From 53bb3952b87ac4c0accee08291f9319a0fe779be Mon Sep 17 00:00:00 2001
From: Paul Beckingham <paul@beckingham.net>
Date: Sun, 20 Dec 2015 21:19:03 -0500
Subject: [PATCH] Lexer: Added hex number support

---
 src/Lexer.cpp    | 35 +++++++++++++++++++++++++++++++++--
 src/Lexer.h      |  4 +++-
 test/lexer.t.cpp |  8 ++++++--
 3 files changed, 42 insertions(+), 5 deletions(-)

diff --git a/src/Lexer.cpp b/src/Lexer.cpp
index 431ddec5..e5052db2 100644
--- a/src/Lexer.cpp
+++ b/src/Lexer.cpp
@@ -51,8 +51,9 @@ bool Lexer::token (std::string& token, Lexer::Type& type)
   if (isEOS ())
     return false;
 
-  if (isString (token, type, "'\"") ||
-      isWord   (token, type))
+  if (isString    (token, type, "'\"") ||
+      isHexNumber (token, type)        ||
+      isWord      (token, type))
     return true;
 
   return false;
@@ -64,6 +65,7 @@ const std::string Lexer::typeName (const Lexer::Type& type)
 {
   switch (type)
   {
+  case Lexer::Type::hex:          return "hex";
   case Lexer::Type::string:       return "string";
   case Lexer::Type::word:         return "word";
   }
@@ -234,6 +236,34 @@ bool Lexer::isString (std::string& token, Lexer::Type& type, const std::string&
   return false;
 }
 
+////////////////////////////////////////////////////////////////////////////////
+// Lexer::Type::hex
+//   0xX+
+bool Lexer::isHexNumber (std::string& token, Lexer::Type& type)
+{
+  std::size_t marker = _cursor;
+
+  if (_eos - marker >= 3 &&
+      _text[marker + 0] == '0' &&
+      _text[marker + 1] == 'x')
+  {
+    marker += 2;
+
+    while (isHexDigit (_text[marker]))
+      ++marker;
+
+    if (marker - _cursor > 2)
+    {
+      token = _text.substr (_cursor, marker - _cursor);
+      type = Lexer::Type::hex;
+      _cursor = marker;
+      return true;
+    }
+  }
+
+  return false;
+}
+
 ////////////////////////////////////////////////////////////////////////////////
 // Lexer::Type::word
 //   [^\s]+
@@ -262,6 +292,7 @@ bool Lexer::isWord (std::string& token, Lexer::Type& type)
 std::string Lexer::typeToString (Lexer::Type type)
 {
        if (type == Lexer::Type::string)       return std::string ("\033[38;5;7m\033[48;5;3m")    + "string"       + "\033[0m";
+  else if (type == Lexer::Type::hex)          return std::string ("\033[38;5;7m\033[48;5;14m")   + "hex"          + "\033[0m";
   else if (type == Lexer::Type::word)         return std::string ("\033[38;5;15m\033[48;5;236m") + "word"         + "\033[0m";
   else                                        return std::string ("\033[37;41m")                 + "unknown"      + "\033[0m";
 }
diff --git a/src/Lexer.h b/src/Lexer.h
index f3c5b20a..ebd5655c 100644
--- a/src/Lexer.h
+++ b/src/Lexer.h
@@ -35,7 +35,8 @@
 class Lexer
 {
 public:
-  enum class Type { string,
+  enum class Type { hex,
+                    string,
                     word };
 
   Lexer (const std::string&);
@@ -60,6 +61,7 @@ public:
   // Stream Classifiers.
   bool isEOS          () const;
   bool isString       (std::string&, Lexer::Type&, const std::string&);
+  bool isHexNumber    (std::string&, Lexer::Type&);
   bool isWord         (std::string&, Lexer::Type&);
 
 private:
diff --git a/test/lexer.t.cpp b/test/lexer.t.cpp
index f90e6fc6..5f483bb2 100644
--- a/test/lexer.t.cpp
+++ b/test/lexer.t.cpp
@@ -34,7 +34,7 @@
 ////////////////////////////////////////////////////////////////////////////////
 int main (int, char**)
 {
-  UnitTest t (124);
+  UnitTest t (131);
 
   std::vector <std::pair <std::string, Lexer::Type>> tokens;
   std::string token;
@@ -77,7 +77,6 @@ int main (int, char**)
   Lexer l1 ("       \t ");
   t.notok (l1.token (token, type), "'       \\t ' --> no tokens");
 
-
   // static bool readWord (const std::string&, const std::string&, std::string::size_type&, std::string&);
   std::string::size_type cursor = 0;
   std::string word;
@@ -149,6 +148,10 @@ int main (int, char**)
     { "\"\tfoo\t\"",                                  { { "\"\tfoo\t\"",                                  Lexer::Type::string       }, NO, NO, NO, NO }, },
     { "\"\\u20A43\"",                                 { { "\"₤3\"",                                       Lexer::Type::string       }, NO, NO, NO, NO }, },
     { "\"U+20AC4\"",                                  { { "\"€4\"",                                       Lexer::Type::string       }, NO, NO, NO, NO }, },
+
+    // Number
+    { "0x2f",                                         { { "0x2f",                                         Lexer::Type::hex          }, NO, NO, NO, NO }, },
+
   };
   #define NUM_TESTS (sizeof (lexerTests) / sizeof (lexerTests[0]))
 
@@ -183,6 +186,7 @@ int main (int, char**)
     }
   }
 
+  t.is (Lexer::typeName (Lexer::Type::hex),          "hex",          "Lexer::typeName (Lexer::Type::hex)");
   t.is (Lexer::typeName (Lexer::Type::string),       "string",       "Lexer::typeName (Lexer::Type::string)");
   t.is (Lexer::typeName (Lexer::Type::word),         "word",         "Lexer::typeName (Lexer::Type::word)");