Compare commits

...
Author SHA1 Message Date
JMR-devandClaude Opus 5 4e88de3045 Stop a permission answer starting a second conversion (#202)
currentInput() answered for Converting, Waiting and Converted as well as Ready. Those three
arms were unreachable by tapping Convert -- the button renders only in the Ready branch --
but they were reachable through the POST_NOTIFICATIONS *result*, which ConverterScreen.kt:91
wires to convert() rather than to the button.

Reaching one enqueued a SECOND job over a live one: activeWorkId was overwritten, and the
first job kept running with its foreground notification orphaned and nothing left holding
its id to cancel it.

#202 decided to narrow rather than to test it as it stood, because a test written against
the old shape would have frozen the double-enqueue as intended behaviour -- the F1/F5 failure
mode docs/coverage-read-findings.md names. currentInput() is now
(_state.value as? ConversionState.Ready)?.input, which is what JoinViewModel.join() has been
all along; the two screens are the same shape and only one of them was over-general.

Four cold refusal arms come with it, all reached the same way -- a system callback arriving
after the screen has moved on, which is what a result redelivered after process death does:

  ConversionViewModel.kt:513   currentInput() ?: return
  ConversionViewModel.kt:600   pendingSave() ?: return
  JoinViewModel.kt:316         (as? Ready)?.inputs ?: return
  JoinViewModel.kt:390         pendingSave() ?: return

One fixture note worth keeping: the second case needs a real staged file in the worker's
output Data. A SUCCEEDED job with no output path maps to Failed rather than Converted, so
Data.EMPTY never reaches the state the case is about -- which cost a timed-out awaitState
before it was spotted.

Mutations, all run and restored:

  restore the over-general four-arm when   1 red  <- the defect this change fixes
  currentInput()!! at :513                 2 red
  pendingSave()!! in save()                1 red
  drop both join guards                    1 red

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-09-05 19:36:38 -05:00
JMR-devandClaude Opus 5 2db0dc65d3 Open the save dialog with the type the job produced (#201)
ConverterScreen.kt:80 -- state.pendingSave()?.mimeType ?: settings.spec.mimeType -- had
never taken its left-hand side. Its comment records what the line is for: a retry after a
failed save must open with the type its FIRST attempt used, because the fallback beside it
is the current picker, which a reattached job never set. So the untested half is the fix and
the tested half is the fallback it was added to stop being used.

This withdraws a named exemption rather than working around it. FailedSaveRetryTest's KDoc
listed the line as not asserted because "it lives in the entry point, above the ScreenContent
seam, and reaching it needs a real ViewModel inside a composition". True when written;
AdaptiveShellTest (#173) then established composing the real screens with real ViewModels,
and #200 added the ShadowActivity mechanics for reading what a launcher launched. The reason
the exemption gave no longer holds, so it is withdrawn in the same change rather than left
to be taken at face value -- the shape of #141 revising #84's boundary.

The job is reattached rather than run because the screen composes its own ViewModel through
viewModel() and nothing can be injected into it. That is also the case the line exists for:
a reattached job's spec was never in these settings at all.

The test asserts the two mime types differ as well as which one is used. Without that, the
assertion would pass just as well against the fallback if the fixture ever drifted onto MP4.

Mutation: collapse :80 to settings.spec.mimeType -- red. Run and restored.

Unrelated, and recorded because it turned up here: #159 now reproduces on this host. The
full suite failed once in six runs on OutputPublisherStagingTest:184, and the isolating
experiment says it is not this change -- three runs WITH the new test all passed, and the
failure occurred on a run with the file removed.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-09-05 19:36:37 -05:00
4 changed files with 307 additions and 11 deletions
@@ -673,13 +673,23 @@ class ConversionViewModel @JvmOverloads constructor(
else -> null
}
private fun currentInput(): InputFile? = when (val s = _state.value) {
is ConversionState.Ready -> s.input
is ConversionState.Converting -> s.input
is ConversionState.Waiting -> s.input
is ConversionState.Converted -> s.input
else -> null
}
/**
* The input `convert()` may act on, which is only ever the one on a `Ready` screen.
*
* This used to answer for `Converting`, `Waiting` and `Converted` as well. Those arms were not
* reachable by tapping Convert -- the button renders only in the `Ready` branch -- but they
* were reachable through the POST_NOTIFICATIONS **result**, which `ConverterScreen.kt:91` wires
* to `convert()` rather than to the button. Reaching one of them enqueued a *second* job over a
* live one: `activeWorkId` was overwritten, and the first job kept running with its foreground
* notification orphaned and nothing left holding its id to cancel it.
*
* Narrowed under #202 rather than tested as it stood, because a test written against the old
* shape would have frozen the double-enqueue as intended behaviour -- the F1/F5 failure mode.
*
* `JoinViewModel.join()` has been `(_state.value as? JoinState.Ready)?.inputs ?: return` all
* along. The two screens are the same shape and only one of them was over-general.
*/
private fun currentInput(): InputFile? = (_state.value as? ConversionState.Ready)?.input
private companion object {
/**
@@ -51,10 +51,13 @@ import java.io.File
* here needs. `OutputPublisherPublishTest` owns what a real publish writes.
* - **The screen's two buttons.** `ConverterStateAffordancesTest` and `JoinStateAffordancesTest`
* own what each state renders; this file owns what each state carries.
* - **`ConverterScreen`'s `destinationMime` line itself.** It lives in the entry point, above the
* `ScreenContent` seam, and reaching it needs a real ViewModel inside a composition. What it
* reads -- `pendingSave()?.mimeType` -- is asserted directly instead, which is why that
* derivation was moved out of the entry point in the first place.
* - ~~**`ConverterScreen`'s `destinationMime` line itself.**~~ **Withdrawn 2026-09-02 (#201).** The
* exemption read: "it lives in the entry point, above the `ScreenContent` seam, and reaching it
* needs a real ViewModel inside a composition". That was true when written and is no longer:
* `AdaptiveShellTest` (#173) established composing the real screens with real ViewModels, and
* #200 added the `ShadowActivity` mechanics for reading what a launcher launched. `RetrySaveMimeTest`
* now asserts the line directly. What this file still owns is the half below the seam -- what each
* state *carries* -- which is why `pendingSave()?.mimeType` is also asserted here.
* - **Picking a new input while a `Failed` carries a file.** `onInputPicked` overwrites the state
* without discarding, from `Converted` exactly as much as from a carrying `Failed`, and neither
* branch renders a picker. It is a pre-existing path this change neither opens nor widens: the
@@ -0,0 +1,136 @@
package org.libremediaconverter.convert
import android.app.Application
import android.content.Intent
import android.net.Uri
import androidx.activity.ComponentActivity
import androidx.compose.ui.test.junit4.v2.createAndroidComposeRule
import androidx.compose.ui.test.onNodeWithTag
import androidx.compose.ui.test.performClick
import androidx.compose.ui.test.performScrollTo
import androidx.media3.common.util.UnstableApi
import androidx.work.WorkManager
import androidx.work.workDataOf
import org.junit.After
import org.junit.Assert.assertEquals
import org.junit.Assert.assertNotEquals
import org.junit.Before
import org.junit.Rule
import org.junit.Test
import org.junit.runner.RunWith
import org.libremediaconverter.model.InputProbe
import org.libremediaconverter.model.OutputFormat
import org.libremediaconverter.ui.TestTags
import org.libremediaconverter.work.ConversionWorker
import org.robolectric.RobolectricTestRunner
import org.robolectric.RuntimeEnvironment
import org.robolectric.Shadows.shadowOf
import java.io.File
/**
* The save dialog opens with the type the *job* produced, not the type the picker is showing now.
*
* `ConverterScreen.kt:80` — `state.pendingSave()?.mimeType ?: settings.spec.mimeType` — had never
* taken its left-hand side. Its comment records what the line is for:
*
* > a retry offered after a failed save opens the dialog with the type its first attempt used —
* > the cast answered null for a `Failed`, and the fallback below is the current picker, which a
* > reattached job never set.
*
* So the untested half is the fix, and the tested half is the fallback it was added to stop being
* used.
*
* ## This revises a named exemption, deliberately
*
* `FailedSaveRetryTest`'s KDoc lists this line under "Not asserted here, so each is a decision
* rather than an omission":
*
* > It lives in the entry point, above the `ScreenContent` seam, and reaching it needs a real
* > ViewModel inside a composition.
*
* That was true when written. `AdaptiveShellTest` (#173) then established exactly that capability,
* and #200 added the two `ShadowActivity` mechanics that let a test read what a launcher launched.
* The reason the exemption gave no longer holds, so the exemption is withdrawn rather than left to
* be taken at face value — the same shape as #141 revising #84's boundary. That KDoc is corrected
* in this change.
*
* ## Why the job is reattached rather than run
*
* The screen composes its own ViewModel through `viewModel()`, so nothing can be injected into it.
* A job finished before the composition is the one route to a `Converted` state carrying output
* `Data` this test chose — and it is also the case the line exists for, since a reattached job's
* spec "was never in these settings at all".
*/
@UnstableApi
@RunWith(RobolectricTestRunner::class)
class RetrySaveMimeTest {
@get:Rule
val composeRule = createAndroidComposeRule<ComponentActivity>()
private lateinit var app: Application
private lateinit var staged: File
@Before
fun setUp() {
app = RuntimeEnvironment.getApplication()
ConversionDependencies.probe = { _, _ -> InputProbe() }
staged = OutputPublisher(app).createStagingFile("holiday.mkv").apply { writeBytes(ByteArray(4096)) }
}
@After
fun tearDown() = ConversionDependencies.reset()
@Test
fun `the save dialog offers the type the job produced, not the one the picker is showing`() {
finishAJobProducing(JOB_MIME_TYPE)
composeRule.setContent { ConverterScreen() }
composeRule.waitForIdle()
composeRule.onNodeWithTag(TestTags.SAVE_FILE).performScrollTo().performClick()
composeRule.waitForIdle()
val intent = requireNotNull(shadowOf(composeRule.activity).nextStartedActivityForResult) {
"the save dialog was never launched"
}.intent
assertEquals(Intent.ACTION_CREATE_DOCUMENT, intent.action)
assertEquals(JOB_MIME_TYPE, intent.type)
// The fixture is only meaningful while the two differ; without this the assertion above
// would pass just as well against the fallback.
assertNotEquals(
"the picker's own type must differ, or this test proves nothing",
JOB_MIME_TYPE,
OutputFormat.MP4_H265.spec.mimeType,
)
}
/**
* A conversion that finished while nothing was watching, which is what `reattach()` picks up.
*
* `SucceedingWorkerFactory` reports this output `Data` for whatever is enqueued, so the job
* lands `SUCCEEDED` carrying a staged path that exists — the two things `Reattachment.choose`
* requires of a finished job.
*/
private fun finishAJobProducing(mimeType: String) {
installTestWorkManager(
app,
workDataOf(
ConversionWorker.KEY_OUTPUT_PATH to staged.absolutePath,
ConversionWorker.KEY_SUGGESTED_NAME to "holiday.mkv",
ConversionWorker.KEY_MIME_TYPE to mimeType,
),
)
WorkManager.getInstance(app).enqueue(
ConversionWorker.request(
inputUri = Uri.parse("content://test/holiday.mkv"),
displayName = "holiday.mkv",
sizeBytes = 4_096L,
),
).result.get()
}
private companion object {
/** Matroska, against the MP4 the picker defaults to. */
const val JOB_MIME_TYPE = "video/x-matroska"
}
}
@@ -0,0 +1,147 @@
package org.libremediaconverter.convert
import android.app.Application
import android.net.Uri
import androidx.media3.common.util.UnstableApi
import androidx.work.WorkManager
import androidx.work.workDataOf
import kotlinx.coroutines.Dispatchers
import org.junit.After
import org.junit.Assert.assertEquals
import org.junit.Before
import org.junit.Test
import org.junit.runner.RunWith
import org.libremediaconverter.join.JoinState
import org.libremediaconverter.join.JoinViewModel
import org.libremediaconverter.model.InputProbe
import org.libremediaconverter.work.ConcatWorker
import org.libremediaconverter.work.ConversionWorker
import org.robolectric.RobolectricTestRunner
import org.robolectric.RuntimeEnvironment
/**
* An answer that arrives after the screen has moved on does nothing.
*
* Four refusal arms, cold before this file:
*
* ```
* convert/ConversionViewModel.kt:513 currentInput() ?: return
* convert/ConversionViewModel.kt:600 pendingSave() ?: return
* join/JoinViewModel.kt:316 (as? Ready)?.inputs ?: return
* join/JoinViewModel.kt:390 pendingSave() ?: return
* ```
*
* They are not merely defensive. `ConverterScreen.kt:91` wires `convert()` to the
* **POST_NOTIFICATIONS result**, and `:83` wires `save()` to the CreateDocument result — so both
* are entered by a system callback rather than by a tap, and a result redelivered after process
* death arrives at a brand-new ViewModel sitting on `Idle`.
*
* ## The production change that came with this
*
* `currentInput()` used to answer for `Converting`, `Waiting` and `Converted` as well as `Ready`.
* Those arms were unreachable by tapping Convert but reachable through that permission callback,
* and reaching one enqueued a **second** job over a live one — `activeWorkId` overwritten, the
* first job still running with an orphaned notification and nothing holding its id.
*
* #202 decided to narrow rather than to test it as it stood, because a test written against the old
* shape would have frozen the double-enqueue as intended behaviour. `JoinViewModel.join()` has been
* `(_state.value as? JoinState.Ready)?.inputs ?: return` all along; the two screens are the same
* shape and only one was over-general.
*/
@UnstableApi
@RunWith(RobolectricTestRunner::class)
class StaleLauncherResultTest {
private lateinit var app: Application
private lateinit var workManager: WorkManager
private lateinit var staged: java.io.File
@Before
fun setUp() {
app = RuntimeEnvironment.getApplication()
val publisher = RecordingPublisher(app)
ConversionDependencies.publisher = { publisher }
ConversionDependencies.probe = { _, _ -> InputProbe() }
// A real staged file, because a SUCCEEDED job with no output path maps to Failed rather
// than Converted -- and Converted is the state this file's second case has to reach.
staged = publisher.createStagingFile("holiday.mp4").apply { writeBytes(ByteArray(4096)) }
installTestWorkManager(
app,
workDataOf(
ConversionWorker.KEY_OUTPUT_PATH to staged.absolutePath,
ConversionWorker.KEY_SUGGESTED_NAME to "holiday.mp4",
ConversionWorker.KEY_MIME_TYPE to "video/mp4",
),
)
workManager = WorkManager.getInstance(app)
}
@After
fun tearDown() = ConversionDependencies.reset()
@Test
fun `a permission answer arriving on an empty screen enqueues nothing`() {
val viewModel = ConversionViewModel(app, Dispatchers.Unconfined)
awaitState(viewModel.state, "Idle") { it is ConversionState.Idle }
viewModel.convert()
assertEquals(ConversionState.Idle, viewModel.state.value)
assertEquals("nothing may be enqueued for a file that is not there", 0, conversionJobs())
}
/**
* The narrowing itself: a permission answer that arrives while a conversion is already running
* must not start a second one.
*
* Reached by converting once — the synchronous test WorkManager finishes it inline, so the
* screen is `Converted`, which is one of the three arms `currentInput()` used to answer for.
* Calling `convert()` again from there is precisely what the permission callback can do.
*/
@Test
fun `a permission answer arriving after the job finished does not start a second one`() {
val viewModel = ConversionViewModel(app, Dispatchers.Unconfined)
viewModel.onInputPicked(Uri.parse("content://test/holiday.mkv"))
awaitState(viewModel.state, "Ready") { it is ConversionState.Ready }
viewModel.convert()
val converted = awaitState(viewModel.state, "Converted") { it is ConversionState.Converted }
assertEquals("the fixture needs exactly one job to start with", 1, conversionJobs())
viewModel.convert()
assertEquals("a second job must not be enqueued over the first", 1, conversionJobs())
assertEquals("and the screen must not move", converted, viewModel.state.value)
}
@Test
fun `a save answer arriving on an empty screen does nothing`() {
val viewModel = ConversionViewModel(app, Dispatchers.Unconfined)
awaitState(viewModel.state, "Idle") { it is ConversionState.Idle }
viewModel.save(DESTINATION)
assertEquals(ConversionState.Idle, viewModel.state.value)
}
@Test
fun `a join answer arriving on an empty screen enqueues nothing`() {
val viewModel = JoinViewModel(app, Dispatchers.Unconfined)
awaitState(viewModel.state, "Idle") { it is JoinState.Idle }
viewModel.join()
viewModel.save(DESTINATION)
assertEquals(JoinState.Idle, viewModel.state.value)
assertEquals(0, joinJobs())
}
private fun conversionJobs() = jobsTagged(ConversionWorker::class.java.name)
private fun joinJobs() = jobsTagged(ConcatWorker::class.java.name)
private fun jobsTagged(tag: String) = workManager.getWorkInfosByTag(tag).get().size
private companion object {
val DESTINATION: Uri = Uri.parse("content://test/destination.mp4")
}
}