295887 - [mozTXTToHTMLConv] word with a colon converted to link - recognition of URL should be more strict

Status: RESOLVED WORKSFORME

(MailNews Core :: Backend, defect)

Description

[ratatosk.se]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.0; sv-SE; rv:1.7.8) Gecko/20050511 Firefox/1.0.4
Build Identifier: Thunderbird version 1.0.2 (20050317) (Windows, lang=en-US)

On saving or sending a mail, a word with a single colon and letters, and only
letters, on either side of it is converted to a link.

Example: "dvd:n" or "dvd:ar" (= Swedish for "the dvd" and "dvd's" respectively).

Reproducible: Always

Steps to Reproduce:
1. Compose an e-mail containing a word with a colon between letters (e.g. "dvd:n")
2. Save or send.
Actual Results:  
Thunderbird converted the word with colon to a link.

Expected Results:  
It should have left the word unchanged. A word with a single colon (and
otherwise only letters) can never be an actual link, as far as I know.

Comment 1

[wintogreen]

Just a little clarification:
This applies especially to html-formatted messages, where the erroneous "link"
gets hard-coded into the outgoing message. E.g., "dvd:ar" gets converted into
"<a class="moz-txt-link-freetext" href="DVD:ar">DVD:ar</a>".

In plain-text messages, it doesn't get converted but it still displays as a link.

Comment 2

[Mike Cowperthwaite]

The behavior is specific to the "dvd" string (and a few other words, such as 
"file") and only when the colon is immediately followed by letters.

xref bug 241573, bug 136782.

English has no instances of a colon as part of a single word.  Is that what's 
actually going on in the Swedish, or is (for instance) ":n" an ASCII 
representation of something like n-with-umlaut?

Comment 3

[ratatosk.se]

The colon in the Swedish is an actual colon, not an ASCII representation of
anything else.

In Swedish, whenever you want to add an ending (e.g. plural, definite article or
a verb ending) to a "word" that is a collection of initials - such as "TV",
"PC", "CD", "DVD", "SMS", "PS", "WC", "EEG", "CRC" etc. - the ending must be
separated from the initials with a colon. Exception for initials pronounceable
as a word rather than a collection letters (e.g. "laser").

This is similar to the use of apostrophe in English ("many DVD's"), except that
the English apostrophe is optional (if am not wrong) whereas the Swedish colon
is mandatory. Plus, of course, the word "the" in Swedish is also an ending, so
this too requires the colon when applied to initials ("DVD:n" or "dvd:n" for
"the DVD"; lowercase is actually preferable in Swedish for very commonly used
abbreviations).

I think this is much more of a problem in Swedish than in English (though still
wouldn't have been a big deal if there were a way to turn off automatic
link-creation entirely - see bug 295899).

Comment 4

[Mike Cowperthwaite]

(In reply to comment #3)
> In Swedish, whenever you want to add an ending (e.g. plural, definite article
> or a verb ending) to a "word" that is a collection of initials - such as "TV",
> "PC", "CD", "DVD", "SMS", "PS", "WC", "EEG", "CRC" etc. - the ending must be
> separated from the initials with a colon. Exception for initials pronounceable
> as a word rather than a collection letters (e.g. "laser").

Hmm.  So if someone were using "ftp" as a verb, it would appear as something
like   ftp:ing   (to use an English declension as an example).  

That would never be a legal URL, of course; URLs for ftp, http, nttp and nfs all 
require a slash following the colon.  The URLs I know of that specifically do 
not require the slash are mailto and news, neither of which are acronyms and so 
not subject to this Swedish peculiarity.

Comment 5

[ratatosk.se]

Yep, "ftp:a" (infinitive) would be a proper Swedish verb.
Thanks for this example! :-)

I now tried googling the Web for "ftp:a", in Swedish only, and found several
hundreds of instances.

See Google: http://www.google.com/search?q=ftp:a&lr=lang_sv

(There are a few instances of "ftp'a" and "ftp-a" too to be seen, but to the
extent that endings are at all used here the preferred separator is colon -
according to regulatory bodies for the Swedish language, such as The Swedish
Language Council (Svenska språknämnden).)

And you are of course right that "mailto" and "news" wouldn't be subject to the
Swedish colon rule. (I think those particular examples would preferably not at
all be used untranslated in Swedish, though "telnetta" is a not entirely
uncommon Swedish verb.)

Comment 6

[ratatosk.se]

Hmm. I must apologize for not researching this bug as well as I should have,
before reporting it. It strikes specifically on "dvd" (as well as on "ftp")
followed by colon and any ending (any letters), but beyond that it doesn't seem
to be at all as general as I had assumed.

When I now tried several other Swedish words with colon in them (e.g. "cd:n",
"cd:ar"), Thunderbird didn't mess with them (no links created). Sorry if I have
mislead anyone with my first report.

Comment 7

[Magnus Melin [:mkmelin]]

->WFM, this is fixed since a long time ago