Fix StringScanner and patterns starting with ^

Author: chrisseaton

CHANGELOG.md

@@ -8,6 +8,7 @@ Bug fixes:
   using XML conversion options and a new destination encoding (#1545).
 * Fixed a bug where a raised cloned exception would be caught as the
   original exception (#1542).
+* Fixed a bug with `StringScanner` and patterns starting with `^` (#1544).
 
 Compatibility:
 

lib/truffle/strscan.rb

@@ -312,18 +312,19 @@ def scan_internal(pattern, advance_pos, getstr, headonly)
     end
     raise ArgumentError, 'uninitialized StringScanner object' unless @string
 
-    @match = nil
-
-    if headonly
-      # NOTE - match_start is an Oniguruma feature that Rubinius exposes.
-      # We use it here to avoid creating a new Regexp with '^' prepended.
-      @match = pattern.match_start @string, @pos
-    else
-      # NOTE - search_from is an Oniguruma feature that Rubinius exposes.
-      # We use it so we can begin the search in the middle of the string
-      @match = pattern.search_from @string, @pos
+    # If the pattern already starts with a ^, and we're not at the start of
+    # the string, then we can't match as normal because match_from still tries
+    # to match the ^ at position 0 even though it's looking from point pos
+    # onwards, even if headonly is set. Instead, remove the ^. This could
+    # possibly be fixed in Joni instead, or maybe there is already some option
+    # we're not using.
+    
+    if pattern.source[0] == '^' && pos > 0
+      pattern = Regexp.new(pattern.source[1..-1])
+      headonly = true
     end
 
+    @match = pattern.match_onwards @string, pos, headonly
     return nil unless @match
 
     fin = @match.byte_end(0)

spec/tags/library/stringscanner/scan_tags.txt

@@ -1,4 +1,3 @@
-fails:StringScanner#scan treats ^ as matching from the beginning of the current position
 graalvm:StringScanner#scan returns the matched string
 graalvm:StringScanner#scan returns nil if there's no match
 graalvm:StringScanner#scan returns nil when there is no more to scan

spec/tags/library/stringscanner/scan_until_tags.txt

@@ -1,3 +1,2 @@
-fails:StringScanner#scan_until can match anchors properly
 graalvm:StringScanner#scan_until returns the substring up to and including the end of the match
 graalvm:StringScanner#scan_until returns nil if there's no match

src/main/java/org/truffleruby/core/regexp/RegexpNodes.java

@@ -168,18 +168,18 @@ public int hash(DynamicObject regexp) {
     }
 
     @NonStandard
-    @CoreMethod(names = "match_start", required = 2, lowerFixnum = 2)
-    public abstract static class MatchStartNode extends CoreMethodArrayArgumentsNode {
+    @CoreMethod(names = "match_onwards", required = 3, lowerFixnum = 2)
+    public abstract static class MatchOnwardsNode extends CoreMethodArrayArgumentsNode {
 
         @Child private TruffleRegexpNodes.MatchNode matchNode = TruffleRegexpNodes.MatchNode.create();
         @Child private RopeNodes.BytesNode bytesNode = RopeNodes.BytesNode.create();
 
         @Specialization(guards = "isRubyString(string)")
-        public Object matchStart(DynamicObject regexp, DynamicObject string, int startPos) {
+        public Object matchOnwards(DynamicObject regexp, DynamicObject string, int startPos, boolean atStart) {
             final Rope rope = StringOperations.rope(string);
             final Matcher matcher = createMatcher(getContext(), regexp, rope, bytesNode.execute(rope), true);
             int range = rope.byteLength();
-            return matchNode.execute(regexp, string, matcher, startPos, range, true);
+            return matchNode.execute(regexp, string, matcher, startPos, range, atStart);
         }
     }
 
@@ -224,23 +224,6 @@ private StringNodes.MakeStringNode getMakeStringNode() {
         }
     }
 
-    @NonStandard
-    @CoreMethod(names = "search_from", required = 2, lowerFixnum = 2)
-    public abstract static class SearchFromNode extends CoreMethodArrayArgumentsNode {
-
-        @Child private TruffleRegexpNodes.MatchNode matchNode = TruffleRegexpNodes.MatchNode.create();
-        @Child private RopeNodes.BytesNode bytesNode = RopeNodes.BytesNode.create();
-
-        @Specialization(guards = "isRubyString(string)")
-        public Object searchFrom(DynamicObject regexp, DynamicObject string, int startPos) {
-            final Rope rope = StringOperations.rope(string);
-            final Matcher matcher = createMatcher(getContext(), regexp, rope, bytesNode.execute(rope), true);
-            int range = StringOperations.rope(string).byteLength();
-
-            return matchNode.execute(regexp, string, matcher, startPos, range, false);
-        }
-    }
-
     @Primitive(name = "regexp_search_from_binary", lowerFixnum = 2)
     public abstract static class SearchFromBinaryNode extends CoreMethodArrayArgumentsNode {