(www-feature-type): Support `*node' and `*sources'.
[chise/est.git] / cwiki-common.el
index 6569ac4..9681fbe 100644 (file)
@@ -4,8 +4,11 @@
 (defvar chise-wiki-view-url "view.cgi")
 (defvar chise-wiki-edit-url "edit/edit.cgi")
 
-(defvar chise-wiki-glyphs-url
-  "http://chise.zinbun.kyoto-u.ac.jp/glyphs/")
+(defvar chise-wiki-bitmap-glyphs-url
+  "http://chise.zinbun.kyoto-u.ac.jp/glyphs")
+
+(defvar chise-wiki-glyph-cgi-url
+  "http://chise.zinbun.kyoto-u.ac.jp/chisewiki/glyph.cgi")
 
 (defun decode-uri-string (string &optional coding-system)
   (if (> (length string) 0)
   (or (char-feature-property feature-name 'type)
       (let ((str (symbol-name feature-name)))
        (cond
+        ((string-match "\\*note\\(@[^*]+\\)?$" str)
+         'stext)
+        ((string-match "\\*sources\\(@[^*]+\\)?$" str)
+         'domain-list)
+        ((string-match "\\*" str)
+         nil)
         ((string-match "^\\(->\\|<-\\)" str)
          'relation)
         ((string-match "^ideographic-structure\\(@\\|$\\)" str)
        (cond ((eq type 'relation)
               'space-separated-char-list)
              ((eq type 'structure)
-              'space-separated-ids)))
+              'space-separated-ids)
+             ((eq type 'stext)
+              'wiki-text)
+             ))
       (if (find-charset feature-name)
          (if (and (= (charset-dimension feature-name) 2)
                   (= (charset-chars feature-name) 94))
                " (" (decimal) ") <" (ku-ten) ">")
            '("0x" (HEX) " (" (decimal) ")")))))
 
+(defun char-feature-name-at-domain (feature-name domain)
+  (let ((name (symbol-name feature-name)))
+    (cond
+     ((string-match "@[^*]+$" name)
+      (intern (format "%s/%s" name domain))
+      )
+     (t
+      (intern (format "%s@%s" name domain))
+      ))))
+
+(defun char-feature-name-sans-versions (feature)
+  (let ((feature-name (symbol-name feature)))
+    (if (string-match "[@/]\\$rev=latest$" feature-name)
+       (intern (substring feature-name 0 (match-beginning 0)))
+      feature)))
+
+(defun www-char-feature (character feature)
+  (let ((latest-feature (char-feature-name-at-domain feature '$rev=latest)))
+    (mount-char-attribute-table latest-feature)
+    (or (char-feature character latest-feature)
+       (char-feature character feature))))
+
 
 ;;; @ URI representation
 ;;;
 (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))
       (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
+                     =>>jis-x0208 =>>jis-x0213-1
+                     =>jis-x0208 =>jis-x0213-1
+                     =>>gt
+                     =big5
+                     =big5-cdp))
+         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))
+           ((and (setq ccs (car (split-char char)))
+                 (setq ret (encode-char char ccs)))
+            (format "%s:0x%X"
+                    (www-uri-encode-feature-name ccs)
+                    ret))
+           (t
+            (format "system-char-id:0x%X"
+                    (encode-char char 'system-char-id))
+            )))))
 
 
 ;;; @ Feature name presentation
           (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)
       )
      (t dest))))
 
-(defun www-format-feature-name (feature-name &optional lang)
+(defun www-format-feature-name* (feature-name &optional lang)
   (let (name)
-    (www-format-encode-string
-     (cond
-      ((or (and lang
-               (char-feature-property
-                feature-name
-                (intern (format "name@%s" lang))))
-          (char-feature-property
-           feature-name 'name)))
-      ((find-charset feature-name)
-       (www-format-feature-name-as-CCS feature-name))
-      ((and (setq name (symbol-name feature-name))
-           (string-match "^\\(->\\)" name))
-       (www-format-feature-name-as-rel-to feature-name))
-      ((string-match "^\\(<-\\)" name)
-       (www-format-feature-name-as-rel-from feature-name))
-      (t
-       (www-format-feature-name-default feature-name))))))
+    (cond
+     ((or (and lang
+              (char-feature-property
+               feature-name
+               (intern (format "name@%s" lang))))
+         (char-feature-property
+          feature-name 'name)))
+     ((find-charset feature-name)
+      (www-format-feature-name-as-CCS feature-name))
+     ((and (setq name (symbol-name feature-name))
+          (string-match "^\\(->\\)" name))
+      (www-format-feature-name-as-rel-to feature-name))
+     ((string-match "^\\(<-\\)" name)
+      (www-format-feature-name-as-rel-from feature-name))
+     (t
+      (www-format-feature-name-default feature-name)))))
+
+(defun www-format-feature-name (feature-name &optional lang)
+  (www-format-encode-string
+   (www-format-feature-name* feature-name lang)))
 
 
 ;;; @ Feature value presentation
 (defun www-format-encode-string (string &optional without-tags)
   (with-temp-buffer
     (insert string)
-    (let (plane code)
+    (let (plane code start end char variants ret)
       (goto-char (point-min))
       (while (search-forward "<" nil t)
        (replace-match "&lt;" nil t))
                '(=jis-x0208@1990       "J90-" 4 X)
                '(=jis-x0212            "JSP-" 4 X)
                '(=cbeta                "CB" 5 d)
-               '(=jef-china3           "JC3-" 4 X)
                '(=jis-x0208@1997       "J97-" 4 X)
                '(=jis-x0208@1978       "J78-" 4 X)
                '(=jis-x0208@1983       "J83-" 4 X)
+               '(=gt                   "GT-" 5 d)
                '(=zinbun-oracle        "ZOB-" 4 d)
+               '(=jef-china3           "JC3-" 4 X)
                '(=daikanwa             "M-" 5 d)
                coded-charset-entity-reference-alist)))
          (encode-coding-region (point-min)(point-max) 'utf-8-mcs-er)
            (replace-match
             (format "<img alt=\"CB%05d\" src=\"%s/cb-gaiji/%02d/CB%05d.gif\">"
                     code
-                    chise-wiki-glyphs-url
+                    chise-wiki-bitmap-glyphs-url
                     (/ code 1000) code)
             t 'literal))
 
            (replace-match
             (format "<img alt=\"J%s-%04X\" src=\"%s/JIS-%s/%02d-%02d.gif\">"
                     plane code
-                    chise-wiki-glyphs-url
+                    chise-wiki-bitmap-glyphs-url
                     plane
                     (- (lsh code -8) 32)
                     (- (logand code 255) 32))
            (replace-match
             (format "<img alt=\"GB%d-%04X\" src=\"%s/GB%d/%02d-%02d.gif\">"
                     plane code
-                    chise-wiki-glyphs-url
+                    chise-wiki-bitmap-glyphs-url
                     plane
                     (- (lsh code -8) 32)
                     (- (logand code 255) 32))
            (replace-match
             (format "<img alt=\"CNS%d-%04X\" src=\"%s/CNS%d/%04X.gif\">"
                     plane code
-                    chise-wiki-glyphs-url
+                    chise-wiki-bitmap-glyphs-url
                     plane code)
             t 'literal))
-         ))
-      (goto-char (point-min))
-      (while (search-forward "&GT-" nil t)
-       (replace-match "&amp;GT-" t 'literal))
 
+         (goto-char (point-min))
+         (while (re-search-forward "&JC3-\\([0-9A-F][0-9A-F][0-9A-F][0-9A-F]\\);" nil t)
+           (setq code (string-to-int (match-string 1) 16))
+           (replace-match
+            (format "<img alt=\"JC3-%04X\" src=\"http://kanji.zinbun.kyoto-u.ac.jp/db/CHINA3/Gaiji/%04x.gif\">"
+                    code code)
+            t 'literal))
+
+         (goto-char (point-min))
+         (while (re-search-forward "&ZOB-\\([0-9]+\\);" nil t)
+           (setq code (string-to-int (match-string 1)))
+           (replace-match
+            (format "<img alt=\"ZOB-%04d\" src=\"%s/ZOB-1968/%04d.png\">"
+                    code
+                    chise-wiki-bitmap-glyphs-url
+                    code)
+            t 'literal))
+
+         (goto-char (point-min))
+         (while (re-search-forward "&\\(G-\\)?GT-\\([0-9]+\\);" nil t)
+           (setq code (string-to-int (match-string 2)))
+           (replace-match
+            (format "<img alt=\"GT-%05d\" src=\"%s?char=GT-%05d\">"
+                    code
+                    chise-wiki-glyph-cgi-url
+                    code)
+            t 'literal))
+
+         (goto-char (point-min))
+         (while (re-search-forward "&B-\\([0-9A-F]+\\);" nil t)
+           (setq code (string-to-int (match-string 1) 16))
+           (replace-match
+            (format "<img alt=\"B-%04X\" src=\"%s?char=B-%04X\">"
+                    code
+                    chise-wiki-glyph-cgi-url
+                    code)
+            t 'literal))
+
+         (goto-char (point-min))
+         (while (re-search-forward "&CDP-\\([0-9A-F]+\\);" nil t)
+           (setq code (string-to-int (match-string 1) 16))
+           (replace-match
+            (format "<img alt=\"CDP-%04X\" src=\"%s?char=CDP-%04X\">"
+                    code
+                    chise-wiki-glyph-cgi-url
+                    code)
+            t 'literal))
+         
+         (goto-char (point-min))
+         (while (re-search-forward "&UU\\+\\([0-9A-F]+\\);" nil t)
+           (setq code (string-to-int (match-string 1) 16))
+           (replace-match
+            (format "<img alt=\"UU+%04X\" src=\"http://www.unicode.org/cgi-bin/refglyph?24-%04X\">"
+                    code
+                    code)
+            t 'literal))
+
+         (goto-char (point-min))
+         (while (re-search-forward "&MCS-\\([0-9A-F]+\\);" nil t)
+           (setq code (string-to-int (match-string 1) 16))
+           (setq start (match-beginning 0)
+                 end (match-end 0))
+           (setq char (decode-char 'system-char-id code))
+           (setq variants (or (www-char-feature char '->subsumptive)
+                              (www-char-feature char '->denotational)))
+           (while (and variants
+                       (setq ret (www-format-encode-string
+                                  (char-to-string (car variants))))
+                       (string-match "&MCS-\\([0-9A-F]+\\);" ret))
+             (setq variants (cdr variants)))
+           (unless (string-match "&MCS-\\([0-9A-F]+\\);" ret)
+             (goto-char start)
+             (delete-region start end)
+             (insert ret)))
+         ))
+      ;; (goto-char (point-min))
+      ;; (while (search-forward "&GT-" nil t)
+      ;;   (replace-match "&amp;GT-" t 'literal))
       (buffer-string))))
 
 (defun www-format-props-to-string (props &optional format)
           ((eq format 'space-separated-ids)
            (www-format-value-as-ids value without-tags))
           (t
-           (setq format 'default)
+            ;; (setq format 'default)
            (www-format-encode-string
             (format (www-format-props-to-string props 'default)
                     value)
                                      feature-name
                                      &optional format lang uri-char value)
   (unless value
-    (setq value (char-feature char feature-name)))
+    (setq value (www-char-feature char feature-name)))
   (unless format
     (setq format (www-feature-value-format feature-name)))
   (cond
 (defun www-format-eval-unit (exp char feature-name
                                 &optional lang uri-char value)
   (unless value
-    (setq value (char-feature char feature-name)))
+    (setq value (www-char-feature char feature-name)))
   (unless uri-char
     (setq uri-char (www-uri-encode-char char)))
   (cond
        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)