Property-based checking. Like QuickCheck.
Install the module with: npm install jsverify
var jsc = require("jsverify");
// forall (f : bool -> bool) (b : bool), f (f (f b)) = f(b).
var boolFnAppliedThrice =
jsc.forall("bool -> bool", "bool", function (f, b) {
return f(f(f(b))) === f(b);
});
jsc.assert(boolFnAppliedThrice);
// OK, passed 100 tests
Using jsverify with mocha is easy, just define the properties and use jsverify.assert.
Starting from version 0.4.3 you can write your specs without any boilerplate:
describe("sort", function () {
jsc.property("idempotent", "array nat", function (arr) {
return _.isEqual(sort(sort(arr)), sort(arr));
});
});
Starting from version 0.8.0 you can write the specs in TypeScript. There are typings provided. The drawback is that you cannot use type DSL:
describe("basic jsverify usage", () => {
jsc.property("(b && b) === b", jsc.bool, b => (b && b) === b);
jsc.property("boolean fn thrice", jsc.fn(jsc.bool), jsc.bool, (f, b) =>
f(f(f(b))) === f(b)
);
});
You can also provide --jsverifyRngState state command line argument, to run tests with particular random generator state.
$ mocha examples/nat.js
1) natural numbers are less than 90:
Error: Failed after 49 tests and 1 shrinks. rngState: 074e9b5f037a8c21d6; Counterexample: 90;
$ mocha examples/nat.js --grep 'are less than' --jsverifyRngState 074e9b5f037a8c21d6
1) natural numbers are less than 90:
Error: Failed after 1 tests and 1 shrinks. rngState: 074e9b5f037a8c21d6; Counterexample: 90;
Erroneous case is found with first try.
Check jasmineHelpers.js and jasmineHelpers2.js for jasmine 1.3 and 2.0 respectively.
Testing shows the presence, not the absence of bugs.
Edsger W. Dijkstra
To show that propositions hold, we need to construct proofs. There are two extremes: proof by example (unit tests) and formal (machine-checked) proof. Property-based testing is somewhere in between. We formulate propositions, invariants or other properties we believe to hold, but only test it to hold for numerous (randomly generated) values.
Types and function signatures are written in Coq/Haskell-influenced style:
C# -style List<T> filter(List<T> v, Func<T, bool> predicate) is represented by
filter(v: array T, predicate: T -> bool): array T in our style.
Methods and objects live in jsc object, e.g. shrink.bless method is used by
var jsc = require("jsverify");
var foo = jsc.shrink.bless(...);
Methods starting with .dot are prototype methods:
var arb = jsc.nat;
var arb2 = jsc.nat.smap(f, g);
jsverify can operate with both synchronous and asynchronous-promise properties.
Generally every property can be wrapped inside functor,
for now in either identity or promise functor, for synchronous and promise properties respectively.
forall(arbs: arbitrary a ..., userenv: (map arbitrary)?, prop : a -> property): property
Property constructor
check (prop: property, opts: checkoptions?): result
Run random checks for given prop. If prop is promise based, result is also wrapped in promise.
Options:
opts.tests - test count to run, default 100opts.size - maximum size of generated values, default 50opts.quiet - do not console.logopts.rngState - state string for the rngThe result is true if check succeeds, otherwise it's an object with various fields:
counterexample - an input for which property fails.tests - number of tests run before failing case is foundshrinks - number of shrinks performedexc - an optional exception thrown by property functionrngState - random number generator's state before execution of the propertyassert(prop: property, opts: checkoptions?) : void
Same as check, but throw exception if property doesn't hold.
property(name: string, ...)
Assuming there is globally defined it, the same as:
it(name, function () {
jsc.assert(jsc.forall(...));
}
You can use property to write facts too:
jsc.property("+0 === -0", function () {
return +0 === -0;
});
compile(desc: string, env: typeEnv?): arbitrary a
Compile the type description in provided type environment, or default one.
sampler(arb: arbitrary a, genSize: nat = 10): (sampleSize: nat?) -> a
Create a sampler for a given arbitrary with an optional size. Handy when used in a REPL:
> jsc = require('jsverify') // or require('./lib/jsverify') w/in the project
...
> jsonSampler = jsc.sampler(jsc.json, 4)
[Function]
> jsonSampler()
0.08467432763427496
> jsonSampler()
[ [ [] ] ]
> jsonSampler()
''
> sampledJson(2)
[-0.4199344692751765, false]
throws(block: () -> a, error: class?, message: string?): bool
Executes nullary function block. Returns true if block throws. See assert.throws
assertForall(arbs: arbitrary a ..., userenv: (map arbitrary)?, prop : a -> property): void
Combines 'assert' and 'forall'. Constructs a property with forall from arguments, then throws an exception if the property doesn't hold. Options for 'assert' cannot be set here - use assert(forall(...)) if you need that.
checkForall(arbs: arbitrary a ..., userenv: (map arbitrary)?, prop : a -> property): result
Combines 'check' and 'forall'. Constructs a property with forall from arguments, and returns a value based on if the property holds or not. See 'check' for description of return value.
Options for 'check' cannot be set here - use check(forall(...)) if you need that.
generator a is a function (size: nat) -> a.show is a function a -> string.shrink is a function a -> [a], returning smaller values.arbitrary a is a triple of generator, shrink and show functions.
{ generator: nat -> a, shrink : a -> array a, show: a -> string }We chose to represent generators and shrinks by functions, yet we would like to have additional methods on them. Thus we bless objects with additional properties.
Usually you don't need to bless anything explicitly, as all combinators return blessed values.
See perldoc for bless.
There is a small DSL to help with forall. For example the two definitions below are equivalent:
var bool_fn_applied_thrice = jsc.forall("bool -> bool", "bool", check);
var bool_fn_applied_thrice = jsc.forall(jsc.fn(jsc.bool), jsc.bool, check);
The DSL is based on a subset of language recognized by typify-parser:
"array bool" is evaluated to jsc.array(jsc.bool)."bool -> bool" is evaluated to jsc.fn(jsc.bool)."[nat]" is evaluated to jsc.array(jsc.nat)."bool | nat" is evaluated to jsc.sum([jsc.bool, jsc.nat]).
oneof cannot be shrunk, because the union is untagged, we don't know which shrink to use."bool & nat" is evaluated to jsc.tuple(jsc.bool, jsc.nat)."{ b: bool; n: nat }" is evaluated to jsc.record({ b: jsc.bool, n: jsc.nat })."rec list -> unit | (nat & list)"..smap(f: a -> b, g: b -> a, newShow: (b -> string)?): arbitrary b
Transform arbitrary a into arbitrary b. For example:
g should be a right inverse of f, but doesn't need to be complete inverse.
i.e. i.e. f doesn't need to be invertible, only surjective.
var positiveIntegersArb = nat.smap(
function (x) { return x + 1; },
function (x) { return x - 1; });
var setNatArb = jsc.array(jsc.nat).smap(_.uniq, _.identity);
Right inverse means that f(g(y)) = y for all y in Y. Here Y is a type of arrays of unique natural numbers. For them
_.uniq(_.identity(y)) = _.uniq(y) = y
Opposite: g(f(x)) for all x in X, doesn't need to hold. X is arrays of natural numbers:
_.identity(_uniq([0, 0])) = [0]] != [0, 0]
We need an inverse for shrinking, and there right inverse is enough. We can always pull back smapped value, shrink the preimage, and map or push forward shrunken preimages again.
bless(arb: {...}): arbitrary a
Bless almost arbitrary structure to be proper arbitrary. Note: this function mutates argument.
var arbTokens = jsc.bless({
generator: function () {
switch (jsc.random(0, 2)) {
case 0: return "foo";
case 1: return "bar";
case 2: return "quux";
}
}
});
integer: arbitrary integer
integer(maxsize: nat): arbitrary integer
integer(minsize: integer, maxsize: integer): arbitrary integer
Integers, ℤ
nat: arbitrary nat
nat(maxsize: nat): arbitrary nat
Natural numbers, ℕ (0, 1, 2...)
number: arbitrary number
number(maxsize: number): arbitrary number
number(min: number, max: number): arbitrary number
JavaScript numbers, "doubles", ℝ. NaN and Infinity are not included.
uint8: arbitrary nat
uint16: arbitrary nat
uint32: arbitrary nat
int8: arbitrary integer
int16: arbitrary integer
int32: arbitrary integer
bool: arbitrary bool
Booleans, true or false.
datetime: arbitrary datetime
Random datetime
elements(args: array a): arbitrary a
Random element of args array.
falsy: arbitrary *
Generates falsy values: false, null, undefined, "", 0, and NaN.
constant(x: a): arbitrary a
Returns an unshrinkable arbitrary that yields the given object.
nonshrink(arb: arbitrary a): arbitrary a
Non shrinkable version of arbitrary arb.
unit: arbitrary ()
either(arbA: arbitrary a, arbB : arbitrary b): arbitrary (either a b)
pair(arbA: arbitrary a, arbB : arbitrary b): arbitrary (pair a b)
If not specified a and b are equal to value().
tuple(arbs: (arbitrary a, arbitrary b...)): arbitrary (a, b...)
sum(arbs: (arbitrary a, arbitrary b...)): arbitrary (a | b ...)
dict(arb: arbitrary a): arbitrary (dict a)
Generates a JavaScript object with properties of type A.
array(arb: arbitrary a): arbitrary (array a)
nearray(arb: arbitrary a): arbitrary (array a)
json: arbitrary json
JavaScript Objects: boolean, number, string, null, array of json values or object with json values.
oneof(gs : array (arbitrary a)...) : arbitrary a
Randomly uses one of the given arbitraries.
letrec(
(tie: key -> (arbitrary a | arbitrary b | ...))
-> { key: arbitrary a, key: arbitrary b, ... }):
{ key: arbitrary a, key: arbitrary b, ... }
Mutually recursive definitions. Every reference to a sibling arbitrary
should go through the tie function.
{ arb1, arb2 } = jsc.letrec(function (tie) {
return {
arb1: jsc.tuple(jsc.int, jsc.oneof(jsc.const(null), tie("arb2"))),
arb2: jsc.tuple(jsc.bool, jsc.oneof(jsc.const(null), tie("arb1"))),
}
});
record(spec: { key: arbitrary a... }, userenv: env?): arbitrary { key: a... }
Generates a javascript object with given record spec.
generator.record(gen: { key: generator a... }): generator { key: a... }
shrink.record(shrs: { key: shrink a... }): shrink { key: a... }
char: arbitrary char — Single character
asciichar: arbitrary char — Single ascii character (0x20-0x7e inclusive, no DEL)
string: arbitrary string
nestring: arbitrary string — Generates strings which are not empty.
asciistring: arbitrary string
asciinestring: arbitrary string
fn(arb: arbitrary a): arbitrary (b -> a)fun(arb: arbitrary a): arbitrary (b -> a)generator.small(gen: generator a): generator asmall(arb: arbitrary a): arbitrary aCreate a generator (abitrary) which will generate smaller values, i.e. generator's size parameter is decreased logarithmically.
jsc.property("small array of small natural numbers", "small (array nat)", function (arr) {
return Array.isArray(arr);
});
jsc.property("small array of normal natural numbers", "(small array) nat", function (arr) {
return Array.isArray(arr);
});
suchthat(arb: arbitrary a, userenv: env?, p : a -> bool): arbitrary a
Arbitrary of values that satisfy p predicate. It's advised that p's accept rate is high.A generator function, generator a, is a function (size: nat) -> a, which generates a value of given size.
Generator combinators are auto-curried:
var xs = jsc.generator.array(jsc.nat.generator, 1); // ≡
var ys = jsc.generator.array(jsc.nat.generator)(1);
In purely functional approach generator a would be explicitly stateful computation:
(size: nat, rng: randomstate) -> (a, randomstate).
JSVerify uses an implicit random number generator state,
but the value generation is deterministic (tests are reproducible),
if the primitives from random module are used.
generator.bless(f: nat -> a): generator a
Bless function with .map and .flatmap properties.
.map(f: a -> b): generator b
Map generator a into generator b. For example:
positiveIntegersGenerator = nat.generator.map(
function (x) { return x + 1; });
.flatmap(f: a -> generator b): generator b
Monadic bind for generators. Also flatMap version is supported.
generator.constant(x: a): generator a
generator.combine(gen: generator a..., f: a... -> b): generator b
generator.oneof(gens: list (generator a)): generator a
generator.recursive(genZ: generator a, genS: generator a -> generator a): generator a
generator.pair(genA: generator a, genB: generator b): generator (a, b)
generator.either(genA: generator a, genB: generator b): generator (either a b)
generator.unit: generator ()
unit is an empty tuple, i.e. empty array in JavaScript representation. This is useful as a building block.
generator.tuple(gens: (generator a, generator b...)): generator (a, b...)
generator.sum(gens: (generator a, generator b...)): generator (a | b...)
generator.array(gen: generator a): generator (array a)
generator.nearray(gen: generator a): generator (array a)
generator.dict(gen: generator a): generator (dict a)
A shrink function, shrink a, is a function a -> [a], returning an array of smaller values.
Shrink combinators are auto-curried:
var xs = jsc.shrink.array(jsc.nat.shrink, [1]); // ≡
var ys = jsc.shrink.array(jsc.nat.shrink)([1]);
shrink.bless(f: a -> [a]): shrink a
Bless function with .smap property.
.smap(f: a -> b, g: b -> a): shrink b
Transform shrink a into shrink b. For example:
positiveIntegersShrink = nat.shrink.smap(
function (x) { return x + 1; },
function (x) { return x - 1; });
shrink.noop: shrink a
shrink.pair(shrA: shrink a, shrB: shrink b): shrink (a, b)
shrink.either(shrA: shrink a, shrB: shrink b): shrink (either a b)
shrink.tuple(shrs: (shrink a, shrink b...)): shrink (a, b...)
shrink.sum(shrs: (shrink a, shrink b...)): shrink (a | b...)
shrink.array(shr: shrink a): shrink (array a)
shrink.nearray(shr: shrink a): shrink (nearray a)
show.def(x : a): string
Currently implemented as JSON.stringify.
show.pair(showA: a -> string, showB: b -> string, x: (a, b)): string
show.either(showA: a -> string, showB: b -> string, e: either a b): string
show.tuple(shrinks: (a -> string, b -> string...), x: (a, b...)): string
show.sum(shrinks: (a -> string, b -> string...), x: (a | b ...)): string
show.array(shrink: a -> string, x: array a): string
random(min: int, max: int): int
Returns random int from [min, max] range inclusively.
getRandomInt(2, 3) // either 2 or 3
random.number(min: number, max: number): number
Returns random number from [min, max) range.
either.left(value: a): either a b
either.right(value: b): either a b
either.either(l: a -> x, r: b -> x): x
either.isEqual(other: either a b): bool
TODO: add eq optional parameter
either.bimap(f: a -> c, g: b -> d): either c d
either.bimap(compose(f, g), compose(h, i)) ≡ either.bimap(g, i).bimap(f, h);
either.first(f: a -> c): either c b
either.first(f) ≡ either.bimap(f, utils.identity)
either.second(g: b -> d): either a d
either.second(g) === either.bimap(utils.identity, g)
Utility functions are exposed (and documented) only to make contributions to jsverify more easy. The changes here don't follow semver, i.e. there might be backward-incompatible changes even in patch releases.
Use underscore.js, lodash, ramda, lazy.js or some other utility belt.
utils.isEqual(x: json, y: json): bool
Equality test for json objects.
utils.isApproxEqual(x: a, y: b, opts: obj): bool
Tests whether two objects are approximately and optimistically equal.
Returns false only if they are distinguishable not equal.
Returns true when x and y are NaN.
This function works with cyclic data.
Takes optional 'opts' parameter with properties:
fnEqual - whether all functions are considered equal (default: yes)depth - how deep to recurse until treating as equal (default: 5)utils.force(x: a | () -> a) : a
Evaluate x as nullary function, if it is one.
utils.merge(x... : obj): obj
Merge two objects, a bit like _.extend({}, x, y).
Does JSC originate with JSCheck?
A: Yes
A: Yes. We don't want to give an easy-to-use interface which forgets shrinking altogether. Note, that right inverse is enough, which is most likely easy to write, even complete inverse doesn't exist.
README.md is generated from the source with ljs, say make literate.jsverify.standalone.js is also generated by the build processmake test, yet travis will do it for you.0.8.3 — 2017-09-11 — Updates
suchthat docssuchthat, and type jsc.record.0.8.2 — 2017-04-01 — Typescript updates
tslinted0.8.1 — 2017-03-31 — Typescript updates
0.8.0 — 2017-03-12 — TypeScript typings
0.7.5 — 2017-03-08 — International Women's Day
0.7.4 — 2016-09-07 — Bless suchthat
suchthat are missing .smap"
#1840.7.3 — 2016-08-26 — Remove lodash
lodash. We have our own isNaN now.0.7.2 — 2016-08-25 — One year since the last release
jsc.utils.isEqual returns true if both arguments are NaN.jsc.assertForall and jsc.checkForall0.7.1 — 2015-08-24 — jsc.throws
jsc.throws #1330.7.0 — 2015-08-23 — More experiments
jsc.sum - generate arbitrary sum types (generalisation of either) #125
|) in DSL generates jsc.sumjsc.forall is given zero generators #128jsc.json has shrink #1220.6.3 — 2015-07-27 — Bug fixes
0.6.2 — 2015-07-13 — Trampolines
0.6.1 — 2015-07-13 — Bug fixes
maxsize = 0 for numeric generators works0.6.0 — 2015-06-19 — Minor but major release!
jsc.utils.isApproxEqual0.6.0-beta.2 — 2015-05-31 — Beta!
0.6.0-beta.1 — 2015-05-04 — Beta!
smap documentationflatmap is also flatMapsmall arbitrariesjsc.generator.record0.6.0-alpha.6 — 2015-04-25 — Fix issues #98
int32 values0.6.0-alpha.5 — 2015-04-23 — Fix issue #99
suchthat shrink)0.6.0-alpha.4 — 2015-04-26 — Fix issue #87
0.6.0-alpha.3 — 2015-04-24 — promise shrink fixed
0.6.0-alpha.2 — 2015-04-24 — jsc.bless
jsc.bless0.6.0-alpha.1 — 2015-04-22 — Preview
jsc.map renamed to jsc.dictjsc.value removed, use jsc.jsonjsc.string() removed, use jsc.stringshrink.isomap renamed to shrink.smap0.5.3 — 2015-04-21 — More algebra
unit and either arbitrariesarbitrary.smap to help creating compound data0.5.2 — 2015-04-10 — show.def -change
0.5.1 — 2015-02-19 — Dependencies bump
0.5.0 — 2014-12-24 — Merry Chrismas 2014!
0.5.0-beta.2 — 2014-12-21 — Beta 2!
CONTRIBUTING.mdjsc.elements shrink0.5.0-beta.1 — 2014-12-20 — Beta!
bless don't close over (uses this)0.4.6 — 2014-11-30 — better shrinks & recursive
0.4.5 — 2014-11-22 — stuff
generator.combine & .flatmapnat, integer, number & and string act as objects too0.4.4 — 2014-11-22 — new generators
nearray, nestringgenerator.constantjsc.property (it ∘ assert)jsc.sampler0.4.3 — 2014-11-08 — jsc.property
0.4.2 — 2014-11-03 — User environments for DSL
map, and shrink prototype isomap0.4.1 Move to own organization in GitHub
0.4.0 — 2014-10-27 — typify-dsl & more arbitraries. Changes from 0.3.6:
forall and suchthatoneof behaves as in QuickCheck (BREAKING CHANGE)elements is new name of old oneof0.4.0-beta.4 generator.oneof
0.4.0-beta.3 Expose shrink and show modules
0.4.0-beta.2 Move everything around
0.4.0-beta.1 Beta!
0.4.0-alpha8 oneof & record -dsl support
jsc.compile0.4.0-alpha7 oneof & record
0.4.0-alpha6 more primitives
0.4.0-alpha5 move david to be devDependency
0.4.0-alpha4 more typify
suchthat supports typify dsloneof → elements to be in line with QuickCheck0.4.0-alpha3 David, npm-freeze and jscs
0.4.0-alpha2 Fix typo in readme
0.4.0-alpha1 typify
DSL for forall
var bool_fn_applied_thrice = jsc.forall("bool -> bool", "bool", check);
generator arguments, which are functions are evaluated. One can now write:
jsc.forall(jsc.nat, check) // previously had to be jsc.nat()
0.3.6 map generator
0.3.5 Fix forgotten rngState in console output
0.3.4 Dependencies update
0.3.3 Dependencies update
0.3.2 fun → fn
0.3.1 Documentation typo fixes
0.3.0 Major changes
--jsverifyRngState parameter value used when run on node0.2.0 Use browserify
0.1.4 Mocha test suite
0.1.3 gen.show and exception catching
0.1.2 Added jsc.assert
0.1.1 Use grunt-literate
0.1.0 Usable library
0.0.2 Documented preview
0.0.1 Initial preview
The MIT License (MIT)
Copyright (c) 2013-2015 Oleg Grenrus
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
FAQs
Property-based testing for JavaScript.
The npm package jsverify receives a total of 22,513 weekly downloads. As such, jsverify popularity was classified as popular.
We found that jsverify demonstrated a not healthy version release cadence and project activity because the last version was released a year ago. It has 1 open source maintainer collaborating on the project.
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Security News
Biden's executive order pushes for AI-driven cybersecurity, software supply chain transparency, and stronger protections for federal and open source systems.
Security News
Fluent Assertions is facing backlash after dropping the Apache license for a commercial model, leaving users blindsided and questioning contributor rights.
Research
Security News
Socket researchers uncover the risks of a malicious Python package targeting Discord developers.