quidproquo-actionprocessor-awslambda 0.0.256 → 0.0.258

package/lib/esm/logic/textract/transformExpenseResponse.js

@@ -0,0 +1,173 @@
+const FIELD_TYPE_MAPPINGS = {
+    'VENDOR_NAME': 'merchantName',
+    'NAME': 'merchantName',
+    'MERCHANT_NAME': 'merchantName',
+    'VENDOR_ADDRESS': 'merchantAddress',
+    'ADDRESS': 'merchantAddress',
+    'MERCHANT_ADDRESS': 'merchantAddress',
+    'INVOICE_RECEIPT_DATE': 'date',
+    'DATE': 'date',
+    'CURRENCY_CODE': 'currency',
+    'PAYMENT_METHOD': 'paymentMethod',
+    'SUBTOTAL': 'subtotal',
+    'TAX': 'tax',
+    'TOTAL': 'total',
+    'AMOUNT_DUE': 'total',
+    'AMOUNT_PAID': 'amountPaid',
+    'INVOICE_RECEIPT_ID': 'receiptNumber',
+    'INVOICE_ID': 'invoiceNumber',
+    'RECEIPT_ID': 'receiptNumber',
+    'TAX_PAYER_ID': 'taxId',
+    'VENDOR_ABN_NUMBER': 'vendorAbn',
+    'VENDOR_PHONE': 'vendorPhone',
+    'VENDOR_URL': 'vendorUrl',
+};
+const LINE_ITEM_FIELD_MAPPINGS = {
+    'ITEM': 'description',
+    'DESCRIPTION': 'description',
+    'PRODUCT_NAME': 'description',
+    'QUANTITY': 'quantity',
+    'UNIT_PRICE': 'unitPrice',
+    'PRICE': 'total',
+    'AMOUNT': 'total',
+};
+function extractFieldValue(field, forceString = false) {
+    const text = field?.ValueDetection?.Text || field?.Type?.Text;
+    if (!text)
+        return undefined;
+    // For certain fields, always return as string (dates, IDs, etc.)
+    if (forceString)
+        return text;
+    // Only try to parse as number if it looks like a price/amount (has $ or is purely numeric)
+    const cleanedText = text.replace(/[$,]/g, '').trim();
+    // Check if it's a pure number or price
+    if (/^\d+\.?\d*$/.test(cleanedText) && !text.includes('/') && !text.includes('-')) {
+        const numericValue = parseFloat(cleanedText);
+        if (!isNaN(numericValue))
+            return numericValue;
+    }
+    return text;
+}
+function extractSummaryFields(summaryFields) {
+    const metadata = {};
+    if (!summaryFields)
+        return metadata;
+    // Fields that should always be strings
+    const stringFields = new Set([
+        'DATE', 'INVOICE_RECEIPT_DATE', 'CURRENCY_CODE', 'PAYMENT_METHOD',
+        'INVOICE_RECEIPT_ID', 'INVOICE_ID', 'RECEIPT_ID', 'TAX_PAYER_ID',
+        'VENDOR_ABN_NUMBER', 'VENDOR_PHONE', 'VENDOR_URL'
+    ]);
+    // Track confidence scores for duplicate fields
+    const fieldConfidence = {};
+    for (const field of summaryFields) {
+        const fieldType = field?.Type?.Text?.toUpperCase();
+        if (!fieldType)
+            continue;
+        const confidence = field?.Type?.Confidence || 0;
+        const forceString = stringFields.has(fieldType) || fieldType.includes('DATE') || fieldType.includes('ID') || fieldType.includes('NUMBER') || fieldType.includes('PHONE') || fieldType.includes('URL');
+        if (FIELD_TYPE_MAPPINGS[fieldType]) {
+            const mappedKey = FIELD_TYPE_MAPPINGS[fieldType];
+            const value = extractFieldValue(field, forceString);
+            // Only update if value exists and has higher confidence than existing
+            if (value !== undefined) {
+                const existingConfidence = fieldConfidence[mappedKey] || 0;
+                if (confidence > existingConfidence) {
+                    metadata[mappedKey] = value;
+                    fieldConfidence[mappedKey] = confidence;
+                }
+            }
+        }
+        else {
+            // Skip fields that are clearly not useful
+            const fieldKey = fieldType.toLowerCase().replace(/_/g, '');
+            // Skip certain field types that aren't useful in our structure
+            if (['street', 'city', 'addressblock', 'expenserow', 'other'].includes(fieldKey)) {
+                continue;
+            }
+            // Store other unmapped fields for extensibility
+            const value = extractFieldValue(field, forceString);
+            if (value !== undefined && !fieldKey.startsWith('vendor_') && !fieldKey.startsWith('amount_')) {
+                // Use more readable field names
+                const cleanKey = fieldType.toLowerCase().replace(/_/g, '');
+                metadata[cleanKey] = value;
+            }
+        }
+    }
+    // Clean up merchant name - take the first line if multiline
+    if (metadata.merchantName && metadata.merchantName.includes('\n')) {
+        metadata.merchantName = metadata.merchantName.split('\n')[0].trim();
+    }
+    return metadata;
+}
+function extractLineItems(lineItemGroups) {
+    if (!lineItemGroups || lineItemGroups.length === 0)
+        return undefined;
+    const items = [];
+    for (const group of lineItemGroups) {
+        if (!group.LineItems)
+            continue;
+        for (const lineItem of group.LineItems) {
+            if (!lineItem.LineItemExpenseFields)
+                continue;
+            const item = { description: '' };
+            for (const field of lineItem.LineItemExpenseFields) {
+                const fieldType = field?.Type?.Text?.toUpperCase();
+                if (!fieldType)
+                    continue;
+                // Determine if this field should be a string
+                const forceString = fieldType === 'ITEM' || fieldType === 'DESCRIPTION' || fieldType === 'PRODUCT_NAME';
+                if (LINE_ITEM_FIELD_MAPPINGS[fieldType]) {
+                    const mappedKey = LINE_ITEM_FIELD_MAPPINGS[fieldType];
+                    const value = extractFieldValue(field, forceString);
+                    if (value !== undefined) {
+                        item[mappedKey] = value;
+                    }
+                }
+                else if (fieldType !== 'EXPENSE_ROW') {
+                    // Store unmapped fields except expense_row which is just raw text
+                    const value = extractFieldValue(field, forceString);
+                    if (value !== undefined) {
+                        const cleanKey = fieldType.toLowerCase().replace(/_/g, '');
+                        item[cleanKey] = value;
+                    }
+                }
+            }
+            // Only add if we have at least a description or some meaningful data
+            if (item.description || (Object.keys(item).length > 1 && (item.total || item.quantity))) {
+                items.push(item);
+            }
+        }
+    }
+    return items.length > 0 ? items : undefined;
+}
+function extractRawText(blocks) {
+    if (!blocks)
+        return undefined;
+    const textBlocks = blocks
+        .filter(block => block.BlockType === 'LINE' && block.Text)
+        .map(block => block.Text)
+        .join('\n');
+    return textBlocks || undefined;
+}
+export function transformTextractExpenseResponse(textractResponse, storageDrive, filePath, includeRaw = true) {
+    // Process the first expense document (most receipts/invoices have just one)
+    const expenseDoc = textractResponse.ExpenseDocuments?.[0];
+    const metadata = extractSummaryFields(expenseDoc?.SummaryFields);
+    const lineItems = extractLineItems(expenseDoc?.LineItemGroups);
+    const rawText = extractRawText(expenseDoc?.Blocks);
+    const result = {
+        metadata,
+        lineItems,
+        rawText,
+        source: {
+            storageDrive: storageDrive,
+            filePath: filePath,
+        },
+    };
+    // Include raw response for debugging (can be disabled in production)
+    if (includeRaw) {
+        result._raw = textractResponse;
+    }
+    return result;
+}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "quidproquo-actionprocessor-awslambda",
-  "version": "0.0.256",
+  "version": "0.0.258",
   "description": "",
   "main": "./lib/commonjs/index.js",
   "module": "./lib/esm/index.js",
@@ -46,7 +46,10 @@
     "@aws-sdk/client-sns": "^3.744.0",
     "@aws-sdk/client-sqs": "^3.744.0",
     "@aws-sdk/client-ssm": "^3.744.0",
+    "@aws-sdk/client-textract": "^3.883.0",
+    "@aws-sdk/s3-presigned-post": "^3.744.0",
     "@aws-sdk/s3-request-presigner": "^3.744.0",
+    "@types/aws-lambda": "^8.10.109",
     "aws-jwt-verify": "^3.4.0",
     "aws-sdk": "^2.1322.0",
     "busboy": "^1.6.0",
@@ -55,17 +58,16 @@
     "lodash": "^4.17.21",
     "node-cache": "^5.1.2",
     "node-match-path": "^0.6.3",
-    "quidproquo-config-aws": "0.0.256",
-    "quidproquo-core": "0.0.256",
-    "quidproquo-webserver": "0.0.256",
-    "@types/aws-lambda": "^8.10.109"
+    "quidproquo-config-aws": "0.0.258",
+    "quidproquo-core": "0.0.258",
+    "quidproquo-webserver": "0.0.258"
   },
   "devDependencies": {
     "@types/busboy": "^1.5.0",
     "@types/jsonwebtoken": "^9.0.2",
     "@types/lodash": "^4.14.194",
     "@types/node": "^22.13.13",
-    "quidproquo-tsconfig": "0.0.256",
+    "quidproquo-tsconfig": "0.0.258",
     "typescript": "^5.8.2"
   }
 }