Lexical Scanning in Go

Intro

The content and motivation of this talk is based on a talk by Rob Pike I stumbled upon on youtube. I also found the slides. I also found a blog post by Eli Bendersky taking the aforementioned talk as a basis for a solution in Python.

My shot at this

So I watched the video and that it was a kinda cool idea, also as a chance to highlight Go’s features nicely. But I didn’t want to dig further for the source code nor was I particularily interested in the exact syntax and semantics of the template engine mentioned in the talk. What I was going for instead was a very simple toy implementation of a potential template engine very loosly influenced from what I saw on the slides, just to get a feel for the idea and Go itself.

The template engine

So here is a description of the toy template eninge:

  • Variable content, as in the talk, are kept between {{ and }}
  • An “action” inside these meta characters can have only one of these forms

    • .ident
    • .ident!mod
    • .guard? .ident
    • .guard? .ident!mod
  • ident stands for identifier, mod for modifier and guard for a guard expression
  • The identifier is a lookup string for a map object called the context. It is used to find the text that has to be substituted for the identifier
  • The guard is a string that, if it reads “true” causes the meta action to be evaluated, otherwise it is ignored
  • The modifier can be used to modifiy the string that is specified by ident, in this simple implementation the only supported modifiers are lower and upper
  • The renderer of the template engine will accept a file path specifying the template, another file path specifying the output file and at last a context (Go map) holding all the variables to look up
  • Everything else is interpreted as text and copied as is to the rendered document

The implementation

So without much more talking here comes the code. As already mentioned above a lot of it is based on Rob Pike’s work. I just filled in some gaps here and there to make the the lexer fully working. In the end follows a little script that tests the library.

lexer.go

parser.go

template_test.go

template.txt

The output