Add char-scanner to libbutl, use in libbpkg and build2

2 files changed, 14 insertions, 129 deletions

diff --git a/build/lexer b/build/lexer
index 1106ed0..5205ae9 100644
--- a/build/lexer
+++ b/build/lexer
@@ -10,6 +10,8 @@
 #include <cstdint> // uint64_t
 #include <exception>
 
+#include <butl/char-scanner>
+
 #include <build/token>
 #include <build/diagnostics>
 
@@ -25,10 +27,11 @@ namespace build
   //
   enum class lexer_mode {normal, value, variable, pairs};
 
-  class lexer
+  class lexer: protected butl::char_scanner
   {
   public:
-    lexer (std::istream& is, const std::string& name): is_ (is), fail (name) {}
+    lexer (std::istream& is, const std::string& name)
+        : char_scanner (is), fail (name) {}
 
     const std::string&
     name () const {return fail.name_;}
@@ -52,30 +55,6 @@ namespace build
     next ();
 
   private:
-    class xchar
-    {
-    public:
-      typedef std::char_traits<char> traits_type;
-      typedef traits_type::int_type int_type;
-      typedef traits_type::char_type char_type;
-
-      xchar (int_type v, std::uint64_t l, std::uint64_t c)
-          : v_ (v), l_ (l), c_ (c) {}
-
-      operator char_type () const {return static_cast<char_type> (v_);}
-
-      int_type
-      value () const {return v_;}
-
-      std::uint64_t line () const {return l_;}
-      std::uint64_t column () const {return c_;}
-
-    private:
-      int_type v_;
-      std::uint64_t l_;
-      std::uint64_t c_;
-    };
-
     token
     name (xchar, bool separated);
 
@@ -89,26 +68,6 @@ namespace build
     xchar
     escape ();
 
-    // Character interface.
-    //
-  private:
-    xchar
-    peek ();
-
-    xchar
-    get ();
-
-    void
-    unget (const xchar&);
-
-    // Tests.
-    //
-    bool
-    is_eos (const xchar& c) const
-    {
-      return c.value () == xchar::traits_type::eof ();
-    }
-
     // Diagnostics.
     //
   private:
@@ -124,16 +83,8 @@ namespace build
     typedef diag_mark<fail_mark_base> fail_mark;
 
   private:
-    std::istream& is_;
     fail_mark fail;
 
-    std::uint64_t l_ {1};
-    std::uint64_t c_ {1};
-
-    bool unget_ {false};
-    xchar buf_ {0, 0, 0};
-
-    bool eos_ {false};
     lexer_mode mode_ {lexer_mode::normal};
     char pair_separator_;
     lexer_mode next_mode_ {lexer_mode::normal}; // Switch to for next token.
diff --git a/build/lexer.cxx b/build/lexer.cxx
index 43c0690..9dce949 100644
--- a/build/lexer.cxx
+++ b/build/lexer.cxx
@@ -20,9 +20,9 @@ namespace build
     bool sep (skip_spaces ());
 
     xchar c (get ());
-    uint64_t ln (c.line ()), cn (c.column ());
+    uint64_t ln (c.line), cn (c.column);
 
-    if (is_eos (c))
+    if (eos (c))
       return token (token_type::eos, sep, ln, cn);
 
     switch (c)
@@ -106,11 +106,11 @@ namespace build
   token lexer::
   name (xchar c, bool sep)
   {
-    uint64_t ln (c.line ()), cn (c.column ());
+    uint64_t ln (c.line), cn (c.column);
     string lexeme;
     lexeme += (c != '\\' ? c : escape ());
 
-    for (c = peek (); !is_eos (c); c = peek ())
+    for (c = peek (); !eos (c); c = peek ())
     {
       bool done (false);
 
@@ -202,9 +202,9 @@ namespace build
     bool r (false);
 
     xchar c (peek ());
-    bool start (c.column () == 1);
+    bool start (c.column == 1);
 
-    for (; !is_eos (c); c = peek ())
+    for (; !eos (c); c = peek ())
     {
       switch (c)
       {
@@ -232,7 +232,7 @@ namespace build
 
           // Read until newline or eos.
           //
-          for (c = peek (); !is_eos (c) && c != '\n'; c = peek ())
+          for (c = peek (); !eos (c) && c != '\n'; c = peek ())
             get ();
 
           r = true;
@@ -266,82 +266,16 @@ namespace build
   {
     xchar c (get ());
 
-    if (is_eos (c))
+    if (eos (c))
       fail (c) << "unterminated escape sequence";
 
     return c;
   }
 
-  lexer::xchar lexer::
-  peek ()
-  {
-    if (unget_)
-      return buf_;
-    else
-    {
-      if (eos_)
-        return xchar (xchar::traits_type::eof (), l_, c_);
-      else
-      {
-        xchar::int_type v (is_.peek ());
-
-        if (v == xchar::traits_type::eof ())
-          eos_ = true;
-
-        return xchar (v, l_, c_);
-      }
-    }
-  }
-
-  lexer::xchar lexer::
-  get ()
-  {
-    if (unget_)
-    {
-      unget_ = false;
-      return buf_;
-    }
-    else
-    {
-      // When is_.get () returns eof, the failbit is also set (stupid,
-      // isn't?) which may trigger an exception. To work around this
-      // we will call peek() first and only call get() if it is not
-      // eof. But we can only call peek() on eof once; any subsequent
-      // calls will spoil the failbit (even more stupid).
-      //
-      xchar c (peek ());
-
-      if (!is_eos (c))
-      {
-        is_.get ();
-
-        if (c == '\n')
-        {
-          l_++;
-          c_ = 1;
-        }
-        else
-          c_++;
-      }
-
-      return c;
-    }
-  }
-
-  void lexer::
-  unget (const xchar& c)
-  {
-    // Because iostream::unget cannot work once eos is reached,
-    // we have to provide our own implementation.
-    //
-    buf_ = c;
-    unget_ = true;
-  }
-
   location_prologue lexer::fail_mark_base::
   operator() (const xchar& c) const
   {
     return build::fail_mark_base<failed>::operator() (
-      location (name_.c_str (), c.line (), c.column ()));
+      location (name_.c_str (), c.line, c.column));
   }
 }