merlin-utils/script.re2c

/* link script support */

/*

	label opcode operand



 */

#include <string>
#include <unordered_map>
#include <stdexcept>
#include <variant>

#include <cctype>
#include <cstdio>
#include <cstdint>

#include <err.h>

#include "script.h"


/*!re2c
    re2c:define:YYCTYPE = char;
    re2c:yyfill:enable = 0;

    // :-~ includes ; which interferes with comments.
	ident = [:<-~][0-~]*;
	ws = [ \t];
	eof = "\x00";
*/

namespace {
	std::unordered_map<std::string, int> opcodes = {
		#define x(op) { #op, OP_##op },

		#include "ops.h"
		#undef x

		/* aliases */
		{ "AUX", OP_ADR },
		{ "REZ", OP_RES },
		{ "LIN", OP_LNK },
		{ "KIN", OP_KND },
		{ "=",   OP_EQ }
	};


	std::unordered_map<std::string, int> types = {

		{ "NON", 0x00 },
		{ "BAD", 0x01 },
		{ "BIN", 0x06 },
		{ "TXT", 0x04 },
		{ "DIR", 0x0f },
		{ "ADB", 0x19 },
		{ "AWP", 0x1a },
		{ "ASP", 0x1b },
		{ "GSB", 0xab },
		{ "TDF", 0xac },
		{ "BDF", 0xad },
		{ "SRC", 0xb0 },
		{ "OBJ", 0xb1 },
		{ "LIB", 0xb2 },
		{ "S16", 0xb3 },
		{ "RTL", 0xb4 },
		{ "EXE", 0xb5 },
		{ "PIF", 0xb6 },
		{ "TIF", 0xb7 },
		{ "NDA", 0xb8 },
		{ "CDA", 0xb9 },
		{ "TOL", 0xba },
		{ "DRV", 0xbb },
		{ "DOC", 0xbf },
		{ "PNT", 0xc0 },
		{ "PIC", 0xc1 },
		{ "FON", 0xcb },
		{ "PAS", 0xef },
		{ "CMD", 0xf0 },
		{ "LNK", 0xf8 },
		{ "BAS", 0xfc },
		{ "VAR", 0xfd },
		{ "REL", 0xfe },
		{ "SYS", 0xff },

	};
}


static uint32_t number_operand(const char *YYCURSOR, bool required = true) {

	const char *iter = YYCURSOR;
	const char *YYMARKER = nullptr;
	uint32_t rv = 0;
	/*!re2c
		* { throw std::invalid_argument("bad operand"); }
		[;*] | eof {
			if (!required) return rv;
			throw std::invalid_argument("missing operand");
		}
		'%' [01]+ {
			++iter;
			for(;iter < YYCURSOR; ++iter) {
				rv <<= 1;
				rv |= *iter - '0';
			}
			goto exit;		
		}

		'$' [A-Fa-f0-9]+ {
			++iter;
			for(;iter < YYCURSOR; ++iter) {
				char c = *iter | 0x20;
				rv <<= 4;
				if (c <= '9') rv |= c - '0';
				else rv |= c - 'a' + 10;
			}
			goto exit;		
		}

		[0-9]+ {
			for(;iter < YYCURSOR; ++iter) {
				rv *= 10;
				rv += *iter - '0';
			}
			goto exit;		
	
		}

		ident {
			std::string s(iter, YYCURSOR);
			//look up symbol, verify it's an absolute value, etc
		}
	*/
exit:
	char c = *YYCURSOR;
	if (isspace(c) || c == 0) return rv;

	throw std::invalid_argument("bad operand");
}

static uint32_t type_operand(const char *YYCURSOR, bool required = true) {

	const char *iter = YYCURSOR;
	const char *YYMARKER = nullptr;
	uint32_t rv = 0;
	/*!re2c
		* { throw std::invalid_argument("bad operand"); }

		[;*] | eof {
			if (!required) return rv;
			throw std::invalid_argument("missing operand");
		}

		'%' [01]+ {
			++iter;
			for(;iter < YYCURSOR; ++iter) {
				rv <<= 1;
				rv |= *iter - '0';
			}
			goto exit;		
		}

		'$' [A-Fa-f0-9]+ {
			++iter;
			for(;iter < YYCURSOR; ++iter) {
				char c = *iter | 0x20;
				rv <<= 4;
				if (c <= '9') rv |= c - '0';
				else rv |= c - 'a' + 10;
			}
			goto exit;		
		}

		[0-9]+ {
			for(;iter < YYCURSOR; ++iter) {
				rv *= 10;
				rv += *iter - '0';
			}
			goto exit;		
	
		}

		[A-Za-z][A-Za-z0-9]{2} {
			std::string s(iter, YYCURSOR);
			for(char &c : s) c = std::toupper(c);
			auto iter = types.find(s);
			if (iter == types.end()) {
				throw std::invalid_argument("bad operand");
			}
			rv = iter->second;
		}
	*/
exit:
	char c = *YYCURSOR;
	if (isspace(c) || c == 0) return rv;

	throw std::invalid_argument("bad operand");
}



static int x_number_operand(const char *YYCURSOR) {
	const char *iter = YYCURSOR;
	const char *YYMARKER = nullptr;
	uint32_t rv = 0;
	/*!re2c
		* { throw std::invalid_argument("bad operand"); }

		'%' [01]+ {
			++iter;
			for(;iter < YYCURSOR; ++iter) {
				rv <<= 1;
				rv |= *iter - '0';
			}
			goto exit;		
		}

		'$' [A-Fa-f0-9]+ {
			++iter;
			for(;iter < YYCURSOR; ++iter) {
				char c = *iter | 0x20;
				rv <<= 4;
				if (c <= '9') rv |= c - '0';
				else rv |= c - 'a' + 10;
			}
			goto exit;		
		}

		[0-9]+ {
			for(;iter < YYCURSOR; ++iter) {
				rv *= 10;
				rv += *iter - '0';
			}
			goto exit;		
	
		}
	*/
exit:
	char c = *YYCURSOR;
	if (isspace(c) || c == 0) return rv;

	throw std::invalid_argument("bad operand");
}

static std::string x_label_operand(const char *YYCURSOR) {
	const char *iter = YYCURSOR;
	const char *YYMARKER = nullptr;
	std::string rv;
	/*!re2c
		* { throw std::invalid_argument("bad operand"); }

		ident {
			std::string s(iter, YYCURSOR);
			goto exit;
		}
	*/
exit:
	char c = *YYCURSOR;
	if (isspace(c) || c == 0) {
		//look up symbol, verify it's an absolute value, etc
		return rv;
	}
	throw std::invalid_argument("bad operand");	
}

static std::string x_string_operand(const char *YYCURSOR) {
	const char *iter = YYCURSOR;
	std::string rv;
	/*!re2c
		* { throw std::invalid_argument("bad operand"); }

		['] [^']* ['] | ["] [^"]* ["] {
			rv = std::string(iter+1, YYCURSOR-1);
			goto exit;
		}

	*/
exit:
	char c = *YYCURSOR;
	if (isspace(c) || c == 0) return rv;
	throw std::invalid_argument("bad operand");	
}



static int ovr_operand(const char *YYCURSOR) {
	int rv = 0;
	const char *YYMARKER = nullptr;

	/*!re2c
		* { throw std::invalid_argument("bad operand"); }
		[;*] | eof {
			return OVR_NONE;
		}
		'ALL' {
			rv = OVR_ALL;
		}
		'OFF' {
			rv = OVR_OFF;
		}
	*/

	char c = *YYCURSOR;
	if (isspace(c) || c == 0) return rv;

	throw std::invalid_argument("bad operand");
}

static std::string label_operand(const char *YYCURSOR, bool required = true) {
	std::string rv;
	const char *iter = YYCURSOR;
	const char *YYMARKER = nullptr;
	/*!re2c
		* { throw std::invalid_argument("bad operand"); }
		[;*] | eof {
			if (!required) return rv;
			throw std::invalid_argument("missing operand");
		}
		ident {
			rv = std::string(iter, YYCURSOR);
			goto exit;
		}
	*/

exit:
	char c = *YYCURSOR;
	if (isspace(c) || c == 0) return rv;

	throw std::invalid_argument("bad operand");
}


static std::string path_operand(const char *YYCURSOR, bool required = true) {
	std::string rv;
	const char *iter = YYCURSOR;
	const char *YYMARKER = nullptr;
	/*!re2c
		* { throw std::invalid_argument("bad operand"); }
		[;*] | eof {
			if (!required) return rv;
			throw std::invalid_argument("missing operand");
		}
		// don't allow leading quotes, eof, or comment chars
		[^;*\x00'"][^ \t]* {
			rv = std::string(iter, YYCURSOR);
			goto exit;
		}
		['] [^']* ['] | ["] [^"]* ["] {
			rv = std::string(iter+1, YYCURSOR-1);
			goto exit;
		}

	*/

exit:
	char c = *YYCURSOR;
	if (isspace(c) || c == 0) return rv;

	throw std::invalid_argument("bad operand");
}


static void no_operand(const char *YYCURSOR) {
	/*!re2c
		* { throw std::invalid_argument("bad operand"); }
		[;*] | eof { return; }
	*/
}


static std::string string_operand(const char *YYCURSOR, bool required = true) {

	std::string rv;
	const char *iter = YYCURSOR;
	const char *YYMARKER = nullptr;
	/*!re2c
		* { throw std::invalid_argument("bad operand"); }
		[;*] | eof {
			if (!required) return rv;
			throw std::invalid_argument("missing operand");
		}
		['] [^']* ['] | ["] [^"]* ["] {
			rv = std::string(iter+1, YYCURSOR-1);
			goto exit;
		}
	*/
exit:
	char c = *YYCURSOR;
	if (isspace(c) || c == 0) return rv;
	throw std::invalid_argument("bad operand");
}



void parse_line(const char *YYCURSOR) {

	label_t label;
	opcode_t opcode = OP_NONE;
	operand_t operand;

	const char *iter = YYCURSOR;
	const char *YYMARKER = nullptr;

	/*!re2c

		* { throw std::invalid_argument("bad label"); }
		[;*] | eof {
			return;
		}
		ws { goto opcode; }
		ident / (ws|eof) {
			label = std::string(iter, YYCURSOR);
			goto opcode;	
		}

	*/


opcode:

	while (isspace(*YYCURSOR)) ++YYCURSOR;
	iter = YYCURSOR;

	/*!re2c

		* { throw std::invalid_argument("bad opcode"); }
		[;*]|eof { return; }

		'=' / (ws|eof) { opcode = OP_EQ; goto operand; }

		[A-Za-z]+ / (ws|eof) {
			size_t l = YYCURSOR - iter;
			if (l > 3) l = 3;
			std::string s(iter, iter + l);
			for (char &c : s) c = std::toupper(c);
			auto iter = opcodes.find(s);
			if (iter == opcodes.end()) {
				throw std::invalid_argument("bad opcode");
			}
			opcode = iter->second;
			goto operand;
		}
	*/

operand:

	while (isspace(*YYCURSOR)) ++YYCURSOR;
	iter = YYCURSOR;


	switch(opcode) {
		case OP_LNK:
		case OP_PUT:
		case OP_ASM:
		case OP_SAV:
		case OP_LIB:
		case OP_IF:
		case OP_PFX:
		case OP_IMP:
		case OP_RES:
		case OP_FIL:
			operand = path_operand(YYCURSOR);
			break;
		case OP_ORG:
		case OP_ADR:
		case OP_DS:
		case OP_KND:
		case OP_VER:
		case OP_ALI:
		case OP_LKV:
		case OP_DO:
		case OP_EQU:
		case OP_EQ:
		case OP_GEQ:
			operand = number_operand(YYCURSOR);
			break;
		case OP_TYP:
			operand = type_operand(YYCURSOR);
			break;
		case OP_OVR:
			operand = ovr_operand(YYCURSOR);
			break;
		case OP_POS:
		case OP_LEN: {
			std::string tmp = label_operand(YYCURSOR, false);
			if (!tmp.empty()) operand = std::move(tmp);
			break;
		}
		case OP_KBD: {
			std::string tmp = string_operand(YYCURSOR, false);
			if (!tmp.empty()) operand = std::move(tmp);
			break;
		}

		case OP_CMD:
			str_operand = std::string(YYCURSOR);
			break;

		default:
			no_operand(YYCURSOR);
			break;
	}


	void evaluate(label_t label, opcode_t opcode, operand_t operand);

	evaluate(label, opcode, opcode);
}