From 9881ef5b12202eb4ad0b6da2cac0ecfbbd765972 Mon Sep 17 00:00:00 2001
From: Zbynek Slajchrt <zbynek.slajchrt@oracle.com>
Date: Mon, 12 Feb 2018 19:56:22 +0100
Subject: [PATCH] '.' matches a new line for non-Perl regexp

---
 .../truffle/r/nodes/builtin/base/GrepFunctions.java      | 9 +++++----
 .../truffle/r/test/builtins/TestBuiltin_grepl.java       | 4 ++++
 .../oracle/truffle/r/test/builtins/TestBuiltin_gsub.java | 5 +++++
 3 files changed, 14 insertions(+), 4 deletions(-)

diff --git a/com.oracle.truffle.r.nodes.builtin/src/com/oracle/truffle/r/nodes/builtin/base/GrepFunctions.java b/com.oracle.truffle.r.nodes.builtin/src/com/oracle/truffle/r/nodes/builtin/base/GrepFunctions.java
index 2799b9c753..c81bcc02d8 100644
--- a/com.oracle.truffle.r.nodes.builtin/src/com/oracle/truffle/r/nodes/builtin/base/GrepFunctions.java
+++ b/com.oracle.truffle.r.nodes.builtin/src/com/oracle/truffle/r/nodes/builtin/base/GrepFunctions.java
@@ -535,10 +535,11 @@ public class GrepFunctions {
                     } else {
                         replacement = convertGroups(replacement);
 
+                        Matcher matcher = Pattern.compile(pattern, Pattern.DOTALL).matcher(input);
                         if (gsub) {
-                            value = input.replaceAll(pattern, replacement);
+                            value = matcher.replaceAll(replacement);
                         } else {
-                            value = input.replaceFirst(pattern, replacement);
+                            value = matcher.replaceFirst(replacement);
                         }
                     }
                     result[i] = value;
@@ -958,7 +959,7 @@ public class GrepFunctions {
             if (pattern.length() > 0 && pattern.charAt(0) == '*') {
                 actualPattern = pattern.substring(1);
             }
-            return Pattern.compile(actualPattern, ignoreCase ? Pattern.CASE_INSENSITIVE : 0).matcher(text);
+            return Pattern.compile(actualPattern, Pattern.DOTALL | (ignoreCase ? Pattern.CASE_INSENSITIVE : 0)).matcher(text);
         }
     }
 
@@ -1069,7 +1070,7 @@ public class GrepFunctions {
 
         @TruffleBoundary
         private static Matcher getPatternMatcher(String pattern, String text, boolean ignoreCase) {
-            return Pattern.compile(pattern, ignoreCase ? Pattern.CASE_INSENSITIVE : 0).matcher(text);
+            return Pattern.compile(pattern, Pattern.DOTALL | (ignoreCase ? Pattern.CASE_INSENSITIVE : 0)).matcher(text);
         }
     }
 
diff --git a/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_grepl.java b/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_grepl.java
index 1e2e745005..be8aa3cbe9 100644
--- a/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_grepl.java
+++ b/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_grepl.java
@@ -79,5 +79,9 @@ public class TestBuiltin_grepl extends TestBase {
         assertEval("{ .Internal(grepl(7, \"7\", F, F, F, F, F, F)) }");
         assertEval("{ .Internal(grepl(character(), \"7\", F, F, F, F, F, F)) }");
         assertEval("{ .Internal(grepl(\"7\", 7, F, F, F, F, F, F)) }");
+        // the dot does not match the new line in a non-Perl regexp
+        assertEval("{ .Internal(grepl('.+X', 'a\nXb', F, F, F, F, F, F)) }");
+        // the dot matches the new line in a Perl regexp
+        assertEval("{ .Internal(grepl('.+X', 'a\nXb', F, F, T, F, F, F)) }");
     }
 }
diff --git a/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_gsub.java b/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_gsub.java
index 05e8cbecd2..e070cbb031 100644
--- a/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_gsub.java
+++ b/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_gsub.java
@@ -200,6 +200,11 @@ public class TestBuiltin_gsub extends TestBase {
         assertEval("{ gsub(pattern = 'a*', replacement = 'x', x = 'ÄaÄ', perl = TRUE) }");
         assertEval("{ gsub(pattern = 'a*', replacement = 'x', x = 'ÄaaaaÄ', perl = TRUE) }");
 
+        // the dot does not match the new line in a non-Perl regexp
+        assertEval("{ gsub('.*X', '', 'a\nXb', perl = FALSE) }");
+        // the dot matches the new line in a Perl regexp
+        assertEval("{ gsub('.*X', '', 'a\nXb', perl = TRUE) }");
+
         // Expected output: [1] "xaxbx"
         // FastR output: [1] "axxxxxb"
         assertEval(Ignored.ImplementationError, "{ gsub(pattern = 'Ä*', replacement = 'x', x = 'aÄÄÄÄÄb', perl = TRUE) }");
-- 
GitLab