dcicutils 8.8.4.1b35__tar.gz → 8.8.4.1b36__tar.gz

{dcicutils-8.8.4.1b35 → dcicutils-8.8.4.1b36}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: dcicutils
-Version: 8.8.4.1b35
+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-8.8.4.1b35 → dcicutils-8.8.4.1b36}/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)]
 
 

{dcicutils-8.8.4.1b35 → dcicutils-8.8.4.1b36}/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-8.8.4.1b35 → dcicutils-8.8.4.1b36}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "dcicutils"
-version = "8.8.4.1b35"  # TODO: To become 8.8.5
+version = "8.8.4.1b36"  # TODO: To become 8.8.5
 description = "Utility package for interacting with the 4DN Data Portal and other 4DN resources"
 authors = ["4DN-DCIC Team <support@4dnucleome.org>"]
 license = "MIT"