From e38fd48aaf41f94eaec39a639a7f6c5db62c22e8 Mon Sep 17 00:00:00 2001
From: Naohisa Goto <ng@bioruby.org>
Date: Tue, 9 Sep 2008 20:22:30 +0900
Subject: [PATCH] Bug fix: GFF2 attributes parser misunderstand semicolons

 * Now, the priority of '"' (double quote) is greater than ';'
   (semicolon). To keep compatibility from older BioRuby,
   the '\;' can still be used anywhere in attributes.
 * Some tests are added.
---
 lib/bio/db/gff.rb            | 31 +++++++++++++++++++---
 test/unit/bio/db/test_gff.rb | 51 ++++++++++++++++++++++++++++++++++++
 2 files changed, 79 insertions(+), 3 deletions(-)

diff --git a/lib/bio/db/gff.rb b/lib/bio/db/gff.rb
index e3a43ecfd..24e1fab1b 100644
--- a/lib/bio/db/gff.rb
+++ b/lib/bio/db/gff.rb
@@ -121,9 +121,34 @@ def initialize(str)
 
       def parse_attributes(attributes)
         hash = Hash.new
-        scanner = StringScanner.new(attributes)
-        while scanner.scan(/(.*[^\\])\;/) or scanner.scan(/(.+)/)
-          key, value = scanner[1].split(' ', 2)
+
+        ary = attributes.split(/\;/)
+        while x = ary.shift
+          i0 = 0
+          while pos = x.index('"', i0)
+            i0 = pos + 1
+            while true
+              if pos = x.index('"', i0) then
+                i0 = pos + 1
+                break unless x[pos - 1, 1] == "\\"
+              else
+                if y = ary.shift
+                  x.concat ';'
+                  x.concat y
+                else
+                  break
+                end
+              end
+            end
+          end #while pos = ...
+
+          # To keep compatibility, the '\;' is concatenated again.
+          if x[-1, 1] == "\\" and ary[0] then
+            x.concat ';'
+            x.concat ary.shift
+          end
+
+          key, value = x.split(' ', 2)
           key.strip!
           value.strip! if value
           hash[key] = value
diff --git a/test/unit/bio/db/test_gff.rb b/test/unit/bio/db/test_gff.rb
index c12662998..7863db7ff 100644
--- a/test/unit/bio/db/test_gff.rb
+++ b/test/unit/bio/db/test_gff.rb
@@ -119,6 +119,57 @@ def test_add_seqname
 
   end # class TestGFFRecordConstruct
 
+
+  class TestGFF2ComplexAttributes < Test::Unit::TestCase
+
+    # The test string is privided by Tomoaki NISHIYAMA.
+    def test_attributes_case1
+      str = "LG_I\tJGI\tCDS\t11052\t11064\t.\t-\t0\tname \"grail3.0116000101\"; proteinId 639579; exonNumber 3\n"
+
+      attributes = {
+        "name"       => "\"grail3.0116000101\"",
+        "proteinId"  => "639579",
+        "exonNumber" => "3"
+      }
+
+      assert_equal(attributes,
+                   Bio::GFF::GFF2::Record.new(str).attributes)
+    end
+
+    # The test string is privided by Tomoaki NISHIYAMA and modified.
+    def test_attributes_case2
+      str = "LG_I\tJGI\tCDS\t11052\t11064\t.\t-\t0\tname \"grail3.0116000101\"; proteinId 639579; exonNumber 3; Note \"Semicolons ; and \;, and quote \\\" can be OK\"; Comment \"This is the \\\"comment\\\"\"\n"
+
+      attributes = {
+        "name"       => "\"grail3.0116000101\"",
+        "proteinId"  => "639579",
+        "exonNumber" => "3",
+        "Note" => "\"Semicolons ; and \;, and quote \\\" can be OK\"",
+        "Comment" => "\"This is the \\\"comment\\\"\""
+      }
+
+      assert_equal(attributes,
+                   Bio::GFF::GFF2::Record.new(str).attributes)
+    end
+
+    def test_attributes_compatibility_backslash_semicolon
+      str =<<END_OF_DATA
+I	sgd	gene	151453	151591	.	+	.	Gene "CEN1" ; Note "Chromosome I Centromere"; Semicolon a "b;c" d "e;f;g" h; Illegal a\\;b c d; Comment "a ; b"
+END_OF_DATA
+
+      attributes = {
+        'Gene' => '"CEN1"',
+        'Note' => '"Chromosome I Centromere"',
+        'Semicolon' => 'a "b;c" d "e;f;g" h',
+        'Illegal' => 'a\;b c d',
+        'Comment' => '"a ; b"'
+      }
+      assert_equal(attributes,
+                   Bio::GFF::GFF2::Record.new(str).attributes)
+    end
+
+  end #class TestGFF2ComplexAttributes
+
   class TestGFF3 < Test::Unit::TestCase
     def setup
       @data =<<END_OF_DATA