scanner.cc

a little DFA compiler.

/* Generated by re2c 0.12.2 on Wed Jun 27 00:25:02 2007 */
/* $Id: scanner.cc 768 2007-06-26 22:26:06Z helly $ */
#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		unsigned 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;
}



int Scanner::echo()
{
    char *cursor = cur;
    bool ignore_eoc = false;
    int  ignore_cnt = 0;

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

    tok = cursor;
echo:
{

	{
		YYCTYPE yych;
		unsigned int yyaccept = 0;

		if((YYLIMIT - YYCURSOR) < 16) YYFILL(16);
		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;
		}
		yyaccept = 0;
		yych = *(YYMARKER = ++YYCURSOR);
		if(yych == '*') goto yy16;
yy3:
		{
					goto echo;
				}
yy4:
		yych = *++YYCURSOR;
		if(yych == '/') goto yy10;
		goto yy3;
yy5:
		++YYCURSOR;
		{
					if (ignore_eoc) {
						ignore_cnt++;
					} else {
						out.write((const char*)(tok), (const char*)(cursor) - (const char*)(tok));
					}
					tok = pos = cursor; cline++;
				  	goto echo;
				}
yy7:
		++YYCURSOR;
		{
					if (!ignore_eoc) {
						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);
					}
				}
yy9:
		yych = *++YYCURSOR;
		goto yy3;
yy10:
		yyaccept = 1;
		yych = *(YYMARKER = ++YYCURSOR);
		if(yych == 0x0A) goto yy14;
		if(yych == 0x0D) goto yy12;
yy11:
		{
					if (ignore_eoc) {
						if (ignore_cnt) {
							out << "\n" << sourceFileInfo;
						}
						ignore_eoc = false;
						ignore_cnt = 0;
					} else {
						out.write((const char*)(tok), (const char*)(cursor) - (const char*)(tok));
					}
					tok = pos = cursor;
					goto echo;
				}
yy12:
		yych = *++YYCURSOR;
		if(yych == 0x0A) goto yy14;
yy13:
		YYCURSOR = YYMARKER;
		if(yyaccept <= 0) {
			goto yy3;
		} else {
			goto yy11;
		}
yy14:
		++YYCURSOR;
		{
					cline++;
					if (ignore_eoc) {
						if (ignore_cnt) {
							out << sourceFileInfo;
						}
						ignore_eoc = false;
						ignore_cnt = 0;
					} else {
						out.write((const char*)(tok), (const char*)(cursor) - (const char*)(tok));
					}
					tok = pos = cursor;
					goto echo;
				}
yy16:
		yych = *++YYCURSOR;
		if(yych != '!') goto yy13;
		yych = *++YYCURSOR;
		switch(yych) {
		case 'g':	goto yy19;
		case 'i':	goto yy18;
		case 'm':	goto yy20;
		case 'r':	goto yy21;
		default:	goto yy13;
		}
yy18:
		yych = *++YYCURSOR;
		if(yych == 'g') goto yy47;
		goto yy13;
yy19:
		yych = *++YYCURSOR;
		if(yych == 'e') goto yy34;
		goto yy13;
yy20:
		yych = *++YYCURSOR;
		if(yych == 'a') goto yy26;
		goto yy13;
yy21:
		yych = *++YYCURSOR;
		if(yych != 'e') goto yy13;
		yych = *++YYCURSOR;
		if(yych != '2') goto yy13;
		yych = *++YYCURSOR;
		if(yych != 'c') goto yy13;
		++YYCURSOR;
		{
					if (bUsedYYMaxFill && bSinglePass) {
						fatal("found scanner block after YYMAXFILL declaration");
					}
					out.write((const char*)(tok), (const char*)(&cursor[-7]) - (const char*)(tok));
					tok = cursor;
					RETURN(1);
				}
yy26:
		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;
		{
					if (bUsedYYMaxFill) {
						fatal("cannot generate YYMAXFILL twice");
					}
					out << "#define YYMAXFILL " << maxFill << std::endl;
					tok = pos = cursor;
					ignore_eoc = true;
					bUsedYYMaxFill = true;
					goto echo;
				}
yy34:
		yych = *++YYCURSOR;
		if(yych != 't') goto yy13;
		yych = *++YYCURSOR;
		if(yych != 's') goto yy13;
		yych = *++YYCURSOR;
		if(yych != 't') goto yy13;
		yych = *++YYCURSOR;
		if(yych != 'a') goto yy13;
		yych = *++YYCURSOR;
		if(yych != 't') goto yy13;
		yych = *++YYCURSOR;
		if(yych != 'e') 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;
		{
					tok = pos = cursor;
					genGetState(out, topIndent, 0);
					ignore_eoc = true;
					goto echo;
				}
yy47:
		yych = *++YYCURSOR;
		if(yych != 'n') goto yy13;
		yych = *++YYCURSOR;
		if(yych != 'o') goto yy13;
		yych = *++YYCURSOR;
		if(yych != 'r') goto yy13;
		yych = *++YYCURSOR;
		if(yych != 'e') 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;
		{
					tok = pos = cursor;
					ignore_eoc = true;
					goto echo;
				}
	}
}

}


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;
    }
{
	static const unsigned char yybm[] = {
		112, 112, 112, 112, 112, 112, 112, 112, 
		112, 116,   0, 112, 112, 112, 112, 112, 
		112, 112, 112, 112, 112, 112, 112, 112, 
		112, 112, 112, 112, 112, 112, 112, 112, 
		116, 112,  48, 112, 112, 112, 112,  80, 
		112, 112, 112, 112, 112, 112, 112, 112, 
		248, 248, 248, 248, 248, 248, 248, 248, 
		248, 248, 112, 112, 112, 112, 112, 112, 
		112, 120, 120, 120, 120, 120, 120, 120, 
		120, 120, 120, 120, 120, 120, 120, 120, 
		120, 120, 120, 120, 120, 120, 120, 120, 
		120, 120, 120, 112,   0,  96, 112, 120, 
		112, 120, 120, 120, 120, 120, 120, 120, 
		120, 120, 120, 120, 120, 120, 120, 120, 
		120, 120, 120, 120, 120, 120, 120, 120, 
		120, 120, 120, 112, 112, 112, 112, 112, 
		112, 112, 112, 112, 112, 112, 112, 112, 
		112, 112, 112, 112, 112, 112, 112, 112, 
		112, 112, 112, 112, 112, 112, 112, 112, 
		112, 112, 112, 112, 112, 112, 112, 112, 
		112, 112, 112, 112, 112, 112, 112, 112, 
		112, 112, 112, 112, 112, 112, 112, 112, 
		112, 112, 112, 112, 112, 112, 112, 112, 
		112, 112, 112, 112, 112, 112, 112, 112, 
		112, 112, 112, 112, 112, 112, 112, 112, 
		112, 112, 112, 112, 112, 112, 112, 112, 
		112, 112, 112, 112, 112, 112, 112, 112, 
		112, 112, 112, 112, 112, 112, 112, 112, 
		112, 112, 112, 112, 112, 112, 112, 112, 
		112, 112, 112, 112, 112, 112, 112, 112, 
		112, 112, 112, 112, 112, 112, 112, 112, 
		112, 112, 112, 112, 112, 112, 112, 112, 
	};

	{
		YYCTYPE yych;
		unsigned int yyaccept = 0;
		if((YYLIMIT - YYCURSOR) < 5) YYFILL(5);
		yych = *YYCURSOR;
		if(yych <= ':') {
			if(yych <= '"') {
				if(yych <= 0x0C) {
					if(yych <= 0x08) goto yy85;
					if(yych <= 0x09) goto yy79;
					if(yych <= 0x0A) goto yy81;
					goto yy85;
				} else {
					if(yych <= 0x1F) {
						if(yych <= 0x0D) goto yy83;
						goto yy85;
					} else {
						if(yych <= ' ') goto yy79;
						if(yych <= '!') goto yy85;
						goto yy66;
					}
				}
			} else {
				if(yych <= '*') {
					if(yych <= '&') goto yy85;
					if(yych <= '\'') goto yy68;
					if(yych <= ')') goto yy72;
					goto yy64;
				} else {
					if(yych <= '-') {
						if(yych <= '+') goto yy73;
						goto yy85;
					} else {
						if(yych <= '.') goto yy77;
						if(yych <= '/') goto yy62;
						goto yy85;
					}
				}
			}
		} else {
			if(yych <= '\\') {
				if(yych <= '>') {
					if(yych == '<') goto yy85;
					if(yych <= '=') goto yy72;
					goto yy85;
				} else {
					if(yych <= '@') {
						if(yych <= '?') goto yy73;
						goto yy85;
					} else {
						if(yych <= 'Z') goto yy76;
						if(yych <= '[') goto yy70;
						goto yy72;
					}
				}
			} else {
				if(yych <= 'q') {
					if(yych == '_') goto yy76;
					if(yych <= '`') goto yy85;
					goto yy76;
				} else {
					if(yych <= 'z') {
						if(yych <= 'r') goto yy74;
						goto yy76;
					} else {
						if(yych <= '{') goto yy60;
						if(yych <= '|') goto yy72;
						goto yy85;
					}
				}
			}
		}
yy60:
		yyaccept = 0;
		yych = *(YYMARKER = ++YYCURSOR);
		if(yych <= '/') {
			if(yych == ',') goto yy126;
		} else {
			if(yych <= '0') goto yy123;
			if(yych <= '9') goto yy124;
		}
yy61:
		{ depth = 1;
				  goto code;
				}
yy62:
		++YYCURSOR;
		if((yych = *YYCURSOR) == '*') goto yy121;
yy63:
		{ RETURN(*tok); }
yy64:
		++YYCURSOR;
		if((yych = *YYCURSOR) == '/') goto yy119;
yy65:
		{ yylval.op = *tok;
				  RETURN(CLOSE); }
yy66:
		yyaccept = 1;
		yych = *(YYMARKER = ++YYCURSOR);
		if(yych != 0x0A) goto yy115;
yy67:
		{ fatal("unterminated string constant (missing \")"); }
yy68:
		yyaccept = 2;
		yych = *(YYMARKER = ++YYCURSOR);
		if(yych != 0x0A) goto yy110;
yy69:
		{ fatal("unterminated string constant (missing ')"); }
yy70:
		yyaccept = 3;
		yych = *(YYMARKER = ++YYCURSOR);
		if(yych == 0x0A) goto yy71;
		if(yych == '^') goto yy101;
		goto yy100;
yy71:
		{ fatal("unterminated range (missing ])"); }
yy72:
		yych = *++YYCURSOR;
		goto yy63;
yy73:
		yych = *++YYCURSOR;
		goto yy65;
yy74:
		++YYCURSOR;
		if((yych = *YYCURSOR) == 'e') goto yy91;
		goto yy90;
yy75:
		{ cur = cursor;
				  yylval.symbol = Symbol::find(token());
				  return ID; }
yy76:
		yych = *++YYCURSOR;
		goto yy90;
yy77:
		++YYCURSOR;
		{ cur = cursor;
				  yylval.regexp = mkDot();
				  return RANGE;
				}
yy79:
		++YYCURSOR;
		yych = *YYCURSOR;
		goto yy88;
yy80:
		{ goto scan; }
yy81:
		++YYCURSOR;
yy82:
		{ if(cursor == eof) RETURN(0);
				  pos = cursor; cline++;
				  goto scan;
	    			}
yy83:
		++YYCURSOR;
		if((yych = *YYCURSOR) == 0x0A) goto yy86;
yy84:
		{ std::ostringstream msg;
				  msg << "unexpected character: ";
				  prtChOrHex(msg, *tok);
				  fatal(msg.str().c_str());
				  goto scan;
				}
yy85:
		yych = *++YYCURSOR;
		goto yy84;
yy86:
		yych = *++YYCURSOR;
		goto yy82;
yy87:
		++YYCURSOR;
		if(YYLIMIT == YYCURSOR) YYFILL(1);
		yych = *YYCURSOR;
yy88:
		if(yybm[0+yych] & 4) {
			goto yy87;
		}
		goto yy80;
yy89:
		++YYCURSOR;
		if(YYLIMIT == YYCURSOR) YYFILL(1);
		yych = *YYCURSOR;
yy90:
		if(yybm[0+yych] & 8) {
			goto yy89;
		}
		goto yy75;
yy91:
		yych = *++YYCURSOR;
		if(yych != '2') goto yy90;
		yych = *++YYCURSOR;
		if(yych != 'c') goto yy90;
		yyaccept = 4;
		yych = *(YYMARKER = ++YYCURSOR);
		if(yych != ':') goto yy90;
yy94:
		++YYCURSOR;
		if(YYLIMIT == YYCURSOR) YYFILL(1);
		yych = *YYCURSOR;
		if(yych <= '^') {
			if(yych <= '@') goto yy95;
			if(yych <= 'Z') goto yy96;
		} else {
			if(yych == '`') goto yy95;
			if(yych <= 'z') goto yy96;
		}
yy95:
		YYCURSOR = YYMARKER;
		if(yyaccept <= 3) {
			if(yyaccept <= 1) {
				if(yyaccept <= 0) {
					goto yy61;
				} else {
					goto yy67;
				}
			} else {
				if(yyaccept <= 2) {
					goto yy69;
				} else {
					goto yy71;
				}
			}
		} else {
			if(yyaccept <= 5) {
				if(yyaccept <= 4) {
					goto yy75;
				} else {
					goto yy98;
				}
			} else {
				goto yy127;
			}
		}
yy96:
		yyaccept = 5;
		YYMARKER = ++YYCURSOR;
		if(YYLIMIT == YYCURSOR) YYFILL(1);
		yych = *YYCURSOR;
		if(yych <= 'Z') {
			if(yych <= '9') {
				if(yych >= '0') goto yy96;
			} else {
				if(yych <= ':') goto yy94;
				if(yych >= 'A') goto yy96;
			}
		} else {
			if(yych <= '_') {
				if(yych >= '_') goto yy96;
			} else {
				if(yych <= '`') goto yy98;
				if(yych <= 'z') goto yy96;
			}
		}
yy98:
		{ cur = cursor;
				  tok+= 5; /* skip "re2c:" */
				  iscfg = 1;
				  yylval.str = new Str(token());
				  return CONFIG;
				}
yy99:
		++YYCURSOR;
		if(YYLIMIT == YYCURSOR) YYFILL(1);
		yych = *YYCURSOR;
yy100:
		if(yybm[0+yych] & 16) {
			goto yy99;
		}
		if(yych <= '[') goto yy95;
		if(yych <= '\\') goto yy103;
		goto yy104;
yy101:
		++YYCURSOR;
		if(YYLIMIT == YYCURSOR) YYFILL(1);
		yych = *YYCURSOR;
		if(yych <= '[') {
			if(yych == 0x0A) goto yy95;
			goto yy101;
		} else {
			if(yych <= '\\') goto yy106;
			if(yych <= ']') goto yy107;
			goto yy101;
		}
yy103: