command line tool testing: capturing stdout and stderr

[Jaakko Suutarla]

Hi,

I would like to write end to end tests for my command line tool and I am trying to figure ways to do it.

What would be the best way to test for code thats checks stdout output? 

Example usage case: 

Code:

...

(defn magical-cli-command [argument1]
  ...
 
  (println "That was nice command and output is: " foober " and " barfoo))

...

Test case:

(deftest test-cli-output
  (testing "a test"
    (magical-cli-command 1 2)
    (is (= (captured-stdout-here-some-how) "That was nice command and output is: 1 and 2"))))

 

Some ideas:

• with-redefs println - feels little dirty
  
• using binding [*out* - not sure if good idea

Any ideas what would be the best way to do it?

Cheers,

Jaakko

[Pavel Prokopenko]

Hi,

you can use https://clojuredocs.org/clojure.core/with-out-str to verify the output which is basically the second option.

Pavel

[Vesa Marttila]

Hi Jaakko,

Would separating the side-effecting functionality (printing) be plausible in this case? If so, it would mean that you could have a unit test for the actual logic.

(defn actual-logic
  [argument1]
  ...
  [foober barfoo])

(defn magical-cli-command [argument1]
  (let [[foober barfoo] (actual-logic argument1)]
    (println "That was nice command and output is: " foober " and " barfoo)))

(deftest test-actual-logic
  (testing "a test"
    (is (= [... ...] (actual-logic 1 2))))

Vesa

[Jaakko Suutarla]

Hi Vesa & Pavel,

Thanks for the quick replies @Vesa and @Pavel

@Vesa

I have already few test sets for internal logic, but I want to ensures that while interacting with it from CLI it will work as I assume. ;)

@Pavel

Thanks I'll try to wrap my test cases with with-out-str

Jaakko