[simon-mundy]

Types of changes

• Bugfix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• QA improvement (unit/integration tests, CS fixes, etc.) that does not change code behavior

Checklist:

• An issue has been created for this pull request.
• I have added tests for my changes.
• All new and existing tests passed.
• Documentation has been updated accordingly.
• This pull request targets the correct branch.

Description

PR 4 of 6 from #189: all 36 throw sites in src/TableGateway and src/RowGateway.

Stacked on #190. GitHub cannot base a cross-fork PR on a fork branch, so this targets 0.6.x and shows both commits. Only the second, 27d546dc, belongs to this PR. Merge #190 first; I will rebase after. It does not depend on #191 or #192.

Message text is byte-identical apart from the magic-access messages below, verified template by template against the original literals.

• 19 templates and factories on TableGateway\Exception\RuntimeException, 4 on its InvalidArgumentException, 5 on RowGateway\Exception\RuntimeException, 2 on its InvalidArgumentException.
• The four 'The table name of the provided X object must match that of the table' copies collapse onto one template taking the statement name.

PrimaryReplicaFeature is covered. #187 added it and left MasterSlaveFeature delegating to it, so MasterSlaveFeature no longer throws at all. Its first guard reuses MISSING_SQL_INSTANCE; its second reports the primary rather than the master, so MISSING_MASTER_SQL became MISSING_PRIMARY_SQL.

The magic-access messages now name the concrete class. AbstractTableGateway interpolated self::class, so a TableGateway instance reported its abstract parent:

Invalid magic property access in PhpDb\TableGateway\AbstractTableGateway::__get()

All three now carry static::class. This changes the message for every subclass, which is the point. The three tests asserted a substring that excluded the class name, so they passed either way; they now assert the full rendered message against $this->table::class, which fails if the code reverts to self::class.

SequenceFeature was throwing the root PhpDb\Exception\RuntimeException from inside the component, bypassing TableGateway\Exception. It now throws the component type, which extends the root one, so nothing catching the old type breaks.

Test tidy-up in the two files this PR rewrites: the test__call, test__get and test__set methods are renamed to testMagicCall, testMagicGet and testMagicSet forms, and the @codingStandardsIgnore pairs they needed are gone. The stale method-name baseline entries go with them.

Motivation and Context

Refs #189. Depends on #190 for the convention, the constant naming rule and the structural guards.

Nothing here is a BC break. The component exception classes already extended the root ones, and the retypes only narrow within that chain.

How has this been tested?

mago format --check, mago lint and mago analyze all exit 0. Full unit suite passes: 1709 tests, 3608 assertions.

Every factory is covered: 19/19, 4/4, 5/5 and 2/2. All 30 templates verified byte-identical against the original literals by a throwaway script before the call sites were touched.

Baseline and configuration changes

• The baselined mixed-operand entry on AbstractRowGateway ("Invalid middle operand: type mixed cannot be reliably used in string concatenation") is re-attributed, not added to: the same unsoundness now reports as a mixed-argument against forMissingPrimaryKeyData, count unchanged at 1.
• src/TableGateway/Exception/ is added to the too-many-methods exclude in mago.toml, as Route the Sql throw sites through named constructors #192 does for src/Sql/Exception/. The threshold is 10 and the catalogue holds 19. This will conflict with Route the Sql throw sites through named constructors #192 on the same line — whichever merges second needs the two paths merged into one array.
• One halstead reading on AbstractTableGateway is baselined. Removing long string literals reduces distinct operands, which raises halstead difficulty, so simplifying the method moved the metric the wrong way.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!