README

ExImageInfo

ElixirLicenseExCoverallsTests

ExImageInfo is an Elixir library to parse images (binaries) and get the dimensions, detected mime-type and overall validity for a set of image formats.

GitHub repo    Docs    Hex.pm package

Table of Contents

  1. Description
  2. Installation
  3. Examples
  4. Design decisions
  5. Acknowledgments
  6. Author
  7. Contributors
  8. ChangeLog
  9. License

Description

Main module that checks and gets if a binary seems to be an image (specific format), the mime-type (and variant detected) and the dimensions of the image (based on the type).

It has convention functions to guess the type of an image by trying the formats supported by the library.

Main features

*Note: both cases as a general overview (partially checked).

Formats

Supported formats (image type to be parsed as):

Mime-types and Variants

The image variant type is an invented string to identify the type of format recognized by this library (more specific than the mime-type).

Each mime-type can be linked to at least one variant type:

mime-type variant type description
image/bmpBMP
image/gifGIF87a 87a gif spec
image/gifGIF89a 89a gif spec
image/x-iconICO
image/jpegbaseJPEG baseline JPEG
image/jpegprogJPEG progressive JPEG
image/jp2JP2 JPEG2000
image/pngPNG
image/x-portable-anymapPNMpbm Portable BitMap
image/x-portable-anymapPNMpgm Portable GrayMap
image/x-portable-anymapPNMppm Portable PixMap
image/psdPSD
image/tiffTIFFII II variant
image/tiffTIFFMM MM variant
image/webpwebpVP8 lossy
image/webpwebpVP8L lossless

The variant type is created just to provide a bit more of information for every image format (if applicable).

Note: :ico returns the dimensions of the largest image contained (not the first found).

The guessing functions try to detect the format of the binary by testing every available type based on its global usage (popularity, usage of image file formats, but still keeping the :png as the first one):

Warnings:

Installation

Add ex_image_info to your list of dependencies in mix.exs.

From Hex:

  def deps do
    [ 
      # ...
      {:ex_image_info, "~> 0.2.0"},
    ]
  end

Or GitHub:

  def deps do
    [ 
      # ...
      {:ex_image_info, github: "rNoz/ex_image_info"},
    ]
  end

Then, use it:

require ExImageInfo
# ExImageInfo.seems? ...

Examples

The following examples are run with the latest version of the library under the next environment:

Erlang/OTP 19 [erts-8.0.2] [source] [64-bit] [smp:8:8] [async-threads:10] [hipe] [kernel-poll:false]

Interactive Elixir (1.3.2) - press Ctrl+C to exit (type h() ENTER for help)
iex(1)> 

Feature seems?

89 50 4E 47 0D 0A 1A 0A are the first 8 bytes in the PNG signature (PNG\\r\\n0x1A\\n).

iex(1)> ExImageInfo.seems? <<0x89504E470D0A1A0A::size(64)>>, :png
true

iex(2)> ExImageInfo.seems? <<0x89504E470D0A1A0A::size(64)>>, :webp
false

ExImageInfo.seems?/2 and ExImageInfo.seems?/1 does not necessarily needs a real image (as it is shown in the previous example) because it just checks the signature of every file format.

Usually it is used as:

iex(1)> ExImageInfo.seems? File.read!("path/to/image.gif"), :gif
true

iex(2)> maybe_png_binary |> ExImageInfo.seems? :png
false

38 42 50 53 are the first 4 bytes in the PSD signature (8BPS).

iex(1)> ExImageInfo.seems? <<0x38425053::size(32)>>
:psd

iex(2)> ExImageInfo.seems? <<0x384250::size(24)>>
nil

ExImageInfo.seems?/2 and ExImageInfo.seems?/1 does not necessarily needs a real image (as it is shown in the previous example) because it just checks the signature of every file format.

Usually it is used as:

iex(1)> ExImageInfo.seems? File.read!("path/to/image.unknown")
:tiff

iex(2)> webp_full_binary |> ExImageInfo.seems?
:webp

Feature type

89 50 4E 47 0D 0A 1A 0A are the first 8 bytes in the PNG signature (PNG\\r\\n0x1A\\n).

iex(1)> ExImageInfo.type <<0x89504E470D0A1A0A::size(64)>>, :png
nil

iex(2)> ExImageInfo.type <<"RIFF", 0::size(32), "WEBPVP8L", 0::size(32), 0x2F7AC07100358683B68D::size(80)>>, :webp
{"image/webp", "webpVP8L"}

The signature part of a png it is now enough to get the type (it check also the IHDR field, just before the width and height).

Usually it is used as:

iex(1)> ExImageInfo.type File.read!("path/to/image.gif"), :gif
{"image/gif", "GIF87a"}

iex(2)> maybe_png_binary |> ExImageInfo.type :png
nil

The guessed version.

iex(1)> ExImageInfo.type <<0x38425053::size(32)>>
{"image/psd", "PSD"}

iex(2)> ExImageInfo.type <<0x384250::size(24)>>
nil

Usually it is used as:

iex(1)> ExImageInfo.type File.read!("path/to/image.unknown")
{"image/tiff", "TIFFMM"}

iex(2)> webp_full_binary |> ExImageInfo.type
{"image/webp", "webpVP8"}

Feature info

89 50 4E 47 0D 0A 1A 0A are the first 8 bytes in the PNG signature (PNG\\r\\n0x1A\\n).

iex(1)> ExImageInfo.info <<0x89504E470D0A1A0A::size(64)>>, :png
nil

iex(2)> ExImageInfo.info <<"RIFF", 0::size(32), "WEBPVP8L", 0::size(32), 0x2F7AC07100358683B68D::size(80)>>, :webp
{"image/webp", 123, 456, "webpVP8L"}

The signature part of a png it is now enough to get the type (it check also the IHDR field, just before the width and height).

Usually it is used as:

iex(1)> ExImageInfo.info File.read!("path/to/image.gif"), :gif
{"image/gif", 1920, 1080, "GIF87a"}

iex(2)> maybe_png_binary |> ExImageInfo.info :png
nil
iex(1)> ExImageInfo.info <<0x38425053::size(32)>>
nil

iex(2)> ExImageInfo.info <<0x38425053::size(32), 0::size(80), 10::size(32), 12::size(32)>>
{"image/psd", 12, 10, "PSD"}

Usually it is used as:

iex(1)> ExImageInfo.info File.read!("path/to/image.unknown")
{"image/tiff", 128, 256, "TIFFMM"}

iex(2)> webp_full_binary |> ExImageInfo.info
{"image/webp", 20, 100, "webpVP8"}

Design decisions

Why seems? and not magic? or signature??

Because for some formats it is enough with the magic number or the signature to get the type (image format that "starts" correctly), but in other cases it is an algorithm a bit more complex to see if the binary seems correct. Therefore, seems it is more general (than getting the magic number) and it will provide a "quick overview" of the validity of the binary.

Why returning the mime-type and variant type when getting the dimensions (info)?

Because both types (variant if applicable) are necessary to obtain the width and height of the binary for a specific format. In case it is required both the type (and variant) and the dimensions it is not necessary to call two functions (and re-parse part or completely the binary). Therefore, to get the dimensions it is obtained the types and all the information is returned in one step.

Renamed from ExImageSize to ExImageInfo

Although it has been released since the very first version with the name ExImageInfo, this library was previously known as ExImageSize, but it is preferable to have a name less restricted. Nowadays it can get information about the type and the dimensions (size), but in a future it could increase the amount of info to extract from an image.

Acknowledgments

This idea comes from libraries that I have used in other platforms and/or languages. Algorithms and some concepts are picked and based on parts of the following:

Thanks to them.

Author

rNoz rnoz.commits@gmail.com.

Contributors

See CONTRIBUTORS for more information.

GitHub readers (repo, no docs): CONTRIBUTORS.md.

ChangeLog

See CHANGELOG for more information.

GitHub readers (repo, no docs): CHANGELOG.md.

License

ExImageInfo source code is released under the MIT License.

See LICENSE for more information.

GitHub readers (repo, no docs): LICENSE.md.