SK-2019 implement de identify files public interfaces for detect node

src/error/codes/index.ts

@@ -196,7 +196,29 @@ const SKYFLOW_ERROR_CODE = {
     INVALID_RESTRICT_REGEX_LIST: { http_code: 400, message: errorMessages.INVALID_RESTRICT_REGEX_LIST },
     INVALID_TOKEN_FORMAT: { http_code: 400, message: errorMessages.INVALID_TOKEN_FORMAT },
     INVALID_TRANSFORMATIONS: { http_code: 400, message: errorMessages.INVALID_TRANSFORMATIONS },
-
+    INVALID_DEIDENTIFY_FILE_REQUEST: { http_code: 400, message: errorMessages.INVALID_DEIDENTIFY_FILE_REQUEST },
+    EMPTY_FILE_OBJECT:{ http_code: 400, message: errorMessages.EMPTY_FILE_OBJECT },
+    INVALID_FILE_FORMAT: { http_code: 400, message: errorMessages.INVALID_FILE_FORMAT },
+    MISSING_FILE_SOURCE: { http_code: 400, message: errorMessages.MISSING_FILE_SOURCE },
+    INVALID_BASE64_STRING: { http_code: 400, message: errorMessages.INVALID_BASE64_STRING },
+    INVALID_DEIDENTIFY_FILE_OPTIONS: { http_code: 400, message: errorMessages.INVALID_DEIDENTIFY_FILE_OPTIONS },
+    INVALID_ENTITIES: { http_code: 400, message: errorMessages.INVALID_ENTITIES },
+    // INVALID_ALLOW_REGEX_LIST: { http_code: 400, message: errorMessages.INVALID_ALLOW_REGEX_LIST },
+    // INVALID_RESTRICT_REGEX_LIST: { http_code: 400, message: errorMessages.INVALID_RESTRICT_REGEX_LIST },
+    // INVALID_TOKEN_FORMAT: { http_code: 400, message: errorMessages.INVALID_TOKEN_FORMAT },
+    // INVALID_TRANSFORMATIONS: { http_code: 400, message: errorMessages.INVALID_TRANSFORMATIONS },
+    INVALID_OUTPUT_PROCESSED_IMAGE: { http_code: 400, message: errorMessages.INVALID_OUTPUT_PROCESSED_IMAGE },
+    INVALID_OUTPUT_OCR_TEXT: { http_code: 400, message: errorMessages.INVALID_OUTPUT_OCR_TEXT },
+    INVALID_MASKING_METHOD: { http_code: 400, message: errorMessages.INVALID_MASKING_METHOD },
+    INVALID_PIXEL_DENSITY: { http_code: 400, message: errorMessages.INVALID_PIXEL_DENSITY },
+    INVALID_MAX_RESOLUTION: { http_code: 400, message: errorMessages.INVALID_MAX_RESOLUTION },
+    INVALID_OUTPUT_PROCESSED_AUDIO: { http_code: 400, message: errorMessages.INVALID_OUTPUT_PROCESSED_AUDIO },
+    INVALID_OUTPUT_TRANSCRIPTION: { http_code: 400, message: errorMessages.INVALID_OUTPUT_TRANSCRIPTION },
+    INVALID_BLEEP:{ http_code: 400, message: errorMessages.INVALID_BLEEP },
+    INVALID_FILE_OR_ENCODED_FILE:{ http_code: 400, message: errorMessages.INVALID_FILE_OR_ENCODED_FILE },
+    INVALID_FILE_TYPE:{ http_code: 400, message: errorMessages.INVALID_FILE_TYPE },
+    FILE_READ_ERROR:{ http_code: 400, message: errorMessages.FILE_READ_ERROR },
+    INVALID_BASE64_HEADER:{ http_code: 400, message: errorMessages.INVALID_BASE64_HEADER },
     INVALID_TEXT_IN_REIDENTIFY: { http_code: 400, message: errorMessages.INVALID_TEXT_IN_REIDENTIFY },
     INVALID_REDACTED_ENTITIES_IN_REIDENTIFY: { http_code: 400, message: errorMessages.INVALID_REDACTED_ENTITIES_IN_REIDENTIFY },
     INVALID_MASKED_ENTITIES_IN_REIDENTIFY: { http_code: 400, message: errorMessages.INVALID_MASKED_ENTITIES_IN_REIDENTIFY },

src/error/messages/index.ts

@@ -201,6 +201,36 @@ const errorMessages = {
     INVALID_RESTRICT_REGEX_LIST: `${errorPrefix} Validation error. The restrictRegexList field must be an array of strings. Specify a valid restrictRegexList.`,
     INVALID_TOKEN_FORMAT: `${errorPrefix} Validation error. The tokenFormat key must be an instance of TokenFormat. Specify a valid token format.`,
     INVALID_TRANSFORMATIONS: `${errorPrefix} Validation error. The transformations key must be an instance of Transformations. Specify a valid transformations.`,
+    INVALID_DEIDENTIFY_FILE_REQUEST: `${errorPrefix} Validation error. Invalid deidentify file request. Specify a valid deidentify file request.`,
+    EMPTY_FILE_OBJECT: `${errorPrefix} Validation error. File object cannot be empty. Specify a valid file object.`,
+    INVALID_FILE_FORMAT: `${errorPrefix} Validation error. Invalid file format. Specify a valid file format.`,
+    MISSING_FILE_SOURCE: `${errorPrefix} Validation error. Provide exactly one of filePath, base64, or fileObject.`,
+    INVALID_FILE_OBJECT: `${errorPrefix} Validation error. Invalid file object. Specify a valid file object.`,
+    INVALID_BASE64_STRING: `${errorPrefix} Validation error. Invalid base64 string. Specify a valid base64 string.`,
+    INVALID_DEIDENTIFY_FILE_OPTIONS: `${errorPrefix} Validation error. Invalid deidentify file options. Specify a valid deidentify file options.`,
+    INVALID_ENTITIES: `${errorPrefix} Validation error. Invalid entities. Specify valid entities as string array.`,
+    EMPTY_ENTITIES: `${errorPrefix} Validation error. Entities cannot be empty. Specify valid entities.`,
+    // INVALID_ALLOW_REGEX_LIST: `${errorPrefix} Validation error. Invalid allow regex list. Specify valid allow regex list as string array.`,
+    EMPTY_ALLOW_REGEX_LIST: `${errorPrefix} Validation error. Allow regex list cannot be empty. Specify valid allow regex list.`,
+    INVALID_ALLOW_REGEX: `${errorPrefix} Validation error. Invalid allow regex. Specify valid allow regex at index %s1.`,
+    // INVALID_RESTRICT_REGEX_LIST: `${errorPrefix} Validation error. Invalid restrict regex list. Specify valid restrict regex list as string array.`,
+    EMPTY_RESTRICT_REGEX_LIST: `${errorPrefix} Validation error. Restrict regex list cannot be empty. Specify valid restrict regex list.`,
+    INVALID_RESTRICT_REGEX: `${errorPrefix} Validation error. Invalid restrict regex. Specify valid restrict regex at index %s1.`,
+    // INVALID_TOKEN_FORMAT: `${errorPrefix} Validation error. Invalid token format. Specify valid token format as string.`,
+    // INVALID_TRANSFORMATIONS: `${errorPrefix} Validation error. Invalid transformations. Specify valid transformations as string array.`,
+    INVALID_OUTPUT_PROCESSED_IMAGE: `${errorPrefix} Validation error. Invalid output processed image. Specify valid output processed image as string.`,
+    INVALID_OUTPUT_OCR_TEXT: `${errorPrefix} Validation error. Invalid output ocr text. Specify valid output ocr text as string.`,
+    INVALID_MASKING_METHOD: `${errorPrefix} Validation error. Invalid masking method. Specify valid masking method as string.`,
+    INVALID_PIXEL_DENSITY: `${errorPrefix} Validation error. Invalid pixel density. Specify valid pixel density as string.`,
+    INVALID_OUTPUT_TRANSCRIPTION: `${errorPrefix} Validation error. Invalid output transcription. Specify valid output transcription as string.`,
+    INVALID_OUTPUT_PROCESSED_AUDIO: `${errorPrefix} Validation error. Invalid output processed audio. Specify valid output processed audio as string.`,
+    INVALID_MAX_RESOLUTION: `${errorPrefix} Validation error. Invalid max resolution. Specify valid max resolution as string.`,
+    INVALID_BLEEP: `${errorPrefix} Validation error. Invalid bleep. Specify valid bleep as object.`,
+    INVALID_FILE_OR_ENCODED_FILE: `${errorPrefix} . Error while decoding base64 and saving file`,
+    INVALID_FILE_TYPE : `${errorPrefix} Validation error. Invalid file type. Specify a valid file type.`,
+    INVALID_FILE_NAME: `${errorPrefix} Validation error. Invalid file name. Specify a valid file name.`,
+    FILE_READ_ERROR: `${errorPrefix} Validation error. Unable to read file. Verify the file path.`, 
+    INVALID_BASE64_HEADER: `${errorPrefix} Validation error. Invalid base64 header. Specify a valid base64 header.`,
 
     INVALID_TEXT_IN_REIDENTIFY: `${errorPrefix} Validation error. The text field is required and must be a non-empty string. Specify a valid text.`,
     INVALID_REDACTED_ENTITIES_IN_REIDENTIFY: `${errorPrefix} Validation error. The redactedEntities field must be an array of DetectEntities enums. Specify a valid redactedEntities.`,

src/index.ts

@@ -36,6 +36,9 @@ import DeidentifyTextOptions from './vault/model/options/deidentify-text';
 import DeidentifyTextResponse from './vault/model/response/deidentify-text';
 import ReidentifyTextRequest from './vault/model/request/reidentify-text';
 import ReidentifyTextOptions from './vault/model/options/reidentify-text';
+import DeidentifyFileOptions from './vault/model/options/deidentify-file';
+import DeidentifyFileRequest from './vault/model/request/deidentify-file';
+import DeidentifyFileResponse from './vault/model/response/deidentify-file';
 import ReidentifyTextResponse from './vault/model/response/reidentify-text';
 import TokenFormat from './vault/model/options/deidentify-text/token-format';
 import Transformations from './vault/model/options/deidentify-text/transformations';
@@ -93,6 +96,9 @@ export {
     DeidentifyTextResponse,
     ReidentifyTextRequest,
     ReidentifyTextOptions,
+    DeidentifyFileOptions,
+    DeidentifyFileRequest,
+    DeidentifyFileResponse,
     ReidentifyTextResponse,
     TokenFormat,
     Transformations

src/utils/index.ts

@@ -99,8 +99,31 @@ export const TYPES = {
     DETECT: 'DETECT',
     INVOKE_CONNECTION: 'INVOKE_CONNECTION',
     DEIDENTIFY_TEXT: 'DEIDENTIFY_TEXT',
+    DEIDENTIFY_FILE: 'DEIDENTIFY_FILE',
 };
 
+export enum DeidenitfyFileRequestTypes {
+    IMAGE= "IMAGE",
+    FILE= 'FILE', 
+    AUDIO= 'AUDIO', 
+    PPT= 'PPT', 
+    DOCUMENT= 'DOCUMENT',
+    PDF= 'PDF', 
+    SPREADSHEET= 'SPREADSHEET', 
+    STRUCTURED_TEXT= 'STRUCTURED_TEXT',
+    TEXT='TEXT'
+}
+export enum DetectTokenTypeWithoutVault {
+    ENTITY_ONLY= "entity_only",
+    ENTITY_UNQ_COUNTER= "entity_unq_counter",
+}
+
+export enum DetectOutputTranscription {
+    DIARIZED_TRANSCRIPTION = "diarized_transcription",
+    MEDICAL_DIARIZED_TRANSCRIPTION = "medical_diarized_transcription",
+    MEDICAL_TRANSCRIPTION = "medical_transcription",
+    PLAINTEXT_TRANSCRIPTION = "plaintext_transcription",
+    TRANSCRIPTION = "transcription",
 export const CONTROLLER_TYPES = {
     DETECT: 'DETECT',
     VAULT: 'VAULT',
@@ -173,6 +196,12 @@ export enum DetectEntities {
     ZODIAC_SIGN = 'zodiac_sign',
 }
 
+export enum MaskingMethod{
+    Blackout= "blackout",
+    Blur= "blur",
+}
+
+
 export interface ISkyflowError {
     http_status?: string | number | null,
     grpc_code?: string | number | null,

src/utils/logs/index.ts

@@ -17,6 +17,7 @@ const logs = {
         CURRENT_LOG_LEVEL: 'Current log level is %s1.',
         CLIENT_INITIALIZED: 'Initialized skyflow client successfully.',
         VALIDATE_INSERT_INPUT: 'Validating insert input.',
+        VALIDATE_DETECT_FILE_INPUT: 'Validating deidentify file input.',
         VALIDATE_DETOKENIZE_INPUT: 'Validating detokenize input.',
         VALIDATE_TOKENIZE_INPUT: 'Validating tokenize input.',
         VALIDATE_FILE_UPLOAD_INPUT: 'Validating detokenize input.',
@@ -39,6 +40,7 @@ const logs = {
         REUSE_API_KEY: 'Reusing api key.',
         CONTROLLER_INITIALIZED: 'SkyflowController initialized.',
         INSERT_TRIGGERED: 'Insert method triggered.',
+        DETECT_FILE_TRIGGERED: 'Deidentify file triggered.',
         DETOKENIZE_TRIGGERED: 'Detokenize method triggered.',
         TOKENIZE_TRIGGERED: 'Tokenize method triggered.',
         GET_BY_ID_TRIGGERED: 'Get by ID triggered.',

src/utils/validations/index.ts

@@ -28,6 +28,9 @@ import DeidentifyTextRequest from "../../vault/model/request/deidentify-text";
 import DeidentifyTextOptions from "../../vault/model/options/deidentify-text";
 import TokenFormat from "../../vault/model/options/deidentify-text/token-format";
 import Transformations from "../../vault/model/options/deidentify-text/transformations";
+import DeidentifyFileOptions from "../../vault/model/options/deidentify-file";
+import DeidentifyFileRequest from "../../vault/model/request/deidentify-file";
+
 import ReidentifyTextRequest from "../../vault/model/request/reidentify-text";
 import ReidentifyTextOptions from "../../vault/model/options/reidentify-text";
 
@@ -1037,4 +1040,96 @@ export const validateInvokeConnectionRequest = (invokeRequest: InvokeConnectionR
     } else {
         throw new SkyflowError(SKYFLOW_ERROR_CODE.INVALID_INVOKE_CONNECTION_REQUEST);
     }
-};
+};
+
+export const validateDeidentifyFileRequest = (deidentifyFileRequest: DeidentifyFileRequest, deidentifyFileOptions?: DeidentifyFileOptions, logLevel: LogLevel = LogLevel.ERROR) => {
+    if (!deidentifyFileRequest) {
+        throw new SkyflowError(SKYFLOW_ERROR_CODE.INVALID_DEIDENTIFY_FILE_REQUEST);
+    }
+
+    // Validate file object
+    const file = deidentifyFileRequest.getFile();
+
+    // Check if the file is a valid File (browser) or Buffer (Node.js)
+    if (!(file instanceof File)) {
+        throw new SkyflowError(SKYFLOW_ERROR_CODE.INVALID_FILE_TYPE);
+    }
+
+    // Additional validation for File (browser)
+    if (file instanceof File) {
+        if (!file.name || !file.size) {
+            throw new SkyflowError(SKYFLOW_ERROR_CODE.INVALID_FILE_TYPE);
+        }
+    }
+
+    // Validate options if provided
+    if (deidentifyFileOptions) {
+        validateDeidentifyFileOptions(deidentifyFileOptions);
+    }
+};
+
+export const validateDeidentifyFileOptions = (deidentifyFileOptions: DeidentifyFileOptions) => {
+    if (!deidentifyFileOptions) {
+        throw new SkyflowError(SKYFLOW_ERROR_CODE.INVALID_DEIDENTIFY_FILE_OPTIONS);
+    }
+
+    // Validate entities
+    if (deidentifyFileOptions.getEntities() !== undefined && (!Array.isArray(deidentifyFileOptions.getEntities()))) {
+        throw new SkyflowError(SKYFLOW_ERROR_CODE.INVALID_ENTITIES);
+    }
+
+    // Validate allowRegexList
+    if (deidentifyFileOptions.getAllowRegexList() && !Array.isArray(deidentifyFileOptions.getAllowRegexList())) {
+        throw new SkyflowError(SKYFLOW_ERROR_CODE.INVALID_ALLOW_REGEX_LIST);
+    }
+
+    // Validate restrictRegexList
+    if (deidentifyFileOptions.getRestrictRegexList() && !Array.isArray(deidentifyFileOptions.getRestrictRegexList())) {
+        throw new SkyflowError(SKYFLOW_ERROR_CODE.INVALID_RESTRICT_REGEX_LIST);
+    }
+
+    // Validate tokenFormat
+    if (deidentifyFileOptions.getTokenFormat() && typeof deidentifyFileOptions.getTokenFormat !== 'object') {
+        throw new SkyflowError(SKYFLOW_ERROR_CODE.INVALID_TOKEN_FORMAT);
+    }
+
+    // Validate transformations
+    if (deidentifyFileOptions.getTransformations() && typeof deidentifyFileOptions.getTransformations() !== 'object') {
+        throw new SkyflowError(SKYFLOW_ERROR_CODE.INVALID_TRANSFORMATIONS);
+    }
+
+    // Validate image-specific options
+    if (deidentifyFileOptions.getOutputProcessedImage() !== undefined && typeof deidentifyFileOptions.getOutputProcessedImage() !== 'boolean') {
+        throw new SkyflowError(SKYFLOW_ERROR_CODE.INVALID_OUTPUT_PROCESSED_IMAGE);
+    }
+
+    if (deidentifyFileOptions.getOutputOcrText() !== undefined && typeof deidentifyFileOptions.getOutputOcrText() !== 'boolean') {
+        throw new SkyflowError(SKYFLOW_ERROR_CODE.INVALID_OUTPUT_OCR_TEXT);
+    }
+
+    if (deidentifyFileOptions.getMaskingMethod() !== undefined && typeof deidentifyFileOptions.getMaskingMethod() !== 'string') {
+        throw new SkyflowError(SKYFLOW_ERROR_CODE.INVALID_MASKING_METHOD);
+    }
+
+    // Validate PDF-specific options
+    if (deidentifyFileOptions.getPixelDensity() !== undefined && typeof deidentifyFileOptions.getPixelDensity() !== 'number') {
+        throw new SkyflowError(SKYFLOW_ERROR_CODE.INVALID_PIXEL_DENSITY);
+    }
+
+    if (deidentifyFileOptions.getMaxResolution() !== undefined && typeof deidentifyFileOptions.getMaxResolution() !== 'number') {
+        throw new SkyflowError(SKYFLOW_ERROR_CODE.INVALID_MAX_RESOLUTION);
+    }
+
+    // Validate audio-specific options
+    if (deidentifyFileOptions.getOutputProcessedAudio() !== undefined && typeof deidentifyFileOptions.getOutputProcessedAudio() !== 'boolean') {
+        throw new SkyflowError(SKYFLOW_ERROR_CODE.INVALID_OUTPUT_PROCESSED_AUDIO);
+    }
+
+    if (deidentifyFileOptions.getOutputTranscription() !== undefined && typeof deidentifyFileOptions.getOutputTranscription() !== 'string') {
+        throw new SkyflowError(SKYFLOW_ERROR_CODE.INVALID_OUTPUT_TRANSCRIPTION);
+    }
+
+    if (deidentifyFileOptions.getBleep() !== undefined && typeof deidentifyFileOptions.getBleep() !== 'object') {
+        throw new SkyflowError(SKYFLOW_ERROR_CODE.INVALID_BLEEP);
+    }
+};

src/vault/client/index.ts

@@ -1,4 +1,5 @@
 // imports 
+import { Files } from "../../ _generated_/rest/api/resources/files/client/Client";
 import { Query } from "../../ _generated_/rest/api/resources/query/client/Client";
 import { Records } from "../../ _generated_/rest/api/resources/records/client/Client";
 import { Strings } from "../../ _generated_/rest/api/resources/strings/client/Client";
@@ -26,6 +27,8 @@ class VaultClient {
 
     stringsAPI!: Strings;
 
+    filesAPI!: Files;
+
     individualCredentials?: Credentials;
 
     skyflowCredentials?: Credentials;
@@ -82,6 +85,10 @@ class VaultClient {
                 break;
             case TYPES.DETECT:
                 this.stringsAPI = new Strings(this.configuration);
+                break
+            case TYPES.DEIDENTIFY_FILE:
+                this.filesAPI = new Files(this.configuration);  
+                break;  
             default:
                 break;
         }