From aa77eb4550d29158c74d479f873d672ba156aa3b Mon Sep 17 00:00:00 2001
From: artemp <artem@mapnik.org>
Date: Tue, 9 Aug 2016 16:13:12 +0100
Subject: [PATCH] Expressions - add more regex tests to cover #3483

---
 src/expression_node.cpp             |  4 ++--
 test/unit/core/expressions_test.cpp | 17 ++++++++++++++---
 2 files changed, 16 insertions(+), 5 deletions(-)

diff --git a/src/expression_node.cpp b/src/expression_node.cpp
index 380c36a4e..549f4c891 100644
--- a/src/expression_node.cpp
+++ b/src/expression_node.cpp
@@ -130,9 +130,9 @@ value regex_replace_node::apply(value const& v) const
     auto const& pattern = impl_.get()->pattern_;
     auto const& format = impl_.get()->format_;
 #if defined(BOOST_REGEX_HAS_ICU)
-    return boost::u32regex_replace(v.to_unicode(),pattern,format);
+    return boost::u32regex_replace(v.to_unicode(), pattern, format);
 #else
-    std::string repl = boost::regex_replace(v.to_string(),pattern,format);
+    std::string repl = boost::regex_replace(v.to_string(), pattern, format);
     transcoder tr_("utf8");
     return tr_.transcode(repl.c_str());
 #endif
diff --git a/test/unit/core/expressions_test.cpp b/test/unit/core/expressions_test.cpp
index e60d6127b..9f6a919f3 100644
--- a/test/unit/core/expressions_test.cpp
+++ b/test/unit/core/expressions_test.cpp
@@ -1,4 +1,3 @@
-
 #include "catch_ext.hpp"
 
 #include <mapnik/expression.hpp>
@@ -176,6 +175,18 @@ TEST_CASE("expressions")
     // regex
     // replace
     TRY_CHECK(eval(" [foo].replace('(\\B)|( )','$1 ') ") == tr.transcode("b a r"));
+
+    // https://en.wikipedia.org/wiki/Chess_symbols_in_Unicode
+    //'\u265C\u265E\u265D\u265B\u265A\u265D\u265E\u265C' - black chess figures
+    // replace black knights with white knights
+    auto val0 = eval(u8"'\u265C\u265E\u265D\u265B\u265A\u265D\u265E\u265C'.replace('\u265E','\u2658')");
+    auto val1 = eval(u8"'♜♞♝♛♚♝♞♜'.replace('♞','♘')"); // ==> expected ♜♘♝♛♚♝♘♜
+    TRY_CHECK(val0 == val1);
+    TRY_CHECK(val0.to_string() == val1.to_string());
+
+    // following test will fail if boost_regex is built without ICU support (unpaired surrogates in output)
+    TRY_CHECK(eval("[name].replace('(\\B)|( )',' ') ") == tr.transcode(u8"Q u é b e c"));
+    TRY_CHECK(eval("'Москва'.replace('(?<!^)(\\B|b)(?!$)',' ')") == tr.transcode(u8"М о с к в а"));
     // 'foo' =~ s:(\w)\1:$1x:r
     TRY_CHECK(eval(" 'foo'.replace('(\\w)\\1', '$1x') ") == tr.transcode("fox"));
     TRY_CHECK(parse_and_dump(" 'foo'.replace('(\\w)\\1', '$1x') ") == "'foo'.replace('(\\w)\\1','$1x')");
@@ -187,8 +198,8 @@ TEST_CASE("expressions")
     TRY_CHECK(parse_and_dump(" [name].match('^Q\\S*$') ") == "[name].match('^Q\\S*$')");
 
     // string & value concatenation
-    // this should evaluate as two strings concatenating, but currently fails
+    // this should evaluate as two strings concatenating
     TRY_CHECK(eval("Hello + '!'") == eval("'Hello!'"));
-    // this should evaulate as a combination of an int value and string, but fails
+    // this should evaulate as a combination of an int value and string
     TRY_CHECK(eval("[int]+m") == eval("'123m'"));
 }