Launchpad links from gerrit are truncating the last digit of the bug number.

[Matthew Gilliard]

See my comments in https://review.openstack.org/#/c/60641 - the last digit of the bug number is being dropped so the links are going to the wrong bugs, usually on the wrong project.

[clayg (clay-gerrard)]

Happened here too:

https://review.openstack.org/#/c/70978/1/etc/swift.conf-sample,unified

[Mat Lowery (mat-lowery)]

Saw this too. Note that this does not occur with https://review-dev.openstack.org.

[Mat Lowery (mat-lowery)]

I think the problem is here: https://github.com/openstack-infra/config/blob/d19e0dd58eea7f075ae4f060872f185205268bcd/modules/openstack_project/manifests/gerrit.pp#L154

I don't understand why the launchpad.net bug URLs need truncating in the first place, but I believe the fix is to change the regex to the following:

<a href=\"(https://bugs\\.launchpad\\.net/[a-zA-Z0-9\\-]+/\\+bug/(\\d+))[^\"]*\">[^<]+</a>

As tested by the following Java:

import java.util.regex.Matcher;
import java.util.regex.Pattern;

public class Test {

    public static void showProblem() {
        System.out.println("Problem");
        String r = "<a href=\"(https://bugs\\.launchpad\\.net/[a-zA-Z0-9\\-]+/\\+bug/(\\d+))[^\"]+\">[^<]+</a>";
        String i = "<a href=\"https://bugs.launchpad.net/trove/+bug/1276858\">https://bugs.launchpad.net/trove/+bug/1276858</a>";
        String i2 = "<a href=\"https://bugs.launchpad.net/trove/+bug/1222891/comments/3\">https://bugs.launchpad.net/trove/+bug/1222891/comments/3</a>";
        Pattern pattern = Pattern.compile(r);
        Matcher matcher = pattern.matcher(i);
        matcher.find();
        System.out.println("Orig: " + i);
        System.out.println(String.format("New: <a href=\"%s\">%s</a>", matcher.group(1), matcher.group(1)));
        matcher = pattern.matcher(i2);
        matcher.find();
        System.out.println("Orig: " + i2);
        System.out.println(String.format("New: <a href=\"%s\">%s</a>", matcher.group(1), matcher.group(1)));
    }

    public static void showSolution() {
        System.out.println("\nSolution");
        String r = "<a href=\"(https://bugs\\.launchpad\\.net/[a-zA-Z0-9\\-]+/\\+bug/(\\d+))[^\"]*\">[^<]+</a>";
        String i = "<a href=\"https://bugs.launchpad.net/trove/+bug/1276858\">https://bugs.launchpad.net/trove/+bug/1276858</a>";
        String i2 = "<a href=\"https://bugs.launchpad.net/trove/+bug/1222891/comments/3\">https://bugs.launchpad.net/trove/+bug/1222891/comments/3</a>";
        Pattern pattern = Pattern.compile(r);
        Matcher matcher = pattern.matcher(i);
        matcher.find();
        System.out.println("Orig: " + i);
        System.out.println(String.format("New: <a href=\"%s\">%s</a>", matcher.group(1), matcher.group(1)));
        matcher = pattern.matcher(i2);
        matcher.find();
        System.out.println("Orig: " + i2);
        System.out.println(String.format("New: <a href=\"%s\">%s</a>", matcher.group(1), matcher.group(1)));
    }

    public static void main(String []args) {
        showProblem();
        showSolution();
    }
}

And its output:

Problem
Orig: <a href="https://bugs.launchpad.net/trove/+bug/1276858">https://bugs.launchpad.net/trove/+bug/1276858</a>
New: <a href="https://bugs.launchpad.net/trove/+bug/127685">https://bugs.launchpad.net/trove/+bug/127685</a>
Orig: <a href="https://bugs.launchpad.net/trove/+bug/1222891/comments/3">https://bugs.launchpad.net/trove/+bug/1222891/comments/3</a>
New: <a href="https://bugs.launchpad.net/trove/+bug/1222891">https://bugs.launchpad.net/trove/+bug/1222891</a>

Solution
Orig: <a href="https://bugs.launchpad.net/trove/+bug/1276858">https://bugs.launchpad.net/trove/+bug/1276858</a>
New: <a href="...

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/71743

[Matthew Gilliard (matthew-gilliard-u)]

Thanks Mat. Possibly the reason for wanting to truncate the URL sometimes is this: https://bugs.launchpad.net/openstack-ci/+bug/1222217 ?

[Mat Lowery (mat-lowery)]

Hi Matthew. Thanks for the link. I should have paid attention to the git blame. Some points:

* I tested that the regex in my patch set works to drop the period.
* Dropping punctuation at the end of the URL is a common problem, yet the fix is only for bugs.launchpad.net.
* With the regex the way it is, URLs like https://bugs.launchpad.net/trove/+bug/1222891/comments/3 will be truncated for no good reason. Granted, I'm not sure how useful links of this type are.
* Seems like Gerrit's linkify is what should be changed so that it doesn't wrap trailing punctuation in the <a> tag, despite it being a legal URL. They've had similar problems in the past: https://code.google.com/p/gerrit/issues/detail?id=276 but I tested on review-dev.openstack.org and while parentheses and brackets are left out, periods, commas, and exclamation points are all wrapped in the <a> tag.

[Alexis Lee (alexisl)]

FWIW I'd much rather have trailing punctuation in the URL than vital digits omitted. Let's not overthink this - it's easy enough to include a separating space, EG http://url.com .

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/71743
Committed: https://git.openstack.org/cgit/openstack-infra/config/commit/?id=f9a59e00380746810e237c3b391983e3d86daa46
Submitter: Jenkins
Branch: master

commit f9a59e00380746810e237c3b391983e3d86daa46
Author: Mat Lowery <email address hidden>
Date: Thu Feb 6 23:19:24 2014 -0600

    Fix launchpadbug comment linking in Gerrit

    Change-Id: I4e3dbd0ebc194097db542cfacb79fc2aece64ac6
    Closes-Bug: #1276449