Use mode stack in lexer

3 files changed, 44 insertions, 24 deletions

diff --git a/build/lexer b/build/lexer
index 9a0582d..1e253fd 100644
--- a/build/lexer
+++ b/build/lexer
@@ -7,7 +7,9 @@
 
 #include <string>
 #include <iosfwd>
+#include <cstddef> // size_t
 #include <cstdint> // uint64_t
+#include <cassert>
 #include <exception>
 
 #include <butl/char-scanner>
@@ -27,13 +29,13 @@ namespace build
   // pairs modes are automatically reset after the end of the line.
   // The variable mode is automatically reset after the name token.
   //
-  enum class lexer_mode {normal, value, variable, pairs};
+  enum class lexer_mode {normal, variable, value, pairs};
 
   class lexer: protected butl::char_scanner
   {
   public:
     lexer (std::istream& is, const std::string& name)
-        : char_scanner (is), fail (name) {}
+        : char_scanner (is), fail (name) {mode_.push (lexer_mode::normal);}
 
     const std::string&
     name () const {return fail.name_;}
@@ -44,12 +46,17 @@ namespace build
     void
     mode (lexer_mode m, char pair_separator = '=')
     {
-      next_mode_ = m;
+      mode_.push (m);
       pair_separator_ = pair_separator;
     }
 
+    // Expire the current mode early.
+    //
+    void
+    expire_mode () {mode_.pop ();}
+
     lexer_mode
-    mode () const {return mode_;}
+    mode () const {return mode_.top ();}
 
     char
     pair_separator () const {return pair_separator_;}
@@ -93,10 +100,23 @@ namespace build
   private:
     fail_mark fail;
 
-    lexer_mode mode_ {lexer_mode::normal};
+    // Currently, the maximum mode nesting is 3: {normal, value, variable}.
+    //
+    struct mode_stack
+    {
+      static const size_t max_size = 3;
+
+      void push (lexer_mode m) {assert (n_ != max_size); d_[n_++] = m;}
+      void pop () {assert (n_ != 0); n_--;}
+      lexer_mode top () const {return d_[n_ - 1];}
+
+    private:
+      size_t n_ = 0;
+      lexer_mode d_[max_size];
+    };
+
+    mode_stack mode_;
     char pair_separator_;
-    lexer_mode next_mode_ {lexer_mode::normal}; // Switch to for next token.
-    lexer_mode prev_mode_; // Return to after current mode expires.
   };
 }
 
diff --git a/build/lexer.cxx b/build/lexer.cxx
index 4151087..f4733be 100644
--- a/build/lexer.cxx
+++ b/build/lexer.cxx
@@ -11,12 +11,6 @@ namespace build
   token lexer::
   next ()
   {
-    if (mode_ != next_mode_)
-    {
-      prev_mode_ = mode_;
-      mode_ = next_mode_;
-    }
-
     bool sep (skip_spaces ());
 
     xchar c (get ());
@@ -25,16 +19,18 @@ namespace build
     if (eos (c))
       return token (token_type::eos, sep, ln, cn);
 
+    lexer_mode m (mode_.top ());
+
     switch (c)
     {
       // NOTE: remember to update name() if adding new punctuations.
       //
     case '\n':
       {
-        // Restore the normal mode at the end of the line.
+        // Expire value/pairs mode at the end of the line.
         //
-        if (mode_ == lexer_mode::value || mode_ == lexer_mode::pairs)
-          mode_ = next_mode_ = lexer_mode::normal;
+        if (m == lexer_mode::value || m == lexer_mode::pairs)
+          mode_.pop ();
 
         return token (token_type::newline, sep, ln, cn);
       }
@@ -62,13 +58,13 @@ namespace build
 
     // Handle pair separator.
     //
-    if (mode_ == lexer_mode::pairs && c == pair_separator_)
+    if (m == lexer_mode::pairs && c == pair_separator_)
       return token (token_type::pair_separator, sep, ln, cn);
 
     // The following characters are not treated as special in the
     // value or pairs mode.
     //
-    if (mode_ != lexer_mode::value && mode_ != lexer_mode::pairs)
+    if (m != lexer_mode::value && m != lexer_mode::pairs)
     {
       // NOTE: remember to update name() if adding new punctuations.
       //
@@ -107,19 +103,21 @@ namespace build
     uint64_t ln (c.line), cn (c.column);
     string lexeme;
 
+    lexer_mode m (mode_.top ());
+
     for (; !eos (c); c = peek ())
     {
       bool done (false);
 
       // Handle pair separator.
       //
-      if (mode_ == lexer_mode::pairs && c == pair_separator_)
+      if (m == lexer_mode::pairs && c == pair_separator_)
         break;
 
       // The following characters are not treated as special in the
       // value or pairs mode.
       //
-      if (mode_ != lexer_mode::value && mode_ != lexer_mode::pairs)
+      if (m != lexer_mode::value && m != lexer_mode::pairs)
       {
         switch (c)
         {
@@ -139,7 +137,7 @@ namespace build
       // While these extra characters are treated as the name end in
       // the variable mode.
       //
-      if (mode_ == lexer_mode::variable)
+      if (m == lexer_mode::variable)
       {
         switch (c)
         {
@@ -198,8 +196,10 @@ namespace build
     //
     assert (c.line != ln || c.column != cn);
 
-    if (mode_ == lexer_mode::variable)
-      next_mode_ = prev_mode_;
+    // Expire variable mode at the end of the name.
+    //
+    if (m == lexer_mode::variable)
+      mode_.pop ();
 
     return token (lexeme, sep, ln, cn);
   }
diff --git a/build/parser.cxx b/build/parser.cxx
index f140bc7..2daf1ce 100644
--- a/build/parser.cxx
+++ b/build/parser.cxx
@@ -1024,7 +1024,7 @@ namespace build
         bool paren (tt == type::lparen);
         if (paren)
         {
-          lexer_->mode (lexer_mode::normal);
+          lexer_->expire_mode ();
           next (t, tt);
         }