Closed Bug 33492 Opened 26 years ago Closed 25 years ago

html-to-plaintext message conversion double-spaces lines

Categories

(MailNews Core :: Composition, defect, P3)

x86
Linux
defect

Tracking

(Not tracked)

VERIFIED WORKSFORME

People

(Reporter: akkzilla, Assigned: rhp)

Details

(Whiteboard: [nsbeta2-])

Send yourself a plaintext message containing a paragraph of wrapped text. Reply to this message in html. A few lines into the paragraph, set the caret near the end of a line and hit return to split the quotation, and type a line of text. Send the message as plaintext The quoted text before the inserted line look fine; but all the quoted lines after the first insertion are double-spaced, i.e. they look like > blah blah [ ... ] blah > > blah blah > OutputText from the compose window does not do this (though it does add some empty quoted lines immediately after the new line), so I think this is happening in mail code after the output system is finished.
Status: NEW → ASSIGNED
Target Milestone: --- → M18
Putting on radar for beta2 consideration, since quoting replies properly is part of mail dogfood. I can probably help with this, at least with tracking it down, if Rich is overloaded.
Keywords: beta2
Akkana, If you want to help with this...feel free, but I will not get to it before Beta 2 because I am focused on FEATURE development at this point. - rhp
akk, 1. I don't think, this is related to my converter, as it shouldn't touch newlines. 2. I *do* see the problem in the editor output. I have a slightly modified viersion (my fix for bug #31906), but I see something like <body>bugzilla-daemon@mozilla.org wrote:<br> <blockquote type="cite" cite="200003272225.OAA05229@lounge.mozilla.org"> <div class="text-plain"><pre> (Tired of this old junk? There is a new experimental email scheme in Bugzilla. If you're feeling brave, you can try it by turnin</pre></div> </blockquote> <br> sorzu epotriz+ep ru zpeuz &nbsp; &nbsp; twwwwwwwwwwwwwwww+wtttttttt zw&uuml;otze+ z&uuml;oezt0wzzzzz&uuml;wwwwww<br> <blockquote type="cite" cite="200003272225.OAA05229@lounge.mozilla.org"> <div class="text-plain"> <pre> g on the<br> new pref at http://bugzilla.mozilla.org/userprefs.cgi?bank=diffs)<br> <br> http://bugzilla.mozilla.org/show_bug.cgi?id=33492<br> <br> *** shadow/33492 Mon Mar 27 14:10:37 2000<br> --- shadow/33492.tmp.5220 Mon Mar 27 14:25:14 2000<br> ***************<br> *** 40,42 ****<br> --- 40,49 ----<br> [...] + 2 because I am focused on FEATURE development at this point.<br> + </pre> </div> </blockquote> et&uuml;iutrzpprutz <blockquote type="cite" cite="200003272225.OAA05229@lounge.mozilla.org"> <div class="text-plain"> <pre> <br> Note, that there are <br>s in the second blockquote, but not in the first - most likely the reason for the bug. There should be no <br>s, I think.
(Ignore the extra newlines in the last comment, this is a bugzilla-with-mozilla-bug.)
Keywords: nsbeta2
Short steps to reproduce this problem: 1. Send yourself a plaintext message containing the following text: --- one two three --- 2. Read it in mozilla, and reply to it in html compose. While the compose window is up, do Debug->OutputHTML and Debug->OutputText to see what the normal output converters do. You get: OutputHTML: <html><head></head><body>akkana@netscape.com (Akkana Peck) wrote:<br> <blockquote type="cite" cite="38FF6417.A6A66529@netscape.com"> <pre wrap> one<br> two<br> three<br> </pre> </blockquote> <br> <br> </body> </html> OutputText: akkana@netscape.com (Akkana Peck) wrote: > > one > two > three 3. Send this message in plaintext. When it arrives, it looks like this: akkana@netscape.com (Akkana Peck) wrote: > > one > > two > > three > Note that the plaintext output is exactly what we want, whereas the html output has both br tags and newlines inside a pre, which is probably incorrect. In other words, plaintext mail didn't go through the normal plaintext output, which would have gotten it right; instead, mail got the html output and then went through some other means to convert it to plaintext. I need to fix the html problem anyway (bug 36188 is filed on ducarroz, but it's probably my bug and I'm going to look at it now) but I'm curious why we're not using the html->text conversion code in nsHTMLToTXTSinkStream that we spent so long trying to get right? Do we have similar code in several different places now that's getting out of synch?
Keywords: beta2
akk, 1. try the following: 1. In Mozilla's HTML composer, mark the word "two" bold. 2. Debug->Output Text shows: ..."*two*"... 3. Now send as plaintext and read it. You'll also see ..."*two*"... 2. nsHTMLToTXTConv is the only HTML->TXT converter in Mozilla, which handles bold this way. 3. 1. && 2. => Mailnews does use nsHTMLToTXTConv. So, should me mark this a dup?
I still don't understand why we're seeing double-spaced lines in plaintext mail, then, when OutputText doesn't show double spaced lines. Obviously something is different ... It's about to become academic, though, because I have a fix for the html problem (bug 36188), so the html output will no longer have <br>NS_LINEBREAK.
Putting on [nsbeta2-] radar due to workaround that akkana will do.
Whiteboard: [nsbeta2-]
Status: ASSIGNED → RESOLVED
Closed: 25 years ago
Resolution: --- → WORKSFORME
This seems to work ok for me now...am I wrong? - rhp
rhp: If you read upward, I checked in a workaround that masked what we thought the real problem was, and this bug was staying open for someone to figure out why we were saving as html in the first place rather than just saving as plaintext since that was the format we knew we were going to.
Linux (2000-07-28-04 M17) Win32 (2000-07-28-12 M17) Mac (2000-07-26-08 M17) I cannot re-produce this bug using the scenario described by the reporter. Akkana, please advise what more needs to be done. thanks
Sigh ... never mind, I guess we'll have to file a separate bug on the performance problems caused by converting to html first and then to plaintext, rather than going straight to plaintext, if it shows up on performance analysis.
Status: RESOLVED → VERIFIED
Product: MailNews → Core
Product: Core → MailNews Core
You need to log in before you can comment on or make changes to this bug.