DataShape

0.5.2 verified Fri May 01 auth: no python maintenance

A data description language for describing and validating the structure of tabular and array data. Current version is 0.5.2. The library is in maintenance mode; no active development expected.

pip install datashape==0.5.2

Common errors

error TypeError: expected shape ... ↓

cause Using 'dshape' with incompatible numpy dtype or array.

fix

Ensure the numpy array's dtype and shape match the datashape string exactly, e.g., '3 * int32' requires an array of length 3 with dtype int32.

error ImportError: cannot import name 'DataShape' from 'datashape' ↓

cause Trying to import DataShape after installing a newer version that moved it, or incorrect import path.

fix

Use 'from datashape import DataShape' (ensure version 0.5.x). If you have an older version, upgrade: pip install --upgrade datashape.

Warnings

breaking The 'var' syntax for variable-length dimensions changed in 0.5.0. Previously 'var' was used directly, now it must be written as 'var * type'. ↓

fix Update shape strings to use 'var * type', e.g., 'var * int32' instead of 'var int32'.

deprecated The 'datashape.predicates' module is deprecated and may be removed in a future version. Use top-level functions like 'isrecord', 'isscalar', etc., from 'datashape' directly. ↓

fix Replace 'from datashape.predicates import isrecord' with 'from datashape import isrecord'.

gotcha DataShape's 'match()' method returns a boolean, not a validation report. It may raise TypeError if the data type is incompatible. ↓

fix Wrap match() in a try-except block for type validation, or use 'validate()' method if available (not in 0.5.2).

Imports

DataShape
wrong
```
import datashape.DataShape
```
correct
```
from datashape import DataShape
```
DataShape is a class in the top-level module, not a submodule.
dshape
wrong
```
from datashape.predicates import dshape
```
correct
```
from datashape import dshape
```
dshape is a function in the top-level module, not in predicates.

Quickstart

Define a data shape and validate a numpy array against it.

from datashape import dshape, DataShape
# Define a simple shape
ds = dshape('3 * int32')
print(ds)
# Validate data
import numpy as np
arr = np.array([1, 2, 3], dtype=np.int32)
result = ds.match(arr)
print(result)