angularjs-test-coverage-inventory
Read-only mining codemod that correlates AngularJS app artifacts (module,
component, controller, service, directive registrations) with test signals
found in the same file, to help estimate test coverage risk before an
AngularJS → Angular migration.
Emits the test_coverage_signal metric. Does not modify any files.
What it detects
For every JS/TS/TSX file scanned:
test_file— the filename matches a common test convention:
*.spec.js,*.test.js,*_test.js,*-test.js.test_marker— a Jasmine/Mocha/angular-mockscall is present in the
file:describe(,xdescribe(,fdescribe(,it(,xit(,fit(,
beforeEach(,afterEach(,beforeAll(,afterAll(,inject(,
module(.angular_artifact_without_test_markers— the file registers an
AngularJSmodule/component/controller/service/directive(via
angular.module('app').<method>(...)), the filename does not look
like a test file, and the file contains no test markers of its own.
Cardinalities
| Field | Description |
|---|---|
file | Relative file path |
line | 1-based line number of the match |
snippet | Truncated source snippet (whitespace-collapsed, 120 chars) |
signal_kind | test_file | test_marker | angular_artifact_without_test_markers |
artifact_type | For test_file: test_file. For test_marker: the marker name (e.g. describe, inject). For the missing-test signal: the AngularJS registration method (module, component, controller, service, directive). |
migration_category | Always testing |
risk | low | medium | high | critical |
automation | full | partial | manual |
effort_points | Fibonacci effort estimate (1, 2, 5, ...) |
recommended_phase | phase-1 .. phase-5 |
Risk/effort heuristics:
- Jasmine/Mocha lifecycle markers (
describe,it,beforeEach, ...) are
low risk / fully automatable (effort_points: 1) — syntax carries over
almost directly to Jest/Karma+Jasmine on Angular. inject(...)andmodule(...)(fromangular-mocks) are medium risk /
partially automatable (effort_points: 2) since they require a real
rewrite toTestBed.configureTestingModule(...).- AngularJS registrations with no test markers in the same file are
flagged high risk / manual (effort_points: 5,phase-1): write or
backfill tests before migrating that code, since regressions will be
invisible otherwise.
Known limitation: file-local analysis only
JSSG codemods run per-file and cannot see sibling files. This codemod
cannot confirm whether a separate *.spec.js exists elsewhere for a
given source file — it can only tell you "this file has no test markers in
it". A source file with describe(/it( in a different file will still
be flagged angular_artifact_without_test_markers here. Cross-reference this
metric's file values against your test file naming/co-location convention
(e.g. foo.js next to foo.spec.js) to filter out true negatives during
assessment.
Each AngularJS method-chain call site (e.g. every .module('app') in
angular.module('app').controller(...)) is counted independently, so a
single logical module can produce multiple module entries — this mirrors
how many risky call sites exist in the source, not how many distinct modules
exist.
Installation
Development
License
MIT