diff --git a/init/Android.mk b/init/Android.mk
index 6ef8dff..fc3f133 100644
--- a/init/Android.mk
+++ b/init/Android.mk
@@ -20,6 +20,27 @@
 
 # --
 
+# If building on Linux, then build unit test for the host.
+ifeq ($(HOST_OS),linux)
+include $(CLEAR_VARS)
+LOCAL_CPPFLAGS := $(init_cflags)
+LOCAL_SRC_FILES:= \
+    parser/tokenizer.cpp \
+
+LOCAL_MODULE := libinit_parser
+LOCAL_CLANG := true
+include $(BUILD_HOST_STATIC_LIBRARY)
+
+include $(CLEAR_VARS)
+LOCAL_MODULE := init_parser_tests
+LOCAL_SRC_FILES := \
+    parser/tokenizer_test.cpp \
+
+LOCAL_STATIC_LIBRARIES := libinit_parser
+LOCAL_CLANG := true
+include $(BUILD_HOST_NATIVE_TEST)
+endif
+
 include $(CLEAR_VARS)
 LOCAL_CPPFLAGS := $(init_cflags)
 LOCAL_SRC_FILES:= \
diff --git a/init/parser/tokenizer.cpp b/init/parser/tokenizer.cpp
new file mode 100644
index 0000000..340e0d9
--- /dev/null
+++ b/init/parser/tokenizer.cpp
@@ -0,0 +1,129 @@
+// Copyright (C) 2015 The Android Open Source Project
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//      http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include "tokenizer.h"
+
+namespace init {
+
+Tokenizer::Tokenizer(const std::string& data)
+    : data_(data), eof_(false), pos_(0), tok_start_(0) {
+  current_.type = TOK_START;
+
+  if (data.size() > 0) {
+    cur_char_ = data[0];
+  } else {
+    eof_ = true;
+    cur_char_ = '\0';
+  }
+}
+
+Tokenizer::~Tokenizer() {}
+
+const Tokenizer::Token& Tokenizer::current() {
+  return current_;
+}
+
+bool Tokenizer::Next() {
+  while (!eof_) {
+    AdvWhiteSpace();
+
+    // Check for comments.
+    if (cur_char_ == '#') {
+      AdvChar();
+      // Skip rest of line
+      while (!eof_ && cur_char_ != '\n') {
+        AdvChar();
+      }
+    }
+
+    if (eof_) {
+      break;
+    }
+
+    if (cur_char_ == '\0') {
+      AdvChar();
+    } else if (cur_char_ == '\n') {
+      current_.type = TOK_NEWLINE;
+      current_.text.clear();
+      AdvChar();
+      return true;
+    } else if (cur_char_ == '\\') {
+      AdvChar();  // skip backslash
+      // This is line continuation so
+      // do not generated TOK_NEWLINE at
+      // the next \n.
+      AdvUntil('\n');
+      AdvChar();  // skip \n
+    } else if (cur_char_ == '\"') {
+      AdvChar();
+      StartText();
+      // Grab everything until the next quote.
+      AdvUntil('\"');
+      EndText();
+      AdvChar();  // skip quote.
+      return true;
+    } else {
+      StartText();
+      AdvText();
+      EndText();
+      return true;
+    }
+  }
+  current_.type = TOK_END;
+  current_.text.clear();
+  return false;
+}
+
+void Tokenizer::AdvChar() {
+  pos_++;
+  if (pos_ < data_.size()) {
+    cur_char_ = data_[pos_];
+  } else {
+    eof_ = true;
+    cur_char_ = '\0';
+  }
+}
+
+void Tokenizer::AdvWhiteSpace() {
+  while (cur_char_ == '\t' || cur_char_ == '\r' || cur_char_ == ' ') {
+    AdvChar();
+  }
+}
+
+void Tokenizer::AdvUntil(char x) {
+  while (!eof_ && cur_char_ != x) {
+    AdvChar();
+  }
+}
+
+void Tokenizer::AdvText() {
+  while (cur_char_ != '\t' && cur_char_ != '\r' && cur_char_ != '\0' &&
+         cur_char_ != ' ' && cur_char_ != '\n' && cur_char_ != '#') {
+    AdvChar();
+  }
+}
+
+void Tokenizer::StartText() {
+  current_.text.clear();
+  tok_start_ = pos_;
+  current_.type = TOK_TEXT;
+}
+
+void Tokenizer::EndText() {
+  if (pos_ != tok_start_) {
+    current_.text.append(data_, tok_start_, pos_ - tok_start_);
+  }
+}
+
+}  // namespace init
\ No newline at end of file
diff --git a/init/parser/tokenizer.h b/init/parser/tokenizer.h
new file mode 100644
index 0000000..40a22b1
--- /dev/null
+++ b/init/parser/tokenizer.h
@@ -0,0 +1,69 @@
+// Copyright (C) 2015 The Android Open Source Project
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//      http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include <string>
+
+namespace init {
+
+// Used to tokenize a std::string.
+// Call Next() to advance through each token until it returns false,
+// indicating there are no more tokens left in the string.
+// The current token can be accessed with current(), which returns
+// a Token.
+// Supported tokens are:
+// TOK_START - Next() has yet to be called
+// TOK_END - At the end of string
+// TOK_NEWLINE - The end of a line denoted by \n.
+// TOK_TEXT - A word.
+// Comments are denoted with '#' and the tokenizer will ignore
+// the rest of the line.
+// Double quotes can be used to insert whitespace into words.
+// A backslash at the end of a line denotes continuation and
+// a TOK_NEWLINE will not be generated for that line.
+class Tokenizer {
+ public:
+  Tokenizer(const std::string& data);
+  ~Tokenizer();
+
+  enum TokenType { TOK_START, TOK_END, TOK_NEWLINE, TOK_TEXT };
+  struct Token {
+    TokenType type;
+    std::string text;
+  };
+
+  // Returns the curret token.
+  const Token& current();
+
+  // Move to the next token, returns false at the end of input.
+  bool Next();
+
+ private:
+  void GetData();
+  void AdvChar();
+  void AdvText();
+  void AdvUntil(char x);
+  void AdvWhiteSpace();
+  void StartText();
+  void EndText();
+
+  const std::string& data_;
+  Token current_;
+
+  bool eof_;
+  size_t pos_;
+  char cur_char_;
+  size_t tok_start_;
+};
+
+}  // namespace init
diff --git a/init/parser/tokenizer_test.cpp b/init/parser/tokenizer_test.cpp
new file mode 100644
index 0000000..c4a48df
--- /dev/null
+++ b/init/parser/tokenizer_test.cpp
@@ -0,0 +1,230 @@
+// Copyright (C) 2015 The Android Open Source Project
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//      http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include "tokenizer.h"
+
+#include <errno.h>
+#include <gtest/gtest.h>
+
+#include <string>
+
+namespace init {
+
+#define SETUP_TEST(test_data)  \
+  std::string data(test_data); \
+  Tokenizer tokenizer(data);   \
+  ASSERT_EQ(Tokenizer::TOK_START, tokenizer.current().type)
+
+#define ASSERT_TEXT_TOKEN(test_text)              \
+  ASSERT_TRUE(tokenizer.Next());                  \
+  ASSERT_EQ(test_text, tokenizer.current().text); \
+  ASSERT_EQ(Tokenizer::TOK_TEXT, tokenizer.current().type)
+
+#define ASSERT_NEWLINE_TOKEN()   \
+  ASSERT_TRUE(tokenizer.Next()); \
+  ASSERT_EQ(Tokenizer::TOK_NEWLINE, tokenizer.current().type)
+
+TEST(Tokenizer, Empty) {
+  SETUP_TEST("");
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, Simple) {
+  SETUP_TEST("test");
+  ASSERT_TEXT_TOKEN("test");
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, LeadingWhiteSpace) {
+  SETUP_TEST(" \t  \r  test");
+  ASSERT_TEXT_TOKEN("test");
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, TrailingWhiteSpace) {
+  SETUP_TEST("test \t  \r  ");
+  ASSERT_TEXT_TOKEN("test");
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, WhiteSpace) {
+  SETUP_TEST(" \t  \r  test \t  \r  ");
+  ASSERT_TEXT_TOKEN("test");
+
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, TwoTokens) {
+  SETUP_TEST("  foo   bar ");
+  ASSERT_TEXT_TOKEN("foo");
+  ASSERT_TEXT_TOKEN("bar");
+
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, MultiToken) {
+  SETUP_TEST("one two three four five");
+  ASSERT_TEXT_TOKEN("one");
+  ASSERT_TEXT_TOKEN("two");
+  ASSERT_TEXT_TOKEN("three");
+  ASSERT_TEXT_TOKEN("four");
+  ASSERT_TEXT_TOKEN("five");
+
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, NewLine) {
+  SETUP_TEST("\n");
+  ASSERT_NEWLINE_TOKEN();
+
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, TextNewLine) {
+  SETUP_TEST("test\n");
+  ASSERT_TEXT_TOKEN("test");
+  ASSERT_NEWLINE_TOKEN();
+
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, MultiTextNewLine) {
+  SETUP_TEST("one\ntwo\nthree\n");
+  ASSERT_TEXT_TOKEN("one");
+  ASSERT_NEWLINE_TOKEN();
+  ASSERT_TEXT_TOKEN("two");
+  ASSERT_NEWLINE_TOKEN();
+  ASSERT_TEXT_TOKEN("three");
+  ASSERT_NEWLINE_TOKEN();
+
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, MultiTextNewLineNoLineEnding) {
+  SETUP_TEST("one\ntwo\nthree");
+  ASSERT_TEXT_TOKEN("one");
+  ASSERT_NEWLINE_TOKEN();
+  ASSERT_TEXT_TOKEN("two");
+  ASSERT_NEWLINE_TOKEN();
+  ASSERT_TEXT_TOKEN("three");
+
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, Comment) {
+  SETUP_TEST("#test");
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, CommentWhiteSpace) {
+  SETUP_TEST(" \t  \r  #test \t  \r  ");
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, CommentNewLine) {
+  SETUP_TEST(" #test   \n");
+  ASSERT_NEWLINE_TOKEN();
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, CommentTwoNewLine) {
+  SETUP_TEST(" #test   \n#test");
+  ASSERT_NEWLINE_TOKEN();
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, CommentWithText) {
+  SETUP_TEST("foo bar #test");
+  ASSERT_TEXT_TOKEN("foo");
+  ASSERT_TEXT_TOKEN("bar");
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, CommentWithTextNoSpace) {
+  SETUP_TEST("foo bar#test");
+  ASSERT_TEXT_TOKEN("foo");
+  ASSERT_TEXT_TOKEN("bar");
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, CommentWithTextLineFeed) {
+  SETUP_TEST("foo bar #test\n");
+  ASSERT_TEXT_TOKEN("foo");
+  ASSERT_TEXT_TOKEN("bar");
+  ASSERT_NEWLINE_TOKEN();
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, CommentWithMultiTextLineFeed) {
+  SETUP_TEST("#blah\nfoo bar #test\n#blah");
+  ASSERT_NEWLINE_TOKEN();
+  ASSERT_TEXT_TOKEN("foo");
+  ASSERT_TEXT_TOKEN("bar");
+  ASSERT_NEWLINE_TOKEN();
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, SimpleEscaped) {
+  SETUP_TEST("fo\\no bar");
+  ASSERT_TEXT_TOKEN("fo\\no");
+  ASSERT_TEXT_TOKEN("bar");
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, EscapedLineContNoLineFeed) {
+  SETUP_TEST("fo\\no bar \\ hello");
+  ASSERT_TEXT_TOKEN("fo\\no");
+  ASSERT_TEXT_TOKEN("bar");
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, EscapedLineContLineFeed) {
+  SETUP_TEST("fo\\no bar \\ hello\n");
+  ASSERT_TEXT_TOKEN("fo\\no");
+  ASSERT_TEXT_TOKEN("bar");
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, EscapedLineCont) {
+  SETUP_TEST("fo\\no bar \\\ntest");
+  ASSERT_TEXT_TOKEN("fo\\no");
+  ASSERT_TEXT_TOKEN("bar");
+  ASSERT_TEXT_TOKEN("test");
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, EscapedLineContWithBadChars) {
+  SETUP_TEST("fo\\no bar \\bad bad bad\ntest");
+  ASSERT_TEXT_TOKEN("fo\\no");
+  ASSERT_TEXT_TOKEN("bar");
+  ASSERT_TEXT_TOKEN("test");
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, SimpleQuotes) {
+  SETUP_TEST("foo \"single token\" bar");
+  ASSERT_TEXT_TOKEN("foo");
+  ASSERT_TEXT_TOKEN("single token");
+  ASSERT_TEXT_TOKEN("bar");
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+TEST(Tokenizer, BadQuotes) {
+  SETUP_TEST("foo \"single token");
+  ASSERT_TEXT_TOKEN("foo");
+  ASSERT_TEXT_TOKEN("single token");
+  ASSERT_FALSE(tokenizer.Next());
+}
+
+}  // namespace init