1

In a file I am saving a file path. I want to read the file path from a file and then convert it to a variable, so everytime I run a function it gets the current file path from the file and then I can use this String in another function. I know variables are immutable but is there a way I can do it?

Here as you can see I am trying to read the file path from the file and then check if it is correct:

checksForCdCommand dir = checkIfPathExists constructedPath $ concat' $ paths sample2 where
    constructedPath = do
        let file = "abc.txt"
        contents <- readFile file
        return (if contents /= "/" then contents ++ "/" ++ dir else contents ++ dir)

Here if the contents is "/" it means that its my root dir so thats why I have this if else clause. But this way the code wont compile due to the following:

* Couldn't match expected type `[Char]'
                  with actual type `IO [Char]'
    * In the first argument of `checkIfPathExists', namely
        `constructedPath'
      In the expression: checkIfPathExists constructedPath
      In the expression:
        checkIfPathExists constructedPath $ concat' $ paths sample2

I want to do something like this:

checksForCdCommand dir = checkIfPathExists constructedPath $ concat' $ paths sample2 where
    constructedPath = if currDir /= "/" then currDir ++ "/" ++ dir else currDir ++ dir

where currDir always has the value from the file.

The idea is that I have a tree that represents a file system and with the func paths I get all of the paths to the children, with checkIfPathExists I pass a String as an argument to see if it's part of all the paths and I am doing a command that is like the cd unix command and when I change directories I save the current dir in the file.

I need this current directory for other things and that's why I need to read it from a file and then change the variable currDir so I can use it for other functions.

Improve this question
2
  • 3
    With no types on any of your custom functions I really have no idea what you're trying to do. The description is very confusing. Commented Dec 31, 2021 at 10:57
  • If I understand the question correctly, the cd command should have a top-level structure: load the tree, load the current path, from the parameter, take a new path, construct an absolute new path, check if the new path exists, write a new path, end. mkdir should be all in the do block again: load the tree, load the current path, load the created path from the parameter, construct an absolute path, check whether it does not exist, add to the tree, write a new tree, end. This allows reliable functions to remain separate from untrusted IO state. Commented Dec 31, 2021 at 11:41

2 Answers 2

Reset to default
5

change the variable currDir

This is the thing you can't do. All the rest is distracting window-dressing. If your approach relies fundamentally on doing this, you need a new approach. At some point, in an IO action, you need to read the contents of this abc.txt file, and save that result into a variable. Then you pass that variable (whose value will not change!) to other functions. If at some point you want to "refresh" that variable, you will need to re-read it and use that new variable.

For example:

addOneToFileContents :: String -> Maybe (Int, Int)
addOneToFileContents c = case reads c of
  [(x, _)] -> Just $ (x, x + 1)
  _ -> Nothing

main = do
  num <- readFile "abc.txt"
  case addOneToFileContents num of
    Nothing -> putStrLn "abc.txt should contain a number" *> main
    Just (x, y) -> putStrLn $ show x ++ " + 1 = " ++ show y

This program is not very well written, but it demonstrates the basic idea. It assumes abc.txt exists; if not, you get some error. But assuming it does, it prints error messages very quickly until you (in another process) modify abc.txt to contain a number.

You will have to structure your program in a similar way: some IO action, probably in main or close to it, reads the relevant file. Then it's passed to some other function, which makes decisions about what to do based on the file contents. Later, you re-read the file in main and continue the loop.

There are tools that make this process simpler; for example, you can use Reader to plumb the contents of abc.txt through the pure parts of your program. It's not really much simpler in isolation, but it combines more cleanly with other bookkeeping you may have to do, like plumbing the results of additional IO actions through. But fundamentally it's no different: you have to do IO in IO actions, and variables don't change.

Improve this answer
Sign up to request clarification or add additional context in comments.

Comments

2

"I know variables are immutable but is there a way I can do it?"

Haskell is very opinionated on the subject of mutable variables - they don't exist, only immutable values exist. Like dominoes, the output of one function is plumbed into the input of the next function. The only changeable things are the values input at the start, and the values that come out of the end, like a sausage machine. It is possible to save values using the state monad, or by using a database, but if you are using a file as an improvised mutable variable then this savours strongly of attempting to subvert the principles of immutability.

But what about this 'Python' code?

    let file = "abc.txt"
    contents <- readFile file

It looks like there are mutable variables happening here, but this is an illusion. This is syntactic sugar which Haskell turns into a sequence of linked lambda expressions.

The error message is caused because you are using the do syntactical sugar:

 * Couldn't match expected type `[Char]'
              with actual type `IO [Char]'

In the absence of your type declaration, which you should really provide, Haskell believes two things:

  1. The use of do and return means that you are doing an IO action, type IO [Char], where [Char] means a string of text
  2. The definition says that you are doing text operations, like concat' hence expected type is [Char], a string

So, it is confused.

There are two things that you need to know about the IO action.

  1. An IO action is like The Golden Goose. Just as the Golden Goose, when cut open, did not contain any gold, so an IO action, when cut open, does not contain the data that you are looking for.
  2. The IO tag taints anything it comes into contact with. So, the correct thing to do is to write functions which don't used IO actions at all, just transforming numbers into strings, strings into lists, and so on. You concentrate the tainted IO actions into the main function.

Hence my recommendations are:

  • Write a main function, which includes IO actions
  • Write as much of the remainder of the code without IO actions
  • Include the type declarations for every function (it stops Haskell getting confused, and allows you to check that you're doing the right thing)
Improve this answer

Comments

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.