jruby · Mar 1, 2015 · Mar 1, 2015 · Mar 2, 2015 · Mar 2, 2015 · Mar 2, 2015
diff --git a/spec/truffle/spec/rubysl/rubysl-strscan/spec/match_start_spec.rb b/spec/truffle/spec/rubysl/rubysl-strscan/spec/match_start_spec.rb
@@ -0,0 +1,21 @@
+require File.expand_path('../../../../../../ruby/spec_helper', __FILE__)
+
+describe "Regexp#match_start" do
+  it "matches a string N bytes from the beginning" do
+    m = /b/.match_start("ab", 1)
+    m[0].should == "b"
+  end
+
+  it "preserves zero length captures" do
+    r = /(a)?/
+    str = "ab"
+
+    m = r.match_start(str, 0)
+    m[0].should == "a"
+    m[1].should == "a"
+
+    m = r.match_start(str, 1)
+    m[0].should == ""
+    m[1].should be_nil
+  end
+end
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/beginning_of_line_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/beginning_of_line_tags.txt
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/bol_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/bol_tags.txt
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/check_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/check_tags.txt
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/check_until_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/check_until_tags.txt
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/dup_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/dup_tags.txt
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/element_reference_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/element_reference_tags.txt
@@ -1,6 +1,2 @@
 fails:StringScanner#[] returns the n-th subgroup in the most recent match
-fails:StringScanner#[] returns nil if index is outside of self
 fails:StringScanner#[] calls to_int on the given index
-fails:StringScanner#[] raises a TypeError if the given index is nil
-fails:StringScanner#[] raises a TypeError when a String is as argument
-fails:StringScanner#[] raises a TypeError when a Range is as argument
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/exist_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/exist_tags.txt
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/get_byte_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/get_byte_tags.txt
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/getbyte_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/getbyte_tags.txt
@@ -1,6 +1 @@
-fails:StringScanner#getbyte scans one byte and returns it
-fails:StringScanner#getbyte is not multi-byte character sensitive
-fails:StringScanner#getbyte returns nil at the end of the string
-fails:StringScanner#getbyte warns in verbose mode that the method is obsolete
-fails:StringScanner#getbyte returns an instance of String when passed a String subclass
 fails:StringScanner#getbyte taints the returned String if the input was tainted
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/getch_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/getch_tags.txt
@@ -1,5 +1,2 @@
-fails:StringScanner#getch scans one character and returns it
 fails:StringScanner#getch is multi-byte character sensitive
-fails:StringScanner#getch returns nil at the end of the string
-fails:StringScanner#getch returns an instance of String when passed a String subclass
 fails:StringScanner#getch taints the returned String if the input was tainted
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/inspect_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/inspect_tags.txt
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/match_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/match_tags.txt
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/matched_size_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/matched_size_tags.txt
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/matched_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/matched_tags.txt
@@ -1,6 +1 @@
-fails:StringScanner#matched returns the last matched string
-fails:StringScanner#matched returns nil if there's no match
-fails:StringScanner#matched returns an instance of String when passed a String subclass
 fails:StringScanner#matched taints the returned String if the input was tainted
-fails:StringScanner#matched? returns true if the last match was successful
-fails:StringScanner#matched? returns false if there's no match
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/pointer_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/pointer_tags.txt
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/pos_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/pos_tags.txt
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/post_match_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/post_match_tags.txt
@@ -1,4 +1 @@
-fails:StringScanner#post_match returns the post-match (in the regular expression sense) of the last scan
-fails:StringScanner#post_match returns nil if there's no match
-fails:StringScanner#post_match returns an instance of String when passed a String subclass
 fails:StringScanner#post_match taints the returned String if the input was tainted
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/pre_match_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/pre_match_tags.txt
@@ -1,6 +1 @@
-fails:StringScanner#pre_match returns the pre-match (in the regular expression sense) of the last scan
-fails:StringScanner#pre_match returns nil if there's no match
-fails:StringScanner#pre_match is more than just the data from the last match
-fails:StringScanner#pre_match is not changed when the scanner's position changes
-fails:StringScanner#pre_match returns an instance of String when passed a String subclass
 fails:StringScanner#pre_match taints the returned String if the input was tainted
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/reset_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/reset_tags.txt
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/rest_size_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/rest_size_tags.txt
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/rest_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/rest_tags.txt
@@ -1,4 +1 @@
-fails:StringScanner#rest returns the rest of the string
-fails:StringScanner#rest returns an instance of String when passed a String subclass
 fails:StringScanner#rest taints the returned String if the input was tainted
-fails:StringScanner#rest? returns true if there is more data in the string
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/restsize_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/restsize_tags.txt
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/scan_full_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/scan_full_tags.txt
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/scan_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/scan_tags.txt
@@ -1,9 +1,2 @@
-fails:StringScanner#scan returns the matched string
 fails:StringScanner#scan returns the first character for a multi byte string with no KCODE
-fails:StringScanner#scan returns the matched string for a multi byte string with KCODE
-fails:StringScanner#scan returns the matched string for a multi byte string with unicode regexp
-fails:StringScanner#scan returns the matched string for a multi byte string
 fails:StringScanner#scan treats ^ as matching from the beginning of the current position
-fails:StringScanner#scan returns nil if there's no match
-fails:StringScanner#scan returns nil when there is no more to scan
-fails:StringScanner#scan returns an empty string when the pattern matches empty
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/scan_until_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/scan_until_tags.txt
@@ -1,3 +1 @@
-fails:StringScanner#scan_until returns the substring up to and including the end of the match
-fails:StringScanner#scan_until returns nil if there's no match
 fails:StringScanner#scan_until can match anchors properly
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/search_full_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/search_full_tags.txt
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/skip_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/skip_tags.txt
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/skip_until_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/skip_until_tags.txt
diff --git a/spec/truffle/tags/rubysl/rubysl-strscan/spec/unscan_tags.txt b/spec/truffle/tags/rubysl/rubysl-strscan/spec/unscan_tags.txt
diff --git a/truffle/src/main/java/org/jruby/truffle/nodes/core/MatchDataNodes.java b/truffle/src/main/java/org/jruby/truffle/nodes/core/MatchDataNodes.java
@@ -44,7 +44,7 @@ public GetIndexNode(GetIndexNode prev) {
         public Object getIndex(RubyMatchData matchData, int index) {
             notDesignedForCompilation();
 
-            if (index >= matchData.getValues().length) {
+            if (index >= matchData.getValues().length || index < 0) {
                 return getContext().getCoreLibrary().getNilObject();
             } else {
                 return matchData.getValues()[index];

diff --git a/truffle/src/main/java/org/jruby/truffle/nodes/core/RegexpNodes.java b/truffle/src/main/java/org/jruby/truffle/nodes/core/RegexpNodes.java
@@ -11,10 +11,12 @@
 
 import static org.jruby.util.StringSupport.CR_7BIT;
 
+import org.joni.Region;
 import org.jruby.truffle.nodes.RubyNode;
 import org.jruby.truffle.runtime.RubyContext;
 import org.jruby.truffle.runtime.control.RaiseException;
 import org.jruby.truffle.runtime.core.RubyBasicObject;
+import org.jruby.truffle.runtime.core.RubyMatchData;
 import org.jruby.truffle.runtime.core.RubyRegexp;
 import org.jruby.truffle.runtime.core.RubyString;
 import org.jruby.truffle.runtime.core.RubySymbol;
@@ -232,6 +234,28 @@ public Object match(RubyRegexp regexp, RubyString string) {
 
     }
 
+    @CoreMethod(names = "match_start", required = 2)
+    public abstract static class MatchStartNode extends CoreMethodNode {
+
+        public MatchStartNode(RubyContext context, SourceSection sourceSection) {
+            super(context, sourceSection);
+        }
+
+        public MatchStartNode(MatchStartNode prev) {
+            super(prev);
+        }
+
+        @Specialization
+        public Object matchStart(RubyRegexp regexp, RubyString string, int startPos) {
+            final Object matchResult = regexp.matchCommon(string, false, false, startPos);
+            if (matchResult instanceof RubyMatchData && ((RubyMatchData) matchResult).getNumberOfRegions() > 0
+                && ((RubyMatchData) matchResult).getRegion().beg[0] == startPos) {
+                return matchResult;
+            }
+            return getContext().getCoreLibrary().getNilObject();
+        }
+    }
+
     @CoreMethod(names = "options")
     public abstract static class OptionsNode extends CoreMethodNode {
 
@@ -292,6 +316,23 @@ public RubyString quote(RubySymbol raw) {
 
     }
 
+    @CoreMethod(names = "search_from", required = 2)
+    public abstract static class SearchFromNode extends CoreMethodNode {
+
+        public SearchFromNode(RubyContext context, SourceSection sourceSection) {
+            super(context, sourceSection);
+        }
+
+        public SearchFromNode(SearchFromNode prev) {
+            super(prev);
+        }
+
+        @Specialization
+        public Object searchFrom(RubyRegexp regexp, RubyString string, int startPos) {
+            return regexp.matchCommon(string, false, false, startPos);
+        }
+    }
+
     @CoreMethod(names = "source")
     public abstract static class SourceNode extends CoreMethodNode {
 

diff --git a/truffle/src/main/java/org/jruby/truffle/runtime/core/RubyMatchData.java b/truffle/src/main/java/org/jruby/truffle/runtime/core/RubyMatchData.java
@@ -121,6 +121,10 @@ public RubyString getGlobal() {
         return global;
     }
 
+    public Region getRegion() {
+        return region;
+    }
+
     public RubyString getSource() {
         return source;
     }

diff --git a/truffle/src/main/java/org/jruby/truffle/runtime/core/RubyRegexp.java b/truffle/src/main/java/org/jruby/truffle/runtime/core/RubyRegexp.java
@@ -98,11 +98,16 @@ public RegexpOptions getOptions() {
 
     @CompilerDirectives.TruffleBoundary
     public Object matchCommon(RubyString source, boolean operator, boolean setNamedCaptures) {
+        return matchCommon(source, operator, setNamedCaptures, 0);
+    }
+
+    @CompilerDirectives.TruffleBoundary
+    public Object matchCommon(RubyString source, boolean operator, boolean setNamedCaptures, int startPos) {
         final byte[] stringBytes = source.getByteList().bytes();
         final Matcher matcher = regex.matcher(stringBytes);
         int range = stringBytes.length;
 
-        return matchCommon(source, operator, setNamedCaptures, matcher, 0, range);
+        return matchCommon(source, operator, setNamedCaptures, matcher, startPos, range);
     }
 
     @CompilerDirectives.TruffleBoundary