Bug #34474 for Text-CSV_XS: wish: integrate row-as-hashref feature from Parse::CSV

Merijn, I needed this functionality and went ahead and wrote my own patch for it, although I later found your GIT respository, where you had also written some patch for this. ( http://repo.or.cz/w/Text-CSV_XS.git?a=shortlog ) I've compared our work, and would like to point out some features of mine: - It uses "column_names", which I think is a clearer name. - It allows the names to be set from an array or an arrayref. The array would be clearer if setting the names by hand, an the arrayref is is nice because it is what getlines() returns. - I include error checking for a number of cases including: - invalid input to column_names() - calling getlines_hr() before column_names() - mismatching the number of colum names and values My patch also includes tests and documentation updates. The one thing in your patch I was curious about was the stub mention in one of the XS files. My implementation doesn't use such a stub, although it appears to function fine with one. Since there are a lot of Text::CSV_XS users, I may get some feedback from some others on the alternatives. Mark

diff -rN -u old-Text-CSV_XS-0.37/ChangeLog new-Text-CSV_XS-0.37/ChangeLog --- old-Text-CSV_XS-0.37/ChangeLog 2008-04-03 13:13:30.000000000 -0400 +++ new-Text-CSV_XS-0.37/ChangeLog 2008-04-03 13:13:30.000000000 -0400 @@ -1,3 +1,7 @@ + + * Added getline_hr() and column_names() to support returning + rows as hashrefs (Mark Stosberg, H.Merijn Brand) + 2008-03-11 0.37 - H.Merijn Brand <h.m.brand@xs4all.nl> * Copied GIT repo to public mirror diff -rN -u old-Text-CSV_XS-0.37/CSV_XS.pm new-Text-CSV_XS-0.37/CSV_XS.pm --- old-Text-CSV_XS-0.37/CSV_XS.pm 2008-04-03 13:13:30.000000000 -0400 +++ new-Text-CSV_XS-0.37/CSV_XS.pm 2008-04-03 13:13:30.000000000 -0400 @@ -380,6 +380,62 @@ $self->{_STATUS}; } # parse +sub column_names { + my $self = shift; + my ($first,@rest) = @_; + my @cols; + # First use? Initialize. + $self->{_COLUMN_NAMES} ||= []; + + # Error checking and unpacking @_ + if (ref $first eq 'ARRAY') { + if (scalar @rest) { + die "column_names() expects only one arg when given an arrayref" + } + else { + $self->{_COLUMN_NAMES} = $first; + } + } + elsif (ref $first) { + die "column_names() was passed a reference which was not of type ARRAY"; + } + # Array of column names expected + elsif ((defined $first) or (scalar @rest)) { + @cols = ($first,@rest); + $self->{_COLUMN_NAMES} = \@cols; + } + else { + # No input processing! + } + + return @{ $self->{_COLUMN_NAMES} }; +} + +sub getline_hr { + my $self = shift; + my $io = shift; + + unless (defined $self->{_COLUMN_NAMES}) { + die "You must call column_names before calling getline_hr"; + } + + my $vals = $self->getline($io); + my @keys = $self->column_names; + + my $n_vals = scalar @$vals; + my $n_keys = scalar @keys; + unless ($n_keys == $n_vals) { + die "Number of column_names ($n_keys) doesn't match the number of entries in this row ($n_vals)" + } + + my %row; + @row{@keys} = @$vals; + return \%row ; +} + + + + bootstrap Text::CSV_XS $VERSION; sub types @@ -431,6 +487,10 @@ $colref = $csv->getline ($io); # Read a line from file $io, # parse it and return an array # ref of fields + + $csv->column_names('Name','Address') # Set array of column names + $href = $csv->getline_hr($io) # Return line as hashref + $eof = $csv->eof (); # Indicate if last parse or # getline () hit End Of File @@ -862,6 +922,29 @@ The I<$csv-E<gt>string ()>, I<$csv-E<gt>fields ()> and I<$csv-E<gt>status ()> methods are meaningless, again. +=item getline_hr + +=item column_names + + $csv->column_names('Name','Address'); + # Or, with an arrayref + $csv->column_names(['Name','Address']); + + # Now, later you can do: + my $href = $csv->getline_hr($io); + +The C<getline_hr> and C<column_names> methods work together to allow you to +have rows returned as hashrefs. You must call C<column_names> first to declare +your column names. + + # Set the column names from a row in the CSV file + $csv->column_names( $csv->getline($io) ); + +Then you can call C<getline_hr()> as you would C<getline()>, and a hashref of +the row will be returned: + + my $href = $csv->getline_hr($io); + =item eof $eof = $csv->eof (); diff -rN -u old-Text-CSV_XS-0.37/MANIFEST new-Text-CSV_XS-0.37/MANIFEST --- old-Text-CSV_XS-0.37/MANIFEST 2008-04-03 13:13:30.000000000 -0400 +++ new-Text-CSV_XS-0.37/MANIFEST 2008-04-03 13:13:30.000000000 -0400 @@ -24,6 +24,7 @@ t/70_rt.t Tests based on RT reports t/80_diag.t Error diagnostics t/util.pl Extra test utilities +t/getline_hr.t Test column_names() and getline_hr() examples/csv2xls Script to onvert CSV files to M$Excel examples/speed.pl Small benchmark script META.yml Module meta-data (added by MakeMaker) diff -rN -u old-Text-CSV_XS-0.37/t/getline_hr.t new-Text-CSV_XS-0.37/t/getline_hr.t --- old-Text-CSV_XS-0.37/t/getline_hr.t 1969-12-31 19:00:00.000000000 -0500 +++ new-Text-CSV_XS-0.37/t/getline_hr.t 2008-04-03 13:13:30.000000000 -0400 @@ -0,0 +1,77 @@ +#!/usr/bin/perl + +use strict; +$^W = 1; + + +use Config; +use Test::More; + +BEGIN { + if (exists $Config{useperlio} && + defined $Config{useperlio} && + $Config{useperlio} eq "define") { + plan tests => 7 ; + } + else { + plan skip_all => "No perlIO available"; + } +} + +BEGIN { + require_ok "Text::CSV_XS"; + plan skip_all => "Cannot load Text::CSV_XS" if $@; +} + +# Setup + my $io; # filehandle + my $io_str; # storage + my $csv = Text::CSV_XS->new; + # Store a row in the $io handle, which uses $io_str for storage. + open $io, "+>", \$io_str or die "_test.csv: $!"; + $csv->print ($io, ['av','bv']); + # reset file handle to beginning. + seek($io,0,0); + +{ + my $test = "calling getline_hr before column_names dies with hint you've done something wrong"; + eval { my $hr = $csv->getline_hr($io); }; + like($@,qr/column_names/, $test); +} + + +{ + my $test = 'reality check column_names'; + $csv->column_names('c','d'); + my @cols = $csv->column_names; + is_deeply(['c','d'],\@cols,$test); +} + +{ + my $test = 'column_names() called a second time with arrayref should work, and clobber current values'; + $csv->column_names(['a','b']); + my @cols = $csv->column_names; + is_deeply(['a','b'],\@cols,$test); +} + +{ + my $hr = $csv->getline_hr($io); + is($hr->{a},'av', "getline_hr test for first key"); + is($hr->{b},'bv', "getline_hr test for last key"); +} + +{ + my $test = "mismatched number of column names and values returns an error"; + # reset file handle to beginning. + seek($io,0,0); + $csv->column_names('a'); + eval { my $hr = $csv->getline_hr($io); }; + like($@, qr/doesn't match/, $test); + +} + + + + + +