diff options
author | Zhangyuan Nie <yuan@znie.org> | 2022-08-05 03:27:32 -0400 |
---|---|---|
committer | Peng Wu <alexepico@gmail.com> | 2022-08-10 14:12:52 +0800 |
commit | fa548f562a97a4662c5910bb98b75eb3e3334e2e (patch) | |
tree | 737e6782d9ec7f09e60c5e6d16689cec0891aca2 | |
parent | 92f3d9e0e99098cbf8c1923eed29ec6d203fb38b (diff) | |
download | ibus-libpinyin-fa548f562a97a4662c5910bb98b75eb3e3334e2e.tar.gz |
migrate create_valid_hanzi.py to python 3
-rw-r--r-- | data/db/android/README | 2 | ||||
-rw-r--r-- | data/db/android/create_valid_hanzi.py | 20 | ||||
-rw-r--r-- | data/db/android/valid_hanzi.py | 1 |
3 files changed, 8 insertions, 15 deletions
diff --git a/data/db/android/README b/data/db/android/README index b7dbaab..459bf20 100644 --- a/data/db/android/README +++ b/data/db/android/README @@ -1,2 +1,2 @@ The data is from android open source project. -http://android.git.kernel.org/?p=platform/packages/inputmethods/PinyinIME.git +https://android.googlesource.com/platform/packages/inputmethods/PinyinIME diff --git a/data/db/android/create_valid_hanzi.py b/data/db/android/create_valid_hanzi.py index 52c473a..51b8d6d 100644 --- a/data/db/android/create_valid_hanzi.py +++ b/data/db/android/create_valid_hanzi.py @@ -1,17 +1,11 @@ - def get_validate_hanzi(): - validate_hanzi = file("valid_utf16.txt").read().decode("utf16") - return set(validate_hanzi) + with open("valid_utf16.txt", encoding="utf16") as file: + return set(file.read()) -def main(): - hanzi = get_validate_hanzi() - hanzi = list(hanzi) - hanzi.sort() - print "# -*- coding: utf-8 -*- " - print "valid_hanzi = set([" - for c in hanzi: - print " u\"%s\"," % c.encode("utf8") - print "])" if __name__ == "__main__": - main() + hanzi = get_validate_hanzi() + print("valid_hanzi = set([") + for c in sorted(hanzi): + print(f' u"{c}",') + print("])") diff --git a/data/db/android/valid_hanzi.py b/data/db/android/valid_hanzi.py index e8ddc74..9f32982 100644 --- a/data/db/android/valid_hanzi.py +++ b/data/db/android/valid_hanzi.py @@ -1,4 +1,3 @@ -# -*- coding: utf-8 -*- valid_hanzi = set([ u"〇", u"一", |