guile-uniseg/scripts/generate-graphemes.in

#!@GUILE@ --no-auto-compile
-*- scheme -*-
!#

;; Can be called with a trailing argument pointing to the file on disk.

(use-modules
 (uniseg internal)
 (ice-9 pretty-print)
 (ice-9 peg)
 (ice-9 format)
 (ice-9 exceptions)
 (ice-9 match)
 (ice-9 hash-table)
 (srfi srfi-1))

(define stdout (current-output-port))

(define url
  "https://www.unicode.org/Public/15.0.0/ucd/auxiliary/GraphemeBreakProperty.txt")

(define grapheme-ht (make-hash-table 13))

(define grapheme-properties
  '(hangul-syllable-l
    hangul-syllable-v
    hangul-syllable-lv
    hangul-syllable-lvt
    prepend
    carriage-return
    line-feed
    control
    extend
    regional-indicator
    spacing-mark
    zero-width-joiner))

(define grapheme-symbols
  (map
   (λ (prop) (symbol-with-prefix "char-set:grapheme-" prop))
   grapheme-properties))

(define (string->property str comment)
  (match str
    ("L" 'hangul-syllable-l)
    ("V" 'hangul-syllable-v)
    ("T" 'hangul-syllable-t)
    ("LV" 'hangul-syllable-lv)
    ("LVT" 'hangul-syllable-lvt)
    ("Prepend" 'prepend)
    ("CR" 'carriage-return)
    ("LF" 'line-feed)
    ("Control" 'control)
    ("Extend" 'extend)
    ("Regional_Indicator" 'regional-indicator)
    ("SpacingMark" 'spacing-mark)
    ("ZWJ" 'zero-width-joiner)))

(define file "uniseg/charsets/graphemes.scm")

(format stdout "Writing to ~a...\n" file)

(with-output-to-file file
  (λ ()
    (format #t ";; Code generated by ~a. DO NOT EDIT\n\n" (basename (current-filename)))

    (pretty-print
     `(define-module (uniseg charsets graphemes)
        #:use-module (ice-9 hash-table)
        #:use-module (srfi srfi-1)
        #:use-module (uniseg internal)
        #:use-module (uniseg charsets emoji)
        #:export (,@grapheme-symbols
                  grapheme-charsets)))

    (define-values (process-line print-to-file)
      (make-line-processor
       grapheme-ht
       string->property
       grapheme-properties
       grapheme-symbols
       'grapheme-charsets
       stdout))

    (for-each process-line (cmdline-wget-or-file url stdout))
    (print-to-file)

    ;; Need emoji in the set as well.
    (pretty-print
     `(set! grapheme-charsets
            (cons (list 'extended-pictographic char-set:emoji-extended-pictographic)
                  grapheme-charsets)))

    (display "Code generation complete.\n" stdout)))

(format stdout "Written to ~a.\n" file)