kodexa 6.0.186__py3-none-any.whl → 6.0.192a0__py3-none-any.whl

kodexa/model/model.py

@@ -79,7 +79,7 @@ class Tag(Dict):
         """A string representing the value that was labelled in the node"""
         self.data: Optional[Any] = data
         """Any data object (JSON serializable) that you wish to associate with the label"""
-        self.uuid: Optional[str] = uuid
+        self.uuid: Optional[str] = uuid or str(uuid.uuid4())
         """The UUID for this tag instance, this allows tags that are on different content nodes to be related through the same UUID"""
         self.confidence: Optional[float] = confidence
         """The confidence of the tag in a range of 0-1"""
@@ -1536,7 +1536,8 @@ class ContentNode(object):
             if not node:
                 if (traverse == traverse.ALL or traverse == traverse.PARENT) and self.get_parent().get_parent():
                     # can now traverse content-areas.. can add traversal of pages if needed, but don't think the scenario exists.
-                    potential_next_node = self.get_parent().get_parent().get_children()[self.get_parent().index + 1].get_children()[0]
+                    potential_next_node = \
+                        self.get_parent().get_parent().get_children()[self.get_parent().index + 1].get_children()[0]
                     if potential_next_node:
                         return potential_next_node
                 return node
@@ -1757,6 +1758,9 @@ class Document(object):
         self.classes: List[ContentClassification] = []
         """A list of the content classifications associated at the document level"""
 
+        self.tag_instances: List[TagInstance] = []
+        """A list of tag instances that contains a set of tag that has a set of nodes"""
+
         # Start persistence layer
         from kodexa.model import PersistenceManager
 
@@ -1765,6 +1769,36 @@ class Document(object):
                                                                                    delete_on_close=delete_on_close)
         self._persistence_layer.initialize()
 
+    def add_tag_instance(self, tag_to_apply, node_list: List[ContentNode]):
+        """
+            This will create a group of a tag with indexes
+        :param tag: name of the tag
+        :param node_indices: contains the list of index of a node
+        :return:
+        """
+        # For each node in the list create/update a feature
+        tag = Tag()
+        for node in node_list:
+            node.add_feature('tag', tag_to_apply, Tag)
+        # Tag Object
+        tag_instance = TagInstance(tag, node_list)
+        self.tag_instances.append(tag_instance)
+
+    def update_tag_instance(self, tag_uuid):
+        for tag_instance in self.tag_instances:
+            if tag_instance.tag.uuid == tag_uuid:
+                # Update attributes of a Tag
+                for node in tag_instance.nodes:
+                    node.get_tag(tag_instance.tag.value, tag_uuid=tag_instance.tag.uuid)
+
+    def get_tag_instance(self, tag):
+        """
+            Get the tag instance based on the tag itself
+        :param tag: name of the tag
+        :return: a list of tag instance
+        """
+        return [tag_instance for tag_instance in self.tag_instances if tag_instance.tag == tag]
+
     def get_persistence(self):
         return self._persistence_layer
 
@@ -2271,6 +2305,15 @@ class Document(object):
         return self.labels
 
 
+class TagInstance:
+    def __init__(self, tag: Tag, nodes):
+        self.tag = tag
+        self.nodes = nodes
+
+    def add_node(self, nodes: List[ContentNode]):
+        self.nodes.extend(nodes)
+
+
 class ContentObjectReference:
     """ """
 

kodexa/platform/client.py

@@ -258,6 +258,35 @@ class ComponentEndpoint(ClientEndpoint, OrganizationOwned):
             return None
         return component_page.content[0]
 
+    def stream_list(self, query="*", page=1, page_size=10, sort=None, filters: List[str] = None):
+        url = f"/api/{self.get_type()}/{self.organization.slug}"
+
+        params = {"query": requests.utils.quote(query),
+                  "page": page,
+                  "pageSize": page_size}
+
+        if sort is not None:
+            params["sort"] = sort
+
+        if filters is not None:
+            params["legacyFilter"] = True
+            params["filter"] = filters
+
+        while True:
+            list_response = self.client.get(url, params=params)
+
+            # If there are no more results, exit the loop
+            if not list_response.json()["content"]:
+                break
+
+            # Yield each endpoint in the current page
+            for endpoint in self.get_page_class(list_response.json()).parse_obj(list_response.json()).set_client(
+                    self.client).to_endpoints():
+                yield endpoint
+
+            # Move to the next page
+            params["page"] += 1
+
     def list(self, query="*", page=1, page_size=10, sort=None, filters: List[str] = None):
         url = f"/api/{self.get_type()}/{self.organization.slug}"
 
@@ -987,6 +1016,28 @@ class ProjectsEndpoint(EntitiesEndpoint):
             return ProjectEndpoint.parse_obj(get_response.json()['content'][0]).set_client(self.client)
         return None
 
+    def stream_query(self, query: str = "*", sort=None):
+        """
+            Stream the query for the project endpoints
+        :param query: the query to run
+        :param sort: sorting order of the query
+        :return:
+            A generator of the project endpoints
+        """
+        page_size = 5
+        page = 1
+
+        if not sort:
+            sort = "id"
+
+        while True:
+            page_response = self.query(query=query, page=page, page_size=page_size, sort=sort)
+            if not page_response.content:
+                break
+            for project_endpoint in page_response.content:
+                yield project_endpoint
+            page += 1
+
     def query(self, query: str = "*", page: int = 1, page_size: int = 100, sort=None) -> Optional[PageProjectEndpoint]:
         params = {
             'page': page,
@@ -2004,6 +2055,28 @@ class DocumentStoreEndpoint(StoreEndpoint):
             f"/api/stores/{self.ref.replace(':', '/')}/families/{document_family_id}")
         return DocumentFamilyEndpoint.parse_obj(document_family_response.json()).set_client(self.client)
 
+    def stream_query(self, query: str = "*", sort=None):
+        """
+            Stream the query for the document family
+        :param query: the query to run
+        :param sort: sorting order of the query
+        :return:
+            A generator of the document families
+        """
+        page_size = 5
+        page = 1
+
+        if not sort:
+            sort = "id"
+
+        while True:
+            page_response = self.query(query=query, page=page, page_size=page_size, sort=sort)
+            if not page_response.content:
+                break
+            for document_family in page_response.content:
+                yield document_family
+            page += 1
+
     def query(self, query: str = "*", page: int = 1, page_size: int = 100, sort=None) -> PageDocumentFamilyEndpoint:
         params = {
             'page': page,
@@ -2019,6 +2092,28 @@ class DocumentStoreEndpoint(StoreEndpoint):
 
         return PageDocumentFamilyEndpoint.parse_obj(get_response.json()).set_client(self.client)
 
+    def stream_filter(self, filter_string: str = "", sort=None):
+        """
+            Stream the filter for the document family
+        :param query: the query to run
+        :param sort: sorting order of the query
+        :return:
+            A generator of the document families
+        """
+        page_size = 5
+        page = 1
+
+        if not sort:
+            sort = "id"
+
+        while True:
+            page_response = self.filter(filter_string=filter_string, page=page, page_size=page_size, sort=sort)
+            if not page_response.content:
+                break
+            for document_family in page_response.content:
+                yield document_family
+            page += 1
+
     def filter(self, filter_string: str = "", page: int = 1, page_size: int = 100,
                sort=None) -> PageDocumentFamilyEndpoint:
         params = {

{kodexa-6.0.186.dist-info → kodexa-6.0.192a0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: kodexa
-Version: 6.0.186
+Version: 6.0.192a0
 Summary: Python SDK for the Kodexa Platform
 Author: Austin Redenbaugh
 Author-email: austin@kodexa.com

{kodexa-6.0.186.dist-info → kodexa-6.0.192a0.dist-info}/RECORD

@@ -5,13 +5,13 @@ kodexa/connectors/__init__.py,sha256=WCUEzFGjHcgPAMFIKLaRTXAkGHx3vUCD8APMhOrNNgM
 kodexa/connectors/connectors.py,sha256=25-TffyGDjxHyp9ITug0qgr1nhqMAekmV5NVvbPGs7o,7722
 kodexa/model/__init__.py,sha256=DyCgkJU7rOfd4SMvPRLaPdklCNlkqCRRWiVPwjYn2GE,720
 kodexa/model/base.py,sha256=6IraEK3RomjPgFpPYkxjuLUriF958AusgJO21Dcopeg,753
-kodexa/model/model.py,sha256=A9v5CRLe-Y28tj0C-oe8L1Gvu3m6bYfPvQQK89tCFzw,88524
+kodexa/model/model.py,sha256=-UTEHt3hDLGgvVqiJ860m2O907gntTZgOtcwusYHXtY,90090
 kodexa/model/objects.py,sha256=c58_U0k5wQdha7zcV3oEO_aHlq1Jfh72kzK_6EapZZE,114121
 kodexa/model/persistence.py,sha256=rRBY_onLcSTFlZZmitU8_FLffP7elDHhcmF8yYT94HE,37655
 kodexa/pipeline/__init__.py,sha256=sA7f5D6qkdMrpp2xTIeefnrUBI6xxEEWostvxfX_1Cs,236
 kodexa/pipeline/pipeline.py,sha256=uzxe7HuSW1CKDZOrnV_LRHj3SHhbs14lvmMGJ_DIVdw,19763
 kodexa/platform/__init__.py,sha256=1O3oiWMg292NPL_NacKDnK1T3_R6cMorrPRue_9e-O4,216
-kodexa/platform/client.py,sha256=nv6o3UDzSv_Sr4FbtsuHr5_Nn5xUYy5YVBaImRDyscw,106727
+kodexa/platform/client.py,sha256=v590s5pHtF4ylAiY67HL0uQV-sXqjFcY0woj8GzDy9M,109778
 kodexa/platform/kodexa.py,sha256=HcwQh1NKkwSzkngKLp_kEsLW0N_JlzORszoknSDFLT0,27604
 kodexa/selectors/__init__.py,sha256=xA9-4vpyaAZWPSk3bh2kvDLkdv6XEmm7PjFbpziiTIk,100
 kodexa/selectors/ast.py,sha256=mTAcX_pjDiDy99ELxbMMwAAkzezOqz5uCyui9qz7Q4A,13499
@@ -34,7 +34,7 @@ kodexa/testing/test_components.py,sha256=i_9M6-bfUBdR1uYAzZZzWiW0M1DGKzE5mkNuHq4
 kodexa/testing/test_utils.py,sha256=HXM3S5FDzarzS6R7jkOHps6d6Ox2UtNqymoK6VCw8Zg,13596
 kodexa/training/__init__.py,sha256=xs2L62YpRkIRfslQwtQZ5Yxjhm7sLzX2TrVX6EuBnZQ,52
 kodexa/training/train_utils.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-kodexa-6.0.186.dist-info/LICENSE,sha256=WNHhf_5RCaeuKWyq_K39vmp9F28LxKsB4SpomwSZ2L0,11357
-kodexa-6.0.186.dist-info/WHEEL,sha256=vVCvjcmxuUltf8cYhJ0sJMRDLr1XsPuxEId8YDzbyCY,88
-kodexa-6.0.186.dist-info/METADATA,sha256=FG9e6VqXiMu6VlyckKaTXTtOpVksyHGDMDyY2BKq8eg,3602
-kodexa-6.0.186.dist-info/RECORD,,
+kodexa-6.0.192a0.dist-info/LICENSE,sha256=WNHhf_5RCaeuKWyq_K39vmp9F28LxKsB4SpomwSZ2L0,11357
+kodexa-6.0.192a0.dist-info/WHEEL,sha256=vVCvjcmxuUltf8cYhJ0sJMRDLr1XsPuxEId8YDzbyCY,88
+kodexa-6.0.192a0.dist-info/METADATA,sha256=dwQ5QImPCW7E4eUbUxBDTVALmn6QgE0-89uKN28X6WQ,3604
+kodexa-6.0.192a0.dist-info/RECORD,,