aspell-sort-pws: rewrite in python

diff --git a/.bin/aspell-sort-pws b/.bin/aspell-sort-pws
index b5573ed3fde07877e54ba052821a5bf8fb4053c1..24c724351e5c8da069a923be3bf3c1b48f854b63 100755 (executable)
--- a/.bin/aspell-sort-pws
+++ b/.bin/aspell-sort-pws
@@ -1,14 +1,51 @@
-#!/bin/bash
-set -ex
-cd ~/
-file=".aspell.en.pws"
-header=$(head -1 ${file} | sed 's: [0-9]* *$::')
-(
-sed 1d ${file}
-printf '%s\n' "$@"
-) | LC_ALL=en_US sort -u | sed '/^[[:space:]]*$/d' > ${file}.tmp
-(
-echo "${header} $(wc -l ${file}.tmp | awk '{print $1}') "
-cat ${file}.tmp
-) > ${file}
-mv ${file}.tmp ".config/google-chrome/Custom Dictionary.txt"
+#!/usr/bin/env python3
+
+"""Update & sort personal dictionaries."""
+
+import argparse
+import locale
+import os
+import sys
+
+
+def get_parser():
+    """Return a CLI parser."""
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument('words', nargs='*',
+                        help='words to add to the dictionary')
+    return parser
+
+def main(argv):
+    parser = get_parser()
+    opts = parser.parse_args(argv)
+
+    locale.setlocale(locale.LC_COLLATE, 'en_US.UTF8')
+
+    home = os.path.expanduser('~')
+
+    # Pull out the current set of words.
+    pws = os.path.join(home, '.aspell.en.pws')
+    with open(pws, encoding='utf-8') as fp:
+        lines = [x.strip() for x in fp.readlines()]
+    header = lines[0]
+    assert header.startswith('personal_ws-1.1 en ')
+    words = set(lines[1:])
+
+    # Add any words from the user.
+    words.update(opts.words)
+
+    # Update the aspell dict.
+    words = sorted(words, key=locale.strxfrm)
+    header = 'personal_ws-1.1 en %s ' % (len(words),)
+    with open(pws, 'w', encoding='utf-8') as fp:
+        fp.write(header + '\n')
+        fp.write('\n'.join(words) + '\n')
+
+    # Update Chrome's dict.
+    chrome_pws = os.path.join(home, '.config', 'google-chrome', 'Custom Dictionary.txt')
+    with open(chrome_pws, 'w', encoding='utf-8') as fp:
+        fp.write('\n'.join(words) + '\n')
+
+
+if __name__ == '__main__':
+    sys.exit(main(sys.argv[1:]))