scanner.fs.c

a little DFA compiler.

/* Generated by re2c */
#line 1 "scanner.fs.re"
/* $Id: scanner.re,v 1.37 2006/02/25 12:41:41 helly Exp $ */
#include <stdlib.h>
#include <string.h>
#include <iostream>
#include <sstream>
#include "scanner.h"
#include "parser.h"
#include "y.tab.h"
#include "globals.h"
#include "dfa.h"

extern YYSTYPE yylval;

#ifndef MAX
#define MAX(a,b) (((a)>(b))?(a):(b))
#endif

#define	BSIZE	8192

#define	YYCTYPE		char
#define	YYCURSOR	cursor
#define	YYLIMIT		lim
#define	YYMARKER	ptr
#define	YYFILL(n)	{cursor = fill(cursor);}

#define	RETURN(i)	{cur = cursor; return i;}

namespace re2c
{

Scanner::Scanner(std::istream& i, std::ostream& o)
	: in(i)
	, out(o)
	, bot(NULL), tok(NULL), ptr(NULL), cur(NULL), pos(NULL), lim(NULL)
	, top(NULL), eof(NULL), tchar(0), tline(0), cline(1), iscfg(0)
{
    ;
}

char *Scanner::fill(char *cursor)
{
	if(!eof)
	{
		uint cnt = tok - bot;
		if(cnt)
		{
			memcpy(bot, tok, lim - tok);
			tok = bot;
			ptr -= cnt;
			cursor -= cnt;
			pos -= cnt;
			lim -= cnt;
		}
		if((top - lim) < BSIZE)
		{
			char *buf = new char[(lim - bot) + BSIZE];
			memcpy(buf, tok, lim - tok);
			tok = buf;
			ptr = &buf[ptr - bot];
			cursor = &buf[cursor - bot];
			pos = &buf[pos - bot];
			lim = &buf[lim - bot];
			top = &lim[BSIZE];
			delete [] bot;
			bot = buf;
		}
		in.read(lim, BSIZE);
		if ((cnt = in.gcount()) != BSIZE )
		{
			eof = &lim[cnt]; *eof++ = '\0';
		}
		lim += cnt;
	}
	return cursor;
}

#line 95 "scanner.fs.re"


int Scanner::echo()
{
    char *cursor = cur;
    bool ignore_eoc = false;

    if (eof && cursor == eof) // Catch EOF
	{
    	return 0;
	}

    tok = cursor;
echo:

#line 96 "<stdout>"
{

	switch(YYGETSTATE()) {
	default: goto yy0;
	case 0: goto yyFillLabel0;
	case 1: goto yyFillLabel1;
	case 2: goto yyFillLabel2;
	case 3: goto yyFillLabel3;
	case 4: goto yyFillLabel4;
	case 5: goto yyFillLabel5;
	case 6: goto yyFillLabel6;
	case 7: goto yyFillLabel7;
	case 8: goto yyFillLabel8;
	case 9: goto yyFillLabel9;
	case 10: goto yyFillLabel10;
	case 11: goto yyFillLabel11;
	case 12: goto yyFillLabel12;
	case 13: goto yyFillLabel13;
	case 14: goto yyFillLabel14;
	case 15: goto yyFillLabel15;
	case 16: goto yyFillLabel16;
	case 17: goto yyFillLabel17;
	case 18: goto yyFillLabel18;
	case 19: goto yyFillLabel19;
	case 20: goto yyFillLabel20;
	case 21: goto yyFillLabel21;
	case 22: goto yyFillLabel22;
	case 23: goto yyFillLabel23;
	case 24: goto yyFillLabel24;
	case 25: goto yyFillLabel25;
	case 26: goto yyFillLabel26;
	case 27: goto yyFillLabel27;
	case 28: goto yyFillLabel28;
	case 29: goto yyFillLabel29;
	case 30: goto yyFillLabel30;
	case 31: goto yyFillLabel31;
	case 32: goto yyFillLabel32;
	case 33: goto yyFillLabel33;
	case 34: goto yyFillLabel34;
	case 35: goto yyFillLabel35;
	}
yy0:
	YYSETSTATE(0);
	if((YYLIMIT - YYCURSOR) < 11) YYFILL(11);
yyFillLabel0:
	yych = *YYCURSOR;
	if(yych <= ')') {
		if(yych <= 0x00) goto yy7;
		if(yych == 0x0A) goto yy5;
		goto yy9;
	} else {
		if(yych <= '*') goto yy4;
		if(yych != '/') goto yy9;
	}
	yych = *(YYMARKER = ++YYCURSOR);
	if(yych == '*') goto yy12;
yy3:
#line 141 "scanner.fs.re"
	{
					goto echo;
				}
#line 158 "<stdout>"
yy4:
	yych = *++YYCURSOR;
	if(yych == '/') goto yy10;
	goto yy3;
yy5:
	++YYCURSOR;
#line 130 "scanner.fs.re"
	{
					out.write((const char*)(tok), (const char*)(cursor) - (const char*)(tok));
					tok = pos = cursor; cline++;
				  	goto echo;
				}
#line 171 "<stdout>"
yy7:
	++YYCURSOR;
#line 135 "scanner.fs.re"
	{
					out.write((const char*)(tok), (const char*)(cursor) - (const char*)(tok) - 1); // -1 so we don't write out the \0
					if(cursor == eof) {
						RETURN(0);
					}
				}
#line 181 "<stdout>"
yy9:
	yych = *++YYCURSOR;
	goto yy3;
yy10:
	++YYCURSOR;
#line 121 "scanner.fs.re"
	{
					if (ignore_eoc) {
						ignore_eoc = false;
					} else {
						out.write((const char*)(tok), (const char*)(cursor) - (const char*)(tok));
					}
					tok = pos = cursor;
					goto echo;
				}
#line 197 "<stdout>"
yy12:
	yych = *++YYCURSOR;
	if(yych == '!') goto yy14;
yy13:
	YYCURSOR = YYMARKER;
	goto yy3;
yy14:
	yych = *++YYCURSOR;
	if(yych == 'm') goto yy15;
	if(yych == 'r') goto yy16;
	goto yy13;
yy15:
	yych = *++YYCURSOR;
	if(yych == 'a') goto yy21;
	goto yy13;
yy16:
	yych = *++YYCURSOR;
	if(yych != 'e') goto yy13;
	yych = *++YYCURSOR;
	if(yych != '2') goto yy13;
	yych = *++YYCURSOR;
	if(yych != 'c') goto yy13;
	++YYCURSOR;
#line 110 "scanner.fs.re"
	{ 
					out.write((const char*)(tok), (const char*)(&cursor[-7]) - (const char*)(tok));
					tok = cursor;
					RETURN(1);
				}
#line 227 "<stdout>"
yy21:
	yych = *++YYCURSOR;
	if(yych != 'x') goto yy13;
	yych = *++YYCURSOR;
	if(yych != ':') goto yy13;
	yych = *++YYCURSOR;
	if(yych != 'r') goto yy13;
	yych = *++YYCURSOR;
	if(yych != 'e') goto yy13;
	yych = *++YYCURSOR;
	if(yych != '2') goto yy13;
	yych = *++YYCURSOR;
	if(yych != 'c') goto yy13;
	++YYCURSOR;
#line 115 "scanner.fs.re"
	{
					out << "#define YYMAXFILL " << maxFill << std::endl;
					tok = pos = cursor;
					ignore_eoc = true;
					goto echo;
				}
#line 249 "<stdout>"
}
#line 144 "scanner.fs.re"

}


int Scanner::scan()
{
    char *cursor = cur;
    uint depth;

scan:
    tchar = cursor - pos;
    tline = cline;
    tok = cursor;
	if (iscfg == 1)
	{
		goto config;
	}
	else if (iscfg == 2)
	{
   		goto value;
    }

#line 274 "<stdout>"
{

	YYSETSTATE(1);
	if((YYLIMIT - YYCURSOR) < 5) YYFILL(5);
yyFillLabel1:
	yych = *YYCURSOR;
	if(yych <= '/') {
		if(yych <= '!') {
			if(yych <= 0x0C) {
				if(yych <= 0x08) goto yy56;
				if(yych <= 0x09) goto yy50;
				if(yych <= 0x0A) goto yy52;
				goto yy56;
			} else {
				if(yych <= 0x0D) goto yy54;
				if(yych == ' ') goto yy50;
				goto yy56;
			}
		} else {
			if(yych <= ')') {
				if(yych <= '"') goto yy37;
				if(yych <= '&') goto yy56;
				if(yych <= '\'') goto yy39;
				goto yy43;
			} else {
				if(yych <= '+') {
					if(yych <= '*') goto yy35;
					goto yy44;
				} else {
					if(yych <= '-') goto yy56;
					if(yych <= '.') goto yy48;
					goto yy33;
				}
			}
		}
	} else {
		if(yych <= 'Z') {
			if(yych <= '=') {
				if(yych == ';') goto yy43;
				if(yych <= '<') goto yy56;
				goto yy43;
			} else {
				if(yych == '?') goto yy44;
				if(yych <= '@') goto yy56;
				goto yy47;
			}
		} else {
			if(yych <= 'q') {
				if(yych <= '[') goto yy41;
				if(yych <= '\\') goto yy43;
				if(yych <= '`') goto yy56;
				goto yy47;
			} else {
				if(yych <= 'z') {
					if(yych <= 'r') goto yy45;
					goto yy47;
				} else {
					if(yych <= '{') goto yy31;
					if(yych <= '|') goto yy43;
					goto yy56;
				}
			}
		}
	}
yy31:
	yyaccept = 0;
	yych = *(YYMARKER = ++YYCURSOR);
	if(yych <= '/') {
		if(yych == ',') goto yy97;
	} else {
		if(yych <= '0') goto yy94;
		if(yych <= '9') goto yy95;
	}
yy32:
#line 166 "scanner.fs.re"
	{ depth = 1;
				  goto code;
				}
#line 353 "<stdout>"
yy33:
	++YYCURSOR;
	if((yych = *YYCURSOR) == '*') goto yy92;
yy34:
#line 196 "scanner.fs.re"
	{ RETURN(*tok); }
#line 360 "<stdout>"
yy35:
	++YYCURSOR;
	if((yych = *YYCURSOR) == '/') goto yy90;
yy36:
#line 198 "scanner.fs.re"
	{ yylval.op = *tok;
				  RETURN(CLOSE); }
#line 368 "<stdout>"
yy37:
	yyaccept = 1;
	yych = *(YYMARKER = ++YYCURSOR);
	if(yych != 0x0A) goto yy86;
yy38:
#line 183 "scanner.fs.re"
	{ fatal("unterminated string constant (missing \")"); }
#line 376 "<stdout>"
yy39:
	yyaccept = 2;
	yych = *(YYMARKER = ++YYCURSOR);
	if(yych != 0x0A) goto yy81;
yy40:
#line 184 "scanner.fs.re"
	{ fatal("unterminated string constant (missing ')"); }
#line 384 "<stdout>"
yy41:
	yyaccept = 3;
	yych = *(YYMARKER = ++YYCURSOR);
	if(yych == 0x0A) goto yy42;
	if(yych == '^') goto yy72;
	goto yy71;
yy42:
#line 194 "scanner.fs.re"
	{ fatal("unterminated range (missing ])"); }
#line 394 "<stdout>"
yy43:
	yych = *++YYCURSOR;
	goto yy34;
yy44:
	yych = *++YYCURSOR;
	goto yy36;
yy45:
	++YYCURSOR;
	if((yych = *YYCURSOR) == 'e') goto yy62;
	goto yy61;
yy46:
#line 225 "scanner.fs.re"
	{ cur = cursor;
				  yylval.symbol = Symbol::find(token());
				  return ID; }
#line 410 "<stdout>"
yy47:
	yych = *++YYCURSOR;
	goto yy61;
yy48:
	++YYCURSOR;
#line 229 "scanner.fs.re"
	{ cur = cursor;
				  yylval.regexp = mkDot();
				  return RANGE;
				}
#line 421 "<stdout>"
yy50:
	++YYCURSOR;
	yych = *YYCURSOR;
	goto yy59;
yy51:
#line 234 "scanner.fs.re"
	{ goto scan; }
#line 429 "<stdout>"
yy52:
	++YYCURSOR;
yy53:
#line 236 "scanner.fs.re"
	{ if(cursor == eof) RETURN(0);
				  pos = cursor; cline++;
				  goto scan;
	    			}
#line 438 "<stdout>"
yy54:
	++YYCURSOR;
	if((yych = *YYCURSOR) == 0x0A) goto yy57;
yy55:
#line 241 "scanner.fs.re"
	{ std::ostringstream msg;
				  msg << "unexpected character: ";
				  prtChOrHex(msg, *tok);
				  fatal(msg.str().c_str());
				  goto scan;
				}
#line 450 "<stdout>"
yy56:
	yych = *++YYCURSOR;
	goto yy55;
yy57:
	yych = *++YYCURSOR;
	goto yy53;
yy58:
	++YYCURSOR;
	YYSETSTATE(2);
	if(YYLIMIT == YYCURSOR) YYFILL(1);
yyFillLabel2:
	yych = *YYCURSOR;
yy59:
	if(yych == 0x09) goto yy58;
	if(yych == ' ') goto yy58;
	goto yy51;
yy60:
	++YYCURSOR;
	YYSETSTATE(3);
	if(YYLIMIT == YYCURSOR) YYFILL(1);
yyFillLabel3:
	yych = *YYCURSOR;
yy61:
	if(yych <= '@') {
		if(yych <= '/') goto yy46;
		if(yych <= '9') goto yy60;
		goto yy46;
	} else {
		if(yych <= 'Z') goto yy60;
		if(yych <= '`') goto yy46;
		if(yych <= 'z') goto yy60;
		goto yy46;
	}
yy62:
	yych = *++YYCURSOR;
	if(yych != '2') goto yy61;
	yych = *++YYCURSOR;
	if(yych != 'c') goto yy61;
	yyaccept = 4;
	yych = *(YYMARKER = ++YYCURSOR);
	if(yych != ':') goto yy61;
yy65:
	++YYCURSOR;
	YYSETSTATE(4);
	if(YYLIMIT == YYCURSOR) YYFILL(1);
yyFillLabel4:
	yych = *YYCURSOR;
	if(yych <= '@') goto yy66;
	if(yych <= 'Z') goto yy67;
	if(yych <= '`') goto yy66;
	if(yych <= 'z') goto yy67;
yy66:
	YYCURSOR = YYMARKER;
	if(yyaccept <= 3) {
		if(yyaccept <= 1) {
			if(yyaccept <= 0) {
				goto yy32;
			} else {
				goto yy38;
			}
		} else {
			if(yyaccept <= 2) {
				goto yy40;
			} else {
				goto yy42;
			}
		}
	} else {
		if(yyaccept <= 5) {
			if(yyaccept <= 4) {
				goto yy46;
			} else {
				goto yy69;
			}
		} else {
			goto yy98;
		}
	}
yy67:
	yyaccept = 5;
	YYMARKER = ++YYCURSOR;
	YYSETSTATE(5);
	if(YYLIMIT == YYCURSOR) YYFILL(1);
yyFillLabel5:
	yych = *YYCURSOR;
	if(yych <= 'Z') {
		if(yych <= '9') {
			if(yych >= '0') goto yy67;
		} else {
			if(yych <= ':') goto yy65;
			if(yych >= 'A') goto yy67;
		}
	} else {
		if(yych <= '_') {
			if(yych >= '_') goto yy67;
		} else {
			if(yych <= '`') goto yy69;
			if(yych <= 'z') goto yy67;
		}
	}
yy69:
#line 218 "scanner.fs.re"
	{ cur = cursor;
				  tok+= 5; /* skip "re2c:" */
				  iscfg = 1;
				  yylval.str = new Str(token());
				  return CONFIG;
				}
#line 559 "<stdout>"
yy70:
	++YYCURSOR;
	YYSETSTATE(6);
	if(YYLIMIT == YYCURSOR) YYFILL(1);
yyFillLabel6:
	yych = *YYCURSOR;
yy71:
	if(yych <= '[') {
		if(yych == 0x0A) goto yy66;
		goto yy70;
	} else {
		if(yych <= '\\') goto yy74;
		if(yych <= ']') goto yy75;
		goto yy70;
	}
yy72:
	++YYCURSOR;
	YYSETSTATE(7);
	if(YYLIMIT == YYCURSOR) YYFILL(1);
yyFillLabel7:
	yych = *YYCURSOR;
	if(yych <= '[') {
		if(yych == 0x0A) goto yy66;
		goto yy72;
	} else {
		if(yych <= '\\') goto yy77;
		if(yych <= ']') goto yy78;
		goto yy72;
	}
yy74:
	++YYCURSOR;
	YYSETSTATE(8);
	if(YYLIMIT == YYCURSOR) YYFILL(1);
yyFillLabel8:
	yych = *YYCURSOR;
	if(yych == 0x0A) goto yy66;
	goto yy70;
yy75:
	++YYCURSOR;
#line 190 "scanner.fs.re"
	{ cur = cursor;
				  yylval.regexp = ranToRE(token());
				  return RANGE; }
#line 603 "<stdout>"
yy77:
	++YYCURSOR;
	YYSETSTATE(9);
	if(YYLIMIT == YYCURSOR) YYFILL(1);
yyFillLabel9:
	yych = *YYCURSOR;
	if(yych == 0x0A) goto yy66;
	goto yy72;
yy78:
	++YYCURSOR;
#line 186 "scanner.fs.re"
	{ cur = cursor;
				  yylval.regexp = invToRE(token());
				  return RANGE; }
#line 618 "<stdout>"
yy80:
	++YYCURSOR;
	YYSETSTATE(10);
	if(YYLIMIT == YYCURSOR) YYFILL(1);
yyFillLabel10:
	yych = *YYCURSOR;
yy81:
	if(yych <= '&') {
		if(yych == 0x0A) goto yy66;
		goto yy80;
	} else {
		if(yych <= '\'') goto yy83;
		if(yych != '\\') goto yy80;
	}
	++YYCURSOR;
	YYSETSTATE(11);
	if(YYLIMIT == YYCURSOR) YYFILL(1);
yyFillLabel11: