dcicutils 8.13.3.1b19__tar.gz → 8.13.3.1b22__tar.gz

{dcicutils-8.13.3.1b19 → dcicutils-8.13.3.1b22}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: dcicutils
-Version: 8.13.3.1b19
+Version: 8.13.3.1b22
 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-8.13.3.1b19 → dcicutils-8.13.3.1b22}/dcicutils/misc_utils.py

@@ -1021,21 +1021,22 @@ _MULTIPLIER_SUFFIXES = {
 
 def to_number(value: str,
               allow_commas: bool = False,
-              allow_suffix: bool = False,
-              allow_float: bool = False) -> Optional[Union[int, float]]:
+              allow_multiplier_suffix: bool = False,
+              as_float: bool = False,
+              fallback: Optional[Union[int, float]] = None) -> Optional[Union[int, float]]:
 
     """
-    Converts the give string value to an int, or possibly float if allow_float is True.
+    Converts the give string value to an int, or float if as_float is True.
     If allow_commas is True (default: False) then allow commas (only) every three digits.
-    If allow_suffix is True (default: False) allow any of K, Kb, KB; or M, Mb, MB; or
-    G, Gb, or GB; or T, Tb, TB, to mean multiply value by one thousand; one million;
-    one billion; or one trillion; respectively. If allow_float is True (default: False)
+    If allow_multiplier_suffix is True (default: False) allow any of K, Kb, KB; or M, Mb, MB;
+    or G, Gb, or GB; or T, Tb, TB, to mean multiply value by one thousand; one million;
+    one billion; or one trillion; respectively. If as_float is True (default: False)
     allow the value to be floating point (i.e. with a decimal point and a fractional part),
     in which case the returned value will be of type float rather than int.
     If the string is not well formated then returns None.
     """
     if not (isinstance(value, str) and (value := value.strip())):
-        return value if isinstance(value, (int, float)) else None
+        return value if isinstance(value, (int, float)) else fallback
 
     value_multiplier = 1
     value_negative = False
@@ -1043,37 +1044,36 @@ def to_number(value: str,
 
     if value.startswith("-"):
         if not (value := value[1:].strip()):
-            return None
+            return fallback
         value_negative = True
     elif value.startswith("+"):
         if not (value := value[1:].strip()):
-            return None
+            return fallback
 
-    if allow_suffix is True:
+    if allow_multiplier_suffix is True:
         for suffix in _MULTIPLIER_SUFFIXES:
             if value.endswith(suffix):
                 value_multiplier *= _MULTIPLIER_SUFFIXES[suffix]
                 if not (value := value[:-len(suffix)].strip()):
-                    return None
+                    return fallback
                 break
 
-    if allow_float is True:
-        if dot_index := value.rindex("."):
+    if as_float is True:
+        if (dot_index := value.rfind(".")) >= 0:
             if value_fraction := value[dot_index + 1:].strip():
                 try:
                     value_fraction = float(f"0.{value_fraction}")
                 except Exception:
-                    return None
+                    return fallback
             value = value[:dot_index].strip()
-        pass
 
-    if allow_commas is True:
+    if (allow_commas is True) and ("," in value):
         if not re.fullmatch(r"(-?\d{1,3}(,\d{3})*)", value):
-            return None
+            return fallback
         value = value.replace(",", "")
 
     if not value.isdigit():
-        return None
+        return fallback
 
     result = int(value) * value_multiplier
 
@@ -1083,7 +1083,7 @@ def to_number(value: str,
     if value_negative:
         result = -result
 
-    return result
+    return float(result) if as_float is True else result
 
 
 def to_float(value: str, fallback: Optional[Any] = None) -> Optional[Any]:

{dcicutils-8.13.3.1b19 → dcicutils-8.13.3.1b22}/dcicutils/structured_data.py

@@ -13,7 +13,7 @@ from dcicutils.data_readers import CsvReader, Excel, RowReader
 from dcicutils.datetime_utils import normalize_date_string, normalize_datetime_string
 from dcicutils.misc_utils import (create_dict, create_readonly_object, is_uuid, load_json_if,
                                   merge_objects, remove_empty_properties, right_trim, split_string,
-                                  to_boolean, to_enum, to_float, to_integer, VirtualApp)
+                                  to_boolean, to_enum, to_integer, to_number, VirtualApp)
 from dcicutils.portal_object_utils import PortalObject
 from dcicutils.portal_utils import Portal as PortalBase
 from dcicutils.submitr.progress_constants import PROGRESS_PARSE as PROGRESS
@@ -380,7 +380,8 @@ class StructuredDataSet:
             structured_row = structured_row_template.create_row()
             for column_name, value in row.items():
                 if self._validator_hook:
-                    value, validator_error = self._validator_hook(self, reader.sheet_name, column_name, value)
+                    value, validator_error = (
+                        self._validator_hook(self, type_name, column_name, reader.row_number, value))
                     if validator_error:
                         self._note_error({
                             "src": create_dict(type=schema_name, row=reader.row_number),
@@ -722,13 +723,23 @@ class Schema(SchemaBase):
         return lambda value, src: map_enum(value, typeinfo.get("enum", []), src)
 
     def _map_function_integer(self, typeinfo: dict) -> Callable:
-        def map_integer(value: str, src: Optional[str]) -> Any:
-            return to_integer(value, value, strict=True)
+        allow_commas = typeinfo.get("allow_commas") is True
+        allow_multiplier_suffix = typeinfo.get("allow_multiplier_suffix") is True
+        def map_integer(value: str, src: Optional[str]) -> Any:  # noqa
+            nonlocal allow_commas, allow_multiplier_suffix
+            return to_number(value, fallback=value, as_float=False,
+                             allow_commas=allow_commas,
+                             allow_multiplier_suffix=allow_multiplier_suffix)
         return map_integer
 
     def _map_function_number(self, typeinfo: dict) -> Callable:
-        def map_number(value: str, src: Optional[str]) -> Any:
-            return to_float(value, value)
+        allow_commas = typeinfo.get("allow_commas") is True
+        allow_multiplier_suffix = typeinfo.get("allow_multiplier_suffix") is True
+        def map_number(value: str, src: Optional[str]) -> Any:  # noqa
+            nonlocal allow_commas, allow_multiplier_suffix
+            return to_number(value, fallback=value, as_float=True,
+                             allow_commas=allow_commas,
+                             allow_multiplier_suffix=allow_multiplier_suffix)
         return map_number
 
     def _map_function_string(self, typeinfo: dict) -> Callable:

{dcicutils-8.13.3.1b19 → dcicutils-8.13.3.1b22}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "dcicutils"
-version = "8.13.3.1b19"  # TODO: To become 8.14.0
+version = "8.13.3.1b22"  # TODO: To become 8.14.0
 description = "Utility package for interacting with the 4DN Data Portal and other 4DN resources"
 authors = ["4DN-DCIC Team <support@4dnucleome.org>"]
 license = "MIT"