This scenario is very common, and inevitably will make the test harder to read, increasing the potential for misunderstanding the intent, which in turn will increase the likelihood of mistakes being made when the test or related code is changed. Before we see how to improve the test, we will learn about the let helper:
describe AddressValidator do
let(:address) { {street: "123 Any Street", city: "Anytown"} }
it "valid? returns false for incomplete address" do
expect(AddressValidator.valid?(address)).to eq(false)
end
it "missing_parts returns an array of missing required parts" do
expect(
AddressValidator.missing_parts(address)
).to eq([:region, :postal_code, :country])
end
context "invalid characters in value" do
let(:address){ {street: "123 Any Street", city: "Any$town%"} }
it "invalid_parts returns keys with invalid values" do
expect(
AddressValidator.invalid_parts(address)
).to eq([:city])
end
end
end
The syntax is simple. The argument for the let helper is the name of the variable to be created, which you can reference within the same context as you would a local variable or function. In this case the argument is :address. The let helper also requires a block, that is is evaluated dynamically at runtime to provide the value for the object. In this case, we just supply a Hash for the value of address.
The secret to let is lambdas, or anonymous functions, which are evaluated at the moment they are called. Our first implementation doesn't show the power of lambdas. Using lambdas is what makes let so flexible and effective. Instead of a single, static let definition, we can create separate definitions for street and city, and reference them in address, allowing us to change inpidual parts of address as needed:
describe AddressValidator do
let(:address) { { street: street, city: city } }
let(:street) { "123 Any Street" }
let(:city) { "Anytown" }
it "valid? returns false for incomplete address" do
expect(AddressValidator.valid?(address)).to eq(false)
end
it "missing_parts returns an array of missing required parts" do
expect(
AddressValidator.missing_parts(address)
).to eq([:region, :postal_code, :country])
end
context "invalid characters in value" do
let(:city) { "Any$town%" }
it "invalid_parts returns keys with invalid values" do
expect(
AddressValidator.invalid_parts(address)
).to eq([:city])
end
end
end
Now we have a test case that clearly shows the differences in address, making the intent of the test case crystal clear. Using this pattern, we can add more cases with ease. We can change one or more nested values (for example city) or redefine address entirely:
describe AddressValidator do
# notice that 'address' is defined as a Hash here
let(:address) { { street: street, city: city } }
let(:street) { "123 Any Street" }
let(:city) { "Anytown" }
it "valid? returns false for incomplete address" do
expect(AddressValidator.valid?(address)).to eq(false)
end
context "address contains invalid characters" do
# here we've redefined 'address' to be a String let(:address) { "$123% A^ St., Anytown, CA, USA 12345" }
it "valid? returns false for incomplete address" do
expect(AddressValidator.valid?(address)).to eq(false)
end
end
context "address is a String" do
let(:address) { "123 Any St., Anytown" }
it "valid? returns false for incomplete address" do
expect(AddressValidator.valid?(address)).to eq(false)
end
end
context "complete address" do
# we define 'address' as a Hash, but with all values
let(:address) do { street: "123 Any Street", city: "Anytown", region: "Anyplace", country: "Anyland", postal_code: "123456" } end
it "valid? returns true" do
expect(AddressValidator.valid?(address)).to eq(true)
end
context "address is a String" do
let(:address) { "123 Any St., Anytown, CA, USA, 12345" }
it "valid? returns true" do
expect(AddressValidator.valid?(address)).to eq(true)
end
end
end
end
We've just seen how let is a simple but effective tool for organizing tests, making them easy to understand and to maintain. We've also started using context in order to organize our tests. In fact, context is just an alias for describe. Often, the outermost grouping of RSpec examples is defined with describe and the inner groups are grouped using context, but both are different names for the exact same function. In the cases we have seen, context gives us a local scope where we can define different versions of our test inputs with let.
Now, let's move on to matchers, which give us a flexible way of making assertions.