How to apply an awk command to a field defined by another awk command?-CodePudding

I have 2 tasks to perform over a data string separated by fields, I could come up with an awk command for each task separatedly, but what I need is to apply a second task to the 3rd field, and have the results of 2nd task inside results of first task.

Data in file data.csv

31;Area A;Language B1 G1-T1-(3343-1-25274) (3343-1-25278) (3345-1-25676);German;;;0;;Wolfgang Mozart

Fisrt task is to generate an xml structure, that's done with next awk command:

awk -F';' -v OFS=';' '{printf "<A>\n\t<T>%s</T>\n\t<S>%s</S>\n\t<AT>%s</AT>\n\t<D>%s</D>\n\t<Id>%s</Id>\n</A>\n", $9,$3,$2,$7,$7,$1}' test.csv

Result is:

<A>
    <T>Wolfgang Mozart</T>
    <S>Language B1 G1-T1-(3343-1-25274) (3343-1-25278) (3345-1-25676)</S>
    <AT>Area A</AT>
    <D>0</D>
    <Id>31</Id>
</A>

Second task is to convert every code in format "[0-9]{4}-[1-5]" into a <AT> tag, which I could manage with the next command:

awk 'BEGIN{FPAT="[0-9]{4}-[1-5]"} {for (i=1; i <= NF; i  ) print "<AT>"$i"</AT>"}' test.csv

Result is (by the way, I could not accomplish to print out only one instance when repeated):

<AT>3343-1</AT>
<AT>3343-1</AT>
<AT>3345-1</AT>

The desired output is:

<A>
    <T>Wolfgang Mozart</T>
    <S>Language B1 G1-T1-(3343-1-25274) (3343-1-25278) (3345-1-25676)</S>
    <AT>Area A</AT>
    <AT>3343-1</AT>
    <AT>3345-1</AT>
    <D>0</D>
    <Id>31</Id>
</A>

The best command I could come up with is the next one, which does not porduce the desired output:

awk -F';' -v OFS=';' 'BEGIN{FPAT="[0-9]{4}-[1-5]"} {for (i=1; i <= NF; i  ) printf "<A>\n\t<T>%s</T>\n\t<S>%s</S>\n\t<AT>%s</AT>\n\t<D>%s</D>\n\t<Id>%s</Id>\n</A>\n", $9,$3,$2,$7,$7,$1}' test.csv

Result is:

<A>
    <T>Wolfgang Mozart</T>
    <S>Language B1 G1-T1-(3343-1-25274) (3343-1-25278) (3345-1-25676)</S>
    <AT>Area A</AT>
    <AT>3343-1</AT>
    <D>0</D>
    <Id>31</Id>
</A>
<A>
    <T>Wolfgang Mozart</T>
    <S>Language B1 G1-T1-(3343-1-25274) (3343-1-25278) (3345-1-25676)</S>
    <AT>Area A</AT>
    <AT>3343-1</AT>
    <D>0</D>
    <Id>31</Id>
</A>
<A>
    <T>Wolfgang Mozart</T>
    <S>Language B1 G1-T1-(3343-1-25274) (3343-1-25278) (3345-1-25676)</S>
    <AT>Area A</AT>
    <AT>3345-1</AT>
    <D>0</D>
    <Id>31</Id>
</A>

CodePudding user response：

One idea using GNU awk:

awk -F';' '
{ patsplit($3,arr,"[0-9]{4}-[1-5]")          # split field #3 into NNNN-N strings
  delete seen                                # clear seen[] array
  for (i in arr)                             # for each NNNN-N string in arr[], store as index in 
      seen[arr[i]]                           # seen[] array; duplicates are effectively eliminated

  printf "<A>\n\t<T>%s</T>\n\t<S>%s</S>\n\t<AT>%s</AT>\n", $9,$3,$2

  PROCINFO["sorted_in"]="@ind_str_asc"       # assuming we want NNNN-N strings displayed in sorted order 

  for (at in seen)
      printf "\t<AT>%s</AT>\n", at

  printf "\t<D>%s</D>\n\t<Id>%s</Id>\n</A>\n", $7,$1
}
' data.csv

This generates:

<A>
        <T>Wolfgang Mozart</T>
        <S>Language B1 G1-T1-(3343-1-25274) (3343-1-25278) (3345-1-25676)</S>
        <AT>Area A</AT>
        <AT>3343-1</AT>
        <AT>3345-1</AT>
        <D>0</D>
        <Id>31</Id>
</A>