feat: Extend parsing to field impressions (#1154)

feat: Extend parsing to field impressions.

Add search_parser_test to PR tests.

Signed-off-by: Roman Gershman <roman@dragonflydb.io>

src/core/search/CMakeLists.txt

@@ -1,5 +1,3 @@
-# apt install libfl-dev
-
 gen_flex(lexer)
 gen_bison(parser)
 
@@ -7,4 +5,4 @@ cur_gen_dir(gen_dir)
 
 add_library(query_parser query_driver.cc ${gen_dir}/parser.cc ${gen_dir}/lexer.cc)
 target_link_libraries(query_parser base absl::strings TRDP::reflex)
-cxx_test(search_parser_test query_parser)
+cxx_test(search_parser_test query_parser LABELS DFLY)

src/core/search/lexer.lex

@@ -53,9 +53,11 @@ term_char [_]|\w
 "("            return Parser::make_LPAREN (loc());
 ")"            return Parser::make_RPAREN (loc());
 "*"            return Parser::make_STAR (loc());
-"~"            return Parser::make_NOT_OP (loc());
+"-"            return Parser::make_NOT_OP (loc());
 ":"            return Parser::make_COLON (loc());
 "=>"           return Parser::make_ARROW (loc());
+"["            return Parser::make_LBRACKET (loc());
+"]"            return Parser::make_RBRACKET (loc());
 
 -?[0-9]+       return make_INT64(matched_view(), loc());
 

src/core/search/parser.y

@@ -46,14 +46,18 @@ using namespace std;
   STAR    "*"
   ARROW   "=>"
   COLON   ":"
-  NOT_OP  "~"
+  LBRACKET "["
+  RBRACKET "]"
 ;
 
-%token YYEOF
+%precedence NOT_OP
+
+// Needed 0 at the end to satisfy bison 3.5.1
+%token YYEOF 0
 %token <std::string> TERM "term" PARAM "param" FIELD "field"
 
 %token <int64_t> INT64 "int64"
-%nterm <AstExpr> search_expr
+%nterm <AstExpr> search_expr field_filter field_cond range_value term_list opt_neg_term
 
 %printer { yyo << $$; } <*>;
 
@@ -64,9 +68,25 @@ query:
   | query search_expr
   ;
 
-search_expr: TERM {
+
-  cout << $1 << endl;
+search_expr:
-}
+ LPAREN search_expr RPAREN { $$ = $2; }
+ | NOT_OP search_expr { $$ = AstExpr{}; };
+ | TERM { }
+ | field_filter;
+
+field_filter:
+   FIELD COLON field_cond { $$ = AstExpr{}; }
+
+field_cond: term_list | range_value
+ range_value: LBRACKET INT64 INT64 RBRACKET { $$ = AstExpr{}; }
+
+term_list:
+  opt_neg_term |
+  LPAREN term_list opt_neg_term RPAREN { };
+
+opt_neg_term:
+  TERM { } | NOT_OP TERM { $$ = AstExpr{}; };
 
 %%
 

src/core/search/search_parser_test.cc

@@ -47,6 +47,16 @@ class SearchParserTest : public ::testing::Test {
     auto tok = Lex();                                   \
     ASSERT_EQ(tok.type_get(), Parser::token::tok_enum); \
   }
+#define NEXT_ERROR()                          \
+  {                                           \
+    bool caught = false;                      \
+    try {                                     \
+      auto tok = Lex();                       \
+    } catch (const Parser::syntax_error& e) { \
+      caught = true;                          \
+    }                                         \
+    ASSERT_TRUE(caught);                      \
+  }
 
 TEST_F(SearchParserTest, Scanner) {
   SetInput("ab cd");
@@ -76,10 +86,18 @@ TEST_F(SearchParserTest, Scanner) {
   SetInput("почтальон Печкин");
   NEXT_EQ(TOK_TERM, string, "почтальон");
   NEXT_EQ(TOK_TERM, string, "Печкин");
+
+  SetInput("18446744073709551616");
+  NEXT_ERROR();
 }
 
 TEST_F(SearchParserTest, Parse) {
-  EXPECT_EQ(0, Parse(" foo "));
+  EXPECT_EQ(0, Parse(" foo bar (baz) "));
+  EXPECT_EQ(0, Parse(" -(foo) @foo:bar @ss:[1 2]"));
+  EXPECT_EQ(1, Parse(" -(foo "));
+  EXPECT_EQ(1, Parse(" foo:bar "));
+  EXPECT_EQ(1, Parse(" @foo:@bar "));
+  EXPECT_EQ(1, Parse(" @foo: "));
 }
 
 }  // namespace search