Further improved parsing of query expressions

Fixes #210 / D4C2DD6F-8967-4FFC-BBBC-A941F9C53475

src/query.cc

@@ -53,25 +53,9 @@ query_t::lexer_t::token_t query_t::lexer_t::next_token()
     }
   }
 
-  if (consume_next_arg) {
-    consume_next_arg = false;
-    token_t tok(token_t::TERM, string(arg_i, arg_end));
-    arg_i = arg_end;
-    return tok;
-  }
-
- resume:
-  bool consume_next = false;
   switch (*arg_i) {
-  case ' ':
-  case '\t':
-  case '\r':
-  case '\n':
-    if (++arg_i == arg_end)
-      return next_token();
-  goto resume;
-
   case '\'':
+  case '"':
   case '/': {
     string pat;
     char   closing	 = *arg_i;
@@ -95,6 +79,25 @@ query_t::lexer_t::token_t query_t::lexer_t::next_token()
 
     return token_t(token_t::TERM, pat);
   }
+  }
+
+  if (multiple_args && consume_next_arg) {
+    consume_next_arg = false;
+    token_t tok(token_t::TERM, string(arg_i, arg_end));
+    arg_i = arg_end;
+    return tok;
+  }
+
+ resume:
+  bool consume_next = false;
+  switch (*arg_i) {
+  case ' ':
+  case '\t':
+  case '\r':
+  case '\n':
+    if (++arg_i == arg_end)
+      return next_token();
+  goto resume;
 
   case '(': ++arg_i; return token_t(token_t::LPAREN);
   case ')': ++arg_i; return token_t(token_t::RPAREN);
@@ -104,7 +107,10 @@ query_t::lexer_t::token_t query_t::lexer_t::next_token()
   case '@': ++arg_i; return token_t(token_t::TOK_PAYEE);
   case '#': ++arg_i; return token_t(token_t::TOK_CODE);
   case '%': ++arg_i; return token_t(token_t::TOK_META);
-  case '=': ++arg_i; return token_t(token_t::TOK_EQ);
+  case '=':
+    ++arg_i;
+    consume_next_arg = true;
+    return token_t(token_t::TOK_EQ);
 
   case '\\':
     consume_next = true;

src/query.h

@@ -62,6 +62,7 @@ public:
 
     bool consume_whitespace;
     bool consume_next_arg;
+    bool multiple_args;
 
   public:
     struct token_t
@@ -177,10 +178,11 @@ public:
     token_t token_cache;
 
     lexer_t(value_t::sequence_t::const_iterator _begin,
-		  value_t::sequence_t::const_iterator _end)
+	    value_t::sequence_t::const_iterator _end,
+	    bool _multiple_args = true)
       : begin(_begin), end(_end),
-	consume_whitespace(false),
+	consume_whitespace(false), consume_next_arg(false),
-	consume_next_arg(false)
+	multiple_args(_multiple_args)
     {
       TRACE_CTOR(query_t::lexer_t, "");
       assert(begin != end);
@@ -192,6 +194,7 @@ public:
 	arg_i(lexer.arg_i), arg_end(lexer.arg_end),
 	consume_whitespace(lexer.consume_whitespace),
 	consume_next_arg(lexer.consume_next_arg),
+	multiple_args(lexer.multiple_args),
 	token_cache(lexer.token_cache)
     {
       TRACE_CTOR(query_t::lexer_t, "copy");
@@ -227,8 +230,8 @@ protected:
     expr_t::ptr_op_t parse_query_expr(lexer_t::token_t::kind_t tok_context);
 
   public:
-    parser_t(const value_t& _args)
+    parser_t(const value_t& _args, bool multiple_args = true)
-      : args(_args), lexer(args.begin(), args.end()) {
+      : args(_args), lexer(args.begin(), args.end(), multiple_args) {
       TRACE_CTOR(query_t::parser_t, "");
     }
     parser_t(const parser_t& parser)
@@ -261,28 +264,30 @@ public:
     TRACE_CTOR(query_t, "copy");
   }
   query_t(const string& arg,
-	  const keep_details_t& _what_to_keep = keep_details_t())
+	  const keep_details_t& _what_to_keep = keep_details_t(),
+	  bool multiple_args = true)
     : predicate_t(_what_to_keep) {
     TRACE_CTOR(query_t, "string, keep_details_t");
     if (! arg.empty()) {
       value_t temp(string_value(arg));
-      parse_args(temp.to_sequence());
+      parse_args(temp.to_sequence(), multiple_args);
     }
   }
   query_t(const value_t& args,
-	  const keep_details_t& _what_to_keep = keep_details_t())
+	  const keep_details_t& _what_to_keep = keep_details_t(),
+	  bool multiple_args = true)
     : predicate_t(_what_to_keep) {
     TRACE_CTOR(query_t, "value_t, keep_details_t");
     if (! args.empty())
-      parse_args(args);
+      parse_args(args, multiple_args);
   }
   virtual ~query_t() {
     TRACE_DTOR(query_t);
   }
 
-  void parse_args(const value_t& args) {
+  void parse_args(const value_t& args, bool multiple_args = true) {
     if (! parser)
-      parser = parser_t(args);
+      parser = parser_t(args, multiple_args);
     ptr = parser->parse();	// expr_t::ptr
   }
 

src/textual.cc

@@ -526,7 +526,7 @@ void instance_t::automated_xact_directive(char * line)
 
   std::auto_ptr<auto_xact_t> ae
     (new auto_xact_t(query_t(string(skip_ws(line + 1)),
-			     keep_details_t(true, true, true))));
+			     keep_details_t(true, true, true), false)));
   ae->pos	    = position_t();
   ae->pos->pathname = pathname;
   ae->pos->beg_pos  = line_beg_pos;

test/unit/t_expr.cc

@@ -158,8 +158,6 @@ void ValueExprTestCase::testPredicateTokenizer7()
 
   assertEqual(query_t::lexer_t::token_t::TOK_EQ, tokens.next_token().kind);
   assertEqual(query_t::lexer_t::token_t::TERM, tokens.next_token().kind);
-  assertEqual(query_t::lexer_t::token_t::TOK_AND, tokens.next_token().kind);
-  assertEqual(query_t::lexer_t::token_t::TERM, tokens.next_token().kind);
   assertEqual(query_t::lexer_t::token_t::END_REACHED, tokens.next_token().kind);
 #endif
 }
@@ -167,7 +165,7 @@ void ValueExprTestCase::testPredicateTokenizer7()
 void ValueExprTestCase::testPredicateTokenizer8()
 {
   value_t args;
-  args.push_back(string_value("expr foo and bar"));
+  args.push_back(string_value("expr 'foo and bar'"));
 
 #ifndef NOT_FOR_PYTHON
   query_t::lexer_t tokens(args.begin(), args.end());
@@ -182,7 +180,7 @@ void ValueExprTestCase::testPredicateTokenizer9()
 {
   value_t args;
   args.push_back(string_value("expr"));
-  args.push_back(string_value("foo and bar"));
+  args.push_back(string_value("'foo and bar'"));
 
 #ifndef NOT_FOR_PYTHON
   query_t::lexer_t tokens(args.begin(), args.end());