From 861555876837449fd8fc480845f642dab4a68b1e Mon Sep 17 00:00:00 2001
From: Karen Arutyunov <karen@codesynthesis.com>
Date: Tue, 12 Nov 2019 15:59:58 +0300
Subject: Use path_name in cc::lexer class

---
 libbuild2/cc/compile-rule.cxx                      |  2 +-
 libbuild2/cc/lexer+char-literal.test.testscript    |  2 +-
 libbuild2/cc/lexer+comment.test.testscript         |  2 +-
 libbuild2/cc/lexer+preprocessor.test.testscript    |  4 +-
 .../cc/lexer+raw-string-literal.test.testscript    | 10 +--
 libbuild2/cc/lexer+string-literal.test.testscript  |  2 +-
 libbuild2/cc/lexer.cxx                             | 73 +++++++++++++++-------
 libbuild2/cc/lexer.hxx                             | 23 ++++---
 libbuild2/cc/lexer.test.cxx                        | 17 +++--
 libbuild2/cc/parser+module.test.testscript         | 14 ++---
 libbuild2/cc/parser.cxx                            |  4 +-
 libbuild2/cc/parser.hxx                            |  2 +-
 libbuild2/cc/parser.test.cxx                       | 12 ++--
 libbuild2/test/script/parser.cxx                   |  2 +-
 libbuild2/test/script/script.hxx                   | 14 ++++-
 15 files changed, 119 insertions(+), 64 deletions(-)

diff --git a/libbuild2/cc/compile-rule.cxx b/libbuild2/cc/compile-rule.cxx
index 98b78e1..f3b2c0d 100644
--- a/libbuild2/cc/compile-rule.cxx
+++ b/libbuild2/cc/compile-rule.cxx
@@ -4218,7 +4218,7 @@ namespace build2
                         fdstream_mode::binary | fdstream_mode::skip);
 
           parser p;
-          unit tu (p.parse (is, *sp));
+          unit tu (p.parse (is, path_name (*sp)));
 
           is.close ();
 
diff --git a/libbuild2/cc/lexer+char-literal.test.testscript b/libbuild2/cc/lexer+char-literal.test.testscript
index afd16dd..e7a44d9 100644
--- a/libbuild2/cc/lexer+char-literal.test.testscript
+++ b/libbuild2/cc/lexer+char-literal.test.testscript
@@ -63,5 +63,5 @@ EOO
 : unterminated
 :
 $* <"'a" 2>>EOE != 0
-stdin:1:1: error: unterminated character literal
+<stdin>:1:1: error: unterminated character literal
 EOE
diff --git a/libbuild2/cc/lexer+comment.test.testscript b/libbuild2/cc/lexer+comment.test.testscript
index bfcc440..5c29af0 100644
--- a/libbuild2/cc/lexer+comment.test.testscript
+++ b/libbuild2/cc/lexer+comment.test.testscript
@@ -41,7 +41,7 @@ $* <<EOI 2>>EOE != 0
 /*
 comment
 EOI
-stdin:1:2: error: unterminated comment
+<stdin>:1:2: error: unterminated comment
 EOE
 
 : cxx-unterminated
diff --git a/libbuild2/cc/lexer+preprocessor.test.testscript b/libbuild2/cc/lexer+preprocessor.test.testscript
index e33eb90..3403588 100644
--- a/libbuild2/cc/lexer+preprocessor.test.testscript
+++ b/libbuild2/cc/lexer+preprocessor.test.testscript
@@ -46,7 +46,7 @@ $* -l <<EOI >>EOO
 # 5 "test.cxx"
 ;
 EOI
-';' stdin:1:1
+';' <stdin>:1:1
 ';' test.cxx:1:1
 ';' test.cxx:2:3
 ';' test.cxx:4:1
@@ -60,7 +60,7 @@ EOO
 $* <<EOI 2>>EOE != 0
 #include <foo/bar>
 EOI
-stdin:1:1: error: unexpected #include directive
+<stdin>:1:1: error: unexpected #include directive
 EOE
 
 : nested
diff --git a/libbuild2/cc/lexer+raw-string-literal.test.testscript b/libbuild2/cc/lexer+raw-string-literal.test.testscript
index 93cddc1..beab7b4 100644
--- a/libbuild2/cc/lexer+raw-string-literal.test.testscript
+++ b/libbuild2/cc/lexer+raw-string-literal.test.testscript
@@ -62,29 +62,29 @@ EOO
 : invalid-no-paren
 :
 $* <'R"a"' 2>>EOE != 0
-stdin:1:2: error: invalid raw string literal
+<stdin>:1:2: error: invalid raw string literal
 EOE
 
 : invalid-paren
 :
 $* <'R")()("' 2>>EOE != 0
-stdin:1:2: error: invalid raw string literal
+<stdin>:1:2: error: invalid raw string literal
 EOE
 
 : invalid-unterminated-paren
 :
 $* <'R"(abc"' 2>>EOE != 0
-stdin:1:2: error: invalid raw string literal
+<stdin>:1:2: error: invalid raw string literal
 EOE
 
 : invalid-unterminated-delimiter
 :
 $* <'R"X(abc)"' 2>>EOE != 0
-stdin:1:2: error: invalid raw string literal
+<stdin>:1:2: error: invalid raw string literal
 EOE
 
 : invalid-unterminated-quote
 :
 $* <'R"X(abc)X' 2>>EOE != 0
-stdin:1:2: error: invalid raw string literal
+<stdin>:1:2: error: invalid raw string literal
 EOE
diff --git a/libbuild2/cc/lexer+string-literal.test.testscript b/libbuild2/cc/lexer+string-literal.test.testscript
index a2509c9..22ddff8 100644
--- a/libbuild2/cc/lexer+string-literal.test.testscript
+++ b/libbuild2/cc/lexer+string-literal.test.testscript
@@ -61,5 +61,5 @@ EOO
 : unterminated
 :
 $* <'"ab' 2>>EOE != 0
-stdin:1:1: error: unterminated string literal
+<stdin>:1:1: error: unterminated string literal
 EOE
diff --git a/libbuild2/cc/lexer.cxx b/libbuild2/cc/lexer.cxx
index 6eba57e..540e48c 100644
--- a/libbuild2/cc/lexer.cxx
+++ b/libbuild2/cc/lexer.cxx
@@ -38,7 +38,7 @@ namespace butl // ADL
     using namespace build2;
 
     assert (data != nullptr); // E.g., must be &lexer::name_.
-    return location (static_cast<const path*> (data), c.line, c.column);
+    return location (*static_cast<const path_name*> (data), c.line, c.column);
   }
 }
 
@@ -146,7 +146,7 @@ namespace build2
     {
       for (;; c = skip_spaces ())
       {
-        t.file = log_file_;
+        t.file = &log_file_;
         t.line = log_line_ ? *log_line_ : c.line;
         t.column = c.column;
 
@@ -156,7 +156,7 @@ namespace build2
           return;
         }
 
-        const location l (&name_, c.line, c.column);
+        const location l (name_, c.line, c.column);
 
         // Hash the token's line. The reason is debug info. In fact, doing
         // this will make quite a few "noop" changes (like adding a newline
@@ -646,7 +646,7 @@ namespace build2
     {
       // note: c is hashed
 
-      const location l (&name_, c.line, c.column);
+      const location l (name_, c.line, c.column);
 
       for (char p (c);;) // Previous character (see below).
       {
@@ -677,7 +677,7 @@ namespace build2
     {
       // note: c is hashed
 
-      const location l (&name_, c.line, c.column);
+      const location l (name_, c.line, c.column);
 
       for (char p (c);;) // Previous character (see below).
       {
@@ -736,7 +736,7 @@ namespace build2
       // Note that the <raw_characters> are not processed in any way, not even
       // for line continuations.
       //
-      const location l (&name_, c.line, c.column);
+      const location l (name_, c.line, c.column);
 
       // As a first step, parse the delimiter (including the openning paren).
       //
@@ -830,7 +830,7 @@ namespace build2
 
       if (c == '\"')
       {
-        const location l (&name_, c.line, c.column);
+        const location l (name_, c.line, c.column);
 
         // It is common to have a large number of #line directives that don't
         // change the file (they seem to be used to track macro locations or
@@ -895,15 +895,42 @@ namespace build2
           }
         }
 
-        if (log_file_.string () == s)
-          return;
+        try
+        {
+          if (s.empty ())
+            throw invalid_path ("");
 
-        // Swap the two string buffers.
-        //
+          // Handle special names (<stdin>, <built-in>, etc).
+          //
+          if (s.front () == '<' && s.back () == '>')
+          {
+            if (log_file_.name)
+            {
+              if (*log_file_.name == s)
+                return;
+
+              log_file_.name->swap (s);
+            }
+            else
+              log_file_.name = move (s);
+
+            log_file_.path.clear ();
+          }
+          else
+          {
+            if (log_file_.path.string () == s)
+              return;
+
+            string r (move (log_file_.path).string ()); // Move string rep out.
+            r.swap (s);
+            log_file_.path = path (move (r)); // Move back in.
+
+            log_file_.name = nullopt;
+          }
+        }
+        catch (const invalid_path&)
         {
-          string r (move (log_file_).string ()); // Move string rep out.
-          r.swap (s);
-          log_file_ = path (move (r)); // Move back in.
+          fail (l) << "invalid path in #line directive";
         }
 
         // If the path is relative, then prefix it with the current working
@@ -926,17 +953,19 @@ namespace build2
         // the part starting from the project root which is immutable. Plus
         // we will need -ffile-prefix-map to deal with __FILE__.
         //
-        if (!log_file_.to_directory ())
-          cs_.append (log_file_.string ());
+        if (!log_file_.path.to_directory ()) // Also covers special names.
+          cs_.append (log_file_.name
+                      ? *log_file_.name
+                      : log_file_.path.string ());
 #if 0
         {
-          using tr = path::traits;
-          const string& f (log_file_.string ());
+          using tr = path::traits_type;
+          const string& f (log_file_.path.string ());
 
-          if (f.find (':') != string::npos            ||
-              (f.front () == '<' && f.back () == '>') ||
-              log_file_.absolute ())
-            cs_.append (f);
+          if (log_file_.name               ||
+              f.find (':') != string::npos ||
+              log_file_.path.absolute ())
+            cs_.append (log_file_.name ? *log_file_.name : f);
           else
           {
             // This gets complicated and slow: the path may contain '..' and
diff --git a/libbuild2/cc/lexer.hxx b/libbuild2/cc/lexer.hxx
index cb2b3a5..c7da135 100644
--- a/libbuild2/cc/lexer.hxx
+++ b/libbuild2/cc/lexer.hxx
@@ -63,9 +63,12 @@ namespace build2
 
       // Logical position.
       //
-      path     file;
-      uint64_t line   = 0;
-      uint64_t column = 0;
+      // Note that file is a shallow pointer to the state maintained by the
+      // lexer.
+      //
+      const path_name* file   = nullptr;
+      uint64_t         line   = 0;
+      uint64_t         column = 0;
 
       // Physical position in the stream, currently only for identifiers.
       //
@@ -80,13 +83,15 @@ namespace build2
     class lexer: protected butl::char_scanner
     {
     public:
-      lexer (ifdstream& is, const path& name)
+      lexer (ifdstream& is, const path_name& name)
           : char_scanner (is, false),
             name_ (name),
             fail ("error", &name_),
-            log_file_ (name) {}
+            log_file_ (name)
+      {
+      }
 
-      const path&
+      const path_name&
       name () const {return name_;}
 
       string
@@ -164,13 +169,13 @@ namespace build2
       geth (const xchar& peeked);
 
     private:
-      const path name_;
+      const path_name& name_;
       const fail_mark fail;
 
       // Logical file and line as set by the #line directives. Note that the
       // lexer diagnostics still uses the physical file/lines.
       //
-      path               log_file_;
+      path_name_value    log_file_;
       optional<uint64_t> log_line_;
 
       string tmp_file_;
@@ -182,7 +187,7 @@ namespace build2
     inline location
     get_location (const token& t, const void* = nullptr)
     {
-      return location (&t.file, t.line, t.column);
+      return location (*t.file, t.line, t.column);
     }
   }
 }
diff --git a/libbuild2/cc/lexer.test.cxx b/libbuild2/cc/lexer.test.cxx
index 0aeadba..68bb152 100644
--- a/libbuild2/cc/lexer.test.cxx
+++ b/libbuild2/cc/lexer.test.cxx
@@ -23,7 +23,7 @@ namespace build2
     main (int argc, char* argv[])
     {
       bool loc (false);
-      const char* file (nullptr);
+      path file;
 
       for (int i (1); i != argc; ++i)
       {
@@ -33,23 +33,28 @@ namespace build2
           loc = true;
         else
         {
-          file = argv[i];
+          file = path (argv[i]);
           break;
         }
       }
 
       try
       {
+        path_name in;
         ifdstream is;
-        if (file != nullptr)
+
+        if (!file.empty ())
+        {
+          in = path_name (file);
           is.open (file);
+        }
         else
         {
-          file = "stdin";
+          in = path_name ("<stdin>");
           is.open (fddup (stdin_fd ()));
         }
 
-        lexer l (is, path (file));
+        lexer l (is, in);
 
         // No use printing eos since we will either get it or loop forever.
         //
@@ -58,7 +63,7 @@ namespace build2
           cout << t;
 
           if (loc)
-            cout << ' ' << t.file << ':' << t.line << ':' << t.column;
+            cout << ' ' << *t.file << ':' << t.line << ':' << t.column;
 
           cout << endl;
         }
diff --git a/libbuild2/cc/parser+module.test.testscript b/libbuild2/cc/parser+module.test.testscript
index d51ac0a..e935180 100644
--- a/libbuild2/cc/parser+module.test.testscript
+++ b/libbuild2/cc/parser+module.test.testscript
@@ -89,7 +89,7 @@ export
   module foo;
 }
 EOI
-stdin:8:1: error: {}-imbalance detected
+<stdin>:8:1: error: {}-imbalance detected
 EOE
 
 : brace-stray
@@ -103,7 +103,7 @@ export
 }
 module foo;
 EOI
-stdin:6:1: error: {}-imbalance detected
+<stdin>:6:1: error: {}-imbalance detected
 EOE
 
 : import-missing-name
@@ -111,7 +111,7 @@ EOE
 $* <<EOI 2>>EOE != 0
 import ;
 EOI
-stdin:1:8: error: module or header name expected instead of ';'
+<stdin>:1:8: error: module or header name expected instead of ';'
 EOE
 
 : module-missing-name
@@ -119,7 +119,7 @@ EOE
 $* <<EOI 2>>EOE != 0
 module ;
 EOI
-stdin:1:1: error: module declaration expected after leading module marker
+<stdin>:1:1: error: module declaration expected after leading module marker
 EOE
 
 : import-missing-semi
@@ -127,7 +127,7 @@ EOE
 $* <<EOI 2>>EOE != 0
 import foo
 EOI
-stdin:2:1: error: ';' expected instead of <end of file>
+<stdin>:2:1: error: ';' expected instead of <end of file>
 EOE
 
 : module-missing-semi
@@ -135,7 +135,7 @@ EOE
 $* <<EOI 2>>EOE != 0
 export module foo
 EOI
-stdin:2:1: error: ';' expected instead of <end of file>
+<stdin>:2:1: error: ';' expected instead of <end of file>
 EOE
 
 : import-missing-header
@@ -143,5 +143,5 @@ EOE
 $* <<EOI 2>>EOE != 0
 import <foo.h;
 EOI
-stdin:2:1: error: closing '>' expected after header name
+<stdin>:2:1: error: closing '>' expected after header name
 EOE
diff --git a/libbuild2/cc/parser.cxx b/libbuild2/cc/parser.cxx
index 179043e..abfe5bd 100644
--- a/libbuild2/cc/parser.cxx
+++ b/libbuild2/cc/parser.cxx
@@ -16,9 +16,9 @@ namespace build2
     using type = token_type;
 
     unit parser::
-    parse (ifdstream& is, const path& name)
+    parse (ifdstream& is, const path_name& in)
     {
-      lexer l (is, name);
+      lexer l (is, in);
       l_ = &l;
 
       unit u;
diff --git a/libbuild2/cc/parser.hxx b/libbuild2/cc/parser.hxx
index 324b62a..cc2eaa3 100644
--- a/libbuild2/cc/parser.hxx
+++ b/libbuild2/cc/parser.hxx
@@ -25,7 +25,7 @@ namespace build2
     {
     public:
       unit
-      parse (ifdstream&, const path& name);
+      parse (ifdstream&, const path_name&);
 
     private:
       void
diff --git a/libbuild2/cc/parser.test.cxx b/libbuild2/cc/parser.test.cxx
index 82c68d1..9d90653 100644
--- a/libbuild2/cc/parser.test.cxx
+++ b/libbuild2/cc/parser.test.cxx
@@ -24,22 +24,26 @@ namespace build2
     {
       try
       {
-        const char* file;
+        path file;
 
+        path_name in;
         ifdstream is;
+
         if (argc > 1)
         {
-          file = argv[1];
+          file = path (argv[1]);
+
+          in = path_name (file);
           is.open (file);
         }
         else
         {
-          file = "stdin";
+          in = path_name ("<stdin>");
           is.open (fddup (stdin_fd ()));
         }
 
         parser p;
-        unit u (p.parse (is, path (file)));
+        unit u (p.parse (is, in));
         unit_type ut (u.type);
 
         for (const module_import& m: u.module_info.imports)
diff --git a/libbuild2/test/script/parser.cxx b/libbuild2/test/script/parser.cxx
index 376d68f..f3f6ffa 100644
--- a/libbuild2/test/script/parser.cxx
+++ b/libbuild2/test/script/parser.cxx
@@ -1028,7 +1028,7 @@ namespace build2
           };
 
           const path_name_value& pn (enter_path (move (n)));
-          const path& p (*pn.path);
+          const path& p (pn.path);
 
           if (include_set_->insert (p).second || !once)
           {
diff --git a/libbuild2/test/script/script.hxx b/libbuild2/test/script/script.hxx
index f4ef32a..73fd696 100644
--- a/libbuild2/test/script/script.hxx
+++ b/libbuild2/test/script/script.hxx
@@ -548,7 +548,19 @@ namespace build2
         // Testscript file paths. Specifically, replay_token::file points to
         // these path names.
         //
-        std::set<path_name_value> paths_;
+        struct compare_paths
+        {
+          bool operator() (const path_name_value& x,
+                           const path_name_value& y) const
+          {
+            // Note that these path names are always paths, so we compare them
+            // as paths.
+            //
+            return x.path < y.path;
+          }
+        };
+
+        std::set<path_name_value, compare_paths> paths_;
       };
     }
   }
-- 
cgit v1.1