From 63fe276dac70f64abf4719493634869c9e559568 Mon Sep 17 00:00:00 2001
From: Chris Nehren <chris@plainblack.com>
Date: Thu, 31 Jul 2008 22:02:09 +0000
Subject: [PATCH] Fixed: splitCSV and joinCSV had issues with complex CSV data.
 They now use Text::CSV_XS internally.

---
 docs/changelog/7.x.x.txt |  1 +
 lib/WebGUI/Text.pm       | 63 +++++++++++++---------------------------
 t/Text.t                 |  3 +-
 3 files changed, 23 insertions(+), 44 deletions(-)

diff --git a/docs/changelog/7.x.x.txt b/docs/changelog/7.x.x.txt
index ee6b0f28a..54dbd7c36 100644
--- a/docs/changelog/7.x.x.txt
+++ b/docs/changelog/7.x.x.txt
@@ -13,6 +13,7 @@
  - fixed: product title problem
  - fixed: adding payment brings me to /
  - fixed: Gallery uses too much disk space. Added a way to change an images pixel density.
+ - fixed: splitCSV and joinCSV had issues with complex CSV data
 
 7.5.18
  - fixed: Collateral Image Manager broken in Firefox 3
diff --git a/lib/WebGUI/Text.pm b/lib/WebGUI/Text.pm
index 4776338a7..4569de3e7 100644
--- a/lib/WebGUI/Text.pm
+++ b/lib/WebGUI/Text.pm
@@ -18,6 +18,7 @@ package WebGUI::Text;
 use strict;
 #use warnings;
 
+use Text::CSV_XS;
 use base 'Exporter';
 
 our @EXPORT_OK = qw(
@@ -31,6 +32,9 @@ our %EXPORT_TAGS = (
 	);
 
 
+# use a single CSV object instead of reconstructing one repeatedly
+my $csv = Text::CSV_XS->new( { binary => 1 } );
+
 =head1 NAME
 
 WebGUI::Text - Routines for manipulating text.
@@ -63,24 +67,14 @@ string according to the de-facto standard outlined by RFC 4180.
 =cut
 
 sub joinCSV {
-	my @input	= @_;
-	my @fixed;	# The properly escaped data
-	for my $i (@input) {
-		# Ignore all characters that aren't ASCII printable characters
-		$i =~ s/[^\x09\x20-\x7e]//g;
-		
-		# All strings with these chars in them must be quoted
-		if ($i =~ /[",\n\t]/ || $i =~ /^\s|\s$/s) {
-			# " must be doubled ("")
-			$i =~ s/"/""/g;
-			
-			$i = qq{"$i"};
-		}
-		
-		push @fixed, $i;
-	}
-	
-	return join ",",@fixed;
+	my @inputColumns    = @_;
+    $csv->combine(@inputColumns);
+    my $joinedLine      = $csv->string;
+    if(my $errorString = $csv->error_diag) {
+        warn "Problems parsing @inputColumns: $errorString";
+        return;
+    }
+    return $joinedLine;
 }
 
 
@@ -94,31 +88,14 @@ Splits a CSV string and fixes any escaping done.
 =cut
 
 sub splitCSV {
-	my $s	= shift;
-	
-	# Split on ,
-	# Negative LIMIT so that empty trailing fields are preserved
-	my @array = split /,/, $s, -1; 
-	
-	for (my $i = 0; $i < @array; $i++) {
-		# Fix quoted strings being used to escape commas.
-		# If it begins with a " but doesn't end with an odd number of "
-		#  shift, add to previous, and try again
-		if ($array[$i] =~ /^"/s && length(($array[$i] =~ m/("*)$/s)[0]) % 2 == 0 ) {
-			# If there are no more elements, this line is erroneous
-			if ($i+1 > @array) { warn "Error parsing CSV line."; return undef; }
-			$array[$i] .= ",".splice(@array,$i+1,1);
-			redo;
-		}
-		
-		# Remove quotes on end of string
-		$array[$i] =~ s/^"|"$//sg;
-		
-		# Fix doubled quotes
-		$array[$i] =~ s/""/"/g;
-	}
-	
-	return @array;
+	my $inputString = shift;
+    $csv->parse($inputString);
+    my @splitColumns = $csv->fields;
+    if(my $errorString = $csv->error_diag) {
+        warn "Problems parsing $inputString: $errorString";
+        return;
+    }
+    return @splitColumns;
 }
 
 
diff --git a/t/Text.t b/t/Text.t
index 8f062cf99..9531855ba 100644
--- a/t/Text.t
+++ b/t/Text.t
@@ -23,7 +23,8 @@ my @tests =
      ['all null', ['', '', ''], ',,'],
      ['single null', [], ''],
      ['escape commas', ['w,x', 'y,z'], '"w,x","y,z"'],
-     ['escape double quotes', ['abc"def', 'ghi-jkl', 'mnop'], '"abc""def",ghi-jkl,mnop']);
+     ['escape double quotes', ['abc"def', 'ghi-jkl', 'mnop'], '"abc""def",ghi-jkl,mnop'],
+     ['cruel embedded newlines', ['foo', 'bar', 'baz', "hello\nworld", 'how are you'], qq{foo,bar,baz,"hello\nworld","how are you"}]);
 plan(tests => scalar(@tests) * 2);
 
 foreach my $testspec (@tests) {