Added remove_empty_objects_from_lists support in structured_data.

CHANGELOG.rst

@@ -14,6 +14,9 @@ Change Log
   for running Foursight checks locally (with local ssh tunnel to ES proxy);
   came up in foursight/checks/audit_checks (2024-04-23).
 * Allow Python 3.12 (pyproject.toml).
+* Added remove_empty_objects_from_lists options to structured_data.StructuredDataSet, defaulting
+  to True, which deletes empty objects from lists; however, only from the *end* of a list; if
+  this flag is True and there are non-empty objects following empty objects then we flag an error.
 * Few general things initially related to and factored out of rclone support in smaht-submitr:
   - Added extract_file_from_zip to zip_utils.
   - Added http_utils with download function.

dcicutils/misc_utils.py

@@ -1155,19 +1155,31 @@ 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):
         for key in list(data.keys()):
             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/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 @@ def __init__(self, file: Optional[str] = None, portal: Optional[Union[VirtualApp
                               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 @@ def portal(self) -> Optional[Portal]:
     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 @@ def _load_reader(self, reader: RowReader, type_name: str) -> None:
                 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 @@ def _load_reader(self, reader: RowReader, type_name: str) -> None:
             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:

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"

test/test_structured_data.py

@@ -943,9 +943,37 @@ def _pytest_kwargs(kwargs: List[dict]) -> List[dict]:
                 }
             }
         ],
-        # "expected": {"Test": [{"arrayofobject": [{"name": "anastasiia", "id": 1234}]}]},
         "expected": {"Test": [{"arrayofobject": [{}, {}, {"name": "olha", "id": 5678},
-                                                 {}, {"name": "anastasiia", "id": 1234}]}]}
+                                                 {}, {"name": "anastasiia", "id": 1234}]}]},
+        "remove_empty_objects_from_lists": False
+    },
+    # ----------------------------------------------------------------------------------------------
+    {
+        "rows": [
+            "arrayofobject#0.name,arrayofobject#0.id,arrayofobject#1.name,arrayofobject#1.id,"
+            "arrayofobject#2.name,arrayofobject#2.id",
+            "anastasiia,1234,olha,5678,,"
+        ],
+        "as_file_name": "test.csv",
+        "schemas": [
+            {
+                "title": "Test",
+                "properties": {
+                    "arrayofobject": {
+                        "type": "array",
+                        "items": {
+                            "type": "object",
+                            "properties": {
+                                "name": {"type": "string"},
+                                "id": {"type": "integer"}
+                            }
+                        }
+                    }
+                }
+            }
+        ],
+        "expected": {"Test": [{"arrayofobject": [{"name": "anastasiia", "id": 1234},
+                                                 {"name": "olha", "id": 5678}]}]}
     },
     # ----------------------------------------------------------------------------------------------
     {
@@ -1332,6 +1360,7 @@ def _test_parse_structured_data(testapp,
                                 schemas: Optional[List[dict]] = None,
                                 autoadd: Optional[dict] = None,
                                 prune: bool = True,
+                                remove_empty_objects_from_lists: bool = True,
                                 ignore: bool = False,
                                 debug: bool = False) -> None:
 
@@ -1360,12 +1389,13 @@ def _test_parse_structured_data(testapp,
     def assert_parse_structured_data():
 
         def call_parse_structured_data(file: str):
-            nonlocal portal, novalidate, autoadd, prune, debug
+            nonlocal portal, novalidate, autoadd, prune, remove_empty_objects_from_lists, debug
             if debug:
                 # import pdb ; pdb.set_trace()
                 pass
             return parse_structured_data(file=file, portal=portal, novalidate=novalidate,
-                                         autoadd=autoadd, prune=True if prune is not False else False)
+                                         autoadd=autoadd, prune=True if prune is not False else False,
+                                         remove_empty_objects_from_lists=remove_empty_objects_from_lists)
 
         nonlocal file, expected, expected_errors, schemas, noschemas, debug
         portal = Portal(testapp, schemas=schemas) if not noschemas else None  # But see mocked_schemas.
@@ -1511,6 +1541,7 @@ def is_accession(value: str) -> bool:
 # Same as in smaht-portal/../ingestion_processors.py
 def parse_structured_data(file: str, portal: Optional[Union[VirtualApp, TestApp, Portal]], novalidate: bool = False,
                           autoadd: Optional[dict] = None, prune: bool = True,
+                          remove_empty_objects_from_lists: bool = True,
                           ref_nocache: bool = False) -> StructuredDataSet:
 
     def ref_lookup_strategy(portal: Portal, type_name: str, schema: dict, value: str) -> Tuple[int, Optional[str]]:
@@ -1528,6 +1559,7 @@ def ref_lookup_strategy(portal: Portal, type_name: str, schema: dict, value: str
 
     structured_data = StructuredDataSet.load(file=file, portal=portal,
                                              autoadd=autoadd, order=ITEM_INDEX_ORDER, prune=prune,
+                                             remove_empty_objects_from_lists=remove_empty_objects_from_lists,
                                              ref_lookup_strategy=ref_lookup_strategy,
                                              ref_lookup_nocache=ref_nocache)
     if not novalidate: