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 "0.22.3")
  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
 114               " <a href=\"http://www.unicode.org/cgi-bin/GetUnihanData.pl?codepoint=%X\">%s</a>"
 115               ucs
 116               (cond ((<= ucs #xFFFF)
 117                      (format "U+%04X" ucs))
 118                     ((<= ucs #x10FFFF)
 119                      (format "U-%08X" ucs))))
 120            "          ")))
 121     (when ucs
 122       (princ
 123        (format " <a href=\"http://geta.mag.keio.ac.jp/chiseperl/map.cgi?code=%X\">(link map)</a>"
 124                ucs)))
 125     (princ " ")
 126     (when is
 127       (princ
 128        (with-temp-buffer
 129          (insert
 130           (encode-coding-string
 131            (ideographic-structure-to-ids is)
 132            'utf-8-jp-er))
 133          (goto-char (point-min))
 134          (while (re-search-forward "&CB\\([0-9]+\\);" nil t)
 135            (setq code (string-to-int (match-string 1)))
 136            (replace-match
 137             (format "<img alt=\"CB%05d\" src=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/glyphs/cb-gaiji/%02d/CB%05d.gif\">"
 138                     code (/ code 1000) code)
 139             t 'literal))
 140          (buffer-string))))
 141     (when (and ucs
 142                (with-current-buffer
 143                    (find-file-noselect
 144                     www-ids-find-tang-chars-file-name)
 145                  (goto-char (point-min))
 146                  (re-search-forward (format "^%d$" ucs) nil t)))
 147       (princ
 148        (format " <a href=\"http://coe21.zinbun.kyoto-u.ac.jp/djvuchar?query=%s\">"
 149                (mapconcat
 150                 (lambda (c)
 151                   (format "%%%02X" (char-int c)))
 152                 (encode-coding-string (char-to-string c)
 153                                       'utf-8-jp)
 154                 "")))
 155       (princ (encode-coding-string "\e$B"M\e(B[\e$BEbBeBsK\\e(B]</a>" 'utf-8-jp-er)))
 156     (princ "<br>\n")))
 157
 158 (defun www-ids-insert-chars-including-components (components
 159                                                   &optional ignored-chars)
 160   (let ((products (copy-list (ideographic-products-find components)))
 161         is as bs)
 162     (dolist (c (cond
 163                 ((> (length products) 10000)
 164                  products)
 165                 ((> (length products) 4096)
 166                  (sort products
 167                        (lambda (a b)
 168                          (< (char-int a)(char-int b))))
 169                  )
 170                 ((> (length products) 512)
 171                  (sort products
 172                        (lambda (a b)
 173                          (if (setq as (char-total-strokes a))
 174                              (if (setq bs (char-total-strokes b))
 175                                  (if (= as bs)
 176                                      (< (char-int a)(char-int b))
 177                                    (< as bs))
 178                                t)
 179                            (< (char-int a)(char-int b)))))
 180                  )
 181                 (t
 182                  (sort products
 183                        (lambda (a b)
 184                          (if (setq as (char-total-strokes a))
 185                              (if (setq bs (char-total-strokes b))
 186                                  (if (= as bs)
 187                                      (ideograph-char< a b)
 188                                    (< as bs))
 189                                t)
 190                            (ideograph-char< a b))))
 191                  )))
 192       (unless (memq c ignored-chars)
 193         (setq is (char-feature c 'ideographic-structure))
 194         (princ "<li>")
 195         (www-ids-find-format-line c is)
 196         (princ "<ul>\n")
 197         (setq ignored-chars
 198               (www-ids-insert-chars-including-components
 199                (char-to-string c)
 200                (cons c ignored-chars)))
 201         (princ "</ul>\n")
 202         )
 203       ))
 204   ignored-chars)
 205
 206 (defun www-batch-ids-find ()
 207   (let ((components (car command-line-args-left))
 208         (coded-charset-entity-reference-alist
 209          (list*
 210           '(=cns11643-1         "C1-" 4 X)
 211           '(=cns11643-2         "C2-" 4 X)
 212           '(=cns11643-3         "C3-" 4 X)
 213           '(=cns11643-4         "C4-" 4 X)
 214           '(=cns11643-5         "C5-" 4 X)
 215           '(=cns11643-6         "C6-" 4 X)
 216           '(=cns11643-7         "C7-" 4 X)
 217           '(=gb2312             "G0-" 4 X)
 218           '(=gb12345            "G1-" 4 X)
 219           '(=jis-x0208@1990     "J90-" 4 X)
 220           '(=jis-x0212          "JSP-" 4 X)
 221           '(=cbeta              "CB" 5 d)
 222           '(=jef-china3         "JC3-" 4 X)
 223           '(=jis-x0208@1978     "J78-" 4 X)
 224           '(=jis-x0208@1983     "J83-" 4 X)
 225           '(=daikanwa           "M-" 5 d)
 226           coded-charset-entity-reference-alist))
 227         )
 228     (setq command-line-args-left (cdr command-line-args-left))
 229     (cond
 230      ((stringp components)
 231       (if (string-match "^components=" components)
 232           (setq components (substring components (match-end 0))))
 233       (setq components
 234             (if (> (length components) 0)
 235                 (decode-url-string components 'utf-8-jp-er)
 236               nil))
 237       )
 238      (t
 239       (setq components nil)
 240       ))
 241     (princ "Content-Type: text/html; charset=UTF-8
 242
 243 <!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\"
 244             \"http://www.w3.org/TR/html4/loose.dtd\">
 245 <html lang=\"ja\">
 246 <head>
 247 <title>CHISE IDS Find</title>
 248 </head>
 249
 250 <body>
 251
 252 <h1>")
 253     (princ (encode-coding-string "CHISE IDS \e$B4A;z8!:w\e(B" 'utf-8-jp-er))
 254     (princ "</h1>")
 255     (princ "
 256 <p>Version ")
 257     (princ www-ids-find-version)
 258     (princ (format-time-string
 259             " (Last-modified: %Y-%m-%d %H:%M:%S)"
 260             (nth 5
 261                  (file-attributes
 262                   www-ids-find-ideographic-products-file-name))))
 263     (princ "
 264 <p>
 265 Copyright (C) 2005 <a href=\"http://kanji.zinbun.kyoto-u.ac.jp/~tomo/\"
 266 >MORIOKA Tomohiko</a>
 267 <hr>
 268 <p>
 269 <form action=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/ids-find\" method=\"GET\">
 270 ")
 271     (princ (encode-coding-string "\e$BItIJJ8;zNs\e(B" 'utf-8-jp-er))
 272     (princ " <input type=\"text\" name=\"components\" size=\"30\" maxlength=\"30\" value=\"")
 273     (if (> (length components) 0)
 274         (princ (encode-coding-string components 'utf-8-jp-er)))
 275     (princ "\">
 276 <input type=\"submit\" value=\"")
 277     (princ (encode-coding-string "\e$B8!:w3+;O\e(B" 'utf-8-jp-er))
 278     (princ "\">
 279 </form>
 280
 281 ")
 282     (unless (file-newer-than-file-p
 283              www-ids-find-ideographic-products-file-name
 284              (locate-file (car command-line-args) exec-path))
 285       (princ (encode-coding-string "<hr>
 286 <p>
 287 \e$B8=:_!"%7%9%F%`$N99?7:n6HCf$G$9!#$7$P$i$/$*BT$A$/$@$5$$!#\e(B
 288 <hr>
 289 " 'utf-8-jp-er))
 290       ;; (setq components nil)
 291       )
 292     (cond
 293      (components
 294       ;; (map-char-attribute
 295       ;;  (lambda (c v)
 296       ;;    (when (every (lambda (p)
 297       ;;                   (ideographic-structure-member p v))
 298       ;;                 components)
 299       ;;      (princ (encode-coding-string
 300       ;;              (ids-find-format-line c v)
 301       ;;              'utf-8-jp-er))
 302       ;;      (princ "<br>\n")
 303       ;;      )
 304       ;;    nil)
 305       ;;  'ideographic-structure)
 306       (when (= (length components) 1)
 307         (www-ids-find-format-line (aref components 0)
 308                                   (char-feature (aref components 0)
 309                                                 'ideographic-structure)))
 310       ;; (dolist (c (ideographic-products-find components))
 311       ;;   (setq is (char-feature c 'ideographic-structure))
 312       ;;   ;; to avoid problems caused by wrong indexes
 313       ;;   (when (every (lambda (c)
 314       ;;                  (ideographic-structure-member c is))
 315       ;;                components)
 316       ;;     (www-ids-find-format-line c is)))
 317       (princ "<ul>\n")
 318       (www-ids-insert-chars-including-components components)
 319       (princ "</ul>\n")
 320       )
 321      (t
 322       (princ (encode-coding-string "<hr>
 323 <p>
 324 \e$B;XDj$7$?ItIJ$rA4$F4^$`4A;z$N0lMw$rI=<($7$^$9!#\e(B
 325 <p>
 326 CHISE \e$B$GMQ$$$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))
 327       (princ (encode-coding-string "
 328 <p>
 329 \[Links\]
 330 <ul>
 331 <li><a href=\"http://www.shuiren.org/chuden/toyoshi/syoseki/chise_ids.html\"
 332 >\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
 333 href=\"http://www.shuiren.org/\">\e$B?g?MDb\e(B</a>\e$B!K$K$h$k2r@b\e(B
 334 </ul>
 335 <ul>
 336 <li><a href=\"http://www.karitsu.org/tools/firefox_plugin.htm\"
 337 >Firefox \e$BMQ\e(B plugin</a> by \e$B=);3M[0lO:$5$s!J\e(B<a href=\"http://www.karitsu.org/\"
 338 >\e$B2aN)c7\e(B</a>\e$B!K\e(B
 339 </ul>
 340 <ul>
 341 <li><a href=\"http://cvs.m17n.org/viewcvs/chise/ids/www/www-ids-find.el?view=markup\"
 342 >www-ids-find.el (source file (Emacs Lisp part))
 343 <li><a href=\"http://kanji.zinbun.kyoto-u.ac.jp/projects/chise/ids/\"
 344 >\e$B!V\e(BCHISE \e$B4A;z9=B$>pJs%G!<%?%Y!<%9!W\e(B</a>
 345 <li><a href=\"http://fonts.jp/chise_linkmap/\"
 346 >\e$B!V\e(Bchise_linkmap : CHISE \e$B4A;zO"4D?^!W\e(B</a> by \e$B>eCO9(0l$5$s\e(B
 347 <li><a href=\"http://kanji.zinbun.kyoto-u.ac.jp/projects/chise/\"
 348 >CHISE Project</a>
 349 </ul>
 350 <ul>
 351 <li><a href=\"http://coe21.zinbun.kyoto-u.ac.jp/djvuchar\"
 352 >\e$B!VBsK\J8;z%G!<%?%Y!<%9!W\e(B</a> by
 353 <a href=\"http://coe21.zinbun.kyoto-u.ac.jp/\"
 354 >\e$B5~ETBg3X\e(B21\e$B@$5*\e(BCOE\e$B!VEl%"%8%"@$3&$N?MJ8>pJs3X8&5f650i5rE@!W\e(B</a>
 355 <li><a href=\"http://www.unicode.org/\"
 356 >Unicode</a>
 357 </ul>"
 358  'utf-8-jp-er))
 359
 360       ))
 361     (princ "<hr>")
 362     (princ
 363      (format
 364       "Powered by <a
 365 href=\"http://kanji.zinbun.kyoto-u.ac.jp/projects/chise/xemacs/\"
 366 >XEmacs CHISE</a> %s."
 367       xemacs-chise-version))
 368     (princ "
 369 </body>
 370 </html>
 371 ")))