[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[elpa] externals/pyim 0b73690671 2/4: 优化 pyim-candidates-get-chief 的速度。
From: |
ELPA Syncer |
Subject: |
[elpa] externals/pyim 0b73690671 2/4: 优化 pyim-candidates-get-chief 的速度。 |
Date: |
Thu, 6 Jan 2022 22:57:48 -0500 (EST) |
branch: externals/pyim
commit 0b73690671cbc59291f35dc183dfed14d3799503
Author: Feng Shu <tumashu@163.com>
Commit: Feng Shu <tumashu@163.com>
优化 pyim-candidates-get-chief 的速度。
* pyim-process.el (pyim-process-create-word): simplify.
* pyim-dhashcache.el (pyim-dhashcache-iword2count-recent1)
(pyim-dhashcache-iword2count-recent2): new variable.
(pyim-dhashcache-init-count-variables)
(pyim-dhashcache-save-personal-dcache-to-file): Use the above two
new variables.
(pyim-dhashcache-update-iword2count-recent): new function.
(pyim-dhashcache-update-iword2count): handle the above two
variables.
* pyim-candidates.el (pyim-candidates-possible-chiefs)
(pyim-candidates-add-possible-chief): removed.
(pyim-candidates-get-chief): simplify.
---
pyim-candidates.el | 64 ++++++++++++++++++++----------------------------------
pyim-dhashcache.el | 29 ++++++++++++++++++++++++-
pyim-process.el | 1 -
3 files changed, 52 insertions(+), 42 deletions(-)
diff --git a/pyim-candidates.el b/pyim-candidates.el
index 639ecd67f5..2d93cbb95a 100644
--- a/pyim-candidates.el
+++ b/pyim-candidates.el
@@ -52,9 +52,6 @@
细节信息请参考 `pyim-page-refresh' 的 docstring.")
-(defvar pyim-candidates-possible-chiefs nil
- "可能做第一位候选词的词条列表。")
-
(pyim-register-local-variables
'(pyim-candidates pyim-candidate-position))
@@ -72,45 +69,32 @@ IMOBJS 获得候选词条。"
(funcall (intern (format "pyim-candidates-create:%S" class))
imobjs scheme-name async)))))
-(defun pyim-candidates-add-possible-chief (word)
- "将 WORD 添加到 `pyim-candidates-possible-chiefs'."
- (push word pyim-candidates-possible-chiefs)
- (setq pyim-candidates-possible-chiefs
- (cl-subseq pyim-candidates-possible-chiefs 0
- (min 100 (length pyim-candidates-possible-chiefs)))))
-
(defun pyim-candidates-get-chief (scheme-name &optional personal-words
common-words)
"选取第一位候选词。"
- (let* ((class (pyim-scheme-get-option scheme-name :class))
- (words pyim-candidates-possible-chiefs)
- (length (length words))
- ;; NOTE: 网上传言,一段话平均70个字,按照一个词两个字估算,100个词大概
- ;; 为两段话。
- (words100 (cl-subseq words 0 (min 100 length)))
- ;; NOTE: 10个词大概1句话。
- (words10 (cl-subseq words 0 (min 10 length))))
- (cond
- ((equal class 'xingma)
- (or
- ;; 如果从公共词库里面获取到的第一个词条是汉字,就选择它。
- (when (= (length (car common-words)) 1)
- (car common-words))
- ;; 从个人词库里面按排列的先后顺序,获取一个汉字。
- (cl-find-if
- (lambda (word)
- (= (length word) 1))
- personal-words)))
- (t (or
- ;; 最近输入的10个词中出现一次以上。
- (cl-find-if (lambda (word)
- (> (cl-count word words10 :test #'equal) 1))
- personal-words)
- ;; 最近输入的100个词中出现过三次以上。
- (cl-find-if (lambda (word)
- (> (cl-count word words100 :test #'equal) 3))
- personal-words)
- ;; 个人词条中的第一个词。
- (car personal-words))))))
+ (let ((class (pyim-scheme-get-option scheme-name :class)))
+ (cond ((equal class 'xingma)
+ (or
+ ;; 如果从公共词库里面获取到的第一个词条是汉字,就选择它。
+ (when (= (length (car common-words)) 1)
+ (car common-words))
+ ;; 从个人词库里面按排列的先后顺序,获取一个汉字。
+ (cl-find-if
+ (lambda (word)
+ (= (length word) 1))
+ personal-words)))
+ (t (or
+ ;; 最近输入的10个不同的词中出现一次以上。
+ (cl-find-if
+ (lambda (word)
+ (> (or (car (pyim-dcache-get word 'iword2count-recent1)) 0)
1))
+ personal-words)
+ ;; 最近输入的50个不同的词中出现过三次以上。
+ (cl-find-if
+ (lambda (word)
+ (> (or (car (pyim-dcache-get word 'iword2count-recent2)) 0)
3))
+ personal-words)
+ ;; 个人词条中的第一个词。
+ (car personal-words))))))
(defun pyim-candidates-create:xingma (imobjs scheme-name &optional async)
"`pyim-candidates-create' 处理五笔仓颉等形码输入法的函数."
diff --git a/pyim-dhashcache.el b/pyim-dhashcache.el
index 1f3796403a..83ee86f071 100644
--- a/pyim-dhashcache.el
+++ b/pyim-dhashcache.el
@@ -59,6 +59,8 @@
(defvar pyim-dhashcache-word2code nil)
(defvar pyim-dhashcache-iword2count nil)
(defvar pyim-dhashcache-iword2count-log nil)
+(defvar pyim-dhashcache-iword2count-recent1 nil)
+(defvar pyim-dhashcache-iword2count-recent2 nil)
(defvar pyim-dhashcache-shortcode2word nil)
(defvar pyim-dhashcache-icode2word nil)
(defvar pyim-dhashcache-ishortcode2word nil)
@@ -469,7 +471,9 @@ code 对应的中文词条了。
(defun pyim-dhashcache-init-count-variables ()
"初始化 count 相关的变量。"
(pyim-dcache-init-variable pyim-dhashcache-iword2count)
- (pyim-dcache-init-variable pyim-dhashcache-iword2count-log))
+ (pyim-dcache-init-variable pyim-dhashcache-iword2count-log)
+ (pyim-dcache-init-variable pyim-dhashcache-iword2count-recent1)
+ (pyim-dcache-init-variable pyim-dhashcache-iword2count-recent2))
(defun pyim-dhashcache-save-personal-dcache-to-file ()
;; 用户选择过的词
@@ -501,8 +505,31 @@ code 对应的中文词条了。
(setq ,new-value (progn ,@body))
(puthash ,key ,new-value ,table))))
+(defun pyim-dhashcache-update-iword2count-recent (word n hash-table)
+ (let (words-need-remove)
+ (pyim-dhashcache-put
+ hash-table :all-words
+ (setq orig-value (remove word orig-value))
+ (push word orig-value)
+ (if (<= (length orig-value) n)
+ orig-value
+ (setq words-need-remove (nthcdr n orig-value))
+ (cl-subseq orig-value 0 n)))
+ (dolist (w words-need-remove)
+ (remhash w hash-table))
+ (pyim-dhashcache-put
+ hash-table word
+ (+ (or orig-value 0) 1))
+ hash-table))
+
(defun pyim-dhashcache-update-iword2count (word &optional wordcount-handler)
"保存词频到缓存."
+ (setq pyim-dhashcache-iword2count-recent1
+ (pyim-dhashcache-update-iword2count-recent
+ word 10 pyim-dhashcache-iword2count-recent1))
+ (setq pyim-dhashcache-iword2count-recent2
+ (pyim-dhashcache-update-iword2count-recent
+ word 50 pyim-dhashcache-iword2count-recent2))
(pyim-dhashcache-put
pyim-dhashcache-iword2count word
(cond
diff --git a/pyim-process.el b/pyim-process.el
index 5bbcd6f6bc..7eb1131441 100644
--- a/pyim-process.el
+++ b/pyim-process.el
@@ -575,7 +575,6 @@ BUG:拼音无法有效地处理多音字。"
(codes (pyim-cstring-to-codes
word scheme-name
(or criteria pyim-cstring-to-code-criteria))))
- (pyim-candidates-add-possible-chief word)
;; 保存对应词条的词频
(when (> (length word) 0)
(pyim-dcache-update-wordcount word (or wordcount-handler #'1+)))