Add support for group/backreferences regex replacement in mapping files.

This is useful to generically remap files from one directory to another.
The following mapping would for instance map all files under the "foo/"
directory to "third_party/foo/", for instance, mapping "foo/bar.h" to
"third_party/foo/bar.h".

[ { include: ['@"(foo/.*)"', private, '"third_party/\1"', public ] } ]

This is useful for large project where different third_party libraries
are compiled together. In Chromium for instance, include paths would
look like:
 -Ithird_party -Ithird_party/v8/include

This allows code in V8 to include its own headers directly. Other
third_parties would include them via "third_party/v8/include/...".
Because files are accessible from two different paths, IWYU can't know
which should be used. Using a mapping file, adding or removing the
"third_party/v8/include" prefix where needed, resolves this problem.

iwyu_include_picker.cc

@@ -1436,10 +1436,14 @@ void IncludePicker::ExpandRegexes() {
   for (const auto& incmap : quoted_includes_to_quoted_includers_) {
     const string& hdr = incmap.first;
     for (const string& regex_key : filepath_include_map_regex_keys) {
+      const string regex = regex_key.substr(1);
       const vector<MappedInclude>& map_to = filepath_include_map_[regex_key];
-      if (RegexMatch(regex_dialect, hdr, regex_key.substr(1)) &&
+      if (RegexMatch(regex_dialect, hdr, regex) &&
           !ContainsQuotedInclude(map_to, hdr)) {
-        Extend(&filepath_include_map_[hdr], filepath_include_map_[regex_key]);
+        for (const MappedInclude& target : map_to) {
+          filepath_include_map_[hdr].push_back(MappedInclude(
+              RegexReplace(regex_dialect, hdr, regex, target.quoted_include)));
+        }
         MarkVisibility(&include_visibility_map_, hdr,
                        include_visibility_map_[regex_key]);
       }

iwyu_regex.cc

@@ -46,4 +46,26 @@ bool RegexMatch(RegexDialect dialect, const std::string& str,
   CHECK_UNREACHABLE_("Unexpected regex dialect");
 }
 
+std::string RegexReplace(RegexDialect dialect, const std::string& str,
+                         const std::string& pattern,
+                         const std::string& replacement) {
+  switch (dialect) {
+    case RegexDialect::LLVM: {
+      // llvm::Regex::sub  has search semantics. Enclose the pattern in ^...$
+      // for start/end anchoring.
+      llvm::Regex r("^" + pattern + "$");
+      return r.sub(replacement, str);
+    }
+
+    case RegexDialect::ECMAScript: {
+      // std::regex_replace has search semantics. Enclose the pattern in ^...$
+      // for start/end anchoring.
+      std::regex r("^" + pattern + "$", std::regex_constants::ECMAScript);
+      return std::regex_replace(str, r, replacement,
+                                std::regex_constants::format_first_only);
+    }
+  }
+  CHECK_UNREACHABLE_("Unexpected regex dialect");
+}
+
 }  // namespace include_what_you_use

iwyu_regex.h

@@ -23,6 +23,12 @@ bool ParseRegexDialect(const char* str, RegexDialect* dialect);
 bool RegexMatch(RegexDialect dialect, const std::string& str,
                 const std::string& pattern);
 
+// Returns input string with the first match of pattern replaced, for the given
+// regex dialect.
+std::string RegexReplace(RegexDialect dialect, const std::string& str,
+                         const std::string& pattern,
+                         const std::string& replacement);
+
 }  // namespace include_what_you_use
 
 #endif  // INCLUDE_WHAT_YOU_USE_IWYU_REGEX_H_

tests/cxx/mapping_replace_regex_ecmascript.cc

@@ -0,0 +1,38 @@
+//===--- mapping_replace_regex_ecmascript.cc - test input file for iwyu ---===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// IWYU_ARGS: -I . \
+//            -Xiwyu --regex=ecmascript \
+//            -Xiwyu --mapping_file=tests/cxx/mapping_replace_regex_ecmascript.imp
+
+// Generically map include paths to a different include directory:
+// * The include of tests/cxx/direct.h should nominally be replaced by
+//   tests/cxx/indirect.h, where IndirectClass is defined
+// * But we provide a mapping adding a "foobar" prefix to any include under
+//   "tests/cxx", resulting in suggesting foobar/tests/cxx/indirect.h.
+
+#include "tests/cxx/direct.h"
+
+void f() {
+  // IWYU: IndirectClass is defined in "foobar/tests/cxx/indirect.h"
+  IndirectClass i;
+}
+
+/**** IWYU_SUMMARY
+
+tests/cxx/mapping_replace_regex_ecmascript.cc should add these lines:
+#include "foobar/tests/cxx/indirect.h"
+
+tests/cxx/mapping_replace_regex_ecmascript.cc should remove these lines:
+- #include "tests/cxx/direct.h"  // lines XX-XX
+
+The full include-list for tests/cxx/mapping_replace_regex_ecmascript.cc:
+#include "foobar/tests/cxx/indirect.h"  // for IndirectClass
+
+***** IWYU_SUMMARY */

tests/cxx/mapping_replace_regex_ecmascript.imp

@@ -0,0 +1,5 @@
+# Maps headers to a different include path.
+# The foobar directory does not exist; we just want IWYU to suggest using it.
+[
+  { include: ['@"(tests/cxx/.*)"', private, '"foobar/$1"', public ] },
+]

tests/cxx/mapping_replace_regex_llvm.cc

@@ -0,0 +1,38 @@
+//===--- mapping_replace_regex_llvm.cc - test input file for iwyu ---------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// IWYU_ARGS: -I . \
+//            -Xiwyu --regex=llvm \
+//            -Xiwyu --mapping_file=tests/cxx/mapping_replace_regex_llvm.imp
+
+// Generically map include paths to a different include directory:
+// * The include of tests/cxx/direct.h should nominally be replaced by
+//   tests/cxx/indirect.h, where IndirectClass is defined
+// * But we provide a mapping adding a "foobar" prefix to any include under
+//   "tests/cxx", resulting in suggesting foobar/tests/cxx/indirect.h.
+
+#include "tests/cxx/direct.h"
+
+void f() {
+  // IWYU: IndirectClass is defined in "foobar/tests/cxx/indirect.h"
+  IndirectClass i;
+}
+
+/**** IWYU_SUMMARY
+
+tests/cxx/mapping_replace_regex_llvm.cc should add these lines:
+#include "foobar/tests/cxx/indirect.h"
+
+tests/cxx/mapping_replace_regex_llvm.cc should remove these lines:
+- #include "tests/cxx/direct.h"  // lines XX-XX
+
+The full include-list for tests/cxx/mapping_replace_regex_llvm.cc:
+#include "foobar/tests/cxx/indirect.h"  // for IndirectClass
+
+***** IWYU_SUMMARY */

tests/cxx/mapping_replace_regex_llvm.imp

@@ -0,0 +1,5 @@
+# Maps headers to a different include path.
+# The foobar directory does not exist; we just want IWYU to suggest using it.
+[
+  { include: ['@"(tests/cxx/.*)"', private, '"foobar/\1"', public ] },
+]