Total coverage no longer easily regexable

[desmondgc]

In coverlet.msbuild v2.4.0, the total coverage output looks like this:

Total Line: 83.3%
Total Branch: 86.4%
Total Method: 83.9%

This functionality was added in #180 to support coverage badges in CI systems. In our GitLab config I can easily report on total branch coverage using the following: coverage: '/^Total Branch: \d+\.\d+/'

With the changes introduced in #271 / coverlet.msbuild v2.5.0, the output now looks like this:

+---------+--------+--------+--------+
|         | Line   | Branch | Method |
+---------+--------+--------+--------+
| Total   | 83.3%  | 86.4%  | 83.9%  |
+---------+--------+--------+--------+
| Average | 41.65% | 43.2%  | 41.95% |
+---------+--------+--------+--------+

This output format is much harder to work with. Can we eliminate the table?

[m4cx]

With this regex /\|\sTotal\s{1,}\|\s[^|]*\|\s\d+.\d+% I can find the string needed. But how exactly does gitlab handle the multiple numeric values? Does anyone know that?

I am not that used to regex. Is it possible with regex to remove the Line value from the string?

[desmondgc]

GitLab's coverage regex will use a capture group, if present. So in this case, I can capture total branch coverage with the following regex: /^\|\s*Total\s*\|[^|]*\|\s*(\d+\.\d+)/

The point is, the output was originally added with an implicit requirement to be easily parsable. What used to be a simple and straightforward regex has become an ugly mess. Maybe we can add a switch for output format: table vs flat?

[m4cx]

Then I think the switch would be the appropriate solution. For humans this table view might quite appealing. But for parsing the simpe text output is perfekt

[anaether]

^\

The problem on your solution is that it dont works if you have full percent values.

This works:
18.00%
18.0%

This dont works:
18%

[anaether]

@desmondgc
My version works and fixes your bug (now it works also with integer values):
^|\sTotal\s|[^|]|\s(\d+.\d+ | [\d.]+)

If you want to use it, test it here:
https://rubular.com/

Sample Values:
+----------------------+-------+--------+--------+
| Module | Line | Branch | Method |
+----------------------+-------+--------+--------+
| Project | 6.3% | 18% | 11.6% |
+----------------------+-------+--------+--------+
| ProjectTest | 97.8% | 100% | 87.5% |
+----------------------+-------+--------+--------+

+---------+-------+--------+--------+
| | Line | Branch | Method |
+---------+-------+--------+--------+
| Total | 17% | 0% | 13.9% |
+---------+-------+--------+--------+
| Average | 3.95% | 9% | 6.95% |
+---------+-------+--------+--------+

[yauhenyaulampyeu]

| ?Total +| +(\d+.\d+|\d+)%
works fine

[gmarokov]

None of the above regex did the trick for me. What actually helped Total.*?(\d+(?:\.\d+)?)%

[joan-grau]

In my case, what separates the decimals from integers is a comma, so What actually helped me si Total.*?(\d+(?:\,\d+)?)%

[MarcoRossignoli]

Close for stale, related feature issue to avoid regex usage #542