@whaly/connector-sdk 0.3.9 → 0.3.10

package/dist/index.d.mts

@@ -147,6 +147,11 @@ declare const MIME_TYPES: Record<string, string>;
  * Returns `application/octet-stream` when the extension is unknown.
  */
 declare function getMimeType(filePathOrExt: string): string;
+/**
+ * Look up the preferred file extension for a MIME type.
+ * Returns `.bin` when the MIME type is unknown.
+ */
+declare function getExtension(mimeType: string): string;
 
 declare enum ReplicationMethod {
     INCREMENTAL = "INCREMENTAL",
@@ -1639,4 +1644,4 @@ declare class GCSStateProvider implements StateProvider {
     writeState(state: string): Promise<void>;
 }
 
-export { ALL_STREAM_ID_KEYWORD, API_CALLS_METRIC_NAME, type AssetEntry, type AssetManifest, type AssetManifestEntry, type AssetReplicationMode, AssetStream, AssetTap, AssetTarget, Authenticator, BATCH_INTERVAL_MS, type BaseConfig, type BigQueryConfig, BigQueryTarget, type BigqueryLocation, type Bookmark, type Bookmarks, CdnAssetTarget, type CdnAssetTargetConfig, type CdnFileMetadata, CdnService, type CdnServiceConfig, type CdnUploadResult, CloudStorageService, type CloudStorageServiceOptions, type ColumnMappingStoreUnsafeToSafe, CounterMetric, CsvExtractionConfigBuilder, type CsvFieldsArrayConfig, type CsvFieldsConfig, type CsvFieldsDictConfig, type CsvFileConfig, type CsvStreamConfig, DEFAULT_MAX_CONCURRENT_STREAMS, DEFAULT_SYNCED_AT_COLUMN, DocumentDownloadSkipError, type DocumentEntry, type DocumentManifest, type DocumentManifestEntry, type DocumentStatus, DocumentStream, type DocumentStreamManifest, type DocumentSummary, DocumentTap, type DownloadFilesOptions, EXECUTION_TIME_METRIC_NAME, type ErrorType, type ExcelCustomExtractorConfig, type ExcelDerivedField, type ExcelExtractionBaseConfig, type ExcelExtractionConfig, ExcelExtractionConfigBuilder, type ExcelFieldMapping, type ExcelFieldSpec, type ExcelSingleSheetExtractionConfig, type ExcelSourceColumn, type ExcelStreamConfig, type FieldType, FileFormat, FilePatterns, FileStream, type FileStreamConfig, type FileStreamEntry, FileTap, type FileTapConfig, type FlattenedSchema, GCSStateProvider, type GetConfig, type HTTPHeaders, type HTTPMethod, ITarget, ImageTransform, type InputTapState, type InternalTableField, type JSONSchemaFieldDefinition, type ListFilesOptions, MIME_TYPES, type Message, type MessageType, type MetricConfiguration, MissingFieldInSchemaError, MissingSchemaError, type PostConfig, type ProcessedAsset, type ProgressMarkers, type PropertiesMetadata, RESTStream, ROWS_SYNCED_METRIC_NAME, RecordMessage, RenameColumnStore, ReplicationMethod, ReplicationMethodMessage, type SafeColumnNameConverterFn, type Schema, SchemaMessage, SchemaValidationError, SftpService, type State, type StateHolder, StateMessage, type StateProvider, StateService, Stream, type StreamId, type StreamManifest, type StreamState, StreamWarehouseSyncService, type StreamWithTempFile, type SyncOptions, Tap, type TapState, type TargetSchemaHook, type TargetSchemaHookInput, type TempFile, type URLParams, VariableExtractors, type WarehouseTableField, type WebpTransformOptions, type WhalyDocument, WhalyDocumentService, type WhalyDocumentServiceConfig, WhalyDocumentTarget, type WhalyPaginatedResponse, type WhalyUploadResult, type WriteGeneratorToCsvOptions, _greaterThan, addDocumentSummaries, addWhalyFields, checkCsvHeaderRow, coerceCellValue, collectPaginated, countCsvLines, createCellReference, createCsvStreamConfig, createExcelGenerator, createExcelStreamConfig, createTemporaryFileStream, csvFieldsToJsonSchema, downloadFiles, emptyDocumentSummary, excelColumnToIndex, excelFieldsToJsonSchema, excelRowToIndex, extractExcelRows, extractPrimaryKeysFromCsvConfig, extractPrimaryKeysFromExcelFields, extractStateForStream, fieldTypeToJsonSchema, finalizeStateProgressMarkers, findAllMatchingConfigs, flattenSchema, getAllMetrics, getApiEndpoint, getAxiosInstance, getCdnId, getCounterMetric, getCounterMetrics, getDownloadFileApiCall, getDryRunLimit, getJSONApiCall, getJSONApiCallWithFullResponse, getMimeType, getRawJSONApiCall, getServiceAccountKey, gracefulExit, haltAndCatchFire, incrementStreamState, indexToExcelColumn, indexToExcelRow, isDryRun, listFilesRecursively, loadJson, loadSchema, logger, optionalEnv, parseCellReference, postFormDataApiCall, postJSONApiCall, postUrlEncodedApiCall, printMemoryFootprint, processFileStreams, processFromAsyncIterable, removeParasiteProperties, requireEnv, rowGeneratorFromCsv, runWithConcurrency, safePath, unzip, validateExcelConfig, writeDataToCsv, writeGeneratorToCSV };
+export { ALL_STREAM_ID_KEYWORD, API_CALLS_METRIC_NAME, type AssetEntry, type AssetManifest, type AssetManifestEntry, type AssetReplicationMode, AssetStream, AssetTap, AssetTarget, Authenticator, BATCH_INTERVAL_MS, type BaseConfig, type BigQueryConfig, BigQueryTarget, type BigqueryLocation, type Bookmark, type Bookmarks, CdnAssetTarget, type CdnAssetTargetConfig, type CdnFileMetadata, CdnService, type CdnServiceConfig, type CdnUploadResult, CloudStorageService, type CloudStorageServiceOptions, type ColumnMappingStoreUnsafeToSafe, CounterMetric, CsvExtractionConfigBuilder, type CsvFieldsArrayConfig, type CsvFieldsConfig, type CsvFieldsDictConfig, type CsvFileConfig, type CsvStreamConfig, DEFAULT_MAX_CONCURRENT_STREAMS, DEFAULT_SYNCED_AT_COLUMN, DocumentDownloadSkipError, type DocumentEntry, type DocumentManifest, type DocumentManifestEntry, type DocumentStatus, DocumentStream, type DocumentStreamManifest, type DocumentSummary, DocumentTap, type DownloadFilesOptions, EXECUTION_TIME_METRIC_NAME, type ErrorType, type ExcelCustomExtractorConfig, type ExcelDerivedField, type ExcelExtractionBaseConfig, type ExcelExtractionConfig, ExcelExtractionConfigBuilder, type ExcelFieldMapping, type ExcelFieldSpec, type ExcelSingleSheetExtractionConfig, type ExcelSourceColumn, type ExcelStreamConfig, type FieldType, FileFormat, FilePatterns, FileStream, type FileStreamConfig, type FileStreamEntry, FileTap, type FileTapConfig, type FlattenedSchema, GCSStateProvider, type GetConfig, type HTTPHeaders, type HTTPMethod, ITarget, ImageTransform, type InputTapState, type InternalTableField, type JSONSchemaFieldDefinition, type ListFilesOptions, MIME_TYPES, type Message, type MessageType, type MetricConfiguration, MissingFieldInSchemaError, MissingSchemaError, type PostConfig, type ProcessedAsset, type ProgressMarkers, type PropertiesMetadata, RESTStream, ROWS_SYNCED_METRIC_NAME, RecordMessage, RenameColumnStore, ReplicationMethod, ReplicationMethodMessage, type SafeColumnNameConverterFn, type Schema, SchemaMessage, SchemaValidationError, SftpService, type State, type StateHolder, StateMessage, type StateProvider, StateService, Stream, type StreamId, type StreamManifest, type StreamState, StreamWarehouseSyncService, type StreamWithTempFile, type SyncOptions, Tap, type TapState, type TargetSchemaHook, type TargetSchemaHookInput, type TempFile, type URLParams, VariableExtractors, type WarehouseTableField, type WebpTransformOptions, type WhalyDocument, WhalyDocumentService, type WhalyDocumentServiceConfig, WhalyDocumentTarget, type WhalyPaginatedResponse, type WhalyUploadResult, type WriteGeneratorToCsvOptions, _greaterThan, addDocumentSummaries, addWhalyFields, checkCsvHeaderRow, coerceCellValue, collectPaginated, countCsvLines, createCellReference, createCsvStreamConfig, createExcelGenerator, createExcelStreamConfig, createTemporaryFileStream, csvFieldsToJsonSchema, downloadFiles, emptyDocumentSummary, excelColumnToIndex, excelFieldsToJsonSchema, excelRowToIndex, extractExcelRows, extractPrimaryKeysFromCsvConfig, extractPrimaryKeysFromExcelFields, extractStateForStream, fieldTypeToJsonSchema, finalizeStateProgressMarkers, findAllMatchingConfigs, flattenSchema, getAllMetrics, getApiEndpoint, getAxiosInstance, getCdnId, getCounterMetric, getCounterMetrics, getDownloadFileApiCall, getDryRunLimit, getExtension, getJSONApiCall, getJSONApiCallWithFullResponse, getMimeType, getRawJSONApiCall, getServiceAccountKey, gracefulExit, haltAndCatchFire, incrementStreamState, indexToExcelColumn, indexToExcelRow, isDryRun, listFilesRecursively, loadJson, loadSchema, logger, optionalEnv, parseCellReference, postFormDataApiCall, postJSONApiCall, postUrlEncodedApiCall, printMemoryFootprint, processFileStreams, processFromAsyncIterable, removeParasiteProperties, requireEnv, rowGeneratorFromCsv, runWithConcurrency, safePath, unzip, validateExcelConfig, writeDataToCsv, writeGeneratorToCSV };

package/dist/index.d.ts

@@ -147,6 +147,11 @@ declare const MIME_TYPES: Record<string, string>;
  * Returns `application/octet-stream` when the extension is unknown.
  */
 declare function getMimeType(filePathOrExt: string): string;
+/**
+ * Look up the preferred file extension for a MIME type.
+ * Returns `.bin` when the MIME type is unknown.
+ */
+declare function getExtension(mimeType: string): string;
 
 declare enum ReplicationMethod {
     INCREMENTAL = "INCREMENTAL",
@@ -1639,4 +1644,4 @@ declare class GCSStateProvider implements StateProvider {
     writeState(state: string): Promise<void>;
 }
 
-export { ALL_STREAM_ID_KEYWORD, API_CALLS_METRIC_NAME, type AssetEntry, type AssetManifest, type AssetManifestEntry, type AssetReplicationMode, AssetStream, AssetTap, AssetTarget, Authenticator, BATCH_INTERVAL_MS, type BaseConfig, type BigQueryConfig, BigQueryTarget, type BigqueryLocation, type Bookmark, type Bookmarks, CdnAssetTarget, type CdnAssetTargetConfig, type CdnFileMetadata, CdnService, type CdnServiceConfig, type CdnUploadResult, CloudStorageService, type CloudStorageServiceOptions, type ColumnMappingStoreUnsafeToSafe, CounterMetric, CsvExtractionConfigBuilder, type CsvFieldsArrayConfig, type CsvFieldsConfig, type CsvFieldsDictConfig, type CsvFileConfig, type CsvStreamConfig, DEFAULT_MAX_CONCURRENT_STREAMS, DEFAULT_SYNCED_AT_COLUMN, DocumentDownloadSkipError, type DocumentEntry, type DocumentManifest, type DocumentManifestEntry, type DocumentStatus, DocumentStream, type DocumentStreamManifest, type DocumentSummary, DocumentTap, type DownloadFilesOptions, EXECUTION_TIME_METRIC_NAME, type ErrorType, type ExcelCustomExtractorConfig, type ExcelDerivedField, type ExcelExtractionBaseConfig, type ExcelExtractionConfig, ExcelExtractionConfigBuilder, type ExcelFieldMapping, type ExcelFieldSpec, type ExcelSingleSheetExtractionConfig, type ExcelSourceColumn, type ExcelStreamConfig, type FieldType, FileFormat, FilePatterns, FileStream, type FileStreamConfig, type FileStreamEntry, FileTap, type FileTapConfig, type FlattenedSchema, GCSStateProvider, type GetConfig, type HTTPHeaders, type HTTPMethod, ITarget, ImageTransform, type InputTapState, type InternalTableField, type JSONSchemaFieldDefinition, type ListFilesOptions, MIME_TYPES, type Message, type MessageType, type MetricConfiguration, MissingFieldInSchemaError, MissingSchemaError, type PostConfig, type ProcessedAsset, type ProgressMarkers, type PropertiesMetadata, RESTStream, ROWS_SYNCED_METRIC_NAME, RecordMessage, RenameColumnStore, ReplicationMethod, ReplicationMethodMessage, type SafeColumnNameConverterFn, type Schema, SchemaMessage, SchemaValidationError, SftpService, type State, type StateHolder, StateMessage, type StateProvider, StateService, Stream, type StreamId, type StreamManifest, type StreamState, StreamWarehouseSyncService, type StreamWithTempFile, type SyncOptions, Tap, type TapState, type TargetSchemaHook, type TargetSchemaHookInput, type TempFile, type URLParams, VariableExtractors, type WarehouseTableField, type WebpTransformOptions, type WhalyDocument, WhalyDocumentService, type WhalyDocumentServiceConfig, WhalyDocumentTarget, type WhalyPaginatedResponse, type WhalyUploadResult, type WriteGeneratorToCsvOptions, _greaterThan, addDocumentSummaries, addWhalyFields, checkCsvHeaderRow, coerceCellValue, collectPaginated, countCsvLines, createCellReference, createCsvStreamConfig, createExcelGenerator, createExcelStreamConfig, createTemporaryFileStream, csvFieldsToJsonSchema, downloadFiles, emptyDocumentSummary, excelColumnToIndex, excelFieldsToJsonSchema, excelRowToIndex, extractExcelRows, extractPrimaryKeysFromCsvConfig, extractPrimaryKeysFromExcelFields, extractStateForStream, fieldTypeToJsonSchema, finalizeStateProgressMarkers, findAllMatchingConfigs, flattenSchema, getAllMetrics, getApiEndpoint, getAxiosInstance, getCdnId, getCounterMetric, getCounterMetrics, getDownloadFileApiCall, getDryRunLimit, getJSONApiCall, getJSONApiCallWithFullResponse, getMimeType, getRawJSONApiCall, getServiceAccountKey, gracefulExit, haltAndCatchFire, incrementStreamState, indexToExcelColumn, indexToExcelRow, isDryRun, listFilesRecursively, loadJson, loadSchema, logger, optionalEnv, parseCellReference, postFormDataApiCall, postJSONApiCall, postUrlEncodedApiCall, printMemoryFootprint, processFileStreams, processFromAsyncIterable, removeParasiteProperties, requireEnv, rowGeneratorFromCsv, runWithConcurrency, safePath, unzip, validateExcelConfig, writeDataToCsv, writeGeneratorToCSV };
+export { ALL_STREAM_ID_KEYWORD, API_CALLS_METRIC_NAME, type AssetEntry, type AssetManifest, type AssetManifestEntry, type AssetReplicationMode, AssetStream, AssetTap, AssetTarget, Authenticator, BATCH_INTERVAL_MS, type BaseConfig, type BigQueryConfig, BigQueryTarget, type BigqueryLocation, type Bookmark, type Bookmarks, CdnAssetTarget, type CdnAssetTargetConfig, type CdnFileMetadata, CdnService, type CdnServiceConfig, type CdnUploadResult, CloudStorageService, type CloudStorageServiceOptions, type ColumnMappingStoreUnsafeToSafe, CounterMetric, CsvExtractionConfigBuilder, type CsvFieldsArrayConfig, type CsvFieldsConfig, type CsvFieldsDictConfig, type CsvFileConfig, type CsvStreamConfig, DEFAULT_MAX_CONCURRENT_STREAMS, DEFAULT_SYNCED_AT_COLUMN, DocumentDownloadSkipError, type DocumentEntry, type DocumentManifest, type DocumentManifestEntry, type DocumentStatus, DocumentStream, type DocumentStreamManifest, type DocumentSummary, DocumentTap, type DownloadFilesOptions, EXECUTION_TIME_METRIC_NAME, type ErrorType, type ExcelCustomExtractorConfig, type ExcelDerivedField, type ExcelExtractionBaseConfig, type ExcelExtractionConfig, ExcelExtractionConfigBuilder, type ExcelFieldMapping, type ExcelFieldSpec, type ExcelSingleSheetExtractionConfig, type ExcelSourceColumn, type ExcelStreamConfig, type FieldType, FileFormat, FilePatterns, FileStream, type FileStreamConfig, type FileStreamEntry, FileTap, type FileTapConfig, type FlattenedSchema, GCSStateProvider, type GetConfig, type HTTPHeaders, type HTTPMethod, ITarget, ImageTransform, type InputTapState, type InternalTableField, type JSONSchemaFieldDefinition, type ListFilesOptions, MIME_TYPES, type Message, type MessageType, type MetricConfiguration, MissingFieldInSchemaError, MissingSchemaError, type PostConfig, type ProcessedAsset, type ProgressMarkers, type PropertiesMetadata, RESTStream, ROWS_SYNCED_METRIC_NAME, RecordMessage, RenameColumnStore, ReplicationMethod, ReplicationMethodMessage, type SafeColumnNameConverterFn, type Schema, SchemaMessage, SchemaValidationError, SftpService, type State, type StateHolder, StateMessage, type StateProvider, StateService, Stream, type StreamId, type StreamManifest, type StreamState, StreamWarehouseSyncService, type StreamWithTempFile, type SyncOptions, Tap, type TapState, type TargetSchemaHook, type TargetSchemaHookInput, type TempFile, type URLParams, VariableExtractors, type WarehouseTableField, type WebpTransformOptions, type WhalyDocument, WhalyDocumentService, type WhalyDocumentServiceConfig, WhalyDocumentTarget, type WhalyPaginatedResponse, type WhalyUploadResult, type WriteGeneratorToCsvOptions, _greaterThan, addDocumentSummaries, addWhalyFields, checkCsvHeaderRow, coerceCellValue, collectPaginated, countCsvLines, createCellReference, createCsvStreamConfig, createExcelGenerator, createExcelStreamConfig, createTemporaryFileStream, csvFieldsToJsonSchema, downloadFiles, emptyDocumentSummary, excelColumnToIndex, excelFieldsToJsonSchema, excelRowToIndex, extractExcelRows, extractPrimaryKeysFromCsvConfig, extractPrimaryKeysFromExcelFields, extractStateForStream, fieldTypeToJsonSchema, finalizeStateProgressMarkers, findAllMatchingConfigs, flattenSchema, getAllMetrics, getApiEndpoint, getAxiosInstance, getCdnId, getCounterMetric, getCounterMetrics, getDownloadFileApiCall, getDryRunLimit, getExtension, getJSONApiCall, getJSONApiCallWithFullResponse, getMimeType, getRawJSONApiCall, getServiceAccountKey, gracefulExit, haltAndCatchFire, incrementStreamState, indexToExcelColumn, indexToExcelRow, isDryRun, listFilesRecursively, loadJson, loadSchema, logger, optionalEnv, parseCellReference, postFormDataApiCall, postJSONApiCall, postUrlEncodedApiCall, printMemoryFootprint, processFileStreams, processFromAsyncIterable, removeParasiteProperties, requireEnv, rowGeneratorFromCsv, runWithConcurrency, safePath, unzip, validateExcelConfig, writeDataToCsv, writeGeneratorToCSV };

package/dist/index.js

@@ -10397,6 +10397,7 @@ __export(index_exports, {
   getCounterMetrics: () => getCounterMetrics,
   getDownloadFileApiCall: () => getDownloadFileApiCall,
   getDryRunLimit: () => getDryRunLimit,
+  getExtension: () => getExtension,
   getJSONApiCall: () => getJSONApiCall,
   getJSONApiCallWithFullResponse: () => getJSONApiCallWithFullResponse,
   getMimeType: () => getMimeType,
@@ -10973,6 +10974,16 @@ function getMimeType(filePathOrExt) {
   const ext = filePathOrExt.startsWith(".") ? filePathOrExt.toLowerCase() : import_node_path.default.extname(filePathOrExt).toLowerCase();
   return MIME_TYPES[ext] ?? "application/octet-stream";
 }
+var EXTENSION_BY_MIME = Object.entries(MIME_TYPES).reduce(
+  (acc, [ext, mime]) => {
+    if (!acc[mime]) acc[mime] = ext;
+    return acc;
+  },
+  {}
+);
+function getExtension(mimeType) {
+  return EXTENSION_BY_MIME[mimeType] ?? ".bin";
+}
 
 // src/sdk/models/replication.ts
 var ReplicationMethod = /* @__PURE__ */ ((ReplicationMethod2) => {
@@ -14174,6 +14185,13 @@ var AssetTap = class {
               const inspectPath = import_node_path4.default.join(this.outputDir, stream.streamId, entry.destinationPath);
               await import_fs_extra4.default.ensureDir(import_node_path4.default.dirname(inspectPath));
               await import_fs_extra4.default.copy(uploadPath, inspectPath);
+              if (wasTransformed) {
+                const baseName = import_node_path4.default.basename(entry.destinationPath, import_node_path4.default.extname(entry.destinationPath));
+                const origExt = getExtension(entry.contentType);
+                const originalPath = import_node_path4.default.join(this.outputDir, "originals", `${baseName}${origExt}`);
+                await import_fs_extra4.default.ensureDir(import_node_path4.default.dirname(originalPath));
+                await import_fs_extra4.default.copy(downloadedPath, originalPath);
+              }
             }
             assetEntries.push({
               sourcePath: entry.sourcePath,
@@ -15594,6 +15612,7 @@ var GCSStateProvider = class {
   getCounterMetrics,
   getDownloadFileApiCall,
   getDryRunLimit,
+  getExtension,
   getJSONApiCall,
   getJSONApiCallWithFullResponse,
   getMimeType,