dcicutils 8.9.0.1b2__py3-none-any.whl → 8.9.0.1b5__py3-none-any.whl

dcicutils/misc_utils.py

@@ -4,6 +4,7 @@ This file contains functions that might be generally useful.
 
 from collections import namedtuple
 import appdirs
+from copy import deepcopy
 import contextlib
 import datetime
 import functools
@@ -2199,28 +2200,58 @@ def merge_key_value_dict_lists(x, y):
     return [key_value_dict(k, v) for k, v in merged.items()]
 
 
-def merge_objects(target: Union[dict, List[Any]], source: Union[dict, List[Any]], full: bool = False) -> dict:
+def merge_objects(target: Union[dict, List[Any]], source: Union[dict, List[Any]],
+                  full: bool = False,  # deprecated
+                  expand_lists: Optional[bool] = None,
+                  primitive_lists: bool = False,
+                  copy: bool = False, _recursing: bool = False) -> Union[dict, List[Any]]:
     """
-    Merges the given source dictionary or list into the target dictionary or list.
-    This MAY well change the given target (dictionary or list) IN PLACE.
-    The the full argument is True then any target lists longer than the
-    source be will be filled out with the last element(s) of the source.
+    Merges the given source dictionary or list into the target dictionary or list and returns the
+    result. This MAY well change the given target (dictionary or list) IN PLACE ... UNLESS the copy
+    argument is True, then the given target will not change as a local copy is made (and returned).
+
+    If the expand_lists argument is True then any target lists longer than the
+    source be will be filled out with the last element(s) of the source; the full
+    argument (is deprecated and) is a synomym for this. The default is False.
+
+    If the primitive_lists argument is True then lists of primitives (i.e. lists in which
+    NONE of its elements are dictionaries, lists, or tuples) will themselves be treated
+    like primitives, meaning the whole of a source list will replace the corresponding
+    target; otherwise they  will be merged normally, meaning each element of a source list
+    will be merged, recursively, into the corresponding target list. The default is False.
     """
+    def is_primitive_list(value: Any) -> bool:  # noqa
+        if not isinstance(value, list):
+            return False
+        for item in value:
+            if isinstance(item, (dict, list, tuple)):
+                return False
+        return True
+
     if target is None:
         return source
+    if expand_lists not in (True, False):
+        expand_lists = full is True
+    if (copy is True) and (_recursing is not True):
+        target = deepcopy(target)
     if isinstance(target, dict) and isinstance(source, dict) and source:
         for key, value in source.items():
-            target[key] = merge_objects(target[key], value, full) if key in target else value
+            if ((primitive_lists is True) and
+                (key in target) and is_primitive_list(target[key]) and is_primitive_list(value)):  # noqa
+                target[key] = value
+            else:
+                target[key] = merge_objects(target[key], value,
+                                            expand_lists=expand_lists, _recursing=True) if key in target else value
     elif isinstance(target, list) and isinstance(source, list) and source:
         for i in range(max(len(source), len(target))):
             if i < len(target):
                 if i < len(source):
-                    target[i] = merge_objects(target[i], source[i], full)
-                elif full:
-                    target[i] = merge_objects(target[i], source[len(source) - 1], full)
+                    target[i] = merge_objects(target[i], source[i], expand_lists=expand_lists, _recursing=True)
+                elif expand_lists is True:
+                    target[i] = merge_objects(target[i], source[len(source) - 1], expand_lists=expand_lists)
             else:
                 target.append(source[i])
-    elif source:
+    elif source not in (None, {}, []):
         target = source
     return target
 

dcicutils/structured_data.py

@@ -74,7 +74,7 @@ class StructuredDataSet:
         self._nrows = 0
         self._autoadd_properties = autoadd if isinstance(autoadd, dict) and autoadd else None
         self._norefs = True if norefs is True else False
-        self._merge = True if merge is True else False
+        self._merge = True if merge is True else False  # New merge functionality (2024-05-25)
         self._debug_sleep = None
         if debug_sleep:
             try:
@@ -347,7 +347,7 @@ class StructuredDataSet:
                 (self._portal.get_schema(schema_name_inferred_from_file_name) is not None)):  # noqa
                 # If the JSON file name looks like a schema name then assume it
                 # contains an object or an array of object of that schema type.
-                if self._merge:
+                if self._merge:  # New merge functionality (2024-05-25)
                     data = self._merge_with_existing_portal_object(data, schema_name_inferred_from_file_name)
                 self._add(Schema.type_name(file), data)
             elif isinstance(data, dict):
@@ -356,7 +356,7 @@ class StructuredDataSet:
                 # which (each property) contains a list of object of that schema type.
                 for schema_name in data:
                     item = data[schema_name]
-                    if self._merge:
+                    if self._merge:  # New merge functionality (2024-05-25)
                         item = self._merge_with_existing_portal_object(item, schema_name)
                     self._add(schema_name, item)
 
@@ -380,8 +380,7 @@ 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)
-            # New merge functionality (2024-05-25).
-            if self._merge:
+            if self._merge:  # New merge functionality (2024-05-25)
                 structured_row = self._merge_with_existing_portal_object(structured_row, schema_name)
             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),
@@ -437,7 +436,7 @@ class StructuredDataSet:
         """
         for identifying_path in self._portal.get_identifying_paths(portal_object, portal_type):
             if existing_portal_object := self._portal.get_metadata(identifying_path, raw=True, raise_exception=False):
-                return merge_objects(existing_portal_object, portal_object)
+                return merge_objects(existing_portal_object, portal_object, primitive_lists=True)
         return portal_object
 
     def _is_ref_lookup_specified_type(ref_lookup_flags: int) -> bool:

{dcicutils-8.9.0.1b2.dist-info → dcicutils-8.9.0.1b5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: dcicutils
-Version: 8.9.0.1b2
+Version: 8.9.0.1b5
 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-8.9.0.1b2.dist-info → dcicutils-8.9.0.1b5.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=zHwsxxEn24muLBP7mDvMa8I9VdMejwW8HMuCL5xbhhw,107690
+dcicutils/misc_utils.py,sha256=-syqTAj8DESiiP_KHoyBv9VvfboFYB03QbBlmXnBZXw,109423
 dcicutils/obfuscation_utils.py,sha256=fo2jOmDRC6xWpYX49u80bVNisqRRoPskFNX3ymFAmjw,5963
 dcicutils/opensearch_utils.py,sha256=V2exmFYW8Xl2_pGFixF4I2Cc549Opwe4PhFi5twC0M8,1017
 dcicutils/portal_object_utils.py,sha256=Az3n1aL-PQkN5gOFE6ZqC2XkYsqiwKlq7-tZggs1QN4,11062
@@ -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=GfG96VyFwQIkmUax-ZdEzbWnfEiJxmeZEaUWz78IQZY,64030
+dcicutils/structured_data.py,sha256=HVe1ruXz0vH4nRBOwq0cNrfR4KtqocC4U940KRXM5zY,64160
 dcicutils/submitr/progress_constants.py,sha256=5bxyX77ql8qEJearfHEvsvXl7D0GuUODW0T65mbRmnE,2895
 dcicutils/submitr/ref_lookup_strategy.py,sha256=VJN-Oo0LLna6Vo2cu47eC-eU-yUC9NFlQP29xajejVU,4741
 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-8.9.0.1b2.dist-info/LICENSE.txt,sha256=qnwSmfnEWMl5l78VPDEzAmEbLVrRqQvfUQiHT0ehrOo,1102
-dcicutils-8.9.0.1b2.dist-info/METADATA,sha256=Mh-FpaQpv_ipkozMQ_Ul_vezgpxdzX4lvp38jaDD8rc,3439
-dcicutils-8.9.0.1b2.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
-dcicutils-8.9.0.1b2.dist-info/entry_points.txt,sha256=51Q4F_2V10L0282W7HFjP4jdzW4K8lnWDARJQVFy_hw,270
-dcicutils-8.9.0.1b2.dist-info/RECORD,,
+dcicutils-8.9.0.1b5.dist-info/LICENSE.txt,sha256=qnwSmfnEWMl5l78VPDEzAmEbLVrRqQvfUQiHT0ehrOo,1102
+dcicutils-8.9.0.1b5.dist-info/METADATA,sha256=YTZR9PRzIbTMoeiRjoZD6mKgIDpZFgQmig1EJXLYoN8,3439
+dcicutils-8.9.0.1b5.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
+dcicutils-8.9.0.1b5.dist-info/entry_points.txt,sha256=51Q4F_2V10L0282W7HFjP4jdzW4K8lnWDARJQVFy_hw,270
+dcicutils-8.9.0.1b5.dist-info/RECORD,,