Adium

Opened 13 years ago

Closed 12 years ago

Last modified 12 years ago

#7910 closed defect (fixed)

URL parser grabs B.Sc.

Reported by: rgovostes Owned by: earthmkii
Milestone: Adium 1.3 Component: AutoHyperlinks
Version: Severity: minor
Keywords: Cc:
Patch Status:

Description

B.Sc. is an abbreviation for Bachelor of Science, but the URL parser thinks it's a site in Seychelles. (No offense to any Seychellois, but I've never hard of it.)

Change History (13)

comment:1 Changed 13 years ago by Ryan Govostes

Nor have I heard of it.

comment:2 Changed 13 years ago by Colin Barrett

Owner: changed from nobody to Stephen Holt

comment:3 Changed 13 years ago by Eric Richie

Milestone: Good idea for "later"

Out of curiosity, how are we supposed to differentiate between an abbreviation like that and a URL? I suppose we could evaluate based on context but that seems like overkill.

comment:4 Changed 13 years ago by Colin Barrett

Typically, nameservers don't give out single letter domains like that (a notable exception being x.org).

I'd say a one letter tld in odd caps shouldn't match (could hardcode in X.org, I guess). I have a number of ideas for ways to improve the url parser but they belong in a separate ticket.

comment:5 Changed 13 years ago by Stephen Holt

Status: newassigned

comment:6 Changed 12 years ago by Carlos Morales

comment:7 Changed 12 years ago by Stephen Holt

Component: Adium CoreAutoHyperlinks
Milestone: Good idea for "later"Needs feedback from users

One solution to fixing this would be to ignore ccTLDs unless flex is in its "cannonical" start state (link has a scheme specifier). This will break autolinking things like del.icio.us, and blo.gs.

The other is to try and implement Mozilla's TLD List <http://wiki.mozilla.org/TLD_List> as closely as we can.

I'm inclined to the second option, though it's much harder.

comment:8 Changed 12 years ago by Robert

Stephen, what kind of feedback are you expecting? Do you want users to give opinions on which option to go for?
Isn't this perhaps something for the other developers to decide?

comment:9 in reply to:  8 Changed 12 years ago by Stephen Holt

Milestone: Needs feedback from usersGood idea for "later"

Replying to Robby:

Stephen, what kind of feedback are you expecting? Do you want users to give opinions on which option to go for?
Isn't this perhaps something for the other developers to decide?

Sorry... selected that by accident.

comment:10 Changed 12 years ago by Andreas Monitzer

Maybe a blacklist would do it for cases like this?

comment:11 Changed 12 years ago by sholt

(In [23567]) http://wiki.mozilla.org/TLD_List defines a number of ccTLDs that will only register third level domains. We can declare these first and second level domains explicitly, and prevent some incorrect linkification in non-strict mode. For example, python filesnames (.py) will no longer be linked. Refs #7910, but not a complete solution yet.

comment:12 Changed 12 years ago by sholt

Resolution: fixed
Status: assignedclosed

(In [23797]) Removed some essentially redundant rules (distinctions without a difference), and added a rule to ignore specific trouble patterns. I don't feel like this is the "right" way to fix these, but it's the best compromise for now.

comment:13 Changed 12 years ago by Stephen Holt

Milestone: Good idea for "later"Adium X 1.3
Note: See TracTickets for help on using tickets.