home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Aminet 24
/
Aminet 24 (1998)(GTI - Schatztruhe)[!][Apr 1998].iso
/
Aminet
/
dev
/
c
/
vbcc.lha
/
vbcc
/
vcpp
/
lex.c
< prev
next >
Wrap
C/C++ Source or Header
|
1997-12-30
|
12KB
|
572 lines
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "cpp.h"
/*
* lexical FSM encoding
* when in state state, and one of the characters
* in ch arrives, enter nextstate.
* States >= S_SELF are either final, or at least require special action.
* In 'fsm' there is a line for each state X charset X nextstate.
* List chars that overwrite previous entries later (e.g. C_ALPH
* can be overridden by '_' by a later entry; and C_XX is the
* the universal set, and should always be first.
* States above S_SELF are represented in the big table as negative values.
* S_SELF and S_SELFB encode the resulting token type in the upper bits.
* These actions differ in that S_SELF doesn't have a lookahead char,
* S_SELFB does.
*
* The encoding is blown out into a big table for time-efficiency.
* Entries have
* nextstate: 6 bits; ?\ marker: 1 bit; tokentype: 9 bits.
*/
#define MAXSTATE 32
#define ACT(tok,act) ((tok<<7)+act)
#define QBSBIT 0100
#define GETACT(st) (st>>7)&0x1ff
/* character classes */
#define C_WS 1
#define C_ALPH 2
#define C_NUM 3
#define C_EOF 4
#define C_XX 5
enum state {
START=0, NUM1, NUM2, NUM3, ID1, ST1, ST2, ST3, COM1, COM2, COM3, COM4,
CC1, CC2, WS1, PLUS1, MINUS1, STAR1, SLASH1, PCT1, SHARP1,
CIRC1, GT1, GT2, LT1, LT2, OR1, AND1, ASG1, NOT1, DOTS1,
S_SELF=MAXSTATE, S_SELFB, S_EOF, S_NL, S_EOFSTR,
S_STNL, S_COMNL, S_EOFCOM, S_COMMENT, S_EOB, S_WS, S_NAME
};
int tottok;
int tokkind[256];
struct fsm {
int state; /* if in this state */
uchar ch[4]; /* and see one of these characters */
int nextstate; /* enter this state if +ve */
};
/*const*/ struct fsm fsm[] = {
/* start state */
START, { C_XX }, ACT(UNCLASS,S_SELF),
START, { ' ', '\t', '\v' }, WS1,
START, { C_NUM }, NUM1,
START, { '.' }, NUM3,
START, { C_ALPH }, ID1,
START, { 'L' }, ST1,
START, { '"' }, ST2,
START, { '\'' }, CC1,
START, { '/' }, COM1,
START, { EOFC }, S_EOF,
START, { '\n' }, S_NL,
START, { '-' }, MINUS1,
START, { '+' }, PLUS1,
START, { '<' }, LT1,
START, { '>' }, GT1,
START, { '=' }, ASG1,
START, { '!' }, NOT1,
START, { '&' }, AND1,
START, { '|' }, OR1,
START, { '#' }, SHARP1,
START, { '%' }, PCT1,
START, { '[' }, ACT(SBRA,S_SELF),
START, { ']' }, ACT(SKET,S_SELF),
START, { '(' }, ACT(LP,S_SELF),
START, { ')' }, ACT(RP,S_SELF),
START, { '*' }, STAR1,
START, { ',' }, ACT(COMMA,S_SELF),
START, { '?' }, ACT(QUEST,S_SELF),
START, { ':' }, ACT(COLON,S_SELF),
START, { ';' }, ACT(SEMIC,S_SELF),
START, { '{' }, ACT(CBRA,S_SELF),
START, { '}' }, ACT(CKET,S_SELF),
START, { '~' }, ACT(TILDE,S_SELF),
START, { '^' }, CIRC1,
/* saw a digit */
NUM1, { C_XX }, ACT(NUMBER,S_SELFB),
NUM1, { C_NUM, C_ALPH, '.' }, NUM1,
NUM1, { 'E', 'e' }, NUM2,
NUM1, { '_' }, ACT(NUMBER,S_SELFB),
/* saw possible start of exponent, digits-e */
NUM2, { C_XX }, ACT(NUMBER,S_SELFB),
NUM2, { '+', '-' }, NUM1,
NUM2, { C_NUM, C_ALPH }, NUM1,
NUM2, { '_' }, ACT(NUMBER,S_SELFB),
/* saw a '.', which could be a number or an operator */
NUM3, { C_XX }, ACT(DOT,S_SELFB),
NUM3, { '.' }, DOTS1,
NUM3, { C_NUM }, NUM1,
DOTS1, { C_XX }, ACT(UNCLASS, S_SELFB),
DOTS1, { C_NUM }, NUM1,
DOTS1, { '.' }, ACT(ELLIPS, S_SELF),
/* saw a letter or _ */
ID1, { C_XX }, ACT(NAME,S_NAME),
ID1, { C_ALPH, C_NUM }, ID1,
/* saw L (start of wide string?) */
ST1, { C_XX }, ACT(NAME,S_NAME),
ST1, { C_ALPH, C_NUM }, ID1,
ST1, { '"' }, ST2,
ST1, { '\'' }, CC1,
/* saw " beginning string */
ST2, { C_XX }, ST2,
ST2, { '"' }, ACT(STRING, S_SELF),
ST2, { '\\' }, ST3,
ST2, { '\n' }, S_STNL,
ST2, { EOFC }, S_EOFSTR,
/* saw \ in string */
ST3, { C_XX }, ST2,
ST3, { '\n' }, S_STNL,
ST3, { EOFC }, S_EOFSTR,
/* saw ' beginning character const */
CC1, { C_XX }, CC1,
CC1, { '\'' }, ACT(CCON, S_SELF),
CC1, { '\\' }, CC2,
CC1, { '\n' }, S_STNL,
CC1, { EOFC }, S_EOFSTR,
/* saw \ in ccon */
CC2, { C_XX }, CC1,
CC2, { '\n' }, S_STNL,
CC2, { EOFC }, S_EOFSTR,
/* saw /, perhaps start of comment */
COM1, { C_XX }, ACT(SLASH, S_SELFB),
COM1, { '=' }, ACT(ASSLASH, S_SELF),
COM1, { '*' }, COM2,
COM1, { '/' }, COM4,
/* saw "/*", start of comment */
COM2, { C_XX }, COM2,
COM2, { '\n' }, S_COMNL,
COM2, { '*' }, COM3,
COM2, { EOFC }, S_EOFCOM,
/* saw the * possibly ending a comment */
COM3, { C_XX }, COM2,
COM3, { '\n' }, S_COMNL,
COM3, { '*' }, COM3,
COM3, { '/' }, S_COMMENT,
/* // comment */
COM4, { C_XX }, COM4,
COM4, { '\n' }, S_NL,
COM4, { EOFC }, S_EOFCOM,
/* saw white space, eat it up */
WS1, { C_XX }, S_WS,
WS1, { ' ', '\t', '\v' }, WS1,
/* saw -, check --, -=, -> */
MINUS1, { C_XX }, ACT(MINUS, S_SELFB),
MINUS1, { '-' }, ACT(MMINUS, S_SELF),
MINUS1, { '=' }, ACT(ASMINUS,S_SELF),
MINUS1, { '>' }, ACT(ARROW,S_SELF),
/* saw +, check ++, += */
PLUS1, { C_XX }, ACT(PLUS, S_SELFB),
PLUS1, { '+' }, ACT(PPLUS, S_SELF),
PLUS1, { '=' }, ACT(ASPLUS, S_SELF),
/* saw <, check <<, <<=, <= */
LT1, { C_XX }, ACT(LT, S_SELFB),
LT1, { '<' }, LT2,
LT1, { '=' }, ACT(LEQ, S_SELF),
LT2, { C_XX }, ACT(LSH, S_SELFB),
LT2, { '=' }, ACT(ASLSH, S_SELF),
/* saw >, check >>, >>=, >= */
GT1, { C_XX }, ACT(GT, S_SELFB),
GT1, { '>' }, GT2,
GT1, { '=' }, ACT(GEQ, S_SELF),
GT2, { C_XX }, ACT(RSH, S_SELFB),
GT2, { '=' }, ACT(ASRSH, S_SELF),
/* = */
ASG1, { C_XX }, ACT(ASGN, S_SELFB),
ASG1, { '=' }, ACT(EQ, S_SELF),
/* ! */
NOT1, { C_XX }, ACT(NOT, S_SELFB),
NOT1, { '=' }, ACT(NEQ, S_SELF),
/* & */
AND1, { C_XX }, ACT(AND, S_SELFB),
AND1, { '&' }, ACT(LAND, S_SELF),
AND1, { '=' }, ACT(ASAND, S_SELF),
/* | */
OR1, { C_XX }, ACT(OR, S_SELFB),
OR1, { '|' }, ACT(LOR, S_SELF),
OR1, { '=' }, ACT(ASOR, S_SELF),
/* # */
SHARP1, { C_XX }, ACT(SHARP, S_SELFB),
SHARP1, { '#' }, ACT(DSHARP, S_SELF),
/* % */
PCT1, { C_XX }, ACT(PCT, S_SELFB),
PCT1, { '=' }, ACT(ASPCT, S_SELF),
/* * */
STAR1, { C_XX }, ACT(STAR, S_SELFB),
STAR1, { '=' }, ACT(ASSTAR, S_SELF),
/* ^ */
CIRC1, { C_XX }, ACT(CIRC, S_SELFB),
CIRC1, { '=' }, ACT(ASCIRC, S_SELF),
-1
};
/* first index is char, second is state */
/* increase #states to power of 2 to encourage use of shift */
short bigfsm[256][MAXSTATE];
void
expandlex(void)
{
/*const*/ struct fsm *fp;
int i, j, nstate;
for (fp = fsm; fp->state>=0; fp++) {
for (i=0; fp->ch[i]; i++) {
nstate = fp->nextstate;
if (nstate >= S_SELF)
nstate = ~nstate;
switch (fp->ch[i]) {
case C_XX: /* random characters */
for (j=0; j<256; j++)
bigfsm[j][fp->state] = nstate;
continue;
case C_ALPH:
for (j=0; j<=256; j++)
if ('a'<=j&&j<='z' || 'A'<=j&&j<='Z'
|| j=='_')
bigfsm[j][fp->state] = nstate;
continue;
case C_NUM:
for (j='0'; j<='9'; j++)
bigfsm[j][fp->state] = nstate;
continue;
default:
bigfsm[fp->ch[i]][fp->state] = nstate;
}
}
}
/* install special cases for ? (trigraphs), \ (splicing), runes, and EOB */
for (i=0; i<MAXSTATE; i++) {
for (j=0; j<0xFF; j++)
if (j=='?' || j=='\\') {
if (bigfsm[j][i]>0)
bigfsm[j][i] = ~bigfsm[j][i];
bigfsm[j][i] &= ~QBSBIT;
}
bigfsm[EOB][i] = ~S_EOB;
if (bigfsm[EOFC][i]>=0)
bigfsm[EOFC][i] = ~S_EOF;
}
}
void
fixlex(void)
{
/* do C++ comments? */
if (Cplusplus==0)
bigfsm['/'][COM1] = bigfsm['x'][COM1];
}
/*
* fill in a row of tokens from input, terminated by NL or END
* First token is put at trp->lp.
* Reset is non-zero when the input buffer can be "rewound."
* The value is a flag indicating that possible macros have
* been seen in the row.
*/
int
gettokens(Tokenrow *trp, int reset)
{
register int c, state, oldstate;
register uchar *ip;
register Token *tp, *maxp;
int runelen;
Source *s = cursource;
int nmac = 0;
extern char outbuf[];
tp = trp->lp;
ip = s->inp;
if (reset) {
s->lineinc = 0;
if (ip>=s->inl) { /* nothing in buffer */
s->inl = s->inb;
fillbuf(s);
ip = s->inp = s->inb;
} else if (ip >= s->inb+(3*INS/4)) {
memmove(s->inb, ip, 4+s->inl-ip);
s->inl = s->inb+(s->inl-ip);
ip = s->inp = s->inb;
}
}
maxp = &trp->bp[trp->max];
runelen = 1;
for (;;) {
continue2:
if (tp>=maxp) {
trp->lp = tp;
tp = growtokenrow(trp);
maxp = &trp->b