Invent quoting modes for to_stream(name)

10 files changed, 86 insertions, 35 deletions

diff --git a/libbuild2/adhoc-rule-buildscript.cxx b/libbuild2/adhoc-rule-buildscript.cxx
index c64dbfb..7174296 100644
--- a/libbuild2/adhoc-rule-buildscript.cxx
+++ b/libbuild2/adhoc-rule-buildscript.cxx
@@ -183,7 +183,7 @@ namespace build2
     {
       os << " [";
       os << "diag=";
-      to_stream (os, name (*script.diag_name), true /* quote */, '@');
+      to_stream (os, name (*script.diag_name), quote_mode::normal, '@');
       os << ']';
     }
   }
diff --git a/libbuild2/build/script/parser.cxx b/libbuild2/build/script/parser.cxx
index dd6fa2d..cba4b88 100644
--- a/libbuild2/build/script/parser.cxx
+++ b/libbuild2/build/script/parser.cxx
@@ -1439,7 +1439,7 @@ namespace build2
               {
                 diag_record dr (fail (l));
                 dr << "depdb dyndep: invalid string value ";
-                to_stream (dr.os, n, true /* quote */);
+                to_stream (dr.os, n, quote_mode::normal);
               }
             }
           }
diff --git a/libbuild2/config/operation.cxx b/libbuild2/config/operation.cxx
index 8ceb4d4..2fb0423 100644
--- a/libbuild2/config/operation.cxx
+++ b/libbuild2/config/operation.cxx
@@ -42,7 +42,7 @@ namespace build2
         ofs << "# Created automatically by the config module." << endl
             << "#" << endl
             << "src_root = ";
-        to_stream (ofs, name (src_root), true /* quote */, '@');
+        to_stream (ofs, name (src_root), quote_mode::normal, '@');
         ofs << endl;
 
         ofs.close ();
@@ -71,7 +71,7 @@ namespace build2
         ofs << "# Created automatically by the config module." << endl
             << "#" << endl
             << "out_root = ";
-        to_stream (ofs, name (out_root), true /* quote */, '@');
+        to_stream (ofs, name (out_root), quote_mode::normal, '@');
         ofs << endl;
 
         ofs.close ();
@@ -539,7 +539,7 @@ namespace build2
             if (!p.first.empty ())
             {
               os << ' ';
-              to_stream (os, p.first, true /* quote */, '@');
+              to_stream (os, p.first, quote_mode::normal, '@');
             }
 
             os << endl;
diff --git a/libbuild2/functions-builtin.cxx b/libbuild2/functions-builtin.cxx
index c013c3b..93f2d0f 100644
--- a/libbuild2/functions-builtin.cxx
+++ b/libbuild2/functions-builtin.cxx
@@ -94,7 +94,7 @@ namespace build2
       ostringstream os;
       to_stream (os,
                  v->as<names> (),
-                 true /* quote */,
+                 quote_mode::normal,
                  '@'  /* pair */,
                  escape && convert<bool> (move (*escape)));
       return os.str ();
diff --git a/libbuild2/name.cxx b/libbuild2/name.cxx
index 1081b5c..a285459 100644
--- a/libbuild2/name.cxx
+++ b/libbuild2/name.cxx
@@ -80,15 +80,21 @@ namespace build2
   }
 
   ostream&
-  to_stream (ostream& os, const name& n, bool quote, char pair, bool escape)
+  to_stream (ostream& os, const name& n, quote_mode q, char pair, bool escape)
   {
     using pattern_type = name::pattern_type;
 
-    auto write_string = [&os, quote, pair, escape] (
+    auto write_string = [&os, q, pair, escape] (
       const string& v,
       optional<pattern_type> pat = nullopt,
       bool curly = false)
     {
+      // We don't expect the effective quoting mode to be specified for the
+      // name patterns or names that need curly braces in their
+      // representations.
+      //
+      assert (q != quote_mode::effective || (!pat && !curly));
+
       // Special characters, path pattern characters, and regex pattern
       // characters. The latter only need to be quoted in the first position
       // and if followed by a non-alphanumeric delimiter. If that's the only
@@ -97,7 +103,7 @@ namespace build2
       // escape leading `+` in the curly braces which is also recognized as a
       // path pattern.
       //
-      char sc[] = {
+      char nsc[] = {
         '{', '}', '[', ']', '$', '(', ')', // Token endings.
         ' ', '\t', '\n', '#',              // Spaces.
         '\\', '"',                         // Escaping and quoting.
@@ -114,6 +120,26 @@ namespace build2
         return (v[0] == '~' || v[0] == '^') && v[1] != '\0' && !alnum (v[1]);
       };
 
+      char esc[] = {
+        '$', '(',             // Token endings.
+        ' ', '\t', '\n', '#', // Spaces.
+        '"',                  // Quoting.
+        pair,                 // Pair separator, if any.
+        '\0'};
+
+      auto ec = [&esc] (const string& v)
+      {
+        for (size_t i (0); i < v.size (); ++i)
+        {
+          char c (v[i]);
+
+          if (strchr (esc, c) != nullptr || (c == '\\' && v[i + 1] == '\\'))
+            return true;
+        }
+
+        return false;
+      };
+
       if (pat)
       {
         switch (*pat)
@@ -124,7 +150,7 @@ namespace build2
         }
       }
 
-      if (quote && v.find ('\'') != string::npos)
+      if (q != quote_mode::none && v.find ('\'') != string::npos)
       {
         // Quote the string with the double quotes rather than with the single
         // one. Escape some of the special characters.
@@ -148,8 +174,10 @@ namespace build2
       // pattern character but not vice-verse. See the parsing logic for
       // details.
       //
-      else if (quote && (v.find_first_of (sc) != string::npos ||
-                         (!pat && v.find_first_of (pc) != string::npos)))
+      else if ((q == quote_mode::normal &&
+                (v.find_first_of (nsc) != string::npos ||
+                 (!pat && v.find_first_of (pc) != string::npos))) ||
+               (q == quote_mode::effective && ec (v)))
       {
         if (escape) os << '\\';
         os << '\'';
@@ -164,8 +192,9 @@ namespace build2
       // details). So we escape it both if it's not a pattern or is a path
       // pattern.
       //
-      else if (quote && ((!pat || *pat == pattern_type::path) &&
-                         ((v[0] == '+' && curly) || rc (v))))
+      else if (q == quote_mode::normal              &&
+               (!pat || *pat == pattern_type::path) &&
+               ((v[0] == '+' && curly) || rc (v)))
       {
         if (escape) os << '\\';
         os << '\\' << v;
@@ -176,12 +205,12 @@ namespace build2
 
     uint16_t dv (stream_verb (os).path); // Directory verbosity.
 
-    auto write_dir = [&os, quote, &write_string, dv] (
+    auto write_dir = [&os, q, &write_string, dv] (
       const dir_path& d,
       optional<pattern_type> pat = nullopt,
       bool curly = false)
     {
-      if (quote)
+      if (q != quote_mode::none)
         write_string (dv < 1 ? diag_relative (d) : d.representation (),
                       pat,
                       curly);
@@ -194,7 +223,7 @@ namespace build2
 
     // If quoted then print empty name as '' rather than {}.
     //
-    if (quote && n.empty ())
+    if (q != quote_mode::none && n.empty ())
       return os << (escape ? "\\'\\'" : "''");
 
     if (n.proj)
@@ -255,7 +284,7 @@ namespace build2
   ostream&
   to_stream (ostream& os,
              const names_view& ns,
-             bool quote,
+             quote_mode q,
              char pair,
              bool escape)
   {
@@ -263,7 +292,7 @@ namespace build2
     {
       const name& n (*i);
       ++i;
-      to_stream (os, n, quote, pair, escape);
+      to_stream (os, n, q, pair, escape);
 
       if (n.pair)
         os << n.pair;
diff --git a/libbuild2/name.hxx b/libbuild2/name.hxx
index 1dd5a9f..795cd5c 100644
--- a/libbuild2/name.hxx
+++ b/libbuild2/name.hxx
@@ -184,8 +184,8 @@ namespace build2
   to_name (string);
 
   // Serialize the name to the stream. If requested, the name components
-  // containing special characters are quoted and/or escaped. The special
-  // characters are:
+  // containing special characters are quoted and/or escaped. In the normal
+  // quoting mode the special characters are:
   //
   // {}[]$() \t\n#\"'%
   //
@@ -199,8 +199,14 @@ namespace build2
   //
   // As well as leading `+` if in the curly braces.
   //
+  // In the effective quoting mode the special characters are:
+  //
+  // $( \t\n#"'
+  //
+  // As well as `\` if followed by any of the above characters or itself.
+  //
   // If the pair argument is not '\0', then it is added to the above special
-  // characters set. If the quote character is present in the component then
+  // characters sets. If the quote character is present in the component then
   // it is double quoted rather than single quoted. In this case the following
   // characters are escaped:
   //
@@ -213,15 +219,23 @@ namespace build2
   // Note that in the quoted mode empty unqualified name is printed as '',
   // not {}.
   //
+  enum class quote_mode
+  {
+    none,
+    normal,
+    effective
+  };
+
   LIBBUILD2_SYMEXPORT ostream&
   to_stream (ostream&,
              const name&,
-             bool quote,
+             quote_mode,
              char pair = '\0',
              bool escape = false);
 
   inline ostream&
-  operator<< (ostream& os, const name& n) {return to_stream (os, n, false);}
+  operator<< (ostream& os, const name& n) {
+    return to_stream (os, n, quote_mode::none);}
 
   // Vector of names.
   //
@@ -240,13 +254,13 @@ namespace build2
   LIBBUILD2_SYMEXPORT ostream&
   to_stream (ostream&,
              const names_view&,
-             bool quote,
+             quote_mode,
              char pair = '\0',
              bool escape = false);
 
   inline ostream&
   operator<< (ostream& os, const names_view& ns) {
-    return to_stream (os, ns, false);}
+    return to_stream (os, ns, quote_mode::none);}
 
   inline ostream&
   operator<< (ostream& os, const names& ns) {return os << names_view (ns);}
diff --git a/libbuild2/name.test.cxx b/libbuild2/name.test.cxx
index 80b830e..c404503 100644
--- a/libbuild2/name.test.cxx
+++ b/libbuild2/name.test.cxx
@@ -46,7 +46,7 @@ namespace build2
     // Test stream representation.
     //
     {
-      auto ts = [] (const name& n, bool quote = true)
+      auto ts = [] (const name& n, quote_mode quote = quote_mode::normal)
       {
         ostringstream os;
         stream_verb (os, stream_verbosity (0, 1));
@@ -54,8 +54,8 @@ namespace build2
         return os.str ();
       };
 
-      assert (ts (name ())        == "''");
-      assert (ts (name (), false) == "{}");
+      assert (ts (name ())                   == "''");
+      assert (ts (name (), quote_mode::none) == "{}");
 
       assert (ts (name ("foo")) == "foo");
 
@@ -70,10 +70,18 @@ namespace build2
       assert (ts (name (dir ("bar/"),     "dir", "foo")) == "bar/dir{foo}");
       assert (ts (name (dir ("bar/baz/"), "dir", "foo")) == "bar/baz/dir{foo}");
 
-      // Quoting.
+      // Normal quoting.
       //
       assert (ts (name (dir ("bar baz/"), "dir", "foo fox")) == "'bar baz/'dir{'foo fox'}");
 
+      // Effective quoting.
+      //
+      assert (ts (name ("bar\\baz"),   quote_mode::effective) == "bar\\baz");
+      assert (ts (name ("bar[baz]"),   quote_mode::effective) == "bar[baz]");
+      assert (ts (name ("bar$baz"),    quote_mode::effective) == "'bar$baz'");
+      assert (ts (name ("bar\\\\baz"), quote_mode::effective) == "'bar\\\\baz'");
+      assert (ts (name ("bar\\$baz"),  quote_mode::effective) == "'bar\\$baz'");
+
       // Relative logic.
       //
 #ifndef _WIN32
diff --git a/libbuild2/parser.cxx b/libbuild2/parser.cxx
index 9fbcd2b..bec3230 100644
--- a/libbuild2/parser.cxx
+++ b/libbuild2/parser.cxx
@@ -42,7 +42,7 @@ namespace build2
     {
       o << '=';
       names storage;
-      to_stream (o, reverse (a.value, storage), true /* quote */, '@');
+      to_stream (o, reverse (a.value, storage), quote_mode::normal, '@');
     }
 
     return o;
diff --git a/libbuild2/script/parser.cxx b/libbuild2/script/parser.cxx
index 7722002..82eb9c8 100644
--- a/libbuild2/script/parser.cxx
+++ b/libbuild2/script/parser.cxx
@@ -1115,7 +1115,7 @@ namespace build2
               {
                 diag_record dr (fail (l));
                 dr << "invalid string value ";
-                to_stream (dr.os, n, true /* quote */);
+                to_stream (dr.os, n, quote_mode::normal);
               }
 
               // If it is a quoted chunk, then we add the word as is.
@@ -1351,7 +1351,7 @@ namespace build2
           {
             diag_record dr (fail (l));
             dr << "invalid string value ";
-            to_stream (dr.os, n, true /* quote */);
+            to_stream (dr.os, n, quote_mode::normal);
           }
         }
 
@@ -1537,7 +1537,7 @@ 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, quote_mode::normal);
 
           dr << info << "exit status is an unsigned integer less than 256";
         }
diff --git a/libbuild2/test/script/parser.cxx b/libbuild2/test/script/parser.cxx
index 9e92f3b..bf04a30 100644
--- a/libbuild2/test/script/parser.cxx
+++ b/libbuild2/test/script/parser.cxx
@@ -1057,7 +1057,7 @@ namespace build2
 
           diag_record dr (fail (dl));
           dr << "invalid testscript include path ";
-          to_stream (dr.os, n, true); // Quote.
+          to_stream (dr.os, n, quote_mode::normal);
         }
       }