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 "0.25.0")
  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://fonts.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$BEbBeBsK\\e(B]</a>" 'utf-8-jp-er)))
 160     (princ "<br>\n")))
 161
 162 (defun www-ids-insert-chars-including-components (components
 163                                                   &optional ignored-chars)
 164   (let ((products (ideographic-products-find components))
 165         is as bs len ignore-children)
 166     (setq len (length products))
 167     (when (>= len 1024)
 168       (setq ignore-children t)
 169       (princ
 170        (encode-coding-string
 171         "<p>\e$B7k2L$,B?$9$.$k$?$a!":F5"E*8!:w$r>JN,$7$^$7$?!#\e(B</p>"
 172         'utf-8-jp-er)))
 173     (if (>= len 2048)
 174         (dolist (c products)
 175           (www-ids-find-format-char c))
 176       (princ "<ul>\n")
 177       (dolist (c (cond
 178                   ;; ((>= len 2048)
 179                   ;;  (setq ignore-children t)
 180                   ;;  products)
 181                   ;; ((>= len 1024)
 182                   ;;  products)
 183                   ((>= len 1024)
 184                    (sort (copy-list products)
 185                          (lambda (a b)
 186                            (< (char-int a)(char-int b))))
 187                    )
 188                   ((>= len 512)
 189                    (sort (copy-list products)
 190                          (lambda (a b)
 191                            (if (setq as (char-total-strokes a))
 192                                (if (setq bs (char-total-strokes b))
 193                                    (if (= as bs)
 194                                        (< (char-int a)(char-int b))
 195                                      (< as bs))
 196                                  t)
 197                              (< (char-int a)(char-int b)))))
 198                    )
 199                   (t
 200                    (sort (copy-list products)
 201                          (lambda (a b)
 202                            (if (setq as (char-total-strokes a))
 203                                (if (setq bs (char-total-strokes b))
 204                                    (if (= as bs)
 205                                        (ideograph-char< a b)
 206                                      (< as bs))
 207                                  t)
 208                              (ideograph-char< a b))))
 209                    )))
 210         (unless (memq c ignored-chars)
 211           (setq is (char-feature c 'ideographic-structure))
 212           (princ "<li>")
 213           (www-ids-find-format-line c is)
 214           (unless ignore-children
 215             ;; (princ "<ul>\n")
 216             (setq ignored-chars
 217                   (www-ids-insert-chars-including-components
 218                    (char-to-string c)
 219                    (cons c ignored-chars)))
 220             ;; (princ "</ul>\n")
 221             ))
 222         )
 223       (princ "</ul>\n")
 224       ))
 225   ignored-chars)
 226
 227 (defun www-batch-ids-find ()
 228   (let ((components (car command-line-args-left))
 229         (coded-charset-entity-reference-alist
 230          (list*
 231           '(=cns11643-1         "C1-" 4 X)
 232           '(=cns11643-2         "C2-" 4 X)
 233           '(=cns11643-3         "C3-" 4 X)
 234           '(=cns11643-4         "C4-" 4 X)
 235           '(=cns11643-5         "C5-" 4 X)
 236           '(=cns11643-6         "C6-" 4 X)
 237           '(=cns11643-7         "C7-" 4 X)
 238           '(=gb2312             "G0-" 4 X)
 239           '(=gb12345            "G1-" 4 X)
 240           '(=jis-x0208@1990     "J90-" 4 X)
 241           '(=jis-x0212          "JSP-" 4 X)
 242           '(=cbeta              "CB" 5 d)
 243           '(=jef-china3         "JC3-" 4 X)
 244           '(=jis-x0208@1978     "J78-" 4 X)
 245           '(=jis-x0208@1983     "J83-" 4 X)
 246           '(=daikanwa           "M-" 5 d)
 247           coded-charset-entity-reference-alist))
 248         )
 249     (setq command-line-args-left (cdr command-line-args-left))
 250     (cond
 251      ((stringp components)
 252       (if (string-match "^components=" components)
 253           (setq components (substring components (match-end 0))))
 254       (setq components
 255             (if (> (length components) 0)
 256                 (decode-url-string components 'utf-8-er)
 257               nil))
 258       )
 259      (t
 260       (setq components nil)
 261       ))
 262     (princ "Content-Type: text/html; charset=UTF-8
 263
 264 <!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\"
 265             \"http://www.w3.org/TR/html4/loose.dtd\">
 266 <html lang=\"ja\">
 267 <head>
 268 <title>CHISE IDS Find</title>
 269 </head>
 270
 271 <body>
 272
 273 <h1>")
 274     (princ (encode-coding-string "CHISE IDS \e$B4A;z8!:w\e(B" 'utf-8-jp-er))
 275     (princ "</h1>")
 276     (princ "
 277 <p>Version ")
 278     (princ www-ids-find-version)
 279     (princ (format-time-string
 280             " (Last-modified: %Y-%m-%d %H:%M:%S)"
 281             (nth 5
 282                  (file-attributes
 283                   www-ids-find-ideographic-products-file-name))))
 284     (princ "
 285 <hr>
 286 <p>
 287 <form action=\"/ids-find\" method=\"GET\">
 288 ")
 289     (princ (encode-coding-string "\e$BItIJJ8;zNs\e(B" 'utf-8-jp-er))
 290     (princ " <input type=\"text\" name=\"components\" size=\"30\" maxlength=\"30\" value=\"")
 291     (if (> (length components) 0)
 292         (princ (encode-coding-string components 'utf-8-er)))
 293     (princ "\">
 294 <input type=\"submit\" value=\"")
 295     (princ (encode-coding-string "\e$B8!:w3+;O\e(B" 'utf-8-jp-er))
 296     (princ "\">
 297 </form>
 298
 299 ")
 300     (unless (file-newer-than-file-p
 301              www-ids-find-ideographic-products-file-name
 302              (locate-file (car command-line-args) exec-path))
 303       (princ (encode-coding-string "<hr>
 304 <p>
 305 \e$B8=:_!"%7%9%F%`$N99?7:n6HCf$G$9!#$7$P$i$/$*BT$A$/$@$5$$!#\e(B
 306 <hr>
 307 " 'utf-8-jp-er))
 308       ;; (setq components nil)
 309       )
 310     (cond
 311      (components
 312       ;; (map-char-attribute
 313       ;;  (lambda (c v)
 314       ;;    (when (every (lambda (p)
 315       ;;                   (ideographic-structure-member p v))
 316       ;;                 components)
 317       ;;      (princ (encode-coding-string
 318       ;;              (ids-find-format-line c v)
 319       ;;              'utf-8-jp-er))
 320       ;;      (princ "<br>\n")
 321       ;;      )
 322       ;;    nil)
 323       ;;  'ideographic-structure)
 324       (when (= (length components) 1)
 325         (www-ids-find-format-line (aref components 0)
 326                                   (char-feature (aref components 0)
 327                                                 'ideographic-structure)))
 328       ;; (dolist (c (ideographic-products-find components))
 329       ;;   (setq is (char-feature c 'ideographic-structure))
 330       ;;   ;; to avoid problems caused by wrong indexes
 331       ;;   (when (every (lambda (c)
 332       ;;                  (ideographic-structure-member c is))
 333       ;;                components)
 334       ;;     (www-ids-find-format-line c is)))
 335       ;; (princ "<ul>\n")
 336       (www-ids-insert-chars-including-components components)
 337       ;; (princ "</ul>\n")
 338       )
 339      (t
 340       (princ (encode-coding-string "<hr>
 341 <p>
 342 \e$B;XDj$7$?ItIJ$rA4$F4^$`4A;z$N0lMw$rI=<($7$^$9!#\e(B
 343 <p>
 344 CHISE \e$B$GMQ$$$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))
 345       (princ (encode-coding-string "
 346 <p>
 347 \[Links\]
 348 <ul>
 349 <li><a href=\"http://www.shuiren.org/chuden/toyoshi/syoseki/chise_ids.html\"
 350 >\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
 351 href=\"http://www.shuiren.org/\">\e$B?g?MDb\e(B</a>\e$B!K$K$h$k2r@b\e(B
 352 </ul>
 353 <ul>
 354 <li><a href=\"http://www.karitsu.org/tools/firefox_plugin.htm\"
 355 >Firefox \e$BMQ\e(B plugin</a> by \e$B=);3M[0lO:$5$s!J\e(B<a href=\"http://www.karitsu.org/\"
 356 >\e$B2aN)c7\e(B</a>\e$B!K\e(B
 357 </ul>
 358 <ul>
 359 <li><a href=\"http://cvs.m17n.org/viewcvs/chise/ids/www/www-ids-find.el?view=markup\"
 360 >www-ids-find.el (source file (Emacs Lisp part))
 361 <li><a href=\"http://chise.zinbun.kyoto-u.ac.jp/chise/ids/\"
 362 >\e$B!V\e(BCHISE \e$B4A;z9=B$>pJs%G!<%?%Y!<%9!W\e(B</a>
 363 <li><a href=\"http://fonts.jp/chise_linkmap/\"
 364 >\e$B!V\e(Bchise_linkmap : CHISE \e$B4A;zO"4D?^!W\e(B</a> by \e$B>eCO9(0l$5$s\e(B
 365 <li><a href=\"http://chise.zinbun.kyoto-u.ac.jp/chise/\"
 366 >CHISE Project</a>
 367 </ul>
 368 <ul>
 369 <li><a href=\"http://coe21.zinbun.kyoto-u.ac.jp/djvuchar\"
 370 >\e$B!VBsK\J8;z%G!<%?%Y!<%9!W\e(B</a> by
 371 <a href=\"http://coe21.zinbun.kyoto-u.ac.jp/\"
 372 >\e$B5~ETBg3X\e(B21\e$B@$5*\e(BCOE\e$B!VEl%"%8%"@$3&$N?MJ8>pJs3X8&5f650i5rE@!W\e(B</a>
 373 <li><a href=\"http://www.unicode.org/\"
 374 >Unicode</a>
 375 </ul>"
 376  'utf-8-jp-er))
 377
 378       ))
 379     (princ "<hr>")
 380     (princ "<p>
 381 Copyright (C) 2005, 2006, 2007, 2008, 2009, 2010 <a href=\"http://kanji.zinbun.kyoto-u.ac.jp/~tomo/\"
 382 >MORIOKA Tomohiko</a>")
 383     (princ
 384      (format
 385       "<p>Powered by <a
 386 href=\"http://chise.zinbun.kyoto-u.ac.jp/chise/xemacs/\"
 387 >XEmacs CHISE</a> %s."
 388       (encode-coding-string xemacs-chise-version 'utf-8-jp-er)))
 389     (princ "
 390 </body>
 391 </html>
 392 ")))