xemacs-beta: lisp/w3/w3-parse.el comparison

comparison lisp/w3/w3-parse.el @ 2:ac2d302a0011 r19-15b2

Import from CVS: tag r19-15b2

author	cvs
date	Mon, 13 Aug 2007 08:46:35 +0200
parents	376386a54a3c
children	9ee227acff29

comparison

equal deleted inserted replaced

-:c0c6a60d29db
+:ac2d302a0011
 ;; Created by: Joe Wells, jbw@csb.bu.edu
 ;; Created on: Sat Sep 30 17:25:40 1995
 ;; Filename: w3-parse.el
 ;; Purpose: Parse HTML and/or SGML for Emacs W3 browser.
-;; Copyright © 1995  Joseph Brian Wells
+;; Copyright © 1995, 1996  Joseph Brian Wells
-;; Copyright © 1993, 1994, 1995 by William M. Perry (wmperry@spry.com)
+;; Copyright © 1993, 1994, 1995 by William M. Perry (wmperry@cs.indiana.edu)
 ;;
 ;; This program is free software; you can redistribute it and/or modify
 ;; it under the terms of the GNU General Public License as published by
 ;; the Free Software Foundation; either version 2 of the License, or
 ;; (at your option) any later version.
 (let ((html-entities w3-html-entities))
 (while html-entities
 (put (car (car html-entities)) 'html-entity-expansion
 	 (cons 'CDATA (if (integerp (cdr (car html-entities)))
-			  (char-to-string (cdr (car html-entities)))
+(char-to-string
+(let ((c (cdr (car html-entities))))
+(cond
+((and (> c 127) (boundp 'MULE))
+(make-character lc-ltn1 c))
+;;((and (> c 127) (featurep 'mule))
+;; What???
+;;)
+(t
+c))))
 			(cdr (car html-entities)))))
 (setq html-entities (cdr html-entities))))
 ;; These are handled differently than the normal HTML entities because
 ;; we need to define the entities with 'nil instead of 'CDATA so
 (cur nil))
 (while html-entities
 (setq cur (car html-entities)
 html-entities (cdr html-entities))
 (put (nth 0 cur) 'html-entity-expansion
-	 (cons 'nil (format "img src=\"%s/%s%s\" alt=\"%s\""
+	 (cons 'nil (format "<img src=\"%s/%s%s\" alt=\"%s\">"
 w3-icon-directory
 (nth 1 cur)
 (if w3-icon-format
 (concat "." (symbol-name w3-icon-format))
 "")
 ;; larger than 255.  I think in MULE it does.  Is this true?
 ;; Bill wants to call w3-resolve-numeric-entity here, but I think
 ;; that functionality belongs in char-to-string.
 ;; The largest valid character in the I18N version of HTML is 65533.
 ;; <URL:ftp://ds.internic.net/internet-drafts/draft-ietf-html-i18n-01.txt>
-(insert (char-to-string w3-p-s-num)))
+;; wrongo!  Apparently, mule doesn't do sane things with char-to-string
+;; -wmp 7/9/96
+(insert (char-to-string
+(cond
+((and (boundp 'MULE) (> w3-p-s-num 127))
+(make-character lc-ltn1 w3-p-s-num))
+;;((and (featurep 'mule) (> w3-p-s-num 127))
+;;what??
+;;)
+(t
+w3-p-s-num)))))
 ((looking-at "&#\\(re\\|rs\\|space\\|tab\\)[\ ;\n]?") ; \n should be \r
 (replace-match (assq (upcase (char-after (+ 3 (point))))
 '(;; *** Strictly speaking, record end should be
 ;; carriage return.
 (?E . "\n") ; RE
 ;;           using its content model.
 ;;
 ;;     NEW-STATE (optional, default *same) is the index of the state to
 ;;     move to after processing the element or one of these:
 ;;       *same: no state change occurs.
-;;       *next: change the the current state + 1.
+;;       *next: change the current state + 1.
 ;;     The initial state is 0.  NEW-STATE does not matter if ACTION is
 ;;     *close.
 ;;
 ;;     ERRORP (optional, default nil) if non-nil indicates this transition
 ;;     represents an error.  The error message includes this value if it
 (eval-when-compile
 (w3-unfold-dtd
 (w3-expand-parameters
 '(
 (%headempty . (link base meta range))
+(%headmisc . (script))
 (%head-deprecated . (nextid))
 ;; client-side imagemaps
 (%imagemaps . (area map))
 ;; special action is taken for %text inside %body.content in the
 ;; content model of each element.
 (%body.content . (%heading %block hr div address %imagemaps))
 (%heading . (h1 h2 h3 h4 h5 h6))
-;; Netscape's CENTER, FONT, and BASEFONT are handled
-;; non-standardly.  In actual psuedo-HTML on the net, these are
-;; used as both text-level constructs and block-level constructs.
-;; They are the only items in both %block and %text in this
-;; definition here.
-;; *** Perhaps add BLINK here too?  But no one uses that as a block
-;; construct.  What about NOBR?
-(%netscape-crud . (center font basefont))
 ;; Emacs-w3 extensions
-(%emacsw3-crud-nonempty . (roach secret wired))
 (%emacsw3-crud  . (pinhead flame cookie yogsothoth hype peek))
-(%block . (p %list dl form %preformatted %netscape-crud font
+(%block . (p %list dl form %preformatted font
 %blockquote isindex fn table fig note
-%block-deprecated %block-obsoleted))
+center %block-deprecated %block-obsoleted))
 (%list . (ul ol))
 (%preformatted . (pre))
 (%blockquote . (bq))
 (%block-deprecated . (dir menu blockquote))
 (%block-obsoleted . (xmp listing))
 ;; Why is IMG in this list?
 (%pre.exclusion . (*include img *discard tab math big small sub sup))
-(%text . (*data b %notmath %netscape-crud sub sup
+(%text . (*data b %notmath sub sup %emacsw3-crud))
-%emacsw3-crud %emacsw3-crud-nonempty))
 (%notmath . (%special %font %phrase %misc))
-(%font . (i u s strike tt big small))   ; B left out for MATH handling
+(%font . (i u s strike tt big small sub sup
-(%phrase . (em strong code samp kbd var cite blink))
+roach secret wired)) ;; B left out for MATH
-;; Don't know if this is right place for EMBED.
+(%phrase . (em strong dfn code samp kbd var cite blink))
-(%special . (a img br wbr nobr tab math embed))
+(%special . (a img applet font br script map math tab))
-(%misc . (q lang au dfn person acronym abbrev ins del))
+(%misc . (q lang au person acronym abbrev ins del))
 (%formula . (*data %math))
 (%math . (box above below %mathvec root sqrt array sub sup
 %mathface))
 (%mathvec . (vec bar dot ddot hat tilde))
 nil
 (*include *same "after FRAMESET"))
 ])
 (end-tag-omissible . t))
 ((head)
-(content-model . [((title isindex %headempty style %head-deprecated)
+(content-model . [((title isindex %headempty %headmisc
+style %head-deprecated)
 nil
 nil
 ;; *** Should only close if tag can
 ;; legitimately follow head.  So many can that
 ;; I haven't bothered to enumerate them.
 (*close))])
 (end-tag-omissible . t))
+;; SCRIPT - - (#PCDATA)
+((script)
+(content-model . CDATA          ; not official, but allows
+; comment hiding of script
+))
 ;; TITLE - - (#PCDATA)
 ((title)
 (content-model . RCDATA         ; not official
 ;; [((*data) include-space nil nil)]
 ))
 ;; STYLE - O (#PCDATA)
 ;; STYLE needs to be #PCDATA to allow omitted end tag.  Bleagh.
 ((style)
-(content-model . [((*data)
+(content-model . CDATA)
-include-space
-nil
-;; *** Should only close if tag can
-;; legitimately follow style.  So many can that
-;; I haven't bothered to enumerate them.
-(*close))])
 (end-tag-omissible . t))
 ((body)
 (content-model . [((banner) nil nil (*retry *next))
 ((bodytext) nil nil (bodytext *next))
 (nil nil (((plaintext) *close)) nil)])
 ;; caused by BQ's sharing of BODYTEXT.  BQ
 ;; should have its own BQTEXT.
 ((credit plaintext) *close))
 nil)])
 (end-tag-omissible . t))
-((div banner)
+((div banner center)
 (content-model . [((%body.content)
 nil
 ;; Push <P> before data characters.  Non-SGML.
 (((%text) p))
 nil)]))
 ((credit) *close))
 nil)])
 (end-tag-omissible . t))
 ((%emacsw3-crud)
 (content-model . EMPTY))
-((%netscape-crud)
-;; Special non-SGML treatment of Netscape's shit.
-(content-model . XINHERIT))
 ;; FORM - - %body.content -(FORM) +(INPUT|KEYGEN|SELECT|TEXTAREA)
 ((form)
 ;; Same as BODY.  Ugh!
 (content-model . [((%body.content)
 nil
 ;; bad formatting if the A is *never* closed.
 ;;((p) *discard *same error)
 )
 nil)])
 (exclusions . (a)))
-((b %font %phrase %misc nobr %emacsw3-crud-nonempty)
+((b font %font %phrase %misc nobr)
 (content-model . [((%text)
 include-space
 ((%in-text-ignore))
 nil)]))
 ((plaintext)

Mercurial > hg > xemacs-beta

comparison lisp/w3/w3-parse.el @ 2:ac2d302a0011 r19-15b2