(www-format-value-as-domain-list): New function.
[chise/est.git] / cwiki-common.el
index 0aa7094..b0e1628 100644 (file)
@@ -7,6 +7,9 @@
 (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)
       (let ((i 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))
          (- (lsh value -8) 32)
          (- (logand value 255) 32)))
 
+(defun www-format-value-default (value &optional without-tags)
+  (if (listp value)
+      (mapconcat
+       (lambda (unit)
+        (www-format-encode-string
+         (format "%S" unit)
+         without-tags))
+       value " ")
+    (www-format-encode-string (format "%S" value) without-tags)))
+  
 (defun www-format-value-as-char-list (value &optional without-tags)
   (if (listp value)
       (mapconcat
        value " ")
     (www-format-encode-string (format "%s" value) without-tags)))
 
+(defun www-format-value-as-domain-list (value &optional without-tags)
+  (let (name source0 source num dest rest unit start end ddest)
+    (if (listp value)
+       (if without-tags
+           (mapconcat
+            (lambda (unit)
+              (format "%s" unit))
+            value " ")
+         (setq rest value)
+         (while rest
+           (setq unit (pop rest))
+           (if (symbolp unit)
+               (setq name (symbol-name unit)))
+           (setq dest
+                 (concat
+                  dest
+                  (cond
+                   ((string-match "^zob1968=" name)
+                    (setq source (intern (substring name 0 (match-end 0)))
+                          num (substring name (match-end 0)))
+                    (if (string-match "^\\([0-9]+\\)-\\([0-9]+\\)$" num)
+                        (setq start (string-to-number
+                                     (match-string 1 num))
+                              end (string-to-number
+                                   (match-string 2 num)))
+                      (setq start (string-to-number num)
+                            end start))
+                    (setq ddest
+                          (if (eq source source0)
+                              (format
+                               ", <a href=\"http://chise.zinbun.kyoto-u.ac.jp/koukotsu/rubbings/%04d\">%04d</a>"
+                               start start)
+                            (setq source0 source)
+                            (format
+                             " <a href=\"http://chise.zinbun.kyoto-u.ac.jp/koukotsu/\">%s</a>=<a href=\"http://chise.zinbun.kyoto-u.ac.jp/koukotsu/rubbings/%04d\">%04d</a>"
+                             (www-format-encode-string "\u4EAC大人\u6587研甲\u9AA8")
+                             start start)))
+                    (setq start (1+ start))
+                    (while (<= start end)
+                      (setq ddest
+                            (concat
+                             ddest
+                             (format
+                              ", <a href=\"http://chise.zinbun.kyoto-u.ac.jp/koukotsu/rubbings/%04d\">%04d</a>"
+                              start start)))
+                      (setq start (1+ start)))
+                    ddest)
+                   (t
+                    (setq source unit)
+                    (if (eq source source0)
+                        ""
+                      (setq source0 source)
+                      (concat " " name))
+                    )))))
+         dest)
+      (www-format-encode-string (format "%s" value) without-tags))))
+
 (defun www-format-value-as-ids (value &optional without-tags)
   (if (listp value)
       (mapconcat
              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-as-kangxi-radical (value)
+  (if (and (integerp value)
+          (<= 0 value)
+          (<= value 214))
+      (www-format-encode-string
+       (format "%c" (ideographic-radical value)))
+    (www-format-value-as-S-exp value)))
+
+(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))
          (encode-coding-region (point-min)(point-max) 'utf-8-mcs-er)
        (let ((coded-charset-entity-reference-alist
               (list*
+               '(=gt                   "GT-" 5 d)
                '(=cns11643-1           "C1-" 4 X)
                '(=cns11643-2           "C2-" 4 X)
                '(=cns11643-3           "C3-" 4 X)
                '(=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)
                '(=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)
                     chise-wiki-bitmap-glyphs-url
                     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 "&\\(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 "&\\(G-\\)?GT-K\\([0-9]+\\);" nil t)
+           (setq code (string-to-int (match-string 2)))
+           (replace-match
+            (format "<img alt=\"GT-K%05d\" src=\"%s?char=GT-K%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)
             without-tags))
           ((eq format 'ku-ten)
            (www-format-value-as-kuten value))
+          ((eq format 'kangxi-radical)
+           (www-format-value-as-kangxi-radical value))
           ((eq format 'space-separated-char-list)
            (www-format-value-as-char-list value without-tags))
           ((eq format 'space-separated-ids)
            (www-format-value-as-ids value without-tags))
+          ((eq format 'space-separated-domain-list)
+           (www-format-value-as-domain-list value without-tags))
+          ((eq format 'string)
+           (www-format-encode-string (format "%s" value) without-tags)
+           )
           (t
-           (setq format 'default)
-           (www-format-encode-string
-            (format (www-format-props-to-string props 'default)
-                    value)
-            without-tags))))
+           (www-format-value-default value without-tags)
+            ))
+         )
     (if (or without-tags (eq (plist-get props :mode) 'peek))
        ret
       (format "%s <a href=\"%s?char=%s&feature=%s&format=%s\"
                                      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
    ((null exp) "")
    ((consp exp)
     (cond
-     ((memq (car exp) '(value decimal hex HEX ku-ten S-exp default))
+     ((memq (car exp) '(value decimal hex HEX ku-ten kangxi-radical
+                             S-exp default))
       (if (eq (car exp) 'value)
          (www-format-eval-feature-value char feature-name
                                         (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
 ;;;
 
   (www-html-display-text text)
   (princ "</p>\n"))
 
+
+;;; @ for GlyphWiki
+;;;
+
+(defvar coded-charset-GlyphWiki-id-alist
+  '((=ucs              "u"     4 x nil)
+    (=ucs@JP           "u"     4 x nil)
+    (=ucs@jis          "u"     4 x nil)
+    (=ucs@gb           "u"     4 x "-g")
+    (=ucs@cns          "u"     4 x "-t")
+    (=ucs@ks           "u"     4 x "-k")
+    (=ucs@iso          "u"     4 x "-u")
+    (=ucs@unicode      "u"     4 x "-us")
+    (=adobe-japan1-6   "aj1-"  5 d nil)
+    (=gt               "gt-"   5 d nil)
+    (=big5-cdp         "cdp-"  4 x nil)
+    (=cbeta            "cb"    5 d nil)
+    (=jis-x0208@1978/1pr "j78-"        4 x nil)
+    (=jis-x0208@1978/-4pr "j78-" 4 x nil)
+    (=jis-x0208@1978   "j78-"  4 x nil)
+    (=jis-x0208@1983   "j83-"  4 x nil)
+    (=jis-x0208@1990   "j90-"  4 x nil)
+    (=jis-x0212                "jsp-"  4 x nil)
+    (=jis-x0213-1@2000 "jx1-2000-" 4 x nil)
+    (=jis-x0213-1@2004 "jx1-2004-" 4 x nil)
+    (=jis-x0213-2      "jx2-"  4 x nil)
+    (=cns11643-1       "c1-"   4 x nil)
+    (=cns11643-2       "c2-"   4 x nil)
+    (=cns11643-3       "c3-"   4 x nil)
+    (=cns11643-4       "c4-"   4 x nil)
+    (=cns11643-5       "c5-"   4 x nil)
+    (=cns11643-6       "c6-"   4 x nil)
+    (=cns11643-7       "c7-"   4 x nil)
+    (=daikanwa         "dkw-"  5 d nil)
+    (=gt-k             "gt-k"  5 d nil)
+    (=jef-china3       "jc3-"  4 x nil)
+    (=big5             "b-"    4 x nil)
+    (=ks-x1001         "k0-"   4 x nil)
+    ))
+
+(defun char-GlyphWiki-id (char)
+  (let ((rest coded-charset-GlyphWiki-id-alist)
+       spec ret code)
+    (while (and rest
+               (setq spec (pop rest))
+               (null (setq ret (char-feature char (car spec))))))
+    (when ret
+      (or
+       (and (memq (car spec) '(=ucs@unicode '=ucs@iso))
+           (cond
+            ((and (or (encode-char char '=jis-x0208@1990)
+                      (encode-char char '=jis-x0212)
+                      (encode-char char '=jis-x0213-1))
+                  (setq code (encode-char char '=ucs@jis)))
+             (format "u%04x" code)
+             )
+            ((and (or (encode-char char '=gb2312)
+                      (encode-char char '=gb12345))
+                  (setq code (encode-char char '=ucs@gb)))
+             (format "u%04x-g" code)
+             )
+            ((and (or (encode-char char '=cns11643-1)
+                      (encode-char char '=cns11643-2)
+                      (encode-char char '=cns11643-3)
+                      (encode-char char '=cns11643-4)
+                      (encode-char char '=cns11643-5)
+                      (encode-char char '=cns11643-6)
+                      (encode-char char '=cns11643-7))
+                  (setq code (encode-char char '=ucs@cns)))
+             (format "u%04x-t" code)
+             )
+            ((and (encode-char char '=ks-x1001)
+                  (setq code (encode-char char '=ucs@ks)))
+             (format "u%04x-k" code)
+             )))
+       (format (format "%s%%0%d%s%s"
+                      (nth 1 spec)
+                      (nth 2 spec)
+                      (nth 3 spec)
+                      (or (nth 4 spec) ""))
+              ret)))))
+
+
+;;; @ End.
+;;;
+
 (provide 'cwiki-common)
+
+;;; cwiki-common.el ends here