www/www-ids-find.el

   1 (require 'ids-find)
   2
   3 (defun decode-url-string (string &optional coding-system)
   4   (if (> (length string) 0)
   5       (let ((i 0)
   6             dest)
   7         (while (string-match "%\\([0-9A-F][0-9A-F]\\)" string i)
   8           (setq dest (concat dest
   9                              (substring string i (match-beginning 0))
  10                              (char-to-string
  11                               (int-char
  12                                (string-to-int (match-string 1 string) 16))))
  13                 i (match-end 0)))
  14         (decode-coding-string
  15          (concat dest (substring string i))
  16          coding-system))))
  17
  18 (defconst www-ids-find-version "0.22.1")
  19
  20 (defvar www-ids-find-ideographic-products-file-name
  21   (expand-file-name "ideographic-products"
  22                     (expand-file-name
  23                      "feature"
  24                      (expand-file-name
  25                       "character"
  26                       chise-system-db-directory))))
  27
  28 (defvar www-ids-find-tang-chars-file-name
  29   "~tomo/projects/chise/ids/www/tang-chars.udd")
  30
  31 (defun www-ids-find-format-line (c is)
  32   (let ((str (encode-coding-string (format "%c" c) 'utf-8-er))
  33         plane code ucs)
  34     (princ
  35      (with-temp-buffer
  36        (cond
  37         ((string-match "&CB\\([0-9]+\\);" str)
  38          (setq code (string-to-int (match-string 1 str)))
  39          (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
  40          (insert str)
  41          (insert (format "\"><img alt=\"CB%05d\" src=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/glyphs/cb-gaiji/%02d/CB%05d.gif\">\n"
  42                          code (/ code 1000) code))
  43          (insert (format "CB%05d</a>" code))
  44          )
  45         ((string-match "&JC3-\\([0-9A-F]+\\);" str)
  46          (setq code (string-to-int (match-string 1 str) 16))
  47          (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
  48          (insert str)
  49          (insert (format "\"><img alt=\"JC3-%04X\" src=\"http://kanji.zinbun.kyoto-u.ac.jp/db/CHINA3/Gaiji/%04x.gif\">\n"
  50                          code code))
  51          (insert (format "JC3-%04X</a>" code))
  52          )
  53         ((string-match "&J\\(78\\|83\\|90\\|SP\\)-\\([0-9A-F]+\\);" str)
  54          (setq plane (match-string 1 str)
  55                code (string-to-int (match-string 2 str) 16))
  56          (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
  57          (insert str)
  58          (insert (format "\"><img alt=\"J%s-%04X\" src=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/glyphs/JIS-%s/%02d-%02d.gif\">\n"
  59                          plane code plane
  60                          (- (lsh code -8) 32)
  61                          (- (logand code 255) 32)))
  62          (insert (format "J%s-%04X</a>" plane code))
  63          )
  64         ((string-match "&G\\([01]\\)-\\([0-9A-F]+\\);" str)
  65          (setq plane (string-to-int (match-string 1 str))
  66                code (string-to-int (match-string 2 str) 16))
  67          (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
  68          (insert str)
  69          (insert (format "\"><img alt=\"G%d-%04X\" src=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/glyphs/GB%d/%02d-%02d.gif\">\n"
  70                          plane code plane
  71                          (- (lsh code -8) 32)
  72                          (- (logand code 255) 32)))
  73          (insert (format "G%d-%04X</a>" plane code))
  74          )
  75         ((string-match "&C\\([1-7]\\)-\\([0-9A-F]+\\);" str)
  76          (setq plane (string-to-int (match-string 1 str))
  77                code (string-to-int (match-string 2 str) 16))
  78          (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
  79          (insert str)
  80          (insert (format "\"><img alt=\"C%d-%04X\" src=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/glyphs/CNS%d/%04X.gif\">\n"
  81                          plane code plane code))
  82          (insert (format "C%d-%04X</a>" plane code))
  83          )
  84         ((string-match "&ZOB-\\([0-9]+\\);" str)
  85          (setq code (string-to-int (match-string 1 str)))
  86          (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
  87          (insert str)
  88          (insert (format "\"><img alt=\"ZOB-%04d\" src=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/glyphs/ZOB-1968/%04d.png\">\n"
  89                          code code))
  90          (insert (format "ZOB-%04d</a>" code))
  91          )
  92         (t
  93          (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
  94          ;; (insert str)
  95          (insert
  96           (mapconcat (lambda (c)
  97                        (if (<= (char-int c) #x7F)
  98                            (char-to-string c)
  99                          (format "%%%02X" c)))
 100                      str ""))
 101          (insert "\">")
 102          (insert str)
 103          (insert "</a>")
 104          ))
 105        (goto-char (point-min))
 106        (while (search-forward "&" nil t)
 107          (replace-match "&amp;" t 'literal))
 108        (buffer-string)
 109        ))
 110     (princ
 111      (or (if (setq ucs (or (char-ucs c)
 112                            (encode-char c 'ucs)))
 113              (format " <a href=\"http://www.unicode.org/cgi-bin/GetUnihanData.pl?codepoint=%X\">%s</a>"
 114                      ucs
 115                      (cond ((<= ucs #xFFFF)
 116                             (format "U+%04X" ucs))
 117                            ((<= ucs #x10FFFF)
 118                             (format "U-%08X" ucs))))
 119            "          ")))
 120     (princ " ")
 121     (when is
 122       (princ
 123        (with-temp-buffer
 124          (insert
 125           (encode-coding-string
 126            (ideographic-structure-to-ids is)
 127            'utf-8-jp-er))
 128          (goto-char (point-min))
 129          (while (re-search-forward "&CB\\([0-9]+\\);" nil t)
 130            (setq code (string-to-int (match-string 1)))
 131            (replace-match
 132             (format "<img alt=\"CB%05d\" src=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/glyphs/cb-gaiji/%02d/CB%05d.gif\">"
 133                     code (/ code 1000) code)
 134             t 'literal))
 135          (buffer-string))))
 136     (when (and ucs
 137                (with-current-buffer
 138                    (find-file-noselect
 139                     www-ids-find-tang-chars-file-name)
 140                  (goto-char (point-min))
 141                  (re-search-forward (format "^%d$" ucs) nil t)))
 142       (princ
 143        (format " <a href=\"http://coe21.zinbun.kyoto-u.ac.jp/djvuchar?query=%s\">"
 144                (mapconcat
 145                 (lambda (c)
 146                   (format "%%%02X" (char-int c)))
 147                 (encode-coding-string (char-to-string c)
 148                                       'utf-8-jp)
 149                 "")))
 150       (princ (encode-coding-string "\e$B"M\e(B[\e$BEbBeBsK\\e(B]</a>" 'utf-8-jp-er)))
 151     (princ "<br>\n")))
 152
 153 (defun www-ids-insert-chars-including-components (components)
 154   (let (is)
 155     (dolist (c (ideographic-products-find components))
 156       (setq is (char-feature c 'ideographic-structure))
 157       ;; to avoid problems caused by wrong indexes
 158       (when (every (lambda (cc)
 159                      (ideographic-structure-member cc is))
 160                    components)
 161         (princ "<li>")
 162         (www-ids-find-format-line c is)
 163         (princ "<ul>\n")
 164         (www-ids-insert-chars-including-components (char-to-string c))
 165         (princ "</ul>\n")
 166         )
 167       )))
 168
 169 (defun www-batch-ids-find ()
 170   (let ((components (car command-line-args-left))
 171         (coded-charset-entity-reference-alist
 172          (list*
 173           '(=cns11643-1         "C1-" 4 X)
 174           '(=cns11643-2         "C2-" 4 X)
 175           '(=cns11643-3         "C3-" 4 X)
 176           '(=cns11643-4         "C4-" 4 X)
 177           '(=cns11643-5         "C5-" 4 X)
 178           '(=cns11643-6         "C6-" 4 X)
 179           '(=cns11643-7         "C7-" 4 X)
 180           '(=gb2312             "G0-" 4 X)
 181           '(=gb12345            "G1-" 4 X)
 182           '(=jis-x0208@1990     "J90-" 4 X)
 183           '(=jis-x0212          "JSP-" 4 X)
 184           '(=cbeta              "CB" 5 d)
 185           '(=jef-china3         "JC3-" 4 X)
 186           '(=jis-x0208@1978     "J78-" 4 X)
 187           '(=jis-x0208@1983     "J83-" 4 X)
 188           '(=daikanwa           "M-" 5 d)
 189           coded-charset-entity-reference-alist))
 190         )
 191     (setq command-line-args-left (cdr command-line-args-left))
 192     (cond
 193      ((stringp components)
 194       (if (string-match "^components=" components)
 195           (setq components (substring components (match-end 0))))
 196       (setq components
 197             (if (> (length components) 0)
 198                 (decode-url-string components 'utf-8-jp-er)
 199               nil))
 200       )
 201      (t
 202       (setq components nil)
 203       ))
 204     (princ "Content-Type: text/html; charset=UTF-8
 205
 206 <!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\"
 207             \"http://www.w3.org/TR/html4/loose.dtd\">
 208 <html lang=\"ja\">
 209 <head>
 210 <title>CHISE IDS Find</title>
 211 </head>
 212
 213 <body>
 214
 215 <h1>")
 216     (princ (encode-coding-string "CHISE IDS \e$B4A;z8!:w\e(B" 'utf-8-jp-er))
 217     (princ "</h1>")
 218     (princ "
 219 <p>Version ")
 220     (princ www-ids-find-version)
 221     (princ (format-time-string
 222             " (Last-modified: %Y-%m-%d %H:%M:%S)"
 223             (nth 5
 224                  (file-attributes
 225                   www-ids-find-ideographic-products-file-name))))
 226     (princ "
 227 <p>
 228 Copyright (C) 2005 <a href=\"http://kanji.zinbun.kyoto-u.ac.jp/~tomo/\"
 229 >MORIOKA Tomohiko</a>
 230 <hr>
 231 <p>
 232 <form action=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/ids-find\" method=\"GET\">
 233 ")
 234     (princ (encode-coding-string "\e$BItIJJ8;zNs\e(B" 'utf-8-jp-er))
 235     (princ " <input type=\"text\" name=\"components\" size=\"30\" maxlength=\"30\" value=\"")
 236     (if (> (length components) 0)
 237         (princ (encode-coding-string components 'utf-8-jp-er)))
 238     (princ "\">
 239 <input type=\"submit\" value=\"")
 240     (princ (encode-coding-string "\e$B8!:w3+;O\e(B" 'utf-8-jp-er))
 241     (princ "\">
 242 </form>
 243
 244 ")
 245     (unless (file-newer-than-file-p
 246              www-ids-find-ideographic-products-file-name
 247              (locate-file (car command-line-args) exec-path))
 248       (princ (encode-coding-string "<hr>
 249 <p>
 250 \e$B8=:_!"%7%9%F%`$N99?7:n6HCf$G$9!#$7$P$i$/$*BT$A$/$@$5$$!#\e(B
 251 <hr>
 252 " 'utf-8-jp-er))
 253       ;; (setq components nil)
 254       )
 255     (cond
 256      (components
 257       ;; (map-char-attribute
 258       ;;  (lambda (c v)
 259       ;;    (when (every (lambda (p)
 260       ;;                   (ideographic-structure-member p v))
 261       ;;                 components)
 262       ;;      (princ (encode-coding-string
 263       ;;              (ids-find-format-line c v)
 264       ;;              'utf-8-jp-er))
 265       ;;      (princ "<br>\n")
 266       ;;      )
 267       ;;    nil)
 268       ;;  'ideographic-structure)
 269       (when (= (length components) 1)
 270         (www-ids-find-format-line (aref components 0)
 271                                   (char-feature (aref components 0)
 272                                                 'ideographic-structure)))
 273       ;; (dolist (c (ideographic-products-find components))
 274       ;;   (setq is (char-feature c 'ideographic-structure))
 275       ;;   ;; to avoid problems caused by wrong indexes
 276       ;;   (when (every (lambda (c)
 277       ;;                  (ideographic-structure-member c is))
 278       ;;                components)
 279       ;;     (www-ids-find-format-line c is)))
 280       (princ "<ul>\n")
 281       (www-ids-insert-chars-including-components components)
 282       (princ "</ul>\n")
 283       )
 284      (t
 285       (princ (encode-coding-string "<hr>
 286 <p>
 287 \e$B;XDj$7$?ItIJ$rA4$F4^$`4A;z$N0lMw$rI=<($7$^$9!#\e(B
 288 <p>
 289 CHISE \e$B$GMQ$$$i$l$k<BBV;2>H7A<0!JNc!'\e(B&amp;M-00256;\e$B!K$GItIJ$r;XDj$9$k;v$b$G$-$^$9!#\e(B" 'utf-8-jp-er))
 290       (princ (encode-coding-string "
 291 <p>
 292 \[Links\]
 293 <ul>
 294 <li><a href=\"http://www.shuiren.org/chuden/toyoshi/syoseki/chise_ids.html\"
 295 >\e$B!V\e(BCHISE IDS FIND\e$B$G4A;z$r8!:w!W\e(B</a> \e$B!=\e(B \e$B;3ED?r?N$5$s!J\e(B<a
 296 href=\"http://www.shuiren.org/\">\e$B?g?MDb\e(B</a>\e$B!K$K$h$k2r@b\e(B
 297 </ul>
 298 <ul>
 299 <li><a href=\"http://www.karitsu.org/tools/firefox_plugin.htm\"
 300 >Firefox \e$BMQ\e(B plugin</a> by \e$B=);3M[0lO:$5$s!J\e(B<a href=\"http://www.karitsu.org/\"
 301 >\e$B2aN)c7\e(B</a>\e$B!K\e(B
 302 </ul>
 303 <ul>
 304 <li><a href=\"http://cvs.m17n.org/viewcvs/chise/ids/www/www-ids-find.el?view=markup\"
 305 >www-ids-find.el (source file (Emacs Lisp part))
 306 <li><a href=\"http://kanji.zinbun.kyoto-u.ac.jp/projects/chise/ids/\"
 307 >\e$B!V\e(BCHISE \e$B4A;z9=B$>pJs%G!<%?%Y!<%9!W\e(B</a>
 308 <li><a href=\"http://kanji.zinbun.kyoto-u.ac.jp/projects/chise/\"
 309 >CHISE Project</a>
 310 </ul>
 311 <ul>
 312 <li><a href=\"http://coe21.zinbun.kyoto-u.ac.jp/djvuchar\"
 313 >\e$B!VBsK\J8;z%G!<%?%Y!<%9!W\e(B</a> by
 314 <a href=\"http://coe21.zinbun.kyoto-u.ac.jp/\"
 315 >\e$B5~ETBg3X\e(B21\e$B@$5*\e(BCOE\e$B!VEl%"%8%"@$3&$N?MJ8>pJs3X8&5f650i5rE@!W\e(B</a>
 316 <li><a href=\"http://www.unicode.org/\"
 317 >Unicode</a>
 318 </ul>"
 319  'utf-8-jp-er))
 320
 321       ))
 322     (princ "<hr>")
 323     (princ
 324      (format
 325       "Powered by <a
 326 href=\"http://kanji.zinbun.kyoto-u.ac.jp/projects/chise/xemacs/\"
 327 >XEmacs CHISE</a> %s."
 328       xemacs-chise-version))
 329     (princ "
 330 </body>
 331 </html>
 332 ")))