🏆Capstone project: Capstone: Prove the Complete Lending Tool

The command functions work separately. Now prove that they form one persistent workflow when every operation receives the same explicit database path. One public test can prove that story without replacing the eight focused database tests you already have.

Open tests/test_lending.py. It contains exactly one unfinished test, test_complete_cli_workflow(tmp_path, capsys). Leave tests/test_catalog.py unchanged. Those eight tests still protect schema setup, queries, foreign keys, checkout rules, commit durability, import rollback and retry, and guarded migration.

Inside the new test, create database_path and csv_path below tmp_path. Write the exact three-row data/items.csv text to the temporary CSV. Then call lending.main in this order:

init
import-items
register-member
checkout
available
activity
return
activity

Use the same arguments as Run: member 17, loan 501, item 1, checkout label 2026-08-26, and return label 2026-08-27. After each call, assert status 0, read capsys, and compare the complete stdout for that command. Require empty stderr every time. Reading the capture after each call keeps one command’s output from hiding inside another’s.

The available output must omit checked-out item 1. The first activity output must include the registered email and render the active return value as NULL. After return, the second activity output must show 2026-08-27 in the same loan row.

Finally open a fresh sqlite3 connection to database_path. Compare schema version 2, counts of three items, one member, and one loan, plus loan 501’s stored return label. Close that connection in finally. This final read proves the workflow’s state survived beyond the command connections.

Press Run. demo.py performs the same workflow in one temporary directory and then prints:

== reopen ==
Schema version: 2
Items: 3
Members: 1
Loans: 1
Loan 501 returned_on: 2026-08-27
Temporary demo database removed: yes

Here, Run demonstrates the workflow through demo.py; Submit checks your test suite. The separate Run project tests toolbar action uses Python Land’s small test runner, which does not supply pytest fixtures such as tmp_path and capsys. The next lesson, Download and Run Your Project Locally, gives the commands to run both test files with pytest on your computer.

A dependable suite can run both test files twice from separate clean directories. The one CLI test should also detect realistic regressions in statuses, headers, imported state, availability, activity email and NULL, and durable return behavior. Keep the test independent by relying only on its arguments and temporary paths.