动态填充多维 awk 数组

Question

我正在编写一个 Awk/Gawk 脚本来解析文件，为每一行填充一个多维数组。第一列是句点分隔的字符串，每个值都是对下一级数组键的引用。第二列是数值

下面是正在解析的内容的示例：

$ echo -e "personal.name.first\t= John\npersonal.name.last\t= Doe\npersonal.other.dob\t= 05/07/87\npersonal.contact.phone\t= 602123456\npersonal.contact.email\t= john.doe@idk\nemployment.jobs.1\t= Company One\nemployment.jobs.2\t= Company Two\nemployment.jobs.3\t= Company Three"
personal.name.first     = John
personal.name.last      = Doe
personal.other.dob      = 05/07/87
personal.contact.phone  = 602123456
personal.contact.email  = john.doe@idk
employment.jobs.1       = Company One
employment.jobs.2       = Company Two
employment.jobs.3       = Company Three

经过解析后，我希望它具有与以下相同的结构：

data["personal"]["name"]["first"]     = "John"
data["personal"]["name"]["last"]      = "Doe"
data["personal"]["other"]["dob"]      = "05/07/87"
data["personal"]["contact"]["phone"]  = "602123456"
data["personal"]["contact"]["email"]  = "john.doe@foo.com"
data["employment"]["jobs"]["1"]       = Company One
data["employment"]["jobs"]["2"]       = Company Two
data["employment"]["jobs"]["3"]       = Company Three

我坚持的部分是如何在构建多维数组时动态填充键。

我找到了 this SO thread that covers a similar issue, which was resolved by using the SUBSEP 变量，起初看起来它可以按我的需要工作，但经过一些测试后，它看起来 arr["foo", "bar"] = "baz" 并没有像真正的数组一样被对待，比如正如 arr["foo"]["bar"] = "baz" 那样。我的意思的一个例子是无法计算数组任何级别的值：arr["foo", "bar"] = "baz"; print length(arr["foo"]) 只会打印一个 0（零）

我发现 this SO thread 有点帮助，可能为我指明了正确的方向。

在讨论帖的一个片段中提到：

BEGIN {
  x=SUBSEP

  a="Red" x "Green" x "Blue"
  b="Yellow" x "Cyan" x "Purple"

  Colors[1][0] = ""
  Colors[2][0] = ""

  split(a, Colors[1], x)
  split(b, Colors[2], x)

  print Colors[2][3]
}

非常接近，但我现在遇到的问题是需要动态指定键（例如：Red、Green 等），并且可能有一个或多个键。

基本上，我如何获取 a_keys 和 b_keys 字符串，将它们拆分为 .，并将 a 和 b 变量填充为多维数组？..

BEGIN {
  x=SUBSEP

  # How can I take these strings...
  a_keys = "Red.Green.Blue"
  b_keys = "Yellow.Cyan.Purple"

  # .. And populate the array, just as this does:
  a="Red" x "Green" x "Blue"
  b="Yellow" x "Cyan" x "Purple"

  Colors[1][0] = ""
  Colors[2][0] = ""

  split(a, Colors[1], x)
  split(b, Colors[2], x)

  print Colors[2][3]
}

任何帮助将不胜感激，谢谢！

Answer 1

您只需要：

BEGIN { FS="\t= " }
{
    split(,d,/\./)
    data[d[1]][d[2]][d[3]] = 
}

看：

$ cat tst.awk
BEGIN { FS="\t= " }
{
    split(,d,/\./)
    data[d[1]][d[2]][d[3]] = 
}
END {
    for (x in data)
        for (y in data[x])
            for (z in data[x][y])
                print x, y, z, "->", data[x][y][z]
}

$ awk -f tst.awk file
personal other dob -> 05/07/87
personal name first -> John
personal name last -> Doe
personal contact email -> john.doe@idk
personal contact phone -> 602123456
employment jobs 1 -> Company One
employment jobs 2 -> Company Two
employment jobs 3 -> Company Three

以上内容当然是特定于 gawk 的，因为没有其他 awk 支持真正的多维数组。

当索引并不总是具有相同的深度（例如上面的 3）时填充多维数组，它相当复杂：

##########
$ cat tst.awk
function rec_populate(a,idxs,curDepth,maxDepth,tmpIdxSet) {
    if ( tmpIdxSet ) {
        delete a[SUBSEP]                # delete scalar a[]
        tmpIdxSet = 0
    }
    if (curDepth < maxDepth) {
        # We need to ensure a[][] exists before calling populate() otherwise
        # inside populate() a[] would be a scalar, but then we need to delete
        # a[][] inside populate() before trying to create a[][][] because
        # creating a[][] below creates IT as scalar. SUBSEP used arbitrarily.

        if ( !( (idxs[curDepth] in a) && (SUBSEP in a[idxs[curDepth]]) ) ) {
            a[idxs[curDepth]][SUBSEP]   # create array a[] + scalar a[][]
            tmpIdxSet = 1
        }
        rec_populate(a[idxs[curDepth]],idxs,curDepth+1,maxDepth,tmpIdxSet)
    }
    else {
        a[idxs[curDepth]] = 
    }
}

function populate(arr,str,sep,  idxs) {
    split(str,idxs,sep)
    rec_populate(arr,idxs,1,length(idxs),0)
}

{ populate(arr,,",") }

END { walk_array(arr, "arr") }

function walk_array(arr, name,      i)
{
    # Mostly copied from the following URL, just added setting of "sorted_in":
    #   https://www.gnu.org/software/gawk/manual/html_node/Walking-Arrays.html
    PROCINFO["sorted_in"] = "@ind_str_asc"
    for (i in arr) {
        if (isarray(arr[i]))
            walk_array(arr[i], (name "[" i "]"))
        else
            printf("%s[%s] = %s\n", name, i, arr[i])
    }
}

.

##########
$ cat file
a uno
b,c dos
d,e,f tres_wan
d,e,g tres_twa
d,e,h,i,j cinco

##########
$ awk -f tst.awk file
arr[a] = uno
arr[b][c] = dos
arr[d][e][f] = tres_wan
arr[d][e][g] = tres_twa
arr[d][e][h][i][j] = cinco

Answer 2

没有真正的 multidim 数组，你可以做更多的簿记工作

awk -F'\t= ' '{split(,k,"."); 
               k1[k[1]]; k2[k[2]]; k3[k[3]]; 
               v[k[1],k[2],k[3]]=}
          END {for(i1 in k1) 
                 for(i2 in k2)
                   for(i3 in k3) 
                     if((i1,i2,i3) in v) 
                       print i1,i2,i3," -> ",v[i1,i2,i3]}' file


personal other dob  ->  05/07/87
personal name first  ->  John
personal name last  ->  Doe
personal contact email  ->  john.doe@idk
personal contact phone  ->  602123456
employment jobs 1  ->  Company One
employment jobs 2  ->  Company Two
employment jobs 3  ->  Company Three

动态填充多维 awk 数组

Dynamically populate multidimensional awk array

arrays

awk

gawk

multidimensional-array