pampipe

A Node.js module to enable chaining of netpbm programs.

Netpbm is a collection of modules to work with graphic images. For example, you can resize, rotate and convert between formats. This module allows to generate commands to be executed in bash. It does not include binaries or bindings, and you need to have netpbm installed on your machine to perform actual manipulations.

Usage

const Pampiple = require('pampipe').Pampipe;

const pampipe = new Pampipe();
const command = pampipe
  .jpegtopnm('image.jpg')
  .pamflip('r180')
  .pnmtojpeg()
  .getCommand();

console.log(command);
// jpegtopnm image.jpg | pamflip -r180 | pnmtojpeg

// save command appends >filename.jpg to the end of the pipe and asynchronously executes it,
// returning a promise fulfilled with the filename.
new Pampipe()
  .jpegtopnm('image.jpg')
  .pamflip('r180')
  .pnmtojpeg()
  .save('filename.jpg')
  .then((data) => {
      console.log(data); // filename.jpg
  });

jpegtopnm

Reads a jpeg file and converts it into a pnm image.

Options:

• dct [int, fast, float, default int]: which DCT method to use
• nosmooth Boolean: faster, lower-quality unsampling route
• maxmemory int|String: how much memory to use, e.g., 2500000 or '4mb'
• adobe Boolean: true to add -adobe arg and false to add -notadobe arg
• dumpexif Boolean: print exif to stderr
• exif String: extract exif header and wirte it to file
• multiple Boolean: read multiple files from stdin
• repair Boolean: try to repair invalid input
• comments Boolean: print comments in file to stderr
• verbose Boolean: print details to stderr
• tracelevel int: debug level, higher level will print more info

Usage:

const pampipe = new Pampipe();
pampipe
  .jpegtopnm('image.jpg', {
    dct: 'fast',
    nosmooth: true,
    maxmemory: '1m',
    adobe: false,
    exif: 'data.exif',
  })
  .getCommand();
// jpegtopnm -dct fast -nosmooth -maxmemory 1m -notadobe -exif=data.exif image.jpg

http://netpbm.sourceforge.net/doc/jpegtopnm.html

pnmtojpeg

Converts a pnm image to a JPEG.

Options:

• exif String: add exif data from specified file
• quality int (0 to 100, default 75): adjust image quality
• grayscale|greyscale Boolean: produce a grayscale image
• rgb Boolean: use RGB colour space instead of YCbCr
• density String (nxn[dpi,dpcm]): resolution information, e.g., 1x1, 3x2, 300x300dpi, 100x200dpcm
• optimise|optimize Boolean: perform entropy optimisation
• progressive Boolean: create a progressive JPEG
• comment String: add a comment marker
• dct [int, fast, float, default int]: which DCT method to use
• arithmetic String: use arithmetic coding instead of Huffman encoding
• restart int: emit a JPEG restart marker every n MCU rows
• smooth int (0 to 100, default 0): eliminate dithering noise
• maxmemory int|String: how much memory to use, e.g., 2500000 or '4mb'
• verbose Boolean: print details to stderr
• baseline Boolean: force baseline-compatible quantisation tables
• qtables String: use quantisation tables from specified file
• qslots String (n[,...]): which quantisation table to use for each colour component
• sample String (HxV[,...]): sampling factors for each colour component
• scans String: use scan script from specified file
• tracelevel int: debug level, higher level will print more info

Usage:

const pampipe = new Pampipe();
pampipe
  .jpegtopnm('image.jpg', {
      exif: 'data.exif',
  })
  .pamflip('r180')
  .pnmtojpeg({
      exif: 'data.exif', // copy exif from original
      quality: 80,
      progressive: true,
      dct: 'fast',
      qtables: 'qtables.txt',
  })
  .getCommand();
// jpegtopnm -exif=data.exif image.jpg | pamflip -r180 | \
// pnmtojpeg -exif=data.exif -quality=80 -progressive -dct=fast -qtables=qtables.txt

http://netpbm.sourceforge.net/doc/pnmtojpeg.html

pamflip

Allows to transform an image, e.g., rotate, flip, transpose, etc. Allowed transforms are:

• leftright (lr)
• topbottom (tb)
• transpose (xy)
• rotate90 (r90, ccw)
• rotate180 (r180)
• rotate270 (r270, cw)
• null

Options:

• memsize int: size in megabytes of memory available for pamflip (see doc)

Usage:

const pampipe = new Pampipe();
pampipe
  .jpegtopnm('image.jpg')
  .pamflip('r90');
  .getCommand();
// jpegtopnm image.jpg | pamflip -r90

// you can also specify transforms as array, as well as memsize
pampipe
  .jpegtopnm('image.jpg')
  .pamflip(['r180', 'leftright'], {
    memsize: 4,
  })
  .getCommand();
// jpegtopnm image.jpg | pamflip -xform=r180,leftright -memsize=4

http://netpbm.sourceforge.net/doc/pamflip.html

pamscale

Scales a pam image by a specific factor, or each dimension individually.

Options:

• xyfit Array|int: scale to fit bounding box. Pass an array to specify width and height, or pass an integer when dimensions are equal, e.g., [480, 640] or 250.
• xyfill Array|int: scale to smallest size to completely fill the box.
• reduce float: how much to scale image
• xsize int: new width
• ysize int: new height
• width int: synonym for xsize
• height int: synonym for ysize
• xscale float: how much to scale x dimension
• yscale float: how much to scale y dimension
• pixels int: maximum number of pixels
• verbose Boolean: print debug information
• nomix Boolean: use discrete sampling
• filter: which filter function to use (see below)
• window: window filter function to use with filter
• linear: skip conversion between these gamma-adjusted numbers and internal intensity-proportional numbers

Filter functions:

• point
• box
• triangle
• quadratic
• cubic
• catrom
• mitchell
• gauss
• sinc
• bessel
• hanning
• hamming
• blackman
• kaiser
• normal
• hermite
• lanczos

Usage:

const pampipe = new Pampipe();
pampipe
  .jpegtopnm('image.jpg')
  .pamscale({
      xyfit: 500,
      filter: 'gauss',
      linear: true,
  })
  .getCommand();
// jpegtopnm image.jpg | pamscale -xyfit=500 -filter=gauss -linear

http://netpbm.sourceforge.net/doc/pamscale.html

Test

To test the package, you need to have netpbm installed, because integration tests will perform transforms and compare them against expected results. To run tests, simply type npm test.

Benchmark

There is a benchmark which compares performance of Netpbm against vips, which is executed via sharp npm module. We have not included the sharp dependency, so to run the benchmark you need to do npm i sharp first (you will need libvips on your machine). Then, type npm run bench to compare the two.

We found that libvips is about 250% faster when resizing a jpeg to 450px wide and rotating it 90 degrees.