projects / chise / ids.git / blobdiff
? search:
summary | shortlog | log | commit | commitdiff | tree
raw | inline | side by side
(www-ids-find-version): Update to 0.24.2.
[chise/ids.git] / www / www-ids-find.el
diff --git a/www/www-ids-find.el b/www/www-ids-find.el
index d743695..f3d0c1d 100644 (file)
--- a/www/www-ids-find.el
+++ b/www/www-ids-find.el
@@ -15,7 +15,7 @@
         (concat dest (substring string i))
         coding-system))))
 
-(defconst www-ids-find-version "0.23.2")
+(defconst www-ids-find-version "0.24.2")
 
 (defvar www-ids-find-ideographic-products-file-name
   (expand-file-name "ideographic-products"
@@ -31,69 +31,75 @@
 (defvar www-ids-find-tang-chars-file-name
   "~tomo/projects/chise/ids/www/tang-chars.udd")
 
-(defun www-ids-find-format-line (c is)
+(defun www-ids-find-format-char (c &optional code-desc)
   (let ((str (encode-coding-string (format "%c" c) 'utf-8-er))
-       plane code ucs)
+       plane code)
     (princ
      (with-temp-buffer
        (cond
        ((string-match "&CB\\([0-9]+\\);" str)
         (setq code (string-to-int (match-string 1 str)))
-        (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
+        (insert "<a href=\"/char-desc?char=")
         (insert str)
-        (insert (format "\"><img alt=\"CB%05d\" src=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/glyphs/cb-gaiji/%02d/CB%05d.gif\">\n"
+        (insert (format "\"><img alt=\"CB%05d\" src=\"/glyphs/cb-gaiji/%02d/CB%05d.gif\">\n"
                         code (/ code 1000) code))
-        (insert (format "CB%05d</a>" code))
+        (when code-desc
+          (insert (format "CB%05d</a>" code)))
         )
        ((string-match "&JC3-\\([0-9A-F]+\\);" str)
         (setq code (string-to-int (match-string 1 str) 16))
-        (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
+        (insert "<a href=\"/char-desc?char=")
         (insert str)
         (insert (format "\"><img alt=\"JC3-%04X\" src=\"http://kanji.zinbun.kyoto-u.ac.jp/db/CHINA3/Gaiji/%04x.gif\">\n"
                         code code))
-        (insert (format "JC3-%04X</a>" code))
+        (when code-desc
+          (insert (format "JC3-%04X</a>" code)))
         )
        ((string-match "&J\\(78\\|83\\|90\\|SP\\)-\\([0-9A-F]+\\);" str)
         (setq plane (match-string 1 str)
               code (string-to-int (match-string 2 str) 16))
-        (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
+        (insert "<a href=\"/char-desc?char=")
         (insert str)
-        (insert (format "\"><img alt=\"J%s-%04X\" src=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/glyphs/JIS-%s/%02d-%02d.gif\">\n"
+        (insert (format "\"><img alt=\"J%s-%04X\" src=\"/glyphs/JIS-%s/%02d-%02d.gif\">\n"
                         plane code plane
                         (- (lsh code -8) 32)
                         (- (logand code 255) 32)))
-        (insert (format "J%s-%04X</a>" plane code))
+        (when code-desc
+          (insert (format "J%s-%04X</a>" plane code)))
         )
        ((string-match "&G\\([01]\\)-\\([0-9A-F]+\\);" str)
         (setq plane (string-to-int (match-string 1 str))
               code (string-to-int (match-string 2 str) 16))
-        (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
+        (insert "<a href=\"/char-desc?char=")
         (insert str)
-        (insert (format "\"><img alt=\"G%d-%04X\" src=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/glyphs/GB%d/%02d-%02d.gif\">\n"
+        (insert (format "\"><img alt=\"G%d-%04X\" src=\"/glyphs/GB%d/%02d-%02d.gif\">\n"
                         plane code plane
                         (- (lsh code -8) 32)
                         (- (logand code 255) 32)))
-        (insert (format "G%d-%04X</a>" plane code))
+        (when code-desc
+          (insert (format "G%d-%04X</a>" plane code)))
         )
        ((string-match "&C\\([1-7]\\)-\\([0-9A-F]+\\);" str)
         (setq plane (string-to-int (match-string 1 str))
               code (string-to-int (match-string 2 str) 16))
-        (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
+        (insert "<a href=\"/char-desc?char=")
         (insert str)
-        (insert (format "\"><img alt=\"C%d-%04X\" src=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/glyphs/CNS%d/%04X.gif\">\n"
+        (insert (format "\"><img alt=\"C%d-%04X\" src=\"/glyphs/CNS%d/%04X.gif\">\n"
                         plane code plane code))
-        (insert (format "C%d-%04X</a>" plane code))
+        (when code-desc
+          (insert (format "C%d-%04X</a>" plane code)))
         )
        ((string-match "&ZOB-\\([0-9]+\\);" str)
         (setq code (string-to-int (match-string 1 str)))
-        (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
+        (insert "<a href=\"/char-desc?char=")
         (insert str)
-        (insert (format "\"><img alt=\"ZOB-%04d\" src=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/glyphs/ZOB-1968/%04d.png\">\n"
+        (insert (format "\"><img alt=\"ZOB-%04d\" src=\"/glyphs/ZOB-1968/%04d.png\">\n"
                         code code))
-        (insert (format "ZOB-%04d</a>" code))
+        (when code-desc
+          (insert (format "ZOB-%04d</a>" code)))
         )
        (t
-        (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
+        (insert "<a href=\"/char-desc?char=")
          ;; (insert str)
         (insert
          (mapconcat (lambda (c)
@@ -107,9 +113,12 @@
         ))
        (goto-char (point-min))
        (while (search-forward "&" nil t)
-        (replace-match "&amp;" t 'literal))
-       (buffer-string)
-       ))
+         (replace-match "&amp;" t 'literal))
+       (buffer-string)))))
+  
+(defun www-ids-find-format-line (c is)
+  (let (ucs len i ids)
+    (www-ids-find-format-char c 'code-desc)
     (princ
      (or (if (setq ucs (or (char-ucs c)
                           (encode-char c 'ucs)))
@@ -127,20 +136,12 @@
               www-ids-find-chise-link-map-url-prefix ucs)))
     (princ " ")
     (when is
-      (princ
-       (with-temp-buffer
-        (insert
-         (encode-coding-string
-          (ideographic-structure-to-ids is)
-          'utf-8-jp-er))
-        (goto-char (point-min))
-        (while (re-search-forward "&CB\\([0-9]+\\);" nil t)
-          (setq code (string-to-int (match-string 1)))
-          (replace-match
-           (format "<img alt=\"CB%05d\" src=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/glyphs/cb-gaiji/%02d/CB%05d.gif\">"
-                   code (/ code 1000) code)
-           t 'literal))
-        (buffer-string))))
+      (setq ids (ideographic-structure-to-ids is))
+      (setq i 0
+           len (length ids))
+      (while (< i len)
+       (www-ids-find-format-char (aref ids i))
+       (setq i (1+ i))))
     (when (and ucs
               (with-current-buffer
                   (find-file-noselect
@@ -161,16 +162,18 @@
 (defun www-ids-insert-chars-including-components (components
                                                  &optional ignored-chars)
   (let ((products (copy-list (ideographic-products-find components)))
-       is as bs) 
+       is as bs len ignore-children)
+    (setq len (length products))
     (dolist (c (cond
-               ((> (length products) 10000)
+               ((> len 8192)
+                (setq ignore-children t)
                 products)
-               ((> (length products) 4096)
+               ((> len 4096)
                 (sort products
                       (lambda (a b)
                         (< (char-int a)(char-int b))))
                 )
-               ((> (length products) 512)
+               ((> len 512)
                 (sort products
                       (lambda (a b)
                         (if (setq as (char-total-strokes a))
@@ -196,12 +199,13 @@
        (setq is (char-feature c 'ideographic-structure))
        (princ "<li>")
        (www-ids-find-format-line c is)
-       (princ "<ul>\n")
-       (setq ignored-chars
-             (www-ids-insert-chars-including-components
-              (char-to-string c)
-              (cons c ignored-chars)))
-       (princ "</ul>\n")
+       (unless ignore-children
+         (princ "<ul>\n")
+         (setq ignored-chars
+               (www-ids-insert-chars-including-components
+                (char-to-string c)
+                (cons c ignored-chars)))
+         (princ "</ul>\n"))
        )
       ))
   ignored-chars)
@@ -235,7 +239,7 @@
          (setq components (substring components (match-end 0))))
       (setq components
            (if (> (length components) 0)
-               (decode-url-string components 'utf-8-jp-er)
+               (decode-url-string components 'utf-8-er)
              nil))
       )
      (t
@@ -266,12 +270,12 @@
     (princ "
 <hr>
 <p>
-<form action=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/ids-find\" method=\"GET\">
+<form action=\"/ids-find\" method=\"GET\">
 ")
     (princ (encode-coding-string "\e$BItIJJ8;zNs\e(B" 'utf-8-jp-er))
     (princ " <input type=\"text\" name=\"components\" size=\"30\" maxlength=\"30\" value=\"")
     (if (> (length components) 0)
-       (princ (encode-coding-string components 'utf-8-jp-er)))
+       (princ (encode-coding-string components 'utf-8-er)))
     (princ "\">
 <input type=\"submit\" value=\"")
     (princ (encode-coding-string "\e$B8!:w3+;O\e(B" 'utf-8-jp-er))
@@ -360,7 +364,7 @@ href=\"http://www.shuiren.org/\">\e$B?g?MDb\e(B</a>\e$B!K$K$h$k2r@b\e(B
       ))
     (princ "<hr>")
     (princ "<p>
-Copyright (C) 2005, 2006, 2007 <a href=\"http://kanji.zinbun.kyoto-u.ac.jp/~tomo/\"
+Copyright (C) 2005, 2006, 2007, 2008, 2009 <a href=\"http://kanji.zinbun.kyoto-u.ac.jp/~tomo/\"
 >MORIOKA Tomohiko</a>")
     (princ
      (format