LCOV reporter for Mocha.
LCOV format can be found in this geninfo manpage. This LCOV reporter was built after Sonar Javascript Plugin LCOVParser class.
The mocha-lcov-reporter is a reporter for mocha. In order to get coverage data, the same instructions are to be followed as for the JSONCov
and HTMLCov
reporters:
- Install jscover or node-jscoverage
- Instrument your library with
jscover
(ornode-jscoverage
) - Run your tests against your instrumented library and save the output
For example, the following script can be part of your build process (add jscover
, mocha
, and mocha-lcov-reporter
as devDependencies
to your package.json
file and run npm install
):
#!/usr/bin/env bash
rm -rf coverage
rm -rf lib-cov
mkdir coverage
node_modules/.bin/jscover lib lib-cov
mv lib lib-orig
mv lib-cov lib
node_modules/.bin/mocha -R mocha-lcov-reporter > coverage/coverage.lcov
rm -rf lib
mv lib-orig lib
This script instruments your sources (source 'lib', target 'lib-cov'), temporarily replaces your library by the instrumented version, run the tests against the instrumented version of your sources, and undoes the replacing of the original library by the instrumented library.
A safer and better approach is to instrument your library (target 'lib-cov'), and include that directory from your tests directly. Instead of doing a 'require("../lib")' do a 'require("../lib-cov")'. This saves the hassle of replacing directory 'lib' with directory 'lib-cov' and undoing it afterwards. You can use an environment variable to check if the instrumented library should be included or the normal version:
var lib = process.env.JSCOV ? require('../lib-cov') : require('../lib');
And to get the test-coverage, run mocha as follows:
JSCOV=1 mocha -R mocha-lcov-reporter > coverage/coverage.lcov
See the SaXPath project for an example on how to do this.
Unfortunately, when the code is instrumented using jscover
or node-jscoverage
, the output of the reporter will have incomplete paths for the covered files. A quick fix is to update the paths after running the tests with the mocha-lcov-reporter, like so:
# run the tests
JS_COV=1 ./node_modules/.bin/mocha -R mocha-lcov-reporter > coverage/coverage_temp.lcov
# fix the paths
sed 's,SF:,SF:lib/,' coverage/coverage_temp.lcov > coverage/coverage.lcov
The reason this is that jscover
runs on the directory you specify (e.g., lib/
) and regards that as the root for the project.
Blanket.js can be used as well. After the lcov file, be sure to fix the paths for the covered files. The path will be an URL, having file:
as its protocol. Using the same manner as above, the path can be fixed using sed
.
What does LCOV output look like? LCOV is meant to be interpreted by other programs and not meant to be readable by humans. This is an example:
SF:base_unit.js
DA:1,1
DA:4,1
DA:5,155
DA:7,155
DA:8,140
DA:9,140
DA:12,155
DA:13,155
DA:16,1
DA:17,1
DA:20,1
DA:21,9
DA:24,1
DA:25,40
DA:28,1
DA:29,26
DA:32,1
DA:33,7
DA:36,1
DA:37,6
DA:40,1
DA:41,45
DA:44,1
DA:45,52
DA:51,1
DA:52,3
DA:55,1
end_of_record
If you are looking for something human readable, the HTMLCov
reporter can be used.