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

3 files changed, 161 insertions, 1 deletions

diff --git a/butl/buildfile b/butl/buildfile
index e9fc900..a1b5a32 100644
--- a/butl/buildfile
+++ b/butl/buildfile
@@ -2,6 +2,6 @@
 # copyright : Copyright (c) 2014-2015 Code Synthesis Ltd
 # license   : MIT; see accompanying LICENSE file
 
-lib{butl}: cxx{fdstream filesystem path process timestamp}
+lib{butl}: cxx{char-scanner fdstream filesystem path process timestamp}
 cxx.poptions += -I$src_root
 lib{butl}: cxx.export.poptions = -I$src_root
diff --git a/butl/char-scanner b/butl/char-scanner
new file mode 100644
index 0000000..3c8cdbe
--- /dev/null
+++ b/butl/char-scanner
@@ -0,0 +1,82 @@
+// file      : butl/char-scanner -*- C++ -*-
+// copyright : Copyright (c) 2014-2015 Code Synthesis Ltd
+// license   : MIT; see accompanying LICENSE file
+
+#ifndef BUTL_CHAR_SCANNER
+#define BUTL_CHAR_SCANNER
+
+#include <string>  // char_traits
+#include <iosfwd>
+#include <cstdint> // uint64_t
+
+namespace butl
+{
+  // Low-level character stream scanner. Normally used as a base for
+  // higher-level lexers.
+  //
+  class char_scanner
+  {
+  public:
+    char_scanner (std::istream& is): is_ (is) {}
+
+    char_scanner (const char_scanner&) = delete;
+    char_scanner& operator= (const char_scanner&) = delete;
+
+    // Scanner interface.
+    //
+  public:
+
+    // Extended character. It includes line/column information
+    // and is capable of representing EOF.
+    //
+    class xchar
+    {
+    public:
+      typedef std::char_traits<char> traits_type;
+      typedef traits_type::int_type int_type;
+      typedef traits_type::char_type char_type;
+
+      int_type value;
+      std::uint64_t line;
+      std::uint64_t column;
+
+      operator char_type () const {return static_cast<char_type> (value);}
+
+      xchar (int_type v, std::uint64_t l = 0, std::uint64_t c = 0)
+          : value (v), line (l), column (c) {}
+    };
+
+    xchar
+    get ();
+
+    void
+    unget (const xchar&);
+
+    // Note that if there is an "ungot" character, peek() will return
+    // that.
+    //
+    xchar
+    peek ();
+
+    // Tests. In the future we can add tests line alpha(), alnum(),
+    // etc.
+    //
+    static bool
+    eos (const xchar& c) {return c.value == xchar::traits_type::eof ();}
+
+    // Line and column of the furthest seen (either via get() or
+    // peek()) character.
+    //
+    std::uint64_t line {1};
+    std::uint64_t column {1};
+
+  protected:
+    std::istream& is_;
+
+    bool unget_ {false};
+    xchar buf_ = '\0';
+    bool eos_ {false};
+  };
+}
+
+#endif // BUTL_CHAR_SCANNER
diff --git a/butl/char-scanner.cxx b/butl/char-scanner.cxx
new file mode 100644
index 0000000..1561ad1
--- /dev/null
+++ b/butl/char-scanner.cxx
@@ -0,0 +1,78 @@
+// file      : butl/char-scanner.cxx -*- C++ -*-
+// copyright : Copyright (c) 2014-2015 Code Synthesis Ltd
+// license   : MIT; see accompanying LICENSE file
+
+#include <butl/char-scanner>
+
+#include <istream>
+
+using namespace std;
+
+namespace butl
+{
+  auto char_scanner::
+  peek () -> xchar
+  {
+    if (unget_)
+      return buf_;
+    else
+    {
+      if (eos_)
+        return xchar (xchar::traits_type::eof (), line, column);
+      else
+      {
+        xchar::int_type v (is_.peek ());
+
+        if (v == xchar::traits_type::eof ())
+          eos_ = true;
+
+        return xchar (v, line, column);
+      }
+    }
+  }
+
+  auto char_scanner::
+  get () -> xchar
+  {
+    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 (!eos (c))
+      {
+        is_.get ();
+
+        if (c == '\n')
+        {
+          line++;
+          column = 1;
+        }
+        else
+          column++;
+      }
+
+      return c;
+    }
+  }
+
+  void char_scanner::
+  unget (const xchar& c)
+  {
+    // Because iostream::unget cannot work once eos is reached,
+    // we have to provide our own implementation.
+    //
+    buf_ = c;
+    unget_ = true;
+  }
+}