Pretty Printer

[xurtis]

Currently, all output is through display and printed on a single line. It'd be good to have a pretty printer that nicely formats the output using the following rules:

• Variable indentation rules (n-spaces or tabs)
• Loose or tight bracing options
• S-expressions with no child lists are printed on one line with either loose or tight bracing
• S-expressions with child sub-lists:
  • Sparse formatting places the first brace on its own line. Each value is placed on its own line
    with indentation advanced beyond the brace.
  • Condensed format places the first brace on the same line as the first value. The first value
    is then indented beyond the brace to line up with the following arguments.
  • The last brace is always on its own line, unindented

Condensed format with tight bracing

( (blah blah blah)
  ( blah
    (blah blah blah)
  )
  blah
)

Sparse formatting with loose bracing

( 
  ( blah blah blah )
  ( 
    blah
    ( blah blah blah )
  )
  blah
)

[jashank]

There is much prior art in s-expression pretty printing, though most of it focuses specifically on rules for Lisp-like languages. If that's your target here, you probably want to implement (something like) the X3J13 pretty-printing style or mimic the Guile or Racket pretty printers.

In summary, the former, tightly-braced style makes a good deal more sense as a starting point, but adding behaviours depending on symbols would be useful. Symbols should also always be tightly-braced before and after, as dangling parens or loose parens tend to look bad in the context of Lisps, so:

((blah blah blah)
  (blah
    (blah blah blah))
  blah)

[xurtis]

Let me propose a few rules.

Classifying Values

First, we classify every value as either code or data in an implicit manner.

• A list is considered data when:
  • it is explicitly marked as data, or
  • it begins with an explicit data value, or
  • it begins with a non-symbol value.
• Any list beginning with a symbol that is either explicitly marked as code or is not
  marked as data is itself considered to be code.
• Symbols are considered code unless marked otherwise.
• All other values are considered data.

Secondly, we classify the depth of each list as such:

• Int, Float, and String all have a depth of 0.
• The depth of a cons pair that is not a list is equal to the depth of its deepest member.
• The depth of a list is equal to the depth of its deepest member + 1

The hard part after this is using this information to know when to indent an break lines...

[fwouts]

Hi there 👋 I'm a few years late to the party, but was wondering if you intended to merge that pretty-print branch? I've been looking for a Rust sexp pretty-printer and this looks like exactly what I need :)