src/parse/html-parser.lisp

   1 (in-package :closure-html)
   2
   3 (defparameter *this-file*
   4   (load-time-value
   5    (or #.*compile-file-pathname* *load-pathname*)))
   6
   7 (defparameter *this-directory*
   8   (make-pathname :directory (pathname-directory *this-file*)))
   9
  10 (defparameter sgml::*simple-catalog*
  11   (loop :for (name . filename)
  12      :in '(("-//W3O//DTD W3 HTML 3.0//EN" . "dtd/HTML-3.0")
  13            ("NETSCAPE-Bookmark-file-1" . "dtd/NETSCAPE-Bookmark-file-1")
  14            ("-//W3C//ENTITIES Special//EN//HTML" . "dtd/Entities-Special")
  15            ("-//W3C//ENTITIES Symbols//EN//HTML" . "dtd/Entities-Symbols")
  16            ("-//W3C//ENTITIES Latin1//EN//HTML" . "dtd/Entities-Latin1")
  17            ("-//W3C//DTD HTML 4.0 Frameset//EN" . "dtd/DTD-HTML-4.0-Frameset")
  18            ("-//W3C//DTD HTML 4.0//EN" . "dtd/DTD-HTML-4.0")
  19            ("-//W3C//DTD HTML 4.0 Transitional//EN" . "dtd/DTD-HTML-4.0-Transitional"))
  20      :collect (cons name (merge-pathnames filename *this-directory*))))
  21
  22 (defparameter *html-dtd* (sgml:parse-dtd '(:public "-//W3C//DTD HTML 4.0 Frameset//EN")))
  23
  24 (defun parse (inputstr)
  25   "given a string, produce a sgml:pt, which would be your toplevel parse tree node"
  26   (let ((dtd *html-dtd*))
  27     (let ((input (runes:make-xstream
  28                   (flexi-streams:make-in-memory-input-stream
  29                    (flexi-streams:string-to-octets
  30                     inputstr :external-format (flexi-streams:make-external-format :utf-8))))))
  31       (setf (sgml::a-stream-scratch input)
  32             (make-array #.(* 2 4096) :element-type 'runes:rune))
  33       (sgml::setup-code-vector input :utf-8)
  34       (let ((r (sgml:sgml-parse dtd input)))
  35         (sgml::post-mortem-heuristic dtd r)))))