From: MORIOKA Tomohiko <tomo@kanji.zinbun.kyoto-u.ac.jp>
Date: Wed, 17 Mar 2010 19:16:02 +0000 (+0900)
Subject: (www-uri-decode-char): Expect char-rep to be encoded as URI.
X-Git-Url: http://git.chise.org/gitweb/?a=commitdiff_plain;h=90bbce1c07115156afb38bfcb8060979fa94e80f;p=chise%2Fest.git

(www-uri-decode-char): Expect char-rep to be encoded as URI.
(www-uri-encode-char): Encode UCS characters as %XX encoding.
(www-format-feature-name-as-CCS): Insert space between prefix (...,
===>, ==>, =>, =>>, =>>>, ..., =) and main-name.
(www-html-display-text): Don't escape "&" except "&GT-".
---

diff --git a/cwiki-common.el b/cwiki-common.el
index 6569ac4..8ef14b7 100644
--- a/cwiki-common.el
+++ b/cwiki-common.el
@@ -141,7 +141,7 @@
 (defun www-uri-decode-char (char-rep)
   (let (ccs cpos)
     (cond
-     ((string-match ":" char-rep)
+     ((string-match "%3A" char-rep)
       (setq ccs (substring char-rep 0 (match-beginning 0))
 	    cpos (substring char-rep (match-end 0)))
       (setq ccs (www-uri-decode-feature-name ccs))
@@ -156,39 +156,47 @@
       (if (numberp cpos)
 	  (decode-char ccs cpos))
       )
-     ((= (length char-rep) 1)
-      (aref char-rep 0)
+     (t
+      (setq char-rep (decode-uri-string char-rep 'utf-8-mcs-er))
+      (when (= (length char-rep) 1)
+	(aref char-rep 0))
       ))))
 
 (defun www-uri-encode-char (char)
-  (let ((ccs-list '(=ucs
-		    =cns11643-1 =cns11643-2 =cns11643-3
-		    =cns11643-4 =cns11643-5 =cns11643-6 =cns11643-7
-		    =gb2312 =gb12345
-		    =jis-x0208 =jis-x0208@1990
-		    =jis-x0212
-		    =cbeta =jef-china3
-		    =jis-x0213-1@2000 =jis-x0213-1@2004
-		    =jis-x0208@1983 =jis-x0208@1978
-		    =zinbun-oracle
-		    =daikanwa
-		    =gt =gt-k
-		    =big5
-		    =big5-cdp
-		    =>>jis-x0208 =>>jis-x0213-1
-		    =>jis-x0208 =>jis-x0213-1))
-	ccs ret)
-    (while (and ccs-list
-		(setq ccs (pop ccs-list))
-		(not (setq ret (encode-char char ccs 'defined-only)))))
-    (cond (ret
-	   (format "%s:0x%X"
-		   (www-uri-encode-feature-name ccs)
-		   ret))
-	  ((setq ccs (car (split-char char)))
-	   (format "%s:0x%X"
-		   (www-uri-encode-feature-name ccs)
-		   (encode-char char ccs))))))
+  (if (encode-char char '=ucs)
+      (mapconcat
+       (lambda (byte)
+	 (format "%%%02X" byte))
+       (encode-coding-string (char-to-string char) 'utf-8-mcs-er)
+       "")
+    (let ((ccs-list '(; =ucs
+		      =cns11643-1 =cns11643-2 =cns11643-3
+		      =cns11643-4 =cns11643-5 =cns11643-6 =cns11643-7
+		      =gb2312 =gb12345
+		      =jis-x0208 =jis-x0208@1990
+		      =jis-x0212
+		      =cbeta =jef-china3
+		      =jis-x0213-1@2000 =jis-x0213-1@2004
+		      =jis-x0208@1983 =jis-x0208@1978
+		      =zinbun-oracle
+		      =daikanwa
+		      =gt =gt-k
+		      =big5
+		      =big5-cdp
+		      =>>jis-x0208 =>>jis-x0213-1
+		      =>jis-x0208 =>jis-x0213-1))
+	  ccs ret)
+      (while (and ccs-list
+		  (setq ccs (pop ccs-list))
+		  (not (setq ret (encode-char char ccs 'defined-only)))))
+      (cond (ret
+	     (format "%s:0x%X"
+		     (www-uri-encode-feature-name ccs)
+		     ret))
+	    ((setq ccs (car (split-char char)))
+	     (format "%s:0x%X"
+		     (www-uri-encode-feature-name ccs)
+		     (encode-char char ccs)))))))
 
 
 ;;; @ Feature name presentation
@@ -214,6 +222,10 @@
 	   (symbol-name feature-name)
 	   "-"))
 	 (dest (upcase (pop rest))))
+    (when (string-match "^=+>*" dest)
+      (setq dest (concat (substring dest 0 (match-end 0))
+			 " "
+			 (substring dest (match-end 0)))))
     (cond
      (rest
       (while (cdr rest)
@@ -581,8 +593,8 @@
 	nil t))
      (encode-coding-region (point-min)(point-max) 'utf-8-mcs-er)
      (goto-char (point-min))
-     (while (search-forward "&" nil t)
-       (replace-match "&amp;" nil t))
+     (while (search-forward "&GT-" nil t)
+       (replace-match "&amp;GT-" nil t))
      (buffer-string))))
 
 (defun www-html-display-paragraph (text)