dcicutils 8.8.4.1b34__py3-none-any.whl → 8.8.4.1b36__py3-none-any.whl

dcicutils/misc_utils.py

@@ -1155,7 +1155,8 @@ def remove_suffix(suffix: str, text: str, required: bool = False):
 
 def remove_empty_properties(data: Optional[Union[list, dict]],
                             isempty: Optional[Callable] = None,
-                            isempty_array_element: Optional[Callable] = None) -> None:
+                            isempty_array_element: Optional[Callable] = None,
+                            raise_exception_on_nonempty_array_element_after_empty: bool = False) -> None:
     def _isempty(value: Any) -> bool:  # noqa
         return isempty(value) if callable(isempty) else value in [None, "", {}, []]
     if isinstance(data, dict):
@@ -1163,11 +1164,22 @@ def remove_empty_properties(data: Optional[Union[list, dict]],
             if _isempty(value := data[key]):
                 del data[key]
             else:
-                remove_empty_properties(value, isempty=isempty, isempty_array_element=isempty_array_element)
+                remove_empty_properties(value, isempty=isempty, isempty_array_element=isempty_array_element,
+                                        raise_exception_on_nonempty_array_element_after_empty=  # noqa
+                                        raise_exception_on_nonempty_array_element_after_empty)
     elif isinstance(data, list):
         for item in data:
-            remove_empty_properties(item, isempty=isempty, isempty_array_element=isempty_array_element)
+            remove_empty_properties(item, isempty=isempty, isempty_array_element=isempty_array_element,
+                                    raise_exception_on_nonempty_array_element_after_empty=  # noqa
+                                    raise_exception_on_nonempty_array_element_after_empty)
         if callable(isempty_array_element):
+            if raise_exception_on_nonempty_array_element_after_empty is True:
+                empty_element_seen = False
+                for item in data:
+                    if not empty_element_seen and isempty_array_element(item):
+                        empty_element_seen = True
+                    elif empty_element_seen and not isempty_array_element(item):
+                        raise Exception("Non-empty element found after empty element.")
             data[:] = [item for item in data if not isempty_array_element(item)]
 
 
@@ -2735,6 +2747,3 @@ def create_short_uuid(length: Optional[int] = None, upper: bool = False):
     if upper is True:
         value = value.upper()
     return value
-
-
-print(format_size(1024 * 1024))

dcicutils/structured_data.py

@@ -53,6 +53,7 @@ class StructuredDataSet:
     def __init__(self, file: Optional[str] = None, portal: Optional[Union[VirtualApp, TestApp, Portal]] = None,
                  schemas: Optional[List[dict]] = None, autoadd: Optional[dict] = None,
                  order: Optional[List[str]] = None, prune: bool = True,
+                 remove_empty_objects_from_lists: bool = True,
                  ref_lookup_strategy: Optional[Callable] = None,
                  ref_lookup_nocache: bool = False,
                  norefs: bool = False,
@@ -65,7 +66,8 @@ class StructuredDataSet:
                               ref_lookup_nocache=ref_lookup_nocache) if portal else None
         self._ref_lookup_strategy = ref_lookup_strategy
         self._order = order
-        self._prune = prune
+        self._prune = prune is True
+        self._remove_empty_objects_from_lists = remove_empty_objects_from_lists is True
         self._warnings = {}
         self._errors = {}
         self._resolved_refs = set()
@@ -93,12 +95,14 @@ class StructuredDataSet:
     def load(file: str, portal: Optional[Union[VirtualApp, TestApp, Portal]] = None,
              schemas: Optional[List[dict]] = None, autoadd: Optional[dict] = None,
              order: Optional[List[str]] = None, prune: bool = True,
+             remove_empty_objects_from_lists: bool = True,
              ref_lookup_strategy: Optional[Callable] = None,
              ref_lookup_nocache: bool = False,
              norefs: bool = False,
              progress: Optional[Callable] = None,
              debug_sleep: Optional[str] = None) -> StructuredDataSet:
         return StructuredDataSet(file=file, portal=portal, schemas=schemas, autoadd=autoadd, order=order, prune=prune,
+                                 remove_empty_objects_from_lists=remove_empty_objects_from_lists,
                                  ref_lookup_strategy=ref_lookup_strategy, ref_lookup_nocache=ref_lookup_nocache,
                                  norefs=norefs, progress=progress, debug_sleep=debug_sleep)
 
@@ -368,7 +372,11 @@ class StructuredDataSet:
                 structured_row_template.set_value(structured_row, column_name, value, reader.file, reader.row_number)
                 if self._autoadd_properties:
                     self._add_properties(structured_row, self._autoadd_properties, schema)
-            self._add(type_name, structured_row)
+            if (prune_error := self._prune_structured_row(structured_row)) is not None:
+                self._note_error({"src": create_dict(type=schema_name, row=reader.row_number),
+                                  "error": prune_error}, "validation")
+            else:
+                self._add(type_name, structured_row)
             if self._progress:
                 self._progress({
                     PROGRESS.LOAD_ITEM: self._nrows,
@@ -385,9 +393,20 @@ class StructuredDataSet:
             self._note_error(schema._unresolved_refs, "ref")
             self._resolved_refs.update(schema._resolved_refs)
 
-    def _add(self, type_name: str, data: Union[dict, List[dict]]) -> None:
-        if self._prune:
+    def _prune_structured_row(self, data: dict) -> Optional[str]:
+        if not self._prune:
+            return None
+        if not self._remove_empty_objects_from_lists:
             remove_empty_properties(data)
+            return None
+        try:
+            remove_empty_properties(data, isempty_array_element=lambda element: element == {},
+                                    raise_exception_on_nonempty_array_element_after_empty=True)
+        except Exception as e:
+            return str(e)
+        return None
+
+    def _add(self, type_name: str, data: Union[dict, List[dict]]) -> None:
         if type_name in self._data:
             self._data[type_name].extend([data] if isinstance(data, dict) else data)
         else:

{dcicutils-8.8.4.1b34.dist-info → dcicutils-8.8.4.1b36.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: dcicutils
-Version: 8.8.4.1b34
+Version: 8.8.4.1b36
 Summary: Utility package for interacting with the 4DN Data Portal and other 4DN resources
 Home-page: https://github.com/4dn-dcic/utils
 License: MIT

{dcicutils-8.8.4.1b34.dist-info → dcicutils-8.8.4.1b36.dist-info}/RECORD

@@ -44,7 +44,7 @@ dcicutils/license_policies/park-lab-gpl-pipeline.jsonc,sha256=vLZkwm3Js-kjV44nug
 dcicutils/license_policies/park-lab-pipeline.jsonc,sha256=9qlY0ASy3iUMQlr3gorVcXrSfRHnVGbLhkS427UaRy4,283
 dcicutils/license_utils.py,sha256=d1cq6iwv5Ju-VjdoINi6q7CPNNL7Oz6rcJdLMY38RX0,46978
 dcicutils/log_utils.py,sha256=7pWMc6vyrorUZQf-V-M3YC6zrPgNhuV_fzm9xqTPph0,10883
-dcicutils/misc_utils.py,sha256=6TWxOOWav47An4X2XCgohYtGEncjU7XdYqnoYSRfyG4,106786
+dcicutils/misc_utils.py,sha256=zHwsxxEn24muLBP7mDvMa8I9VdMejwW8HMuCL5xbhhw,107690
 dcicutils/obfuscation_utils.py,sha256=fo2jOmDRC6xWpYX49u80bVNisqRRoPskFNX3ymFAmjw,5963
 dcicutils/opensearch_utils.py,sha256=V2exmFYW8Xl2_pGFixF4I2Cc549Opwe4PhFi5twC0M8,1017
 dcicutils/portal_object_utils.py,sha256=gDXRgPsRvqCFwbC8WatsuflAxNiigOnqr0Hi93k3AgE,15422
@@ -64,7 +64,7 @@ dcicutils/secrets_utils.py,sha256=8dppXAsiHhJzI6NmOcvJV5ldvKkQZzh3Fl-cb8Wm7MI,19
 dcicutils/sheet_utils.py,sha256=VlmzteONW5VF_Q4vo0yA5vesz1ViUah1MZ_yA1rwZ0M,33629
 dcicutils/snapshot_utils.py,sha256=ymP7PXH6-yEiXAt75w0ldQFciGNqWBClNxC5gfX2FnY,22961
 dcicutils/ssl_certificate_utils.py,sha256=F0ifz_wnRRN9dfrfsz7aCp4UDLgHEY8LaK7PjnNvrAQ,9707
-dcicutils/structured_data.py,sha256=BQuIMv6OPySsn6YxtXE2Er-zLE2QJuCYhEQ3V0u_UXY,61238
+dcicutils/structured_data.py,sha256=klpphnw4-mIZk8rGd5YBsaYkwS-YJ-6c1OasvWnqYhE,62284
 dcicutils/submitr/progress_constants.py,sha256=5bxyX77ql8qEJearfHEvsvXl7D0GuUODW0T65mbRmnE,2895
 dcicutils/submitr/ref_lookup_strategy.py,sha256=Js2cVznTmgjciLWBPLCvMiwLIHXjDn3jww-gJPjYuFw,3467
 dcicutils/task_utils.py,sha256=MF8ujmTD6-O2AC2gRGPHyGdUrVKgtr8epT5XU8WtNjk,8082
@@ -73,8 +73,8 @@ dcicutils/trace_utils.py,sha256=g8kwV4ebEy5kXW6oOrEAUsurBcCROvwtZqz9fczsGRE,1769
 dcicutils/validation_utils.py,sha256=cMZIU2cY98FYtzK52z5WUYck7urH6JcqOuz9jkXpqzg,14797
 dcicutils/variant_utils.py,sha256=2H9azNx3xAj-MySg-uZ2SFqbWs4kZvf61JnK6b-h4Qw,4343
 dcicutils/zip_utils.py,sha256=_Y9EmL3D2dUZhxucxHvrtmmlbZmK4FpSsHEb7rGSJLU,3265
-dcicutils-8.8.4.1b34.dist-info/LICENSE.txt,sha256=qnwSmfnEWMl5l78VPDEzAmEbLVrRqQvfUQiHT0ehrOo,1102
-dcicutils-8.8.4.1b34.dist-info/METADATA,sha256=hkWY8lqw4aHKXxtqn81dHbQpU2PeJ_3a-ZurdInqtB4,3440
-dcicutils-8.8.4.1b34.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
-dcicutils-8.8.4.1b34.dist-info/entry_points.txt,sha256=51Q4F_2V10L0282W7HFjP4jdzW4K8lnWDARJQVFy_hw,270
-dcicutils-8.8.4.1b34.dist-info/RECORD,,
+dcicutils-8.8.4.1b36.dist-info/LICENSE.txt,sha256=qnwSmfnEWMl5l78VPDEzAmEbLVrRqQvfUQiHT0ehrOo,1102
+dcicutils-8.8.4.1b36.dist-info/METADATA,sha256=LrZoMbnXJRhcUEnTg2Z8KIBARN_dA_8e_cI4CH8CWKE,3440
+dcicutils-8.8.4.1b36.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
+dcicutils-8.8.4.1b36.dist-info/entry_points.txt,sha256=51Q4F_2V10L0282W7HFjP4jdzW4K8lnWDARJQVFy_hw,270
+dcicutils-8.8.4.1b36.dist-info/RECORD,,