Clojure - Convert string into vector of vector of integers - clojure

I'm trying to parse through a file and use each line to execute any number of functions and parameters. The functions I want to call accept two vectors of vectors of integers for matrix multiplication. I'm able to parse the arguments into one vector so I can call apply on it and the resolved function symbol. But I still need to convert the arguments from strings into the appropriate type. How can I achieve this?
Function header example:
(defn ijk [[& matrixA] [& matrixB]]
...
)
Input file example: (splitting string by commas)
ijk,[[1 2] [3 4]],[[1 2] [3 4]]
kij,[[2 2] [3 4]],[[1 2] [3 4]]
How I'm reading the file so far:
(defn get-lines [fname]
(with-open [r (reader fname)]
(loop [file (line-seq r)]
(if-let [[line & file] file]
(do (let [[command & args] (str/split line #",")]
;apply (resolve (symbol command)) (vec args))
)
(recur file))
file))))
Format of (vec args):
[[[1 2] [3 4]] [[1 2] [3 4]]]
[[[2 2] [3 4]] [[1 2] [3 4]]]
I need to convert each matrix in the args vector into a vector of vectors of integers above. Any and all help is much appreciated by this Clojure noob!

You could use clojure.edn/read-string to parse the strings into data structures:
(def args ["[[1 2] [3 4]]"
"[[1 2] [3 4]]"])
(mapv clojure.edn/read-string args)
=> [[[1 2] [3 4]] [[1 2] [3 4]]]

Related

How do I get Clojure maro to output the original code without substitutions?

There is macros. This macro is convenient because it allows you
to see the result of executing the Clojure script.
But often it gives out a result that does not suit me.
I use Babashka for running scripts.
(defn symbol-several
"returns a symbol with the concatenation of the str values of the args"
[& x]
(symbol (apply str x)))
(defmacro ! [& forms]
(cons
`symbol-several
(for [form forms]
`(str '~form " ;-> " ~form "\n"))))
,that outputs:
c:\clj>bb "~#.clj"
(do (def v [3 4]) (def l (quote (1 2))) (clojure.core/sequence (clojure.core/seq (clojure.core/concat (clojure.core/list 0) l v)))) ;-> (0 1 2 3 4)
The ~#.clj file contains the macro "!" and the code:
"(! (do (def v [3 4]) (def l '(1 2)) `(0 ~#l ~#v)))"
How to rewrite a macro so that it outputs the original code, i.e. something like this:
(do (def v [3 4]) (def l '(1 2)) `(0 ~#l ~#v))) ;-> (0 1 2 3 4)
Also, instead of the result (list), the macro outputs LazySeq:
c:\clj>bb map.clj
(apply map vector [[1 2] [3 4]]) ;-> clojure.lang.LazySeq#8041
I use Babashka (bb.exe) for running script.
The map.clj file contains the macro "!" and the code:
(apply map vector [[1 2] [3 4]])
Using pr-str helps printing Clojure data structures properly.
(defn symbol-several
"returns a symbol with the concatenation of the str values of the args"
[& x]
(symbol (apply str x)))
(defmacro ! [& forms]
`(symbol-several
~#(for [form forms]
`(str '~form " ;-> " (pr-str ~form) "\n"))))
(! (apply map vector [[1 2] [3 4]]))
;; => (apply map vector [[1 2] [3 4]]) ;-> ([1 3] [2 4])
Also see https://clojuredocs.org/clojure.core/pr-str

Create 2-dimensional vectors from 1-dimensionals

Although my question seems to be concerned with a rather trivial task, I was not yet able to sucessfully create a single 2-dimensional vector
[[1 2] [3 4]]
fom two individual vectors,
(def a [1 2]) and (def b [3 4]).
Leaving the functions conj and cons aside, I ran into the problem that vec or into-array both expect single input values.
Another workaround would be pre-filling a two dimenstional vector
(vec (replicate 2 (vec (replicate 2 nil))))
but I it is still a complicated option.
Just use vector:
(def a [1 2])
(def b [3 4])
(vector a b) => [[1 2] [3 4]]
[a b] => [[1 2] [3 4]]
There are several ways to approach this:
As #MartinPůda says, use vector: (vector a b) => [[1 2] [3 4]])
Use conj: (conj [] a b) => [[1 2] [3 4]]
Use a vector literal: [a b] => [[1 2] [3 4]]
If Clojure has a problem it is an excess of riches. :-)

Tree traversal in Clojure

Consider a tree, which is defined by the following recursive definition:
a tree should be a vector of two elements: The first one is a number, the second one is either a list of trees or nil.
The following clojure data structure would be an example:
(def tree '[9 ([4 nil]
[6 nil]
[2 ([55 nil]
[22 nil]
[3 ([5 nil])])]
[67 ([44 nil])])])
This structure should be transformed into a list of all possible downwards connections, that can be made from any node to its childs. The connections should be represented as vectors containing the value of the parent node followed by the value of the child node. The order is not important:
(def result '([9 4]
[9 6]
[9 2]
[2 55]
[2 22]
[2 3]
[3 5]
[9 67]
[67 44])
I came up with this solution:
(defn get-connections [[x xs]]
(concat (map #(vector x (first %)) xs)
(mapcat get-connections xs)))
And, indeed:
(= (sort result)
(sort (get-connections tree)))
;; true
However, are there better ways to do so with just plain clojure? In the approach, I'm traversing each node's childs twice, this should be avoided. In this particular case, tail-recursion is not necessary, so a simple recursive version would be ok.
Moreover, I'd be interested which higher level abstractions could be used for solving this. What about Zippers or Clojure/Walk? And finally: Which of those techniques would be available in ClojureScript as well?
you could try combination of list comprehension + tree-seq:
user> (for [[value children] (tree-seq coll? second tree)
[child-value] children]
[value child-value])
;;=> ([9 4] [9 6] [9 2] [9 67] [2 55] [2 22] [2 3] [3 5] [67 44])
this should be available in cljs.
as far as i know, both zippers and clojure.walk are available in clojurescript, but in fact you don't need them for this trivial task. I guess tree-seq is rather idiomatic.
What about the double traversal, you could easily rearrange it into a single one like this:
(defn get-connections [[x xs]]
(mapcat #(cons [x (first %)] (get-connections %)) xs))
user> (get-connections tree)
;;=> ([9 4] [9 6] [9 2] [2 55] [2 22] [2 3] [3 5] [9 67] [67 44])
then you could add laziness, to make this solution truly idiomatic:
(defn get-connections [[x xs]]
(mapcat #(lazy-seq (cons [x (first %)] (get-connections %))) xs))

How to split an input sequence according to the input number given

I'm writing a clojure function like:
(defn area [n locs]
(let [a1 (first locs)]
(vrp (rest locs))))
I basically want to input like: (area 3 '([1 2] [3 5] [3 1] [4 2])) But when I do that it gives me an error saying Wrong number of args (1) passed. But I'm passing two arguments.
What I actually want to do with this function is that whatever value of n is inputted (say 3 is inputted), then a1 should store [1 2], a2 should store [3 5], a3 should store ([3 1] [4 2]). What should I add in the function to get that?
clojue's build in split-at function is very close to solving this. It splits a sequence "at" a given point. So if we first split the data apart and then wrap the second half in a list and concatenate it back together again it should solve this problem:
user> (let [[start & end] (split-at 2 sample-data)]
(concat start end))
([1 2] [3 5] ([3 1] [4 2]))
the & before end in let causes the last item to be rolled up in a list. It's equivalent to:
user> (let [[start end] (split-at 2 sample-data)]
(concat start (list end)))
([1 2] [3 5] ([3 1] [4 2]))
If I recklessly assume you have some function called vrp that needs data in this form then you could finish your function with something like this:
(defn area [n locs]
(let [[start end] (split-at (dec n) sample-data)
a (concat start (list end))]
(apply vrp a)))
Though please forgive me to making wild guesses as to the nature of vrp, I could be totally off base here.

Clojure: Semi-Flattening a nested Sequence

I have a list with embedded lists of vectors, which looks like:
(([1 2]) ([3 4] [5 6]) ([7 8]))
Which I know is not ideal to work with. I'd like to flatten this to ([1 2] [3 4] [5 6] [7 8]).
flatten doesn't work: it gives me (1 2 3 4 5 6 7 8).
How do I do this? I figure I need to create a new list based on the contents of each list item, not the items, and it's this part I can't find out how to do from the docs.
If you only want to flatten it one level you can use concat
(apply concat '(([1 2]) ([3 4] [5 6]) ([7 8])))
=> ([1 2] [3 4] [5 6] [7 8])
To turn a list-of-lists into a single list containing the elements of every sub-list, you want apply concat as nickik suggests.
However, there's usually a better solution: don't produce the list-of-lists to begin with! For example, let's imagine you have a function called get-names-for which takes a symbol and returns a list of all the cool things you could call that symbol:
(get-names-for '+) => (plus add cross junction)
If you want to get all the names for some list of symbols, you might try
(map get-names-for '[+ /])
=> ((plus add cross junction) (slash divide stroke))
But this leads to the problem you were having. You could glue them together with an apply concat, but better would be to use mapcat instead of map to begin with:
(mapcat get-names-for '[+ /])
=> (plus add cross junction slash divide stroke)
The code for flatten is fairly short:
(defn flatten
[x]
(filter (complement sequential?)
(rest (tree-seq sequential? seq x))))
It uses tree-seq to walk through the data structure and return a sequence of the atoms. Since we want all the bottom-level sequences, we could modify it like this:
(defn almost-flatten
[x]
(filter #(and (sequential? %) (not-any? sequential? %))
(rest (tree-seq #(and (sequential? %) (some sequential? %)) seq x))))
so we return all the sequences that don't contain sequences.
Also you may found useful this general 1 level flatten function I found on clojuremvc:
(defn flatten-1
"Flattens only the first level of a given sequence, e.g. [[1 2][3]] becomes
[1 2 3], but [[1 [2]] [3]] becomes [1 [2] 3]."
[seq]
(if (or (not (seqable? seq)) (nil? seq))
seq ; if seq is nil or not a sequence, don't do anything
(loop [acc [] [elt & others] seq]
(if (nil? elt) acc
(recur
(if (seqable? elt)
(apply conj acc elt) ; if elt is a sequence, add each element of elt
(conj acc elt)) ; if elt is not a sequence, add elt itself
others)))))
Example:
(flatten-1 (([1 2]) ([3 4] [5 6]) ([7 8])))
=>[[1 2] [3 4] [5 6] [7 8]]
concat exampe surely do job for you, but this flatten-1 is also allowing non seq elements inside a collection:
(flatten-1 '(1 2 ([3 4] [5 6]) ([7 8])))
=>[1 2 [3 4] [5 6] [7 8]]
;whereas
(apply concat '(1 2 ([3 4] [5 6]) ([7 8])))
=> java.lang.IllegalArgumentException:
Don't know how to create ISeq from: java.lang.Integer
Here's a function that will flatten down to the sequence level, regardless of uneven nesting:
(fn flt [s] (mapcat #(if (every? coll? %) (flt %) (list %)) s))
So if your original sequence was:
'(([1 2]) (([3 4]) ((([5 6])))) ([7 8]))
You'd still get the same result:
([1 2] [3 4] [5 6] [7 8])