From a473abe80f4c42a366f0573bbbc762fa440b7fe6 Mon Sep 17 00:00:00 2001
From: Boris Kolpackov <boris@codesynthesis.com>
Date: Tue, 26 Apr 2022 10:39:03 +0200
Subject: Use new cmdline type for canned command lines in {Build,Test}script

---
 libbuild2/script/parser.cxx | 47 +++++++++++++++++++++++++++++++++++----------
 libbuild2/script/parser.hxx | 19 ++++++++++++++++--
 2 files changed, 54 insertions(+), 12 deletions(-)

(limited to 'libbuild2/script')

diff --git a/libbuild2/script/parser.cxx b/libbuild2/script/parser.cxx
index 82eb9c8..d5cabe2 100644
--- a/libbuild2/script/parser.cxx
+++ b/libbuild2/script/parser.cxx
@@ -3,6 +3,8 @@
 
 #include <libbuild2/script/parser.hxx>
 
+#include <cstring> // strchr()
+
 #include <libbuild2/variable.hxx>
 #include <libbuild2/script/run.hxx>   // exit
 #include <libbuild2/script/lexer.hxx>
@@ -15,6 +17,33 @@ namespace build2
   {
     using type = token_type;
 
+    bool parser::
+    need_cmdline_relex (const string& s)
+    {
+      for (auto i (s.begin ()), e (s.end ()); i != e; ++i)
+      {
+        char c (*i);
+
+        if (c == '\\')
+        {
+          if (++i != e)
+            return false;
+
+          c = *i;
+
+          if (c == '\\' || c == '\'' || c == '\"')
+            return true;
+
+          // Fall through.
+        }
+
+        if (strchr ("|<>&\"'", c) != nullptr)
+          return true;
+      }
+
+      return false;
+    }
+
     value parser::
     parse_variable_line (token& t, type& tt)
     {
@@ -1092,16 +1121,17 @@ namespace build2
 
             // Process what we got.
             //
-            // First see if this is a value that should not be re-lexed. The
-            // long term plan is to only re-lex values of a special type
-            // representing a canned command line.
+            // First see if this is a value that should not be re-lexed. We
+            // only re-lex values of the special `cmdline` type that
+            // represents a canned command line.
             //
             // Otherwise, determine whether anything inside was quoted (note
             // that the current token is "next" and is not part of this).
             //
-            bool q (
-              (pr.value && !relex_) ||
-              (quoted () - (t.qtype != quote_type::unquoted ? 1 : 0)) != 0);
+            bool lex (
+              pr.value
+              ? pr.type != nullptr && pr.type->is_a<cmdline> ()
+              : (quoted () - (t.qtype != quote_type::unquoted ? 1 : 0)) == 0);
 
             for (name& n: ns)
             {
@@ -1123,10 +1153,7 @@ namespace build2
               // interesting characters (operators plus quotes/escapes),
               // then no need to re-lex.
               //
-              // NOTE: update quoting (script.cxx:to_stream_q()) if adding
-              // any new characters.
-              //
-              if (q || s.find_first_of ("|&<>\'\"\\") == string::npos)
+              if (!lex || !need_cmdline_relex (s))
                 add_word (move (s), l);
               else
               {
diff --git a/libbuild2/script/parser.hxx b/libbuild2/script/parser.hxx
index 6e24d37..d8e5dbf 100644
--- a/libbuild2/script/parser.hxx
+++ b/libbuild2/script/parser.hxx
@@ -25,7 +25,7 @@ namespace build2
     class parser: protected build2::parser
     {
     public:
-      parser (context& c, bool relex): build2::parser (c), relex_ (relex) {}
+      parser (context& c): build2::parser (c) {}
 
       // Helpers.
       //
@@ -42,6 +42,15 @@ namespace build2
 
       using build2::parser::apply_value_attributes;
 
+      // Return true if a command line element needs to be re-lexed.
+      //
+      // Specifically, it needs to be re-lexed if it contains any of the
+      // special characters (|<>&), quotes ("') or effective escape sequences
+      // (\", \', \\).
+      //
+      static bool
+      need_cmdline_relex (const string&);
+
       // Commonly used parsing functions. Issue diagnostics and throw failed
       // in case of an error.
       //
@@ -200,6 +209,13 @@ namespace build2
       // something that requires re-lexing, for example `foo|bar`, which won't
       // be easy to translate but which are handled by the parser.
       //
+      // Note that the chunk could be of the special cmdline type in which
+      // case the names may need to be "preprocessed" (at least unquoted or
+      // potentially fully re-lexed) before being analyzed/consumed. Note also
+      // that in this case any names left unconsumed must remain of the
+      // cmdline type.
+      //
+      //
       // During the pre-parsing phase the returned process path and names
       // (that must still be parsed) are discarded. The main purpose of the
       // call is to allow implementations to perform static script analysis,
@@ -229,7 +245,6 @@ namespace build2
       size_t replay_quoted_;
 
     protected:
-      bool relex_;
       lexer* lexer_ = nullptr;
     };
   }
-- 
cgit v1.1