www/www-hng-ids-find.el

   1 (require 'ids-find)
   2 (require 'cwiki-common)
   3
   4 (defvar hng-ccs-list
   5   (let (dest)
   6     (dolist (ccs (charset-list))
   7       (when (string-match "^===hng-" (symbol-name ccs))
   8         (setq dest (cons ccs dest))))
   9     dest))
  10
  11 (defun char-hng-p (char)
  12   (or (get-char-attribute char '->HNG)
  13       (char-have-hng-p char)))
  14
  15 (defun char-have-hng-p (char)
  16   (or (some (lambda (ccs)
  17               (and (encode-char char ccs)
  18                    char))
  19             hng-ccs-list)
  20       (some #'char-have-hng-p
  21             (get-char-attribute char '->subsumptive))
  22       (some #'char-have-hng-p
  23             (get-char-attribute char '->denotational))))
  24
  25 (defun decode-url-string (string &optional coding-system)
  26   (if (> (length string) 0)
  27       (let ((i 0)
  28             dest)
  29         (while (string-match "%\\([0-9A-F][0-9A-F]\\)" string i)
  30           (setq dest (concat dest
  31                              (substring string i (match-beginning 0))
  32                              (char-to-string
  33                               (int-char
  34                                (string-to-int (match-string 1 string) 16))))
  35                 i (match-end 0)))
  36         (decode-coding-string
  37          (concat dest (substring string i))
  38          coding-system))))
  39
  40 (defconst www-hng-ids-find-version "0.26")
  41
  42 (defvar www-ids-find-ideographic-products-file-name
  43   (expand-file-name "ideographic-products"
  44                     (expand-file-name
  45                      "feature"
  46                      (expand-file-name
  47                       "character"
  48                       chise-system-db-directory))))
  49
  50 (defvar www-ids-find-char-viewer-url
  51   "/est/view/character/")
  52
  53 (defvar www-ids-find-chise-link-map-url-prefix
  54   "http://fonts.jp/chise_linkmap/map.cgi?code=")
  55
  56 (defvar www-ids-find-tang-chars-file-name
  57   "~tomo/projects/chise/ids/www/tang-chars.udd")
  58
  59 (defun www-ids-find-format-char (c &optional code-desc)
  60   (princ
  61    (format "<a href=\"%s%s\">%s</a>"
  62            www-ids-find-char-viewer-url
  63            (www-uri-encode-object c)
  64            (www-format-encode-string (char-to-string c))))
  65   ;; (let ((str (encode-coding-string (format "%c" c) 'utf-8-er))
  66   ;;       plane code)
  67   ;;   (princ
  68   ;;    (with-temp-buffer
  69   ;;      (cond
  70   ;;       ((string-match "&CB\\([0-9]+\\);" str)
  71   ;;        (setq code (string-to-int (match-string 1 str)))
  72   ;;        (insert (format "<a href=\"%s"
  73   ;;                        www-ids-find-char-viewer-url))
  74   ;;        (insert str)
  75   ;;        (insert (format "\"><img alt=\"CB%05d\" src=\"/glyphs/cb-gaiji/%02d/CB%05d.gif\">\n"
  76   ;;                        code (/ code 1000) code))
  77   ;;        (when code-desc
  78   ;;          (insert (format "CB%05d</a>" code)))
  79   ;;        )
  80   ;;       ((string-match "&JC3-\\([0-9A-F]+\\);" str)
  81   ;;        (setq code (string-to-int (match-string 1 str) 16))
  82   ;;        (insert (format "<a href=\"%s"
  83   ;;                        www-ids-find-char-viewer-url))
  84   ;;        (insert str)
  85   ;;        (insert (format "\"><img alt=\"JC3-%04X\" src=\"http://kanji.zinbun.kyoto-u.ac.jp/db/CHINA3/Gaiji/%04x.gif\">\n"
  86   ;;                        code code))
  87   ;;        (when code-desc
  88   ;;          (insert (format "JC3-%04X</a>" code)))
  89   ;;        )
  90   ;;       ((string-match "&J\\(78\\|83\\|90\\|SP\\)-\\([0-9A-F]+\\);" str)
  91   ;;        (setq plane (match-string 1 str)
  92   ;;              code (string-to-int (match-string 2 str) 16))
  93   ;;        (insert (format "<a href=\"%s"
  94   ;;                        www-ids-find-char-viewer-url))
  95   ;;        (insert str)
  96   ;;        (insert (format "\"><img alt=\"J%s-%04X\" src=\"/glyphs/JIS-%s/%02d-%02d.gif\">\n"
  97   ;;                        plane code plane
  98   ;;                        (- (lsh code -8) 32)
  99   ;;                        (- (logand code 255) 32)))
 100   ;;        (when code-desc
 101   ;;          (insert (format "J%s-%04X</a>" plane code)))
 102   ;;        )
 103   ;;       ((string-match "&G\\([01]\\)-\\([0-9A-F]+\\);" str)
 104   ;;        (setq plane (string-to-int (match-string 1 str))
 105   ;;              code (string-to-int (match-string 2 str) 16))
 106   ;;        (insert (format "<a href=\"%s"
 107   ;;                        www-ids-find-char-viewer-url))
 108   ;;        (insert str)
 109   ;;        (insert (format "\"><img alt=\"G%d-%04X\" src=\"/glyphs/GB%d/%02d-%02d.gif\">\n"
 110   ;;                        plane code plane
 111   ;;                        (- (lsh code -8) 32)
 112   ;;                        (- (logand code 255) 32)))
 113   ;;        (when code-desc
 114   ;;          (insert (format "G%d-%04X</a>" plane code)))
 115   ;;        )
 116   ;;       ((string-match "&C\\([1-7]\\)-\\([0-9A-F]+\\);" str)
 117   ;;        (setq plane (string-to-int (match-string 1 str))
 118   ;;              code (string-to-int (match-string 2 str) 16))
 119   ;;        (insert (format "<a href=\"%s"
 120   ;;                        www-ids-find-char-viewer-url))
 121   ;;        (insert str)
 122   ;;        (insert (format "\"><img alt=\"C%d-%04X\" src=\"/glyphs/CNS%d/%04X.gif\">\n"
 123   ;;                        plane code plane code))
 124   ;;        (when code-desc
 125   ;;          (insert (format "C%d-%04X</a>" plane code)))
 126   ;;        )
 127   ;;       ((string-match "&ZOB-\\([0-9]+\\);" str)
 128   ;;        (setq code (string-to-int (match-string 1 str)))
 129   ;;        (insert (format "<a href=\"%s"
 130   ;;                        www-ids-find-char-viewer-url))
 131   ;;        (insert str)
 132   ;;        (insert (format "\"><img alt=\"ZOB-%04d\" src=\"/glyphs/ZOB-1968/%04d.png\">\n"
 133   ;;                        code code))
 134   ;;        (when code-desc
 135   ;;          (insert (format "ZOB-%04d</a>" code)))
 136   ;;        )
 137   ;;       (t
 138   ;;        (insert (format "<a href=\"%s"
 139   ;;                        www-ids-find-char-viewer-url))
 140   ;;        ;; (insert str)
 141   ;;        (insert
 142   ;;         (mapconcat (lambda (c)
 143   ;;                      (if (<= (char-int c) #x7F)
 144   ;;                          (char-to-string c)
 145   ;;                        (format "%%%02X" c)))
 146   ;;                    str ""))
 147   ;;        (insert "\">")
 148   ;;        (insert str)
 149   ;;        (insert "</a>")
 150   ;;        ))
 151   ;;      (goto-char (point-min))
 152   ;;      (while (search-forward "&" nil t)
 153   ;;        (replace-match "&amp;" t 'literal))
 154   ;;      (buffer-string))))
 155   )
 156
 157 (defun www-ids-find-format-line (c is)
 158   (let (ucs len i ids)
 159     (www-ids-find-format-char c 'code-desc)
 160     (princ
 161      (or (if (setq ucs (or (char-ucs c)
 162                            (encode-char c 'ucs)))
 163              (format
 164               " <a href=\"http://www.unicode.org/cgi-bin/GetUnihanData.pl?codepoint=%X\">%s</a>"
 165               ucs
 166               (cond ((<= ucs #xFFFF)
 167                      (format "U+%04X" ucs))
 168                     ((<= ucs #x10FFFF)
 169                      (format "U-%08X" ucs))))
 170            "          ")))
 171     (when ucs
 172       (princ
 173        (format " <a href=\"%s%X\">(link map)</a>"
 174                www-ids-find-chise-link-map-url-prefix ucs)))
 175     (princ " ")
 176     (when is
 177       (setq ids (ideographic-structure-to-ids is))
 178       (setq i 0
 179             len (length ids))
 180       (while (< i len)
 181         (www-ids-find-format-char (aref ids i))
 182         (setq i (1+ i))))
 183     (when (and ucs
 184                (with-current-buffer
 185                    (find-file-noselect
 186                     www-ids-find-tang-chars-file-name)
 187                  (goto-char (point-min))
 188                  (re-search-forward (format "^%d$" ucs) nil t)))
 189       (princ
 190        (format " <a href=\"http://coe21.zinbun.kyoto-u.ac.jp/djvuchar?query=%s\">"
 191                (mapconcat
 192                 (lambda (c)
 193                   (format "%%%02X" (char-int c)))
 194                 (encode-coding-string (char-to-string c)
 195                                       'utf-8-jp)
 196                 "")))
 197       (princ (encode-coding-string "\e$B"M\e(B[\e$BEbBeBsK\\e(B]</a>" 'utf-8-jp-er)))
 198     (princ "<br>\n")))
 199
 200 (defun www-ids-insert-chars-including-components (components
 201                                                   &optional ignored-chars)
 202   (let ((ret (ideographic-products-find components))
 203         products
 204         is as bs len ignore-children)
 205     (dolist (char ret)
 206       (if (char-hng-p char)
 207           (setq products (cons char products))))
 208     (setq len (length products))
 209     (when (>= len 1024)
 210       (setq ignore-children t)
 211       (princ
 212        (encode-coding-string
 213         "<p>\e$B7k2L$,B?$9$.$k$?$a!":F5"E*8!:w$r>JN,$7$^$7$?!#\e(B</p>"
 214         'utf-8-jp-er)))
 215     (if (>= len 2048)
 216         (dolist (c products)
 217           (www-ids-find-format-char c))
 218       (princ "<ul>\n")
 219       (dolist (c (cond
 220                   ;; ((>= len 2048)
 221                   ;;  (setq ignore-children t)
 222                   ;;  products)
 223                   ;; ((>= len 1024)
 224                   ;;  products)
 225                   ((>= len 1024)
 226                    (sort (copy-list products)
 227                          (lambda (a b)
 228                            (< (char-int a)(char-int b))))
 229                    )
 230                   ((>= len 512)
 231                    (sort (copy-list products)
 232                          (lambda (a b)
 233                            (if (setq as (char-total-strokes a))
 234                                (if (setq bs (char-total-strokes b))
 235                                    (if (= as bs)
 236                                        (< (char-int a)(char-int b))
 237                                      (< as bs))
 238                                  t)
 239                              (< (char-int a)(char-int b)))))
 240                    )
 241                   (t
 242                    (sort (copy-list products)
 243                          (lambda (a b)
 244                            (if (setq as (char-total-strokes a))
 245                                (if (setq bs (char-total-strokes b))
 246                                    (if (= as bs)
 247                                        (ideograph-char< a b)
 248                                      (< as bs))
 249                                  t)
 250                              (ideograph-char< a b))))
 251                    )))
 252         (unless (memq c ignored-chars)
 253           (setq is (char-feature c 'ideographic-structure))
 254           (princ "<li>")
 255           (www-ids-find-format-line c is)
 256           (unless ignore-children
 257             ;; (princ "<ul>\n")
 258             (setq ignored-chars
 259                   (www-ids-insert-chars-including-components
 260                    (char-to-string c)
 261                    (cons c ignored-chars)))
 262             ;; (princ "</ul>\n")
 263             ))
 264         )
 265       (princ "</ul>\n")
 266       ))
 267   ignored-chars)
 268
 269 (defun www-batch-ids-find ()
 270   (let ((components (car command-line-args-left))
 271         (coded-charset-entity-reference-alist
 272          (list*
 273           '(=cns11643-1         "C1-" 4 X)
 274           '(=cns11643-2         "C2-" 4 X)
 275           '(=cns11643-3         "C3-" 4 X)
 276           '(=cns11643-4         "C4-" 4 X)
 277           '(=cns11643-5         "C5-" 4 X)
 278           '(=cns11643-6         "C6-" 4 X)
 279           '(=cns11643-7         "C7-" 4 X)
 280           '(=gb2312             "G0-" 4 X)
 281           '(=gb12345            "G1-" 4 X)
 282           '(=jis-x0208@1990     "J90-" 4 X)
 283           '(=jis-x0212          "JSP-" 4 X)
 284           '(=cbeta              "CB" 5 d)
 285           '(=jef-china3         "JC3-" 4 X)
 286           '(=jis-x0208@1978     "J78-" 4 X)
 287           '(=jis-x0208@1983     "J83-" 4 X)
 288           '(=daikanwa           "M-" 5 d)
 289           coded-charset-entity-reference-alist))
 290         )
 291     (setq command-line-args-left (cdr command-line-args-left))
 292     (cond
 293      ((stringp components)
 294       (if (string-match "^components=" components)
 295           (setq components (substring components (match-end 0))))
 296       (setq components
 297             (if (> (length components) 0)
 298                 (decode-url-string components 'utf-8-er)
 299               nil))
 300       )
 301      (t
 302       (setq components nil)
 303       ))
 304     (princ "Content-Type: text/html; charset=UTF-8
 305
 306 <!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\"
 307             \"http://www.w3.org/TR/html4/loose.dtd\">
 308 <html lang=\"ja\">
 309 <head>
 310 <title>CHISE IDS Find</title>
 311 </head>
 312
 313 <body>
 314
 315 <h1>")
 316     (princ (encode-coding-string "CHISE-IDS HNG \e$B4A;z8!:w\e(B" 'utf-8-jp-er))
 317     (princ "</h1>")
 318     (princ "
 319 <p>Version ")
 320     (princ www-hng-ids-find-version)
 321     (princ (format-time-string
 322             " (Last-modified: %Y-%m-%d %H:%M:%S)"
 323             (nth 5
 324                  (file-attributes
 325                   www-ids-find-ideographic-products-file-name))))
 326     (princ "
 327 <hr>
 328 <p>
 329 <form action=\"/hng-ids-find\" method=\"GET\">
 330 ")
 331     (princ (encode-coding-string "\e$BItIJJ8;zNs\e(B" 'utf-8-jp-er))
 332     (princ " <input type=\"text\" name=\"components\" size=\"30\" maxlength=\"30\" value=\"")
 333     (if (> (length components) 0)
 334         (princ (encode-coding-string components 'utf-8-er)))
 335     (princ "\">
 336 <input type=\"submit\" value=\"")
 337     (princ (encode-coding-string "\e$B8!:w3+;O\e(B" 'utf-8-jp-er))
 338     (princ "\">
 339 </form>
 340
 341 ")
 342     (unless (file-newer-than-file-p
 343              www-ids-find-ideographic-products-file-name
 344              (locate-file (car command-line-args) exec-path))
 345       (princ (encode-coding-string "<hr>
 346 <p>
 347 \e$B8=:_!"%7%9%F%`$N99?7:n6HCf$G$9!#$7$P$i$/$*BT$A$/$@$5$$!#\e(B
 348 <hr>
 349 " 'utf-8-jp-er))
 350       ;; (setq components nil)
 351       )
 352     (cond
 353      (components
 354       ;; (map-char-attribute
 355       ;;  (lambda (c v)
 356       ;;    (when (every (lambda (p)
 357       ;;                   (ideographic-structure-member p v))
 358       ;;                 components)
 359       ;;      (princ (encode-coding-string
 360       ;;              (ids-find-format-line c v)
 361       ;;              'utf-8-jp-er))
 362       ;;      (princ "<br>\n")
 363       ;;      )
 364       ;;    nil)
 365       ;;  'ideographic-structure)
 366       (when (= (length components) 1)
 367         (www-ids-find-format-line (aref components 0)
 368                                   (char-feature (aref components 0)
 369                                                 'ideographic-structure)))
 370       ;; (dolist (c (ideographic-products-find components))
 371       ;;   (setq is (char-feature c 'ideographic-structure))
 372       ;;   ;; to avoid problems caused by wrong indexes
 373       ;;   (when (every (lambda (c)
 374       ;;                  (ideographic-structure-member c is))
 375       ;;                components)
 376       ;;     (www-ids-find-format-line c is)))
 377       ;; (princ "<ul>\n")
 378       (www-ids-insert-chars-including-components components)
 379       ;; (princ "</ul>\n")
 380       )
 381      (t
 382       (princ (encode-coding-string "<hr>
 383 <p>
 384 \e$B;XDj$7$?ItIJ$rA4$F4^$`4A;z$N0lMw$rI=<($7$^$9!#\e(B
 385 <p>
 386 CHISE \e$B$GMQ$$$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))
 387       (princ (encode-coding-string "
 388 <p>
 389 \[Links\]
 390 <ul>
 391 <li><a href=\"http://www.shuiren.org/chuden/toyoshi/syoseki/chise_ids.html\"
 392 >\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
 393 href=\"http://www.shuiren.org/\">\e$B?g?MDb\e(B</a>\e$B!K$K$h$k2r@b\e(B
 394 </ul>
 395 <ul>
 396 <li><a href=\"http://www.karitsu.org/tools/firefox_plugin.htm\"
 397 >Firefox \e$BMQ\e(B plugin</a> by \e$B=);3M[0lO:$5$s!J\e(B<a href=\"http://www.karitsu.org/\"
 398 >\e$B2aN)c7\e(B</a>\e$B!K\e(B
 399 </ul>
 400 <ul>
 401 <li><a href=\"http://cvs.m17n.org/viewcvs/chise/ids/www/www-ids-find.el?view=markup\"
 402 >www-ids-find.el (source file (Emacs Lisp part))
 403 <li><a href=\"http://www.chise.org/ids/\"
 404 >\e$B!V\e(BCHISE \e$B4A;z9=B$>pJs%G!<%?%Y!<%9!W\e(B</a>
 405 <li><a href=\"http://fonts.jp/chise_linkmap/\"
 406 >\e$B!V\e(Bchise_linkmap : CHISE \e$B4A;zO"4D?^!W\e(B</a> by \e$B>eCO9(0l$5$s\e(B
 407 <li><a href=\"http://www.chise.org/\"
 408 >CHISE Project</a>
 409 </ul>
 410 <ul>
 411 <li><a href=\"http://coe21.zinbun.kyoto-u.ac.jp/djvuchar\"
 412 >\e$B!VBsK\J8;z%G!<%?%Y!<%9!W\e(B</a> by
 413 <a href=\"http://coe21.zinbun.kyoto-u.ac.jp/\"
 414 >\e$B5~ETBg3X\e(B21\e$B@$5*\e(BCOE\e$B!VEl%"%8%"@$3&$N?MJ8>pJs3X8&5f650i5rE@!W\e(B</a>
 415 <li><a href=\"http://www.unicode.org/\"
 416 >Unicode</a>
 417 </ul>"
 418  'utf-8-jp-er))
 419
 420       ))
 421     (princ "<hr>")
 422     (princ "<p>
 423 Copyright (C) 2005, 2006, 2007, 2008, 2009, 2010, 2015 <a href=\"http://kanji.zinbun.kyoto-u.ac.jp/~tomo/\"
 424 >MORIOKA Tomohiko</a>")
 425     (princ
 426      (format
 427       "<p>Powered by <a
 428 href=\"http://www.chise.org/xemacs/\"
 429 >XEmacs CHISE</a> %s."
 430       (encode-coding-string xemacs-chise-version 'utf-8-jp-er)))
 431     (princ "
 432 </body>
 433 </html>
 434 ")))