@danielarndt0/cnpj-db-loader 2.3.1 → 2.4.0-beta.2

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/dist/index.d.ts CHANGED
@@ -582,6 +582,8 @@ declare function showQuarantineRow(id: number, options?: {
582
582
  dbUrl?: string;
583
583
  }): Promise<QuarantineRecord>;
584
584
 
585
+ type SanitizeSourceEncoding = "WIN1252" | "LATIN1" | "UTF8";
586
+
585
587
  type SanitizeDatasetType = Exclude<DatasetType, "zip-archive" | "unknown">;
586
588
  type SanitizeFilePlan = {
587
589
  dataset: SanitizeDatasetType;
@@ -607,7 +609,10 @@ type SanitizeSummary = {
607
609
  totalBytes: number;
608
610
  processedFiles: number;
609
611
  processedRows: number;
612
+ sourceEncoding: SanitizeSourceEncoding;
610
613
  nulBytesRemoved: number;
614
+ invalidBytesRemoved: number;
615
+ controlCharsRemoved: number;
611
616
  changedFiles: number;
612
617
  unchangedFiles: number;
613
618
  datasets: SanitizeDatasetType[];
@@ -618,6 +623,8 @@ type SanitizeSummary = {
618
623
  lineCount: number;
619
624
  changed: boolean;
620
625
  nulBytesRemoved: number;
626
+ invalidBytesRemoved: number;
627
+ controlCharsRemoved: number;
621
628
  }>;
622
629
  warnings: string[];
623
630
  nextStep?: string | undefined;
@@ -629,6 +636,7 @@ type SanitizeProgressEvent = {
629
636
  totalFiles: number;
630
637
  totalBytes: number;
631
638
  datasets: SanitizeDatasetType[];
639
+ sourceEncoding: SanitizeSourceEncoding;
632
640
  } | {
633
641
  kind: "progress";
634
642
  currentFileDisplayPath: string;
@@ -640,12 +648,16 @@ type SanitizeProgressEvent = {
640
648
  currentFileSize: number;
641
649
  processedRows: number;
642
650
  nulBytesRemoved: number;
651
+ invalidBytesRemoved: number;
652
+ controlCharsRemoved: number;
643
653
  changedFiles: number;
644
654
  } | {
645
655
  kind: "finish";
646
656
  totalFiles: number;
647
657
  processedRows: number;
648
658
  nulBytesRemoved: number;
659
+ invalidBytesRemoved: number;
660
+ controlCharsRemoved: number;
649
661
  changedFiles: number;
650
662
  totalBytes: number;
651
663
  };
@@ -653,6 +665,7 @@ type SanitizeProgressListener = (event: SanitizeProgressEvent) => void;
653
665
  type SanitizeOptions = {
654
666
  outputPath?: string | undefined;
655
667
  dataset?: SanitizeDatasetType | undefined;
668
+ sourceEncoding?: string | undefined;
656
669
  onProgress?: SanitizeProgressListener | undefined;
657
670
  };
658
671
 
@@ -967,4 +980,124 @@ declare function finalizeFederalRevenueManifest(outputPath: string, lastStatus:
967
980
 
968
981
  declare function syncFederalRevenueDataset(options?: FederalRevenueSyncOptions): Promise<FederalRevenueSyncSummary>;
969
982
 
970
- export { type AppConfig, type AppEnvironment, AppError, type CheckpointCleanupPhase, DEFAULT_FEDERAL_REVENUE_DOWNLOAD_ROOT, DEFAULT_FEDERAL_REVENUE_SHARE_TOKEN, DEFAULT_FEDERAL_REVENUE_USER_AGENT, DEFAULT_FEDERAL_REVENUE_WEBDAV_URL, type DatabaseCleanupSummary, type DatabaseConfig, type DatasetBlock, type DatasetType, type ExtractionEntry, type ExtractionProgressEvent, type ExtractionProgressListener, type ExtractionSummary, FEDERAL_REVENUE_CONTROL_DIR, FEDERAL_REVENUE_CONTROL_SCOPE, FEDERAL_REVENUE_MANIFEST_VERSION, type Failure, type FederalRevenueCheckOptions, type FederalRevenueCheckSummary, type FederalRevenueCleanMode, type FederalRevenueCleanOptions, type FederalRevenueCleanSummary, type FederalRevenueClientOptions, type FederalRevenueDownloadEntry, type FederalRevenueDownloadOptions, type FederalRevenueDownloadProgressEvent, type FederalRevenueDownloadProgressListener, type FederalRevenueDownloadStatus, type FederalRevenueDownloadSummary, type FederalRevenueFile, type FederalRevenueLocalFileStatus, type FederalRevenueLocalStatusEntry, type FederalRevenueLockFile, type FederalRevenueManifest, type FederalRevenueManifestFile, type FederalRevenueManifestLastCommand, type FederalRevenueManifestLastStatus, type FederalRevenueReference, type FederalRevenueReferenceMode, type FederalRevenueReferenceSelection, type FederalRevenueRetryOptions, type FederalRevenueStatusOptions, type FederalRevenueStatusSummary, type FederalRevenueSyncLockOptions, type FederalRevenueSyncOptions, type FederalRevenueSyncSummary, type FileInspection, type ImportCheckpointRecord, type ImportCheckpointStatus, type ImportDatasetPlan, type ImportDatasetType, type ImportFilePlan, type ImportOptions, type ImportPerformanceSummary, type ImportPhaseStatus, type ImportPlanRecord, type ImportProgressEvent, type ImportProgressListener, type ImportSchemaCapabilities, type ImportSummary, type InputDetectionMode, type InputMode, type InspectSummary, type LogLevel, type LogStatus, type QuarantineListFilters, type QuarantineListSummary, type QuarantineRecord, type QuarantineStatsFilters, type QuarantineStatsSummary, type Result, type SanitizeDatasetType, type SanitizeOptions, type SanitizePlan, type SanitizeProgressEvent, type SanitizeProgressListener, type SanitizeSummary, type SchemaGenerationOptions, type SchemaProfile, ServiceError, type StructuredLogEntry, type SupportedOs, ValidationError, type ValidationSummary, appendJsonLinesLog, assertPostgresUrl, buildFederalRevenueDownloadHeaders, buildFederalRevenueReferenceOutputPath, checkFederalRevenueDataset, cleanFederalRevenueDataset, cleanupDatabaseCheckpointsData, cleanupDatabaseMaterializedData, cleanupDatabasePlansData, cleanupDatabaseStagingData, createFederalRevenueManifest, createJsonLinesLog, defaultExtractedOutputPath, detectOs, downloadFederalRevenueDataset, ensureDirectory, evaluateFederalRevenueManifestFile, evaluateFederalRevenueManifestFiles, extractArchives, finalizeFederalRevenueManifest, generateSchemaSql, getAllLayouts, getCurrentFederalRevenueReference, getFederalRevenueControlDirectory, getFederalRevenueManifestPath, getFederalRevenueStatus, getFederalRevenueSyncLockPath, getLayoutSummary, getLogsDirectoryPath, getQuarantineStats, getUserAppDirectoryPath, importDataToDatabase, inspectFiles, listFederalRevenueFiles, listFederalRevenueReferences, listQuarantineRows, loadImportDataToStaging, materializeImportedData, prettyJson, readDatabaseConfig, readFederalRevenueManifest, resetDefaultDbUrl, resolveDatabaseUrl, resolveDbUrl, resolveFederalRevenueReference, resolveInputMode, resolveSchemaProfile, retryFederalRevenueDataset, runDoctor, safeReadText, safeWriteText, sanitizeInputDirectory, setDefaultDbUrl, showQuarantineRow, syncFederalRevenueDataset, testDatabaseConnection, toTitleCase, updateFederalRevenueManifestFile, validateFederalRevenueReference, validateInputDirectory, withFederalRevenueSyncLock, writeCommandFailureLog, writeCommandLog, writeDatabaseConfig, writeFederalRevenueManifest, writeSchemaFile };
983
+ type PostgresCsvDatasetSummary = {
984
+ dataset: ImportDatasetType;
985
+ files: number;
986
+ rows: number;
987
+ outputFiles: string[];
988
+ };
989
+ type PostgresCsvExportProgressEvent = {
990
+ kind: "start";
991
+ inputPath: string;
992
+ validatedPath: string;
993
+ outputPath: string;
994
+ totalFiles: number;
995
+ datasets: ImportDatasetType[];
996
+ } | {
997
+ kind: "file_start";
998
+ dataset: ImportDatasetType;
999
+ fileIndex: number;
1000
+ totalFiles: number;
1001
+ inputFile: string;
1002
+ outputFile: string;
1003
+ } | {
1004
+ kind: "file_finish";
1005
+ dataset: ImportDatasetType;
1006
+ fileIndex: number;
1007
+ totalFiles: number;
1008
+ inputFile: string;
1009
+ outputFile: string;
1010
+ rows: number;
1011
+ } | {
1012
+ kind: "finish";
1013
+ outputPath: string;
1014
+ scriptPath: string;
1015
+ totalFiles: number;
1016
+ totalRows: number;
1017
+ };
1018
+ type PostgresCsvExportProgressListener = (event: PostgresCsvExportProgressEvent) => void;
1019
+ type PostgresCsvExportOptions = {
1020
+ outputPath?: string | undefined;
1021
+ dataset?: ImportDatasetType | undefined;
1022
+ scriptName?: string | undefined;
1023
+ onProgress?: PostgresCsvExportProgressListener | undefined;
1024
+ };
1025
+ type PostgresCsvExportSummary = {
1026
+ inputPath: string;
1027
+ validatedPath: string;
1028
+ outputPath: string;
1029
+ scriptPath: string;
1030
+ manifestPath: string;
1031
+ totalFiles: number;
1032
+ totalRows: number;
1033
+ datasets: PostgresCsvDatasetSummary[];
1034
+ warnings: string[];
1035
+ nextStep?: string | undefined;
1036
+ };
1037
+ type PostgresCsvFile = {
1038
+ dataset: ImportDatasetType;
1039
+ absolutePath: string;
1040
+ relativePath: string;
1041
+ rowCount: number;
1042
+ };
1043
+ type PostgresDirectSourceFile = {
1044
+ dataset: ImportDatasetType;
1045
+ absolutePath: string;
1046
+ relativePath: string;
1047
+ fileSize: number;
1048
+ };
1049
+ type PostgresDirectScriptDatasetSummary = {
1050
+ dataset: ImportDatasetType;
1051
+ files: number;
1052
+ totalBytes: number;
1053
+ sourceFiles: string[];
1054
+ };
1055
+ type PostgresDirectScriptProgressEvent = {
1056
+ kind: "start";
1057
+ inputPath: string;
1058
+ validatedPath: string;
1059
+ outputPath: string;
1060
+ totalFiles: number;
1061
+ datasets: ImportDatasetType[];
1062
+ sourceEncoding: string;
1063
+ } | {
1064
+ kind: "file_registered";
1065
+ dataset: ImportDatasetType;
1066
+ fileIndex: number;
1067
+ totalFiles: number;
1068
+ inputFile: string;
1069
+ fileSize: number;
1070
+ } | {
1071
+ kind: "finish";
1072
+ outputPath: string;
1073
+ scriptPath: string;
1074
+ totalFiles: number;
1075
+ totalBytes: number;
1076
+ };
1077
+ type PostgresDirectScriptProgressListener = (event: PostgresDirectScriptProgressEvent) => void;
1078
+ type PostgresDirectScriptOptions = {
1079
+ outputPath?: string | undefined;
1080
+ dataset?: ImportDatasetType | undefined;
1081
+ scriptName?: string | undefined;
1082
+ sourceEncoding?: string | undefined;
1083
+ onProgress?: PostgresDirectScriptProgressListener | undefined;
1084
+ };
1085
+ type PostgresDirectScriptSummary = {
1086
+ inputPath: string;
1087
+ validatedPath: string;
1088
+ outputPath: string;
1089
+ scriptPath: string;
1090
+ manifestPath: string;
1091
+ sourceEncoding: string;
1092
+ totalFiles: number;
1093
+ totalBytes: number;
1094
+ datasets: PostgresDirectScriptDatasetSummary[];
1095
+ warnings: string[];
1096
+ nextStep?: string | undefined;
1097
+ };
1098
+
1099
+ declare function exportPostgresCsvDataset(inputPath: string, options?: PostgresCsvExportOptions): Promise<PostgresCsvExportSummary>;
1100
+
1101
+ declare function generatePostgresDirectScript(inputPath: string, options?: PostgresDirectScriptOptions): Promise<PostgresDirectScriptSummary>;
1102
+
1103
+ export { type AppConfig, type AppEnvironment, AppError, type CheckpointCleanupPhase, DEFAULT_FEDERAL_REVENUE_DOWNLOAD_ROOT, DEFAULT_FEDERAL_REVENUE_SHARE_TOKEN, DEFAULT_FEDERAL_REVENUE_USER_AGENT, DEFAULT_FEDERAL_REVENUE_WEBDAV_URL, type DatabaseCleanupSummary, type DatabaseConfig, type DatasetBlock, type DatasetType, type ExtractionEntry, type ExtractionProgressEvent, type ExtractionProgressListener, type ExtractionSummary, FEDERAL_REVENUE_CONTROL_DIR, FEDERAL_REVENUE_CONTROL_SCOPE, FEDERAL_REVENUE_MANIFEST_VERSION, type Failure, type FederalRevenueCheckOptions, type FederalRevenueCheckSummary, type FederalRevenueCleanMode, type FederalRevenueCleanOptions, type FederalRevenueCleanSummary, type FederalRevenueClientOptions, type FederalRevenueDownloadEntry, type FederalRevenueDownloadOptions, type FederalRevenueDownloadProgressEvent, type FederalRevenueDownloadProgressListener, type FederalRevenueDownloadStatus, type FederalRevenueDownloadSummary, type FederalRevenueFile, type FederalRevenueLocalFileStatus, type FederalRevenueLocalStatusEntry, type FederalRevenueLockFile, type FederalRevenueManifest, type FederalRevenueManifestFile, type FederalRevenueManifestLastCommand, type FederalRevenueManifestLastStatus, type FederalRevenueReference, type FederalRevenueReferenceMode, type FederalRevenueReferenceSelection, type FederalRevenueRetryOptions, type FederalRevenueStatusOptions, type FederalRevenueStatusSummary, type FederalRevenueSyncLockOptions, type FederalRevenueSyncOptions, type FederalRevenueSyncSummary, type FileInspection, type ImportCheckpointRecord, type ImportCheckpointStatus, type ImportDatasetPlan, type ImportDatasetType, type ImportFilePlan, type ImportOptions, type ImportPerformanceSummary, type ImportPhaseStatus, type ImportPlanRecord, type ImportProgressEvent, type ImportProgressListener, type ImportSchemaCapabilities, type ImportSummary, type InputDetectionMode, type InputMode, type InspectSummary, type LogLevel, type LogStatus, type PostgresCsvDatasetSummary, type PostgresCsvExportOptions, type PostgresCsvExportProgressEvent, type PostgresCsvExportProgressListener, type PostgresCsvExportSummary, type PostgresCsvFile, type PostgresDirectScriptDatasetSummary, type PostgresDirectScriptOptions, type PostgresDirectScriptProgressEvent, type PostgresDirectScriptProgressListener, type PostgresDirectScriptSummary, type PostgresDirectSourceFile, type QuarantineListFilters, type QuarantineListSummary, type QuarantineRecord, type QuarantineStatsFilters, type QuarantineStatsSummary, type Result, type SanitizeDatasetType, type SanitizeOptions, type SanitizePlan, type SanitizeProgressEvent, type SanitizeProgressListener, type SanitizeSourceEncoding, type SanitizeSummary, type SchemaGenerationOptions, type SchemaProfile, ServiceError, type StructuredLogEntry, type SupportedOs, ValidationError, type ValidationSummary, appendJsonLinesLog, assertPostgresUrl, buildFederalRevenueDownloadHeaders, buildFederalRevenueReferenceOutputPath, checkFederalRevenueDataset, cleanFederalRevenueDataset, cleanupDatabaseCheckpointsData, cleanupDatabaseMaterializedData, cleanupDatabasePlansData, cleanupDatabaseStagingData, createFederalRevenueManifest, createJsonLinesLog, defaultExtractedOutputPath, detectOs, downloadFederalRevenueDataset, ensureDirectory, evaluateFederalRevenueManifestFile, evaluateFederalRevenueManifestFiles, exportPostgresCsvDataset, extractArchives, finalizeFederalRevenueManifest, generatePostgresDirectScript, generateSchemaSql, getAllLayouts, getCurrentFederalRevenueReference, getFederalRevenueControlDirectory, getFederalRevenueManifestPath, getFederalRevenueStatus, getFederalRevenueSyncLockPath, getLayoutSummary, getLogsDirectoryPath, getQuarantineStats, getUserAppDirectoryPath, importDataToDatabase, inspectFiles, listFederalRevenueFiles, listFederalRevenueReferences, listQuarantineRows, loadImportDataToStaging, materializeImportedData, prettyJson, readDatabaseConfig, readFederalRevenueManifest, resetDefaultDbUrl, resolveDatabaseUrl, resolveDbUrl, resolveFederalRevenueReference, resolveInputMode, resolveSchemaProfile, retryFederalRevenueDataset, runDoctor, safeReadText, safeWriteText, sanitizeInputDirectory, setDefaultDbUrl, showQuarantineRow, syncFederalRevenueDataset, testDatabaseConnection, toTitleCase, updateFederalRevenueManifestFile, validateFederalRevenueReference, validateInputDirectory, withFederalRevenueSyncLock, writeCommandFailureLog, writeCommandLog, writeDatabaseConfig, writeFederalRevenueManifest, writeSchemaFile };