Restore original bundle file names

Allow extra pair metadata in bundle checks
2026-05-16 16:49:18 +08:00 · 2026-05-11 16:12:26 +08:00
3 changed files with 331 additions and 14 deletions
@@ -50,6 +50,13 @@ OutputRoot/
  restore.log
 ```

+恢复时会尽量去掉去重打包时增加的名称前缀：
+
+- `data/scene_<hash>_<原始数据名>/` 只作为 bundle 内部存储目录，恢复后不会在 `master/` 或 `slave/` 下额外保留这一层目录。
+- `master/` 和 `slave/` 下直接放数据文件；如果数据文件名为 `scene_<hash>_<原始文件名>`，会恢复为 `<原始文件名>`。
+- `orbit/orbit_<hash>_<原始轨道文件名>` 会恢复为 `Task_*/orbit/<原始轨道文件名>`。
+- 如果 `pairs.json` 中提供了 `master_orbit_source_path` / `slave_orbit_source_path`，轨道文件会优先使用这些原始路径的文件名部分。
+
 Task 目录名优先使用：

 1. `task_alias`
@@ -47,6 +47,10 @@ type pair struct {
 	SlaveData         string `json:"slave_data"`
 	MasterOrbit       string `json:"master_orbit"`
 	SlaveOrbit        string `json:"slave_orbit"`
+	MasterSourcePath  string `json:"master_source_path"`
+	SlaveSourcePath   string `json:"slave_source_path"`
+	MasterOrbitSource string `json:"master_orbit_source_path"`
+	SlaveOrbitSource  string `json:"slave_orbit_source_path"`
 	MasterImagingDate string `json:"master_imaging_date"`
 	SlaveImagingDate  string `json:"slave_imaging_date"`
 	TimeBaselineDays  int    `json:"time_baseline_days"`
@@ -60,6 +64,10 @@ type pairMetadata struct {
 	SlaveData         string `json:"slave_data"`
 	MasterOrbit       string `json:"master_orbit"`
 	SlaveOrbit        string `json:"slave_orbit"`
+	MasterSourcePath  string `json:"master_source_path,omitempty"`
+	SlaveSourcePath   string `json:"slave_source_path,omitempty"`
+	MasterOrbitSource string `json:"master_orbit_source_path,omitempty"`
+	SlaveOrbitSource  string `json:"slave_orbit_source_path,omitempty"`
 	MasterImagingDate string `json:"master_imaging_date"`
 	SlaveImagingDate  string `json:"slave_imaging_date"`
 	TimeBaselineDays  int    `json:"time_baseline_days"`
@@ -426,7 +434,6 @@ func loadAndValidateInput(inputRoot string) (pairsDocument, error) {

 	var doc pairsDocument
 	decoder := json.NewDecoder(file)
-	decoder.DisallowUnknownFields()
 	if err := decoder.Decode(&doc); err != nil {
 		return pairsDocument{}, fmt.Errorf("parse pairs.json: %w", err)
 	}
@@ -523,10 +530,10 @@ func restorePair(cfg config, inputRoot string, outputRoot string, p pair, logger
 		}
 	}()

-	if err := copyDirContents(masterSource, filepath.Join(tempDir, "master")); err != nil {
+	if err := copyDataTree(masterSource, filepath.Join(tempDir, "master"), p.MasterData, p.MasterSourcePath); err != nil {
 		return nil, fmt.Errorf("copy master data: %w", err)
 	}
-	if err := copyDirContents(slaveSource, filepath.Join(tempDir, "slave")); err != nil {
+	if err := copyDataTree(slaveSource, filepath.Join(tempDir, "slave"), p.SlaveData, p.SlaveSourcePath); err != nil {
 		return nil, fmt.Errorf("copy slave data: %w", err)
 	}

@@ -540,6 +547,10 @@ func restorePair(cfg config, inputRoot string, outputRoot string, p pair, logger
 		SlaveData:         p.SlaveData,
 		MasterOrbit:       p.MasterOrbit,
 		SlaveOrbit:        p.SlaveOrbit,
+		MasterSourcePath:  p.MasterSourcePath,
+		SlaveSourcePath:   p.SlaveSourcePath,
+		MasterOrbitSource: p.MasterOrbitSource,
+		SlaveOrbitSource:  p.SlaveOrbitSource,
 		MasterImagingDate: p.MasterImagingDate,
 		SlaveImagingDate:  p.SlaveImagingDate,
 		TimeBaselineDays:  p.TimeBaselineDays,
@@ -575,15 +586,23 @@ func taskDirectoryName(p pair) (string, error) {
 func copyOrbitFiles(inputRoot string, tempDir string, taskDirName string, p pair, logger *log.Logger) []string {
 	var warnings []string
 	for _, entry := range []struct {
-		label string
-		value string
+		label      string
+		value      string
+		sourcePath string
 	}{
-		{label: "master_orbit", value: p.MasterOrbit},
-		{label: "slave_orbit", value: p.SlaveOrbit},
+		{label: "master_orbit", value: p.MasterOrbit, sourcePath: p.MasterOrbitSource},
+		{label: "slave_orbit", value: p.SlaveOrbit, sourcePath: p.SlaveOrbitSource},
 	} {
 		if strings.TrimSpace(entry.value) == "" {
 			continue
 		}
+		destName, err := restoredOrbitFileName(entry.value, entry.sourcePath)
+		if err != nil {
+			warning := fmt.Sprintf("task %s %s cannot determine restored file name: %v", taskDirName, entry.label, err)
+			warnings = append(warnings, warning)
+			logger.Printf("warning %s", warning)
+			continue
+		}
 		source, err := safeJoin(inputRoot, entry.value)
 		if err != nil {
 			warning := fmt.Sprintf("task %s %s invalid: %v", taskDirName, entry.label, err)
@@ -611,7 +630,7 @@ func copyOrbitFiles(inputRoot string, tempDir string, taskDirName string, p pair
 			logger.Printf("warning %s", warning)
 			continue
 		}
-		if err := copyFile(source, filepath.Join(orbitDir, filepath.Base(source)), stat.Mode()); err != nil {
+		if err := copyFile(source, filepath.Join(orbitDir, destName), stat.Mode()); err != nil {
 			warning := fmt.Sprintf("task %s cannot copy %s %s: %v", taskDirName, entry.label, entry.value, err)
 			warnings = append(warnings, warning)
 			logger.Printf("warning %s", warning)
@@ -624,15 +643,20 @@ func copyOrbitFiles(inputRoot string, tempDir string, taskDirName string, p pair
 func orbitWarnings(inputRoot string, p pair, taskDirName string) []string {
 	var warnings []string
 	for _, entry := range []struct {
-		label string
-		value string
+		label      string
+		value      string
+		sourcePath string
 	}{
-		{label: "master_orbit", value: p.MasterOrbit},
-		{label: "slave_orbit", value: p.SlaveOrbit},
+		{label: "master_orbit", value: p.MasterOrbit, sourcePath: p.MasterOrbitSource},
+		{label: "slave_orbit", value: p.SlaveOrbit, sourcePath: p.SlaveOrbitSource},
 	} {
 		if strings.TrimSpace(entry.value) == "" {
 			continue
 		}
+		if _, err := restoredOrbitFileName(entry.value, entry.sourcePath); err != nil {
+			warnings = append(warnings, fmt.Sprintf("task %s %s cannot determine restored file name: %v", taskDirName, entry.label, err))
+			continue
+		}
 		source, err := safeJoin(inputRoot, entry.value)
 		if err != nil {
 			warnings = append(warnings, fmt.Sprintf("task %s %s invalid: %v", taskDirName, entry.label, err))
@@ -647,7 +671,158 @@ func orbitWarnings(inputRoot string, p pair, taskDirName string) []string {
 	return warnings
 }

+func copyDataTree(sourceDir string, destRoleDir string, bundlePath string, sourcePath string) error {
+	rootName, hasRootName, err := restoredDataRootName(bundlePath, sourcePath)
+	if err != nil {
+		return err
+	}
+	if hasRootName {
+		if nestedSource, ok, err := singleNestedDataRoot(sourceDir, rootName); err != nil {
+			return err
+		} else if ok {
+			sourceDir = nestedSource
+		}
+	}
+	return copyDirContentsRenamed(sourceDir, destRoleDir, restoredDataEntryName)
+}
+
+func singleNestedDataRoot(sourceDir string, rootName string) (string, bool, error) {
+	entries, err := os.ReadDir(sourceDir)
+	if err != nil {
+		return "", false, err
+	}
+	if len(entries) != 1 {
+		return "", false, nil
+	}
+	entry := entries[0]
+	if !entry.IsDir() {
+		return "", false, nil
+	}
+	if restoredDataEntryName(entry.Name()) != rootName {
+		return "", false, nil
+	}
+	return filepath.Join(sourceDir, entry.Name()), true, nil
+}
+
+func restoredDataRootName(bundlePath string, sourcePath string) (string, bool, error) {
+	if name, ok, err := originalNameFromSourcePath(sourcePath); err != nil {
+		return "", false, err
+	} else if ok {
+		return name, true, nil
+	}
+
+	name := pathLeafName(bundlePath)
+	if stripped := stripBundlePrefix(name, "scene"); stripped != name {
+		if err := validateRestoredLeafName(stripped); err != nil {
+			return "", false, err
+		}
+		return stripped, true, nil
+	}
+	return "", false, nil
+}
+
+func restoredDataEntryName(name string) string {
+	return stripBundlePrefix(name, "scene")
+}
+
+func restoredOrbitFileName(bundlePath string, sourcePath string) (string, error) {
+	if name, ok, err := originalNameFromSourcePath(sourcePath); err != nil {
+		return "", err
+	} else if ok {
+		return name, nil
+	}
+
+	name := stripBundlePrefix(pathLeafName(bundlePath), "orbit")
+	if err := validateRestoredLeafName(name); err != nil {
+		return "", err
+	}
+	return name, nil
+}
+
+func originalNameFromSourcePath(sourcePath string) (string, bool, error) {
+	name := pathLeafName(sourcePath)
+	if name == "" {
+		return "", false, nil
+	}
+	switch strings.ToLower(name) {
+	case "master", "slave", "data", "orbit":
+		return "", false, nil
+	}
+	if err := validateRestoredLeafName(name); err != nil {
+		return "", false, err
+	}
+	return name, true, nil
+}
+
+func stripBundlePrefix(name string, prefix string) string {
+	rest, ok := strings.CutPrefix(name, prefix+"_")
+	if !ok {
+		return name
+	}
+	hashEnd := strings.IndexByte(rest, '_')
+	if hashEnd <= 0 || hashEnd == len(rest)-1 {
+		return name
+	}
+	if !looksLikeHashToken(rest[:hashEnd]) {
+		return name
+	}
+	return rest[hashEnd+1:]
+}
+
+func looksLikeHashToken(value string) bool {
+	if len(value) < 6 {
+		return false
+	}
+	for _, r := range value {
+		switch {
+		case r >= '0' && r <= '9':
+		case r >= 'a' && r <= 'f':
+		case r >= 'A' && r <= 'F':
+		default:
+			return false
+		}
+	}
+	return true
+}
+
+func pathLeafName(value string) string {
+	value = strings.TrimSpace(value)
+	value = strings.TrimRight(value, `/\`)
+	if value == "" {
+		return ""
+	}
+	index := strings.LastIndexAny(value, `/\`)
+	if index >= 0 {
+		return value[index+1:]
+	}
+	return value
+}
+
+func validateRestoredLeafName(name string) error {
+	if name == "" || name == "." || name == ".." {
+		return fmt.Errorf("invalid empty file name")
+	}
+	if strings.ContainsAny(name, `<>:"/\|?*`) {
+		return fmt.Errorf("invalid file name %q", name)
+	}
+	if strings.HasSuffix(name, " ") || strings.HasSuffix(name, ".") {
+		return fmt.Errorf("invalid file name %q", name)
+	}
+	for _, r := range name {
+		if r < 32 {
+			return fmt.Errorf("invalid file name %q", name)
+		}
+	}
+	return nil
+}
+
 func copyDirContents(sourceDir string, destDir string) error {
+	return copyDirContentsRenamed(sourceDir, destDir, func(name string) string {
+		return name
+	})
+}
+
+func copyDirContentsRenamed(sourceDir string, destDir string, renameEntry func(string) string) error {
 	sourceEntries, err := os.ReadDir(sourceDir)
 	if err != nil {
 		return err
@@ -657,7 +832,11 @@ func copyDirContents(sourceDir string, destDir string) error {
 	}
 	for _, entry := range sourceEntries {
 		sourcePath := filepath.Join(sourceDir, entry.Name())
-		destPath := filepath.Join(destDir, entry.Name())
+		destName := renameEntry(entry.Name())
+		if err := validateRestoredLeafName(destName); err != nil {
+			return err
+		}
+		destPath := filepath.Join(destDir, destName)
 		info, err := entry.Info()
 		if err != nil {
 			return err
@@ -667,7 +846,7 @@ func copyDirContents(sourceDir string, destDir string) error {
 		case mode&os.ModeSymlink != 0:
 			return fmt.Errorf("symlinks are not supported: %s", sourcePath)
 		case info.IsDir():
-			if err := copyDirContents(sourcePath, destPath); err != nil {
+			if err := copyDirContentsRenamed(sourcePath, destPath, renameEntry); err != nil {
 				return err
 			}
 		case mode.IsRegular():
@@ -59,6 +59,95 @@ func TestRunRestoresPair(t *testing.T) {
 	}
 }

+func TestRunRestoresOriginalDataAndOrbitNames(t *testing.T) {
+	root := t.TempDir()
+	input := filepath.Join(root, "bundle")
+	output := filepath.Join(root, "out")
+
+	masterName := "LT1B_MONO_SYC_STRIP1_019663_E131.0_N44.8_20251010_SLC_HH_S2A_0000898998"
+	slaveName := "LT1A_MONO_SYC_STRIP1_019664_E131.0_N44.8_20251022_SLC_HH_S2A_0000899001"
+	masterBundleName := "scene_e1e6ef658c_" + masterName
+	slaveBundleName := "scene_abc123def4_" + slaveName
+
+	writeTestFile(t, filepath.Join(input, "data", masterBundleName, "scene_e1e6ef658c_"+masterName+".tiff"), "master")
+	writeTestFile(t, filepath.Join(input, "data", slaveBundleName, "scene_abc123def4_"+slaveName+".tiff"), "slave")
+	writeTestFile(t, filepath.Join(input, "orbit", "orbit_0123abcd_LT1A_GpsData_GAS_C_20230508.txt"), "master orbit")
+	writeTestFile(t, filepath.Join(input, "orbit", "orbit_abcd1234_LT1B_GpsData_GAS_C_20230508.txt"), "slave orbit")
+	writeTestJSON(t, filepath.Join(input, pairsFileName), pairsDocument{
+		Pairs: []pair{{
+			PairID:      "pair_0001",
+			TaskName:    "Task_20251010_20251022",
+			MasterData:  filepath.ToSlash(filepath.Join("data", masterBundleName)),
+			SlaveData:   filepath.ToSlash(filepath.Join("data", slaveBundleName)),
+			MasterOrbit: "orbit/orbit_0123abcd_LT1A_GpsData_GAS_C_20230508.txt",
+			SlaveOrbit:  "orbit/orbit_abcd1234_LT1B_GpsData_GAS_C_20230508.txt",
+		}},
+	})
+
+	result, err := run(config{inputRoot: input, outputRoot: output, skipExisting: true, logWriter: io.Discard})
+	if err != nil {
+		t.Fatalf("run failed: %v", err)
+	}
+	if result.report.Restored != 1 {
+		t.Fatalf("restored = %d, want 1", result.report.Restored)
+	}
+
+	task := filepath.Join(output, "Task_20251010_20251022")
+	for _, path := range []string{
+		filepath.Join(task, "master", masterName+".tiff"),
+		filepath.Join(task, "slave", slaveName+".tiff"),
+		filepath.Join(task, "orbit", "LT1A_GpsData_GAS_C_20230508.txt"),
+		filepath.Join(task, "orbit", "LT1B_GpsData_GAS_C_20230508.txt"),
+	} {
+		if _, err := os.Stat(path); err != nil {
+			t.Fatalf("expected %s to exist: %v", path, err)
+		}
+	}
+	for _, path := range []string{
+		filepath.Join(task, "master", masterBundleName),
+		filepath.Join(task, "slave", slaveBundleName),
+		filepath.Join(task, "master", "scene_e1e6ef658c_"+masterName+".tiff"),
+		filepath.Join(task, "slave", "scene_abc123def4_"+slaveName+".tiff"),
+		filepath.Join(task, "orbit", "orbit_0123abcd_LT1A_GpsData_GAS_C_20230508.txt"),
+		filepath.Join(task, "orbit", "orbit_abcd1234_LT1B_GpsData_GAS_C_20230508.txt"),
+	} {
+		if _, err := os.Stat(path); !os.IsNotExist(err) {
+			t.Fatalf("expected %s to be absent, stat err=%v", path, err)
+		}
+	}
+}
+
+func TestRunCollapsesNestedPrefixedDataRoot(t *testing.T) {
+	root := t.TempDir()
+	input := filepath.Join(root, "bundle")
+	output := filepath.Join(root, "out")
+
+	originalName := "LT1B_MONO_SYC_STRIP1_019663_E131.0_N44.8_20251010_SLC_HH_S2A_0000898998"
+	bundleName := "scene_e1e6ef658c_" + originalName
+	writeTestFile(t, filepath.Join(input, "data", bundleName, bundleName, "m.txt"), "master")
+	writeTestFile(t, filepath.Join(input, "data", "scene_abc123def4_slave", "s.txt"), "slave")
+	writeTestJSON(t, filepath.Join(input, pairsFileName), pairsDocument{
+		Pairs: []pair{{
+			PairID:     "pair_0001",
+			TaskName:   "Task_20251010_20251022",
+			MasterData: filepath.ToSlash(filepath.Join("data", bundleName)),
+			SlaveData:  "data/scene_abc123def4_slave",
+		}},
+	})
+
+	if _, err := run(config{inputRoot: input, outputRoot: output, skipExisting: true, logWriter: io.Discard}); err != nil {
+		t.Fatalf("run failed: %v", err)
+	}
+
+	task := filepath.Join(output, "Task_20251010_20251022")
+	if _, err := os.Stat(filepath.Join(task, "master", "m.txt")); err != nil {
+		t.Fatalf("expected nested data root to be collapsed: %v", err)
+	}
+	if _, err := os.Stat(filepath.Join(task, "master", originalName)); !os.IsNotExist(err) {
+		t.Fatalf("expected duplicated nested data root to be absent, stat err=%v", err)
+	}
+}
+
 func TestRunSkipsExistingCompleteTask(t *testing.T) {
 	root := t.TempDir()
 	input := filepath.Join(root, "bundle")
@@ -194,6 +283,48 @@ func TestRunCheckTreatsMissingOrbitAsWarning(t *testing.T) {
 	}
 }

+func TestRunCheckAllowsExtraPairMetadata(t *testing.T) {
+	root := t.TempDir()
+	input := filepath.Join(root, "bundle")
+
+	writeTestFile(t, filepath.Join(input, "data", "scene_master", "m.txt"), "master")
+	writeTestFile(t, filepath.Join(input, "data", "scene_slave", "s.txt"), "slave")
+	writeTestFile(t, filepath.Join(input, "orbit", "master.txt"), "master orbit")
+	writeTestFile(t, filepath.Join(input, "orbit", "slave.txt"), "slave orbit")
+	writeTestFile(t, filepath.Join(input, pairsFileName), `{
+  "schema": "dinsar_source_bundle_pairs.v1",
+  "exported_at": "2026-05-11T07:20:15Z",
+  "pairs": [
+    {
+      "pair_id": "pair_0001",
+      "identity_key": "uid:test",
+      "task_name": "Task_20250101_20250113",
+      "task_alias": "Task_20250101_20250113",
+      "pair_key": "lt1_test",
+      "scene_pair_uid": "test",
+      "master_data": "data/scene_master",
+      "slave_data": "data/scene_slave",
+      "master_orbit": "orbit/master.txt",
+      "slave_orbit": "orbit/slave.txt",
+      "master_source_path": "D:\\source\\master",
+      "slave_source_path": "D:\\source\\slave",
+      "scene_center_distance_meters": 123.4
+    }
+  ]
+}`)
+
+	result, err := runCheck(config{inputRoot: input, logWriter: io.Discard})
+	if err != nil {
+		t.Fatalf("runCheck failed: %v", err)
+	}
+	if result.report.Failed != 0 {
+		t.Fatalf("failed = %d, want 0; errors=%v", result.report.Failed, result.report.Errors)
+	}
+	if result.report.Valid != 1 {
+		t.Fatalf("valid = %d, want 1", result.report.Valid)
+	}
+}
+
 func TestParseFlagsRejectsSkipExistingWithOverwrite(t *testing.T) {
 	_, err := parseFlags([]string{"--input", "in", "--output", "out", "--skip-existing", "--overwrite"})
 	if err == nil {
Author	SHA1	Message	Date
Harmon	27fb516d1b	Restore original bundle file names	2026-05-16 16:49:18 +08:00
Harmon	5a2248667d	Allow extra pair metadata in bundle checks	2026-05-11 16:12:26 +08:00