From 74918fea2ae0ce55fb6541726ec2ab607d7e2d6e Mon Sep 17 00:00:00 2001
From: Boris Kolpackov <boris@codesynthesis.com>
Date: Wed, 4 Jan 2017 17:00:15 +0200
Subject: Change lexer modes to be semantically accurate

---
 build2/test/script/lexer      |  4 ++--
 build2/test/script/lexer.cxx  | 32 ++++++++++++++++----------------
 build2/test/script/parser.cxx | 10 ++++++----
 3 files changed, 24 insertions(+), 22 deletions(-)

diff --git a/build2/test/script/lexer b/build2/test/script/lexer
index a6a96ca..4a6cd14 100644
--- a/build2/test/script/lexer
+++ b/build2/test/script/lexer
@@ -24,11 +24,11 @@ namespace build2
 
         enum
         {
-          script_line = base_type::value_next,
+          command_line = base_type::value_next,
           first_token,     // Expires at the end of the token.
           second_token,    // Expires at the end of the token.
           variable_line,   // Expires at the end of the line.
-          command_line,
+          command_expansion,
           here_line_single,
           here_line_double,
           description_line // Expires at the end of the line.
diff --git a/build2/test/script/lexer.cxx b/build2/test/script/lexer.cxx
index 8bd3484..ce6ed5d 100644
--- a/build2/test/script/lexer.cxx
+++ b/build2/test/script/lexer.cxx
@@ -32,7 +32,7 @@ namespace build2
 
         switch (m)
         {
-        case lexer_mode::script_line:
+        case lexer_mode::command_line:
           {
             s1 = ":;=!|&<> $(#\t\n";
             s2 = "  ==          ";
@@ -40,9 +40,9 @@ namespace build2
           }
         case lexer_mode::first_token:
           {
-            // First token on the script line. Like script_line but recognizes
-            // leading '+-{}' as tokens as well as variable assignments as
-            // separators.
+            // First token on the script line. Like command_line but
+            // recognizes leading '+-{}' as tokens as well as variable
+            // assignments as separators.
             //
             // Note that to recognize only leading '+-{}' we shouldn't add
             // them to the separator strings.
@@ -53,12 +53,12 @@ namespace build2
           }
         case lexer_mode::second_token:
           {
-            // Second token on the script line. Like script_line but
+            // Second token on the script line. Like command_line but
             // recognizes leading variable assignments.
             //
             // Note that to recognize only leading assignments we shouldn't
             // add them to the separator strings (so this is identical to
-            // script_line).
+            // command_line).
             //
             s1 = ":;=!|&<> $(#\t\n";
             s2 = "  ==          ";
@@ -75,7 +75,7 @@ namespace build2
             break;
           }
 
-        case lexer_mode::command_line:
+        case lexer_mode::command_expansion:
           {
             // Note that whitespaces are not word separators in this mode.
             //
@@ -145,11 +145,11 @@ namespace build2
 
         switch (state_.top ().mode)
         {
-        case lexer_mode::script_line:
+        case lexer_mode::command_line:
         case lexer_mode::first_token:
         case lexer_mode::second_token:
         case lexer_mode::variable_line:
-        case lexer_mode::command_line:
+        case lexer_mode::command_expansion:
         case lexer_mode::here_line_single:
         case lexer_mode::here_line_double:
           r = next_line ();
@@ -224,7 +224,7 @@ namespace build2
 
         // NOTE: remember to update mode() if adding new special characters.
 
-        if (m != lexer_mode::command_line)
+        if (m != lexer_mode::command_expansion)
         {
           switch (c)
           {
@@ -266,7 +266,7 @@ namespace build2
 
         // Line separators.
         //
-        if (m == lexer_mode::script_line  ||
+        if (m == lexer_mode::command_line ||
             m == lexer_mode::first_token  ||
             m == lexer_mode::second_token ||
             m == lexer_mode::variable_line)
@@ -277,7 +277,7 @@ namespace build2
           }
         }
 
-        if (m == lexer_mode::script_line  ||
+        if (m == lexer_mode::command_line ||
             m == lexer_mode::first_token  ||
             m == lexer_mode::second_token)
         {
@@ -289,8 +289,8 @@ namespace build2
 
         // Command line operator/separators.
         //
-        if (m == lexer_mode::script_line ||
-            m == lexer_mode::first_token ||
+        if (m == lexer_mode::command_line ||
+            m == lexer_mode::first_token  ||
             m == lexer_mode::second_token)
         {
           switch (c)
@@ -311,10 +311,10 @@ namespace build2
 
         // Command operators/separators.
         //
-        if (m == lexer_mode::script_line  ||
+        if (m == lexer_mode::command_line ||
             m == lexer_mode::first_token  ||
             m == lexer_mode::second_token ||
-            m == lexer_mode::command_line)
+            m == lexer_mode::command_expansion)
         {
           switch (c)
           {
diff --git a/build2/test/script/parser.cxx b/build2/test/script/parser.cxx
index da61c64..99e4a66 100644
--- a/build2/test/script/parser.cxx
+++ b/build2/test/script/parser.cxx
@@ -57,7 +57,7 @@ namespace build2
 
         pre_parse_ = true;
 
-        lexer l (is, *path_, lexer_mode::script_line);
+        lexer l (is, *path_, lexer_mode::command_line);
         set_lexer (&l);
 
         id_prefix_.clear ();
@@ -989,7 +989,7 @@ namespace build2
             try
             {
               ifdstream ifs (p);
-              lexer l (ifs, p, lexer_mode::script_line);
+              lexer l (ifs, p, lexer_mode::command_line);
 
               const path* op (path_);
               path_ = &p;
@@ -1908,7 +1908,7 @@ namespace build2
                     name = path (move (n));
                   }
 
-                  // When re-parsing we do "effective escaping" and only for
+                  // When re-lexing we do "effective escaping" and only for
                   // ['"\] (quotes plus the backslash itself). In particular,
                   // there is no way to escape redirects, operators, etc. The
                   // idea is to prefer quoting except for passing literal
@@ -1921,7 +1921,9 @@ namespace build2
                   // cmd $args               # cmd x="foo bar"
                   //
                   istringstream is (s);
-                  lexer lex (is, name, lexer_mode::command_line, "\'\"\\");
+                  lexer lex (is, name,
+                             lexer_mode::command_expansion,
+                             "\'\"\\");
 
                   // Treat the first "sub-token" as always separated from what
                   // we saw earlier.
-- 
cgit v1.1