kodexa 6.0.180__py3-none-any.whl → 6.0.184__py3-none-any.whl

kodexa/model/model.py

@@ -2016,10 +2016,7 @@ class Document(object):
                 'content_node': self.content_node.to_dict() if self.content_node else None,
                 'source': clean_none_values(dataclasses.asdict(self.source)),
                 'mixins': self._mixins,
-                'taxonomies': self.taxonomies,
                 'classes': [content_class.to_dict() for content_class in self.classes],
-                'exceptions': self.exceptions,
-                'log': self.log,
                 'labels': self.labels,
                 'uuid': self.uuid}
 
@@ -2039,8 +2036,6 @@ class Document(object):
         new_document = Document(DocumentMetadata(doc_dict['metadata']))
         new_document.version = doc_dict['version'] if 'version' in doc_dict and doc_dict[
             'version'] else Document.PREVIOUS_VERSION  # some older docs don't have a version or it's None
-        new_document.log = doc_dict['log'] if 'log' in doc_dict else []
-        new_document.exceptions = doc_dict['exceptions'] if 'exceptions' in doc_dict else []
         new_document.uuid = doc_dict['uuid'] if 'uuid' in doc_dict else str(
             uuid.uuid5(uuid.NAMESPACE_DNS, 'kodexa.com'))
 
@@ -2051,11 +2046,6 @@ class Document(object):
             new_document.source = SourceMetadata.from_dict(doc_dict['source'])
         if 'labels' in doc_dict and doc_dict['labels']:
             new_document.labels = doc_dict['labels']
-        if 'taxomomies' in doc_dict and doc_dict['taxomomies']:
-            new_document.labels = doc_dict['taxomomies']
-        if 'classes' in doc_dict and doc_dict['classes']:
-            new_document.classes = [ContentClassification.from_dict(content_class) for content_class in
-                                    doc_dict['classes']]
 
         new_document.get_persistence().update_metadata()
         return new_document

kodexa/pipeline/pipeline.py

@@ -603,7 +603,6 @@ class PipelineStatistics:
 
     def __init__(self):
         self.documents_processed = 0
-        self.document_exceptions = 0
 
     def processed_document(self, document):
         """Update statistics based on this document completing processing
@@ -615,6 +614,3 @@ class PipelineStatistics:
 
         """
         self.documents_processed += 1
-
-        if document and document.exceptions:
-            self.document_exceptions += 1

kodexa/spatial/azure_models.py

@@ -68,6 +68,8 @@ def create_kddb_from_azure(azure_data, keep_azure_lines=True, overlap_percentage
         if issue_found:
             return None
 
+
+
     document.content_node = root_node
     document.add_mixin('spatial')
 
@@ -231,6 +233,22 @@ def get_azure_next_line(document_lines, ref_line, direction='right', overlap_per
 
         return sorted_next_up_lines[0]
 
+    elif direction == 'up_left':
+        # Get all the lines above of the cell, where the x is to the left of the cell
+        up_left_lines = [up_line for up_line in possible_lines if
+                         up_line.get_bbox()[1] >= ref_bbox[3] and
+                         ref_bbox[2] > up_line.get_x() and ref_bbox[0] - up_line.get_bbox()[2] <= 0.75]
+
+        if not up_left_lines:
+            return None
+
+        # Sort by y (decreasing since 0 is at the bottom of the page)
+        sorted_next_up_left_lines = [up_left_lines[0]]
+        [sorted_next_up_left_lines.insert(0, up_left_line) for up_left_line in up_left_lines
+         if up_left_line.get_bbox()[1] < sorted_next_up_left_lines[0].get_bbox()[1]]
+
+        return sorted_next_up_left_lines[0]
+
     return None
 
 

{kodexa-6.0.180.dist-info → kodexa-6.0.184.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: kodexa
-Version: 6.0.180
+Version: 6.0.184
 Summary: Python SDK for the Kodexa Platform
 Author: Austin Redenbaugh
 Author-email: austin@kodexa.com

{kodexa-6.0.180.dist-info → kodexa-6.0.184.dist-info}/RECORD

@@ -5,11 +5,11 @@ 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=Zh1Ug_oy8i_ePL30rA3ag1JAS1YkQf_9t13Xr-B_9H0,89156
+kodexa/model/model.py,sha256=A9v5CRLe-Y28tj0C-oe8L1Gvu3m6bYfPvQQK89tCFzw,88524
 kodexa/model/objects.py,sha256=1cOpzFUriSQjIx3snJ4TNgRRo-8GN54cPanqnsSOmQ4,114246
 kodexa/model/persistence.py,sha256=rRBY_onLcSTFlZZmitU8_FLffP7elDHhcmF8yYT94HE,37655
 kodexa/pipeline/__init__.py,sha256=sA7f5D6qkdMrpp2xTIeefnrUBI6xxEEWostvxfX_1Cs,236
-kodexa/pipeline/pipeline.py,sha256=fofj2wpkbDEyWeyPuebKsKROQeRcfpQWEHZG_GSXWkw,19888
+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/kodexa.py,sha256=HcwQh1NKkwSzkngKLp_kEsLW0N_JlzORszoknSDFLT0,27604
@@ -24,7 +24,7 @@ kodexa/selectors/parserules.pyi,sha256=UQrLMI_bYxdyGjwd4wJDfJevi5lpku8LSbHXGmNpx
 kodexa/selectors/parsetab.py,sha256=JFQMAOjcGu-a5QBJvp77xpQ4Y8J6hQAzDgzl6tMIjYw,21267
 kodexa/selectors/parsetab.pyi,sha256=UQrLMI_bYxdyGjwd4wJDfJevi5lpku8LSbHXGmNpx_g,60
 kodexa/spatial/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-kodexa/spatial/azure_models.py,sha256=1NeXqPloxfnDb6iz4eHXrxfOm_kJzO1SYmYJhmJYdzE,18229
+kodexa/spatial/azure_models.py,sha256=nxpu1QAduZIeym3k9Wf5A1rOHFE3ahheOWEuNepF6gM,18991
 kodexa/spatial/bbox_common.py,sha256=WArS8zv-swd7w6devmCAzZF7MXhD1bzIOZAAFKLVzdE,2738
 kodexa/spatial/table_form_common.py,sha256=K015yXdsK3higBBmt3Kkk3sUwOAlh-1i79ei6Sz2ea4,34222
 kodexa/steps/__init__.py,sha256=crCQCfwjg5QpqRjD8kSNI6QuUvc6O_an6ZKhRgKfShU,160
@@ -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-6.0.180.dist-info/LICENSE,sha256=WNHhf_5RCaeuKWyq_K39vmp9F28LxKsB4SpomwSZ2L0,11357
-kodexa-6.0.180.dist-info/WHEEL,sha256=vxFmldFsRN_Hx10GDvsdv1wroKq8r5Lzvjp6GZ4OO8c,88
-kodexa-6.0.180.dist-info/METADATA,sha256=fG-iZkx0ypuzpIaPWy7oE5dZKpEzDrOZVqh81QlvFa0,3602
-kodexa-6.0.180.dist-info/RECORD,,
+kodexa-6.0.184.dist-info/LICENSE,sha256=WNHhf_5RCaeuKWyq_K39vmp9F28LxKsB4SpomwSZ2L0,11357
+kodexa-6.0.184.dist-info/WHEEL,sha256=vVCvjcmxuUltf8cYhJ0sJMRDLr1XsPuxEId8YDzbyCY,88
+kodexa-6.0.184.dist-info/METADATA,sha256=E_9yTsoZ87ZHtIngeCrom1FRjL20vU1am9_0YN-6V-Y,3602
+kodexa-6.0.184.dist-info/RECORD,,

{kodexa-6.0.180.dist-info → kodexa-6.0.184.dist-info}/WHEEL

@@ -1,4 +1,4 @@
 Wheel-Version: 1.0
-Generator: poetry-core 1.3.2
+Generator: poetry-core 1.4.0
 Root-Is-Purelib: true
 Tag: py3-none-any