Sort associative array with AWK

Question

Here's my array (gawk script) :

myArray["peter"] = 32
myArray["bob"] = 5
myArray["john"] = 463
myArray["jack"] = 11

After sort, I need the following result :

bob    5
jack   11
peter  32
john   463

When i use "asort", indices are lost. How to sort by array value without losing indices ? (I need ordered indices based on their values)

(I need to obtain this result with awk/gawk only, not shell script, perl, etc)

If my post isn't clear enough, here is an other post explaining the same issue : http://www.experts-exchange.com/Programming/Languages/Scripting/Shell/Q_26626841.html )

Thanks in advance

Update :

Thanks to you both, but i need to sort by values, not indices (i want ordered indices according to their values).

In other terms, i need this result :

bob    5
jack   11
peter  32
john   463

not :

bob 5
jack 11
john 463
peter 32

(I agree, my example is confusing, the chosen values are pretty bad)

From the code of Catcall, I wrote a quick implementation that works, but it's rather ugly (I concatenate keys & values before sort and split during comparison). Here's what it looks like :

function qsort(A, left, right,   i, last) {
  if (left >= right)
    return
  swap(A, left, left+int((right-left+1)*rand()))
  last = left
  for (i = left+1; i <= right; i++)
    if (getPart(A[i], "value") < getPart(A[left], "value"))
      swap(A, ++last, i)
  swap(A, left, last)
  qsort(A, left, last-1)
  qsort(A, last+1, right)
}

function swap(A, i, j,   t) {
  t = A[i]; A[i] = A[j]; A[j] = t
}

function getPart(str, part) {
  if (part == "key")
    return substr(str, 1, index(str, "#")-1)
  if (part == "value")
    return substr(str, index(str, "#")+1, length(str))+0
  return
}

BEGIN {  }
      {  }
END {

  myArray["peter"] = 32
  myArray["bob"] = 5
  myArray["john"] = 463
  myArray["jack"] = 11

  for (key in myArray)
    sortvalues[j++] = key "#" myArray[key]

  qsort(sortvalues, 0, length(myArray));

  for (i = 1; i <= length(myArray); i++)
    print getPart(sortvalues[i], "key"), getPart(sortvalues[i], "value")
}

Of course I'm interested if you have something more clean...

Thanks for your time

Answer 1

Edit:

Sort by values

Oh! To sort the values, it's a bit of a kludge, but you can create a temporary array using a concatenation of the values and the indices of the original array as indices in the new array. Then you can asorti() the temporary array and split the concatenated values back into indices and values. If you can't follow that convoluted description, the code is much easier to understand. It's also very short.

# right justify the integers into space-padded strings and cat the index
# to create the new index
for (i in myArray) tmpidx[sprintf("%12s", myArray[i]),i] = i
num = asorti(tmpidx)
j = 0
for (i=1; i<=num; i++) {
    split(tmpidx[i], tmp, SUBSEP)
    indices[++j] = tmp[2]  # tmp[2] is the name
}
for (i=1; i<=num; i++) print indices[i], myArray[indices[i]]

Edit 2:

If you have GAWK 4, you can traverse the array by order of values without performing an explicit sort:

#!/usr/bin/awk -f
BEGIN {
    myArray["peter"] = 32
    myArray["bob"] = 5
    myArray["john"] = 463
    myArray["jack"] = 11

    PROCINFO["sorted_in"] = "@val_num_asc"

    for (i in myArray) {
        {print i, myArray[i]}}
    }

 }

There are settings for traversing by index or value, ascending or descending and other options. You can also specify a custom function.

Previous answer:

Sort by indices

If you have an AWK, such as gawk 3.1.2 or greater, which supports asorti():

#!/usr/bin/awk -f
BEGIN {
    myArray["peter"] = 32
    myArray["bob"] = 5
    myArray["john"] = 463
    myArray["jack"] = 11

    num = asorti(myArray, indices)
    for (i=1; i<=num; i++) print indices[i], myArray[indices[i]]
}

If you don't have asorti():

#!/usr/bin/awk -f
BEGIN {
    myArray["peter"] = 32
    myArray["bob"] = 5
    myArray["john"] = 463
    myArray["jack"] = 11

    for (i in myArray) indices[++j] = i
    num = asort(indices)
    for (i=1; i<=num; i++) print i, indices[i], myArray[indices[i]]
}