Update app.py

app.py

@@ -430,175 +430,67 @@ The output should be formatted as a JSON instance that conforms to the JSON sche
                      }
                      ```"""
             ),
-             "Aramco test": (
-                """You are a high-performance document parsing assistant, optimized for speed and accuracy. Your primary objective is to extract structured data from the provided document and return it in valid JSON format with minimal processing time.
-
-                    Guidelines for Speed Optimization:
-                    1. Process the document with minimal computation and only extract the required fields.
-                    2. Use null for any fields that are missing or not clearly identifiable.
-                    3. Avoid redundant checks or deep parsing; rely on the most straightforward extraction methods.
-                    4. Always return ONLY valid JSON—no additional text, explanations, or formatting errors.
-                    5. Focus on precision for key-value pairs; skip over ambiguous or irrelevant information.
-                    
-                    Document-Specific JSON Structures:
-                    
-                    1. **Invoice**:
-                       - Extract financial and customer details efficiently.
-                       - JSON format:
-                         ```json
-                         {
-                             "invoice": {
-                                 "date": null,
-                                 "dueDate": null,
-                                 "accountNumber": null,
-                                 "invoiceNumber": null,
-                                 "customerContact": null,
-                                 "kintecContact": null,
-                                 "accountsContact": null,
-                                 "periodEnd": null,
-                                 "contractNo": null,
-                                 "specialistsName": null,
-                                 "rpoNumber": null,
-                                 "assignmentProject": null,
-                                 "workLocation": null,
-                                 "expenses": null,
-                                 "regularHours": null,
-                                 "overtime": null,
-                                 "mobilisationAllowance": null,
-                                 "dailyHousing": null,
-                                 "opPipTechnical": null,
-                                 "code": null,
-                                 "vatBasis": null,
-                                 "vatRate": null,
-                                 "vatAmount": null,
-                                 "totalExclVat": null,
-                                 "totalInclVat": null
-                             }
-                         }
-                         ```
-                    
-                    2. **Timesheet**:
-                       - Extract time tracking and approval data swiftly.
-                       - JSON format:
-                         ```json
-                         {
-                             "timesheet": {
-                                 "Year": null,
-                                 "RPO_Number": null,
-                                 "PMC_Name": null,
-                                 "Project_Location": null,
-                                 "Project_and_Package": null,
-                                 "Month": null,
-                                 "Timesheet_Details": [
-                                     {
-                                         "Week": null,
-                                         "Regular_Hours": null,
-                                         "Overtime_Hours": null,
-                                         "Total_Hours": null,
-                                         "Comments": null
-                                     },
-                                     {
-                                         "Week": null,
-                                         "Regular_Hours": null,
-                                         "Overtime_Hours": null,
-                                         "Total_Hours": null,
-                                         "Comments": null
-                                     }
-                                 ],
-                                 "Monthly_Totals": {
-                                     "Regular_Hours": null,
-                                     "Overtime_Hours": null,
-                                     "Total_Hours": null
-                                 },
-                                 "reviewedBy": {
-                                     "name": null,
-                                     "position": null,
-                                     "date": null
-                                 },
-                                 "approvedBy": {
-                                     "name": null,
-                                     "position": null,
-                                     "date": null
-                                 }
-                             }
-                         }
-                         ```
-                    
-                    3. **Purchase Order**:
-                       - Extract contract and pricing details with minimal overhead.
-                       - JSON format:
-                         ```json
-                         {
-                             "purchaseOrder": {
-                                 "contractNo": null,
-                                 "relPoNo": null,
-                                 "version": null,
-                                 "title": null,
-                                 "startDate": null,
-                                 "endDate": null,
-                                 "costCenter": null,
-                                 "purchasingGroup": null,
-                                 "contractor": null,
-                                 "location": null,
-                                 "workDescription": null,
-                                 "pricing": {
-                                     "regularRate": null,
-                                     "overtimeRate": null,
-                                     "totalBudget": null
-                                 }
-                             }
-                         }
-                         ```
-                    
-                    4. **Travel Booking**:
-                       - Extract essential travel and employee data efficiently.
-                       - JSON format:
-                         ```json
-                         {
-                             "travelBooking": {
-                                 "requestId": null,
-                                 "approvalStatus": null,
-                                 "employee": {
-                                     "name": null,
-                                     "id": null,
-                                     "email": null,
-                                     "firstName": null,
-                                     "lastName": null,
-                                     "gradeCodeGroup": null
-                                 },
-                                 "defaultManager": {
-                                     "name": null,
-                                     "email": null
-                                 },
-                                 "sender": {
-                                     "name": null,
-                                     "email": null
-                                 },
-                                 "travel": {
-                                     "startDate": null,
-                                     "endDate": null,
-                                     "requestPolicy": null,
-                                     "requestType": null,
-                                     "employeeType": null,
-                                     "travelActivity": null,
-                                     "tripType": null
-                                 },
-                                 "cost": {
-                                     "companyCode": null,
-                                     "costObject": null,
-                                     "costObjectId": null
-                                 },
-                                 "transport": {
-                                     "type": null,
-                                     "comments": null
-                                 },
-                                 "changeRequired": null,
-                                 "comments": null
-                             }
-                         }
-                         ```
-                    
-                    Ensure your parsing method balances accuracy and speed, prioritizing quick turnaround without compromising JSON validity or structural integrity.
+             "NOC Invoice": (
+                """You are a highly accurate data extraction system.  Your task is to analyze the provided image of an invoice and extract all data, paying close attention to the structure and formatting of the document.  Organize the extracted data in a clear, structured format, such as JSON.  Do not invent any information. If a field cannot be read with high confidence, indicate that with "UNCLEAR" or a similar designation.  Be as specific as possible, and do not summarize or combine fields unless explicitly indicated.  Do not provide any additional information that is not present in the document.
+
+Here's the expected output format, in JSON, with all required fields:
+
+```json
+{
+  "invoiceDetails": {
+    "pleaseQuote": "string", // Text related to quoting
+    "invoiceNumber": "string",
+    "workPeriod": "string",
+    "invoiceDate": "string",
+    "assignmentReference": "string"
+  },
+  "from": {
+    "companyName": "string",
+    "addressLine1": "string",
+    "addressLine2": "string",
+    "city": "string",
+    "postalCode": "string",
+    "country": "string"
+  },
+  "to": {
+    "companyName": "string",
+    "office": "string",
+    "floor": "string",
+	"building":"string",
+    "addressLine1": "string",
+    "poBox": "string",
+    "city": "string"
+  },
+  "services": [
+    {
+      "serviceDetails": "string",
+      "fromDate": "string",
+      "toDate": "string",
+      "currency": "string",
+      "fx": "string", //If it is present
+      "noOfDays": "number or string (if range)",
+      "rate": "number",
+      "total": "number"
+    },
+   // ... more service entries
+  ],
+  "totals": {
+    "subTotal": "number",
+    "tax": "number",
+    "totalDue": "number"
+  },
+  "bankDetails": {
+    "bankName": "string",
+    "descriptionReferenceField": "string",
+    "bankAddress": "string",
+    "swiftBicCode": "string",
+    "ibanNumber": "string",
+    "accountNumber": "string",
+    "beneficiaryName": "string",
+    "accountCurrency": "string",
+    "expectedAmount": "string"
+  }
+}
                     """
             )
         }
@@ -675,10 +567,9 @@ with gr.Blocks() as demo:
             choices=[
                 "NOC Timesheet",
                 "NOC Basic",
-                "NOC Structured test",
                 "Aramco Full structured",
                 "Aramco Timesheet only",
-                "Aramco test"
+                "NOC Invoice"
             ],
             value="Options"
         )