246505 - simple html displays the ; from entities

Status: RESOLVED FIXED

(SeaMonkey :: MailNews: Message Display, defect)

Description

[Michiel van Leeuwen (email: mvl+moz@)]

A mail message displayed as simple html will show the ; of entities. For example
  displays as  ;, and & displays as &;

Seeing this with linux gtk2 2004061218 (self made cvs build). Didn't see this
with my previous build, 2004060321.

steps to reproduce:
1. send yourself a html+plaintext mail, with at least an & in it.
2. view the message as simple html
3. notice the extra ;
4. check in plain text and originial html that the ; shouldn't be there

Comment 1

[Andreas Lange]

I looked through likely bugs, and found bug 243040 "HTML sanitizer misinterprets
attribute values". By backing out attachment 150102 [details] [diff] [review] this problem went away.

Comment 2

[Michiel van Leeuwen (email: mvl+moz@)]

cc-ing ben bucksch, who wrote that patch

Comment 3

[Ben Bucksch (:BenB)]

That's probably because I added a ";" after outputting entities. I thought that
was necessary (&entityname;)? Maybe Gecko already gives us the ";" in the entity
name?
Or that's because of the escaping, but I don't think so.
Will have to test.

Comment 4

[Ben Bucksch (:BenB)]

Attachment: Fix, v1

> Maybe Gecko already gives us the ";" in the entity name?

Indeed, it gives e.g. "nspr;" as entity name. Freaky.

Comment 5

[Ben Bucksch (:BenB)]

Comment on attachment 150923 [details] [diff] [review]
Fix, v1

trivial patch, akk and jst, please review

Comment 6

[Ben Bucksch (:BenB)]

trivial patch, akk and jst, please review

Comment 7

[Johnny Stenback (:jst)]

Comment on attachment 150923 [details] [diff] [review]
Fix, v1

Hmm, odd. One could maybe check if the terminating ';' is there and add it if
it isn't, but one could also hope that we're consistent (I'm assuming you
checked)...

sr=jst

Comment 8

[Akkana Peck]

Comment on attachment 150923 [details] [diff] [review]
Fix, v1

r=akkana -- obviously needed to fix the regression.

I wonder if it would be better in the long run to use nsIParserService's
HTMLConvertEntityToUnicode method instead, like the plaintext serializer does? 
Anyone know what the implications are of using one vs. the other?    

The html and xml serializers don't seem to do anything for this case -- in
fact, they no longer seem to support serializing from parser tags at all -- no
matches for eHTMLTag_ at all.  Weird.

Comment 9

[Mike Cowperthwaite]

The patch has reviews, can it get checked in or is a rewrite per comment 8 
necessary?

Comment 10

[Ben Bucksch (:BenB)]

yes, once I get to it. low pri.

Comment 11

[Ere Maijala (slow)]

Comment on attachment 150923 [details] [diff] [review]
Fix, v1

I'd like to get this in. Getting tired of reading html mail with umlaut
entities..

Comment 12

[Ere Maijala (slow)]

Comment on attachment 150923 [details] [diff] [review]
Fix, v1

I checked this patch in. Can this be marked fixed or will there be more
changes?

Comment 13

[Mike Cowperthwaite]

Marking Fixed.  Thanks, Ere.