src/parse/html-parser.lisp

   1 (in-package :closure-html)
   2
   3 ;;; FIXME: I liked the old SLURP-CATALOG code better than the LOOP below.
   4 ;;; (Except for the use of NETLIB and URI, which we don't have here.)
   5
   6 #||
   7
   8 (defun slurp-catalog (catalog-url)
   9   ;; Really dirty implementation
  10   (setf *simple-catalog* nil)
  11   (multiple-value-bind (io header) (netlib::open-document-2 catalog-url)
  12     (declare (ignore header))
  13     (unwind-protect
  14         (let ((str (glisp::gstream-as-string io)))
  15           (with-input-from-string (input str)
  16             (do ((x (read input nil nil) (read input nil nil)))
  17                 ((null x))
  18               (assert (equal (symbol-name x) "PUBLIC"))
  19               (let ((name (read input))
  20                     (file (read input)))
  21                 (assert (stringp name))
  22                 (assert (stringp file))
  23                 (push (cons name (url:merge-url (url:parse-url file) catalog-url))
  24                       *simple-catalog*)))))
  25       (g/close io))))
  26
  27 (format T "~&;; Parsing DTD~% ")
  28 (sgml:slurp-catalog (url:parse-url "file://closure/resources/dtd/catalog"))
  29 (setf cl-user::*html-dtd* (sgml:parse-dtd '(:public "-//W3C//DTD HTML 4.0 Frameset//EN")))
  30 (format T "~&;; done~%")
  31
  32 ||#
  33
  34 (defparameter sgml::*simple-catalog*
  35   (let ((base
  36          (merge-pathnames
  37           "resources/"
  38           (asdf:component-relative-pathname
  39            (asdf:find-system :closure-html)))))
  40     (loop
  41        :for (name . filename)
  42        :in '(("-//W3O//DTD W3 HTML 3.0//EN" . "dtd/HTML-3.0")
  43              ("NETSCAPE-Bookmark-file-1" . "dtd/NETSCAPE-Bookmark-file-1")
  44              ("-//W3C//ENTITIES Special//EN//HTML" . "dtd/Entities-Special")
  45              ("-//W3C//ENTITIES Symbols//EN//HTML" . "dtd/Entities-Symbols")
  46              ("-//W3C//ENTITIES Latin1//EN//HTML" . "dtd/Entities-Latin1")
  47              ("-//W3C//DTD HTML 4.0 Frameset//EN" . "dtd/DTD-HTML-4.0-Frameset")
  48              ("-//W3C//DTD HTML 4.0//EN" . "dtd/DTD-HTML-4.0")
  49              ("-//W3C//DTD HTML 4.0 Transitional//EN" . "dtd/DTD-HTML-4.0-Transitional"))
  50        :collect (cons name (merge-pathnames filename base)))))
  51
  52 (defparameter *html-dtd*
  53   (sgml:parse-dtd '(:public "-//W3C//DTD HTML 4.0 Frameset//EN")))
  54
  55 (defun parse-xstream (input handler)
  56   (setf (sgml::a-stream-scratch input)
  57         (make-array #.(* 2 4096) :element-type 'runes:rune))
  58   (sgml::setup-code-vector input :utf-8)
  59   (let* ((dtd *html-dtd*)
  60          (sgml::*unmungle-attribute-case* t)
  61          (r (sgml:sgml-parse dtd input))
  62          (pt (sgml::post-mortem-heuristic dtd r)))
  63     (if handler
  64         (serialize-pt pt handler)
  65         pt)))
  66
  67 (defun parse (input handler &key pathname)
  68   (declare (ignore pathname))
  69   (etypecase input
  70     (xstream
  71      (parse-xstream input handler))
  72     (rod
  73      (let ((xstream (make-rod-xstream (string-rod input))))
  74 ;;;        (setf (xstream-name xstream)
  75 ;;;          (make-stream-name
  76 ;;;           :entity-name "main document"
  77 ;;;           :entity-kind :main
  78 ;;;           :uri nil))
  79        (parse-xstream xstream handler)))
  80     (array
  81      (parse (make-octet-input-stream input) handler))
  82     (pathname
  83      (with-open-file (s input :element-type '(unsigned-byte 8))
  84        (parse s handler :pathname input)))
  85     (stream
  86      (let ((xstream (make-xstream input :speed 8192)))
  87 ;;;        (setf (xstream-name xstream)
  88 ;;;          (make-stream-name
  89 ;;;           :entity-name "main document"
  90 ;;;           :entity-kind :main
  91 ;;;           :uri (pathname-to-uri
  92 ;;;                 (merge-pathnames (or pathname (pathname input))))))
  93        (parse-xstream xstream handler)))))
  94
  95 (defun serialize-pt-attributes (plist)
  96   (loop
  97      for (name value) on plist by #'cddr
  98      collect
  99      (let ((n (coerce (symbol-name name) 'rod))
 100            (v (etypecase value
 101                 (symbol (coerce (string-downcase (symbol-name value)) 'rod))
 102                 (rod value)
 103                 (string (coerce value 'rod)))))
 104        (hax:make-attribute n v t))))
 105
 106 (defun serialize-pt (document handler &key (name "HTML") public-id system-id)
 107   (hax:start-document handler name public-id system-id)
 108   (labels ((recurse (pt)
 109              (cond
 110                ((eq (gi pt) :pcdata)
 111                 (hax:characters handler (pt-attrs pt)))
 112                (t
 113                 (let ((name (coerce (symbol-name (pt-name pt)) 'rod))
 114                       (attrs (serialize-pt-attributes (pt-attrs pt))))
 115                   (hax:start-element handler name attrs)
 116                   (mapc #'recurse (pt-children pt))
 117                   (hax:end-element handler name))))))
 118     (recurse document))
 119   (hax:end-document handler))
 120
 121 (defclass pt-builder (hax:abstract-handler)
 122   ((current :initform nil :accessor current)
 123    (root :initform nil :accessor root)))
 124
 125 (defun make-pt-builder ()
 126   (make-instance 'pt-builder))
 127
 128 (defmethod hax:start-document ((handler pt-builder) name pubid sysid)
 129   (declare (ignore name pubid sysid))
 130   nil)
 131
 132 (defun unserialize-pt-attributes (attrs)
 133   (loop
 134      for a in attrs
 135      collect (intern (string-upcase (hax:attribute-name a)) :keyword)
 136      collect (hax:attribute-value a)))
 137
 138 (defmethod hax:start-element ((handler pt-builder) name attrs)
 139   (let* ((parent (current handler))
 140          (this (sgml::make-pt/low
 141                 :name (intern (string-upcase name) :keyword)
 142                 :attrs (unserialize-pt-attributes attrs)
 143                 :parent parent)))
 144     (setf (current handler) this)
 145     (if parent
 146         (push this (pt-children parent))
 147         (setf (root handler) this))))
 148
 149 (defmethod hax:characters ((handler pt-builder) data)
 150   (push data (pt-children (current handler))))
 151
 152 (defmethod hax:comment ((handler pt-builder) data)
 153   ;; zzz haven't found out what the representation of comments is...
 154   data)
 155
 156 (defmethod hax:end-element ((handler pt-builder) name)
 157   (let ((current (current handler)))
 158     (setf (pt-children current) (nreverse (pt-children current)))
 159     (setf (current handler) (pt-parent current))))
 160
 161 (defmethod hax:end-document ((handler pt-builder))
 162   (root handler))