Update to ruby/spec@944ea57

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@66622 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

3 files changed, 186 insertions, 34 deletions

diff --git a/spec/ruby/core/matchdata/begin_spec.rb b/spec/ruby/core/matchdata/begin_spec.rb
index b791018633..85c454da56 100644
--- a/spec/ruby/core/matchdata/begin_spec.rb
+++ b/spec/ruby/core/matchdata/begin_spec.rb
@@ -3,28 +3,102 @@
 require_relative '../../spec_helper'
 
 describe "MatchData#begin" do
-  it "returns the offset of the start of the nth element" do
-    match_data = /(.)(.)(\d+)(\d)/.match("THX1138.")
-    match_data.begin(0).should == 1
-    match_data.begin(2).should == 2
-  end
+  context "when passed an integer argument" do
+    it "returns the character offset of the start of the nth element" do
+      match_data = /(.)(.)(\d+)(\d)/.match("THX1138.")
+      match_data.begin(0).should == 1
+      match_data.begin(2).should == 2
+    end
 
-  it "returns nil when the nth match isn't found" do
-    match_data = /something is( not)? (right)/.match("something is right")
-    match_data.begin(1).should be_nil
+    it "returns nil when the nth match isn't found" do
+      match_data = /something is( not)? (right)/.match("something is right")
+      match_data.begin(1).should be_nil
+    end
+
+    it "returns the character offset for multi-byte strings" do
+      match_data = /(.)(.)(\d+)(\d)/.match("TñX1138.")
+      match_data.begin(0).should == 1
+      match_data.begin(2).should == 2
+    end
+
+    not_supported_on :opal do
+      it "returns the character offset for multi-byte strings with unicode regexp" do
+        match_data = /(.)(.)(\d+)(\d)/u.match("TñX1138.")
+        match_data.begin(0).should == 1
+        match_data.begin(2).should == 2
+      end
+    end
+
+    it "tries to convert the passed argument to an Integer using #to_int" do
+      obj = mock('to_int')
+      obj.should_receive(:to_int).and_return(2)
+
+      match_data = /(.)(.)(\d+)(\d)/.match("THX1138.")
+      match_data.begin(obj).should == 2
+    end
   end
 
-  it "returns the offset for multi byte strings" do
-    match_data = /(.)(.)(\d+)(\d)/.match("TñX1138.")
-    match_data.begin(0).should == 1
-    match_data.begin(2).should == 2
+  context "when passed a String argument" do
+    it "return the character offset of the start of the named capture" do
+      match_data = /(?<a>.)(.)(?<b>\d+)(\d)/.match("THX1138.")
+      match_data.begin("a").should == 1
+      match_data.begin("b").should == 3
+    end
+
+    it "returns the character offset for multi byte strings" do
+      match_data = /(?<a>.)(.)(?<b>\d+)(\d)/.match("TñX1138.")
+      match_data.begin("a").should == 1
+      match_data.begin("b").should == 3
+    end
+
+    not_supported_on :opal do
+      it "returns the character offset for multi byte strings with unicode regexp" do
+        match_data = /(?<a>.)(.)(?<b>\d+)(\d)/u.match("TñX1138.")
+        match_data.begin("a").should == 1
+        match_data.begin("b").should == 3
+      end
+    end
+
+    it "returns the character offset for the farthest match when multiple named captures use the same name" do
+      match_data = /(?<a>.)(.)(?<a>\d+)(\d)/.match("THX1138.")
+      match_data.begin("a").should == 3
+    end
+
+    it "returns the character offset for multi-byte names" do
+      match_data = /(?<æ>.)(.)(?<b>\d+)(\d)/.match("THX1138.")
+      match_data.begin("æ").should == 1
+    end
   end
 
-  not_supported_on :opal do
-    it "returns the offset for multi byte strings with unicode regexp" do
-      match_data = /(.)(.)(\d+)(\d)/u.match("TñX1138.")
-      match_data.begin(0).should == 1
-      match_data.begin(2).should == 2
+  context "when passed a Symbol argument" do
+    it "return the character offset of the start of the named capture" do
+      match_data = /(?<a>.)(.)(?<b>\d+)(\d)/.match("THX1138.")
+      match_data.begin(:a).should == 1
+      match_data.begin(:b).should == 3
+    end
+
+    it "returns the character offset for multi byte strings" do
+      match_data = /(?<a>.)(.)(?<b>\d+)(\d)/.match("TñX1138.")
+      match_data.begin(:a).should == 1
+      match_data.begin(:b).should == 3
+    end
+
+    not_supported_on :opal do
+      it "returns the character offset for multi byte strings with unicode regexp" do
+        match_data = /(?<a>.)(.)(?<b>\d+)(\d)/u.match("TñX1138.")
+        match_data.begin(:a).should == 1
+        match_data.begin(:b).should == 3
+      end
+    end
+
+    it "returns the character offset for the farthest match when multiple named captures use the same name" do
+      match_data = /(?<a>.)(.)(?<a>\d+)(\d)/.match("THX1138.")
+      match_data.begin(:a).should == 3
+    end
+
+    it "returns the character offset for multi-byte names" do
+      match_data = /(?<æ>.)(.)(?<b>\d+)(\d)/.match("THX1138.")
+      match_data.begin(:æ).should == 1
     end
   end
 end
diff --git a/spec/ruby/core/matchdata/end_spec.rb b/spec/ruby/core/matchdata/end_spec.rb
index f6f3e1a281..d01b0a8b30 100644
--- a/spec/ruby/core/matchdata/end_spec.rb
+++ b/spec/ruby/core/matchdata/end_spec.rb
@@ -3,28 +3,102 @@
 require_relative '../../spec_helper'
 
 describe "MatchData#end" do
-  it "returns the offset of the end of the nth element" do
-    match_data = /(.)(.)(\d+)(\d)/.match("THX1138.")
-    match_data.end(0).should == 7
-    match_data.end(2).should == 3
-  end
+  context "when passed an integer argument" do
+    it "returns the character offset of the end of the nth element" do
+      match_data = /(.)(.)(\d+)(\d)/.match("THX1138.")
+      match_data.end(0).should == 7
+      match_data.end(2).should == 3
+    end
 
-  it "returns nil when the nth match isn't found" do
-    match_data = /something is( not)? (right)/.match("something is right")
-    match_data.end(1).should be_nil
+    it "returns nil when the nth match isn't found" do
+      match_data = /something is( not)? (right)/.match("something is right")
+      match_data.end(1).should be_nil
+    end
+
+    it "returns the character offset for multi-byte strings" do
+      match_data = /(.)(.)(\d+)(\d)/.match("TñX1138.")
+      match_data.end(0).should == 7
+      match_data.end(2).should == 3
+    end
+
+    not_supported_on :opal do
+      it "returns the character offset for multi-byte strings with unicode regexp" do
+        match_data = /(.)(.)(\d+)(\d)/u.match("TñX1138.")
+        match_data.end(0).should == 7
+        match_data.end(2).should == 3
+      end
+    end
+
+    it "tries to convert the passed argument to an Integer using #to_int" do
+      obj = mock('to_int')
+      obj.should_receive(:to_int).and_return(2)
+
+      match_data = /(.)(.)(\d+)(\d)/.match("THX1138.")
+      match_data.end(obj).should == 3
+    end
   end
 
-  it "returns the offset for multi byte strings" do
-    match_data = /(.)(.)(\d+)(\d)/.match("TñX1138.")
-    match_data.end(0).should == 7
-    match_data.end(2).should == 3
+  context "when passed a String argument" do
+    it "return the character offset of the start of the named capture" do
+      match_data = /(?<a>.)(.)(?<b>\d+)(\d)/.match("THX1138.")
+      match_data.end("a").should == 2
+      match_data.end("b").should == 6
+    end
+
+    it "returns the character offset for multi byte strings" do
+      match_data = /(?<a>.)(.)(?<b>\d+)(\d)/.match("TñX1138.")
+      match_data.end("a").should == 2
+      match_data.end("b").should == 6
+    end
+
+    not_supported_on :opal do
+      it "returns the character offset for multi byte strings with unicode regexp" do
+        match_data = /(?<a>.)(.)(?<b>\d+)(\d)/u.match("TñX1138.")
+        match_data.end("a").should == 2
+        match_data.end("b").should == 6
+      end
+    end
+
+    it "returns the character offset for the farthest match when multiple named captures use the same name" do
+      match_data = /(?<a>.)(.)(?<a>\d+)(\d)/.match("THX1138.")
+      match_data.end("a").should == 6
+    end
+
+    it "returns the character offset for multi-byte names" do
+      match_data = /(?<æ>.)(.)(?<b>\d+)(\d)/.match("THX1138.")
+      match_data.end("æ").should == 2
+    end
   end
 
-  not_supported_on :opal do
-    it "returns the offset for multi byte strings with unicode regexp" do
-      match_data = /(.)(.)(\d+)(\d)/u.match("TñX1138.")
-      match_data.end(0).should == 7
-      match_data.end(2).should == 3
+  context "when passed a Symbol argument" do
+    it "return the character offset of the start of the named capture" do
+      match_data = /(?<a>.)(.)(?<b>\d+)(\d)/.match("THX1138.")
+      match_data.end(:a).should == 2
+      match_data.end(:b).should == 6
+    end
+
+    it "returns the character offset for multi byte strings" do
+      match_data = /(?<a>.)(.)(?<b>\d+)(\d)/.match("TñX1138.")
+      match_data.end(:a).should == 2
+      match_data.end(:b).should == 6
+    end
+
+    not_supported_on :opal do
+      it "returns the character offset for multi byte strings with unicode regexp" do
+        match_data = /(?<a>.)(.)(?<b>\d+)(\d)/u.match("TñX1138.")
+        match_data.end(:a).should == 2
+        match_data.end(:b).should == 6
+      end
+    end
+
+    it "returns the character offset for the farthest match when multiple named captures use the same name" do
+      match_data = /(?<a>.)(.)(?<a>\d+)(\d)/.match("THX1138.")
+      match_data.end(:a).should == 6
+    end
+
+    it "returns the character offset for multi-byte names" do
+      match_data = /(?<æ>.)(.)(?<b>\d+)(\d)/.match("THX1138.")
+      match_data.end(:æ).should == 2
     end
   end
 end
diff --git a/spec/ruby/core/matchdata/named_captures_spec.rb b/spec/ruby/core/matchdata/named_captures_spec.rb
index 588b607a44..0b0771355f 100644
--- a/spec/ruby/core/matchdata/named_captures_spec.rb
+++ b/spec/ruby/core/matchdata/named_captures_spec.rb
@@ -9,5 +9,9 @@ ruby_version_is '2.4' do
     it 'prefers later captures' do
       /\A(?<a>.)(?<b>.)(?<b>.)(?<a>.)\z/.match('0123').named_captures.should == { 'a' => '3', 'b' => '2' }
     end
+
+    it 'returns the latest matched capture, even if a later one that does not match exists' do
+      /\A(?<a>.)(?<b>.)(?<b>.)(?<a>.)?\z/.match('012').named_captures.should == { 'a' => '0', 'b' => '2' }
+    end
   end
 end