branch: externals/pyim
commit 6eb5c47b2fcd64944958edd095d4f2c2f18a8526
Author: Feng Shu <tuma...@163.com>
Commit: Feng Shu <tuma...@163.com>
Add pyim-cregexp-build-xingma-regexp-from-words
---
pyim-cregexp.el | 59 +++++++++++++++++++++++++++++------------------------
tests/pyim-tests.el | 5 ++++-
2 files changed, 36 insertions(+), 28 deletions(-)
diff --git a/pyim-cregexp.el b/pyim-cregexp.el
index 095dba6336..d8ff011ec1 100644
--- a/pyim-cregexp.el
+++ b/pyim-cregexp.el
@@ -210,34 +210,39 @@ regexp, 所以搜索单字的时候一般可以搜到生僻字，但搜索句子
(imobj (scheme pyim-scheme-xingma)
&optional match-beginning first-equal _all-equal _char-level-num)
"从 IMOBJ 创建一个搜索中文的 regexp, 适用于形码输入法。"
- (cl-flet ((build-regexp
- (list)
- (let* ((n (apply #'max (mapcar #'length list)))
- results)
- (dotimes (i n)
- (push (format "[%s]%s"
- (mapconcat
- (lambda (x)
- (if (> i (- (length x) 1))
- ""
- (char-to-string
- (elt x i))))
- list "")
- (if (> i 0) "?" ""))
- results))
- (string-join (reverse results)))))
- (let* ((code-prefix (pyim-scheme-code-prefix scheme))
- (regexp (mapconcat
- (lambda (x)
- (let ((code (concat (or code-prefix "")
- (if first-equal
- (substring x 0 1)
- x))))
- (build-regexp (pyim-dcache-get code '(code2word)))))
- imobj "")))
- (unless (equal regexp "")
- (concat (if match-beginning "^" "") regexp)))))
+ (let* ((code-prefix (pyim-scheme-code-prefix scheme))
+ (regexp (mapconcat
+ (lambda (x)
+ (let ((code (concat (or code-prefix "")
+ (if first-equal
+ (substring x 0 1)
+ x))))
+ (pyim-cregexp-build-xingma-regexp-from-words
+ (pyim-dcache-get code '(code2word)))))
+ imobj "")))
+ (unless (equal regexp "")
+ (concat (if match-beginning "^" "") regexp))))
+(defun pyim-cregexp-build-xingma-regexp-from-words (words)
+ "根据 WORDS, 创建一个可以搜索这些 WORDS 的 regexp.
+
+比如：工, 恭恭敬敬 => [工恭][恭]?[敬]?[敬]?
+
+通过 \"[工恭][恭]?[敬]?[敬]?\" 可以搜索 \"工\" 和 \"恭恭敬敬\"."
+ (let ((n (apply #'max (mapcar #'length words)))
+ results)
+ (dotimes (i n)
+ (push (format "[%s]%s"
+ (mapconcat
+ (lambda (x)
+ (if (> i (- (length x) 1))
+ ""
+ (char-to-string
+ (elt x i))))
+ words "")
+ (if (> i 0) "?" ""))
+ results))
+ (string-join (reverse results))))
;; * Footer
(provide 'pyim-cregexp)
diff --git a/tests/pyim-tests.el b/tests/pyim-tests.el
index 7c25ff4385..90cbb6c48b 100644
--- a/tests/pyim-tests.el
+++ b/tests/pyim-tests.el
@@ -975,7 +975,10 @@
(should (equal (pyim-cregexp-build "aaaa'aaaa")
"\\(?:\\(?:aaaa'\\|aaaa\\|[工恭]恭?敬?敬?\\)\\(?:aaaa\\|[工恭]恭?敬?敬?\\)\\)"))
(should (equal (pyim-cregexp-create-1 "aaaa'aaaa" wubi)
-
"\\(?:aaaa'\\|aaaa\\|[工恭][恭]?[敬]?[敬]?\\)\\(?:aaaa\\|[工恭][恭]?[敬]?[敬]?\\)")))
+
"\\(?:aaaa'\\|aaaa\\|[工恭][恭]?[敬]?[敬]?\\)\\(?:aaaa\\|[工恭][恭]?[敬]?[敬]?\\)"))
+ (should (equal (pyim-cregexp-build-xingma-regexp-from-words '("工" "恭恭敬敬"))
+ "[工恭][恭]?[敬]?[敬]?"))
+ )
(with-temp-buffer
(insert "haha nihao")
