data-sitter 0.1.1__py3-none-any.whl → 0.1.3__py3-none-any.whl

data_sitter/Contract.py

@@ -3,6 +3,7 @@ from functools import cached_property
 
 from pydantic import BaseModel
 
+from .Validation import Validation
 from .field_types import BaseField
 from .FieldResolver import FieldResolver
 from .rules import MatchedRule, RuleRegistry, RuleParser
@@ -68,10 +69,13 @@ class Contract:
         return rules
 
     def model_validate(self, item: dict):
-        pydantic_model = self.get_pydantic_model()
-        return pydantic_model.model_validate(item).model_dump()
+        return self.pydantic_model.model_validate(item).model_dump()
 
-    def get_pydantic_model(self) -> BaseModel:
+    def validate(self, item: dict) -> Validation:
+        return Validation.validate(self.pydantic_model, item)
+
+    @cached_property
+    def pydantic_model(self) -> BaseModel:
         return type(self.name, (BaseModel,), {
             "__annotations__": {
                 field_name: field_validator.get_annotation()
@@ -97,5 +101,6 @@ class Contract:
                     ]
                 }
                 for field_name, field_validator in self.field_validators.items()
-            ]
+            ],
+            "values": self.rule_parser.values
         }

data_sitter/FieldResolver.py

@@ -32,7 +32,8 @@ class FieldResolver:
         return matched_rules
 
     def get_field_validator(self, field_name: str, parsed_rules: List[str]) -> BaseField:
-        validator = self.field_class(field_name)
+        is_optional = "Validate Not Null" not in parsed_rules
+        validator = self.field_class(field_name, is_optional)
         matched_rules = self.get_matched_rules(parsed_rules)
         for matched_rule in matched_rules:
             matched_rule.add_to_instance(validator)

data_sitter/Validation.py

@@ -0,0 +1,30 @@
+
+from collections import defaultdict
+from typing import Any, Dict, List, Type
+
+from pydantic import BaseModel, ValidationError
+
+
+class Validation():
+    row: Dict[str, Any]
+    errors: Dict[str, List[str]]
+
+    def __init__(self, row: dict, errors: dict = None):
+        self.row = row
+        self.errors = errors or {}
+
+    def to_dict(self) -> dict:
+        return {"row": self.row, "errors": self.errors}
+    
+    @classmethod
+    def validate(cls, model: Type[BaseModel], item: dict) -> "Validation":
+        try:
+            row = model(**item)  # Validate the row
+            return Validation(row=row.model_dump())
+        except ValidationError as e:
+            errors = defaultdict(list)
+            for error in e.errors():
+                field = error['loc'][0]  # Extract the field name
+                msg = error['msg']
+                errors[field].append(msg)
+            return Validation(row=item, errors=dict(errors))

data_sitter/cli.py

@@ -23,7 +23,7 @@ def main():
     contract_path = Path(args.contract)
     contract_dict = json.loads(contract_path.read_text(encoding))
     contract = Contract.from_dict(contract_dict)
-    pydantic_contract = contract.get_pydantic_model()
+    pydantic_contract = contract.pydantic_model
 
     if file_path.suffix == '.csv':
         with open(file_path, encoding=encoding) as f:

data_sitter/field_types/BaseField.py

@@ -1,12 +1,14 @@
 from abc import ABC
-from typing import Annotated, List, Type
+from typing import Annotated, List, Optional, Type
 
 from pydantic import AfterValidator
 from ..rules import register_rule, register_field
 
 
-def aggregated_validator(validators: List[callable]):
+def aggregated_validator(validators: List[callable], is_optional: bool):
     def _validator(value):
+        if is_optional and value is None:
+            return value
         for validator_func in validators:
             validator_func(value)
         return value
@@ -15,18 +17,22 @@ def aggregated_validator(validators: List[callable]):
 @register_field
 class BaseField(ABC):
     name: str
+    is_optional: bool
     validators = None
     field_type = None
 
-    def __init__(self, name) -> None:
+    def __init__(self, name: str, is_optional: bool) -> None:
         self.name = name
+        self.is_optional = is_optional
         self.validators = []
 
     @register_rule("Validate Not Null")
     def validator_not_null(self):
         def _validator(value):
+            if self.is_optional:
+                return value
             if value is None:
-                raise ValueError()
+                raise ValueError("Value cannot be null")
             return value
 
         self.validators.append(_validator)
@@ -36,7 +42,8 @@ class BaseField(ABC):
             validator(value)
 
     def get_annotation(self):
-        return Annotated[self.field_type, AfterValidator(aggregated_validator(self.validators))]
+        field_type = Optional[self.field_type] if self.is_optional else self.field_type
+        return Annotated[field_type, AfterValidator(aggregated_validator(self.validators, self.is_optional))]
 
     @classmethod
     def get_parents(cls: Type["BaseField"]) -> List[Type["BaseField"]]:

{data_sitter-0.1.1.dist-info → data_sitter-0.1.3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: data-sitter
-Version: 0.1.1
+Version: 0.1.3
 Summary: A Python library that reads data contracts and generates Pydantic models for seamless data validation.
 Author-email: Lázaro Pereira Candea <lazaro@candea.es>
 Requires-Dist: python-dotenv==1.0.1

{data_sitter-0.1.1.dist-info → data_sitter-0.1.3.dist-info}/RECORD

@@ -1,8 +1,9 @@
-data_sitter/Contract.py,sha256=sxzLM8CmZQyelw76lANyGfioL3nGiFMD62e-rfWtkVs,3328
-data_sitter/FieldResolver.py,sha256=RX3JcN8yHWtwb0xuv0sh0ztFI8pRPLlIreXDP8UvXKY,1820
+data_sitter/Contract.py,sha256=E3VYrCQZhGk79coHRTV0hCvLAV8uEhYRBMCnCuF_e48,3494
+data_sitter/FieldResolver.py,sha256=aSavmk3V8QCphLRL6i3T_V2DIsWfEKBdcCnZC71hrx0,1895
+data_sitter/Validation.py,sha256=MHwPMK06XO9YqdWMQMZ0QSFSk7UfkfmE19TmCTUfi3c,940
 data_sitter/__init__.py,sha256=qbE-wU8ELMFwOMG4UTK0lmzn5XF2MK3rc22E8ROgypo,113
-data_sitter/cli.py,sha256=J_FzO8qnF5AVS4PH2AY5ljs4IpCrIg2l2z6PpFsjjWk,1672
-data_sitter/field_types/BaseField.py,sha256=sPUQ49nQaL-4LIMlYFXIyY3at8yNulgXitKaHt2G1u8,1431
+data_sitter/cli.py,sha256=1ICrtokqV5RvvWhzWKAeS5ZUSUpiviQyy2JSK71ER10,1666
+data_sitter/field_types/BaseField.py,sha256=_Pg6a7gdmQFwb4f7LDyOxElX8j0NnTYZGOJJr_jddt8,1797
 data_sitter/field_types/FloatField.py,sha256=pWU449uUFzlpnIpZI-2WxN1YKv7PxIiYe_c7W91VqCc,147
 data_sitter/field_types/IntegerField.py,sha256=o__5z3bg6wsx7FIfJbBYZW5b760-WSZw_05J-OSKXR0,147
 data_sitter/field_types/NumericField.py,sha256=ncPSrUOOgU5MPcA7in_lR6Luqb1xOo8m7UL6TW1qqGw,2665
@@ -18,8 +19,8 @@ data_sitter/rules/Parser/alias_parameters_parser.py,sha256=jsx_JWzkA4lY2nq4hzc4f
 data_sitter/rules/Parser/parser_utils.py,sha256=ypI021uYJTsHAoKGShAfnhd5xQGtqqTGTHozleefsLQ,642
 data_sitter/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 data_sitter/utils/logger_config.py,sha256=w9E4jWfGJnkC9tZz4qrolSqglKm4jEB8l6vjC-qfj8A,1215
-data_sitter-0.1.1.dist-info/METADATA,sha256=uSpkSz85168EciQCs-b37dOzyLcTCybBywRDyjezJLs,324
-data_sitter-0.1.1.dist-info/WHEEL,sha256=52BFRY2Up02UkjOa29eZOS2VxUrpPORXg1pkohGGUS8,91
-data_sitter-0.1.1.dist-info/entry_points.txt,sha256=1I7xxqFZvA78wmDx7NGavttAb8JFWM3Wxgehftx_5C4,53
-data_sitter-0.1.1.dist-info/top_level.txt,sha256=Q7N21PYeqIdRbDvZQCJXhbbv0PFIf876gu1_DpInH_E,12
-data_sitter-0.1.1.dist-info/RECORD,,
+data_sitter-0.1.3.dist-info/METADATA,sha256=cKlUMxk_rNht0G8bHiotyRn7w6Ho6Vila8-0N7yjVhs,324
+data_sitter-0.1.3.dist-info/WHEEL,sha256=52BFRY2Up02UkjOa29eZOS2VxUrpPORXg1pkohGGUS8,91
+data_sitter-0.1.3.dist-info/entry_points.txt,sha256=1I7xxqFZvA78wmDx7NGavttAb8JFWM3Wxgehftx_5C4,53
+data_sitter-0.1.3.dist-info/top_level.txt,sha256=Q7N21PYeqIdRbDvZQCJXhbbv0PFIf876gu1_DpInH_E,12
+data_sitter-0.1.3.dist-info/RECORD,,