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 "0.24.2")
  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-chise-link-map-url-prefix
  29   "http://kamichi.jp/chise_linkmap/map.cgi?code=")
  30
  31 (defvar www-ids-find-tang-chars-file-name
  32   "~tomo/projects/chise/ids/www/tang-chars.udd")
  33
  34 (defun www-ids-find-format-char (c &optional code-desc)
  35   (let ((str (encode-coding-string (format "%c" c) 'utf-8-er))
  36         plane code)
  37     (princ
  38      (with-temp-buffer
  39        (cond
  40         ((string-match "&CB\\([0-9]+\\);" str)
  41          (setq code (string-to-int (match-string 1 str)))
  42          (insert "<a href=\"/char-desc?char=")
  43          (insert str)
  44          (insert (format "\"><img alt=\"CB%05d\" src=\"/glyphs/cb-gaiji/%02d/CB%05d.gif\">\n"
  45                          code (/ code 1000) code))
  46          (when code-desc
  47            (insert (format "CB%05d</a>" code)))
  48          )
  49         ((string-match "&JC3-\\([0-9A-F]+\\);" str)
  50          (setq code (string-to-int (match-string 1 str) 16))
  51          (insert "<a href=\"/char-desc?char=")
  52          (insert str)
  53          (insert (format "\"><img alt=\"JC3-%04X\" src=\"http://kanji.zinbun.kyoto-u.ac.jp/db/CHINA3/Gaiji/%04x.gif\">\n"
  54                          code code))
  55          (when code-desc
  56            (insert (format "JC3-%04X</a>" code)))
  57          )
  58         ((string-match "&J\\(78\\|83\\|90\\|SP\\)-\\([0-9A-F]+\\);" str)
  59          (setq plane (match-string 1 str)
  60                code (string-to-int (match-string 2 str) 16))
  61          (insert "<a href=\"/char-desc?char=")
  62          (insert str)
  63          (insert (format "\"><img alt=\"J%s-%04X\" src=\"/glyphs/JIS-%s/%02d-%02d.gif\">\n"
  64                          plane code plane
  65                          (- (lsh code -8) 32)
  66                          (- (logand code 255) 32)))
  67          (when code-desc
  68            (insert (format "J%s-%04X</a>" plane code)))
  69          )
  70         ((string-match "&G\\([01]\\)-\\([0-9A-F]+\\);" str)
  71          (setq plane (string-to-int (match-string 1 str))
  72                code (string-to-int (match-string 2 str) 16))
  73          (insert "<a href=\"/char-desc?char=")
  74          (insert str)
  75          (insert (format "\"><img alt=\"G%d-%04X\" src=\"/glyphs/GB%d/%02d-%02d.gif\">\n"
  76                          plane code plane
  77                          (- (lsh code -8) 32)
  78                          (- (logand code 255) 32)))
  79          (when code-desc
  80            (insert (format "G%d-%04X</a>" plane code)))
  81          )
  82         ((string-match "&C\\([1-7]\\)-\\([0-9A-F]+\\);" str)
  83          (setq plane (string-to-int (match-string 1 str))
  84                code (string-to-int (match-string 2 str) 16))
  85          (insert "<a href=\"/char-desc?char=")
  86          (insert str)
  87          (insert (format "\"><img alt=\"C%d-%04X\" src=\"/glyphs/CNS%d/%04X.gif\">\n"
  88                          plane code plane code))
  89          (when code-desc
  90            (insert (format "C%d-%04X</a>" plane code)))
  91          )
  92         ((string-match "&ZOB-\\([0-9]+\\);" str)
  93          (setq code (string-to-int (match-string 1 str)))
  94          (insert "<a href=\"/char-desc?char=")
  95          (insert str)
  96          (insert (format "\"><img alt=\"ZOB-%04d\" src=\"/glyphs/ZOB-1968/%04d.png\">\n"
  97                          code code))
  98          (when code-desc
  99            (insert (format "ZOB-%04d</a>" code)))
 100          )
 101         (t
 102          (insert "<a href=\"/char-desc?char=")
 103          ;; (insert str)
 104          (insert
 105           (mapconcat (lambda (c)
 106                        (if (<= (char-int c) #x7F)
 107                            (char-to-string c)
 108                          (format "%%%02X" c)))
 109                      str ""))
 110          (insert "\">")
 111          (insert str)
 112          (insert "</a>")
 113          ))
 114        (goto-char (point-min))
 115        (while (search-forward "&" nil t)
 116          (replace-match "&amp;" t 'literal))
 117        (buffer-string)))))
 118
 119 (defun www-ids-find-format-line (c is)
 120   (let (ucs len i ids)
 121     (www-ids-find-format-char c 'code-desc)
 122     (princ
 123      (or (if (setq ucs (or (char-ucs c)
 124                            (encode-char c 'ucs)))
 125              (format
 126               " <a href=\"http://www.unicode.org/cgi-bin/GetUnihanData.pl?codepoint=%X\">%s</a>"
 127               ucs
 128               (cond ((<= ucs #xFFFF)
 129                      (format "U+%04X" ucs))
 130                     ((<= ucs #x10FFFF)
 131                      (format "U-%08X" ucs))))
 132            "          ")))
 133     (when ucs
 134       (princ
 135        (format " <a href=\"%s%X\">(link map)</a>"
 136                www-ids-find-chise-link-map-url-prefix ucs)))
 137     (princ " ")
 138     (when is
 139       (setq ids (ideographic-structure-to-ids is))
 140       (setq i 0
 141             len (length ids))
 142       (while (< i len)
 143         (www-ids-find-format-char (aref ids i))
 144         (setq i (1+ i))))
 145     (when (and ucs
 146                (with-current-buffer
 147                    (find-file-noselect
 148                     www-ids-find-tang-chars-file-name)
 149                  (goto-char (point-min))
 150                  (re-search-forward (format "^%d$" ucs) nil t)))
 151       (princ
 152        (format " <a href=\"http://coe21.zinbun.kyoto-u.ac.jp/djvuchar?query=%s\">"
 153                (mapconcat
 154                 (lambda (c)
 155                   (format "%%%02X" (char-int c)))
 156                 (encode-coding-string (char-to-string c)
 157                                       'utf-8-jp)
 158                 "")))
 159       (princ (encode-coding-string "\e$B"M\e(B[\e$BEbBeBsK\\e(B]</a>" 'utf-8-jp-er)))
 160     (princ "<br>\n")))
 161
 162 (defun www-ids-insert-chars-including-components (components
 163                                                   &optional ignored-chars)
 164   (let ((products (copy-list (ideographic-products-find components)))
 165         is as bs len ignore-children)
 166     (setq len (length products))
 167     (dolist (c (cond
 168                 ((> len 8192)
 169                  (setq ignore-children t)
 170                  products)
 171                 ((> len 4096)
 172                  (sort products
 173                        (lambda (a b)
 174                          (< (char-int a)(char-int b))))
 175                  )
 176                 ((> len 512)
 177                  (sort products
 178                        (lambda (a b)
 179                          (if (setq as (char-total-strokes a))
 180                              (if (setq bs (char-total-strokes b))
 181                                  (if (= as bs)
 182                                      (< (char-int a)(char-int b))
 183                                    (< as bs))
 184                                t)
 185                            (< (char-int a)(char-int b)))))
 186                  )
 187                 (t
 188                  (sort products
 189                        (lambda (a b)
 190                          (if (setq as (char-total-strokes a))
 191                              (if (setq bs (char-total-strokes b))
 192                                  (if (= as bs)
 193                                      (ideograph-char< a b)
 194                                    (< as bs))
 195                                t)
 196                            (ideograph-char< a b))))
 197                  )))
 198       (unless (memq c ignored-chars)
 199         (setq is (char-feature c 'ideographic-structure))
 200         (princ "<li>")
 201         (www-ids-find-format-line c is)
 202         (unless ignore-children
 203           (princ "<ul>\n")
 204           (setq ignored-chars
 205                 (www-ids-insert-chars-including-components
 206                  (char-to-string c)
 207                  (cons c ignored-chars)))
 208           (princ "</ul>\n"))
 209         )
 210       ))
 211   ignored-chars)
 212
 213 (defun www-batch-ids-find ()
 214   (let ((components (car command-line-args-left))
 215         (coded-charset-entity-reference-alist
 216          (list*
 217           '(=cns11643-1         "C1-" 4 X)
 218           '(=cns11643-2         "C2-" 4 X)
 219           '(=cns11643-3         "C3-" 4 X)
 220           '(=cns11643-4         "C4-" 4 X)
 221           '(=cns11643-5         "C5-" 4 X)
 222           '(=cns11643-6         "C6-" 4 X)
 223           '(=cns11643-7         "C7-" 4 X)
 224           '(=gb2312             "G0-" 4 X)
 225           '(=gb12345            "G1-" 4 X)
 226           '(=jis-x0208@1990     "J90-" 4 X)
 227           '(=jis-x0212          "JSP-" 4 X)
 228           '(=cbeta              "CB" 5 d)
 229           '(=jef-china3         "JC3-" 4 X)
 230           '(=jis-x0208@1978     "J78-" 4 X)
 231           '(=jis-x0208@1983     "J83-" 4 X)
 232           '(=daikanwa           "M-" 5 d)
 233           coded-charset-entity-reference-alist))
 234         )
 235     (setq command-line-args-left (cdr command-line-args-left))
 236     (cond
 237      ((stringp components)
 238       (if (string-match "^components=" components)
 239           (setq components (substring components (match-end 0))))
 240       (setq components
 241             (if (> (length components) 0)
 242                 (decode-url-string components 'utf-8-er)
 243               nil))
 244       )
 245      (t
 246       (setq components nil)
 247       ))
 248     (princ "Content-Type: text/html; charset=UTF-8
 249
 250 <!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\"
 251             \"http://www.w3.org/TR/html4/loose.dtd\">
 252 <html lang=\"ja\">
 253 <head>
 254 <title>CHISE IDS Find</title>
 255 </head>
 256
 257 <body>
 258
 259 <h1>")
 260     (princ (encode-coding-string "CHISE IDS \e$B4A;z8!:w\e(B" 'utf-8-jp-er))
 261     (princ "</h1>")
 262     (princ "
 263 <p>Version ")
 264     (princ www-ids-find-version)
 265     (princ (format-time-string
 266             " (Last-modified: %Y-%m-%d %H:%M:%S)"
 267             (nth 5
 268                  (file-attributes
 269                   www-ids-find-ideographic-products-file-name))))
 270     (princ "
 271 <hr>
 272 <p>
 273 <form action=\"/ids-find\" method=\"GET\">
 274 ")
 275     (princ (encode-coding-string "\e$BItIJJ8;zNs\e(B" 'utf-8-jp-er))
 276     (princ " <input type=\"text\" name=\"components\" size=\"30\" maxlength=\"30\" value=\"")
 277     (if (> (length components) 0)
 278         (princ (encode-coding-string components 'utf-8-er)))
 279     (princ "\">
 280 <input type=\"submit\" value=\"")
 281     (princ (encode-coding-string "\e$B8!:w3+;O\e(B" 'utf-8-jp-er))
 282     (princ "\">
 283 </form>
 284
 285 ")
 286     (unless (file-newer-than-file-p
 287              www-ids-find-ideographic-products-file-name
 288              (locate-file (car command-line-args) exec-path))
 289       (princ (encode-coding-string "<hr>
 290 <p>
 291 \e$B8=:_!"%7%9%F%`$N99?7:n6HCf$G$9!#$7$P$i$/$*BT$A$/$@$5$$!#\e(B
 292 <hr>
 293 " 'utf-8-jp-er))
 294       ;; (setq components nil)
 295       )
 296     (cond
 297      (components
 298       ;; (map-char-attribute
 299       ;;  (lambda (c v)
 300       ;;    (when (every (lambda (p)
 301       ;;                   (ideographic-structure-member p v))
 302       ;;                 components)
 303       ;;      (princ (encode-coding-string
 304       ;;              (ids-find-format-line c v)
 305       ;;              'utf-8-jp-er))
 306       ;;      (princ "<br>\n")
 307       ;;      )
 308       ;;    nil)
 309       ;;  'ideographic-structure)
 310       (when (= (length components) 1)
 311         (www-ids-find-format-line (aref components 0)
 312                                   (char-feature (aref components 0)
 313                                                 'ideographic-structure)))
 314       ;; (dolist (c (ideographic-products-find components))
 315       ;;   (setq is (char-feature c 'ideographic-structure))
 316       ;;   ;; to avoid problems caused by wrong indexes
 317       ;;   (when (every (lambda (c)
 318       ;;                  (ideographic-structure-member c is))
 319       ;;                components)
 320       ;;     (www-ids-find-format-line c is)))
 321       (princ "<ul>\n")
 322       (www-ids-insert-chars-including-components components)
 323       (princ "</ul>\n")
 324       )
 325      (t
 326       (princ (encode-coding-string "<hr>
 327 <p>
 328 \e$B;XDj$7$?ItIJ$rA4$F4^$`4A;z$N0lMw$rI=<($7$^$9!#\e(B
 329 <p>
 330 CHISE \e$B$GMQ$$$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))
 331       (princ (encode-coding-string "
 332 <p>
 333 \[Links\]
 334 <ul>
 335 <li><a href=\"http://www.shuiren.org/chuden/toyoshi/syoseki/chise_ids.html\"
 336 >\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
 337 href=\"http://www.shuiren.org/\">\e$B?g?MDb\e(B</a>\e$B!K$K$h$k2r@b\e(B
 338 </ul>
 339 <ul>
 340 <li><a href=\"http://www.karitsu.org/tools/firefox_plugin.htm\"
 341 >Firefox \e$BMQ\e(B plugin</a> by \e$B=);3M[0lO:$5$s!J\e(B<a href=\"http://www.karitsu.org/\"
 342 >\e$B2aN)c7\e(B</a>\e$B!K\e(B
 343 </ul>
 344 <ul>
 345 <li><a href=\"http://cvs.m17n.org/viewcvs/chise/ids/www/www-ids-find.el?view=markup\"
 346 >www-ids-find.el (source file (Emacs Lisp part))
 347 <li><a href=\"http://kanji.zinbun.kyoto-u.ac.jp/projects/chise/ids/\"
 348 >\e$B!V\e(BCHISE \e$B4A;z9=B$>pJs%G!<%?%Y!<%9!W\e(B</a>
 349 <li><a href=\"http://fonts.jp/chise_linkmap/\"
 350 >\e$B!V\e(Bchise_linkmap : CHISE \e$B4A;zO"4D?^!W\e(B</a> by \e$B>eCO9(0l$5$s\e(B
 351 <li><a href=\"http://kanji.zinbun.kyoto-u.ac.jp/projects/chise/\"
 352 >CHISE Project</a>
 353 </ul>
 354 <ul>
 355 <li><a href=\"http://coe21.zinbun.kyoto-u.ac.jp/djvuchar\"
 356 >\e$B!VBsK\J8;z%G!<%?%Y!<%9!W\e(B</a> by
 357 <a href=\"http://coe21.zinbun.kyoto-u.ac.jp/\"
 358 >\e$B5~ETBg3X\e(B21\e$B@$5*\e(BCOE\e$B!VEl%"%8%"@$3&$N?MJ8>pJs3X8&5f650i5rE@!W\e(B</a>
 359 <li><a href=\"http://www.unicode.org/\"
 360 >Unicode</a>
 361 </ul>"
 362  'utf-8-jp-er))
 363
 364       ))
 365     (princ "<hr>")
 366     (princ "<p>
 367 Copyright (C) 2005, 2006, 2007, 2008, 2009 <a href=\"http://kanji.zinbun.kyoto-u.ac.jp/~tomo/\"
 368 >MORIOKA Tomohiko</a>")
 369     (princ
 370      (format
 371       "<p>Powered by <a
 372 href=\"http://kanji.zinbun.kyoto-u.ac.jp/projects/chise/xemacs/\"
 373 >XEmacs CHISE</a> %s."
 374       xemacs-chise-version))
 375     (princ "
 376 </body>
 377 </html>
 378 ")))