#!/bin/sh
# Writes what `forward-word` needs from Emacs to Sources/OrgCore/Semantic/WordSyntax.swift:
# the characters without word syntax in an org-mode buffer, and `char-script-table`.
# Usage: scripts/update-word-syntax.sh [emacs]
set -eu
cd "$(dirname "$0")/.."
emacs=${1:-emacs}
out=Sources/OrgCore/Semantic/WordSyntax.swift
{
  echo "// Generated by scripts/update-word-syntax.sh from $("$emacs" --version | head -1) and Org $("$emacs" -Q --batch --eval '(progn (require (quote org)) (princ (org-version)))')."
  echo
  echo "extension TextStats {"
  echo "    /// Code points without word syntax in org-mode, sorted."
  echo "    static let nonWord: [ClosedRange<UInt32>] = ["
  "$emacs" -Q --batch --eval '
(progn
  (require (quote org))
  (with-temp-buffer
    (org-mode)
    (let ((c 0) start)
      (while (<= c #x110000)
        (let ((word (and (< c #x110000) (eq (char-syntax c) ?w))))
          (cond ((and (not word) (not start) (< c #x110000)) (setq start c))
                ((and (or word (= c #x110000)) start)
                 (princ (format "        0x%X...0x%X,\n" start (1- c)))
                 (setq start nil))))
        (setq c (1+ c))))))'
  echo "    ]"
  echo
  echo "    /// \`char-script-table\`: where each script's range starts, and its script."
  echo "    static let scripts: [(start: UInt32, script: UInt16)] = ["
  "$emacs" -Q --batch --eval '
(let ((c 0) (ids (make-hash-table)) prev)
  (while (< c #x110000)
    (let ((s (aref char-script-table c)))
      (unless (and (> c 0) (eq s prev))
        (unless (gethash s ids) (puthash s (hash-table-count ids) ids))
        (princ (format "        (0x%X, %d),\n" c (gethash s ids)))
        (setq prev s)))
    (setq c (1+ c))))'
  echo "    ]"
  echo "}"
} > "$out"