static inline void
lexwhite(fz_stream *f)
{
int c;
do
{
c = fz_readbyte(f);
}
while ((c <= 32) && (iswhite(c)));
if (c != EOF)
fz_unreadbyte(f);
}
static void
closelzwd(fz_stream *stm)
{
fz_lzwd *lzw = stm->state;
int i;
/* if we read any extra bytes, try to put them back */
i = (32 - lzw->bidx) / 8;
while (i--)
fz_unreadbyte(lzw->chain);
fz_close(lzw->chain);
fz_free(lzw);
}
static int
lexnumber(fz_stream *f, char *s, int n, int *tok)
{
char *buf = s;
*tok = PDF_TINT;
/* Initially we might have +, -, . or a digit */
if (n > 1)
{
int c = fz_readbyte(f);
switch (c)
{
case '.':
*tok = PDF_TREAL;
*s++ = c;
n--;
goto loop_after_dot;
case '+':
case '-':
case RANGE_0_9:
*s++ = c;
n--;
goto loop_after_sign;
default:
fz_unreadbyte(f);
goto end;
case EOF:
goto end;
}
}
/* We can't accept a sign from here on in, just . or a digit */
loop_after_sign:
while (n > 1)
{
int c = fz_readbyte(f);
switch (c)
{
case '.':
*tok = PDF_TREAL;
*s++ = c;
n--;
goto loop_after_dot;
case RANGE_0_9:
*s++ = c;
break;
default:
fz_unreadbyte(f);
goto end;
case EOF:
goto end;
}
n--;
}
/* In here, we've seen a dot, so can accept just digits */
loop_after_dot:
while (n > 1)
{
int c = fz_readbyte(f);
switch (c)
{
case RANGE_0_9:
*s++ = c;
break;
default:
fz_unreadbyte(f);
goto end;
case EOF:
goto end;
}
n--;
}
end:
*s = '\0';
return s-buf;
}
fz_error
pdf_lex(int *tok, fz_stream *f, char *buf, int n, int *sl)
{
while (1)
{
int c = fz_readbyte(f);
switch (c)
{
case EOF:
*tok = PDF_TEOF;
return fz_okay;
case ISWHITE:
lexwhite(f);
break;
case '%':
lexcomment(f);
break;
case '/':
lexname(f, buf, n);
*sl = strlen(buf);
*tok = PDF_TNAME;
return fz_okay;
case '(':
*sl = lexstring(f, buf, n);
*tok = PDF_TSTRING;
return fz_okay;
case ')':
*tok = PDF_TERROR;
goto cleanuperror;
case '<':
c = fz_readbyte(f);
if (c == '<')
{
*tok = PDF_TODICT;
}
else
{
fz_unreadbyte(f);
*sl = lexhexstring(f, buf, n);
*tok = PDF_TSTRING;
}
return fz_okay;
case '>':
c = fz_readbyte(f);
if (c == '>')
{
*tok = PDF_TCDICT;
return fz_okay;
}
*tok = PDF_TERROR;
goto cleanuperror;
case '[':
*tok = PDF_TOARRAY;
return fz_okay;
case ']':
*tok = PDF_TCARRAY;
return fz_okay;
case '{':
*tok = PDF_TOBRACE;
return fz_okay;
case '}':
*tok = PDF_TCBRACE;
return fz_okay;
case ISNUMBER:
fz_unreadbyte(f);
*sl = lexnumber(f, buf, n, tok);
return fz_okay;
default: /* isregular: !isdelim && !iswhite && c != EOF */
fz_unreadbyte(f);
lexname(f, buf, n);
*sl = strlen(buf);
*tok = pdf_tokenfromkeyword(buf);
return fz_okay;
}
}
cleanuperror:
*tok = PDF_TERROR;
return fz_throw("lexical error");
}
static int
lexstring(fz_stream *f, char *buf, int n)
{
char *s = buf;
char *e = buf + n;
int bal = 1;
int oct;
int c;
while (s < e)
{
c = fz_readbyte(f);
switch (c)
{
case EOF:
goto end;
case '(':
bal++;
*s++ = c;
break;
case ')':
bal --;
if (bal == 0)
goto end;
*s++ = c;
break;
case '\\':
c = fz_readbyte(f);
switch (c)
{
case EOF:
goto end;
case 'n':
*s++ = '\n';
break;
case 'r':
*s++ = '\r';
break;
case 't':
*s++ = '\t';
break;
case 'b':
*s++ = '\b';
break;
case 'f':
*s++ = '\f';
break;
case '(':
*s++ = '(';
break;
case ')':
*s++ = ')';
break;
case '\\':
*s++ = '\\';
break;
case RANGE_0_9:
oct = c - '0';
c = fz_readbyte(f);
if (c >= '0' && c <= '9')
{
oct = oct * 8 + (c - '0');
c = fz_readbyte(f);
if (c >= '0' && c <= '9')
oct = oct * 8 + (c - '0');
else if (c != EOF)
fz_unreadbyte(f);
}
else if (c != EOF)
fz_unreadbyte(f);
*s++ = oct;
break;
case '\n':
break;
case '\r':
c = fz_readbyte(f);
if ((c != '\n') && (c != EOF))
fz_unreadbyte(f);
break;
default:
*s++ = c;
}
break;
default:
*s++ = c;
break;
}
}
end:
return s - buf;
}
static void
lexname(fz_stream *f, char *s, int n)
{
while (n > 1)
{
int c = fz_readbyte(f);
switch (c)
{
case ISWHITE:
case ISDELIM:
fz_unreadbyte(f);
goto end;
case EOF:
goto end;
case '#':
{
int d;
c = fz_readbyte(f);
switch (c)
{
case RANGE_0_9:
d = (c - '0') << 4;
break;
case RANGE_a_f:
d = (c - 'a' + 10) << 4;
break;
case RANGE_A_F:
d = (c - 'A' + 10) << 4;
break;
default:
fz_unreadbyte(f);
/* fallthrough */
case EOF:
goto end;
}
c = fz_readbyte(f);
switch (c)
{
case RANGE_0_9:
c -= '0';
break;
case RANGE_a_f:
c -= 'a' - 10;
break;
case RANGE_A_F:
c -= 'A' - 10;
break;
default:
fz_unreadbyte(f);
/* fallthrough */
case EOF:
*s++ = d;
n--;
goto end;
}
*s++ = d + c;
n--;
break;
}
default:
*s++ = c;
n--;
break;
}
}
end:
*s = '\0';
}
