It seems that SARB is only aware of the concept of Error; it doesn't know anything about warnings or what it should do with them.

In its most simplistic form, I could say that a warning (or a list of warnings) should be shown, but the tool should return success instead of failure, in case the baseline contains only warnings.

Since this is a fairly complex feature to implement and extra caution must be taken in order not to introduce any BC-breaking changes, I thought it would be better to ask, is submitting a Pull Request that introduces the concept of severity to SARB, has any meaning.

Thanks in advance, Alex.

Hi, i am using PHPStan (tried either with text and json format) on file with error, and just copied it as latest errors, created sarb baseline and got false positive.

Steps to reproduce:

Output of vendor/bin/phpstan analyse --error-format=raw > reports/phpstan_errors_baseline.txt (phpstan_errors_baseline.txt):

/app/presenters/ExitPresenter.php:165:Cannot access property $brand on bool|Klarka\Item\Item.

cp reports/phpstan_errors_baseline.txt reports/phpstan_errors_latest.txt to simulate error was not fixed.

Create baseline: vendor/bin/sarb create-baseline reports/phpstan_errors_baseline.txt reports/sarb_baseline.json phpstan-text-tmp

And then compare: vendor/bin/sarb remove-baseline-results reports/phpstan_errors_latest.txt reports/sarb_baseline.json reports/errors_since_baseline.txt -f

(false positive) Output is:

Baseline uses ResultsParser [phpstan-text-tmp] and HistoryAnalyser [git]
Errors before baseline 1
Errors in baseline 1
Errors introduced since baseline 1

FILE: app/presenters/ExitPresenter.php
+------+---------------------------------------------------------+
| Line | Description                                             |
+------+---------------------------------------------------------+
| 165  | Cannot access property $brand on bool|Klarka\Item\Item. |
+------+---------------------------------------------------------+

Here is content of sarb_baseline.json if that helps:

{"historyAnalyser":"git","historyMarker":"8f7184af1d682f5ba2acff3e06c2b2196d6f5afc","resultsParser":"phpstan-text-tmp","analysisResults":[{"lineNumber":165,"fileName":"app\/presenters\/ExitPresenter.php","type":"Cannot access property $brand on","message":"Cannot access property $brand on bool|Klarka\\Item\\Item.","fullDetails":"\/app\/presenters\/ExitPresenter.php:165:Cannot access property $brand on bool|Klarka\\Item\\Item."}]}

Do you have any idea what am i doing wrong?

Add support for php code sniffer json format.

@DaveLiddament Got any better idea how to deal with the closures in PhpCodeSnifferJsonResultsParser::convertToString()? I need information from the full details, and didn't find anything else...

It will be nice to add Code_Sniffer here. Why? The same thing as phpstan. If applying for old project, it can show you tons of issues that you have in your code.

In v1.4.0 webmozart/path-util was removed - https://github.com/DaveLiddament/sarb/pull/110

Usages however still exist: https://github.com/DaveLiddament/sarb/search?q=PathUtil

Resulting in:

$ cat /tmp/phpcpd-sarb.json | php vendor/bin/sarb remove phpcpd.baseline
Unexpected critical error: Class 'Webmozart\PathUtil\Path' not found

Hi, I'm using phpcs with sarb on windows and encountered an issue today.

Steps to reproduce:

1. Create a file with an error
2. Create baseline
3. Add some lines (error now on a different line)
4. Run phpcs and remove errors with sarb

Expected behavior

Still no errors (the history feature works in my pipeline and my docker containers, which run on linux)

Actual behavior

Errors showing, despite being in the baseline

I'm working on a windows machine. Despite that git commands return \n as line breaks. This means that DaveLiddament\StaticAnalysisResultsBaseliner\Domain\HistoryAnalyser\UnifiedDiffParser\Parser on line 62 doesn't work for me. As far as I know git returns \n on all OS and not just linux. I attached the simplest change in a PR. I'm not quite sure if git is outputting somewhere something else than \n, but if we would want to make sure that would work too, the other proposal would be something like this: preg_split("/\r\n|\n|\r/", $diffAsString)

Long directory paths in phpcs text reports are truncated

FILE: ...eran/test/foo/bar/quz/test.php
--------------------------------------------------------------------------------
FOUND 1 ERROR AFFECTING 1 LINE

Links:

• https://github.com/squizlabs/PHP_CodeSniffer/blob/9c90cbfafe32be55a0b00e2bda4035d48f76a7d9/src/Reports/Full.php#L85
• https://github.com/squizlabs/PHP_CodeSniffer/wiki/Configuration-Options#setting-the-default-report-width

I'd be grateful if the PHPCS documentation was updated to account for this:

vendor/bin/phpcs src/ > /tmp/phpcs.txt

becomes

vendor/bin/phpcs --report-width=auto src/ > /tmp/phpcs.txt

Perhaps with a note explaining what --report-width is for in case the user has to run a different command, for example I used a modified command:

./vendor/bin/phpcs --report-width=auto --standard=phpcs.xml . > /tmp/phpcs.txt

Edit

Alternative workaround - use the undocumented JSON parser https://github.com/DaveLiddament/sarb/pull/60:

$ ./vendor/bin/phpcs --report=json --standard=phpcs.xml . > /tmp/phpcs.txt
$ ./vendor/bin/sarb create-baseline /tmp/phpcs.txt phpcs.baseline phpcodesniffer-json
Baseline created
Errors in baseline 12617

Creating baseline for PHPCS results in empty one for me, vendor/bin/sarb create-baseline /tmp/phpcs.txt phpcs.baseline phpcodesniffer-full:

Baseline created
Errors in baseline 0

I was going by https://github.com/DaveLiddament/sarb/blob/master/docs/PhpCodeSniffer.md

Tried also json output for PHPCS but baseline just stays empty. My git repository is clean, everything committed.

{"historyAnalyser":"git","historyMarker":"97fe7e1fb7c5be68bccb3f86ff97cfb4fe633f1e","resultsParser":"phpcodesniffer-full","analysisResults":[]}

I have suspicion maybe I'm missing something.

Hi @DaveLiddament, you were in contact with some colleagues of mine at the Symfony Live in London. They told me you are interested in supporting deptrac with SARB. Deptrac supports the jUnit format we might could adapt here. There is still an issue https://github.com/sensiolabs-de/deptrac/issues/205 but should be fixable soon. Otherwise we could introduce another formatter. WDYT?

Hello!

We had following error reported by Phpstan:

ProfilePresenter.php:0:ParseError (syntax error, unexpected end of file, expecting function (T_FUNCTION) or const (T_CONST)) thrown while autoloading class Klarka\TagHide\TagHide.

I understand what the error says and it is completely legitimate. Though when running SARB: vendor/bin/sarb remove-baseline-results reports/phpstseline.json reports/errors_since_baseline.txt

There is following error: Unexpected critical error: Line number must be positive integer. Got: 0

I think SARB should allow number 0 in line, because it is legit error and should be reported.

What do you think?

$ composer why webmozart/path-util
vimeo/psalm  4.13.1  requires  webmozart/path-util (^2.3)  

$ composer why vimeo/psalm
dave-liddament/sarb   dev-master  requires (for development)  vimeo/psalm (^4.13)

Please consider phar file distribution https://phar.io/

1. create a phar file with https://github.com/humbug/box
2. add GPG signature
3. upload them as release attachments
4. upload your public key to keyservers
5. add a "sarb" alias to https://github.com/phar-io/phar.io/blob/master/data/repositories.xml

Example output is below. The failures are listed in the 'Escaped mutants' section of the infection-log.txt file. For the example below the failure would be file /projects/danack/github/params/lib/Params/Params.php on line 36 and the error type would be OneZeroInteger.

Escaped mutants:
================


1) /projects/danack/github/params/lib/Params/Params.php:36    [M] OneZeroInteger

--- Original
+++ New
@@ @@
     {
         foreach ($namedRules as $parameterName => $rules) {
             // TODO - test for packed array?
-            if (count($rules) === 0) {
+            if (count($rules) === 1) {
                 throw new RulesEmptyException();
             }
             $firstRule = $rules[0];

PHPUnit 8.1.2 by Sebastian Bergmann and contributors.

...                                                                 3 / 3 (100%)

Time: 29 ms, Memory: 4.00 MB

OK (3 tests, 13 assertions)

Timed Out mutants:
==================

Killed mutants:
===============


1) /projects/danack/github/params/lib/Params/Exception/ValidationException.php:29    [M] OneZeroInteger