Add support for extra arguments

8 files changed, 93 insertions, 7 deletions

diff --git a/src/generator.cpp b/src/generator.cpp
index 31dac2a..16f08d0 100644
--- a/src/generator.cpp
+++ b/src/generator.cpp
@@ -120,9 +120,10 @@ void generator_t::generate_itemsets() {
 generator_t::generator_t(const std::vector<std::tuple<item_t, std::vector<std::string>, std::string>> &rules0,
 			 const std::map<std::string, std::string> &nonterm_types0,
 			 const std::map<symbol_t, std::pair<std::string, std::string>> &term_types0,
-			 const std::string &header_block0, const std::string &source_block0)
+			 const std::string &header_block0, const std::string &source_block0,
+			 const std::vector<std::pair<std::string, std::string>> &extra_args0)
 	: rules(rules0), nonterm_types(nonterm_types0), term_types(term_types0),
-	  header_block(header_block0), source_block(source_block0) {
+	  header_block(header_block0), source_block(source_block0), extra_args(extra_args0) {
 	for (size_t i = 0; i < rules.size(); i++) {
 		item_t rule = std::get<0>(rules[i]);
 
diff --git a/src/generator.hpp b/src/generator.hpp
index 23f4f9d..b0bd4c9 100644
--- a/src/generator.hpp
+++ b/src/generator.hpp
@@ -66,6 +66,8 @@ private:
 	std::string header_block;
 	std::string source_block;
 
+	std::vector<std::pair<std::string, std::string>> extra_args;
+
 
 	void close_set(std::set<item_t> *set);
 	std::set<item_t> get_set(const std::string &nonterm);
@@ -164,11 +166,16 @@ public:
 		return source_block;
 	}
 
+	const std::vector<std::pair<std::string, std::string>> & get_extra_args() const {
+		return extra_args;
+	}
+
 
 	generator_t(const std::vector<std::tuple<item_t, std::vector<std::string>, std::string>> &rules0,
 		    const std::map<std::string, std::string> &nonterm_types0,
 		    const std::map<symbol_t, std::pair<std::string, std::string>> &term_types0,
-		    const std::string &header_block0, const std::string &source_block0);
+		    const std::string &header_block0, const std::string &source_block0,
+		    const std::vector<std::pair<std::string, std::string>> &extra_args0);
 };
 
 }
diff --git a/src/lex.cpp b/src/lex.cpp
index 7dfc013..c97e03d 100644
--- a/src/lex.cpp
+++ b/src/lex.cpp
@@ -42,6 +42,7 @@ struct keyword_t {
 
 /* the keyword list must be sorted */
 static const keyword_t keywords[] = {
+	{"%extra_arg", TOK_EXTRA_ARG},
 	{"%header", TOK_HEADER},
 	{"%source", TOK_SOURCE},
 	{"%type", TOK_TYPE},
@@ -229,7 +230,7 @@ int lex_t::lex_keyword(parser_value_t *value) {
 	while (next(false)) {
 		char cur = current();
 
-		if (!((cur >= 'a' && cur <= 'z') || (cur >= '0' && cur <= '9') || cur == '-'))
+		if (!((cur >= 'a' && cur <= 'z') || (cur >= '0' && cur <= '9') || cur == '_'))
 			break;
 	}
 
diff --git a/src/output.cpp b/src/output.cpp
index 7f074ab..33c6bb1 100644
--- a/src/output.cpp
+++ b/src/output.cpp
@@ -101,7 +101,12 @@ void output_t::emit_header() {
 	std::fprintf(header_file, "%scontext_t * %salloc(void *(*alloc_func)(size_t));\n", prefix(), prefix());
 	std::fprintf(header_file, "void %sfree(%scontext_t *parser, void (*free_func)(void *));\n\n", prefix(), prefix());
 
-	std::fprintf(header_file, "int %spush(%scontext_t *parser, int token, const %stoken_value_t *value);\n", prefix(), prefix(), prefix());
+	std::fprintf(header_file, "int %spush(%scontext_t *parser, int token, const %stoken_value_t *value", prefix(), prefix(), prefix());
+
+	for (const auto &arg : generator->get_extra_args())
+		std::fprintf(header_file, ", %s %s", arg.first.c_str(), arg.second.c_str());
+
+	std::fprintf(header_file, ");\n");
 }
 
 void output_t::emit_reduction(unsigned rule_id) {
@@ -133,6 +138,15 @@ void output_t::emit_reduction(unsigned rule_id) {
 		empty = false;
 	}
 
+	for (const auto &arg : generator->get_extra_args()) {
+		if (!empty)
+			std::fprintf(source_file, ", ");
+
+		std::fprintf(source_file, "%s %s", arg.first.c_str(), arg.second.c_str());
+
+		empty = false;
+	}
+
 	if (empty)
 		std::fprintf(source_file, "void");
 
@@ -207,6 +221,15 @@ void output_t::emit_state_reduce(const item_t &item, int rule_id) {
 			empty = false;
 		}
 
+		for (const auto &arg : generator->get_extra_args()) {
+			if (!empty)
+				std::fprintf(source_file, ", ");
+
+			std::fprintf(source_file, "%s", arg.second.c_str());
+
+			empty = false;
+		}
+
 		std::fprintf(source_file, ");\n");
 	}
 
@@ -312,7 +335,11 @@ void output_t::emit_source() {
 
 	emit_reductions();
 
-	std::fprintf(source_file, "int %spush(%scontext_t *parser, int token, const %stoken_value_t *value) {\n", prefix(), prefix(), prefix());
+	std::fprintf(source_file, "int %spush(%scontext_t *parser, int token, const %stoken_value_t *value", prefix(), prefix(), prefix());
+	for (const auto &arg : generator->get_extra_args())
+		std::fprintf(source_file, ", %s %s", arg.first.c_str(), arg.second.c_str());
+	std::fprintf(source_file, ") {\n");
+
 	std::fprintf(source_file, "\twhile (1) {\n");
 	std::fprintf(source_file, "\t\tswitch (parser->stack[parser->top].state) {\n");
 
diff --git a/src/parser.cpp b/src/parser.cpp
index 84159dc..0333243 100644
--- a/src/parser.cpp
+++ b/src/parser.cpp
@@ -45,6 +45,8 @@ enum parser_state {
 	STATE_TYPE_TERM_BLOCK,
 	STATE_HEADER,
 	STATE_SOURCE,
+	STATE_EXTRA_ARG,
+	STATE_EXTRA_ARG_NAME,
 };
 
 struct parser {
@@ -80,6 +82,10 @@ int parser_push(parser_t *parser, int token, const parser_value_t *value, parser
 			parser->state = STATE_HEADER;
 			return 1;
 
+		case TOK_EXTRA_ARG:
+			parser->state = STATE_EXTRA_ARG;
+			return 1;
+
 		case 0:
 			return 0;
 		}
@@ -234,6 +240,7 @@ int parser_push(parser_t *parser, int token, const parser_value_t *value, parser
 			parser->state = STATE_INIT;
 			return 1;
 		}
+		break;
 
 	case STATE_HEADER:
 		if (token == TOK_BLOCK) {
@@ -244,6 +251,7 @@ int parser_push(parser_t *parser, int token, const parser_value_t *value, parser
 		}
 
 		break;
+
 	case STATE_SOURCE:
 		if (token == TOK_BLOCK) {
 			state->set_source_block(value->str);
@@ -253,6 +261,26 @@ int parser_push(parser_t *parser, int token, const parser_value_t *value, parser
 		}
 
 		break;
+
+	case STATE_EXTRA_ARG:
+		if (token == TOK_BLOCK) {
+			state->add_extra_arg(value->str);
+			free(value->str);
+			parser->state = STATE_EXTRA_ARG_NAME;
+			return 1;
+		}
+		break;
+
+	case STATE_EXTRA_ARG_NAME:
+		switch (token) {
+		case TOK_SYMBOL:
+		case TOK_SYMBOL_UC:
+		case TOK_SYMBOL_LC:
+			state->set_extra_arg_name(value->str);
+			free(value->str);
+			parser->state = STATE_INIT;
+			return 1;
+		}
 	}
 
 	switch (token) {
diff --git a/src/parser.hpp b/src/parser.hpp
index 9117914..61bfab4 100644
--- a/src/parser.hpp
+++ b/src/parser.hpp
@@ -43,6 +43,7 @@ enum parser_token_t {
 	TOK_TYPE,
 	TOK_SOURCE,
 	TOK_HEADER,
+	TOK_EXTRA_ARG,
 };
 
 typedef struct parser_value {
diff --git a/src/parser_state.hpp b/src/parser_state.hpp
index b183898..5dcce54 100644
--- a/src/parser_state.hpp
+++ b/src/parser_state.hpp
@@ -47,6 +47,9 @@ private:
 	std::string header_block;
 	std::string source_block;
 
+	std::string current_extra_arg;
+	std::vector<std::pair<std::string, std::string>> extra_args;
+
 public:
 	parser_state_t() : current("") {}
 
@@ -78,6 +81,19 @@ public:
 		source_block = value;
 	}
 
+	const std::vector<std::pair<std::string, std::string>> & get_extra_args() const {
+		return extra_args;
+	}
+
+	void add_extra_arg(const char *type) {
+		current_extra_arg = type;
+	}
+
+	void set_extra_arg_name(const char *name) {
+		extra_args.emplace_back(current_extra_arg, name);
+	}
+
+
 	void new_rule(const char *nonterm);
 	void add_rule_nonterminal(const char *nonterm);
 	void add_rule_terminal(const char *term);
diff --git a/src/solar.cpp b/src/solar.cpp
index 46c7fad..7e211a5 100644
--- a/src/solar.cpp
+++ b/src/solar.cpp
@@ -87,7 +87,12 @@ int main(int argc, char *argv[]) {
 	if (!read_grammar(argv[1], &state))
 		return 1;
 
-	generator_t generator(state.get_rules(), state.get_nonterm_types(), state.get_term_types(), state.get_header_block(), state.get_source_block());
+	generator_t generator(state.get_rules(),
+			      state.get_nonterm_types(),
+			      state.get_term_types(),
+			      state.get_header_block(),
+			      state.get_source_block(),
+			      state.get_extra_args());
 
 	output_t output(&generator, argv[3], argv[2]);
 	output.write();