CODEWITHSUNDEEP
user-interfaceclojurelazy-evaluationread-eval-print-loop
user1922460
user1922460

Reputation:

Different results when a function is evaluated in the REPL than in a program

I have a feeling the answer to my question is something to do with Clojure's lazy evaluation (which I am still fuzzy on...)

So I have a function:

(defn fix-str-old [string]
  (let [words (->> string
                   (clojure.string/split-lines) 
                   (map #(clojure.string/replace % #"\W" "")))]
    (apply str (interleave words (repeat " ")))))

Basically it just takes a wacky sentence with non-alphanumeric chars, chars, return characters, line feeds etc in place of a space and turns it into a regular sentence. The reason for this if you're curious is that whenever I try to copy out of certain PDFs, it puts line feeds and other mysterious characters in between the words.

Here is an example:

(fix-str "A 
  block 
  of    
  SQL   
  statements    
  that  
  must  
  all   
  complete  
   successfully 
  before    
  returning 
  or    
  changing  
  anything  ")

==> "A block of SQL statements that must all complete  successfully before returning or changing anything"

It works fine in the REPL but when it is evaluated inside of a little swing gui you get this:

"AblockofSQLstatementsthatmustallcompletesuccessfullybeforereturningorchanginganything "

(note the space at the end of the string)

I was pretty sure that this was because of some gap in my understanding of how Clojure handles lazy seqs so I whipped up this function that just does regex operations.

(defn fix-str [string]
  (-> string
      (clojure.string/replace #"[ \t\n\r]+" " ")
      (clojure.string/replace #"[^a-zA-Z0-9 ]" "")
      (clojure.string/trimr)))

which isn't lazy and works fine in both the REPL and in the GUI.

Note: I also tried putting doall statements in various places in the original function that I though might make sense to make sure it forced evaluation of the lazy seqs but I couldn't get it to work either.

So my question isn't really if the first way is a good way to fix the strings, but rather why am I getting a different result in the REPL and in my GUI.

Upvotes: 2

Views: 110

Answers (1)

Alex
Alex

Reputation: 13961

Laziness should not be your problem here, because (apply str ...) forces the output from map to be realized (and because there's no bindings here, which is usually your first clue that laziness is the culprit).

Looks to me like there's something funky going on with the line-endings coming from the GUI, and that split-lines is not splitting anything. That function splits on \n or \r\n - maybe somehow you're getting \r as line-endings from the GUI? You can verify this by adding this to the beginning of your fix-str function:

(doseq [c string] (println (int c)))

Upvotes: 2

Related Questions