Hi,

I need to build a bash script to access and convert a text file header 
based on a sql match in order to copy data to another table.  As so, I 
thought of awk's split as the best alternative.

Consider a bash script where:

header=$(head -1 $CAMINHO/$FICHEIRO | sed -e 's/"//g')
# header is a ';' separated CSV like line: "nad huje;am;tre saim"

I want to have: 1st element of array = nad huje
                2nd elem = am
                3rd elem = tre saim
I want to acess the header variable inside awk command.
awk '{
    search=";";
    n=split('$header',array,search);
    for (i=1;i<=n;i++) {
        printf("Word[%d]=%s\n",i,array[i]);
    }
    exit;
}  '

The problem is:
awk: cmd. line:3:     n=split(nad_huje
awk: cmd. line:3:                     ^ unexpected newline or end of string

In the search field I tried to write any character 'a' for example, but 
the split function doesn't seem to be working.

I also tried to play with FS and RS but with no success.

Can't I access the header variable inside awk like that?  Is the array 
accessible to bash after awk finishes its job?

Thank you for any help,

Luis

On Tuesday 24 February 2009 14:13, Luis P. Mendes wrote:

> Hi,
> 
> I need to build a bash script to access and convert a text file header
> based on a sql match in order to copy data to another table.  As so, I
> thought of awk's split as the best alternative.
> 
> Consider a bash script where:
> 
> header=$(head -1 $CAMINHO/$FICHEIRO | sed -e 's/"//g')
> # header is a ';' separated CSV like line: "nad huje;am;tre saim"
> 
> I want to have: 1st element of array = nad huje
>                 2nd elem = am
>                 3rd elem = tre saim
> I want to acess the header variable inside awk command.
> awk '{
>     search=";";
>     n=split('$header',array,search);
>     for (i=1;i<=n;i++) {
>         printf("Word[%d]=%s\n",i,array[i]);
>     }
>     exit;
> }  '
> 
> The problem is:
> awk: cmd. line:3:     n=split(nad_huje
> awk: cmd. line:3:                     ^ unexpected newline or end of
> string

What you need is to use the -v option:

awk -v h="$header" '{
   search=";";
   n=split(header,array,search);
   for (i=1;i<=n;i++) {
     printf("Word[%d]=%s\n",i,array[i]);
   }
   exit;
}'

But of course you can also do all in awk:

awk '
NR==1{
   gsub(/"/,"");
   search=";";
   n=split($0,array,search);
   for (i=1;i<=n;i++) {
     printf("Word[%d]=%s\n",i,array[i]);
   }
   exit;
}' "$CAMINHO/$FICHEIRO"

Tue, 24 Feb 2009 14:22:25 +0100, pk escreveu:

> On Tuesday 24 February 2009 14:13, Luis P. Mendes wrote:
> 
>> Hi,
>> 
>> I need to build a bash script to access and convert a text file header
>> based on a sql match in order to copy data to another table.  As so, I
>> thought of awk's split as the best alternative.
>> 
>> Consider a bash script where:
>> 
>> header=$(head -1 $CAMINHO/$FICHEIRO | sed -e 's/"//g') # header is a
>> ';' separated CSV like line: "nad huje;am;tre saim"
>> 
>> I want to have: 1st element of array = nad huje
>>                 2nd elem = am
>>                 3rd elem = tre saim
>> I want to acess the header variable inside awk command. awk '{
>>     search=";";
>>     n=split('$header',array,search);
>>     for (i=1;i<=n;i++) {
>>         printf("Word[%d]=%s\n",i,array[i]);
>>     }
>>     exit;
>> }  '
>> 
>> The problem is:
>> awk: cmd. line:3:     n=split(nad_huje awk: cmd. line:3:               
>>      ^ unexpected newline or end of string
> 
> What you need is to use the -v option:
> 
> awk -v h="$header" '{
>    search=";";
>    n=split(header,array,search);
>    for (i=1;i<=n;i++) {
>      printf("Word[%d]=%s\n",i,array[i]);
>    }
>    exit;
> }'

Thank you pk, but now I have two other questions:
1- now the scripts pauses and needs a <return> to execute the awk part.  
Why is that?
2- I'd need the array Word back in bash as a bash variable.  How can I do 
it?

Luis

On Tuesday 24 February 2009 16:12, Luis P. Mendes wrote:

>> What you need is to use the -v option:
>> 
>> awk -v h="$header" '{
>>    search=";";
>>    n=split(header,array,search);
>>    for (i=1;i<=n;i++) {
>>      printf("Word[%d]=%s\n",i,array[i]);
>>    }
>>    exit;
>> }'
> 
> Thank you pk, but now I have two other questions:
> 1- now the scripts pauses and needs a <return> to execute the awk part.
> Why is that?

Because you did not specify an input file:

awk -v h="$header" '.....' "$CAMINHO/$FICHEIRO"

> 2- I'd need the array Word back in bash as a bash variable.  How can I do
> it?

Then (OT here but anyway) you can just use bash only:

oifs=$IFS
IFS=';'
array=( $(head -n 1 "$CAMINHO/$FICHEIRO" | sed -e 's/"//g') )
IFS=$oifs

On Feb 24, 7:22=A0am, pk <p...@pk.invalid> wrote:
> On Tuesday 24 February 2009 14:13, Luis P. Mendes wrote:
>
>
>
> > Hi,
>
> > I need to build a bash script to access and convert a text file header
> > based on a sql match in order to copy data to another table. =A0As so, =
I
> > thought of awk's split as the best alternative.
>
> > Consider a bash script where:
>
> > header=3D$(head -1 $CAMINHO/$FICHEIRO | sed -e 's/"//g')
> > # header is a ';' separated CSV like line: "nad huje;am;tre saim"
>
> > I want to have: 1st element of array =3D nad huje
> > =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 2nd elem =3D am
> > =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 3rd elem =3D tre saim
> > I want to acess the header variable inside awk command.
> > awk '{
> > =A0 =A0 search=3D";";
> > =A0 =A0 n=3Dsplit('$header',array,search);
> > =A0 =A0 for (i=3D1;i<=3Dn;i++) {
> > =A0 =A0 =A0 =A0 printf("Word[%d]=3D%s\n",i,array[i]);
> > =A0 =A0 }
> > =A0 =A0 exit;
> > } =A0'
>
> > The problem is:
> > awk: cmd. line:3: =A0 =A0 n=3Dsplit(nad_huje
> > awk: cmd. line:3: =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 ^ unexpected =
newline or end of
> > string
>
> What you need is to use the -v option:
>
> awk -v h=3D"$header" '{
> =A0 =A0search=3D";";
> =A0 =A0n=3Dsplit(header,array,search);
> =A0 =A0for (i=3D1;i<=3Dn;i++) {
> =A0 =A0 =A0printf("Word[%d]=3D%s\n",i,array[i]);
> =A0 =A0}
> =A0 =A0exit;
>
> }'
>
> But of course you can also do all in awk:
>
> awk '
> NR=3D=3D1{
> =A0 =A0gsub(/"/,"");
> =A0 =A0search=3D";";
> =A0 =A0n=3Dsplit($0,array,search);
> =A0 =A0for (i=3D1;i<=3Dn;i++) {
> =A0 =A0 =A0printf("Word[%d]=3D%s\n",i,array[i]);
> =A0 =A0}
> =A0 =A0exit;
>
> }' "$CAMINHO/$FICHEIRO"
>
>

No need for split() (or spurious semicolons and braces):

awk -F';' '
NR=3D=3D1{
   gsub(/"/,"")
   for (i=3D1;i<=3DNF;i++)
     printf "Word[%d]=3D%s\n",i,$i
   exit
}' "$CAMINHO/$FICHEIRO"

   Ed.

On Feb 24, 9:30=A0am, pk <p...@pk.invalid> wrote:
> On Tuesday 24 February 2009 16:12, Luis P. Mendes wrote:
>
> >> What you need is to use the -v option:
>
> >> awk -v h=3D"$header" '{
> >> =A0 =A0search=3D";";
> >> =A0 =A0n=3Dsplit(header,array,search);
> >> =A0 =A0for (i=3D1;i<=3Dn;i++) {
> >> =A0 =A0 =A0printf("Word[%d]=3D%s\n",i,array[i]);
> >> =A0 =A0}
> >> =A0 =A0exit;
> >> }'
>
> > Thank you pk, but now I have two other questions:
> > 1- now the scripts pauses and needs a <return> to execute the awk part.
> > Why is that?
>
> Because you did not specify an input file:
>
> awk -v h=3D"$header" '.....' "$CAMINHO/$FICHEIRO"
>
> > 2- I'd need the array Word back in bash as a bash variable. =A0How can =
I do
> > it?
>
> Then (OT here but anyway) you can just use bash only:
>
> oifs=3D$IFS
> IFS=3D';'
> array=3D( $(head -n 1 "$CAMINHO/$FICHEIRO" | sed -e 's/"//g') )
> IFS=3D$oifs

No need for head and pipe:

oifs=3D$IFS
IFS=3D';'
array=3D( $(sed -n '1 s/"//g p' "$CAMINHO/$FICHEIRO") )
IFS=3D$oifs

or, more topically:

oifs=3D$IFS
IFS=3D';'
array=3D( $(awk 'NR=3D=3D1{gsub(/"/,""); print; exit}' "$CAMINHO/
$FICHEIRO") )
IFS=3D$oifs

I expect there's a way to make sed exit after processing that first
line if efficiency's a concern - follow up at comp.unix.shell or
similar if necessary.

   Ed.

On Tuesday 24 February 2009 20:07, Ed Morton wrote:

>> Then (OT here but anyway) you can just use bash only:
>>
>> oifs=$IFS
>> IFS=';'
>> array=( $(head -n 1 "$CAMINHO/$FICHEIRO" | sed -e 's/"//g') )
>> IFS=$oifs
> 
> No need for head and pipe:
> 
> oifs=$IFS
> IFS=';'
> array=( $(sed -n '1 s/"//g p' "$CAMINHO/$FICHEIRO") )
> IFS=$oifs
> 
> or, more topically:
> 
> oifs=$IFS
> IFS=';'
> array=( $(awk 'NR==1{gsub(/"/,""); print; exit}' "$CAMINHO/
> $FICHEIRO") )
> IFS=$oifs

Right, thanks (also for the other correction). I was more focused on the
array part of the question than fixing other things.

pk wrote:
> On Tuesday 24 February 2009 14:13, Luis P. Mendes wrote:
> 
> 
>>Hi,
>>
>>I need to build a bash script to access and convert a text file header
>>based on a sql match in order to copy data to another table.  As so, I
>>thought of awk's split as the best alternative.
>>
>>Consider a bash script where:
>>
>>header=$(head -1 $CAMINHO/$FICHEIRO | sed -e 's/"//g')
>># header is a ';' separated CSV like line: "nad huje;am;tre saim"
>>
>>I want to have: 1st element of array = nad huje
>>                2nd elem = am
>>                3rd elem = tre saim
>>I want to acess the header variable inside awk command.
>>awk '{
>>    search=";";
>>    n=split('$header',array,search);
>>    for (i=1;i<=n;i++) {
>>        printf("Word[%d]=%s\n",i,array[i]);
>>    }
>>    exit;
>>}  '
>>
>>The problem is:
>>awk: cmd. line:3:     n=split(nad_huje
>>awk: cmd. line:3:                     ^ unexpected newline or end of
>>string
> 
> 
> What you need is to use the -v option:
> 
> awk -v h="$header" '{
>    search=";";
>    n=split(header,array,search);

I suppose you meant...

      n=split(h,array,search)


Janis

>    for (i=1;i<=n;i++) {
>      printf("Word[%d]=%s\n",i,array[i]);
>    }
>    exit;
> }'
> 
> But of course you can also do all in awk:
> 
> awk '
> NR==1{
>    gsub(/"/,"");
>    search=";";
>    n=split($0,array,search);
>    for (i=1;i<=n;i++) {
>      printf("Word[%d]=%s\n",i,array[i]);
>    }
>    exit;
> }' "$CAMINHO/$FICHEIRO"
>