Break URLs at separator character and remove the separator

I just added a lot of pattern trimming to both the 5.x-2.x and 6.x-1.x branches. Could please confirm this is still happening with the 5.x-2.x-dev version? (You may have to wait a bit though, until the "Last updated" field says "June 13" (or later).)

Steps taken:

1. Create a node with the title Often times, we are forced to have a very long title name. When that happens, PathAuto often trunkates the path as it should, bu
2. Resulting alias was often-times-we-are-forced-to-have-very-long-title-name-when-that-happens-pathauto-often-trunkates-th

That seems as expected to me.

So, I tried
1. Create a node with the title Often times, we are forced to have a very long title name. When that happens, PathAuto often trunkates the path as it (note all of the spaces after trunkates, I expected pathauto to cut the line there and replace the space with hyphen) but...
2. Resulting alias was often-times-we-are-forced-to-have-very-long-title-name-when-that-happens-pathauto-often-trunkates-th

So I tried
1. Create a node with the title Often times, we are forced to have a very long title name. When that happens, PathAuto often trunkates t he path as it (note the space in the middle of the word "the" in "trunkates t he")
2. Resulting alias was often-times-we-are-forced-to-have-very-long-title-name-when-that-happens-pathauto-often-trunkates-t-

So, I'd say that is the bug, right?

Hohum. I'm inclined to re-categorise into a feature request for what's described in #4. This also means that I'm inclined to not want to spend time fixing this for 5.x, but have it moved to 6.x-2.x instead.

@Freso - Agreed.

I've also received several requests that when urls are shortened for whatever reason they should be shortened to a good logical point (i.e. separators) instead of just the exact point where 100 characters lands. Since those seem similar in my mind, I'll just bundle that one in here.

Something like http://api.drupal.org/api/function/truncate_utf8 will sure help when we need to do this.

This should work, but I haven't tested it. So please do.

Just tested it, and it doesn't seem to work. Bugger.

How about this?

Yay! It works! :D

It had a line of pure whitespace though, so fixed that. I also added a line documenting what that block of code does. Should be good to go.

Awesome - thanks, Freso.

http://drupal.org/cvs?commit=131526

Per Freso, this needs a little more testing...

The problem was that if you created a node with a short title this still lopped off the last word.

1. Create a node with the title "this is a short title"
Expect results:
aliased as content/this-short-title

Actual results:
aliased as content/this-short

The attached patch fixes that problem, generalizes this code so it can be re-used elsewhere in pathauto, and then re-uses it where individual tokens are built and shortened to the appropriate length.

Attached...now...

+1 for separating the logic to its own function, but perhaps call it _pathauto_truncate_url() or truncate_chars instead? (FWIW, the 7.x version of truncate_utf8() will be drupal_truncate_chars().)

Also "A Pathauto friendly version of truncate_utf8" should be finished off with a ".". ;)

Apart from this, the patch looks good. Looking forward to test it in the morrow!

Good points.

Works now with both long and short aliases! Yay! And I see no more nits to pick at (well, perhaps truncate_utf8 could be referenced with a pair of parentheses... but I'm somewhat indifferent on that). :p

It breaks the tests though, but this is likely because the tests themselves need to be updated for this new logic.

Actually, without this follow-up patch, it'll break two tests, but with it, it'll only break one. So it's an improvement. The test still breaking is "[testPathAuto]: Node accessible through alias at [[...]/sites/all/modules/pathauto/tests/pathauto.test line 84]", which tries to fetch the node using the alias.

And even reverting the originally committed patch from this issue doesn't take the fail away.

And fixed - http://drupal.org/cvs?commit=131753

Thanks for the reviews/testing.

I know this is a year old, but did this ever make it into the official release?
I cant seem to get this functionality to work. my url titles are being cut off in mid word.
Is there an option to enable this that im missing?

thanks.

There is a similar issue at the beginning of an url - if you have a page who's title starts with a short word (ie. 'to') it will remove it, but leave the separator. Creating a node titled 'To Something' creates an url of '/-something', instead of just '/something'.

@msiama - is that problem still present in Pathauto or did this patch fix it (this patch is in the tip of 6.x-2.x dev and in 6.x-1.2

I hate to be a pain but I am experiencing this issue with the 6.x-1.4 release.

This: Broker Commissions Will Get Squeezed by Health Reform, but a New 'Navigator' Role Is Seen Growing

translates to this: ...enews/archive/2010/07/07/Broker-Commissions-Will-Get-Squeezed-Health-Reform-New-Navigator-Role-Seen-

I didn't think I should have to mess with the patch since it's from like 2 years ago. What am I doing wrong?

@interestingaftermath: Can you please file a new issue detailing exactly what's going wrong for you instead of opening tickets that have been closed for several months or years?

Dave, this is the second post you've made that comment on but I really don't understand your reasoning... The issue is DIRECTLY related to this one (as was the other issue). I will open a new issue as requested but it really seems pointless and unorganized. People can read timestamps, that's what they're for.