sgml-mode skip tag forward/back bug fix -- followup

sgml-mode skip tag forward/back bug fix -- followup

[Mallory, Rich]

This is a followup to the email to you included at the end of this 
email, on the above subject.  

Further testing revealed that the bug fix proposed in that email 
was incomplete.  The complete set of diffs follows, including the 
diffs from the original email.  The proposed fix works without the 
new additions.   However, the change to line 687 is needed to 
keep up-list from beeping each time function sgml-skip-tag-forward 
is executed.  (My machine at work has no speaker, so I didn't hear
a beep and ignored the message until I got home.)  The original 
code for line 683 produces no beep, but the change to line 683 
seems advisable.

Diffs between old sgml-mode.el (<) and new (>)
In sgml-skip-tag-backward --
656c656
< 	      (re (concat "</?" (regexp-quote (match-string 1)))))
---
> 	      (re (concat "</?" (regexp-quote (match-string 1)) "[/ \n\t>]")))

In sgml-skip-tag-forward --
676c676
< 	      (re (concat "</?" (regexp-quote (match-string 1))))
---
> 	      (re (concat "</?" (regexp-quote (match-string 1)) "[/ \n\t>]"))
683c683
< 		      (not (up-list -1))
---
> 		      (not (progn (backward-char 1) (up-list -1)))
687c687
> 	      (up-list 1)
---
< 	      (progn (backward-char 1) (up-list 1))

The changes to lines 683 and 687 are required by the change to 
line 676.  The latter causes the regexp 're' to match one 
additional character, leaving the point after that character.  If this 
character is '>', the point is outside the tag matched by 're'.  In 
the original code, the point was inside the tag, and up-list moved 
the point to the beginning of the tag.  With the point after the 
closing '>', up-list can't do that.

-- Rich Mallory



-----Original Message-----
From: Mallory, Rich 
Sent: Friday, April 26, 2002 12:43 PM
To: 'bug-gnu-emacs@gnu.org'
Subject: sgml-mode skip tag forward/back bug fix

Your bug report will be posted to the bug-gnu-emacs@gnu.org mailing list,
and to the gnu.emacs.bug news group.

In GNU Emacs 21.2.1 (i386-msvc-nt5.0.2195) of 2002-03-19 on buffy
configured using `configure --with-msvc (12.00)'
Important settings:
  value of $LC_ALL: nil
  value of $LC_COLLATE: nil
  value of $LC_CTYPE: nil
  value of $LC_MESSAGES: nil
  value of $LC_MONETARY: nil
  value of $LC_NUMERIC: nil
  value of $LC_TIME: nil
  value of $LANG: ENU
  locale-coding-system: iso-latin-1
  default-enable-multibyte-characters: t

Please describe exactly what actions triggered the bug
and the precise symptoms of the bug:

In editing an XML file in sgml-mode, the commands sgml-skip-tag-forward and
sgml-skip-tag-backward failed as follows.  When an outer element contained an
inner element and the inner element's tag name began with the outer element's
tag name but was longer, these commands did not skip the whole outer element,
from start tag to end tag (or back), but only part of it.

For example, in the XML --
  <foo>
    <whatever .../>
    <fooBar note="Tag name begins with 'foo', the outer tag name." />
    <something .../>
  </foo>
 -- with the cursor before <foo>, sgml-skip-tag-forward would only skip to the 
end of tag <foo>, not to the end of tag </foo>, and 
 -- with the the cursor after </foo>, sgml-skip-tag-backward would only skip to 
the beginning of <fooBar .../>, not to the beginning of <foo>.

Proposed fix: The changelog and proposed changes to the code are shown below.
The added elements of the regexps cause the search code to look for the end of
the tag name, not just its beginning.  The new code seems to work correctly.
However, I am not sure that the added regexp elements are are sufficiently
general to cover all cases.

Change log
2002-04-26  Rich Mallory  <mallory@ausinfo.com>
	* lisp/textmodes/sgml-mode.el (sgml-skip-tag-backward, sgml-skip-tag-forward):
	Fix bug where an outer element containing an inner element whose name
	began with the outer name (e.g., <foo> ... <fooBar ... /> ... </foo>)
	was not properly skipped.

Diffs between old sgml-mode.el (<) and new (>)
In sgml-skip-tag-backward --
656c656
< 	      (re (concat "</?" (regexp-quote (match-string 1)))))
---
> 	      (re (concat "</?" (regexp-quote (match-string 1)) "[/ \n\t>]")))

In sgml-skip-tag-forward --
676c676
< 	      (re (concat "</?" (regexp-quote (match-string 1))))
---
> 	      (re (concat "</?" (regexp-quote (match-string 1)) "[/ \n\t>]"))

(I haven't actually modified sgml-mode.el but put the changes in another file.
The diffs above show what has been changed.)

-- Rich Mallory

Re: sgml-mode skip tag forward/back bug fix -- followup

[Kevin Rodgers]

"Mallory, Rich" <Rich.Mallory@ausinfo.com> writes:

> Further testing revealed that the bug fix proposed in that email
> was incomplete.  The complete set of diffs follows, including the
> diffs from the original email.  The proposed fix works without the
> new additions.   However, the change to line 687 is needed to
> keep up-list from beeping each time function sgml-skip-tag-forward
> is executed.  (My machine at work has no speaker, so I didn't hear
> a beep and ignored the message until I got home.)  The original
> code for line 683 produces no beep, but the change to line 683
> seems advisable.
...
> The changes to lines 683 and 687 are required by the change to
> line 676.  The latter causes the regexp 're' to match one
> additional character, leaving the point after that character.  If this
> character is '>', the point is outside the tag matched by 're'.  In
> the original code, the point was inside the tag, and up-list moved
> the point to the beginning of the tag.  With the point after the
> closing '>', up-list can't do that.

Why don't you make use of the syntax table, and append "\\>" instead
to the tag regexp?  That would prevent matching element names that are
a prefix of the one to be skipped as desired, and wouldn't match the
extra character that necessitates the call to backward-char before
up-list.

-- 
Kevin Rodgers <kevinr@ihs.com>