Avro Validator

A pure python avro schema validator.

The default avro library for Python provide validation of data against the schema, the problem is that the output of this validation doesn't provide information about the error. All you get is the the datum is not an example of the schema error message.

When working with bigger avro schemas, sometimes is not easy to visually find the field that has an issue.

This library provide clearer exceptions when validating data against the avro schema, in order to be easier to identify the field that is not compliant with the schema and the problem with that field.

Installing

Install using pip:

$ pip install -U avro_validator

Validating data against Avro schema

The validator can be used as a console application. It receives a schema file, and a data file, validating the data and returning the error message in case of failure.

The avro_validator can also be used as a library in python code.

Console usage

In order to validate the data_to_validate.json file against the schema.avsc using the avro_validator callable, just type:

$ avro_validator schema.avsc data_to_valdate.json
OK

Since the data is valid according to the schema, the return message is OK.

Error validating the data

If the data is not valid, the program returns an error message:

$ avro_validator schema.avsc data_to_valdate.json
Error validating value for field [data,my_boolean_value]: The value [123] is not from one of the following types: [[NullType, BooleanType]]

This message indicates that the field my_boolean_value inside the data dictionary has value 123, which is not compatible with the bool type.

Command usage

It is possible to get information about usage of the avro_validator using the help:

$ avro_validator -h

Library usage

Using schema file

When using the avr_validator as a library, it is possible to pass the schema as a file:

from avro_validator.schema import Schema

schema_file = 'schema.avsc'

schema = Schema(schema_file)
parsed_schema = schema.parse()

data_to_validate = {
    'name': 'My Name'
}

parsed_schema.validate(data_to_validate)

In this example, if the data_to_validate is valid according to the schema, then the parsed_schema.validate(data_to_validate) call will return True.

Using a dict as schema

It is also possible to provide the schema as a json string:

import json
from avro_validator.schema import Schema

schema = json.dumps({
    'name': 'test schema',
    'type': 'record',
    'doc': 'schema for testing avro_validator',
    'fields': [
        {
            'name': 'name',
            'type': 'string'
        }
    ]
})

schema = Schema(schema)
parsed_schema = schema.parse()

data_to_validate = {
    'name': 'My Name'
}

parsed_schema.validate(data_to_validate)

In this example, the parsed_schema.validate(data_to_validate) call will return True, since the data is valid according to the schema.

Invalid data

If the data is not valid, the parsed_schema.validate will raise a ValueError, with the message containing the error description.

import json
from avro_validator.schema import Schema

schema = json.dumps({
    'name': 'test schema',
    'type': 'record',
    'doc': 'schema for testing avro_validator',
    'fields': [
        {
            'name': 'name',
            'type': 'string',
            'doc': 'Field that stores the name'
        }
    ]
})

schema = Schema(schema)
parsed_schema = schema.parse()

data_to_validate = {
    'my_name': 'My Name'
}

parsed_schema.validate(data_to_validate)

The schema defined expects only one field, named name, but the data contains only the field name_2, making it invalid according to the schema. In this case, the validate method will return the following error:

Traceback (most recent call last):
  File "/Users/leonardo.almeida/.pyenv/versions/avro_validator_venv/lib/python3.7/site-packages/IPython/core/interactiveshell.py", line 3326, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "<ipython-input-3-a5e8ce95d21c>", line 23, in <module>
    parsed_schema.validate(data_to_validate)
  File "/opt/dwh/avro_validator/avro_validator/avro_types.py", line 563, in validate
    raise ValueError(f'The fields from value [{value}] differs from the fields '
ValueError: The fields from value [{'my_name': 'My Name'}] differs from the fields of the record type [{'name': RecordTypeField <name: name, type: StringType, doc: Field that stores the name, default: None, order: None, aliases: None>}]

The message detailed enough to enable the developer to pinpoint the error in the data.

Invalid schema

If the schema is not valid according to avro specifications, the parse method will also return a ValueError.

import json
from avro_validator.schema import Schema

schema = json.dumps({
    'name': 'test schema',
    'type': 'record',
    'doc': 'schema for testing avro_validator',
    'fields': [
        {
            'name': 'name',
            'type': 'invalid_type',
            'doc': 'Field that stores the name'
        }
    ]
})

schema = Schema(schema)
parsed_schema = schema.parse()

Since the schema tries to define the name field as invalid_type, the schema declaration is invalid, thus the following exception will be raised:

Traceback (most recent call last):
  File "/Users/leonardo.almeida/.pyenv/versions/avro_validator_venv/lib/python3.7/site-packages/IPython/core/interactiveshell.py", line 3326, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "<ipython-input-2-7f3f77000f08>", line 18, in <module>
    parsed_schema = schema.parse()
  File "/opt/dwh/avro_validator/avro_validator/schema.py", line 28, in parse
    return RecordType.build(schema)
  File "/opt/dwh/avro_validator/avro_validator/avro_types.py", line 588, in build
    record_type.__fields = {field['name']: RecordTypeField.build(field) for field in json_repr['fields']}
  File "/opt/dwh/avro_validator/avro_validator/avro_types.py", line 588, in <dictcomp>
    record_type.__fields = {field['name']: RecordTypeField.build(field) for field in json_repr['fields']}
  File "/opt/dwh/avro_validator/avro_validator/avro_types.py", line 419, in build
    field.__type = cls.__build_field_type(json_repr)
  File "/opt/dwh/avro_validator/avro_validator/avro_types.py", line 401, in __build_field_type
    raise ValueError(f'Error parsing the field [{fields}]: {actual_error}')
ValueError: Error parsing the field [name]: The type [invalid_type] is not recognized by Avro

The message is clearly indicating that the the invalid_type is not recognized by avro.