[Core] Implement cucumber expressions

[mpkorstanje]

Summary

• Implements Cucumber expressions
• Removes XStream
• Adds DataTables from add datatable/java common#291
• Introduces the concept of StepExpressions which combines a CucumberExpression and DataTable expression. Currently data table expressions are limited to a name or java type. Currently only the implicit type derived from the method argument is used.

For a working example check the calculator examples.

How to build

Install (mvn install) datatables/java from cucumber/cucumber
Then install this branch. If it fails you may need to remove pax-exam from examples/pom,xml. This is okay.

Todo

• Fix android examples
  * [ ] Add link to configuration docs in exceptions
• Test SnippetGenerators and StepdefGenerator icm Idea
  * [ ] Update documentation
• Write blog post
  * [ ] File IDEA bug about "Pattern expected" warning
• Undefined parameter type error should be similarly friendly as missing table type

This closes #1041

[mlvandijk]

Update:
Replacing parameter Integer with Byte in step def:

    @Given("I have {byte} cucumbers in my belly/stomach")
    public void i_have_byte_cucumbers_in_my_belly(Byte byte1) {
        // Write code here that turns the phrase above into concrete actions
        throw new PendingException();
    }

This works for valid values (e.g. 0x7F), but not for invalid values (same stacktrace as above).

Note: Generated snippets contain {int}, e.g.:

@Given("I have {int}x{int}F cucumbers in my belly")
public void i_have_x_F_cucumbers_in_my_belly(Integer int1, Integer int2) {
   // Write code here that turns the phrase above into concrete actions
   throw new PendingException();
}

Some examples here: mlvandijk/cucumber-expression-test@1d58b06

[mpkorstanje]

I've had to look up how those worked again and the rules downright rather complicated because java treats them as 2s-complement numbers.

As a normal person I generally expect 0xff to give me a bit set of 11111111. In java you get 011111111. We could make it follow that expectation but that might just make it confusing for people who are familiar with the regular java behavior.

So as this is all a bit annoying. So lets byte the bullet and drop it. The people who want to use it can sink their teeth in it.

[aslakhellesoy]

I pushed cucumber/common@ecab2ab where {byte} use the same decimal pattern as {short}, {int} and {long}. That's less confusing.

I considered adding a {hex} type with pattern 0[xX][0-9a-fA-F]{1,16} that would resolve to a Long. It could then be down-cast to any integer type when the stepdef gets called. But I resisted the temptation since I should add it to the 3 other implementations as well. If people really want to write hex in Gherkin they can write their own bloody stepdefs! 😈

[mpkorstanje]

Hallelujah!

[mpkorstanje]

Explain cucumber.api.Configuration in error message (unless it already exists)

Got the infra structure set up for that. A full blown explanation depending on the existence of the configuration adds too much complexity. As does combining all the messages in a completely sensible way. People can read a stack trace.

Just need to add a link to the documentation.

[mpkorstanje]

Step expression should be completely hidden from the user. They'll only ever see the cucumber.api.TypeRegistry, they won't ever see io.cucumber.stepexpression.TypeRegistry.

The purpose of the package is to abstract the cucumber expressions and data table types and all the transformations involved from the java and java8 step definitions and the step definition match.

This means:

• Creating a step expresssion from a cucumber expression + optional data table or doc string type
• Combining the pickle parameters and pickle arguments into a list of method arguments (combining these makes things much simpler)
• Transforming these arguments into java objects.

This used to be done in PickleStepDefinitionMatch.transformedArgs and ParameterInfo .convert neither of which should be concerned with transformations. Now it has been contained.

It also required LocalizedXStreams to be passed around everywhere. The TypeRegistry goes from the Runtime to the Backend to the StepDefinition and that is it.

[mpkorstanje]

easier for us to version.

I've automated it a while ago. Break the api, break the build.

https://github.com/cucumber/cucumber-jvm/blob/master/pom.xml#L866

edit: Ofcourse it doesn't catch protocol level changes such as the formatter event streams.

edit; And if you're looking to test revapi. Set the version back to 2.x. Breaks are allowed on majors.

[mpkorstanje]

Blog post drafted: https://github.com/cucumber/website/pull/290

[mpkorstanje]

@aslakhellesoy StepdefGenerator doesn't appear to be used by IDEA. Or anybody on Github. You sure it's not dead code?

[aslakhellesoy]

Yeah just get rid of StepdefGenerator. It's old dead code.

[mpkorstanje]

I'm going to split writing the docs regarding the configuration from this issue. The current docs are talking about a completely different configurations. This will take some work. And while that is being done I'd rather see another PR come in that is going to conflict with everything.

The IDEA people will need a release to work with so that also needs a merge.

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.