Pydantic Validators

Read a good introductions to validators in Pydantic here. Even though Pydantic is gonna deprecate validator and root_valiator decerators in v3, but it’s still widely used, like LangChain still stick with Pydantic v1.

1 validator

Validator can take two arguments, first being value, which is the value of one field. The values is a dictionary containing all previous validated fields in the model.

class Student(BaseModel):
    id: uuid.UUID
    student_name: str = Field(alias="name")
    GPA: confloat(ge=0, le=4)

    @validator('GPA')
    def validate_gpa(cls, value, values):
        ## will print id and student_name
        print(values)
        ## will return GPA value
        return value

2 root_validator

In case you want to refer to some fields which is not listed before, so you can NOT get from the values in validator. Root validators are a solution to this. These run validation on the entire model’s data, after the validator functions have run for each individual field.

  @root_validator
  def validate_gpa(cls, values):
    ## values holds ALL filed values
    valid_gpa = values.get('GPA') >= 3.5
    if not valid_gpa:
      raise ValueError("GPA low")
    return values

3 Pre-validator

Pydantic’s default field validation occurs before custom @validator functions are called. So the data type check is happening before any validators

class Student(BaseModel):
  ...
  tags: list[str]

  @valiador('tags')
  def split_tags(cls, value):
        return value.split(",")

stu = Student(tags="a,b,c",...)

This would leads to following error due to ‘tags’ expects a List.
This is can be fixed by adding pre=True argument

@validator('tags', pre=True)
def split_tags(cls, value):
    #split a string into List
    return value.split(",")

### OR we can use roo_validator ###
@root_validator(pre=True)
# This runs FIRST, values['tags']=="a,b,c"
def validate_all(cls, values):
    print("-->root_validator FIRST")
    print(cls, values)
    values['tags']=values['tags'].split(',')
    return values

@root_validator()
# This runs LAST, values['tags']==[a,b,c]
def validate_all(cls, values):
    print("-->root_validator LAST")
    print(cls, values)
    return values

Use ‘pre=True’ for root_validator will move it above all validator, including fields validators

4 Per-Item Validators

Use the ‘tags’ example, if we want to apply validator to each item in the list, we can use each_item argument.

@validator('tags', each_item=True)
## remove_slackers will be called MULTIPLE times
## each time, value is an element from the tags list
def remove_slackers(cls, value):
    if value == 'slacker':
        raise ValueError("Student is a slacker and cannot be enrolled!")
    return value

So each_item argument controls the type of value is element or the whole List

@validatorv1('modules',each_item=False)
#validate_modules will ONLY called ONCE
# The whole list is in value (type List)
def validate_modules(cls, value, values):
    print(value, type(value), values)
    if type(value)==list:
        for v in value:
            print(v)
    return value