lex: change treatment of symbols which are neither all-uppercase or all-lowercase

4 files changed, 83 insertions, 142 deletions

diff --git a/src/lex.cpp b/src/lex.cpp
index 3c86f25..29bdd4e 100644
--- a/src/lex.cpp
+++ b/src/lex.cpp
@@ -241,17 +241,17 @@ int lex_t::lex_symbol(parse_token_value_t *value) {
 	if (needspace)
 		return syntax_error(value);
 
-	bool uc = true;
-	bool lc = true;
+	bool uc = false;
+	bool lc = false;
 
 	do {
 		switch (current()) {
 		case 'A' ... 'Z':
-			lc = false;
+			uc = true;
 			continue;
 
 		case 'a' ... 'z':
-			uc = false;
+			lc = true;
 			continue;
 
 		case '0' ... '9':
@@ -264,10 +264,8 @@ int lex_t::lex_symbol(parse_token_value_t *value) {
 
 	value->str = get_token();
 
-	if (uc)
+	if (uc && !lc)
 		return TOK_SYMBOL_UC;
-	else if (lc)
-		return TOK_SYMBOL_LC;
 	else
 		return TOK_SYMBOL;
 }
@@ -420,6 +418,7 @@ int lex_t::lex(parse_token_value_t *value) {
 
 		case 'a' ... 'z':
 		case 'A' ... 'Z':
+		case '_':
 			return lex_symbol(value);
 
 		case '%':
diff --git a/src/parse.cpp b/src/parse.cpp
index fe3a026..9a24c03 100644
--- a/src/parse.cpp
+++ b/src/parse.cpp
@@ -141,8 +141,6 @@ static inline std::string * parse_reduce_23(std::string *v, __attribute__((unuse
 
 static inline std::string * parse_reduce_24(std::string *v, __attribute__((unused)) solar::grammar_t *grammar) {return v;}
 
-static inline std::string * parse_reduce_25(std::string *v, __attribute__((unused)) solar::grammar_t *grammar) {return v;}
-
 
 static int parse_do_push(parse_context_t *parser, int token, __attribute__((unused)) solar::grammar_t *grammar) {
 	parse_symbol_value_t result;
@@ -161,7 +159,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			case 0:
 				return 0;
 
-			case TOK_SYMBOL_LC:
+			case TOK_SYMBOL:
 				parser->stack[++parser->top].state = 4;
 				return 1;
 
@@ -378,7 +376,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 				parser->stack[++parser->top].state = 30;
 				return 1;
 
-			case TOK_SYMBOL_LC:
+			case TOK_SYMBOL:
 				parser->stack[++parser->top].state = 31;
 				return 1;
 
@@ -489,8 +487,8 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 					parser->stack[++parser->top].state = 25;
 					break;
 
-				case 69:
-					parser->stack[++parser->top].state = 71;
+				case 68:
+					parser->stack[++parser->top].state = 70;
 					break;
 
 				}
@@ -515,7 +513,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 				parser->stack[parser->top].value.symbol_term = result.symbol_term;
 				switch (parser->stack[parser->top].state) {
 				case 18:
-				case 69:
+				case 68:
 					parser->stack[++parser->top].state = 26;
 					break;
 
@@ -561,8 +559,8 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 					parser->stack[++parser->top].state = 25;
 					break;
 
-				case 69:
-					parser->stack[++parser->top].state = 71;
+				case 68:
+					parser->stack[++parser->top].state = 70;
 					break;
 
 				}
@@ -578,7 +576,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 				parser->stack[parser->top].value.symbol_term = result.symbol_term;
 				switch (parser->stack[parser->top].state) {
 				case 18:
-				case 69:
+				case 68:
 					parser->stack[++parser->top].state = 26;
 					break;
 
@@ -654,7 +652,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 				parser->stack[++parser->top].state = 32;
 				return 1;
 
-			case TOK_SYMBOL_LC:
+			case TOK_SYMBOL:
 				parser->stack[++parser->top].state = 45;
 				return 1;
 
@@ -669,16 +667,12 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 				parser->stack[++parser->top].state = 47;
 				return 1;
 
-			case TOK_SYMBOL_LC:
-				parser->stack[++parser->top].state = 48;
-				return 1;
-
 			case TOK_SYMBOL_UC:
-				parser->stack[++parser->top].state = 49;
+				parser->stack[++parser->top].state = 48;
 				return 1;
 
 			case '=':
-				parser->stack[++parser->top].state = 50;
+				parser->stack[++parser->top].state = 49;
 				return 1;
 
 			default:
@@ -689,7 +683,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 		case 40:
 			switch (token) {
 			case 'u':
-				parser->stack[++parser->top].state = 51;
+				parser->stack[++parser->top].state = 50;
 				return 1;
 
 			default:
@@ -700,7 +694,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 		case 41:
 			switch (token) {
 			case '_':
-				parser->stack[++parser->top].state = 52;
+				parser->stack[++parser->top].state = 51;
 				return 1;
 
 			default:
@@ -711,7 +705,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 		case 42:
 			switch (token) {
 			case 'r':
-				parser->stack[++parser->top].state = 53;
+				parser->stack[++parser->top].state = 52;
 				return 1;
 
 			default:
@@ -722,7 +716,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 		case 43:
 			switch (token) {
 			case 'e':
-				parser->stack[++parser->top].state = 54;
+				parser->stack[++parser->top].state = 53;
 				return 1;
 
 			default:
@@ -733,7 +727,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 		case 44:
 			switch (token) {
 			case TOK_BLOCK:
-				parser->stack[++parser->top].state = 55;
+				parser->stack[++parser->top].state = 54;
 				return 1;
 
 			default:
@@ -744,7 +738,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 		case 45:
 			switch (token) {
 			case TOK_BLOCK:
-				parser->stack[++parser->top].state = 56;
+				parser->stack[++parser->top].state = 55;
 				return 1;
 
 			default:
@@ -755,7 +749,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 		case 46:
 			switch (token) {
 			case ')':
-				parser->stack[++parser->top].state = 57;
+				parser->stack[++parser->top].state = 56;
 				return 1;
 
 			default:
@@ -767,27 +761,27 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			switch (token) {
 			default:
 				parser->top -= 1;
-				result.symbol_varname = parse_reduce_25(parser->stack[parser->top + 0].value.token.str, grammar);
+				result.symbol_varname = parse_reduce_24(parser->stack[parser->top + 0].value.token.str, grammar);
 				parser->stack[parser->top].value.symbol_varname = result.symbol_varname;
 				switch (parser->stack[parser->top].state) {
 				case 39:
 					parser->stack[++parser->top].state = 46;
 					break;
 
-				case 50:
-					parser->stack[++parser->top].state = 58;
+				case 49:
+					parser->stack[++parser->top].state = 57;
 					break;
 
-				case 55:
-					parser->stack[++parser->top].state = 63;
+				case 54:
+					parser->stack[++parser->top].state = 62;
 					break;
 
-				case 70:
-					parser->stack[++parser->top].state = 72;
+				case 69:
+					parser->stack[++parser->top].state = 71;
 					break;
 
-				case 71:
-					parser->stack[++parser->top].state = 73;
+				case 70:
+					parser->stack[++parser->top].state = 72;
 					break;
 
 				}
@@ -805,69 +799,34 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 					parser->stack[++parser->top].state = 46;
 					break;
 
-				case 50:
-					parser->stack[++parser->top].state = 58;
+				case 49:
+					parser->stack[++parser->top].state = 57;
 					break;
 
-				case 55:
-					parser->stack[++parser->top].state = 63;
+				case 54:
+					parser->stack[++parser->top].state = 62;
 					break;
 
-				case 70:
-					parser->stack[++parser->top].state = 72;
-					break;
-
-				case 71:
-					parser->stack[++parser->top].state = 73;
-					break;
-
-				}
-			}
-			break;
-
-		case 49:
-			switch (token) {
-			default:
-				parser->top -= 1;
-				result.symbol_varname = parse_reduce_24(parser->stack[parser->top + 0].value.token.str, grammar);
-				parser->stack[parser->top].value.symbol_varname = result.symbol_varname;
-				switch (parser->stack[parser->top].state) {
-				case 39:
-					parser->stack[++parser->top].state = 46;
-					break;
-
-				case 50:
-					parser->stack[++parser->top].state = 58;
-					break;
-
-				case 55:
-					parser->stack[++parser->top].state = 63;
+				case 69:
+					parser->stack[++parser->top].state = 71;
 					break;
 
 				case 70:
 					parser->stack[++parser->top].state = 72;
 					break;
 
-				case 71:
-					parser->stack[++parser->top].state = 73;
-					break;
-
 				}
 			}
 			break;
 
-		case 50:
+		case 49:
 			switch (token) {
 			case TOK_SYMBOL:
 				parser->stack[++parser->top].state = 47;
 				return 1;
 
-			case TOK_SYMBOL_LC:
-				parser->stack[++parser->top].state = 48;
-				return 1;
-
 			case TOK_SYMBOL_UC:
-				parser->stack[++parser->top].state = 49;
+				parser->stack[++parser->top].state = 48;
 				return 1;
 
 			default:
@@ -875,10 +834,10 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 51:
+		case 50:
 			switch (token) {
 			case 'c':
-				parser->stack[++parser->top].state = 59;
+				parser->stack[++parser->top].state = 58;
 				return 1;
 
 			default:
@@ -886,10 +845,10 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 52:
+		case 51:
 			switch (token) {
 			case 'a':
-				parser->stack[++parser->top].state = 60;
+				parser->stack[++parser->top].state = 59;
 				return 1;
 
 			default:
@@ -897,10 +856,10 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 53:
+		case 52:
 			switch (token) {
 			case TOK_BLOCK:
-				parser->stack[++parser->top].state = 61;
+				parser->stack[++parser->top].state = 60;
 				return 1;
 
 			default:
@@ -908,10 +867,10 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 54:
+		case 53:
 			switch (token) {
 			case TOK_BLOCK:
-				parser->stack[++parser->top].state = 62;
+				parser->stack[++parser->top].state = 61;
 				return 1;
 
 			default:
@@ -919,18 +878,14 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 55:
+		case 54:
 			switch (token) {
 			case TOK_SYMBOL:
 				parser->stack[++parser->top].state = 47;
 				return 1;
 
-			case TOK_SYMBOL_LC:
-				parser->stack[++parser->top].state = 48;
-				return 1;
-
 			case TOK_SYMBOL_UC:
-				parser->stack[++parser->top].state = 49;
+				parser->stack[++parser->top].state = 48;
 				return 1;
 
 			default:
@@ -938,7 +893,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 56:
+		case 55:
 			switch (token) {
 			default:
 				parser->top -= 7;
@@ -949,7 +904,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 57:
+		case 56:
 			switch (token) {
 			default:
 				parser->top -= 5;
@@ -961,10 +916,10 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 58:
+		case 57:
 			switch (token) {
 			case ')':
-				parser->stack[++parser->top].state = 64;
+				parser->stack[++parser->top].state = 63;
 				return 1;
 
 			default:
@@ -972,10 +927,10 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 59:
+		case 58:
 			switch (token) {
 			case 't':
-				parser->stack[++parser->top].state = 65;
+				parser->stack[++parser->top].state = 64;
 				return 1;
 
 			default:
@@ -983,10 +938,10 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 60:
+		case 59:
 			switch (token) {
 			case 'r':
-				parser->stack[++parser->top].state = 66;
+				parser->stack[++parser->top].state = 65;
 				return 1;
 
 			default:
@@ -994,7 +949,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 61:
+		case 60:
 			switch (token) {
 			default:
 				parser->top -= 8;
@@ -1004,7 +959,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 62:
+		case 61:
 			switch (token) {
 			default:
 				parser->top -= 8;
@@ -1014,7 +969,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 63:
+		case 62:
 			switch (token) {
 			default:
 				parser->top -= 8;
@@ -1026,7 +981,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 64:
+		case 63:
 			switch (token) {
 			default:
 				parser->top -= 6;
@@ -1038,10 +993,10 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 65:
+		case 64:
 			switch (token) {
 			case 'o':
-				parser->stack[++parser->top].state = 67;
+				parser->stack[++parser->top].state = 66;
 				return 1;
 
 			default:
@@ -1049,10 +1004,10 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 66:
+		case 65:
 			switch (token) {
 			case 'g':
-				parser->stack[++parser->top].state = 68;
+				parser->stack[++parser->top].state = 67;
 				return 1;
 
 			default:
@@ -1060,10 +1015,10 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 67:
+		case 66:
 			switch (token) {
 			case 'r':
-				parser->stack[++parser->top].state = 69;
+				parser->stack[++parser->top].state = 68;
 				return 1;
 
 			default:
@@ -1071,10 +1026,10 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 68:
+		case 67:
 			switch (token) {
 			case TOK_BLOCK:
-				parser->stack[++parser->top].state = 70;
+				parser->stack[++parser->top].state = 69;
 				return 1;
 
 			default:
@@ -1082,13 +1037,13 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 69:
+		case 68:
 			switch (token) {
 			case TOK_CHAR:
 				parser->stack[++parser->top].state = 28;
 				return 1;
 
-			case TOK_SYMBOL_LC:
+			case TOK_SYMBOL:
 				parser->stack[++parser->top].state = 31;
 				return 1;
 
@@ -1101,18 +1056,14 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 70:
+		case 69:
 			switch (token) {
 			case TOK_SYMBOL:
 				parser->stack[++parser->top].state = 47;
 				return 1;
 
-			case TOK_SYMBOL_LC:
-				parser->stack[++parser->top].state = 48;
-				return 1;
-
 			case TOK_SYMBOL_UC:
-				parser->stack[++parser->top].state = 49;
+				parser->stack[++parser->top].state = 48;
 				return 1;
 
 			default:
@@ -1120,18 +1071,14 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 71:
+		case 70:
 			switch (token) {
 			case TOK_SYMBOL:
 				parser->stack[++parser->top].state = 47;
 				return 1;
 
-			case TOK_SYMBOL_LC:
-				parser->stack[++parser->top].state = 48;
-				return 1;
-
 			case TOK_SYMBOL_UC:
-				parser->stack[++parser->top].state = 49;
+				parser->stack[++parser->top].state = 48;
 				return 1;
 
 			default:
@@ -1139,7 +1086,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 72:
+		case 71:
 			switch (token) {
 			default:
 				parser->top -= 12;
@@ -1150,7 +1097,7 @@ static int parse_do_push(parse_context_t *parser, int token, __attribute__((unus
 			}
 			break;
 
-		case 73:
+		case 72:
 			switch (token) {
 			default:
 				parser->top -= 13;
diff --git a/src/parse.hpp b/src/parse.hpp
index 2d73755..e080ceb 100644
--- a/src/parse.hpp
+++ b/src/parse.hpp
@@ -10,8 +10,7 @@ typedef enum parse_token {
 	TOK_SQBLOCK = 258,
 	TOK_STRING = 259,
 	TOK_SYMBOL = 260,
-	TOK_SYMBOL_LC = 261,
-	TOK_SYMBOL_UC = 262,
+	TOK_SYMBOL_UC = 261,
 } parse_token_t;
 
 typedef struct parse_token_value {
diff --git a/src/parse.y b/src/parse.y
index f0ff040..56fc567 100644
--- a/src/parse.y
+++ b/src/parse.y
@@ -31,9 +31,6 @@ static inline void free_rhs(rhs_t *v) {
 %type SYMBOL_UC {std::string *} str
 %destructor SYMBOL_UC free_string
 
-%type SYMBOL_LC {std::string *} str
-%destructor SYMBOL_LC free_string
-
 %type BLOCK {std::string *} str
 %destructor BLOCK free_string
 
@@ -72,7 +69,7 @@ grammar |=;
 grammar |= grammar directive;
 
 
-directive |= "%type" SYMBOL_LC(nonterm) BLOCK(type) {
+directive |= "%type" SYMBOL(nonterm) BLOCK(type) {
 	grammar->nonterm_types.insert(std::make_pair(*nonterm, *type));
 }
 
@@ -107,7 +104,7 @@ directive |= rule(rule) {
 }
 
 
-rule |= SYMBOL_LC(lhs) "|=" rhs(rhs) action(action)
+rule |= SYMBOL(lhs) "|=" rhs(rhs) action(action)
 	[new solar::rule_t {solar::item_t(*lhs, rhs->first), rhs->second, *action}]
 
 
@@ -149,12 +146,11 @@ action |= BLOCK(=v) [v]
 action |= SQBLOCK(v) [new std::string("return " + *v + ";")]
 
 
-symbol |= SYMBOL_LC(v) [new solar::symbol_t(solar::symbol_t::make_nonterm(*v))]
+symbol |= SYMBOL(v) [new solar::symbol_t(solar::symbol_t::make_nonterm(*v))]
 symbol |= term(=v) [v]
 
 term |= SYMBOL_UC(v) [new solar::symbol_t(solar::symbol_t::make_term(*v))]
 term |= CHAR(v) [new solar::symbol_t(solar::symbol_t::make_char(v))]
 
-varname |= SYMBOL_LC(=v) [v]
 varname |= SYMBOL_UC(=v) [v]
 varname |= SYMBOL(=v) [v]