www/www-ids-find.el

   1 (require 'ids-find)
   2 (require 'cwiki-common)
   3
   4 (defun decode-url-string (string &optional coding-system)
   5   (if (> (length string) 0)
   6       (let ((i 0)
   7             dest)
   8         (while (string-match "%\\([0-9A-F][0-9A-F]\\)" string i)
   9           (setq dest (concat dest
  10                              (substring string i (match-beginning 0))
  11                              (char-to-string
  12                               (int-char
  13                                (string-to-int (match-string 1 string) 16))))
  14                 i (match-end 0)))
  15         (decode-coding-string
  16          (concat dest (substring string i))
  17          coding-system))))
  18
  19 (defconst www-ids-find-version "0.26")
  20
  21 (defvar www-ids-find-ideographic-products-file-name
  22   (expand-file-name "ideographic-products"
  23                     (expand-file-name
  24                      "feature"
  25                      (expand-file-name
  26                       "character"
  27                       chise-system-db-directory))))
  28
  29 (defvar www-ids-find-char-viewer-url
  30   "/est/view/character/")
  31
  32 (defvar www-ids-find-chise-link-map-url-prefix
  33   "http://fonts.jp/chise_linkmap/map.cgi?code=")
  34
  35 (defvar www-ids-find-tang-chars-file-name
  36   "~tomo/projects/chise/ids/www/tang-chars.udd")
  37
  38 (defun www-ids-find-format-char (c &optional code-desc)
  39   (princ
  40    (format "<a href=\"%s%s\">%s</a>"
  41            www-ids-find-char-viewer-url
  42            (www-uri-encode-object c)
  43            (www-format-encode-string (char-to-string c))))
  44   ;; (let ((str (encode-coding-string (format "%c" c) 'utf-8-er))
  45   ;;       plane code)
  46   ;;   (princ
  47   ;;    (with-temp-buffer
  48   ;;      (cond
  49   ;;       ((string-match "&CB\\([0-9]+\\);" str)
  50   ;;        (setq code (string-to-int (match-string 1 str)))
  51   ;;        (insert (format "<a href=\"%s"
  52   ;;                        www-ids-find-char-viewer-url))
  53   ;;        (insert str)
  54   ;;        (insert (format "\"><img alt=\"CB%05d\" src=\"/glyphs/cb-gaiji/%02d/CB%05d.gif\">\n"
  55   ;;                        code (/ code 1000) code))
  56   ;;        (when code-desc
  57   ;;          (insert (format "CB%05d</a>" code)))
  58   ;;        )
  59   ;;       ((string-match "&JC3-\\([0-9A-F]+\\);" str)
  60   ;;        (setq code (string-to-int (match-string 1 str) 16))
  61   ;;        (insert (format "<a href=\"%s"
  62   ;;                        www-ids-find-char-viewer-url))
  63   ;;        (insert str)
  64   ;;        (insert (format "\"><img alt=\"JC3-%04X\" src=\"http://kanji.zinbun.kyoto-u.ac.jp/db/CHINA3/Gaiji/%04x.gif\">\n"
  65   ;;                        code code))
  66   ;;        (when code-desc
  67   ;;          (insert (format "JC3-%04X</a>" code)))
  68   ;;        )
  69   ;;       ((string-match "&J\\(78\\|83\\|90\\|SP\\)-\\([0-9A-F]+\\);" str)
  70   ;;        (setq plane (match-string 1 str)
  71   ;;              code (string-to-int (match-string 2 str) 16))
  72   ;;        (insert (format "<a href=\"%s"
  73   ;;                        www-ids-find-char-viewer-url))
  74   ;;        (insert str)
  75   ;;        (insert (format "\"><img alt=\"J%s-%04X\" src=\"/glyphs/JIS-%s/%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 "J%s-%04X</a>" plane code)))
  81   ;;        )
  82   ;;       ((string-match "&G\\([01]\\)-\\([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 (format "<a href=\"%s"
  86   ;;                        www-ids-find-char-viewer-url))
  87   ;;        (insert str)
  88   ;;        (insert (format "\"><img alt=\"G%d-%04X\" src=\"/glyphs/GB%d/%02d-%02d.gif\">\n"
  89   ;;                        plane code plane
  90   ;;                        (- (lsh code -8) 32)
  91   ;;                        (- (logand code 255) 32)))
  92   ;;        (when code-desc
  93   ;;          (insert (format "G%d-%04X</a>" plane code)))
  94   ;;        )
  95   ;;       ((string-match "&C\\([1-7]\\)-\\([0-9A-F]+\\);" str)
  96   ;;        (setq plane (string-to-int (match-string 1 str))
  97   ;;              code (string-to-int (match-string 2 str) 16))
  98   ;;        (insert (format "<a href=\"%s"
  99   ;;                        www-ids-find-char-viewer-url))
 100   ;;        (insert str)
 101   ;;        (insert (format "\"><img alt=\"C%d-%04X\" src=\"/glyphs/CNS%d/%04X.gif\">\n"
 102   ;;                        plane code plane code))
 103   ;;        (when code-desc
 104   ;;          (insert (format "C%d-%04X</a>" plane code)))
 105   ;;        )
 106   ;;       ((string-match "&ZOB-\\([0-9]+\\);" str)
 107   ;;        (setq code (string-to-int (match-string 1 str)))
 108   ;;        (insert (format "<a href=\"%s"
 109   ;;                        www-ids-find-char-viewer-url))
 110   ;;        (insert str)
 111   ;;        (insert (format "\"><img alt=\"ZOB-%04d\" src=\"/glyphs/ZOB-1968/%04d.png\">\n"
 112   ;;                        code code))
 113   ;;        (when code-desc
 114   ;;          (insert (format "ZOB-%04d</a>" code)))
 115   ;;        )
 116   ;;       (t
 117   ;;        (insert (format "<a href=\"%s"
 118   ;;                        www-ids-find-char-viewer-url))
 119   ;;        ;; (insert str)
 120   ;;        (insert
 121   ;;         (mapconcat (lambda (c)
 122   ;;                      (if (<= (char-int c) #x7F)
 123   ;;                          (char-to-string c)
 124   ;;                        (format "%%%02X" c)))
 125   ;;                    str ""))
 126   ;;        (insert "\">")
 127   ;;        (insert str)
 128   ;;        (insert "</a>")
 129   ;;        ))
 130   ;;      (goto-char (point-min))
 131   ;;      (while (search-forward "&" nil t)
 132   ;;        (replace-match "&amp;" t 'literal))
 133   ;;      (buffer-string))))
 134   )
 135
 136 (defun www-ids-find-format-line (c is)
 137   (let (ucs len i ids)
 138     (www-ids-find-format-char c 'code-desc)
 139     (princ
 140      (or (if (setq ucs (or (char-ucs c)
 141                            (encode-char c 'ucs)))
 142              (format
 143               " <a href=\"http://www.unicode.org/cgi-bin/GetUnihanData.pl?codepoint=%X\">%s</a>"
 144               ucs
 145               (cond ((<= ucs #xFFFF)
 146                      (format "U+%04X" ucs))
 147                     ((<= ucs #x10FFFF)
 148                      (format "U-%08X" ucs))))
 149            "          ")))
 150     (when ucs
 151       (princ
 152        (format " <a href=\"%s%X\">(link map)</a>"
 153                www-ids-find-chise-link-map-url-prefix ucs)))
 154     (princ " ")
 155     (when is
 156       (setq ids (ideographic-structure-to-ids is))
 157       (setq i 0
 158             len (length ids))
 159       (while (< i len)
 160         (www-ids-find-format-char (aref ids i))
 161         (setq i (1+ i))))
 162     (when (and ucs
 163                (with-current-buffer
 164                    (find-file-noselect
 165                     www-ids-find-tang-chars-file-name)
 166                  (goto-char (point-min))
 167                  (re-search-forward (format "^%d$" ucs) nil t)))
 168       (princ
 169        (format " <a href=\"http://coe21.zinbun.kyoto-u.ac.jp/djvuchar?query=%s\">"
 170                (mapconcat
 171                 (lambda (c)
 172                   (format "%%%02X" (char-int c)))
 173                 (encode-coding-string (char-to-string c)
 174                                       'utf-8-jp)
 175                 "")))
 176       (princ (encode-coding-string "\e$B"M\e(B[\e$BEbBeBsK\\e(B]</a>" 'utf-8-jp-er)))
 177     (princ "<br>\n")))
 178
 179 (defun www-ids-insert-chars-including-components (components
 180                                                   &optional ignored-chars)
 181   (let ((products (ideographic-products-find components))
 182         is as bs len ignore-children)
 183     (setq len (length products))
 184     (when (>= len 1024)
 185       (setq ignore-children t)
 186       (princ
 187        (encode-coding-string
 188         "<p>\e$B7k2L$,B?$9$.$k$?$a!":F5"E*8!:w$r>JN,$7$^$7$?!#\e(B</p>"
 189         'utf-8-jp-er)))
 190     (if (>= len 2048)
 191         (dolist (c products)
 192           (www-ids-find-format-char c))
 193       (princ "<ul>\n")
 194       (dolist (c (cond
 195                   ;; ((>= len 2048)
 196                   ;;  (setq ignore-children t)
 197                   ;;  products)
 198                   ;; ((>= len 1024)
 199                   ;;  products)
 200                   ((>= len 1024)
 201                    (sort (copy-list products)
 202                          (lambda (a b)
 203                            (< (char-int a)(char-int b))))
 204                    )
 205                   ((>= len 512)
 206                    (sort (copy-list products)
 207                          (lambda (a b)
 208                            (if (setq as (char-total-strokes a))
 209                                (if (setq bs (char-total-strokes b))
 210                                    (if (= as bs)
 211                                        (< (char-int a)(char-int b))
 212                                      (< as bs))
 213                                  t)
 214                              (< (char-int a)(char-int b)))))
 215                    )
 216                   (t
 217                    (sort (copy-list products)
 218                          (lambda (a b)
 219                            (if (setq as (char-total-strokes a))
 220                                (if (setq bs (char-total-strokes b))
 221                                    (if (= as bs)
 222                                        (ideograph-char< a b)
 223                                      (< as bs))
 224                                  t)
 225                              (ideograph-char< a b))))
 226                    )))
 227         (unless (memq c ignored-chars)
 228           (setq is (char-feature c 'ideographic-structure))
 229           (princ "<li>")
 230           (www-ids-find-format-line c is)
 231           (unless ignore-children
 232             ;; (princ "<ul>\n")
 233             (setq ignored-chars
 234                   (www-ids-insert-chars-including-components
 235                    (char-to-string c)
 236                    (cons c ignored-chars)))
 237             ;; (princ "</ul>\n")
 238             ))
 239         )
 240       (princ "</ul>\n")
 241       ))
 242   ignored-chars)
 243
 244 (defun www-batch-ids-find ()
 245   (let ((components (car command-line-args-left))
 246         (coded-charset-entity-reference-alist
 247          (list*
 248           '(=cns11643-1         "C1-" 4 X)
 249           '(=cns11643-2         "C2-" 4 X)
 250           '(=cns11643-3         "C3-" 4 X)
 251           '(=cns11643-4         "C4-" 4 X)
 252           '(=cns11643-5         "C5-" 4 X)
 253           '(=cns11643-6         "C6-" 4 X)
 254           '(=cns11643-7         "C7-" 4 X)
 255           '(=gb2312             "G0-" 4 X)
 256           '(=gb12345            "G1-" 4 X)
 257           '(=jis-x0208@1990     "J90-" 4 X)
 258           '(=jis-x0212          "JSP-" 4 X)
 259           '(=cbeta              "CB" 5 d)
 260           '(=jef-china3         "JC3-" 4 X)
 261           '(=jis-x0208@1978     "J78-" 4 X)
 262           '(=jis-x0208@1983     "J83-" 4 X)
 263           '(=daikanwa           "M-" 5 d)
 264           coded-charset-entity-reference-alist))
 265         )
 266     (setq command-line-args-left (cdr command-line-args-left))
 267     (cond
 268      ((stringp components)
 269       (if (string-match "^components=" components)
 270           (setq components (substring components (match-end 0))))
 271       (setq components
 272             (if (> (length components) 0)
 273                 (decode-url-string components 'utf-8-er)
 274               nil))
 275       )
 276      (t
 277       (setq components nil)
 278       ))
 279     (princ "Content-Type: text/html; charset=UTF-8
 280
 281 <!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\"
 282             \"http://www.w3.org/TR/html4/loose.dtd\">
 283 <html lang=\"ja\">
 284 <head>
 285 <title>CHISE IDS Find</title>
 286 </head>
 287
 288 <body>
 289
 290 <h1>")
 291     (princ (encode-coding-string "CHISE IDS \e$B4A;z8!:w\e(B" 'utf-8-jp-er))
 292     (princ "</h1>")
 293     (princ "
 294 <p>Version ")
 295     (princ www-ids-find-version)
 296     (princ (format-time-string
 297             " (Last-modified: %Y-%m-%d %H:%M:%S)"
 298             (nth 5
 299                  (file-attributes
 300                   www-ids-find-ideographic-products-file-name))))
 301     (princ "
 302 <hr>
 303 <p>
 304 <form action=\"/ids-find\" method=\"GET\">
 305 ")
 306     (princ (encode-coding-string "\e$BItIJJ8;zNs\e(B" 'utf-8-jp-er))
 307     (princ " <input type=\"text\" name=\"components\" size=\"30\" maxlength=\"30\" value=\"")
 308     (if (> (length components) 0)
 309         (princ (encode-coding-string components 'utf-8-er)))
 310     (princ "\">
 311 <input type=\"submit\" value=\"")
 312     (princ (encode-coding-string "\e$B8!:w3+;O\e(B" 'utf-8-jp-er))
 313     (princ "\">
 314 </form>
 315
 316 ")
 317     (unless (file-newer-than-file-p
 318              www-ids-find-ideographic-products-file-name
 319              (locate-file (car command-line-args) exec-path))
 320       (princ (encode-coding-string "<hr>
 321 <p>
 322 \e$B8=:_!"%7%9%F%`$N99?7:n6HCf$G$9!#$7$P$i$/$*BT$A$/$@$5$$!#\e(B
 323 <hr>
 324 " 'utf-8-jp-er))
 325       ;; (setq components nil)
 326       )
 327     (cond
 328      (components
 329       ;; (map-char-attribute
 330       ;;  (lambda (c v)
 331       ;;    (when (every (lambda (p)
 332       ;;                   (ideographic-structure-member p v))
 333       ;;                 components)
 334       ;;      (princ (encode-coding-string
 335       ;;              (ids-find-format-line c v)
 336       ;;              'utf-8-jp-er))
 337       ;;      (princ "<br>\n")
 338       ;;      )
 339       ;;    nil)
 340       ;;  'ideographic-structure)
 341       (when (= (length components) 1)
 342         (www-ids-find-format-line (aref components 0)
 343                                   (char-feature (aref components 0)
 344                                                 'ideographic-structure)))
 345       ;; (dolist (c (ideographic-products-find components))
 346       ;;   (setq is (char-feature c 'ideographic-structure))
 347       ;;   ;; to avoid problems caused by wrong indexes
 348       ;;   (when (every (lambda (c)
 349       ;;                  (ideographic-structure-member c is))
 350       ;;                components)
 351       ;;     (www-ids-find-format-line c is)))
 352       ;; (princ "<ul>\n")
 353       (www-ids-insert-chars-including-components components)
 354       ;; (princ "</ul>\n")
 355       )
 356      (t
 357       (princ (encode-coding-string "<hr>
 358 <p>
 359 \e$B;XDj$7$?ItIJ$rA4$F4^$`4A;z$N0lMw$rI=<($7$^$9!#\e(B
 360 <p>
 361 CHISE \e$B$GMQ$$$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))
 362       (princ (encode-coding-string "
 363 <p>
 364 \[Links\]
 365 <ul>
 366 <li><a href=\"http://www.shuiren.org/chuden/toyoshi/syoseki/chise_ids.html\"
 367 >\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
 368 href=\"http://www.shuiren.org/\">\e$B?g?MDb\e(B</a>\e$B!K$K$h$k2r@b\e(B
 369 </ul>
 370 <ul>
 371 <li><a href=\"http://www.karitsu.org/tools/firefox_plugin.htm\"
 372 >Firefox \e$BMQ\e(B plugin</a> by \e$B=);3M[0lO:$5$s!J\e(B<a href=\"http://www.karitsu.org/\"
 373 >\e$B2aN)c7\e(B</a>\e$B!K\e(B
 374 </ul>
 375 <ul>
 376 <li><a href=\"http://cvs.m17n.org/viewcvs/chise/ids/www/www-ids-find.el?view=markup\"
 377 >www-ids-find.el (source file (Emacs Lisp part))
 378 <li><a href=\"http://www.chise.org/ids/\"
 379 >\e$B!V\e(BCHISE \e$B4A;z9=B$>pJs%G!<%?%Y!<%9!W\e(B</a>
 380 <li><a href=\"http://fonts.jp/chise_linkmap/\"
 381 >\e$B!V\e(Bchise_linkmap : CHISE \e$B4A;zO"4D?^!W\e(B</a> by \e$B>eCO9(0l$5$s\e(B
 382 <li><a href=\"http://www.chise.org/\"
 383 >CHISE Project</a>
 384 </ul>
 385 <ul>
 386 <li><a href=\"http://coe21.zinbun.kyoto-u.ac.jp/djvuchar\"
 387 >\e$B!VBsK\J8;z%G!<%?%Y!<%9!W\e(B</a> by
 388 <a href=\"http://coe21.zinbun.kyoto-u.ac.jp/\"
 389 >\e$B5~ETBg3X\e(B21\e$B@$5*\e(BCOE\e$B!VEl%"%8%"@$3&$N?MJ8>pJs3X8&5f650i5rE@!W\e(B</a>
 390 <li><a href=\"http://www.unicode.org/\"
 391 >Unicode</a>
 392 </ul>"
 393  'utf-8-jp-er))
 394
 395       ))
 396     (princ "<hr>")
 397     (princ "<p>
 398 Copyright (C) 2005, 2006, 2007, 2008, 2009, 2010, 2015 <a href=\"http://kanji.zinbun.kyoto-u.ac.jp/~tomo/\"
 399 >MORIOKA Tomohiko</a>")
 400     (princ
 401      (format
 402       "<p>Powered by <a
 403 href=\"http://www.chise.org/xemacs/\"
 404 >XEmacs CHISE</a> %s."
 405       (encode-coding-string xemacs-chise-version 'utf-8-jp-er)))
 406     (princ "
 407 </body>
 408 </html>
 409 ")))