parze 0.2.3 → 0.2.5

package/README.md

@@ -34,6 +34,19 @@ const extraction2 = await client.extract(parseResult.text, {
   date: { type: 'string', description: 'Invoice date' }
 }, parseResult.job_id);
 
+// Validate document quality (pre-validation)
+const validation = await client.validate('invoice.pdf', {
+  validationType: 'pre',
+  validationRules: {
+    quality_checks: {
+      min_resolution: 150,
+      check_readability: true,
+      check_completeness: true
+    }
+  }
+});
+console.log(validation);
+
 // Get AI-suggested schema
 const suggested = await client.suggestSchema(parseResult.text);
 console.log(suggested);
@@ -49,7 +62,6 @@ Parse a document into structured text.
 - `options` (object, optional):
   - `outputFormat`: "structured" | "markdown" | "json"
   - `preserveTables`: boolean
-  - `preserveLayout`: boolean
   - `extractionMode`: "auto" | "ocr_only" | "llm_only" | "identity_doc"
 
 **Returns:** Promise with parsed text and metadata
@@ -70,11 +82,23 @@ Extract structured data from a file. Parse runs internally.
 - `extractionSchema` (object): Schema defining fields to extract
 - `options` (object, optional):
   - `preserveTables`: boolean
-  - `preserveLayout`: boolean
   - `extractionMode`: "auto" | "ocr_only" | "llm_only" | "identity_doc"
 
 **Returns:** Promise with extracted data and confidence scores
 
+### `validate(filePath(s), options?)`
+Validate document quality (pre) or extracted data (post).
+
+**Parameters:**
+- `filePath(s)` (string | string[]): Path(s) to document file(s)
+- `options` (object, optional):
+  - `validationType`: "pre" | "post"
+  - `validationRules`: object
+  - `extractionSchema`: object (required for post)
+  - `jobId`: string
+
+**Returns:** Promise with validation results
+
 ### `suggestSchema(text)`
 Get AI-suggested extraction schema based on document text.
 

package/dist/index.d.ts

@@ -5,14 +5,18 @@ export interface ParzeClientOptions {
 export interface ParseOptions {
     outputFormat?: 'structured' | 'markdown' | 'json';
     preserveTables?: boolean;
-    preserveLayout?: boolean;
     extractionMode?: 'auto' | 'ocr_only' | 'llm_only' | 'identity_doc';
 }
 export interface ExtractFileOptions {
     preserveTables?: boolean;
-    preserveLayout?: boolean;
     extractionMode?: 'auto' | 'ocr_only' | 'llm_only' | 'identity_doc';
 }
+export interface ValidateOptions {
+    validationType?: 'pre' | 'post';
+    validationRules?: object;
+    extractionSchema?: object;
+    jobId?: string;
+}
 export declare class ParzeClient {
     private apiKey;
     private baseUrl;
@@ -21,6 +25,7 @@ export declare class ParzeClient {
     parse(filePath: string, options?: ParseOptions): Promise<any>;
     extract(text: string, extractionSchema: object, jobId: string): Promise<any>;
     extractFile(filePath: string, extractionSchema: object, options?: ExtractFileOptions): Promise<any>;
+    validate(filePaths: string | string[], options?: ValidateOptions): Promise<any>;
     suggestSchema(text: string): Promise<any>;
     textToSchema(description: string): Promise<any>;
 }

package/dist/index.js

@@ -58,8 +58,6 @@ class ParzeClient {
             form.append('output_format', options.outputFormat);
         if ((options === null || options === void 0 ? void 0 : options.preserveTables) !== undefined)
             form.append('preserve_tables', String(options.preserveTables));
-        if ((options === null || options === void 0 ? void 0 : options.preserveLayout) !== undefined)
-            form.append('preserve_layout', String(options.preserveLayout));
         if (options === null || options === void 0 ? void 0 : options.extractionMode)
             form.append('extraction_mode', options.extractionMode);
         const response = await this.http.post('/parse', form, {
@@ -81,8 +79,6 @@ class ParzeClient {
         form.append('extraction_schema', JSON.stringify(extractionSchema));
         if ((options === null || options === void 0 ? void 0 : options.preserveTables) !== undefined)
             form.append('preserve_tables', String(options.preserveTables));
-        if ((options === null || options === void 0 ? void 0 : options.preserveLayout) !== undefined)
-            form.append('preserve_layout', String(options.preserveLayout));
         if (options === null || options === void 0 ? void 0 : options.extractionMode)
             form.append('extraction_mode', options.extractionMode);
         const response = await this.http.post('/extract', form, {
@@ -90,6 +86,25 @@ class ParzeClient {
         });
         return response.data;
     }
+    async validate(filePaths, options) {
+        const form = new form_data_1.default();
+        const paths = Array.isArray(filePaths) ? filePaths : [filePaths];
+        paths.forEach((path) => {
+            form.append('files', fs.createReadStream(path));
+        });
+        if (options === null || options === void 0 ? void 0 : options.validationType)
+            form.append('validation_type', options.validationType);
+        if (options === null || options === void 0 ? void 0 : options.validationRules)
+            form.append('validation_rules', JSON.stringify(options.validationRules));
+        if (options === null || options === void 0 ? void 0 : options.extractionSchema)
+            form.append('extraction_schema', JSON.stringify(options.extractionSchema));
+        if (options === null || options === void 0 ? void 0 : options.jobId)
+            form.append('job_id', options.jobId);
+        const response = await this.http.post('/validate', form, {
+            headers: form.getHeaders(),
+        });
+        return response.data;
+    }
     async suggestSchema(text) {
         const response = await this.http.post('/suggest-schema', { text });
         return response.data;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "parze",
-  "version": "0.2.3",
+  "version": "0.2.5",
   "description": "TypeScript SDK for the Parze API",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",