syrupy 4.6.3

Pytest Snapshot Test Utility

syrupy

Overview

Syrupy is a zero-dependency pytest snapshot plugin. It enables developers to write tests which assert immutability of computed results.

Motivation

The most popular snapshot test plugin compatible with pytest has some core limitations which this package attempts to address by upholding some key values:

• Extensible: If a particular data type is not supported, users should be able to easily and quickly add support.
• Idiomatic: Snapshot testing should fit naturally among other test cases in pytest, e.g. assert x == snapshot vs. snapshot.assert_match(x).
• Soundness: Snapshot tests should uncover even the most minute issues. Unlike other snapshot libraries, Syrupy will fail a test suite if a snapshot does not exist, not just on snapshot differences.

Installation

python -m pip install syrupy

Migration from snapshottest

You cannot use syrupy alongside snapshottest due to argument conflicts. To ease migration, we've made syrupy aware of snapshottest call syntax. Simply uninstall snapshottest and remove old snapshots:

pip uninstall snapshottest -y;
find . -type d ! -path '*/\.*' -name 'snapshots' | xargs rm -r

Pytest and Python Compatibility

Syrupy will always be compatible with the latest version of Python and Pytest. If you're running an old version of Python or Pytest, you will need to use an older major version of Syrupy:

Syrupy Version	Python Support	Pytest Support
4.x.x	>3.8.1	>=7
3.x.x	>=3.7, <4	>=5.1, <8
2.x.x	>=3.6, <4	>=5.1, <8

Usage

Basic Usage

In a pytest test file test_file.py:

def test_foo(snapshot):
    actual = "Some computed value!"
    assert actual == snapshot

when you run pytest, the above test should fail due to a missing snapshot. Re-run pytest with the update snapshots flag like so:

pytest --snapshot-update

A snapshot file should be generated under a __snapshots__ directory in the same directory as test_file.py. The __snapshots__ directory and all its children should be committed along with your test code.

Custom Objects

The default serializer supports all python built-in types and provides a sensible default for custom objects.

Representation

If you need to customise your object snapshot, it is as easy as overriding the default __repr__ implementation.

def __repr__(self) -> str:
    return "MyCustomClass(...)"

If you need bypass a custom object representation to use the amber standard, it is easy using the following helpers.

def test_object_as_named_tuple(snapshot):
    assert snapshot == AmberDataSerializer.object_as_named_tuple(obj_with_custom_repr)

See test_snapshot_object_as_named_tuple_class for an example on automatically doing this for all nested properties

Attributes

If you want to limit what properties are serialized at a class type level you could either:

A. Provide a filter function to the snapshot exclude configuration option.

def limit_foo_attrs(prop, path):
    allowed_foo_attrs = {"do", "not", "serialize", "these", "attrs"}
    return isinstance(path[-1][1], Foo) and prop in allowed_foo_attrs

def test_bar(snapshot):
    actual = Foo(...)
    assert actual == snapshot(exclude=limit_foo_attrs)

B. Provide a filter function to the snapshot include configuration option.

def limit_foo_attrs(prop, path):
    allowed_foo_attrs = {"only", "serialize", "these", "attrs"}
    return isinstance(path[-1][1], Foo) and prop in allowed_foo_attrs

def test_bar(snapshot):
    actual = Foo(...)
    assert actual == snapshot(include=limit_foo_attrs)

C. Or override the __dir__ implementation to control the attribute list.

class Foo:
    def __dir__(self):
        return ["only", "serialize", "these", "attrs"]

def test_bar(snapshot):
    actual = Foo(...)
    assert actual == snapshot

Both options will generate equivalent snapshots but the latter is only viable when you have control over the class implementation and do not need to share the exclusion logic with other objects.

CLI Options

These are the cli options exposed to pytest by the plugin.

Option	Description	Default
--snapshot-update	Snapshots will be updated to match assertions and unused snapshots will be deleted.	False
--snapshot-details	Includes details of unused snapshots (test name and snapshot location) in the final report.	False
--snapshot-warn-unused	Prints a warning on unused snapshots rather than fail the test suite.	False
--snapshot-default-extension	Use to change the default snapshot extension class.	AmberSnapshotExtension
--snapshot-no-colors	Disable test results output highlighting. Equivalent to setting the environment variables ANSI_COLORS_DISABLED or NO_COLOR	Disabled by default if not in terminal.

Assertion Options

These are the options available on the snapshot assertion fixture. Use of these options are one shot and do not persist across assertions. For more persistent options see advanced usage.

matcher

This allows you to match on a property path and value to control how specific object shapes are serialized.

The matcher is a function that takes two keyword arguments. It should return the replacement value to be serialized or the original unmutated value.

Argument	Description
data	Current serializable value being matched on
path	Ordered path traversed to the current value e.g. (("a", dict), ("b", dict)) from { "a": { "b": { "c": 1 } } }}

NOTE: Do not mutate the value received as it could cause unintended side effects.

Built-In Matchers

Syrupy comes with built-in helpers that can be used to make easy work of using property matchers.

path_type(mapping=None, *, types=(), strict=True, regex=False)

Easy way to build a matcher that uses the path and value type to replace serialized data. When strict, this will raise a ValueError if the types specified are not matched.

Argument	Description
mapping	Dict of path string to tuples of class types, including primitives e.g. (MyClass, UUID, datetime, int, str)
types	Tuple of class types used if none of the path strings from the mapping are matched
strict	If a path is matched but the value at the path does not match one of the class types in the tuple then a PathTypeError is raised
regex	If true, the mapping key is treated as a regular expression when matching paths
replacer	Called with any matched value and result is used as the replacement that is serialized. Defaults to the object type when not given

from syrupy.matchers import path_type

def test_bar(snapshot):
    actual = {
      "date_created": datetime.now(),
      "value": "Some computed value!!",
    }
    assert actual == snapshot(matcher=path_type({
      "date_created": (datetime,),
      "nested.path.id": (int,),
    }))

# name: test_bar
  dict({
    'date_created': datetime,
    'value': 'Some computed value!!',
  })
# ---

NOTE: When regex is True all matcher mappings are treated as regex patterns

path_value(mapping=None, *, **kwargs)

Shares the same kwargs as path_type matcher, with the exception of the mapping argument type. Only runs replacement for objects at a matching path where the value of the mapping also matches the object data string repr.

Argument	Description
mapping	Dict of path string to object value string representations

See test_regex_matcher_str_value for example usage.

exclude

This allows you to filter out object properties from the serialized snapshot.

The exclude parameter takes a filter function that accepts two keyword arguments. It should return true if the property should be excluded, or false if the property should be included.

Argument	Description
prop	Current property on the object, could be any hashable value that can be used to retrieve a value e.g. 1, "prop_str", SomeHashableObject
path	Ordered path traversed to the current value e.g. (("a", dict), ("b", dict)) from { "a": { "b": { "c": 1 } } }}

Built-In Filters

Syrupy comes with built-in helpers that can be used to make easy work of using the filter options.

props(prop_name, *prop_name)

Easy way to build a filter that excludes based on string based property names.

Takes an argument list of property names, with support for indexed iterables.

from syrupy.filters import props

def test_bar(snapshot):
    actual = {
      "id": uuid.uuid4(),
      "list": [1,2,3],
    }
    assert actual == snapshot(exclude=props("id", "1"))

# name: test_bar
  dict({
    'list': list([
      1,
      3,
    ]),
  })
# ---

paths(path_string, *path_strings)

Easy way to build a filter that uses full path strings delimited with ..

Takes an argument list of path strings.

from syrupy.filters import paths

def test_bar(snapshot):
    actual = {
      "date": datetime.now(),
      "list": [1,2,3],
    }
    assert actual == snapshot(exclude=paths("date", "list.1"))

# name: test_bar
  dict({
    'list': list([
      1,
      3,
    ]),
  })
# ---

include

This allows you filter an object's properties to a subset using a predicate. This is the opposite of exclude. All the same property filters supporterd by exclude are supported for include.

The include parameter takes a filter function that accepts two keyword arguments. It should return true if the property should be include, or false if the property should not be included.

Argument	Description
prop	Current property on the object, could be any hashable value that can be used to retrieve a value e.g. 1, "prop_str", SomeHashableObject
path	Ordered path traversed to the current value e.g. (("a", dict), ("b", dict)) from { "a": { "b": { "c": 1 } } }}

Note that include has some caveats which make it a bit more difficult to use than exclude. Both include and exclude are evaluated for each key of an object before traversing down nested paths. This means if you want to include a nested path, you must include all parents of the nested path, otherwise the nested child will never be reached to be evaluated against the include predicate. For example:

obj = {
    "nested": { "key": True }
}
assert obj == snapshot(include=paths("nested", "nested.key"))

The extra "nested" is required, otherwise the nested dictionary will never be searched -- it'd get pruned too early.

To avoid adding duplicate path parts, we provide a convenient paths_include which supports a list/tuple instead of a string for each path to match:

obj = {
    "other": False,
    "nested": { "key": True }
}
assert obj == snapshot(include=paths_include(["other"], ["nested", "key"]))

extension_class

This is a way to modify how the snapshot matches and serializes your data in a single assertion.

def test_foo(snapshot):
    actual_svg = "<svg></svg>"
    assert actual_svg == snapshot(extension_class=SVGImageSnapshotExtension)

diff

This is an option to snapshot only the diff between the actual object and a previous snapshot, with the diff argument being the previous snapshot index/name.

def test_diff(snapshot):
    actual0 = [1,2,3,4]
    actual1 = [0,1,3,4]

    assert actual0 == snapshot
    assert actual1 == snapshot(diff=0)
    # This is equivalent to the lines above
    # Must use the index name to diff when given
    assert actual0 == snapshot(name='snap_name')
    assert actual1 == snapshot(diff='snap_name')

Built-In Extensions

Syrupy comes with a few built-in preset configurations for you to choose from. You should also feel free to extend the AbstractSyrupyExtension if your project has a need not captured by one our built-ins.

• AmberSnapshotExtension: This is the default extension which generates .ambr files. Serialization of most data types are supported.
  • Line control characters are normalised when snapshots are generated i.e. \r and \n characters are all written as \n. This is to allow interoperability of snapshots between operating systems that use disparate line control characters.
• SingleFileSnapshotExtension: Unlike the AmberSnapshotExtension, which groups all tests within a single test file into a singular snapshot file, this extension creates one .raw file per test case.
• PNGImageSnapshotExtension: An extension of single file, this should be used to produce .png files from a byte string.
• SVGImageSnapshotExtension: Another extension of single file. This produces .svg files from an svg string.
• JSONSnapshotExtension: Another extension of single file. This produces .json files from dictionaries and lists.

name

By default, if you make multiple snapshot assertions within a single test case, an auto-increment identifier will be used to index the snapshots. You can override this behaviour by specifying a custom snapshot name to use in place of the auto-increment number.

def test_case(snapshot):
    assert "actual" == snapshot(name="case_a")
    assert "other" == snapshot(name="case_b")

Warning: If you use a custom name, you must make sure the name is not re-used within a test case.

Advanced Usage

By overriding the provided AbstractSyrupyExtension you can implement varied custom behaviours.

See examples of how syrupy can be used and extended in the test examples.

Overriding defaults

It is possible to override include, exclude, matchers and extension_class on a more global level just once, instead of every time per test. By default, after every assertion the modified values per snapshot assert are reverted to their default values. However, it is possible to override those default values with ones you would like persisted, which will be treated as the new defaults.

To achieve that you can use snapshot.with_defaults, which will create new instance of SnapshotAssertion with the provided values.

snapshot.use_extension is retained for compatibility. However, it is limited to only overriding the default extension class.

JSONSnapshotExtension

This extension can be useful when testing API responses, or when you have to deal with long dictionaries that are cumbersome to validate inside a test. For example:

import pytest

from syrupy.extensions.json import JSONSnapshotExtension

@pytest.fixture
def snapshot_json(snapshot):
    return snapshot.with_defaults(extension_class=JSONSnapshotExtension)
    # or return snapshot.use_extension(JSONSnapshotExtension)


def test_api_call(client, snapshot_json):
    resp = client.post("/endpoint")
    assert resp.status_code == 200
    assert snapshot_json == resp.json()

API responses often contain dynamic data, like IDs or dates. You can still validate and store other data of a response by leveraging syrupy matchers. For example:

from datetime import datetime

from syrupy.matchers import path_type

def test_api_call(client, snapshot_json):
    resp = client.post("/user", json={"name": "Jane"})
    assert resp.status_code == 201

    matcher = path_type({
      "id": (int,),
      "registeredAt": (datetime,)
    })
    assert snapshot_json(matcher=matcher) == resp.json()

The generated snapshot:

{
  "id": "<class 'int'>",
  "registeredAt": "<class 'datetime'>",
  "name": "Jane"
}

Or a case where the value needs to be replaced using a condition e.g. file path string

import re

from syrupy.matchers import path_type

def test_matches_generated_string_value(snapshot, tmp_file):
    matcher = path_value(
        mapping={"file_path": r"\w+://(.*/)+dir/filename.txt"},
        replacer=lambda _, match: match[0].replace(match[1], "<tmp-file-path>/"),
        types=(str,),
    )

    assert snapshot(matcher=matcher) == tmp_file

The generated snapshot:

{
  "name": "Temp Files",
  "file_path": "scheme://<tmp-file-path>/dir/filename.txt"
}

Extending Syrupy

• Custom defaults
• Custom snapshot directory 1
• Custom snapshot directory 2
• Custom snapshot name
• Custom object snapshots
• Custom comparator
• JPEG image extension
• Built-in image extensions

Uninstalling

pip uninstall syrupy

If you have decided not to use Syrupy for your project after giving us a try, we'd love to get your feedback. Please create a GitHub issue if applicable.

Known Limitations

• pytest-xdist support only partially exists. There is no issue when it comes to reads however when you attempt to run pytest --snapshot-update, if running with more than 1 process, the ability to detect unused snapshots is disabled. See #535 for more information.

We welcome contributions to patch these known limitations.

Contributing

Feel free to open a PR or GitHub issue. Contributions welcome!

To develop locally, clone this repository and run . script/bootstrap to install test dependencies. You can then use invoke --list to see available commands.

See contributing guide

Contributors

Noah 🚇 🤔 💻 📖 ⚠️	Emmanuel Ogbizi 💻 🎨 🚇 📖 ⚠️	Adam Lazzarato 📖	Marc Cataford 💻 ⚠️	Michael Rose 💻 ⚠️	Jimmy Jia 💻 ⚠️	Steven Loria 🚇
Artur Balabanov 💻	Huon Wilson 💻 🐛	Elizabeth Culbertson 💻 ⚠️	Joakim Nordling 🐛	Ouail 💻	Denis 💻	N0124 💻
dtczest 🐛	Eddie Darling 📖	darrenburns 📖	Magnus Heskestad Waage 🐛	Herbert Ho 🐛	Tolga Eren 🐛	John Kurkowski 🐛
Atharva Arya 💻	Michał Jelonek 💻	ManiacDC 💻	Dmitry Dygalo 📖	Allan Chain 🐛

This section is automatically generated via tagging the all-contributors bot in a PR:

@all-contributors please add <username> for <contribution type>

License

Syrupy is licensed under Apache License Version 2.0.