[edridudi]

utils and factory were the two thinnest packages in the codebase, and HexUtil had no tests at all.

Coverage

Measured on the same basis before and after (the preview suite plus the offline tests, which is what the Codecov job runs):

	Before	After
rest.koios.client.utils	57.8%	92.2%
rest.koios.client.backend.factory	56.2%	90.6%
Overall lines	76.9%	81.5%
Overall branches	63.4%	70.6%

Branch coverage gaining 7 points is the part that matters — those are untested decisions, not untested getters.

What the 29 tests actually assert

HexUtil (11) — had zero coverage. Round trips, 0x prefix handling, odd-length and non-hex rejection with the expected messages, and negative bytes not sign-extending (Byte.MIN_VALUE → 80, -1 → ff). Plus a round trip across all 256 byte values, which would catch any sign or masking regression in one assertion.

Bech32Util (7) — the existing test only proves malformed input is rejected; nothing proved valid input is accepted. These validate real mainnet stake addresses, pool ids and drep ids, and reject a deliberately corrupted checksum that still passes the charset check — the case that separates checksum verification from character validation.

Tuple (3) — values, nulls, mixed types.

BackendFactory (8) — every overload resolves to the expected base URL (https://api.koios.rest/api/v1/ and the guild/preview/preprod equivalents) rather than merely returning non-null, custom RPC honours the given URL, all ten services are wired, and the utility class cannot be instantiated.

Cost

All 29 run offline in under a second. No CI time, and nothing that can flake on a live Koios instance.

Note on the target

This does not chase 100%, deliberately. Lombok-generated code is already excluded by JaCoCo, and the remaining gap is mostly the seven System.getenv branches in the BaseService constructor — getenv is immutable in-process, so covering those needs either subprocesses per permutation or a seam added to production code purely for tests. Not worth it for the number.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 68.67%. Comparing base (5a617dc) to head (e6922b5).
⚠️ Report is 15 commits behind head on main.

Additional details and impacted files

@@             Coverage Diff              @@
##               main     #213      +/-   ##
============================================
- Coverage     69.39%   68.67%   -0.72%     
- Complexity      246      248       +2     
============================================
  Files            37       37              
  Lines           941      961      +20     
  Branches         73       79       +6     
============================================
+ Hits            653      660       +7     
- Misses          249      256       +7     
- Partials         39       45       +6     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.