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 "0.25.1")
  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-char-viewer-url
  29   "/chisewiki/view.cgi?char=")
  30
  31 (defvar www-ids-find-chise-link-map-url-prefix
  32   "http://fonts.jp/chise_linkmap/map.cgi?code=")
  33
  34 (defvar www-ids-find-tang-chars-file-name
  35   "~tomo/projects/chise/ids/www/tang-chars.udd")
  36
  37 (defun www-ids-find-format-char (c &optional code-desc)
  38   (let ((str (encode-coding-string (format "%c" c) 'utf-8-er))
  39         plane code)
  40     (princ
  41      (with-temp-buffer
  42        (cond
  43         ((string-match "&CB\\([0-9]+\\);" str)
  44          (setq code (string-to-int (match-string 1 str)))
  45          (insert (format "<a href=\"%s"
  46                          www-ids-find-char-viewer-url))
  47          (insert str)
  48          (insert (format "\"><img alt=\"CB%05d\" src=\"/glyphs/cb-gaiji/%02d/CB%05d.gif\">\n"
  49                          code (/ code 1000) code))
  50          (when code-desc
  51            (insert (format "CB%05d</a>" code)))
  52          )
  53         ((string-match "&JC3-\\([0-9A-F]+\\);" str)
  54          (setq code (string-to-int (match-string 1 str) 16))
  55          (insert (format "<a href=\"%s"
  56                          www-ids-find-char-viewer-url))
  57          (insert str)
  58          (insert (format "\"><img alt=\"JC3-%04X\" src=\"http://kanji.zinbun.kyoto-u.ac.jp/db/CHINA3/Gaiji/%04x.gif\">\n"
  59                          code code))
  60          (when code-desc
  61            (insert (format "JC3-%04X</a>" code)))
  62          )
  63         ((string-match "&J\\(78\\|83\\|90\\|SP\\)-\\([0-9A-F]+\\);" str)
  64          (setq plane (match-string 1 str)
  65                code (string-to-int (match-string 2 str) 16))
  66          (insert (format "<a href=\"%s"
  67                          www-ids-find-char-viewer-url))
  68          (insert str)
  69          (insert (format "\"><img alt=\"J%s-%04X\" src=\"/glyphs/JIS-%s/%02d-%02d.gif\">\n"
  70                          plane code plane
  71                          (- (lsh code -8) 32)
  72                          (- (logand code 255) 32)))
  73          (when code-desc
  74            (insert (format "J%s-%04X</a>" plane code)))
  75          )
  76         ((string-match "&G\\([01]\\)-\\([0-9A-F]+\\);" str)
  77          (setq plane (string-to-int (match-string 1 str))
  78                code (string-to-int (match-string 2 str) 16))
  79          (insert (format "<a href=\"%s"
  80                          www-ids-find-char-viewer-url))
  81          (insert str)
  82          (insert (format "\"><img alt=\"G%d-%04X\" src=\"/glyphs/GB%d/%02d-%02d.gif\">\n"
  83                          plane code plane
  84                          (- (lsh code -8) 32)
  85                          (- (logand code 255) 32)))
  86          (when code-desc
  87            (insert (format "G%d-%04X</a>" plane code)))
  88          )
  89         ((string-match "&C\\([1-7]\\)-\\([0-9A-F]+\\);" str)
  90          (setq plane (string-to-int (match-string 1 str))
  91                code (string-to-int (match-string 2 str) 16))
  92          (insert (format "<a href=\"%s"
  93                          www-ids-find-char-viewer-url))
  94          (insert str)
  95          (insert (format "\"><img alt=\"C%d-%04X\" src=\"/glyphs/CNS%d/%04X.gif\">\n"
  96                          plane code plane code))
  97          (when code-desc
  98            (insert (format "C%d-%04X</a>" plane code)))
  99          )
 100         ((string-match "&ZOB-\\([0-9]+\\);" str)
 101          (setq code (string-to-int (match-string 1 str)))
 102          (insert (format "<a href=\"%s"
 103                          www-ids-find-char-viewer-url))
 104          (insert str)
 105          (insert (format "\"><img alt=\"ZOB-%04d\" src=\"/glyphs/ZOB-1968/%04d.png\">\n"
 106                          code code))
 107          (when code-desc
 108            (insert (format "ZOB-%04d</a>" code)))
 109          )
 110         (t
 111          (insert (format "<a href=\"%s"
 112                          www-ids-find-char-viewer-url))
 113          ;; (insert str)
 114          (insert
 115           (mapconcat (lambda (c)
 116                        (if (<= (char-int c) #x7F)
 117                            (char-to-string c)
 118                          (format "%%%02X" c)))
 119                      str ""))
 120          (insert "\">")
 121          (insert str)
 122          (insert "</a>")
 123          ))
 124        (goto-char (point-min))
 125        (while (search-forward "&" nil t)
 126          (replace-match "&amp;" t 'literal))
 127        (buffer-string)))))
 128
 129 (defun www-ids-find-format-line (c is)
 130   (let (ucs len i ids)
 131     (www-ids-find-format-char c 'code-desc)
 132     (princ
 133      (or (if (setq ucs (or (char-ucs c)
 134                            (encode-char c 'ucs)))
 135              (format
 136               " <a href=\"http://www.unicode.org/cgi-bin/GetUnihanData.pl?codepoint=%X\">%s</a>"
 137               ucs
 138               (cond ((<= ucs #xFFFF)
 139                      (format "U+%04X" ucs))
 140                     ((<= ucs #x10FFFF)
 141                      (format "U-%08X" ucs))))
 142            "          ")))
 143     (when ucs
 144       (princ
 145        (format " <a href=\"%s%X\">(link map)</a>"
 146                www-ids-find-chise-link-map-url-prefix ucs)))
 147     (princ " ")
 148     (when is
 149       (setq ids (ideographic-structure-to-ids is))
 150       (setq i 0
 151             len (length ids))
 152       (while (< i len)
 153         (www-ids-find-format-char (aref ids i))
 154         (setq i (1+ i))))
 155     (when (and ucs
 156                (with-current-buffer
 157                    (find-file-noselect
 158                     www-ids-find-tang-chars-file-name)
 159                  (goto-char (point-min))
 160                  (re-search-forward (format "^%d$" ucs) nil t)))
 161       (princ
 162        (format " <a href=\"http://coe21.zinbun.kyoto-u.ac.jp/djvuchar?query=%s\">"
 163                (mapconcat
 164                 (lambda (c)
 165                   (format "%%%02X" (char-int c)))
 166                 (encode-coding-string (char-to-string c)
 167                                       'utf-8-jp)
 168                 "")))
 169       (princ (encode-coding-string "\e$B"M\e(B[\e$BEbBeBsK\\e(B]</a>" 'utf-8-jp-er)))
 170     (princ "<br>\n")))
 171
 172 (defun www-ids-insert-chars-including-components (components
 173                                                   &optional ignored-chars)
 174   (let ((products (ideographic-products-find components))
 175         is as bs len ignore-children)
 176     (setq len (length products))
 177     (when (>= len 1024)
 178       (setq ignore-children t)
 179       (princ
 180        (encode-coding-string
 181         "<p>\e$B7k2L$,B?$9$.$k$?$a!":F5"E*8!:w$r>JN,$7$^$7$?!#\e(B</p>"
 182         'utf-8-jp-er)))
 183     (if (>= len 2048)
 184         (dolist (c products)
 185           (www-ids-find-format-char c))
 186       (princ "<ul>\n")
 187       (dolist (c (cond
 188                   ;; ((>= len 2048)
 189                   ;;  (setq ignore-children t)
 190                   ;;  products)
 191                   ;; ((>= len 1024)
 192                   ;;  products)
 193                   ((>= len 1024)
 194                    (sort (copy-list products)
 195                          (lambda (a b)
 196                            (< (char-int a)(char-int b))))
 197                    )
 198                   ((>= len 512)
 199                    (sort (copy-list products)
 200                          (lambda (a b)
 201                            (if (setq as (char-total-strokes a))
 202                                (if (setq bs (char-total-strokes b))
 203                                    (if (= as bs)
 204                                        (< (char-int a)(char-int b))
 205                                      (< as bs))
 206                                  t)
 207                              (< (char-int a)(char-int b)))))
 208                    )
 209                   (t
 210                    (sort (copy-list products)
 211                          (lambda (a b)
 212                            (if (setq as (char-total-strokes a))
 213                                (if (setq bs (char-total-strokes b))
 214                                    (if (= as bs)
 215                                        (ideograph-char< a b)
 216                                      (< as bs))
 217                                  t)
 218                              (ideograph-char< a b))))
 219                    )))
 220         (unless (memq c ignored-chars)
 221           (setq is (char-feature c 'ideographic-structure))
 222           (princ "<li>")
 223           (www-ids-find-format-line c is)
 224           (unless ignore-children
 225             ;; (princ "<ul>\n")
 226             (setq ignored-chars
 227                   (www-ids-insert-chars-including-components
 228                    (char-to-string c)
 229                    (cons c ignored-chars)))
 230             ;; (princ "</ul>\n")
 231             ))
 232         )
 233       (princ "</ul>\n")
 234       ))
 235   ignored-chars)
 236
 237 (defun www-batch-ids-find ()
 238   (let ((components (car command-line-args-left))
 239         (coded-charset-entity-reference-alist
 240          (list*
 241           '(=cns11643-1         "C1-" 4 X)
 242           '(=cns11643-2         "C2-" 4 X)
 243           '(=cns11643-3         "C3-" 4 X)
 244           '(=cns11643-4         "C4-" 4 X)
 245           '(=cns11643-5         "C5-" 4 X)
 246           '(=cns11643-6         "C6-" 4 X)
 247           '(=cns11643-7         "C7-" 4 X)
 248           '(=gb2312             "G0-" 4 X)
 249           '(=gb12345            "G1-" 4 X)
 250           '(=jis-x0208@1990     "J90-" 4 X)
 251           '(=jis-x0212          "JSP-" 4 X)
 252           '(=cbeta              "CB" 5 d)
 253           '(=jef-china3         "JC3-" 4 X)
 254           '(=jis-x0208@1978     "J78-" 4 X)
 255           '(=jis-x0208@1983     "J83-" 4 X)
 256           '(=daikanwa           "M-" 5 d)
 257           coded-charset-entity-reference-alist))
 258         )
 259     (setq command-line-args-left (cdr command-line-args-left))
 260     (cond
 261      ((stringp components)
 262       (if (string-match "^components=" components)
 263           (setq components (substring components (match-end 0))))
 264       (setq components
 265             (if (> (length components) 0)
 266                 (decode-url-string components 'utf-8-er)
 267               nil))
 268       )
 269      (t
 270       (setq components nil)
 271       ))
 272     (princ "Content-Type: text/html; charset=UTF-8
 273
 274 <!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\"
 275             \"http://www.w3.org/TR/html4/loose.dtd\">
 276 <html lang=\"ja\">
 277 <head>
 278 <title>CHISE IDS Find</title>
 279 </head>
 280
 281 <body>
 282
 283 <h1>")
 284     (princ (encode-coding-string "CHISE IDS \e$B4A;z8!:w\e(B" 'utf-8-jp-er))
 285     (princ "</h1>")
 286     (princ "
 287 <p>Version ")
 288     (princ www-ids-find-version)
 289     (princ (format-time-string
 290             " (Last-modified: %Y-%m-%d %H:%M:%S)"
 291             (nth 5
 292                  (file-attributes
 293                   www-ids-find-ideographic-products-file-name))))
 294     (princ "
 295 <hr>
 296 <p>
 297 <form action=\"/ids-find\" method=\"GET\">
 298 ")
 299     (princ (encode-coding-string "\e$BItIJJ8;zNs\e(B" 'utf-8-jp-er))
 300     (princ " <input type=\"text\" name=\"components\" size=\"30\" maxlength=\"30\" value=\"")
 301     (if (> (length components) 0)
 302         (princ (encode-coding-string components 'utf-8-er)))
 303     (princ "\">
 304 <input type=\"submit\" value=\"")
 305     (princ (encode-coding-string "\e$B8!:w3+;O\e(B" 'utf-8-jp-er))
 306     (princ "\">
 307 </form>
 308
 309 ")
 310     (unless (file-newer-than-file-p
 311              www-ids-find-ideographic-products-file-name
 312              (locate-file (car command-line-args) exec-path))
 313       (princ (encode-coding-string "<hr>
 314 <p>
 315 \e$B8=:_!"%7%9%F%`$N99?7:n6HCf$G$9!#$7$P$i$/$*BT$A$/$@$5$$!#\e(B
 316 <hr>
 317 " 'utf-8-jp-er))
 318       ;; (setq components nil)
 319       )
 320     (cond
 321      (components
 322       ;; (map-char-attribute
 323       ;;  (lambda (c v)
 324       ;;    (when (every (lambda (p)
 325       ;;                   (ideographic-structure-member p v))
 326       ;;                 components)
 327       ;;      (princ (encode-coding-string
 328       ;;              (ids-find-format-line c v)
 329       ;;              'utf-8-jp-er))
 330       ;;      (princ "<br>\n")
 331       ;;      )
 332       ;;    nil)
 333       ;;  'ideographic-structure)
 334       (when (= (length components) 1)
 335         (www-ids-find-format-line (aref components 0)
 336                                   (char-feature (aref components 0)
 337                                                 'ideographic-structure)))
 338       ;; (dolist (c (ideographic-products-find components))
 339       ;;   (setq is (char-feature c 'ideographic-structure))
 340       ;;   ;; to avoid problems caused by wrong indexes
 341       ;;   (when (every (lambda (c)
 342       ;;                  (ideographic-structure-member c is))
 343       ;;                components)
 344       ;;     (www-ids-find-format-line c is)))
 345       ;; (princ "<ul>\n")
 346       (www-ids-insert-chars-including-components components)
 347       ;; (princ "</ul>\n")
 348       )
 349      (t
 350       (princ (encode-coding-string "<hr>
 351 <p>
 352 \e$B;XDj$7$?ItIJ$rA4$F4^$`4A;z$N0lMw$rI=<($7$^$9!#\e(B
 353 <p>
 354 CHISE \e$B$GMQ$$$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))
 355       (princ (encode-coding-string "
 356 <p>
 357 \[Links\]
 358 <ul>
 359 <li><a href=\"http://www.shuiren.org/chuden/toyoshi/syoseki/chise_ids.html\"
 360 >\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
 361 href=\"http://www.shuiren.org/\">\e$B?g?MDb\e(B</a>\e$B!K$K$h$k2r@b\e(B
 362 </ul>
 363 <ul>
 364 <li><a href=\"http://www.karitsu.org/tools/firefox_plugin.htm\"
 365 >Firefox \e$BMQ\e(B plugin</a> by \e$B=);3M[0lO:$5$s!J\e(B<a href=\"http://www.karitsu.org/\"
 366 >\e$B2aN)c7\e(B</a>\e$B!K\e(B
 367 </ul>
 368 <ul>
 369 <li><a href=\"http://cvs.m17n.org/viewcvs/chise/ids/www/www-ids-find.el?view=markup\"
 370 >www-ids-find.el (source file (Emacs Lisp part))
 371 <li><a href=\"http://www.chise.org/ids/\"
 372 >\e$B!V\e(BCHISE \e$B4A;z9=B$>pJs%G!<%?%Y!<%9!W\e(B</a>
 373 <li><a href=\"http://fonts.jp/chise_linkmap/\"
 374 >\e$B!V\e(Bchise_linkmap : CHISE \e$B4A;zO"4D?^!W\e(B</a> by \e$B>eCO9(0l$5$s\e(B
 375 <li><a href=\"http://www.chise.org/\"
 376 >CHISE Project</a>
 377 </ul>
 378 <ul>
 379 <li><a href=\"http://coe21.zinbun.kyoto-u.ac.jp/djvuchar\"
 380 >\e$B!VBsK\J8;z%G!<%?%Y!<%9!W\e(B</a> by
 381 <a href=\"http://coe21.zinbun.kyoto-u.ac.jp/\"
 382 >\e$B5~ETBg3X\e(B21\e$B@$5*\e(BCOE\e$B!VEl%"%8%"@$3&$N?MJ8>pJs3X8&5f650i5rE@!W\e(B</a>
 383 <li><a href=\"http://www.unicode.org/\"
 384 >Unicode</a>
 385 </ul>"
 386  'utf-8-jp-er))
 387
 388       ))
 389     (princ "<hr>")
 390     (princ "<p>
 391 Copyright (C) 2005, 2006, 2007, 2008, 2009, 2010 <a href=\"http://kanji.zinbun.kyoto-u.ac.jp/~tomo/\"
 392 >MORIOKA Tomohiko</a>")
 393     (princ
 394      (format
 395       "<p>Powered by <a
 396 href=\"http://www.chise.org/xemacs/\"
 397 >XEmacs CHISE</a> %s."
 398       (encode-coding-string xemacs-chise-version 'utf-8-jp-er)))
 399     (princ "
 400 </body>
 401 </html>
 402 ")))