scripts/update-word-syntax.sh

 1#!/bin/sh
 2# Writes what `forward-word` needs from Emacs to Sources/OrgCore/Semantic/WordSyntax.swift:
 3# the characters without word syntax in an org-mode buffer, and `char-script-table`.
 4# Usage: scripts/update-word-syntax.sh [emacs]
 5set -eu
 6cd "$(dirname "$0")/.."
 7emacs=${1:-emacs}
 8out=Sources/OrgCore/Semantic/WordSyntax.swift
 9{
10  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)))')."
11  echo
12  echo "extension TextStats {"
13  echo "    /// Code points without word syntax in org-mode, sorted."
14  echo "    static let nonWord: [ClosedRange<UInt32>] = ["
15  "$emacs" -Q --batch --eval '
16(progn
17  (require (quote org))
18  (with-temp-buffer
19    (org-mode)
20    (let ((c 0) start)
21      (while (<= c #x110000)
22        (let ((word (and (< c #x110000) (eq (char-syntax c) ?w))))
23          (cond ((and (not word) (not start) (< c #x110000)) (setq start c))
24                ((and (or word (= c #x110000)) start)
25                 (princ (format "        0x%X...0x%X,\n" start (1- c)))
26                 (setq start nil))))
27        (setq c (1+ c))))))'
28  echo "    ]"
29  echo
30  echo "    /// \`char-script-table\`: where each script's range starts, and its script."
31  echo "    static let scripts: [(start: UInt32, script: UInt16)] = ["
32  "$emacs" -Q --batch --eval '
33(let ((c 0) (ids (make-hash-table)) prev)
34  (while (< c #x110000)
35    (let ((s (aref char-script-table c)))
36      (unless (and (> c 0) (eq s prev))
37        (unless (gethash s ids) (puthash s (hash-table-count ids) ids))
38        (princ (format "        (0x%X, %d),\n" c (gethash s ids)))
39        (setq prev s)))
40    (setq c (1+ c))))'
41  echo "    ]"
42  echo "}"
43} > "$out"