in reply to Re^3: The best way to split tab delimited file
in thread The best way to split tab delimited file

my $var='474627    asidase ta sidase ala,"lpha-D- ctoside gtohydrolase
+","razyme","arazyme    (enz Corp)","Melie","lagal","idase bta",    ru
+g    00103';
my @each=split(/(?<!,)\t/,$var);

for my $eachvar(@each)
    {
       print "$eachvar\n";
     }

[download]

This is the output i get

474627
asidase ta
sidase ala,"lpha-D- ctoside gtohydrolase","razyme","arazyme
(enz Corp)","Melie","lagal","idase bta",        rug
00103

[download]

This is the output i expect

474627
asidase ta
sidase ala,"lpha-D- ctoside gtohydrolase","razyme","arazyme    (enz Co
+rp)","Melie","lagal","idase bta",
rug
00103

[download]

Replies are listed 'Best First'.
Re^5: The best way to split tab delimited file
by gmargo (Hermit) on Nov 23, 2009 at 18:59 UTC

    So in addition to ignoring a tab after a comma, you also want to ignore tabs within quoted strings. This revises the original spec just a bit.

    Do you really require both of these things? Or is the real requirement only the latter (ignore tabs within strings), and the original example just happened to have been derived from a string with a comma-tab?

    Also, I get this output, different from yours. (no tab after 'ta') (And you have a comma-tab, just before 'rug', that you do want to split.) 

    474627
asidase ta sidase ala,"lpha-D- ctoside gtohydrolase","razyme","arazyme
(enz Corp)","Melie","lagal","idase bta",        rug
00103

    [download]
[reply]
[d/l]

      want to ignore the tab within the quoted string only,this data just happened to have a comma before the next(column tabbed data)data.

      Note:After "asidase ta" there is a tab.so whatever output i got is correct.

[reply]

        If you don't mind losing the tab within the quotes, pre-process the string to remove those tabs. Here I replaced the embedded tabs with spaces, then just split on tab: 

        my $var='474627    asidase ta    sidase ala,"lpha-D- ctoside gtohydrol
+ase","razyme","arazyme    (enz Corp)","Melie","lagal","idase bta",   
+ rug    00103';

my $tmp;
$var =~ s{ ("[^"]+") }{ ($tmp = $1) =~ s/\t/ /g; $tmp }xge;

my @each=split(/\t/,$var);

for my $eachvar(@each)
{
    print "$eachvar\n";
}

        [download]

        Update 1: Oops, I made a mistake in the pattern. The quotes belong on the inside of the capture. (Was: "([^"]+)", Now: ("[^"]+").

        Update 2: In response to a private message, here's a little better explanation of the pattern: 

        # Using s{}{} form of substitute.
# Substitute supports using several different separator formats
# which helps one avoid having to escape things (like '/') within the 
+pattern.
# The 'x' option which means ignore whitespace so that comments can be
+ easily inserted.
# The 'g' option is global obviously.
# The 'e' option says that the replacement part of the pattern is a pe
+rl expression.

$var =~
    s{
        ("[^"]+")       # Matches two quotes and content between them.
                        # Capture the match for use in the replacement
+.
                        #
                        # Disection of pattern:
                        # ("[^"]+")  = full pattern
                        # (       )  = capture everything between pare
+ntheses.
                        #  "     "   = quotes at start and end of patt
+ern.
                        #   [^"]+    = one or more non-quote character
+s
    }
    {                       # The replacement part is a perl expressio
+n.
                            # Original: ($tmp = $1) =~ s/\t/ /g; 
                            # is same as next 2 lines:
        $tmp = $1;          # Make a copy of the captured match.
        $tmp =~ s/\t/ /g;   # Replace tabs with spaces throughout the 
+match.
        $tmp;               # Use resultant value for replacement.
    }xge;
    # x = ignore white space and comments
    # g = global
    # e = expression

        [download]
[reply]
[d/l]
[select]

In Section Seekers of Perl Wisdom