hyperguild/ingestion/internal/legacymove/legacymove_test.go

package legacymove_test

import (
	"os"
	"path/filepath"
	"strings"
	"testing"
	"time"

	"github.com/stretchr/testify/assert"
	"github.com/stretchr/testify/require"

	"github.com/mathiasbq/hyperguild/ingestion/internal/legacymove"
)

func page(t *testing.T, dir, rel, body string) {
	t.Helper()
	full := filepath.Join(dir, rel)
	require.NoError(t, os.MkdirAll(filepath.Dir(full), 0o755))
	require.NoError(t, os.WriteFile(full, []byte(body), 0o644))
}

func src(typ string) string {
	return "---\ntitle: T\ntype: " + typ + "\n---\n\n## Summary\n\nBody with [[a-link|A]] kept as is.\n"
}

func byFrom(items []legacymove.Item) map[string]legacymove.Item {
	m := map[string]legacymove.Item{}
	for _, it := range items {
		m[it.From] = it
	}
	return m
}

func TestPlan_VideosAndArticlesMoveToTheDefaultWing(t *testing.T) {
	dir := t.TempDir()
	page(t, dir, "wiki/sources/a-talk.md", src("video"))
	page(t, dir, "wiki/sources/an-article.md", src("article"))

	items, err := legacymove.Plan(dir, "web-clip")
	require.NoError(t, err)
	m := byFrom(items)
	assert.Equal(t, legacymove.Move, m["wiki/sources/a-talk.md"].Action)
	assert.Equal(t, "wiki/web-clip/sources/a-talk.md", m["wiki/sources/a-talk.md"].To)
	assert.Equal(t, legacymove.Move, m["wiki/sources/an-article.md"].Action)
}

func TestPlan_NotesProjectsAndUntypedPagesStayWhereTheyAre(t *testing.T) {
	dir := t.TempDir()
	page(t, dir, "wiki/sources/a-note.md", src("note"))
	page(t, dir, "wiki/sources/a-project.md", src("project"))
	page(t, dir, "wiki/sources/untyped.md", "---\ntitle: T\n---\nbody")
	page(t, dir, "wiki/sources/_index.md", "index")

	items, err := legacymove.Plan(dir, "web-clip")
	require.NoError(t, err)
	m := byFrom(items)
	for _, f := range []string{"a-note", "a-project", "untyped"} {
		assert.Equal(t, legacymove.Skip, m["wiki/sources/"+f+".md"].Action, f)
	}
	_, listed := m["wiki/sources/_index.md"]
	assert.False(t, listed, "the index is not a page")
}

func TestPlan_AutoTunnelLogsAreDeleted(t *testing.T) {
	dir := t.TempDir()
	page(t, dir, "wiki/sources/tunnel-candidates-2026-06-14.md", src("video"))

	items, err := legacymove.Plan(dir, "web-clip")
	require.NoError(t, err)
	assert.Equal(t, legacymove.Delete, byFrom(items)["wiki/sources/tunnel-candidates-2026-06-14.md"].Action,
		"never moved, whatever the type says")
}

func TestPlan_ASlugAlreadyUsedElsewhereIsSkippedNotOverwritten(t *testing.T) {
	dir := t.TempDir()
	page(t, dir, "wiki/sources/a-talk.md", src("video"))
	page(t, dir, "wiki/entities/a-talk.md", "an entity with the same slug")

	items, err := legacymove.Plan(dir, "web-clip")
	require.NoError(t, err)
	it := byFrom(items)["wiki/sources/a-talk.md"]
	assert.Equal(t, legacymove.Skip, it.Action)
	assert.Contains(t, it.Reason, "already used")
}

func TestPlan_ADefaultWingThatIsNotAPlainNameIsRefused(t *testing.T) {
	_, err := legacymove.Plan(t.TempDir(), "../x")
	assert.Error(t, err)
}

func TestApply_MovesStampsAndArchivesEverythingItTouches(t *testing.T) {
	dir := t.TempDir()
	talk := src("video")
	junk := "tunnel log"
	page(t, dir, "wiki/sources/a-talk.md", talk)
	page(t, dir, "wiki/sources/tunnel-candidates-2026-06-14.md", junk)
	page(t, dir, "wiki/sources/a-note.md", src("note"))
	items, err := legacymove.Plan(dir, "web-clip")
	require.NoError(t, err)

	res, err := legacymove.Apply(dir, items, time.Date(2026, 10, 20, 0, 0, 0, 0, time.UTC))
	require.NoError(t, err)
	assert.Equal(t, 1, res.Moved)
	assert.Equal(t, 1, res.Deleted)

	moved, err := os.ReadFile(filepath.Join(dir, "wiki/web-clip/sources/a-talk.md"))
	require.NoError(t, err)
	assert.Contains(t, string(moved), "wing: web-clip")
	assert.Contains(t, string(moved), "hall: sources")
	assert.Contains(t, string(moved), "Body with [[a-link|A]] kept as is.", "body and links are byte-preserved")

	for _, gone := range []string{"wiki/sources/a-talk.md", "wiki/sources/tunnel-candidates-2026-06-14.md"} {
		_, statErr := os.Stat(filepath.Join(dir, gone))
		assert.ErrorIs(t, statErr, os.ErrNotExist, gone)
	}
	_, statErr := os.Stat(filepath.Join(dir, "wiki/sources/a-note.md"))
	assert.NoError(t, statErr, "a skipped page is left exactly where it is")

	again, err := legacymove.Plan(dir, "web-clip")
	require.NoError(t, err)
	for _, it := range again {
		assert.Equal(t, legacymove.Skip, it.Action, "nothing is left to do: %s", it.From)
	}
}

func TestRollback_RestoresEveryTouchedPageByteForByte(t *testing.T) {
	dir := t.TempDir()
	talk := src("video")
	junk := "tunnel log"
	page(t, dir, "wiki/sources/a-talk.md", talk)
	page(t, dir, "wiki/sources/tunnel-candidates-2026-06-14.md", junk)
	items, err := legacymove.Plan(dir, "web-clip")
	require.NoError(t, err)
	res, err := legacymove.Apply(dir, items, time.Date(2026, 10, 20, 0, 0, 0, 0, time.UTC))
	require.NoError(t, err)

	require.NoError(t, legacymove.Rollback(dir, res.ArchiveDir))

	got, err := os.ReadFile(filepath.Join(dir, "wiki/sources/a-talk.md"))
	require.NoError(t, err)
	assert.Equal(t, talk, string(got))
	got, err = os.ReadFile(filepath.Join(dir, "wiki/sources/tunnel-candidates-2026-06-14.md"))
	require.NoError(t, err)
	assert.Equal(t, junk, string(got))
	_, statErr := os.Stat(filepath.Join(dir, "wiki/web-clip/sources/a-talk.md"))
	assert.ErrorIs(t, statErr, os.ErrNotExist, "the moved copy is removed")
}

func TestApply_NeverOverwritesAndKeepsTheSourceIfTheDestinationAppearsLate(t *testing.T) {
	dir := t.TempDir()
	page(t, dir, "wiki/sources/a-talk.md", src("video"))
	items, err := legacymove.Plan(dir, "web-clip")
	require.NoError(t, err)
	page(t, dir, "wiki/web-clip/sources/a-talk.md", "appeared after the plan")

	_, err = legacymove.Apply(dir, items, time.Date(2026, 10, 20, 0, 0, 0, 0, time.UTC))
	require.Error(t, err)
	got, _ := os.ReadFile(filepath.Join(dir, "wiki/web-clip/sources/a-talk.md"))
	assert.Equal(t, "appeared after the plan", string(got))
	_, statErr := os.Stat(filepath.Join(dir, "wiki/sources/a-talk.md"))
	assert.NoError(t, statErr, "the source survives")
}

func TestApply_TheArchiveIsOutsideRawSoTheWatcherNeverIngestsIt(t *testing.T) {
	dir := t.TempDir()
	page(t, dir, "wiki/sources/tunnel-candidates-2026-06-14.md", "x")
	items, _ := legacymove.Plan(dir, "web-clip")
	res, err := legacymove.Apply(dir, items, time.Date(2026, 10, 20, 0, 0, 0, 0, time.UTC))
	require.NoError(t, err)
	assert.False(t, strings.Contains(res.ArchiveDir, "/raw"), res.ArchiveDir)
	assert.True(t, strings.HasPrefix(res.ArchiveDir, "archive/"), res.ArchiveDir)
}

// --- raw/ leftovers from before inbox zero ---

func rawFile(t *testing.T, dir, rel, body string) { page(t, dir, rel, body) }

func TestPlanRaw_AnArchivedOriginalAndItsMarkerAreDeleted(t *testing.T) {
	dir := t.TempDir()
	rawFile(t, dir, "raw/clip.md", "the clip text")
	rawFile(t, dir, "raw/clip.md.processed", "2026-10-19")
	rawFile(t, dir, "raw/processed/2026-10-19/clip.md", "the clip text")

	items, err := legacymove.PlanRaw(dir)
	require.NoError(t, err)
	m := byFrom(items)
	assert.Equal(t, legacymove.Delete, m["raw/clip.md"].Action)
	assert.Equal(t, legacymove.Delete, m["raw/clip.md.processed"].Action)
}

func TestPlanRaw_AnOriginalWhoseArchiveCopyDiffersIsLeftAlone(t *testing.T) {
	dir := t.TempDir()
	rawFile(t, dir, "raw/clip.md", "the clip text, edited since")
	rawFile(t, dir, "raw/clip.md.processed", "2026-10-19")
	rawFile(t, dir, "raw/processed/2026-10-19/clip.md", "the clip text")

	items, err := legacymove.PlanRaw(dir)
	require.NoError(t, err)
	m := byFrom(items)
	assert.Equal(t, legacymove.Skip, m["raw/clip.md"].Action, "no identical archive copy: not ours to delete")
	assert.Contains(t, m["raw/clip.md"].Reason, "no identical copy")
	_, hasMarker := m["raw/clip.md.processed"]
	assert.False(t, hasMarker, "the marker of a note we keep is kept")
}

func TestPlanRaw_ACopyUnderATimestampedNameCounts(t *testing.T) {
	dir := t.TempDir()
	rawFile(t, dir, "raw/clip.md", "same text")
	rawFile(t, dir, "raw/clip.md.processed", "d")
	rawFile(t, dir, "raw/processed/2026-10-19/clip-180200.md", "same text")
	rawFile(t, dir, "raw/processed/2026-10-19/clip-2.md", "other text")

	items, err := legacymove.PlanRaw(dir)
	require.NoError(t, err)
	assert.Equal(t, legacymove.Delete, byFrom(items)["raw/clip.md"].Action)
}

func TestPlanRaw_AnOrphanMarkerIsDeleted(t *testing.T) {
	dir := t.TempDir()
	rawFile(t, dir, "raw/gone.md.processed", "2026-06-16")

	items, err := legacymove.PlanRaw(dir)
	require.NoError(t, err)
	assert.Equal(t, legacymove.Delete, byFrom(items)["raw/gone.md.processed"].Action)
}

func TestPlanRaw_TheAutoTunnelReviewQueueAndANoteWithoutAMarkerAreNotTouched(t *testing.T) {
	dir := t.TempDir()
	rawFile(t, dir, "raw/tunnel-candidates-2026-06-14.md", "AutoTunnel's own queue")
	rawFile(t, dir, "raw/tunnel-candidates-2026-06-14.md.processed", "x")
	rawFile(t, dir, "raw/fresh.md", "an unprocessed clip")
	rawFile(t, dir, "raw/processed/2026-10-19/fresh.md", "an unprocessed clip") // same bytes, but no marker: still in the inbox

	items, err := legacymove.PlanRaw(dir)
	require.NoError(t, err)
	m := byFrom(items)
	for _, f := range []string{"raw/tunnel-candidates-2026-06-14.md", "raw/tunnel-candidates-2026-06-14.md.processed", "raw/fresh.md"} {
		_, listed := m[f]
		assert.False(t, listed, "%s is not a leftover", f)
	}
}

func TestApplyAndRollback_HandleRawLeftoversAndNeverCollideWithSourcesOfTheSameName(t *testing.T) {
	dir := t.TempDir()
	rawFile(t, dir, "raw/same.md", "raw text")
	rawFile(t, dir, "raw/same.md.processed", "d")
	rawFile(t, dir, "raw/processed/2026-10-19/same.md", "raw text")
	page(t, dir, "wiki/sources/same.md", src("video")) // a wiki source with the same basename
	rawItems, err := legacymove.PlanRaw(dir)
	require.NoError(t, err)
	srcItems, err := legacymove.Plan(dir, "web-clip")
	require.NoError(t, err)

	res, err := legacymove.Apply(dir, append(rawItems, srcItems...), time.Date(2026, 10, 20, 0, 0, 0, 0, time.UTC))
	require.NoError(t, err)
	for _, gone := range []string{"raw/same.md", "raw/same.md.processed", "wiki/sources/same.md"} {
		_, statErr := os.Stat(filepath.Join(dir, gone))
		assert.ErrorIs(t, statErr, os.ErrNotExist, gone)
	}
	_, statErr := os.Stat(filepath.Join(dir, "raw/processed/2026-10-19/same.md"))
	assert.NoError(t, statErr, "the archive copy in raw/processed is never touched")

	require.NoError(t, legacymove.Rollback(dir, res.ArchiveDir))
	got, err := os.ReadFile(filepath.Join(dir, "raw/same.md"))
	require.NoError(t, err)
	assert.Equal(t, "raw text", string(got), "the raw note and the wiki source of one name did not overwrite each other in the archive")
	got, err = os.ReadFile(filepath.Join(dir, "wiki/sources/same.md"))
	require.NoError(t, err)
	assert.Equal(t, src("video"), string(got))
}

// --- review of #213 ---

var day = time.Date(2026, 10, 20, 0, 0, 0, 0, time.UTC)

// F1: a second Apply on the same day must not touch the first run's archive.
func TestApply_ASecondRunTheSameDayGetsItsOwnArchiveAndKeepsTheFirst(t *testing.T) {
	dir := t.TempDir()
	page(t, dir, "wiki/sources/a-talk.md", src("video"))
	page(t, dir, "wiki/sources/tunnel-candidates-1.md", "FIRST tunnel bytes")
	items, err := legacymove.Plan(dir, "web-clip")
	require.NoError(t, err)
	first, err := legacymove.Apply(dir, items, day)
	require.NoError(t, err)

	page(t, dir, "wiki/sources/tunnel-candidates-1.md", "SECOND tunnel bytes") // same name reappears
	items2, err := legacymove.Plan(dir, "web-clip")
	require.NoError(t, err)
	second, err := legacymove.Apply(dir, items2, day)
	require.NoError(t, err)

	assert.NotEqual(t, first.ArchiveDir, second.ArchiveDir)
	got, err := os.ReadFile(filepath.Join(dir, first.ArchiveDir, "originals", "wiki__sources__tunnel-candidates-1.md"))
	require.NoError(t, err)
	assert.Equal(t, "FIRST tunnel bytes", string(got), "the first run's archived original is intact")

	require.NoError(t, legacymove.Rollback(dir, first.ArchiveDir))
	_, statErr := os.Stat(filepath.Join(dir, "wiki/sources/a-talk.md"))
	assert.NoError(t, statErr, "the first run's move is still rolled back by its own manifest")
}

// F2: Plan decided on the bytes it saw. Apply must not delete different bytes.
func TestApply_ARawNoteThatChangedSincePlanningIsNotDeleted(t *testing.T) {
	dir := t.TempDir()
	rawFile(t, dir, "raw/a.md", "bytes that have an archive copy")
	rawFile(t, dir, "raw/a.md.processed", "d")
	rawFile(t, dir, "raw/processed/2026-10-19/a.md", "bytes that have an archive copy")
	items, err := legacymove.PlanRaw(dir)
	require.NoError(t, err)

	rawFile(t, dir, "raw/a.md", "NEW UNPROCESSED BYTES") // dropped again before Apply runs
	res, err := legacymove.Apply(dir, items, day)
	require.NoError(t, err)

	got, err := os.ReadFile(filepath.Join(dir, "raw/a.md"))
	require.NoError(t, err, "the changed note is still in the inbox")
	assert.Equal(t, "NEW UNPROCESSED BYTES", string(got))
	assert.Contains(t, strings.Join(res.Changed, ","), "raw/a.md")
}

func TestApply_AWikiSourceThatChangedSincePlanningIsNotMoved(t *testing.T) {
	dir := t.TempDir()
	page(t, dir, "wiki/sources/a-talk.md", src("video"))
	items, err := legacymove.Plan(dir, "web-clip")
	require.NoError(t, err)
	page(t, dir, "wiki/sources/a-talk.md", src("video")+"\nan edit made after the plan\n")

	res, err := legacymove.Apply(dir, items, day)
	require.NoError(t, err)
	assert.Zero(t, res.Moved)
	got, _ := os.ReadFile(filepath.Join(dir, "wiki/sources/a-talk.md"))
	assert.Contains(t, string(got), "an edit made after the plan")
	assert.NotEmpty(t, res.Changed)
}

// F3: rollback checks everything first, changes nothing on a conflict, can be
// repeated, and never throws away an edit.
func TestRollback_RefusesBeforeChangingAnythingWhenATargetExists(t *testing.T) {
	dir := t.TempDir()
	page(t, dir, "wiki/sources/a-talk.md", src("video"))
	page(t, dir, "wiki/sources/b-talk.md", src("video"))
	items, _ := legacymove.Plan(dir, "web-clip")
	res, err := legacymove.Apply(dir, items, day)
	require.NoError(t, err)

	page(t, dir, "wiki/sources/b-talk.md", "someone put a different b here")
	err = legacymove.Rollback(dir, res.ArchiveDir)
	require.Error(t, err)
	_, statErr := os.Stat(filepath.Join(dir, "wiki/sources/a-talk.md"))
	assert.ErrorIs(t, statErr, os.ErrNotExist, "nothing was restored: a was not touched either")
	_, statErr = os.Stat(filepath.Join(dir, "wiki/web-clip/sources/a-talk.md"))
	assert.NoError(t, statErr, "and the moved copies are still there")
}

func TestRollback_CanBeRunTwice(t *testing.T) {
	dir := t.TempDir()
	page(t, dir, "wiki/sources/a-talk.md", src("video"))
	items, _ := legacymove.Plan(dir, "web-clip")
	res, err := legacymove.Apply(dir, items, day)
	require.NoError(t, err)
	require.NoError(t, legacymove.Rollback(dir, res.ArchiveDir))
	require.NoError(t, legacymove.Rollback(dir, res.ArchiveDir), "a repeat finds everything already restored")
	got, _ := os.ReadFile(filepath.Join(dir, "wiki/sources/a-talk.md"))
	assert.Equal(t, src("video"), string(got))
}

func TestRollback_DoesNotDeleteAMovedPageThatWasEditedSinceTheMove(t *testing.T) {
	dir := t.TempDir()
	page(t, dir, "wiki/sources/a-talk.md", src("video"))
	items, _ := legacymove.Plan(dir, "web-clip")
	res, err := legacymove.Apply(dir, items, day)
	require.NoError(t, err)
	moved := filepath.Join(dir, "wiki/web-clip/sources/a-talk.md")
	b, _ := os.ReadFile(moved)
	require.NoError(t, os.WriteFile(moved, append(b, []byte("\nmy edit\n")...), 0o644))

	err = legacymove.Rollback(dir, res.ArchiveDir)
	require.Error(t, err)
	got, readErr := os.ReadFile(moved)
	require.NoError(t, readErr)
	assert.Contains(t, string(got), "my edit", "the edit survives")
	_, statErr := os.Stat(filepath.Join(dir, "wiki/sources/a-talk.md"))
	assert.ErrorIs(t, statErr, os.ErrNotExist, "nothing was changed")
}

// F4: a name whose archive key cannot fit must be skipped at planning time.
func TestPlanRaw_ANameTooLongToArchiveIsSkippedNotAbortedOn(t *testing.T) {
	dir := t.TempDir()
	long := strings.Repeat("a", 240) + ".md"
	rawFile(t, dir, "raw/"+long, "text")
	rawFile(t, dir, "raw/"+long+".processed", "d")
	rawFile(t, dir, "raw/processed/2026-10-19/"+long, "text")
	rawFile(t, dir, "raw/ok.md", "fine")
	rawFile(t, dir, "raw/ok.md.processed", "d")
	rawFile(t, dir, "raw/processed/2026-10-19/ok.md", "fine")

	items, err := legacymove.PlanRaw(dir)
	require.NoError(t, err)
	m := byFrom(items)
	assert.Equal(t, legacymove.Skip, m["raw/"+long].Action)
	assert.Contains(t, m["raw/"+long].Reason, "too long")
	assert.Equal(t, legacymove.Delete, m["raw/ok.md"].Action)

	_, err = legacymove.Apply(dir, items, day)
	require.NoError(t, err, "the long name does not wedge the run")
	_, statErr := os.Stat(filepath.Join(dir, "raw/"+long))
	assert.NoError(t, statErr, "and it is left where it is")
}

// F6: do not stamp a page that already carries a wing or hall; only a top-level
// type inside a closed frontmatter block decides.
func TestPlan_AMovedPageIsNeverGivenDuplicateWingKeys(t *testing.T) {
	dir := t.TempDir()
	page(t, dir, "wiki/sources/has-wing.md", "---\ntitle: T\ntype: video\nwing: old\nhall: facts\n---\nbody")
	items, err := legacymove.Plan(dir, "web-clip")
	require.NoError(t, err)
	it := byFrom(items)["wiki/sources/has-wing.md"]
	assert.Equal(t, legacymove.Skip, it.Action)
	assert.Contains(t, it.Reason, "already has")
}

func TestPlan_OnlyATopLevelTypeInAClosedBlockClassifiesAPage(t *testing.T) {
	dir := t.TempDir()
	page(t, dir, "wiki/sources/nested.md", "---\ntitle: T\nmeta:\n  type: video\n---\nbody")
	page(t, dir, "wiki/sources/blockscalar.md", "---\ntitle: T\ndesc: |\n  type: article\n---\nbody")
	page(t, dir, "wiki/sources/unterminated.md", "---\ntitle: T\ntype: video\nbody with no closing fence")
	items, err := legacymove.Plan(dir, "web-clip")
	require.NoError(t, err)
	m := byFrom(items)
	for _, f := range []string{"nested", "blockscalar", "unterminated"} {
		assert.Equal(t, legacymove.Skip, m["wiki/sources/"+f+".md"].Action, f)
	}
}

func TestPlan_AWikiSourceWithATooLongNameIsSkipped(t *testing.T) {
	dir := t.TempDir()
	long := strings.Repeat("a", 240) + ".md"
	page(t, dir, "wiki/sources/"+long, src("video"))
	items, err := legacymove.Plan(dir, "web-clip")
	require.NoError(t, err)
	it := byFrom(items)["wiki/sources/"+long]
	assert.Equal(t, legacymove.Skip, it.Action)
	assert.Contains(t, it.Reason, "too long")
}