[GR-32617] Warn when TRegex needs to use backtracking for a Regexp

PullRequest: truffleruby/2836

Author: eregon

mx.truffleruby/suite.py

@@ -7,7 +7,7 @@
             {
                 "name": "regex",
                 "subdir": True,
-                "version": "a1346dffebe541333754360e825b2185d60cb005",
+                "version": "bc35c63a6db0aa8faf5471b74f9771c7be8f7a9b",
                 "urls": [
                     {"url": "https://github.com/oracle/graal.git", "kind": "git"},
                     {"url": "https://curio.ssw.jku.at/nexus/content/repositories/snapshots", "kind": "binary"},
@@ -16,7 +16,7 @@
             {
                 "name": "sulong",
                 "subdir": True,
-                "version": "a1346dffebe541333754360e825b2185d60cb005",
+                "version": "bc35c63a6db0aa8faf5471b74f9771c7be8f7a9b",
                 "urls": [
                     {"url": "https://github.com/oracle/graal.git", "kind": "git"},
                     {"url": "https://curio.ssw.jku.at/nexus/content/repositories/snapshots", "kind": "binary"},

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

@@ -28,6 +28,8 @@
 import org.truffleruby.core.rope.Rope;
 import org.truffleruby.core.rope.RopeBuilder;
 import org.truffleruby.core.rope.RopeOperations;
+import org.truffleruby.interop.InteropNodes;
+import org.truffleruby.interop.TranslateInteropExceptionNode;
 import org.truffleruby.language.Nil;
 import org.truffleruby.language.control.DeferredRaiseException;
 
@@ -75,6 +77,15 @@ public Object compile(RubyContext context, RubyRegexp regexp, boolean atStart, R
                         atStart,
                         encoding);
             }
+        } else if (isBacktracking(tregex)) {
+            if (context.getOptions().WARN_TRUFFLE_REGEX_COMPILE_FALLBACK) {
+                node.getWarnOnFallbackNode().call(
+                        context.getCoreLibrary().truffleRegexpOperationsModule,
+                        "warn_backtracking",
+                        regexp,
+                        atStart,
+                        encoding);
+            }
         }
 
         if (encoding == Encodings.US_ASCII) {
@@ -108,6 +119,14 @@ public Object compile(RubyContext context, RubyRegexp regexp, boolean atStart, R
         return tregex;
     }
 
+    private static boolean isBacktracking(Object tregex) {
+        return (boolean) InteropNodes.readMember(
+                InteropLibrary.getUncached(),
+                tregex,
+                "isBacktracking",
+                TranslateInteropExceptionNode.getUncached());
+    }
+
     public static String toTRegexEncoding(Encoding encoding) {
         if (encoding == UTF8Encoding.INSTANCE) {
             return "UTF-8";

src/main/ruby/truffleruby/core/truffle/regexp_operations.rb

@@ -101,6 +101,10 @@ def self.warn_fallback_regex(re, at_start, encoding)
       warn "Regexp #{re.inspect} at_start=#{at_start} encoding=#{encoding} cannot be compiled to a Truffle regexp and fell back to Joni", uplevel: 1
     end
 
+    def self.warn_backtracking(re, at_start, encoding)
+      warn "Regexp #{re.inspect} at_start=#{at_start} encoding=#{encoding} requires backtracking and will not match in linear time", uplevel: 1
+    end
+
     def self.match_args_to_string(re, str, from, to, at_start, start, suffix)
       "match_in_region(#{re.inspect}, #{str.inspect}@#{str.encoding}, #{from}, #{to}, #{at_start}, #{start}) #{suffix}"
     end