Text::CSV::Slurp losing in getline

throop has asked for the wisdom of the Perl Monks concerning the following question:

I'm trying to read in a CSV data file using Text::CSV::Slurp. It's not working, and my code is so close to the code in Slurp's own example, that I suspect I might have stumbled into some odd configuration problem. Ideas?

In the following test file:

use strict;
use Text::CSV::Slurp;
use vars qw($file $data);

$file = "/orion-reqs/SaferHtml/CradleCSV/2012_03_30/Doc_Sections_SAFER
+_Group_Trace.csv";
$data = Text::CSV::Slurp->load(file => $file);

1
[download]

$data is an empty array. Going into the debugger, I find that the file is opened correctly, and the header is read. On the first line of data, Slurp calls

$csv->getline_hr($io)
[download]

Single step shows that this goes into Text::CSV::AUTOLOAD. That routine expects a method-name that starts with an underscore, which getline_hr doesn't.

    my $attr = $Text::CSV::AUTOLOAD;   # bound to 'Text::CSV::getline_
+hr'
    $attr =~ s/.*:://;             # Now it's just 'getline_hr'
    return unless $attr =~ /^_/;   # returns with no value
[download]

FWIW, my %INC

 DB<11> x \%INC
0  HASH(0x98c4780)
   'AutoLoader.pm' => '/usr/share/perl5/AutoLoader.pm'
   'Carp.pm' => '/usr/share/perl5/Carp.pm'
   'Config.pm' => '/usr/lib/perl5/Config.pm'
   'Config_git.pl' => '/usr/lib/perl5/Config_git.pl'
   'Config_heavy.pl' => '/usr/lib/perl5/Config_heavy.pl'
   'DynaLoader.pm' => '/usr/lib/perl5/DynaLoader.pm'
   'Exporter.pm' => '/usr/share/perl5/Exporter.pm'
   'IO.pm' => '/usr/lib/perl5/IO.pm'
   'IO/Handle.pm' => '/usr/lib/perl5/IO/Handle.pm'
   'SelectSaver.pm' => '/usr/share/perl5/SelectSaver.pm'
   'Symbol.pm' => '/usr/share/perl5/Symbol.pm'
   'Term/Cap.pm' => '/usr/share/perl5/Term/Cap.pm'
   'Term/ReadLine.pm' => '/usr/share/perl5/Term/ReadLine.pm'
   'Text/CSV.pm' => '/home/dthroop/lib/perl/Text/CSV.pm'
   'Text/CSV/Slurp.pm' => '/usr/local/share/perl5/Text/CSV/Slurp.pm'
   'Text/CSV_XS.pm' => '/usr/lib/perl5/Text/CSV_XS.pm'
   'XSLoader.pm' => '/usr/share/perl5/XSLoader.pm'
   'dumpvar.pl' => '/usr/share/perl5/dumpvar.pl'
   'perl5db.pl' => '/usr/share/perl5/perl5db.pl'
   'strict.pm' => '/usr/share/perl5/strict.pm'
   'vars.pm' => '/usr/share/perl5/vars.pm'
   'warnings.pm' => '/usr/share/perl5/warnings.pm'
   'warnings/register.pm' => '/usr/share/perl5/warnings/register.pm'
[download]

and what the args to the AUTOLOAD look like:

 DB<16> x @_
0  Text::CSV=HASH(0x8ee2f90)
   '_AHEAD' => undef
   '_BOUND_COLUMNS' => undef
   '_CACHE' => "\"\",\cA\c@\c@\c@\c@\c@\c@\c@\c@\c@\c@\c@\c@\c@\c@\c@\
+c@\c@\c@\c@\c@\c@\cA\c@\c@\c@\c@\c@\cA"
   '_COLUMN_NAMES' => undef
   '_EOF' => ''
   '_FFLAGS' => undef
   '_FIELDS' => undef
   '_MODULE' => 'Text::CSV_XS'
   '_STATUS' => undef
   '_STRING' => undef
   'allow_loose_escapes' => 0
   'allow_loose_quotes' => 0
   'allow_whitespace' => 0
   'always_quote' => 0
   'auto_diag' => 0
   'binary' => 1
   'blank_is_undef' => 0
   'empty_is_undef' => 0
   'eol' => ''
   'escape_char' => '"'
   'keep_meta_info' => 0
   'quote_char' => '"'
   'quote_null' => 1
   'quote_space' => 1
   'sep_char' => ','
   'types' => undef
   'verbatim' => 0
1  IO::Handle=GLOB(0x8cb42e8)
   -> *Symbol::GEN0
         FileHandle({*Symbol::GEN0}) => fileno(6)
[download]

Does something look fishy?

Comment on Text::CSV::Slurp losing in getline_hr Select or Download Code

Replies are listed 'Best First'.
Re: Text::CSV::Slurp losing in getline_hr by Tux (Canon) on Apr 11, 2012 at 06:29 UTC
FWIW Text::CSV_XS (and Text::CSV - which uses Text::CSV_XS if installed - used in Text::CSV::Slurp), have a native method that comes very close to what you are using: getline_hr_all since version 0.80 (2010-12-24). Note that the documentation of Text::CSV::Slurp does not mention that a header line is required in the CSV file. Your statement "On the first line of data, Slurp calls `$csv->getline_hr ($io)`" is not true. It does so for the second line. The first line is read with `if (my $head = $csv->getline ($io)) { $csv->column_names ($head); ...` In Text::CSV_XS speak, your code would be something like: `use strict; use warnings; use Text::CSV_XS; my $file = "/orion-reqs/SaferHtml/CradleCSV/2012_03_30/Doc_Sections_SA +FER_Group_Trace.csv"; open my $fh, "<:encoding(utf-8)", $file or die "$file: $!"; my $csv = Text::CSV_XS->new ({ binary => 1, auto_diag => 1 }); $csv->column_names ($csv->getline ($fh)); my $data = $csv->getline_hr_all ($fh); close ($fh);` [download] Enjoy, Have FUN! H.Merijn /b	[reply] [d/l] [select]
Re: Text::CSV::Slurp losing in getline_hr by Khen1950fx (Canon) on Apr 10, 2012 at 19:56 UTC
I can't replicate your problem, but I tried it a little differently. First, don't use `use vars`. That creates package-scoped global variables which are more trouble than their worth. Just keep it simple. `#!/usr/bin/perl use strict; use warnings; use Text::CSV::Slurp; use Data::Dumper::Concise; my $file = '/orion-regs/SaferHtml/CradleCSV/2012_03_30/ Doc_Sections/SAFER_Group_Trace.csv'; my $data = Text::CSV::Slurp->load( file => $file, binary => 1 ); print Dumper($data);` [download]	[reply] [d/l] [select]