From ea66709a853255c7957a8a7907fd21fa7f6cfd3f Mon Sep 17 00:00:00 2001
From: Boris Kolpackov <boris@codesynthesis.com>
Date: Wed, 9 Sep 2015 15:56:54 +0200
Subject: Add support for quoting directive names

Now only unquoted, literal names are recognized as directives, for
example:

'print' = abc
print $print
---
 build/b.cxx               | 10 +++++-----
 build/file.cxx            |  6 +++---
 build/lexer.cxx           |  7 ++++++-
 build/parser.cxx          | 27 ++++++++++++++-------------
 build/token               | 39 +++++++++++++++------------------------
 build/token.cxx           |  4 ++--
 tests/directive/buildfile |  8 ++++++++
 tests/directive/test.out  |  1 +
 tests/directive/test.sh   |  3 +++
 9 files changed, 57 insertions(+), 48 deletions(-)
 create mode 100644 tests/directive/buildfile
 create mode 100644 tests/directive/test.out
 create mode 100755 tests/directive/test.sh

diff --git a/build/b.cxx b/build/b.cxx
index bbfc635..665fac8 100644
--- a/build/b.cxx
+++ b/build/b.cxx
@@ -146,24 +146,24 @@ main (int argc, char* argv[])
       lexer l (is, "<cmdline>");
       token t (l.next ());
 
-      if (t.type () == token_type::eos)
+      if (t.type == token_type::eos)
         continue; // Whitespace-only argument.
 
       // Unless this is a name followed by = or +=, assume it is
       // a start of the buildspec.
       //
-      if (t.type () != token_type::name)
+      if (t.type != token_type::name)
         break;
 
-      token_type tt (l.next ().type ());
+      token_type tt (l.next ().type);
 
       if (tt != token_type::equal && tt != token_type::plus_equal)
         break;
 
       parser p;
-      t = p.parse_variable (l, *global_scope, t.name (), tt);
+      t = p.parse_variable (l, *global_scope, t.value, tt);
 
-      if (t.type () != token_type::eos)
+      if (t.type != token_type::eos)
         fail << "unexpected " << t << " in variable " << s;
     }
 
diff --git a/build/file.cxx b/build/file.cxx
index f8f62f9..f6fdd10 100644
--- a/build/file.cxx
+++ b/build/file.cxx
@@ -274,8 +274,8 @@ namespace build
       token t (lex.next ());
       token_type tt;
 
-      if (t.type () != token_type::name || t.name () != var ||
-          ((tt = lex.next ().type ()) != token_type::equal &&
+      if (t.type != token_type::name || t.value != var ||
+          ((tt = lex.next ().type) != token_type::equal &&
            tt != token_type::plus_equal))
       {
         error << "variable '" << var << "' expected as first line in " << rbf;
@@ -284,7 +284,7 @@ namespace build
 
       parser p;
       temp_scope tmp (*global_scope);
-      p.parse_variable (lex, tmp, t.name (), tt);
+      p.parse_variable (lex, tmp, t.value, tt);
 
       auto l (tmp.vars[var]);
       assert (l.defined ());
diff --git a/build/lexer.cxx b/build/lexer.cxx
index 133375b..e083b4a 100644
--- a/build/lexer.cxx
+++ b/build/lexer.cxx
@@ -152,6 +152,7 @@ namespace build
     string lexeme;
 
     lexer_mode m (mode_.top ());
+    bool quoted (m == lexer_mode::quoted);
 
     for (; !eos (c); c = peek ())
     {
@@ -247,6 +248,7 @@ namespace build
               if (eos (c))
                 fail (c) << "unterminated single-quoted sequence";
 
+              quoted = true;
               continue;
             }
           }
@@ -281,7 +283,10 @@ namespace build
             if (m == lexer_mode::quoted)
               mode_.pop ();
             else
+            {
               mode_.push (lexer_mode::quoted);
+              quoted = true;
+            }
 
             m = mode_.top ();
             continue;
@@ -307,7 +312,7 @@ namespace build
     if (m == lexer_mode::variable)
       mode_.pop ();
 
-    return token (lexeme, sep, ln, cn);
+    return token (lexeme, sep, quoted, ln, cn);
   }
 
   bool lexer::
diff --git a/build/parser.cxx b/build/parser.cxx
index 42b9a98..1ceb193 100644
--- a/build/parser.cxx
+++ b/build/parser.cxx
@@ -90,11 +90,12 @@ namespace build
           tt != type::colon)     // Empty name: ': ...'
         break; // Something else. Let our caller handle that.
 
-      // See if this is one of the keywords.
+      // See if this is one of the directives. This should be an
+      // unquoted literal name.
       //
-      if (tt == type::name)
+      if (tt == type::name && !t.quoted)
       {
-        const string& n (t.name ());
+        const string& n (t.value);
 
         if (n == "print")
         {
@@ -642,7 +643,7 @@ namespace build
 
       if (at == token_type::equal || at == token_type::plus_equal)
       {
-        var = &variable_pool.find (t.name ());
+        var = &variable_pool.find (t.value);
         val = at == token_type::equal
           ? &scope_->assign (*var)
           : &scope_->append (*var);
@@ -857,10 +858,10 @@ namespace build
       if (!concat.empty () &&
           ((tt != type::name   &&
             tt != type::dollar &&
-            tt != type::lparen) || peeked ().separated ()))
+            tt != type::lparen) || peeked ().separated))
       {
         tt = type::name;
-        t = token (move (concat), true, t.line (), t.column ());
+        t = token (move (concat), true, false, t.line, t.column);
         concat.clear ();
       }
       else if (!first)
@@ -870,7 +871,7 @@ namespace build
       //
       if (tt == type::name)
       {
-        string name (t.name ()); //@@ move?
+        string name (t.value); //@@ move?
         tt = peek ();
 
         // Should we accumulate? If the buffer is not empty, then
@@ -881,7 +882,7 @@ namespace build
         //
         if (!concat.empty () ||                                // Continue.
             ((tt == type::dollar ||
-              tt == type::lparen) && !peeked ().separated ())) // Start.
+              tt == type::lparen) && !peeked ().separated))    // Start.
         {
           concat += name;
           continue;
@@ -1062,7 +1063,7 @@ namespace build
 
           string n;
           if (tt == type::name)
-            n = t.name ();
+            n = t.value;
           else if (tt == type::lparen)
           {
             lexer_->expire_mode ();
@@ -1159,7 +1160,7 @@ namespace build
         if (!concat.empty () ||                       // Continue.
             ((tt == type::name   ||                   // Start.
               tt == type::dollar ||
-              tt == type::lparen) && !peeked ().separated ()))
+              tt == type::lparen) && !peeked ().separated))
         {
           // This should be a simple value or a simple directory. The
           // token still points to the name (or closing paren).
@@ -1642,7 +1643,7 @@ namespace build
       peeked_ = false;
     }
 
-    tt = t.type ();
+    tt = t.type;
     return tt;
   }
 
@@ -1655,7 +1656,7 @@ namespace build
       peeked_ = true;
     }
 
-    return peek_.type ();
+    return peek_.type;
   }
 
   static location
@@ -1663,6 +1664,6 @@ namespace build
   {
     assert (data != nullptr);
     const string& p (**static_cast<const string* const*> (data));
-    return location (p.c_str (), t.line (), t.column ());
+    return location (p.c_str (), t.line, t.column);
   }
 }
diff --git a/build/token b/build/token
index 3b9edb1..fedaba0 100644
--- a/build/token
+++ b/build/token
@@ -9,7 +9,6 @@
 #include <iosfwd>
 #include <cstddef> // size_t
 #include <cstdint> // uint64_t
-#include <cassert>
 #include <utility> // move
 
 namespace build
@@ -33,34 +32,26 @@ namespace build
   class token
   {
   public:
-    token_type
-    type () const {return t_;}
+    token_type type;
+    bool separated; // Whitespace-separated from the previous token.
+    bool quoted;    // Name (or some part of it) was quoted.
 
-    // Token is whitespace-separated from the previous token.
-    //
-    bool
-    separated () const {return s_;}
+    std::string value; // Only valid for name.
 
-    std::string const&
-    name () const {assert (t_ == token_type::name); return n_;}
-
-    std::uint64_t line () const {return l_;}
-    std::uint64_t column () const {return c_;}
+    std::uint64_t line;
+    std::uint64_t column;
 
   public:
     token (token_type t, bool s, std::uint64_t l, std::uint64_t c)
-        : t_ (t), s_ (s), l_ (l), c_ (c) {}
-
-    token (std::string n, bool s, std::uint64_t l, std::uint64_t c)
-        : t_ (token_type::name), s_ (s), n_ (std::move (n)), l_ (l), c_ (c) {}
-
-  private:
-    token_type t_;
-    bool s_;
-    std::string n_;
-
-    std::uint64_t l_;
-    std::uint64_t c_;
+        : type (t), separated (s), line (l), column (c) {}
+
+    token (std::string n, bool s, bool q, std::uint64_t l, std::uint64_t c)
+        : type (token_type::name),
+          separated (s),
+          quoted (q),
+          value (std::move (n)),
+          line (l),
+          column (c) {}
   };
 
   // Output the token value in a format suitable for diagnostics.
diff --git a/build/token.cxx b/build/token.cxx
index 22f5b75..35a7f89 100644
--- a/build/token.cxx
+++ b/build/token.cxx
@@ -13,7 +13,7 @@ namespace build
   ostream&
   operator<< (ostream& os, const token& t)
   {
-    switch (t.type ())
+    switch (t.type)
     {
     case token_type::eos:            os << "<end-of-file>"; break;
     case token_type::newline:        os << "<newline>"; break;
@@ -26,7 +26,7 @@ namespace build
     case token_type::dollar:         os << "$"; break;
     case token_type::lparen:         os << "("; break;
     case token_type::rparen:         os << ")"; break;
-    case token_type::name:           os << t.name (); break;
+    case token_type::name:           os << t.value; break;
     }
 
     return os;
diff --git a/tests/directive/buildfile b/tests/directive/buildfile
new file mode 100644
index 0000000..0f83717
--- /dev/null
+++ b/tests/directive/buildfile
@@ -0,0 +1,8 @@
+"print"  = a
+'print' += b
+pr"int" += c
+print'' += d
+
+print $print
+
+./:
diff --git a/tests/directive/test.out b/tests/directive/test.out
new file mode 100644
index 0000000..8e13e46
--- /dev/null
+++ b/tests/directive/test.out
@@ -0,0 +1 @@
+a b c d
diff --git a/tests/directive/test.sh b/tests/directive/test.sh
new file mode 100755
index 0000000..b898b3c
--- /dev/null
+++ b/tests/directive/test.sh
@@ -0,0 +1,3 @@
+#!/bin/sh
+
+valgrind -q b -q | diff -u test.out -
-- 
cgit v1.1