Vladiate helps you write explicit assertions for every field of your CSV file.
- Write validation schemas in plain-old Python
- No UI, no XML, no JSON, just code.
- Write your own validators
- Vladiate comes with a few by default, but there's no reason you can't write your own.
- Validate multiple files at once
- Either with the same schema, or different ones.
Installing:
$ pip install vladiate
Below is an example of a vladfile.py
from vladiate import Vlad
from vladiate.validators import UniqueValidator, SetValidator
from vladiate.inputs import LocalFile
class YourFirstValidator(Vlad):
source = LocalFile('vampires.csv')
validators = {
'Column A': [
UniqueValidator()
],
'Column B': [
SetValidator(['Vampire', 'Not A Vampire'])
]
}
Here we define a number of validators for a local file vampires.csv
,
which would look like this:
Column A,Column B Vlad the Impaler,Not A Vampire Dracula,Vampire Count Chocula,Vampire
We then run vladiate
in the same directory as your .csv
file:
$ vladiate
And get the following output:
Validating YourFirstValidator(source=LocalFile('vampires.csv')) Passed! :)
Let's imagine that you've gotten a new CSV file,
potential_vampires.csv
, that looks like this:
Column A,Column B Vlad the Impaler,Not A Vampire Dracula,Vampire Count Chocula,Vampire Ronald Reagan,Maybe A Vampire
If we were to update our first validator to use this file as follows:
- class YourFirstValidator(Vlad): - source = LocalFile('vampires.csv') + class YourFirstFailingValidator(Vlad): + source = LocalFile('potential_vampires.csv')
we would get the following error:
Validating YourFirstFailingValidator(source=LocalFile('potential_vampires.csv')) Failed :( SetValidator failed 1 time(s) (25.0%) on field: 'Column B' Invalid fields: ['Maybe A Vampire']
And we would know that we'd either need to sanitize this field, or add
it to the SetValidator
.
To make writing a new vladfile.py
easy, Vladiate will give
meaningful error messages.
Given the following as real_vampires.csv
:
Column A,Column B,Column C Vlad the Impaler,Not A Vampire Dracula,Vampire Count Chocula,Vampire Ronald Reagan,Maybe A Vampire
We could write a bare-bones validator as follows:
class YourFirstEmptyValidator(Vlad):
source = LocalFile('real_vampires.csv')
validators = {}
Running this with vladiate
would give the following error:
Validating YourFirstEmptyValidator(source=LocalFile('real_vampires.csv')) Missing... Missing validators for: 'Column A': [], 'Column B': [], 'Column C': [],
Vladiate expects something to be specified for every column, even if it
is an empty list (more on this later). We can easily copy and paste
from the error into our vladfile.py
to make it:
class YourFirstEmptyValidator(Vlad):
source = LocalFile('real_vampires.csv')
validators = {
'Column A': [],
'Column B': [],
'Column C': [],
}
When we run this with vladiate
, we get:
Validating YourSecondEmptyValidator(source=LocalFile('real_vampires.csv')) Failed :( EmptyValidator failed 4 time(s) (100.0%) on field: 'Column A' Invalid fields: ['Dracula', 'Vlad the Impaler', 'Count Chocula', 'Ronald Reagan'] EmptyValidator failed 4 time(s) (100.0%) on field: 'Column B' Invalid fields: ['Maybe A Vampire', 'Not A Vampire', 'Vampire'] EmptyValidator failed 4 time(s) (100.0%) on field: 'Column C' Invalid fields: ['Real', 'Not Real']
This is because Vladiate interprets an empty list of validators for a
field as an EmptyValidator
, which expects an empty string in every
field. This helps us make meaningful decisions when adding validators to
our vladfile.py
. It also ensures that we are not forgetting about a
column or field which is not empty.
Vladiate comes with a few common validators built-in:
class Validator
Generic validator. Should be subclassed by any custom validators. Not to be used directly.
class CastValidator
Generic "can-be-cast-to-x" validator. Should be subclassed by any cast-test validator. Not to be used directly.
class IntValidator
Validates whether a field can be cast to an
int
type or not.
empty_ok=False
:Specify whether a field which is an empty string should be ignored.
class FloatValidator
Validates whether a field can be cast to an
float
type or not.
empty_ok=False
:Specify whether a field which is an empty string should be ignored.
class SetValidator
Validates whether a field is in the specified set of possible fields.
valid_set=[]
:List of valid possible fields empty_ok=False
:Implicity adds the empty string to the specified set. ignore_case=False
:Ignore the case between values in the column and valid set
class UniqueValidator
Ensures that a given field is not repeated in any other column. Can optionally determine "uniqueness" with other fields in the row as well via
unique_with
.
unique_with=[]
:List of field names to make the primary field unique with. empty_ok=False
:Specify whether a field which is an empty string should be ignored.
class RegexValidator
Validates whether a field matches the given regex using re.match().
pattern=r'di^'
:The regex pattern. Fails for all fields by default. full=False
:Specify whether we should use a fullmatch() or match(). empty_ok=False
:Specify whether a field which is an empty string should be ignored.
class RangeValidator
Validates whether a field falls within a given range (inclusive). Can handle integers or floats.
low
:The low value of the range. high
:The high value of the range. empty_ok=False
:Specify whether a field which is an empty string should be ignored.
class EmptyValidator
Ensure that a field is always empty. Essentially the same as an empty
SetValidator
. This is used by default when a field has no
validators.
class NotEmptyValidator
The opposite of an EmptyValidator
. Ensure that a field is never empty.
class Ignore
Always passes validation. Used to explicity ignore a given column.
class RowValidator
Generic row validator. Should be subclassed by any custom validators. Not to be used directly.
class RowLengthValidator
Validates that each row has the expected number of fields. The expected
number of fields is inferred from the CSV header row read by
csv.DictReader
.
Vladiate comes with the following input types:
class VladInput
Generic input. Should be subclassed by any custom inputs. Not to be used directly.
class LocalFile
Read from a file local to the filesystem.
filename
:Path to a local CSV file.
class S3File
Read from a file in S3. Optionally can specify either a full path, or a bucket/key pair.
Requires the boto library, which should be installed via
pip install vladiate[s3]
.
path=None
:A full S3 filepath (e.g., s3://foo.bar/path/to/file.csv
)bucket=None
:S3 bucket. Must be specified with a key
.key=None
:S3 key. Must be specified with a bucket
.
class String
Read CSV from a string. Can take either an
str
or aStringIO
.
- :
string_input=None
- Regular Python string input.
- :
string_io=None
StringIO
input.
class Vlad
Initialize a Vlad programatically
source
:Required. Any VladInput. validators={}
:List of validators. Optional, defaults to the class variable validators if set, otherwise uses EmptyValidator for all fields. row_validators=[]
:List of row-level validators. Validators provided here operate on entire rows and can be used to define constraints that involve more than one field. Optional, defaults to the class variable row_validators if set, otherwise [], which does not perform any row-level validation. delimiter=','
:The delimiter used within your csv source. Optional, defaults to ,. ignore_missing_validators=False
:Whether to fail validation if there are fields in the file for which the Vlad does not have validators. Optional, defaults to False. quiet=False
:Whether to disable log output generated by validations. Optional, defaults to False. file_validation_failure_threshold=None
:Stops validating the file after this failure threshold is reached. Input a value between 0.0 and 1.0. 1.0`(100%) validates the entire file. Optional, defaults to `None. For example:
from vladiate import Vlad
from vladiate.inputs import LocalFile
Vlad(source=LocalFile('path/to/local/file.csv')).validate()
To run the tests:
make test
To run the linter:
make lint
Usage: vladiate [options] [VladClass [VladClass2 ... ]]
Options:
-h, --help show this help message and exit
-f VLADFILE, --vladfile=VLADFILE
Python module file to import, e.g. '../other.py'.
Default: vladfile
-l, --list Show list of possible vladiate classes and exit
-V, --version show version number and exit
-p PROCESSES, --processes=PROCESSES
attempt to use this number of processes, Default: 1
-q, --quiet disable console log output generated by validations
- Dustin Ingram
- Clara Bennett
- Aditya Natraj
- Sterling Petersen
- Aleix
- Bob Lannon
- Santi
- David Park
- Jon Banafato
- haritha-ravi
Open source MIT license.