577006 - Can't access phone book.

Status: RESOLVED FIXED

(mozilla.org Graveyard :: Webdev, task)

Description

[John Hopkins (:jhopkins)]

My LDAP login ([e-mail]) is failing against https://ldap.mozilla.org/phonebook/

Can you check my account and/or reset my password, please?

Comment 1

[Aravind Gottipati [:aravind]]

Your account did not have access to the phonebook.  I enabled it, please try now and let me know if it doesn't work.

Comment 2

[John Hopkins (:jhopkins)]

Aravind, I still cannot login to the phonebook.

Comment 3

[John Hopkins (:jhopkins)]

justdave says my LDAP record is corrupted on at least 1 of the LDAP servers.

Comment 4

[Aravind Gottipati [:aravind]]

(In reply to comment #3)
> justdave says my LDAP record is corrupted on at least 1 of the LDAP servers.

The ldap entry looks fine.  The app is somehow getting confused and throwing all sorts of errors..

To replicate the problem, go to "https://ldap.mozilla.org/phonebook/#search/john%20hopkins"

Comment 5

[Wilson Lee [:kourge]]

Attachment: Make output-html.inc tread carefully

Comment 6

[Alex Buchanan [:abuchanan]]

Comment on attachment 458446 [details] [diff] [review]
Make output-html.inc tread carefully

Looks ok, but I didn't setup phonebook to test, so please test before pushing to production

Comment 7

[Wilson Lee [:kourge]]

I tested it on my side and it seems to run fine.

Comment 8

[Wilson Lee [:kourge]]

Just realized that the issue here is two-fold:
1. Some people have so little information on their entry that they cause the "offset undefined" error message(s) when they show up in a search result. The existing patch fixes that already.
2. The dot (.) character was not considered to be a valid username. A patch will shortly be attached that fixes this.

Comment 9

[Wilson Lee [:kourge]]

Attachment: Include dot (.) as a valid character in usernames

Comment 10

[Alex Buchanan [:abuchanan]]

Comment on attachment 458798 [details] [diff] [review]
Include dot (.) as a valid character in usernames

I haven't tested, but if it fixes the problem then r+

It seems we have a lot of issues with this part of the app, and that parsing email addresses is more complicated than a custom regex can handle.

Check this out,
http://www.iamcal.com/publish/articles/php/parsing_email/

It explains just how difficult email regexps are, and includes code to properly split them up.  You might consider using that instead.

Comment 11

[Wilson Lee [:kourge]]

A proper way to parse email addresses would only be needed when the email addresses allowed are not under our direct control. For example, we wouldn't need to parse route-addrs per RFC 822. The only reason we're even parsing email addresses is because we need some way to represent algorithmically how IT places an entry under which O of the mozilla Domain Component, which is so far reliably predictable.

Comment 12

[Wilson Lee [:kourge]]

Committed r71838.