Provides a way to test textmate grammars against a vscode engine using user-friendly plaintext files.
Demo:
Inspired by Sublime Text syntax tests
As a project dependency:
npm i --save-dev vscode-tmgrammar-test
Or as a standalone command line tool:
npm i -g vscode-tmgrammar-test
vscode-tmgrammar-test --help
// SYNTAX TEST "source.scala" "sample testcase"
// line can start with a <comment token> and not have a valid assertion
class Stack[A] {
// <----- keyword.declaration.scala
// ^ - keyword.declaration.scala entity.name.class.declaration
// ^^^^^ entity.name.class.declaration
// ^ source.scala meta.bracket.scala
// ^ entity.name.class
// ^ meta.bracket.scala
// ^ punctuation.section.block.begin.scala
To write a unit test:
- include a header line:
<comment token> SYNTAX TEST "<language scope>" "optional description"
- Require tokens to have specific scope by using
^
:
private var elements: List[A] = Nil
// ^^^^^^^^ variable.other.declaration.scala
- Get into those pesky first few characters by using
<-
:
var x = 3
// <--- keyword.declaration.volatile.scala
// the length of '-' determine how many characters are matched from the start of the line
x=5
// <~- keyword.operator.comparison.scala
// you specify offset from start by using '~' character, just in case
- To ensure that tokens don't have undesired scopes put
-
symbol before them:
/ ensure comment start with two double slashes
^ - comment.line.double slash.scala
/ or you can combine both positive and negative scopes
^ source.scala - comment.line.double slash.scala
Lines which start with a <comment token>
and assertion symbol are ignored by the textmate grammar.
Note, that scope comparison takes into account relative scope's position.
So, if required scopes are 'scope1 scope2'
, the test will report an error if a grammar returns them as 'scope2 scope1'
.
To run a unit test:
vscode-tmgrammar-test 'tests/unit/**/*.test.scala'
Snapshot tests are like functional tests
but you don't have to write outputs explicitly.
All you have to do is to provide a source files, scopes of which you want to test. Then on
the first run vscode-tmgrammar-snap
will generate a set of .snap
files which are an
instant snapshot of lines of the source files together with corresponding scopes.
Then if you change the grammar and run the test again, the program will output the changes between
the .snap
file and the real output.
If you satisfied with the changes you can commit
them by running
vscode-tmgrammar-snap --updateSnapshot ....
this will overwrite the existing .snap
files with a new ones.
After this you should commit them alongside with the source code test cases.
You can read more about them at snapshot testing
To run snapshot test:
vscode-tmgrammar-snap 'tests/snap/**/*.scala'
The configuration follows the format of vscode:
{
"contributes": {
"languages": [
{
"id": "scala",
"extensions": [
".scala",
".sbt",
".sc"
]
}
],
"grammars": [
{
"language": "scala",
"scopeName": "source.scala",
"path": "./syntaxes/Scala.tmLanguage.json"
}
]
}
}
The idea is that for the average language extension all necessary information for tests are already included in the package.json
.
It is optional, though. If the configuration is missing it is necessary to specify grammars and scopeName of testcases via command line options.
Right now only regular grammars and Injection Grammars via injectTo
directive are supported.
Unit tests:
Usage: vscode-tmgrammar-test [options] <testcases...>
Run Textmate grammar test cases using vscode-textmate
Arguments:
testcases A glob pattern(s) which specifies testcases to run, e.g. "./tests/**/test*.dhall". Quotes are important!
Options:
-g, --grammar <grammar> Path to a grammar file. Multiple options supported. 'scopeName' is taken from the grammar (default: [])
--config <configuration.json> Path to the language configuration, package.json by default
-c, --compact Display output in the compact format, which is easier to use with VSCode problem matchers
--xunit-report <report.xml> Path to directory where test reports in the XUnit format will
be emitted in addition to console output
--xunit-format <generic|gitlab> Format of XML reports generated when --xunit-report is used.
`gitlab` format is suitable for viewing the results in GitLab
-V, --version output the version number
-h, --help display help for command
Snapshot tests:
Usage: vscode-tmgrammar-snap [options] <testcases...>
Run VSCode textmate grammar snapshot tests
Arguments:
testcases A glob pattern(s) which specifies testcases to run, e.g. "./tests/**/test*.dhall". Quotes are important!
Options:
-u, --updateSnapshot overwrite all snap files with new changes
--config <configuration.json> Path to the language configuration, package.json by default
--printNotModified include not modified scopes in the output (default: false)
--expandDiff produce each diff on two lines prefixed with "++" and "--" (default: false)
-g, --grammar <grammar> Path to a grammar file. Multiple options supported. 'scopeName' is taken from the grammar (default: [])
-s, --scope <scope> Explicitly specify scope of testcases, e.g. source.dhall
-V, --version output the version number
-h, --help display help for command
You can setup a vscode unit test task for convenience:
{
"label": "Run tests",
"type": "shell",
"command": "vscode-tmgrammar-test -c -g testcase/dhall.tmLanguage.json '**/*.dhall'",
"group": "test",
"presentation": {
"reveal": "always",
"panel":"new",
},
"problemMatcher": {
"owner": "vscode-tmgrammar-test",
"fileLocation": [
"relative",
"${workspaceFolder}",
],
"pattern": [
{
"regexp": "^(ERROR)\\s([^:]+):(\\d+):(\\d+):(\\d+)\\s(.*)$",
"severity": 1,
"file": 2,
"line": 3,
"column": 4,
"endColumn": 5,
"message": 6,
},
],
},
},
Notice the -c
option that will output messages in a handy format for the problemMatcher.
Result: