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 "0.23.3")
  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-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-line (c is)
  35   (let ((str (encode-coding-string (format "%c" c) 'utf-8-er))
  36         plane code ucs)
  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=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
  43          (insert str)
  44          (insert (format "\"><img alt=\"CB%05d\" src=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/glyphs/cb-gaiji/%02d/CB%05d.gif\">\n"
  45                          code (/ code 1000) code))
  46          (insert (format "CB%05d</a>" code))
  47          )
  48         ((string-match "&JC3-\\([0-9A-F]+\\);" str)
  49          (setq code (string-to-int (match-string 1 str) 16))
  50          (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
  51          (insert str)
  52          (insert (format "\"><img alt=\"JC3-%04X\" src=\"http://kanji.zinbun.kyoto-u.ac.jp/db/CHINA3/Gaiji/%04x.gif\">\n"
  53                          code code))
  54          (insert (format "JC3-%04X</a>" code))
  55          )
  56         ((string-match "&J\\(78\\|83\\|90\\|SP\\)-\\([0-9A-F]+\\);" str)
  57          (setq plane (match-string 1 str)
  58                code (string-to-int (match-string 2 str) 16))
  59          (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
  60          (insert str)
  61          (insert (format "\"><img alt=\"J%s-%04X\" src=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/glyphs/JIS-%s/%02d-%02d.gif\">\n"
  62                          plane code plane
  63                          (- (lsh code -8) 32)
  64                          (- (logand code 255) 32)))
  65          (insert (format "J%s-%04X</a>" plane code))
  66          )
  67         ((string-match "&G\\([01]\\)-\\([0-9A-F]+\\);" str)
  68          (setq plane (string-to-int (match-string 1 str))
  69                code (string-to-int (match-string 2 str) 16))
  70          (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
  71          (insert str)
  72          (insert (format "\"><img alt=\"G%d-%04X\" src=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/glyphs/GB%d/%02d-%02d.gif\">\n"
  73                          plane code plane
  74                          (- (lsh code -8) 32)
  75                          (- (logand code 255) 32)))
  76          (insert (format "G%d-%04X</a>" plane code))
  77          )
  78         ((string-match "&C\\([1-7]\\)-\\([0-9A-F]+\\);" str)
  79          (setq plane (string-to-int (match-string 1 str))
  80                code (string-to-int (match-string 2 str) 16))
  81          (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
  82          (insert str)
  83          (insert (format "\"><img alt=\"C%d-%04X\" src=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/glyphs/CNS%d/%04X.gif\">\n"
  84                          plane code plane code))
  85          (insert (format "C%d-%04X</a>" plane code))
  86          )
  87         ((string-match "&ZOB-\\([0-9]+\\);" str)
  88          (setq code (string-to-int (match-string 1 str)))
  89          (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
  90          (insert str)
  91          (insert (format "\"><img alt=\"ZOB-%04d\" src=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/glyphs/ZOB-1968/%04d.png\">\n"
  92                          code code))
  93          (insert (format "ZOB-%04d</a>" code))
  94          )
  95         (t
  96          (insert "<a href=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/char-desc?char=")
  97          ;; (insert str)
  98          (insert
  99           (mapconcat (lambda (c)
 100                        (if (<= (char-int c) #x7F)
 101                            (char-to-string c)
 102                          (format "%%%02X" c)))
 103                      str ""))
 104          (insert "\">")
 105          (insert str)
 106          (insert "</a>")
 107          ))
 108        (goto-char (point-min))
 109        (while (search-forward "&" nil t)
 110          (replace-match "&amp;" t 'literal))
 111        (buffer-string)
 112        ))
 113     (princ
 114      (or (if (setq ucs (or (char-ucs c)
 115                            (encode-char c 'ucs)))
 116              (format
 117               " <a href=\"http://www.unicode.org/cgi-bin/GetUnihanData.pl?codepoint=%X\">%s</a>"
 118               ucs
 119               (cond ((<= ucs #xFFFF)
 120                      (format "U+%04X" ucs))
 121                     ((<= ucs #x10FFFF)
 122                      (format "U-%08X" ucs))))
 123            "          ")))
 124     (when ucs
 125       (princ
 126        (format " <a href=\"%s%X\">(link map)</a>"
 127                www-ids-find-chise-link-map-url-prefix ucs)))
 128     (princ " ")
 129     (when is
 130       (princ
 131        (with-temp-buffer
 132          (insert
 133           (encode-coding-string
 134            (ideographic-structure-to-ids is)
 135            'utf-8-jp-er))
 136          (goto-char (point-min))
 137          (while (re-search-forward "&CB\\([0-9]+\\);" nil t)
 138            (setq code (string-to-int (match-string 1)))
 139            (replace-match
 140             (format "<img alt=\"CB%05d\" src=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/glyphs/cb-gaiji/%02d/CB%05d.gif\">"
 141                     code (/ code 1000) code)
 142             t 'literal))
 143          (buffer-string))))
 144     (when (and ucs
 145                (with-current-buffer
 146                    (find-file-noselect
 147                     www-ids-find-tang-chars-file-name)
 148                  (goto-char (point-min))
 149                  (re-search-forward (format "^%d$" ucs) nil t)))
 150       (princ
 151        (format " <a href=\"http://coe21.zinbun.kyoto-u.ac.jp/djvuchar?query=%s\">"
 152                (mapconcat
 153                 (lambda (c)
 154                   (format "%%%02X" (char-int c)))
 155                 (encode-coding-string (char-to-string c)
 156                                       'utf-8-jp)
 157                 "")))
 158       (princ (encode-coding-string "\e$B"M\e(B[\e$BEbBeBsK\\e(B]</a>" 'utf-8-jp-er)))
 159     (princ "<br>\n")))
 160
 161 (defun www-ids-insert-chars-including-components (components
 162                                                   &optional ignored-chars)
 163   (let ((products (copy-list (ideographic-products-find components)))
 164         is as bs)
 165     (dolist (c (cond
 166                 ((> (length products) 10000)
 167                  products)
 168                 ((> (length products) 4096)
 169                  (sort products
 170                        (lambda (a b)
 171                          (< (char-int a)(char-int b))))
 172                  )
 173                 ((> (length products) 512)
 174                  (sort products
 175                        (lambda (a b)
 176                          (if (setq as (char-total-strokes a))
 177                              (if (setq bs (char-total-strokes b))
 178                                  (if (= as bs)
 179                                      (< (char-int a)(char-int b))
 180                                    (< as bs))
 181                                t)
 182                            (< (char-int a)(char-int b)))))
 183                  )
 184                 (t
 185                  (sort products
 186                        (lambda (a b)
 187                          (if (setq as (char-total-strokes a))
 188                              (if (setq bs (char-total-strokes b))
 189                                  (if (= as bs)
 190                                      (ideograph-char< a b)
 191                                    (< as bs))
 192                                t)
 193                            (ideograph-char< a b))))
 194                  )))
 195       (unless (memq c ignored-chars)
 196         (setq is (char-feature c 'ideographic-structure))
 197         (princ "<li>")
 198         (www-ids-find-format-line c is)
 199         (princ "<ul>\n")
 200         (setq ignored-chars
 201               (www-ids-insert-chars-including-components
 202                (char-to-string c)
 203                (cons c ignored-chars)))
 204         (princ "</ul>\n")
 205         )
 206       ))
 207   ignored-chars)
 208
 209 (defun www-batch-ids-find ()
 210   (let ((components (car command-line-args-left))
 211         (coded-charset-entity-reference-alist
 212          (list*
 213           '(=cns11643-1         "C1-" 4 X)
 214           '(=cns11643-2         "C2-" 4 X)
 215           '(=cns11643-3         "C3-" 4 X)
 216           '(=cns11643-4         "C4-" 4 X)
 217           '(=cns11643-5         "C5-" 4 X)
 218           '(=cns11643-6         "C6-" 4 X)
 219           '(=cns11643-7         "C7-" 4 X)
 220           '(=gb2312             "G0-" 4 X)
 221           '(=gb12345            "G1-" 4 X)
 222           '(=jis-x0208@1990     "J90-" 4 X)
 223           '(=jis-x0212          "JSP-" 4 X)
 224           '(=cbeta              "CB" 5 d)
 225           '(=jef-china3         "JC3-" 4 X)
 226           '(=jis-x0208@1978     "J78-" 4 X)
 227           '(=jis-x0208@1983     "J83-" 4 X)
 228           '(=daikanwa           "M-" 5 d)
 229           coded-charset-entity-reference-alist))
 230         )
 231     (setq command-line-args-left (cdr command-line-args-left))
 232     (cond
 233      ((stringp components)
 234       (if (string-match "^components=" components)
 235           (setq components (substring components (match-end 0))))
 236       (setq components
 237             (if (> (length components) 0)
 238                 (decode-url-string components 'utf-8-er ;'utf-8-jp-er
 239                                    )
 240               nil))
 241       )
 242      (t
 243       (setq components nil)
 244       ))
 245     (princ "Content-Type: text/html; charset=UTF-8
 246
 247 <!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\"
 248             \"http://www.w3.org/TR/html4/loose.dtd\">
 249 <html lang=\"ja\">
 250 <head>
 251 <title>CHISE IDS Find</title>
 252 </head>
 253
 254 <body>
 255
 256 <h1>")
 257     (princ (encode-coding-string "CHISE IDS \e$B4A;z8!:w\e(B" 'utf-8-jp-er))
 258     (princ "</h1>")
 259     (princ "
 260 <p>Version ")
 261     (princ www-ids-find-version)
 262     (princ (format-time-string
 263             " (Last-modified: %Y-%m-%d %H:%M:%S)"
 264             (nth 5
 265                  (file-attributes
 266                   www-ids-find-ideographic-products-file-name))))
 267     (princ "
 268 <hr>
 269 <p>
 270 <form action=\"http://mousai.kanji.zinbun.kyoto-u.ac.jp/ids-find\" method=\"GET\">
 271 ")
 272     (princ (encode-coding-string "\e$BItIJJ8;zNs\e(B" 'utf-8-jp-er))
 273     (princ " <input type=\"text\" name=\"components\" size=\"30\" maxlength=\"30\" value=\"")
 274     (if (> (length components) 0)
 275         (princ (encode-coding-string components 'utf-8-jp-er)))
 276     (princ "\">
 277 <input type=\"submit\" value=\"")
 278     (princ (encode-coding-string "\e$B8!:w3+;O\e(B" 'utf-8-jp-er))
 279     (princ "\">
 280 </form>
 281
 282 ")
 283     (unless (file-newer-than-file-p
 284              www-ids-find-ideographic-products-file-name
 285              (locate-file (car command-line-args) exec-path))
 286       (princ (encode-coding-string "<hr>
 287 <p>
 288 \e$B8=:_!"%7%9%F%`$N99?7:n6HCf$G$9!#$7$P$i$/$*BT$A$/$@$5$$!#\e(B
 289 <hr>
 290 " 'utf-8-jp-er))
 291       ;; (setq components nil)
 292       )
 293     (cond
 294      (components
 295       ;; (map-char-attribute
 296       ;;  (lambda (c v)
 297       ;;    (when (every (lambda (p)
 298       ;;                   (ideographic-structure-member p v))
 299       ;;                 components)
 300       ;;      (princ (encode-coding-string
 301       ;;              (ids-find-format-line c v)
 302       ;;              'utf-8-jp-er))
 303       ;;      (princ "<br>\n")
 304       ;;      )
 305       ;;    nil)
 306       ;;  'ideographic-structure)
 307       (when (= (length components) 1)
 308         (www-ids-find-format-line (aref components 0)
 309                                   (char-feature (aref components 0)
 310                                                 'ideographic-structure)))
 311       ;; (dolist (c (ideographic-products-find components))
 312       ;;   (setq is (char-feature c 'ideographic-structure))
 313       ;;   ;; to avoid problems caused by wrong indexes
 314       ;;   (when (every (lambda (c)
 315       ;;                  (ideographic-structure-member c is))
 316       ;;                components)
 317       ;;     (www-ids-find-format-line c is)))
 318       (princ "<ul>\n")
 319       (www-ids-insert-chars-including-components components)
 320       (princ "</ul>\n")
 321       )
 322      (t
 323       (princ (encode-coding-string "<hr>
 324 <p>
 325 \e$B;XDj$7$?ItIJ$rA4$F4^$`4A;z$N0lMw$rI=<($7$^$9!#\e(B
 326 <p>
 327 CHISE \e$B$GMQ$$$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))
 328       (princ (encode-coding-string "
 329 <p>
 330 \[Links\]
 331 <ul>
 332 <li><a href=\"http://www.shuiren.org/chuden/toyoshi/syoseki/chise_ids.html\"
 333 >\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
 334 href=\"http://www.shuiren.org/\">\e$B?g?MDb\e(B</a>\e$B!K$K$h$k2r@b\e(B
 335 </ul>
 336 <ul>
 337 <li><a href=\"http://www.karitsu.org/tools/firefox_plugin.htm\"
 338 >Firefox \e$BMQ\e(B plugin</a> by \e$B=);3M[0lO:$5$s!J\e(B<a href=\"http://www.karitsu.org/\"
 339 >\e$B2aN)c7\e(B</a>\e$B!K\e(B
 340 </ul>
 341 <ul>
 342 <li><a href=\"http://cvs.m17n.org/viewcvs/chise/ids/www/www-ids-find.el?view=markup\"
 343 >www-ids-find.el (source file (Emacs Lisp part))
 344 <li><a href=\"http://kanji.zinbun.kyoto-u.ac.jp/projects/chise/ids/\"
 345 >\e$B!V\e(BCHISE \e$B4A;z9=B$>pJs%G!<%?%Y!<%9!W\e(B</a>
 346 <li><a href=\"http://fonts.jp/chise_linkmap/\"
 347 >\e$B!V\e(Bchise_linkmap : CHISE \e$B4A;zO"4D?^!W\e(B</a> by \e$B>eCO9(0l$5$s\e(B
 348 <li><a href=\"http://kanji.zinbun.kyoto-u.ac.jp/projects/chise/\"
 349 >CHISE Project</a>
 350 </ul>
 351 <ul>
 352 <li><a href=\"http://coe21.zinbun.kyoto-u.ac.jp/djvuchar\"
 353 >\e$B!VBsK\J8;z%G!<%?%Y!<%9!W\e(B</a> by
 354 <a href=\"http://coe21.zinbun.kyoto-u.ac.jp/\"
 355 >\e$B5~ETBg3X\e(B21\e$B@$5*\e(BCOE\e$B!VEl%"%8%"@$3&$N?MJ8>pJs3X8&5f650i5rE@!W\e(B</a>
 356 <li><a href=\"http://www.unicode.org/\"
 357 >Unicode</a>
 358 </ul>"
 359  'utf-8-jp-er))
 360
 361       ))
 362     (princ "<hr>")
 363     (princ "<p>
 364 Copyright (C) 2005, 2006, 2007 <a href=\"http://kanji.zinbun.kyoto-u.ac.jp/~tomo/\"
 365 >MORIOKA Tomohiko</a>")
 366     (princ
 367      (format
 368       "<p>Powered by <a
 369 href=\"http://kanji.zinbun.kyoto-u.ac.jp/projects/chise/xemacs/\"
 370 >XEmacs CHISE</a> %s."
 371       xemacs-chise-version))
 372     (princ "
 373 </body>
 374 </html>
 375 ")))