Add env script pseudo-builtin

Also disable C++ recipe tests when cross-testing.

9 files changed, 430 insertions, 23 deletions

diff --git a/libbuild2/build/script/parser+diag.test.testscript b/libbuild2/build/script/parser+diag.test.testscript
index 60683bc..5b4e64a 100644
--- a/libbuild2/build/script/parser+diag.test.testscript
+++ b/libbuild2/build/script/parser+diag.test.testscript
@@ -55,3 +55,39 @@ $* <<EOI >>~%EOO%
       buildfile:12:1: info: previous call is here
     EOE
 }
+
+: inside-if
+:
+$* <<EOI 2>>EOE != 0
+  if true
+    diag copy >= $>
+  fi
+  EOI
+  buildfile:12:3: error: 'diag' call inside flow control construct
+  EOE
+
+: inside-if-cond
+:
+$* <<EOI 2>>EOE != 0
+  if diag copy >= $>
+    true
+  fi
+  EOI
+  buildfile:11:4: error: 'diag' call inside flow control construct
+  EOE
+
+: second-command
+:
+$* <<EOI 2>>EOE != 0
+  true && diag copy >= $>
+  EOI
+  buildfile:11:9: error: 'diag' call must be the only command
+  EOE
+
+: via-env
+:
+$* <<EOI 2>>EOE != 0
+  env -- diag copy >= $>
+  EOI
+  buildfile:11:8: error: 'diag' call via 'env' builtin
+  EOE
diff --git a/libbuild2/build/script/parser.cxx b/libbuild2/build/script/parser.cxx
index 2c41ac1..8f2c46d 100644
--- a/libbuild2/build/script/parser.cxx
+++ b/libbuild2/build/script/parser.cxx
@@ -330,6 +330,7 @@ namespace build2
       optional<process_path> parser::
       parse_program (token& t, build2::script::token_type& tt,
                      bool first,
+                     bool env,
                      names& ns)
       {
         const location l (get_location (t));
@@ -369,13 +370,16 @@ namespace build2
           // Verify that the special builtin is not called inside an improper
           // context (flow control construct or complex expression).
           //
-          auto verify = [first, &v, &l, this] ()
+          auto verify = [first, env, &v, &l, this] ()
           {
             if (level_ != 0)
               fail (l) << "'" << v << "' call inside flow control construct";
 
             if (!first)
               fail (l) << "'" << v << "' call must be the only command";
+
+            if (env)
+              fail (l) << "'" << v << "' call via 'env' builtin";
           };
 
           if (v == "diag")
diff --git a/libbuild2/build/script/parser.hxx b/libbuild2/build/script/parser.hxx
index 73bcd09..5ada8be 100644
--- a/libbuild2/build/script/parser.hxx
+++ b/libbuild2/build/script/parser.hxx
@@ -98,13 +98,16 @@ namespace build2
         //
         // During pre-parsing try to deduce the low-verbosity script
         // diagnostics name as a program/builtin name or obtain the custom
-        // low-verbosity diagnostics specified with the diag builtin. Note
-        // that the diag builtin can only appear at the beginning of the
-        // command line.
+        // low-verbosity diagnostics specified with the diag builtin. Also
+        // handle the depdb builtin calls.
+        //
+        // Note that the diag and depdb builtins can only appear at the
+        // beginning of the command line.
         //
         virtual optional<process_path>
         parse_program (token&, build2::script::token_type&,
                        bool first,
+                       bool env,
                        names&) override;
 
       protected:
diff --git a/libbuild2/script/parser.cxx b/libbuild2/script/parser.cxx
index a00651c..d5cff1a 100644
--- a/libbuild2/script/parser.cxx
+++ b/libbuild2/script/parser.cxx
@@ -97,7 +97,7 @@ namespace build2
     }
 
     optional<process_path> parser::
-    parse_program (token& t, type& tt, bool, names& ns)
+    parse_program (token& t, type& tt, bool, bool, names& ns)
     {
       parse_names (t, tt,
                    ns,
@@ -1019,6 +1019,18 @@ namespace build2
               }
             }
 
+            bool prog (p == pending::program_first ||
+                       p == pending::program_next);
+
+            // Check if this is the env pseudo-builtin.
+            //
+            bool env (false);
+            if (prog && tt == type::word && t.value == "env")
+            {
+              c.variables = parse_env_builtin (t, tt);
+              env = true;
+            }
+
             // Parse the next chunk as names to get expansion, etc. Note that
             // we do it in the chunking mode to detect whether anything in
             // each chunk is quoted. If we are waiting for the command
@@ -1033,10 +1045,10 @@ namespace build2
             //
             reset_quoted (t);
 
-            if (p == pending::program_first || p == pending::program_next)
+            if (prog)
             {
               optional<process_path> pp (
-                parse_program (t, tt, p == pending::program_first, ns));
+                parse_program (t, tt, p == pending::program_first, env, ns));
 
               // During pre-parsing we are not interested in the
               // parse_program() call result, so just discard the potentially
@@ -1088,7 +1100,7 @@ namespace build2
               {
                 diag_record dr (fail (l));
                 dr << "invalid string value ";
-                to_stream (dr.os, n, true); // Quote.
+                to_stream (dr.os, n, true /* quote */);
               }
 
               // If it is a quoted chunk, then we add the word as is.
@@ -1275,6 +1287,146 @@ namespace build2
       return make_pair (move (expr), move (hd));
     }
 
+    environment_vars parser::
+    parse_env_builtin (token& t, token_type& tt)
+    {
+      // enter: 'env' word token
+      // leave: first token of the program name
+
+      next (t, tt); // Skip 'env'.
+
+      // Note that the -u option and its value can belong to the different
+      // name chunks. That's why we parse the env builtin arguments in the
+      // chunking mode into the argument/location pair list up to the '--'
+      // separator and parse this list into the variable sets/unsets
+      // afterwords.
+      //
+      // Align the size with environment_vars (double because of -u <var>
+      // which is two arguments).
+      //
+      using args = small_vector<pair<string, location>, 4>;
+
+      args as;
+      names ns; // Reuse to reduce allocations.
+      while (tt != type::word || t.value != "--")
+      {
+        location l (get_location (t));
+
+        if (!start_names (tt))
+          fail (l) << "env: expected option, variable, or '--' separator "
+                   << "instead of " << t;
+
+        parse_names (t, tt,
+                     ns,
+                     pattern_mode::ignore,
+                     true /* chunk */,
+                     "env builtin argument",
+                     nullptr);
+
+        if (pre_parse_)
+          continue;
+
+        for (name& n: ns)
+        {
+          try
+          {
+            as.emplace_back (
+              value_traits<string>::convert (move (n), nullptr), l);
+          }
+          catch (const invalid_argument&)
+          {
+            diag_record dr (fail (l));
+            dr << "invalid string value ";
+            to_stream (dr.os, n, true /* quote */);
+          }
+        }
+
+        ns.clear ();
+      }
+
+      location l (get_location (t)); // '--' location.
+      next (t, tt);                  // Skip '--'.
+
+      if (tt == type::newline || tt == type::eos)
+        fail (t) << "env: expected program name instead of " << t;
+
+      // Parse the env builtin options and arguments.
+      //
+      environment_vars r;
+
+      // Note: args is empty in the pre-parse mode.
+      //
+      auto i (as.begin ()), e (as.end ());
+
+      // Parse the variable unsets (from options).
+      //
+      for (; i != e; ++i)
+      {
+        string& o (i->first);
+
+        // Bail out if the options and arguments separator is encountered.
+        //
+        if (o == "-")
+        {
+          ++i;
+          break;
+        }
+
+        // Unset the variable, adding its name to the resulting variable list.
+        //
+        auto unset = [&r, &i, this] (string&& v, const char* o)
+        {
+          if (v.empty ())
+            fail (i->second) << "env: empty value for option '" << o << "'";
+
+          if (v.find ('=') != string::npos)
+            fail (i->second) << "env: invalid value '" << v << "' for "
+                             << "option '" << o << "': contains '='";
+
+          r.push_back (move (v));
+        };
+
+        // If this is the --unset|-u option then add the variable unset and
+        // bail out to parsing the variable sets otherwise.
+        //
+        if (o == "--unset" || o == "-u")
+        {
+          if (++i == e)
+            fail (l) << "env: missing value for option '" << o << "'";
+
+          unset (move (i->first), o.c_str ());
+        }
+        else if (o.compare (0, 8, "--unset=") == 0)
+          unset (string (o, 8), "--unset");
+        else
+          break;
+      }
+
+      // Parse the variable sets (from arguments).
+      //
+      for (; i != e; ++i)
+      {
+        string& a (i->first);
+
+        // Validate the variable assignment.
+        //
+        size_t p (a.find ('='));
+
+        if (p == string::npos)
+          fail (i->second)
+            << "env: expected variable assignment instead of '" << a << "'";
+
+        if (p == 0)
+          fail (i->second) << "env: empty variable name";
+
+        // Add the variable set to the resulting list.
+        //
+        r.push_back (move (a));
+      }
+
+      return r;
+    }
+
     command_exit parser::
     parse_command_exit (token& t, type& tt)
     {
@@ -1310,7 +1462,7 @@ namespace build2
           diag_record dr;
 
           dr << fail (l) << "expected exit status instead of ";
-          to_stream (dr.os, ns, true); // Quote.
+          to_stream (dr.os, ns, true /* quote */);
 
           dr << info << "exit status is an unsigned integer less than 256";
         }
diff --git a/libbuild2/script/parser.hxx b/libbuild2/script/parser.hxx
index bec6867..da69591 100644
--- a/libbuild2/script/parser.hxx
+++ b/libbuild2/script/parser.hxx
@@ -129,6 +129,15 @@ namespace build2
       line_type
       pre_parse_line_start (token&, token_type&, lexer_mode);
 
+      // Parse the env pseudo-builtin arguments up to the program name. Return
+      // the list of the variables that should be unset ("name") and/or set
+      // ("name=value") in the command environment and the token/type that
+      // starts the program name. Note that the variable unsets come first, if
+      // present.
+      //
+      environment_vars
+      parse_env_builtin (token&, token_type&);
+
       // Execute.
       //
     protected:
@@ -166,7 +175,8 @@ namespace build2
       //
     protected:
       // Parse the command's leading name chunk. The argument first is true if
-      // this is the first command in the line.
+      // this is the first command in the line. The argument env is true if
+      // the command is executed via the env pseudo-builtin.
       //
       // During the execution phase try to parse and translate the leading
       // names into the process path and return nullopt if choose not to do
@@ -189,7 +199,7 @@ namespace build2
       // recognize and execute certain directives, or some such.
       //
       virtual optional<process_path>
-      parse_program (token&, token_type&, bool first, names&);
+      parse_program (token&, token_type&, bool first, bool env, names&);
 
       // Set lexer pointers for both the current and the base classes.
       //
diff --git a/libbuild2/script/run.cxx b/libbuild2/script/run.cxx
index 46c061c..b90ba48 100644
--- a/libbuild2/script/run.cxx
+++ b/libbuild2/script/run.cxx
@@ -1619,16 +1619,19 @@ namespace build2
                            ? process::path_search (args[0])
                            : process_path ());
 
-          // Note: the builtin-escaping character '^' is not printed.
+          // Note that CWD and builtin-escaping character '^' are not printed.
           //
+          process_env pe (resolve ? pp : c.program, c.variables);
+
           if (verb >= 2)
-            print_process (args);
+            print_process (pe, args);
 
           process pr (
-            resolve ? pp : c.program,
+            *pe.path,
             args.data (),
             {ifd.get (), -1}, process::pipe (ofd), {-1, efd.get ()},
-            env.work_dir.path->string ().c_str ());
+            env.work_dir.path->string ().c_str (),
+            pe.vars);
 
           ifd.reset ();
           ofd.out.reset ();
diff --git a/libbuild2/script/script.cxx b/libbuild2/script/script.cxx
index d0d3304..ee238cc 100644
--- a/libbuild2/script/script.cxx
+++ b/libbuild2/script/script.cxx
@@ -229,22 +229,57 @@ namespace build2
       }
     }
 
-    // Quote if empty or contains spaces or any of the special characters.
-    // Note that we use single quotes since double quotes still allow
-    // expansion.
+    // Quote a string unconditionally, assuming it contains some special
+    // characters.
     //
-    // @@ What if it contains single quotes?
+    // If the quote character is present in the string then it is double
+    // quoted rather than single quoted. In this case the following characters
+    // are escaped:
+    //
+    // \"
+    //
+    static void
+    to_stream_quoted (ostream& o, const char* s)
+    {
+      if (strchr (s, '\'') != nullptr)
+      {
+        o << '"';
+
+        for (; *s != '\0'; ++s)
+        {
+          // Escape characters special inside double quotes.
+          //
+          if (strchr ("\\\"", *s) != nullptr)
+            o << '\\';
+
+          o << *s;
+        }
+
+        o << '"';
+      }
+      else
+        o << '\'' << s << '\'';
+    }
+
+    static inline void
+    to_stream_quoted (ostream& o, const string& s)
+    {
+      to_stream_quoted (o, s.c_str ());
+    }
+
+    // Quote if empty or contains spaces or any of the command line special
+    // characters.
     //
     static void
     to_stream_q (ostream& o, const string& s)
     {
       // NOTE: update dump(line) if adding any new special character.
       //
-      if (s.empty () || s.find_first_of (" |&<>=\\\"") != string::npos)
-        o << '\'' << s << '\'';
+      if (s.empty () || s.find_first_of (" |&<>=\\\"'") != string::npos)
+        to_stream_quoted (o, s);
       else
         o << s;
-    };
+    }
 
     void
     to_stream (ostream& o, const command& c, command_to_stream m)
@@ -373,6 +408,87 @@ namespace build2
 
       if ((m & command_to_stream::header) == command_to_stream::header)
       {
+        // Print the env builtin arguments, if any environment variable
+        // (un)sets are present.
+        //
+        if (!c.variables.empty ())
+        {
+          o << "env";
+
+          auto b (c.variables.begin ()), i (b), e (c.variables.end ());
+
+          // Print a variable name or assignment to the stream, quoting it if
+          // necessary.
+          //
+          auto print = [&o] (const string& v, bool name)
+          {
+            size_t p (v.find_first_of (" \\\"'"));
+
+            // Print the variable name/assignment as is if it doesn't contain
+            // any special characters.
+            //
+            if (p == string::npos)
+            {
+              o << v;
+              return;
+            }
+
+            // If the variable name contains any special characters, then
+            // quote the name/assignment as a whole.
+            //
+            size_t eq;
+            if (name || (eq = v.find ('=')) > p)
+            {
+              to_stream_quoted (o, v);
+              return;
+            }
+
+            // Finally, if the variable value contains any special characters,
+            // then we quote only the value.
+            //
+            assert (eq != string::npos);
+
+            o.write (v.c_str (), eq + 1);              // Includes '='.
+            to_stream_quoted (o, v.c_str () + eq + 1);
+          };
+
+          // Variable unsets.
+          //
+          // Print the variable unsets as the -u options until a variable set
+          // is encountered (contains '=') or the end of the variable list is
+          // reached. In the former case, to avoid a potential ambiguity add
+          // the '-' separator, if there are any options.
+          //
+          // Note that we rely on the fact that unsets come first, which is
+          // guaranteed by parser::parse_env_builtin().
+          //
+          for (; i != e; ++i)
+          {
+            const string& v (*i);
+
+            if (v.find ('=') == string::npos) // Variable unset.
+            {
+              o << " -u "; print (v, true /* name*/);
+            }
+            else                              // Variable set.
+            {
+              if (i != b)
+                o << " -";
+
+              break;
+            }
+          }
+
+          // Variable sets.
+          //
+          for (; i != e; ++i)
+          {
+            o << ' '; print (*i, false /* name */);
+          }
+
+          o << " -- ";
+        }
+
         // Program.
         //
         to_stream_q (o, c.program.recall_string ());
diff --git a/libbuild2/script/script.hxx b/libbuild2/script/script.hxx
index d751169..8e1c852 100644
--- a/libbuild2/script/script.hxx
+++ b/libbuild2/script/script.hxx
@@ -298,6 +298,11 @@ namespace build2
 
     // command
     //
+    // Align with butl::process_env, assuming it is not very common to (un)set
+    // more than two variables.
+    //
+    using environment_vars = small_vector<string, 2>;
+
     struct command
     {
       // We use NULL initial as an indication that the path stored in recall
@@ -306,7 +311,8 @@ namespace build2
       //
       process_path program;
 
-      strings arguments;
+      strings          arguments;
+      environment_vars variables;
 
       optional<redirect> in;
       optional<redirect> out;
diff --git a/libbuild2/test/script/parser+env.test.testscript b/libbuild2/test/script/parser+env.test.testscript
new file mode 100644
index 0000000..b1e864c
--- /dev/null
+++ b/libbuild2/test/script/parser+env.test.testscript
@@ -0,0 +1,77 @@
+# file      : libbuild2/test/script/parser+env.test.testscript
+# license   : MIT; see accompanying LICENSE file
+
+: unset
+:
+{
+  $* <'env -u a -- cmd'      >'env -u a -- cmd'      : short-opt
+  $* <'env --unset a -- cmd' >'env -u a -- cmd'      : long-opt
+  $* <'env --unset=a -- cmd' >'env -u a -- cmd'      : long-opt-eq
+  $* <'env -u a -u b -- cmd' >'env -u a -u b -- cmd' : mult-opt
+  $* <'env -u "a b" -- cmd'  >"env -u 'a b' -- cmd"  : quote
+
+  : invalid-opt
+  :
+  $* <'env -w a -- cmd' 2>>EOE != 0
+    testscript:1:5: error: env: expected variable assignment instead of '-w'
+    EOE
+
+  : no-val
+  :
+  $* <'env -u -- cmd' 2>>EOE != 0
+    testscript:1:8: error: env: missing value for option '-u'
+    EOE
+
+  : empty-val
+  :
+  $* <'env --unset= -- cmd' 2>>EOE != 0
+    testscript:1:5: error: env: empty value for option '--unset'
+    EOE
+
+  : invalid-val
+  :
+  $* <'env --unset=a=b -- cmd' 2>>EOE != 0
+    testscript:1:5: error: env: invalid value 'a=b' for option '--unset': contains '='
+    EOE
+
+  : no-sep
+  :
+  $* <'env -u a cmd' 2>>EOE != 0
+    testscript:1:13: error: env: expected option, variable, or '--' separator instead of <newline>
+    EOE
+
+  $* <'env && cmd' 2>>EOE != 0
+    testscript:1:5: error: env: expected option, variable, or '--' separator instead of '&&'
+    EOE
+}
+
+: set
+:
+{
+  $* <'env a=b -- cmd'      >'env a=b -- cmd'        : var
+  $* <'env -u a b=c -- cmd' >'env -u a - b=c -- cmd' : opt-var
+  $* <'env a="b c" -- cmd'  >"env a='b c' -- cmd"    : quote
+  $* <'env "a b"=c -- cmd'  >"env 'a b=c' -- cmd"    : quote-name
+
+  : double-quote
+  :
+  $* <<EOF >>EOF
+    env a="'a\"'" -- cmd
+    EOF
+
+  : expected-assign
+  :
+  $* <'env a -- cmd' 2>>EOE != 0
+    testscript:1:5: error: env: expected variable assignment instead of 'a'
+    EOE
+}
+
+: non-first
+:
+$* <'cmd1 && env -u a b=c -- cmd2' >'cmd1 && env -u a - b=c -- cmd2'
+
+: no-cmd
+:
+$* <'env -u a --' 2>>EOE != 0
+  testscript:1:12: error: env: expected program name instead of <newline>
+  EOE