Supported Test Result Formats¶

adacovex can parse test results in several standard formats. Parsing is additive and best-effort: every line is inspected for any recognised format, so a file mixing styles (or wrapping a third-party runner’s log) still produces correct pass/fail totals.

Every format contributes to the same three outputs:

totals – Total_Passed / Total_Failed, reported as tests: N passed, M failed and consumed by the DAL “Tests passing” criterion;
per-category metrics – Categories (name, count, status), consumed by the dashboard’s Tests chart and the --emit-metrics export;
the summary lines below.

Markdown Table (native test runner)¶

The built-in test_runner executable produces this format, and any Markdown table with the same shape is accepted:

  | Category            |  Tests | Status   |
  |---------------------|--------|----------|
  | Types conversions   |  67 | PASS     |
  | Server routing      |  25 | PASS     |
  | DAL compliance      |   2 | FAIL     |
  |-----------------------------------------|--------|----------|

  Passed: 92  Failed: 2

Format rules¶

Columns separated by |. The header row and |---| separator rows are ignored (they have no numeric count cell).
Two row layouts are accepted (the count cell is detected by its digits in either):
- with a leading index cell: | - | Category | N | PASS | (the AUnit-report style),
- plain: | Category | N | PASS | (the native test_runner style).
Category cell: arbitrary text (leading/trailing spaces are trimmed).
Tests cell: a positive integer, optionally space-padded to the column width (| 67 |). Header and separator rows have no such cell.
Status cell: PASS (anything else counts as FAIL).
Each row ADDITIONALLY adds its count to the totals.
A summary line containing Passed: and Failed: followed by numbers overrides the totals (the last such line wins). This is the footer of the native test_result.md, so files with both rows and a footer report the footer’s totals.

TAP (Test Anything Protocol)¶

The protocol used by Perl’s prove/Test::Harness toolchain (also emitted by many dynamic-language test frameworks):

1..5
ok 1 - add works
not ok 2 - remove edge case
ok 3 - merge works
ok - unnamed pass
not ok - unnamed fail

ok ... lines increment the passed counter.
not ok ... lines increment the failed counter.
The 1..N plan line and anything else are ignored.

Automake test-suite¶

The PASS:/FAIL:/SKIP: lines emitted by GNU Automake’s make check (testsuite.log / test-suite.log style):

PASS: basic
FAIL: edge case
SKIP: needs network

PASS: increments passed; FAIL: increments failed; SKIP: is ignored.

Maven Surefire¶

The summary line of Maven’s Surefire JUnit runner (“Tests run:” line from the console/log output):

Tests run: 5, Failures: 1, Errors: 0, Skipped: 0, Time elapsed: 0.1 s

Failed = Failures + Errors; passed = Tests run - failed. The last summary line wins.

Unity¶

The summary line of the Unity C test framework:

-----------------------
5 Tests 1 Failures 0 Ignored
OK

Failed = Failures; passed = Tests - Failures. The N Failures count may be separated from the number by spaces.

AUnit (Ada)¶

AdaCore’s AUnit runner output – either its tabular report (the indexed | N | Category | Count | PASS | layout with a leading index or - cell, including the optional Passed:/Failed: footer) or a Markdown table with the same columns. Both are handled by the generic Markdown-table parser above; AUnit report lines that match TAP/Automake patterns are additionally accepted.

Usage¶

The parser (Adacovex.Parsers.Tests.Parse_Test_Result) reads from a file; Parse_Test_Stdout reads from standard input. When no explicit file is given, Parse_Test_Result_From_Project auto-discovers the test summary in the target project, trying conventional file names (test_result.md, test_results.md, test-result.md, test_report.md, test_output.md, .txt/.log variants, and docs/ mirrors) in order.

# Parse a test result file
adacovex --target=../Ada_CRDT

# The parser auto-discovers test_result.md (or any conventional variant) from the target project