[bookpath], [menupath], [*path] tokens not cleaned: aliases without punctuation removed, lower casing, etc.

Tokens that end in 'url', 'path', 'alias' or those words plus the -raw suffix will not be modified by pathauto since they are assumed to be URL aliases already.

Ok we're going to have to specialcase some token names that end with path-raw as the violate the naming convention enforced by pathauto.

I've committed a fix to make sure pathauto_cleanstring() is applied to the [bookpath], [bookpath-raw], [menupath], and [menupath-raw] tokens. Those are the only ones I could find that would be affected. [termpath] and [catpath] are not affected since they already apply pathauto_cleanstring() internally.

http://drupal.org/cvs?commit=406132
http://drupal.org/cvs?commit=406134

The fix will be available whenever the -dev releases regenerate, which is always within 12 hours.

I think this is a major bug and should be addressed asap as it breaks sites working with Pathauto 1.3.

The attached patch implements Dave's suggestion from #3.

Marked #880744: latest commits makes [bookpath-raw] not get lowercase or separator as a duplicate of this issue.

Note that if you are using [bookpath-raw] I would highly recommend changing it to [bookpathalias-raw].

@platch: That's is the approach that was committed to CVS. You can check out the actual code I committed with the links I put up in #5.

Changing title to help others find this issue

Not quite to sure what's going on... comments are saying the problem is fixed, but also saying it'll take twelve hours?

Is the patch (pathauto-881270-14.patch) the fix, or is it that supposed to be temporary?

Not new to Drupal, but definitely new to using patches in .inc files... not sure what to do with the patch, is there going to be a new release instead?

@steven_kropp

As I understand it you can use the patch now, or wait for the next dev-release which will be released within 12 hours (which apparently is an automated process). This dev-release will include the patch.

For applying patches yourself (in case you can’t wait for the dev-release) see http://drupal.org/patch/apply.

@Dave Reid:

They won't solve the problem cited in #18, I am afraid.

subscribe

I've applied the patch which works great, except that it's not scrubbing punctuation (at least not apostrophes, like in don't or wasn't). I am using a [menupath-raw] token with one of my content types. Does this bug exist with the new 6.x-2.0 revision candidate?

I have the same problem with a token called [field_alias-raw] as described here:
#882166: Not 'cleaning' tokens after upgrade to 6.x-1.4

6.x-1.x-dev doesn't solve the problem so I went back to 6.x-1.3.

No because that the slash character is configurable now. For instance the default for character replacement in pathauto is to remove. If we were to run cleanstring on a [menupath] token like 'Parent menu item/Clild parent item/My node title' would end up like 'parent-menu-itemchild-parent-itemmy-node-title'. The only way forward is to solve #22.

To anyone else posting "I'm having the problem too" comments, they're not helpful at this point, so please don't. Let's focus on solutions at this point. And that requires constructive, valid input.

From talking this over with Dave Reid. I think for 6.x we should try to take steps that will be minimally disruptive and then fix it "for real" for Drupal 7.

@_double - you mean from http://drupal.org/project/content_taxonomy ? That module would need to fix it's naming.

Ok, so patch against 6.x-2.x that will apply pathauto cleanstring'ing to all -path* tokens except for node-path.

The last submitted patch, 881270-pathauto-stupid-path-tokens.patch, failed testing.

The last submitted patch, 881270-pathauto-stupid-path-tokens.patch, failed testing.

subscribing.

can someone pls change the title to something like
"Pathauto bulk generating/updating doesn't work - wrong aliases are generated"
as in my opinion present title doesn't reflect the problem to someone who's searching for solution to it and we're getting duplicates

cheers

I'm experiencing the same problem after updating.

Same thing here I rolled back/sideways to 1.x-dev and the problem cleared up. subscribing

subscribing.

For example, a Webform page with [menupath-raw].htm:

Title: Coupon
Placement: Primary Links: Homes

Result: Homes/Self%20Qualifying%20Coupon.htm

[menupath].htm:

Result: Homes/Self%20Qualifying%20Coupon.htm

For me I typically just follow what Drupal report and drush say is available as a stable release. I am more comfortable doing that than throwing on .dev, unless I have read through issue queues.

subscribe

This is super critical, because using the [menupath] (very very common on all my customer drupal installations) will not be rewritten (no lower case, transliteration at all).

I had to correct all lot of aliases because I updated to 1.4. Please release a new stable version as soon as possible.

Does the dev release resolve the problem? I just went back to 1.3 since 1.4 broke my setup...

I tried installing the dev version, but it didn't help -- I just created a page and the path didn't use the separators. The pattern is: [fullhierarchytitlepath-raw]

Subscribe. I love using [save-lineage-vid-termpath-raw].

I have had problem described first by psmith when I used Token 6.x-1.14 and Pathauto 6.x-1.4 I got for pattern [menupath-raw] this output: "úvod" was title of page and generated url was "úvod" too. But I tested Pathauto 6.x-2.x-dev and same Token version and problem is now like this: "úvod" was title of page and generated url was "vod". It looks like the problem described in title of this topic is solved, but now in devel version is another problem (in short - pattern removes [not substitutes] letters written in local language).

Thanks for the fix in 6.x-1.x-dev. A 1.5 release would be nice so I don't have to explain why a "development version" is running live.

The 6.x-1.x-dev version still doesn't clean some tokens : I encounter the issue with [menu-trail-parents-path-raw] as well as the non-raw version [menu-trail-parents-path], which are tokens from the Menu Trails module. They were working fine with Pathauto 6.x-1.3.

subscribe

subscribe

subscribing

Following #66 the attached patches (branches 1.x and 2.x) introduce support for segment cleaning for tokens ending with "hierarchy"/"hierarchy-raw".

This is needed by #867288-5: Pathauto path generation.

@Dave Reid:

Honestly the best approach forward here for D7 is to *stop* using any kind of tree tokens at all. Do not allow them. Enforce use of [mything:parent:alias]/[mything:name].

I ain't sure I follow you: would the [mything:parent:alias] token return a string like 'foo/bar/baz'? If so it seems you are proposing to call those kind of tokens 'parent'. If not how would I get a string with a dynamic number of levels?

I'm still at a loss on how to address properly for D6.

Cannot we agree on a "hierarchy"/"parent"/"tree" suffix to trigger the segment-cleaning mode?

Maybe that is a good question to ask. Which is more of a priority: having the ability to convert the forward slash or being able to use forward slash delimited segments?

::

subscribe

@Dave Reid:

I really like solution #2, I'd make only a simple change to it: modules implementing hook_token_values() could return their favorite segment delimiter through the following (or similar) data structure array('segments' => $term_tree, 'delimiter' => ':').

That said, IMHO the right way to proceed here is roll everything back for the 6.x-1.x-dev branch, since there are lots of broken tokens out there right now, and implement #2 in the 6.x-2.x-dev and 7.x-1.x-dev branches.

Edit: Perhaps $options['context'] = 'pathauto' would be more general.

I think $options['context'] = 'path' is a more generic and still descriptive enough. I'm not prideful enough to think pathauto deserves it's own context ;)

I could go either way on whether we make this change to the 1.x branch. I think it's up to Dave since he's doing the work. My impression is that the most popular tokens affected by this are in pathauto/token module itself and therefore we know we can fix those quickly.

subscribing

A 'paths' context doesn't really make sense to me because this is a special case that Pathauto only needs an array. Although html and plain text contexts make sense - URLs would just be 'plain text' though I think - in which case it would make more sense to use $options['html'] = TRUE just like we do elsewhere in Drupal hooks. Normally we would never *ever* want that. We can bikeshed this as well: maybe use 'path_array' => TRUE to show it's explicit that fake paths need to return an array. I'm still partial to 'pathauto' => TRUE.

I'm looking into whether this change would break backwards compatibility with 6.x-1.x, but I have a feeling it will be fairly easy to fix the major modules (we can coordinate a pathauto/token release).

subscribing

Another consideration: if use an option like 'path_array' => TRUE or 'segments' => TRUE we won't need to make Token tokens aware of Pathauto existence. Theorically any module could exploit that option.

subscribe

subscribing

Attached are the current patches for both pathauto and token to implement #86-2.

$options was not being accepted by pathauto_token_values so I added that.

Then I got hung up on _token_array_merge which very handily navigates the token tree, looks for tokens that are an array, and squashes the array down to a single string assuming that this is an error.

We could add in logic to that function that looks to see if it's a strpos($token, 'path') !== FALSE token and, if so, allow it to have an array. That doesn't guarantee that it's behaving properly, however, so really we should also test !empty($options['pathauto']).

That would fix the problem but it puts pathauto specific code into token which feels wrong.

Can we use a more generic name for the $options value now? ;)

The last submitted patch, 881270-token-path-tokens_109.patch, failed testing.

#109 has been committed to CVS for 6.x-2.x. Will need to be ported to 6.x-1.x. Likely it will need to be brought forward to 7.x-1.x as well, but there are currently no tokens which require this behavior.

http://drupal.org/cvs?commit=427086

Follow-up patch to cleanup and revise the inline docs.

The proposal in #116 doesn't look intuitive to me.

subscribing

Trailing whitespaces.

Forgive my english: is "an value" correct?

Maybe "Token"?

Powered by Dreditor.

#115 and fixes in #121 committed to 6.x-2.x.

I have the drupal 6 dev version and [menupath-raw] is still not getting cleaned. Is there a particular version of Tokens I need to have?

Or Pathauto 6.x-1.x-dev tarball from tomorrow or CVS today.

I'm having this bug on a live site after upgrading as well. menupath-raw.
I just tested 6.x-1.x-dev from 2010-Sep-27. It did not work.

Update: Per suggestion in #132, I also upgraded Token to the dev version and now it works.

Latest 6.x-1.x-dev fixed issue that I had with [term-raw]/[bookpath-raw]/[title-raw] token...

@Dave Reid #132: I didn't realize I had to update Token as well. I've done that and now the menupath-raw tokens are working again. I should have read the whole thread I guess. Sorry. Thanks for the quick reply and the help!

I tested with Token dev tarball (Sep 25) and Pathauto dev (Sep 28). Translitteration is on and Pathauto settings is to change aliases to lowercase.

While built-in tokens are cleaned correctly ([title-raw], [menupath-raw]), tokens set by third-party modules are not cleaned at all (for example, [menu-trail-parents-path-raw] from Menutrails module). Will every third party module exporting tokens have to update its code, unless users will get incorrect aliases containing upper-case and non-translitterated characters?

Thanks Network. If modules exposing tokens that can be used by Pathauto require updating, I suggest to put a notice on Pathauto's module page.

Is this confirmed? I need to know so I can add an issue for modules that need to update their tokens.

I tried upgrading to the latest PathAuto 6.x-1.x-dev and Token 6.x-1.x-dev releases, but it still isn't replacing spaces and such correctly like it did in version 1.3. I'm using this for my pattern:

[fullhierarchytitlepath-raw]

This is what it did with the path:
/Student%20Resources/Clubs%20%2526%20Activities/Phi%20Theta%20Kappa/test%20test%20page

It should have put dashes where it found spaces.

I'm using the latest release of Node Hierarchy: 6.x-2.x-dev

@Dave Reid

Where is the most complete documentation of the new format?

::

Thanks.

::

@pianomansam - You must also upgrade Token to the latest dev.

Reading comprehension ftl.

The security fix was not the problem, it was a combination of bug fixes and feature requests that would have been difficult to roll back and pissed more people off. This has *already* been fixed in 6.x-1.x-dev and 6.x-2.x-dev and we're rolling new releases today.

I just tried with Pathauto 6.x-1.5 and Token 6.x-1.15 but to no avail: My problem from #29 stays the same. I use a CCK field ("alias") to create the path but

* Spaces are not converted into dashes (the separator)
* Special characters are not replaced at all (should be as defined in i18n-ascii.txt)
* "Reduce strings to letters and numbers from ASCII-96" has no effect

Going back to insecure 6.x-1.3 again.

@xeontheme - this is your first comment in the issue and all you have to say is a complaint? You also haven't contributed (in the form of code, documentation, simpletests, money, etc. as far as I can tell) to the improvement of module quality. And yet you expect me to somehow provide you with free services *and* listen to your complaints? Your karma just went down in my book. Note that the GPL comes with no warranty, but I offer a money back warranty on all my modules. If you have a bug I will fully refund the price you paid me for them.

@yan - I answered you already in http://drupal.org/node/881270#comment-3490014 - you seem to be ignoring it. If you change the status again I will see about having your account blocked for inappropriate use of the issue queue. If my explanation is insufficient for you, open a new issue.

btw, if there are names that don't work with pathauto, shouldn't there be a warning or a hint? Or did I just not see it?

updating to pathauto 1.5 and Token 1.15 solved my issues with [menupath-raw] . Thank you !

just remember to delete and regenerate your aliases.

Thanks Dave. That means that modules shouldn't 'send' tokens that have the mentioned patterns to Pathauto, right? Now, CCK allows free naming of fields which can easily lead to a situation like mine (naming a field 'alias' or so). What would be the steps to prevent that those fields are treated as 'special' fields? Should the CCK module take care of that, i.e. for example by adding a -cck or whatever? Or would it be a task for the Token module? Or maybe a warning on the Pathauto settings page if such a pattern is detected?

Can't remember seeing you contribute the solution..
I also don't understand what you mean by '...the general way in which Drupal is carried along.." ?

In another issue, please.

@yan - The answer is to add a new field to the node, do a database level update query, and then delete the old field.

Upgraded Pathauto 6.15 and Token to 6.15, my aliases still aren't changed to lowercase for [menupath] or [menupath-raw].

Is this something I will be able to do with the recent changes? I like to have it set as menupath so the alias looks like path_section_title/parent_page_title/child_page_title.htm.

Is there another token that could facilitate this?

@Kevin: That usually means you have a sneaky version of Token hiding somewhere else. You can check your {system} table for the filepath of the module.

subscribing

Thanks for the in-depth knowledge passed on here. The karma is great just a couple of entries from those not perceptive of the work and thinking going on here.

I am new here nothing loaded yet just reading the extensive documentation.

As a newbie should I load Pathauto 6 or 7?

Menupath token seems to be working but hierarchypath is not. Honestly speaking I never know the difference between the two :)
For previous pathauto versions I guess hierarchypath worked for me thats why I used it. Anybody knows the difference?

Yes I just got the hold of the issue and finished reading the node hierarchy issue. Came here to update my post that Its a nodehierarchy issue but you were prompt and replied before me. My apologies, I should research more before posting, Probably the reason why I don't post in issues. Great work Dave!
Regards

if it's nodes, go to admin/content/node, select the nodes that you like to update and select Update URL alias from the select list form to update.

If there are hundreds or more of nodes use Views Bulk Operations.

Hi,

I am not entirely sure if I should be posting my question here or in the token issue queue. However, this seems to be the cause for the problem I am seeing. I am fairly new to Drupal and have never build my own tokens. Could someone please clarify the impact of this discussion on custom tokens?

In the site I maintain a custom token is generated in a module. These tokens stopped working properly after the upgrade to Token 6.x-1.15 and Pathauto 6.x-1.5 (punctuation and spaces not removed, not converted to lower case etc.). Below is the code to create a limited menupath (2 levels only). The site uses the workflow module and hence needed some extra checks because workflow influenced the values that some of the functions returned.

Based on Dave's comment in 142, is it enough to just replace

Sorry about this lengthy post, but with my limited Drupal experience I do not get the full picture of this discussion and would appreciate your help. Thank you.

Has anyone else experienced this issue for FileField Paths while using PathAuto 6.x-1.5 and Token 6.x-1.15 (and FileField Paths version 6.x-1.4)?

I am using [menupath]. Prior to the security update for PathAuto, my FileField Paths for file attachments would look like:

"animals/cats-dogs/kitty.jpg"

Since the upgrade, a new attachment to the same node is created in a new folder called:

"animals/cats and dogs/kitty.jpg"

Most Pathauto cleanup settings seem to be ignored.

- The letters ARE transformed to lower case
- Separator "-" is ignored, an empty space is shown between the words
- Special characters are not being stripped out as they previously were

This issue started occurring after we upgraded PathAuto to version 6.x-1.5 in late October '10. I assume we jumped right from 6.x-1.3 to 1.5.

We do use Menu Trails on our site, but I wasn't sure if this mattered being we are using [menupath] and not the custom tokens from the Menu Trails module?

Any help or insight is appreciated.