kodexa 7.4.413466578435__py3-none-any.whl → 7.4.413508355193__py3-none-any.whl

kodexa/model/model.py

@@ -1244,6 +1244,7 @@ class ContentNode(object):
             status=None,
             owner_uri=None,
             is_dirty=None,
+            sort_by_bbox: bool=False,
     ):
         """
         This will tag (see Feature Tagging) the expression groups identified by the regular expression.
@@ -1305,7 +1306,7 @@ class ContentNode(object):
             return str(uuid.uuid4())
 
         def tag_node_position(
-                node_to_check, start, end, node_data, tag_uuid, offset=0, value=None
+                node_to_check, start, end, node_data, tag_uuid, offset=0, value=None, sort_by_bbox: bool=False
         ):
             """
             This function tags a node position in a given data structure. It iterates over the content parts of the node to check,
@@ -1420,6 +1421,7 @@ class ContentNode(object):
                         tag_uuid,
                         offset=offset,
                         value=value,
+                        sort_by_bbox=sort_by_bbox,
                     )
 
                     if result < 0 or (end - result) <= 0:
@@ -1434,7 +1436,16 @@ class ContentNode(object):
                     raise Exception("Invalid part?")
 
             # We need to determine if we have missing children and add them to the end
-            for child_idx, child_node in enumerate(node_to_check.get_children()):
+            node_children = node_to_check.get_children()
+            if node_children and sort_by_bbox:
+                # Sort nodes by x-coordinate if they have bboxes, otherwise use index
+                try:
+                    node_children.sort(key=lambda x: x.get_bbox()[0] if hasattr(x, 'get_bbox') else x.index if hasattr(x, 'index') else 0)
+                except (AttributeError, TypeError, IndexError):
+                    # If sorting fails, keep original order
+                    pass
+            
+            for child_idx, child_node in enumerate(node_children):
                 if child_node.index not in node_to_check.get_content_parts():
                     if content_length > 0:
                         end = end - len(separator)
@@ -1452,6 +1463,7 @@ class ContentNode(object):
                         tag_uuid,
                         offset=offset,
                         value=value,
+                        sort_by_bbox=sort_by_bbox,
                     )
 
                     if result < 0 or (end - result) <= 0:
@@ -1487,6 +1499,7 @@ class ContentNode(object):
                     get_tag_uuid(tag_uuid),
                     0,
                     value=value,
+                    sort_by_bbox=sort_by_bbox,
                 )
 
             else:
@@ -1558,6 +1571,7 @@ class ContentNode(object):
                                             data,
                                             get_tag_uuid(tag_uuid),
                                             value=value,
+                                            sort_by_bbox=sort_by_bbox,
                                         )
 
                         else:
@@ -1572,6 +1586,7 @@ class ContentNode(object):
                                     data,
                                     get_tag_uuid(tag_uuid),
                                     value=value,
+                                    sort_by_bbox=sort_by_bbox,
                                 )
 
     def get_tags(self):

kodexa/model/objects.py

@@ -3108,6 +3108,10 @@ class Task(BaseModel):
         protected_namespaces=("model_config",),
     )
 
+    id: Optional[str] = Field(None)
+    uuid: Optional[str] = None
+    created_on: Optional[StandardDateTime] = Field(None, alias="createdOn")
+    updated_on: Optional[StandardDateTime] = Field(None, alias="updatedOn")
     project: Optional['Project'] = None
     title: Optional[str] = None
     description: Optional[str] = None

kodexa/platform/client.py

@@ -4634,14 +4634,18 @@ class DocumentFamilyEndpoint(DocumentFamily, ClientEndpoint):
 
     def get_json(
             self,
-            project_id: str,
+            project_id: Optional[str] = None,
             friendly_names=False,
+            include_ids=True,
+            include_exceptions=False,          
     ) -> str:
         """Get the JSON export for the document family
 
         Args:
             project_id str: The project ID
             friendly_names (bool): Whether to use friendly names. Defaults to False
+            include_ids (bool): Whether to include the IDs. Defaults to True
+            include_exceptions (bool): Whether to include the exceptions. Defaults to False
 
         Returns:
             str: The JSON
@@ -4656,6 +4660,8 @@ class DocumentFamilyEndpoint(DocumentFamily, ClientEndpoint):
             "format": "json",
             "friendlyNames": friendly_names,
             "projectId": project_id,
+            "includeIds": include_ids,
+            "includeExceptions": include_exceptions,
         }
 
         response = self.client.get(url, params=params)

{kodexa-7.4.413466578435.dist-info → kodexa-7.4.413508355193.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: kodexa
-Version: 7.4.413466578435
+Version: 7.4.413508355193
 Summary: Python SDK for the Kodexa Platform
 Author: Austin Redenbaugh
 Author-email: austin@kodexa.com

{kodexa-7.4.413466578435.dist-info → kodexa-7.4.413508355193.dist-info}/RECORD

@@ -12,14 +12,14 @@ kodexa/model/entities/check_response.py,sha256=eqBHxO6G2OAziL3p9bHGI-oiPkAG82H6C
 kodexa/model/entities/product.py,sha256=StUhTEeLXmc05cj6XnZppQfeJsqCPbX1jdhsysHH--Q,5787
 kodexa/model/entities/product_group.py,sha256=540fRGyUf34h1BzAN1DiWu6rGgvaj3xDFhZ2k-RvSFY,3617
 kodexa/model/entities/product_subscription.py,sha256=UcmWR-qgLfdV7VCtJNwzgkanoS8nBSL6ngVuxQUK1M8,3810
-kodexa/model/model.py,sha256=qh1YUew3UgtjU0t4fAwSXYYuzQjXTOZWZkafyFp_w8M,118801
-kodexa/model/objects.py,sha256=ZEKSGugy7qmVbmW9y6haDCF_Zs1uOZdB5bn8eqyw2Zk,189319
+kodexa/model/model.py,sha256=q3zEm6pPOB-xPCKbOxmTMqLALzmQr2Ppam8knApoSEE,119645
+kodexa/model/objects.py,sha256=C1zdBU7DmnhLwioCd8OEKEGEcDDZbCorolJ4bnTKVg0,189538
 kodexa/model/persistence.py,sha256=jUgQ8xwsAFIoZ_bEynxCDEWhUII42eN0e0Mum0dkQPg,72043
 kodexa/model/utils.py,sha256=6R-3rFiW9irBwj0Mq5yhp7EDXkNUFaeFhr3bWmnlW4g,2961
 kodexa/pipeline/__init__.py,sha256=sA7f5D6qkdMrpp2xTIeefnrUBI6xxEEWostvxfX_1Cs,236
 kodexa/pipeline/pipeline.py,sha256=zyNEpA7KlGhPs_l-vgV6m-OCb16dbxQhl8QezeylugA,25540
 kodexa/platform/__init__.py,sha256=1O3oiWMg292NPL_NacKDnK1T3_R6cMorrPRue_9e-O4,216
-kodexa/platform/client.py,sha256=ydZrRP3tBNM9E2khjkoHhkG2UphEQx2tbteni4ekhkw,231089
+kodexa/platform/client.py,sha256=e-MZV4chLYVQj24EADsOdYXbRJ6dgdJJmZ0fqfujSQw,231445
 kodexa/platform/interaction.py,sha256=6zpcwXKNZstUGNS6m4JsoRXAqCZPJHWI-ZN3co8nnF0,1055
 kodexa/platform/kodexa.py,sha256=tPXHO500q3S75GhKGDcaxO51Viq2PNlHmAzpBZlahgo,34857
 kodexa/selectors/__init__.py,sha256=xA9-4vpyaAZWPSk3bh2kvDLkdv6XEmm7PjFbpziiTIk,100
@@ -44,7 +44,7 @@ kodexa/testing/test_utils.py,sha256=v44p__gE7ia67W7WeHN2HBFCWSCUrCZt7G4xBNCmwf8,
 kodexa/training/__init__.py,sha256=xs2L62YpRkIRfslQwtQZ5Yxjhm7sLzX2TrVX6EuBnZQ,52
 kodexa/training/train_utils.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 kodexa/utils/__init__.py,sha256=Pnim1o9_db5YEnNvDTxpM7HG-qTlL6n8JwFwOafU9wo,5928
-kodexa-7.4.413466578435.dist-info/LICENSE,sha256=WNHhf_5RCaeuKWyq_K39vmp9F28LxKsB4SpomwSZ2L0,11357
-kodexa-7.4.413466578435.dist-info/METADATA,sha256=VHhK3bW8tZV7UsLYKVd654E5pFoRaW6ZcmL-o69oC_Q,2813
-kodexa-7.4.413466578435.dist-info/WHEEL,sha256=Zb28QaM1gQi8f4VCBhsUklF61CTlNYfs9YAZn-TOGFk,88
-kodexa-7.4.413466578435.dist-info/RECORD,,
+kodexa-7.4.413508355193.dist-info/LICENSE,sha256=WNHhf_5RCaeuKWyq_K39vmp9F28LxKsB4SpomwSZ2L0,11357
+kodexa-7.4.413508355193.dist-info/METADATA,sha256=YNvguxKfF49HqQ-wHPTPoD3WrWRN9r29FWSURSO1GOk,2813
+kodexa-7.4.413508355193.dist-info/WHEEL,sha256=Zb28QaM1gQi8f4VCBhsUklF61CTlNYfs9YAZn-TOGFk,88
+kodexa-7.4.413508355193.dist-info/RECORD,,