Design Pattern for Image and Text Composition in Go

Go Conference Tokyo 2019 Spring

Transcript

1. Design pattern 

   for Image and text
   composition in Go
   Go Conference 2019 Spring

   @__timakin__
   

   View Slide

2. SEIJI TAKAHASHI
   Software Engineer at Gunosy Inc.
   

   View Slide

3. Agenda
   > Image and font packages
   > Design pattern for composition
   > Conclusions
   

   View Slide

4. There are so much stuff to do when you
   composite image resources and
   generate a new one.
   

   View Slide

5. Scope for this talk is …
   > Workflow of image composition
   > How to implement them along with the separation of
   concerns.
   

   View Slide

6. Scope for this talk is …
   Timakin
   Timakin
   

   View Slide

7. Not scope for this talk ...
   > The detail of the image package
   > Image processing except composition
   

   View Slide

8. Image and font packages
   

   View Slide

9. Quick Question
   Have you ever used the image package?
   

   View Slide

10. package “image”
    > Package image implements a basic 2-D image library.
    • Encodes/decodes formats such as GIF, JPEG or PNG.
    > Provides below interfaces
    • An inspection of data in the pixels
    • Adds a color to the specified range
    • Composition, mask and quantization.
    

    View Slide

11. e.g. Draw a red rectangle
    

    View Slide

12. golang.org/x/image/font
    > Package which defines an interface for font faces, for
    drawing text on an image.
    > font.Face is an interface that represents a content of
    the file of font face.
    > font.Drawer reads it and draws text on a destination
    image.
    > Opentype is not suitable for font.Drawer.
    

    View Slide

13. github.com/golang/freetype/truetype
    > Package which provides a parser for the TTF and TTC
    file formats.
    > When you would like to flexibly set a font face to
    font.Drawer, this package will really help you.
    > A lot of .ttf/.ttc files don’t pass the parse process.
    • PostScript font type
    • Hiragino Sans…
    

    View Slide

14. Design pattern of image composition
    

    View Slide

15. Workflow of composition
    > To generate this name card, 

    you have to
    • Initialize a frame
    • Draw images
    • Attach labels
    • Encode a jpeg output
    

    View Slide

16. Interfaces
    Processor
    Initialize a frame
    Draw images
    Attach labels
    Encode a jpeg output
    Framer
    Drawer
    Labeler
    Encoder
    

    View Slide

17. Roles of each structs
    Framer Drawer Labeler Encoder
    

    View Slide

18. Processor
    > Loads fixture images and pass

    them to Drawer.
    > Includes Framer, Drawer,

    Encoder as fields.
    

    View Slide

19. Loading fixtures
    > github.com/rakyll/statik is better if your fixtures are not
    so large to save the loading time.
    

    View Slide

20. Insertion of dependent structs
    > Optional values
    • Main Thumbnail
    • Sub image
    • Background image
    • Labels
    • Color
    • Font family/size
    • Output path
    

    View Slide

21. Framer
    > Initializes *image.RGBA 

    as a canvas.
    > Width & height, and 

    a background color are required.
    

    View Slide

22. Fill the RGBA frame with colors
    

    View Slide

23. Drawer
    > Receiving the frame,
    drawer draws fixture
    and specified images.
    

    View Slide

24. Call drawing and labeling functions inside
    > Labeler is called
    inside of Drawer.
    • The positions of the
    texts are calculated
    by relative value to
    the images.
    

    View Slide

25. Call drawing and labeling functions inside
    Sometimes you have to guard an
    unexpected-sized images, so drawing
    function often contains a resizer.
    

    View Slide

26. Labeler
    > Labeler attaches a
    text label with the
    options.
    (coordinate, font-
    family, text
    alignment etc)
    

    View Slide

27. Loading assets and font.Drawer
    > After loading font assets,
    call truetype font parser
    with options of size and
    color.
    

    View Slide

28. A minor adjustment of text position 

    with fixed.PointXX_xx
    

    View Slide

29. Encoder
    > Encoder just calls encode
    function the image package.
    > The role of this interface is to
    abstract the difference between
    each formats.
    

    View Slide

30. View Slide

31. Conclusion
    > Image and text compression becomes really messy if you
    don’t care about the separation of concerns.
    • Separation by the role becomes easy if you simplify the
    process.
    > Composition process deeply depends on a coordinate
    calculation. Designing optional values is the essential for
    a simple code.
    

    View Slide

32. Thank you for listening
    

    View Slide