(www-xml-format-unit): Escape `&'.
[chise/est.git] / cwiki-common.el
index 5a8cfda..4a30be9 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
 ;;;
                      =gt =gt-k
                      =>>jis-x0208 =>>jis-x0213-1
                      =>jis-x0208 =>jis-x0213-1
+                     =>>gt
                      =big5
                      =big5-cdp))
          ccs ret)
     "-")
    " "))
 
+(defun www-format-feature-name-as-metadata (feature-name &optional lang)
+  (let ((str (symbol-name feature-name))
+       base meta)
+    (cond
+     ((string-match "\\*[^*]+$" str)
+      (setq base (substring str 0 (match-beginning 0))
+           meta (substring str (match-beginning 0)))
+      (concat (www-format-feature-name* (intern base) lang)
+             meta))
+     (t
+      (www-format-feature-name-default feature-name)
+      ))))
+
 (defun www-format-feature-name-as-rel-to (feature-name)
   (concat "\u2192" (substring (symbol-name feature-name) 2)))
 
      ((find-charset feature-name)
       (www-format-feature-name-as-CCS feature-name))
      ((and (setq name (symbol-name feature-name))
-          (string-match "^\\(->\\)" name))
+          (string-match "\\*" name))
+      (www-format-feature-name-as-metadata feature-name lang))
+     ((string-match "^\\(->\\)" name)
       (www-format-feature-name-as-rel-to feature-name))
      ((string-match "^\\(<-\\)" name)
       (www-format-feature-name-as-rel-from feature-name))
              value)
     (www-format-value-as-S-exp value)))
 
-(defun www-format-value (value &optional feature-name format without-tags)
-  ;; (cond
-  ;;  ((find-charset feature-name)
-  ;;   (cond
-  ;;    ((and (= (charset-chars feature-name) 94)
-  ;;          (= (charset-dimension feature-name) 2))
-  ;;     (www-format-value-as-CCS-94x94 value))
-  ;;    (t
-  ;;     (www-format-value-as-CCS-default value)))
-  ;;   )
-  ;;  (t
-  ;;   (www-format-value-as-S-exp value)))
-  (www-format-apply-value format nil value nil nil without-tags)
+(defun www-format-value (object feature-name
+                               &optional value format without-tags)
+  (unless value
+    (setq value (www-char-feature object feature-name)))
+  (www-format-apply-value object feature-name
+                         format nil value nil nil without-tags)
   )
 
 
 (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\\+\\|U-\\)\\([0-9A-F]+\\);" nil t)
+           (setq code (string-to-int (match-string 2) 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 'S-exp) "S")
           (t "s"))))      
 
-(defun www-format-apply-value (format props value
+(defun www-format-apply-value (object feature-name
+                                     format props value
                                      &optional uri-char uri-feature
                                      without-tags)
   (let (ret)
               (format "%s" value)
               without-tags))
            )
+          ((eq format 'wiki-text)
+           (if without-tags
+               (www-xml-format-list value)
+             (www-format-eval-list value object feature-name nil uri-char))
+           )
           ((eq format 'S-exp)
            (www-format-encode-string
             (format (www-format-props-to-string props 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
    ((symbolp format)
     (www-format-apply-value
+     char feature-name
      format nil value
      uri-char (www-uri-encode-feature-name feature-name))
     )
     (cond ((null (cdr format))
           (setq format (car format))
           (www-format-apply-value
+           char feature-name
            (car format) (nth 1 format) value
            uri-char (www-uri-encode-feature-name feature-name))
           )
 (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
                                         (plist-get (nth 1 exp) :format)
                                         lang uri-char value)
        (www-format-apply-value
+        char feature-name
         (car exp) (nth 1 exp) value
         uri-char (www-uri-encode-feature-name feature-name)))
       )
     (www-format-eval-unit format-list char feature-name lang uri-char)))
 
 
+;;; @ XML generator
+;;;
+
+(defun www-xml-format-props (props)
+  (let ((dest "")
+       key val)
+    (while props
+      (setq key (pop props)
+           val (pop props))
+      (if (symbolp key)
+         (setq key (symbol-name key)))
+      (if (eq (aref key 0) ?:)
+         (setq key (substring key 1)))
+      (setq dest
+           (format "%s %s=\"%s\""
+                   dest key
+                   (www-format-encode-string
+                    (format "%s" val) 'without-tags))))
+    dest))
+
+(defun www-xml-format-unit (format-unit)
+  (let (name props children ret)
+    (cond
+     ((stringp format-unit)
+      (mapconcat (lambda (c)
+                  (cond
+                   ((eq c ?&) "&amp;")
+                    ;; ((eq c ?<) "&amp;lt;")
+                    ;; ((eq c ?>) "&amp;gt;")
+                   (t
+                    (char-to-string c))))
+                (www-format-encode-string format-unit 'without-tags)
+                "")
+      )
+     ((consp format-unit)
+      (setq name (car format-unit)
+           props (nth 1 format-unit)
+           children (nthcdr 2 format-unit))
+      (when (eq name 'link)
+       (setq ret (plist-get props :ref))
+       (unless (stringp ret)
+         (setq props (plist-remprop (copy-list props) :ref))
+         (setq children
+               (cons (list* 'ref nil ret)
+                     children))))
+      (if children
+         (format "<%s%s>%s</%s>"
+                 name
+                 (if props
+                     (www-xml-format-props props)
+                   "")
+                 (www-xml-format-list children)
+                 name)
+       (format "<%s%s/>"
+               name (www-xml-format-props props)))
+      )
+     (t
+      (format "%s" format-unit)))))
+
+(defun www-xml-format-list (format-list)
+  (if (atom format-list)
+      (www-xml-format-unit format-list)
+    (mapconcat #'www-xml-format-unit
+              format-list "")))
+
+
 ;;; @ HTML generator
 ;;;