data-science-document-ai 1.43.5__tar.gz → 1.43.7__tar.gz

{data_science_document_ai-1.43.5 → data_science_document_ai-1.43.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: data-science-document-ai
-Version: 1.43.5
+Version: 1.43.7
 Summary: "Document AI repo for data science"
 Author: Naomi Nguyen
 Author-email: naomi.nguyen@forto.com

{data_science_document_ai-1.43.5 → data_science_document_ai-1.43.7}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "data-science-document-ai"
-version = "1.43.5"
+version = "1.43.7"
 description = "\"Document AI repo for data science\""
 authors = ["Naomi Nguyen <naomi.nguyen@forto.com>", "Kumar Rajendrababu <kumar.rajendrababu@forto.com>", "Igor Tonko <igor.tonko@forto.com>", "Osman Demirel <osman.demirel@forto.com>"]
 packages = [

{data_science_document_ai-1.43.5 → data_science_document_ai-1.43.7}/src/prompts/library/partnerInvoice/other/prompt.txt

@@ -53,7 +53,7 @@ Your role is to accurately extract specific entities from these invoices to supp
     - totalAmount: The total amount for the item. It can be in different currencies, so ensure to capture the currency as well for the totalAmountCurrency.
     - totalAmountEuro: Few line items contains a total amount in Euro. You can find it by looking for the term "Total EUR" or "Amount in Euro" in the line item but it's always in the EURO / € currency. Sometimes, it can be same as totalAmount if the line item is already in Euro.
     - quantity: The quantity of the item or service provided in the line item. Pay attention to 2x40HC. It means, quantity is 2 and containerSize is 40HC but not 240.
-    - containerNumber: Container Number always starts with 4 letters and is followed by 7 digits (e.g., ABCD1234567).
+    - containerNumber: Container Number always starts with 4 letters and is followed by 7 digits (e.g., ABCD1234567, XALU 8593678).
 
 - hblNumber and mblNumber:
     - The Master Bill of Lading number. Commonly known as "Bill of Lading Number", "BILL OF LADING NO.", "BL Number", "BL No.", "B/L No.", "BL-Nr.", "B/L", or "HBL No.".
@@ -81,6 +81,7 @@ Your role is to accurately extract specific entities from these invoices to supp
 
 IMPORTANT NOTE:
 - Ensure all extracted values are directly from the document. Do not make assumptions, modifications or calculations.
+- Do not split the quantity into different line items. e.g., if quantity is 2 or 2 CTR or 2 BIL, do not create 2 separate line items with quantity 1 each.
 - Do not normalize or modify any entity values.
 - Pay attention to the line item details and paymentInformation, as they may vary significantly across different invoices.
 

{data_science_document_ai-1.43.5 → data_science_document_ai-1.43.7}/src/utils.py

@@ -443,12 +443,23 @@ def transform_schema_strings(schema):
     Returns:
         dict: The transformed schema dictionary.
     """
-    # Base case: if the current schema definition is for a string
-    if isinstance(schema, dict) and schema.get("type").upper() == "STRING":
-        new_schema = {
+    if not isinstance(schema, dict):
+        return schema
+
+    schema_type = schema.get("type")
+    if not schema_type:
+        return schema
+
+    # Base case: STRING → OBJECT (only if not already transformed)
+    if schema_type.upper() == "STRING":
+        return {
             "type": "OBJECT",
             "properties": {
-                "value": {"type": "STRING"},
+                "value": {
+                    "type": "STRING",
+                    "nullable": schema.get("nullable", False),
+                    "description": schema.get("description", ""),
+                },
                 "page_number": {
                     "type": "STRING",
                     "description": "Number of a page where the value was found in the document starting from 0.",
@@ -457,29 +468,29 @@ def transform_schema_strings(schema):
             "required": [],
         }
 
-        # Preserve original properties like nullable and description on the new 'value' key
-        if "nullable" in schema:
-            new_schema["properties"]["value"]["nullable"] = schema["nullable"]
-        if "description" in schema:
-            new_schema["properties"]["value"]["description"] = schema["description"]
+    # Skip already transformed OBJECT (has both 'value' & 'page_number')
+    if (
+        schema_type.upper() == "OBJECT"
+        and "properties" in schema
+        and {"value", "page_number"}.issubset(schema["properties"].keys())
+    ):
+        return schema
 
+    # Recursive case for OBJECT
+    if schema_type.upper() == "OBJECT" and "properties" in schema:
+        new_schema = schema.copy()
+        new_schema["properties"] = {
+            k: transform_schema_strings(v) for k, v in schema["properties"].items()
+        }
         return new_schema
 
-    # Recursive case: if the schema is a dictionary
-    elif isinstance(schema, dict) and schema.get("type").upper() == "OBJECT":
-        transformed_schema = schema.copy()
-        for key, value in schema.get("properties").items():
-            transformed_schema["properties"][key] = transform_schema_strings(value)
-        return transformed_schema
-
-    # Recursive case: if the schema is a list
-    elif isinstance(schema, dict) and schema.get("type").upper() == "ARRAY":
-        schema["items"] = transform_schema_strings(schema["items"])
-        return schema
+    # Recursive case for ARRAY
+    if schema_type.upper() == "ARRAY" and "items" in schema:
+        new_schema = schema.copy()
+        new_schema["items"] = transform_schema_strings(schema["items"])
+        return new_schema
 
-    # Base case: for non-dict/list values (e.g., None, bool, str)
-    else:
-        return schema
+    return schema
 
 
 def estimate_page_count(sheet):