main.c

另一版的词法分析器

	printstats = syntaxerror = trace = spprdflt = caseins = false;
	lex_compat = C_plus_plus = backing_up_report = ddebug = fulltbl = false;
	fullspd = long_align = nowarn = yymore_used = continued_action = false;
	do_yylineno = yytext_is_array = in_rule = reject = do_stdinit = false;
	yymore_really_used = reject_really_used = unspecified;
	interactive = csize = unspecified;
	do_yywrap = gen_line_dirs = usemecs = useecs = true;
	performance_report = 0;
	did_outfilename = 0;
	prefix = "yy";
	yyclass = 0;
	use_read = use_stdout = false;

	sawcmpflag = false;

	/* Initialize dynamic array for holding the rule actions. */
	action_size = 2048;	/* default size of action array in bytes */
	action_array = allocate_character_array( action_size );
	defs1_offset = prolog_offset = action_offset = action_index = 0;
	action_array[0] = '\0';

	program_name = argv[0];

	if ( program_name[0] != '\0' &&
	     program_name[strlen( program_name ) - 1] == '+' )
		C_plus_plus = true;

	/* read flags */
	for ( --argc, ++argv; argc ; --argc, ++argv )
		{
		arg = argv[0];

		if ( arg[0] != '-' || arg[1] == '\0' )
			break;

		if ( arg[1] == '-' )
			{ /* --option */
			if ( ! strcmp( arg, "--help" ) )
				arg = "-h";

			else if ( ! strcmp( arg, "--version" ) )
				arg = "-V";

			else if ( ! strcmp( arg, "--" ) )
				{ /* end of options */
				--argc;
				++argv;
				break;
				}
			}

		for ( i = 1; arg[i] != '\0'; ++i )
			switch ( arg[i] )
				{
				case '+':
					C_plus_plus = true;
					break;

				case 'B':
					interactive = false;
					break;

				case 'b':
					backing_up_report = true;
					break;

				case 'c':
					break;

				case 'C':
					if ( i != 1 )
						flexerror(
				_( "-C flag must be given separately" ) );

					if ( ! sawcmpflag )
						{
						useecs = false;
						usemecs = false;
						fulltbl = false;
						sawcmpflag = true;
						}

					for ( ++i; arg[i] != '\0'; ++i )
						switch ( arg[i] )
							{
							case 'a':
								long_align =
									true;
								break;

							case 'e':
								useecs = true;
								break;

							case 'F':
								fullspd = true;
								break;

							case 'f':
								fulltbl = true;
								break;

							case 'm':
								usemecs = true;
								break;

							case 'r':
								use_read = true;
								break;

							default:
								lerrif(
						_( "unknown -C option '%c'" ),
								(int) arg[i] );
								break;
							}

					goto get_next_arg;

				case 'd':
					ddebug = true;
					break;

				case 'f':
					useecs = usemecs = false;
					use_read = fulltbl = true;
					break;

				case 'F':
					useecs = usemecs = false;
					use_read = fullspd = true;
					break;

				case '?':
				case 'h':
					usage();
					exit( 0 );

				case 'I':
					interactive = true;
					break;

				case 'i':
					caseins = true;
					break;

				case 'l':
					lex_compat = true;
					break;

				case 'L':
					gen_line_dirs = false;
					break;

				case 'n':
					/* Stupid do-nothing deprecated
					 * option.
					 */
					break;

				case 'o':
					if ( i != 1 )
						flexerror(
				_( "-o flag must be given separately" ) );

					outfilename = arg + i + 1;
					did_outfilename = 1;
					goto get_next_arg;

				case 'P':
					if ( i != 1 )
						flexerror(
				_( "-P flag must be given separately" ) );

					prefix = arg + i + 1;
					goto get_next_arg;

				case 'p':
					++performance_report;
					break;

				case 'S':
					if ( i != 1 )
						flexerror(
				_( "-S flag must be given separately" ) );

					skelname = arg + i + 1;
					goto get_next_arg;

				case 's':
					spprdflt = true;
					break;

				case 't':
					use_stdout = true;
					break;

				case 'T':
					trace = true;
					break;

				case 'v':
					printstats = true;
					break;

				case 'V':
					printf( _( "%s version %s\n" ),
						program_name, flex_version );
					exit( 0 );

				case 'w':
					nowarn = true;
					break;

				case '7':
					csize = 128;
					break;

				case '8':
					csize = CSIZE;
					break;

				default:
					fprintf( stderr,
		_( "%s: unknown flag '%c'.  For usage, try\n\t%s --help\n" ),
						program_name, (int) arg[i],
						program_name );
					exit( 1 );
				}

		/* Used by -C, -S, -o, and -P flags in lieu of a "continue 2"
		 * control.
		 */
		get_next_arg: ;
		}

	num_input_files = argc;
	input_files = argv;
	set_input_file( num_input_files > 0 ? input_files[0] : NULL );

	lastccl = lastsc = lastdfa = lastnfa = 0;
	num_rules = num_eof_rules = default_rule = 0;
	numas = numsnpairs = tmpuses = 0;
	numecs = numeps = eps2 = num_reallocs = hshcol = dfaeql = totnst = 0;
	numuniq = numdup = hshsave = eofseen = datapos = dataline = 0;
	num_backing_up = onesp = numprots = 0;
	variable_trailing_context_rules = bol_needed = false;

	out_linenum = linenum = sectnum = 1;
	firstprot = NIL;

	/* Used in mkprot() so that the first proto goes in slot 1
	 * of the proto queue.
	 */
	lastprot = 1;

	set_up_initial_allocations();
	}


/* readin - read in the rules section of the input file(s) */

void readin()
	{
	static char yy_stdinit[] = "FILE *yyin = stdin, *yyout = stdout;";
	static char yy_nostdinit[] =
		"FILE *yyin = (FILE *) 0, *yyout = (FILE *) 0;";

	line_directive_out( (FILE *) 0, 1 );

	if ( yyparse() )
		{
		pinpoint_message( _( "fatal parse error" ) );
		flexend( 1 );
		}

	if ( syntaxerror )
		flexend( 1 );

	if ( backing_up_report )
		{
		backing_up_file = fopen( backing_name, "w" );
		if ( backing_up_file == NULL )
			lerrsf(
			_( "could not create backing-up info file %s" ),
				backing_name );
		}

	else
		backing_up_file = NULL;

	if ( yymore_really_used == true )
		yymore_used = true;
	else if ( yymore_really_used == false )
		yymore_used = false;

	if ( reject_really_used == true )
		reject = true;
	else if ( reject_really_used == false )
		reject = false;

	if ( performance_report > 0 )
		{
		if ( lex_compat )
			{
			fprintf( stderr,
_( "-l AT&T lex compatibility option entails a large performance penalty\n" ) );
			fprintf( stderr,
_( " and may be the actual source of other reported performance penalties\n" ) );
			}

		else if ( do_yylineno )
			{
			fprintf( stderr,
	_( "%%option yylineno entails a large performance penalty\n" ) );
			}

		if ( performance_report > 1 )
			{
			if ( interactive )
				fprintf( stderr,
	_( "-I (interactive) entails a minor performance penalty\n" ) );

			if ( yymore_used )
				fprintf( stderr,
		_( "yymore() entails a minor performance penalty\n" ) );
			}

		if ( reject )
			fprintf( stderr,
			_( "REJECT entails a large performance penalty\n" ) );

		if ( variable_trailing_context_rules )
			fprintf( stderr,
_( "Variable trailing context rules entail a large performance penalty\n" ) );
		}

	if ( reject )
		real_reject = true;

	if ( variable_trailing_context_rules )
		reject = true;

	if ( (fulltbl || fullspd) && reject )
		{
		if ( real_reject )
			flexerror(
				_( "REJECT cannot be used with -f or -F" ) );
		else if ( do_yylineno )
			flexerror(
			_( "%option yylineno cannot be used with -f or -F" ) );
		else
			flexerror(
	_( "variable trailing context rules cannot be used with -f or -F" ) );
		}

	if ( reject )
		outn( "\n#define YY_USES_REJECT" );

	if ( ! do_yywrap )
		{
		outn( "\n#define yywrap() 1" );
		outn( "#define YY_SKIP_YYWRAP" );
		}

	if ( ddebug )
		outn( "\n#define FLEX_DEBUG" );

	if ( csize == 256 )
		outn( "typedef unsigned char YY_CHAR;" );
	else
		outn( "typedef char YY_CHAR;" );

	if ( C_plus_plus )
		{
		outn( "#define yytext_ptr yytext" );

		if ( interactive )
			outn( "#define YY_INTERACTIVE" );
		}

	else
		{
		if ( do_stdinit )
			{
			outn( "#ifdef VMS" );
			outn( "#ifndef __VMS_POSIX" );
			outn( yy_nostdinit );
			outn( "#else" );
			outn( yy_stdinit );
			outn( "#endif" );
			outn( "#else" );
			outn( yy_stdinit );
			outn( "#endif" );
			}

		else
			outn( yy_nostdinit );
		}

	if ( fullspd )
		outn( "typedef yyconst struct yy_trans_info *yy_state_type;" );
	else if ( ! C_plus_plus )
		outn( "typedef int yy_state_type;" );

	if ( ddebug )
		outn( "\n#define FLEX_DEBUG" );

	if ( lex_compat )
		outn( "#define YY_FLEX_LEX_COMPAT" );

	if ( do_yylineno && ! C_plus_plus )
		{
		outn( "extern int yylineno;" );
		outn( "int yylineno = 1;" );
		}

	if ( C_plus_plus )
		{
		outn( "\n#include <FlexLexer.h>" );

		if ( yyclass )
			{
			outn( "int yyFlexLexer::yylex()" );
			outn( "\t{" );
			outn(
"\tLexerError( \"yyFlexLexer::yylex invoked but %option yyclass used\" );" );
			outn( "\treturn 0;" );
			outn( "\t}" );
	
			out_str( "\n#define YY_DECL int %s::yylex()\n",
				yyclass );
			}
		}

	else
		{
		if ( yytext_is_array )
			outn( "extern char yytext[];\n" );

		else
			{
			outn( "extern char *yytext;" );
			outn( "#define yytext_ptr yytext" );
			}

		if ( yyclass )
			flexerror(
		_( "%option yyclass only meaningful for C++ scanners" ) );
		}

	if ( useecs )
		numecs = cre8ecs( nextecm, ecgroup, csize );
	else
		numecs = csize;

	/* Now map the equivalence class for NUL to its expected place. */
	ecgroup[0] = ecgroup[csize];
	NUL_ec = ABS( ecgroup[0] );

	if ( useecs )
		ccl2ecl();
	}


/* set_up_initial_allocations - allocate memory for internal tables */

void set_up_initial_allocations()
	{
	current_mns = INITIAL_MNS;
	firstst = allocate_integer_array( current_mns );
	lastst = allocate_integer_array( current_mns );
	finalst = allocate_integer_array( current_mns );
	transchar = allocate_integer_array( current_mns );
	trans1 = allocate_integer_array( current_mns );
	trans2 = allocate_integer_array( current_mns );
	accptnum = allocate_integer_array( current_mns );
	assoc_rule = allocate_integer_array( current_mns );
	state_type = allocate_integer_array( current_mns );

	current_max_rules = INITIAL_MAX_RULES;
	rule_type = allocate_integer_array( current_max_rules );
	rule_linenum = allocate_integer_array( current_max_rules );
	rule_useful = allocate_integer_array( current_max_rules );

	current_max_scs = INITIAL_MAX_SCS;
	scset = allocate_integer_array( current_max_scs );
	scbol = allocate_integer_array( current_max_scs );
	scxclu = allocate_integer_array( current_max_scs );
	sceof = allocate_integer_array( current_max_scs );
	scname = allocate_char_ptr_array( current_max_scs );

	current_maxccls = INITIAL_MAX_CCLS;
	cclmap = allocate_integer_array( current_maxccls );
	ccllen = allocate_integer_array( current_maxccls );
	cclng = allocate_integer_array( current_maxccls );

	current_max_ccl_tbl_size = INITIAL_MAX_CCL_TBL_SIZE;
	ccltbl = allocate_Character_array( current_max_ccl_tbl_size );

	current_max_dfa_size = INITIAL_MAX_DFA_SIZE;

	current_max_xpairs = INITIAL_MAX_XPAIRS;
	nxt = allocate_integer_array( current_max_xpairs );
	chk = allocate_integer_array( current_max_xpairs );

	current_max_template_xpairs = INITIAL_MAX_TEMPLATE_XPAIRS;
	tnxt = allocate_integer_array( current_max_template_xpairs );

	current_max_dfas = INITIAL_MAX_DFAS;
	base = allocate_integer_array( current_max_dfas );
	def = allocate_integer_array( current_max_dfas );
	dfasiz = allocate_integer_array( current_max_dfas );
	accsiz = allocate_integer_array( current_max_dfas );
	dhash = allocate_integer_array( current_max_dfas );
	dss = allocate_int_ptr_array( current_max_dfas );
	dfaacc = allocate_dfaacc_union( current_max_dfas );

	nultrans = (int *) 0;
	}


void usage()
	{
	FILE *f = stdout;

	fprintf( f,
_( "%s [-bcdfhilnpstvwBFILTV78+? -C[aefFmr] -ooutput -Pprefix -Sskeleton]\n" ),
		program_name );
	fprintf( f, _( "\t[--help --version] [file ...]\n" ) );

	fprintf( f, _( "\t-b  generate backing-up information to %s\n" ),
		backing_name );
	fprintf( f, _( "\t-c  do-nothing POSIX option\n" ) );
	fprintf( f, _( "\t-d  turn on debug mode in generated scanner\n" ) );
	fprintf( f, _( "\t-f  generate fast, large scanner\n" ) );
	fprintf( f, _( "\t-h  produce this help message\n" ) );
	fprintf( f, _( "\t-i  generate case-insensitive scanner\n" ) );
	fprintf( f, _( "\t-l  maximal compatibility with original lex\n" ) );
	fprintf( f, _( "\t-n  do-nothing POSIX option\n" ) );
	fprintf( f, _( "\t-p  generate performance report to stderr\n" ) );
	fprintf( f,
		_( "\t-s  suppress default rule to ECHO unmatched text\n" ) );

	if ( ! did_outfilename )
		{
		sprintf( outfile_path, outfile_template,
			prefix, C_plus_plus ? "cc" : "c" );
		outfilename = outfile_path;
		}

	fprintf( f,
		_( "\t-t  write generated scanner on stdout instead of %s\n" ),
		outfilename );

	fprintf( f,
		_( "\t-v  write summary of scanner statistics to f\n" ) );
	fprintf( f, _( "\t-w  do not generate warnings\n" ) );
	fprintf( f, _( "\t-B  generate batch scanner (opposite of -I)\n" ) );
	fprintf( f,
		_( "\t-F  use alternative fast scanner representation\n" ) );
	fprintf( f,
		_( "\t-I  generate interactive scanner (opposite of -B)\n" ) );
	fprintf( f, _( "\t-L  suppress #line directives in scanner\n" ) );
	fprintf( f, _( "\t-T  %s should run in trace mode\n" ), program_name );
	fprintf( f, _( "\t-V  report %s version\n" ), program_name );
	fprintf( f, _( "\t-7  generate 7-bit scanner\n" ) );
	fprintf( f, _( "\t-8  generate 8-bit scanner\n" ) );
	fprintf( f, _( "\t-+  generate C++ scanner class\n" ) );
	fprintf( f, _( "\t-?  produce this help message\n" ) );
	fprintf( f,
_( "\t-C  specify degree of table compression (default is -Cem):\n" ) );
	fprintf( f,
_( "\t\t-Ca  trade off larger tables for better memory alignment\n" ) );
	fprintf( f, _( "\t\t-Ce  construct equivalence classes\n" ) );
	fprintf( f,
_( "\t\t-Cf  do not compress scanner tables; use -f representation\n" ) );
	fprintf( f,
_( "\t\t-CF  do not compress scanner tables; use -F representation\n" ) );
	fprintf( f, _( "\t\t-Cm  construct meta-equivalence classes\n" ) );
	fprintf( f,
	_( "\t\t-Cr  use read() instead of stdio for scanner input\n" ) );
	fprintf( f, _( "\t-o  specify output filename\n" ) );
	fprintf( f, _( "\t-P  specify scanner prefix other than \"yy\"\n" ) );
	fprintf( f, _( "\t-S  specify skeleton file\n" ) );
	fprintf( f, _( "\t--help     produce this help message\n" ) );
	fprintf( f, _( "\t--version  report %s version\n" ), program_name );
	}