branch: externals/pyim
commit 892cf95d35e3da858e48b731e23c5db13e51a8ed
Author: Feng Shu <tuma...@163.com>
Commit: Feng Shu <tuma...@163.com>
Sort lines when export.
---
pyim-dhashcache.el | 101 +++++++++++++++++++++++++++++++---------------------
tests/pyim-tests.el | 20 +++++++++--
2 files changed, 79 insertions(+), 42 deletions(-)
diff --git a/pyim-dhashcache.el b/pyim-dhashcache.el
index 07ad13ce3f..962a0abd5b 100644
--- a/pyim-dhashcache.el
+++ b/pyim-dhashcache.el
@@ -36,6 +36,7 @@
(require 'cl-lib)
(require 'async nil t)
(require 'pyim-common)
+(require 'pyim-cstring)
(require 'pyim-dcache)
(require 'pyim-dict)
(require 'pyim-scheme)
@@ -716,51 +717,71 @@ pyim 使用的词库文件是简单的文本文件，编码 *强制* 为 \\='utf
如果 CONFIRM 为 non-nil，文件存在时将会提示用户是否覆盖，
默认为覆盖模式"
(with-temp-buffer
- (maphash
- (lambda (key value)
- (let ((value (cl-remove-if
- (lambda (x)
- ;; 如果某个词条的 text 属性 :noexport 设置为 t, 在导出的
- ;; 时候自动忽略这个词条。
- (and (stringp x)
- (get-text-property 0 :noexport x)))
- (if (listp value)
- value
- (list value)))))
- (when value
- (insert (format "%s %s\n" key (mapconcat #'identity value " "))))))
- dcache)
- (sort-lines nil (point-min) (point-max))
- (goto-char (point-min))
- (insert ";;; -*- coding: utf-8-unix -*-\n")
- (pyim-dcache-write-file file confirm)))
+ (let (export-lines)
+ (maphash
+ (lambda (key value)
+ (let ((value (cl-remove-if
+ (lambda (x)
+ ;; 如果某个词条的 text 属性 :noexport 设置为 t, 在导出的
+ ;; 时候自动忽略这个词条。
+ (and (stringp x)
+ (get-text-property 0 :noexport x)))
+ (if (listp value)
+ value
+ (list value)))))
+ (when value
+ (push
+ (format "%s %s\n" key (mapconcat #'identity value " "))
+ export-lines))))
+ dcache)
+ (setq export-lines (sort export-lines #'string<))
+ (goto-char (point-min))
+ (insert ";;; -*- coding: utf-8-unix -*-\n")
+ (dolist (line export-lines)
+ (insert line))
+ (pyim-dcache-write-file file confirm))))
(cl-defmethod pyim-dcache-export-words-and-counts
(file &context ((pyim-dcache-backend) (eql pyim-dhashcache))
&optional confirm ignore-counts)
(with-temp-buffer
- (insert ";;; -*- coding: utf-8-unix -*-\n")
- (maphash
- (lambda (key value)
- (insert
- (if ignore-counts
- (format "%s\n" key)
- (format "%s %s\n" key value))))
- pyim-dhashcache-iword2count)
- ;; 在默认情况下，用户选择过的词生成的缓存中存在的词条，
- ;; `pyim-dhashcache-iword2count' 中也一定存在，但如果用户
- ;; 使用了特殊的方式给用户选择过的词生成的缓存中添加了
- ;; 词条，那么就需要将这些词条也导出，且设置词频为 0
- (maphash
- (lambda (_ words)
- (dolist (word words)
- (unless (gethash word pyim-dhashcache-iword2count)
- (insert
- (if ignore-counts
- (format "%s\n" word)
- (format "%s %s\n" word 0))))))
- pyim-dhashcache-icode2word)
- (pyim-dcache-write-file file confirm)))
+ (let (export-lines)
+ (maphash
+ (lambda (key value)
+ (push
+ (if ignore-counts
+ (format "%s\n" key)
+ (format "%s %s\n" key value))
+ export-lines))
+ pyim-dhashcache-iword2count)
+ ;; 在默认情况下，用户选择过的词生成的缓存中存在的词条，
+ ;; `pyim-dhashcache-iword2count' 中也一定存在，但如果用户
+ ;; 使用了特殊的方式给用户选择过的词生成的缓存中添加了
+ ;; 词条，那么就需要将这些词条也导出，且设置词频为 0
+ (maphash
+ (lambda (_ words)
+ (dolist (word words)
+ (unless (gethash word pyim-dhashcache-iword2count)
+ (push
+ (if ignore-counts
+ (format "%s\n" word)
+ (format "%s %s\n" word 0))
+ export-lines))))
+ pyim-dhashcache-icode2word)
+ (setq export-lines
+ (sort export-lines
+ #'pyim-dhashcache-pinyin-string<))
+ (goto-char (point-min))
+ (insert ";;; -*- coding: utf-8-unix -*-\n")
+ (dolist (line export-lines)
+ (insert line))
+ (pyim-dcache-write-file file confirm))))
+
+(defun pyim-dhashcache-pinyin-string< (a b)
+ "比较 A 和 B 两个字符串的拼音的大小。"
+ (let ((pinyin1 (pyim-cstring-to-pinyin-simple a))
+ (pinyin2 (pyim-cstring-to-pinyin-simple b)))
+ (string< pinyin1 pinyin2)))
;; * Footer
(provide 'pyim-dhashcache)
diff --git a/tests/pyim-tests.el b/tests/pyim-tests.el
index 90cbb6c48b..cba1482c96 100644
--- a/tests/pyim-tests.el
+++ b/tests/pyim-tests.el
@@ -1083,6 +1083,7 @@
(pyim-dhashcache-icode2word (make-hash-table :test #'equal))
(file (pyim-tests-make-temp-file)))
(puthash "你好" 10 pyim-dhashcache-iword2count)
+ (puthash "锕系" 10 pyim-dhashcache-iword2count)
(puthash "尼耗" 1 pyim-dhashcache-iword2count)
(puthash "wo-hao" (list "我好") pyim-dhashcache-icode2word)
(puthash "ni-hao" (list "你好" "尼耗") pyim-dhashcache-icode2word)
@@ -1091,8 +1092,9 @@
(insert-file-contents file)
(should (equal (buffer-string)
";;; -*- coding: utf-8-unix -*-
-你好 10
+锕系 10
尼耗 1
+你好 10
我好 0
")))
(pyim-dcache-export-words-and-counts file nil t)
@@ -1100,8 +1102,9 @@
(insert-file-contents file)
(should (equal (buffer-string)
";;; -*- coding: utf-8-unix -*-
-你好
+锕系
尼耗
+你好
我好
")))
(pyim-dcache-export-personal-words file)
@@ -1160,6 +1163,19 @@ wo-hao 我好
'("/home/user/test3.pyim" "/home/user/test2.pyim")))))
;; ** pyim-dhashcache 相关单元测试
+(ert-deftest pyim-tests-pyim-dhashcache-pinyin-string< ()
+ (should (pyim-dhashcache-pinyin-string< "啊" "波"))
+ (should-not (string< "锕" "波"))
+ (should (pyim-dhashcache-pinyin-string< "锕" "波"))
+ (should-not (pyim-dhashcache-pinyin-string< "波" "啊"))
+ (should (pyim-dhashcache-pinyin-string< "a" "b"))
+ (should-not (pyim-dhashcache-pinyin-string< "b" "a"))
+ (should (pyim-dhashcache-pinyin-string< "aa" "ab"))
+ (should-not (pyim-dhashcache-pinyin-string< "ab" "aa"))
+ (should (pyim-dhashcache-pinyin-string< "你不好" "你好"))
+ (should-not (pyim-dhashcache-pinyin-string< "你好" "你不好"))
+ )
+
(ert-deftest pyim-tests-pyim-dhashcache-get-shortcodes ()
(should (equal (pyim-dhashcache-get-shortcodes ".abcde") nil))
(should (equal (pyim-dhashcache-get-shortcodes "wubi/abcde")
