decompression de-interlacing.

There's sure to be others.

Currently giving noticeable benefits on larger PDFs.