int
pdf_lex(fz_stream *f, pdf_lexbuf *buf)
{
while (1)
{
int c = fz_read_byte(f);
switch (c)
{
case EOF:
return PDF_TOK_EOF;
case IS_WHITE:
lex_white(f);
break;
case '%':
lex_comment(f);
break;
case '/':
lex_name(f, buf);
return PDF_TOK_NAME;
case '(':
return lex_string(f, buf);
case ')':
fz_warn(f->ctx, "lexical error (unexpected ')')");
continue;
case '<':
c = fz_read_byte(f);
if (c == '<')
{
return PDF_TOK_OPEN_DICT;
}
else
{
fz_unread_byte(f);
return lex_hex_string(f, buf);
}
case '>':
c = fz_read_byte(f);
if (c == '>')
{
return PDF_TOK_CLOSE_DICT;
}
fz_warn(f->ctx, "lexical error (unexpected '>')");
continue;
case '[':
return PDF_TOK_OPEN_ARRAY;
case ']':
return PDF_TOK_CLOSE_ARRAY;
case '{':
return PDF_TOK_OPEN_BRACE;
case '}':
return PDF_TOK_CLOSE_BRACE;
case IS_NUMBER:
return lex_number(f, buf, c);
default: /* isregular: !isdelim && !iswhite && c != EOF */
fz_unread_byte(f);
lex_name(f, buf);
return pdf_token_from_keyword(buf->scratch);
}
}
}
static void
lex_white(fz_stream *f)
{
int c;
do {
c = fz_read_byte(f);
} while ((c <= 32) && (iswhite(c)));
if (c != EOF)
fz_unread_byte(f);
}
static int
lex_number(fz_stream *f, pdf_lexbuf *buf, int c)
{
int neg = 0;
int i = 0;
int n;
int d;
float v;
/* Initially we might have +, -, . or a digit */
switch (c)
{
case '.':
goto loop_after_dot;
case '-':
neg = 1;
break;
case '+':
break;
default: /* Must be a digit */
i = c - '0';
break;
}
while (1)
{
c = fz_read_byte(f);
switch (c)
{
case '.':
goto loop_after_dot;
case RANGE_0_9:
i = 10*i + c - '0';
/* FIXME: Need overflow check here; do we care? */
break;
default:
fz_unread_byte(f);
/* Fallthrough */
case EOF:
if (neg)
i = -i;
buf->i = i;
return PDF_TOK_INT;
}
}
/* In here, we've seen a dot, so can accept just digits */
loop_after_dot:
n = 0;
d = 1;
while (1)
{
c = fz_read_byte(f);
switch (c)
{
case RANGE_0_9:
if (d >= INT_MAX/10)
goto underflow;
n = n*10 + (c - '0');
d *= 10;
break;
default:
fz_unread_byte(f);
/* Fallthrough */
case EOF:
v = (float)i + ((float)n / (float)d);
if (neg)
v = -v;
buf->f = v;
return PDF_TOK_REAL;
}
}
underflow:
/* Ignore any digits after here, because they are too small */
while (1)
{
c = fz_read_byte(f);
switch (c)
{
case RANGE_0_9:
break;
default:
fz_unread_byte(f);
/* Fallthrough */
case EOF:
v = (float)i + ((float)n / (float)d);
if (neg)
v = -v;
buf->f = v;
return PDF_TOK_REAL;
}
}
}
pdf_token
pdf_lex(fz_stream *f, pdf_lexbuf *buf)
{
while (1)
{
int c = fz_read_byte(f);
switch (c)
{
case EOF:
return PDF_TOK_EOF;
case IS_WHITE:
lex_white(f);
break;
case '%':
lex_comment(f);
break;
case '/':
lex_name(f, buf);
return PDF_TOK_NAME;
case '(':
return lex_string(f, buf);
case ')':
fz_warn(f->ctx, "lexical error (unexpected ')')");
continue;
case '<':
c = fz_read_byte(f);
if (c == '<')
{
return PDF_TOK_OPEN_DICT;
}
else
{
fz_unread_byte(f);
return lex_hex_string(f, buf);
}
case '>':
c = fz_read_byte(f);
if (c == '>')
{
return PDF_TOK_CLOSE_DICT;
}
fz_warn(f->ctx, "lexical error (unexpected '>')");
if (c == EOF)
{
return PDF_TOK_EOF;
}
fz_unread_byte(f);
continue;
case '[':
return PDF_TOK_OPEN_ARRAY;
case ']':
return PDF_TOK_CLOSE_ARRAY;
case '{':
return PDF_TOK_OPEN_BRACE;
case '}':
return PDF_TOK_CLOSE_BRACE;
case IS_NUMBER:
/* cf. https://code.google.com/p/sumatrapdf/issues/detail?id=2231 */
{
int tok = lex_number(f, buf, c);
while (1)
{
c = fz_peek_byte(f);
switch (c)
{
case IS_NUMBER:
fz_warn(f->ctx, "ignoring invalid character after number: '%c'", c);
fz_read_byte(f);
continue;
default:
return tok;
}
}
}
default: /* isregular: !isdelim && !iswhite && c != EOF */
fz_unread_byte(f);
lex_name(f, buf);
return pdf_token_from_keyword(buf->scratch);
}
}
}
static int
lex_string(fz_stream *f, pdf_lexbuf *lb)
{
char *s = lb->scratch;
char *e = s + lb->size;
int bal = 1;
int oct;
int c;
while (1)
{
if (s == e)
{
s += pdf_lexbuf_grow(lb);
e = lb->scratch + lb->size;
}
c = fz_read_byte(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_read_byte(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_7:
oct = c - '0';
c = fz_read_byte(f);
if (c >= '0' && c <= '7')
{
oct = oct * 8 + (c - '0');
c = fz_read_byte(f);
if (c >= '0' && c <= '7')
oct = oct * 8 + (c - '0');
else if (c != EOF)
fz_unread_byte(f);
}
else if (c != EOF)
fz_unread_byte(f);
*s++ = oct;
break;
case '\n':
break;
case '\r':
c = fz_read_byte(f);
if ((c != '\n') && (c != EOF))
fz_unread_byte(f);
break;
default:
*s++ = c;
}
break;
default:
*s++ = c;
break;
}
}
end:
lb->len = s - lb->scratch;
return PDF_TOK_STRING;
}
static void
lex_name(fz_stream *f, pdf_lexbuf *buf)
{
char *s = buf->scratch;
int n = buf->size;
while (n > 1)
{
int c = fz_read_byte(f);
switch (c)
{
case IS_WHITE:
case IS_DELIM:
fz_unread_byte(f);
goto end;
case EOF:
goto end;
case '#':
{
int d;
c = fz_read_byte(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;
/* cf. https://code.google.com/p/sumatrapdf/issues/detail?id=2300 */
case '#':
fz_unread_byte(f);
*s++ = '#';
n--;
continue;
default:
fz_unread_byte(f);
/* fallthrough */
case EOF:
goto end;
}
c = fz_read_byte(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_unread_byte(f);
/* fallthrough */
case EOF:
*s++ = d;
n--;
goto end;
}
*s++ = d + c;
n--;
break;
}
default:
*s++ = c;
n--;
break;
}
}
end:
*s = '\0';
buf->len = s - buf->scratch;
}
static int
lex_number(fz_context *ctx, fz_stream *f, pdf_lexbuf *buf, int c)
{
int neg = 0;
fz_off_t i = 0;
int n;
int d;
float v;
float fd;
/* Initially we might have +, -, . or a digit */
switch (c)
{
case '.':
goto loop_after_dot;
case '-':
neg = 1;
break;
case '+':
break;
default: /* Must be a digit */
i = c - '0';
break;
}
while (1)
{
c = fz_read_byte(ctx, f);
switch (c)
{
case '.':
goto loop_after_dot;
case RANGE_0_9:
/* We deliberately ignore overflow here. We tried
* code that returned INT_MIN/MAX as appropriate,
* but this causes loss of data (see Bug695950.pdf
* for an example). Tests show that Acrobat handles
* overflows in exactly the same way we do (i.e.
* 123450000000000000000678 is read as 678). */
i = 10*i + c - '0';
break;
default:
fz_unread_byte(ctx, f);
/* Fallthrough */
case EOF:
if (neg)
i = -i;
buf->i = i;
return PDF_TOK_INT;
}
}
/* In here, we've seen a dot, so can accept just digits */
loop_after_dot:
n = 0;
d = 1;
while (1)
{
c = fz_read_byte(ctx, f);
switch (c)
{
case RANGE_0_9:
if (d >= INT_MAX/10)
goto underflow;
n = n*10 + (c - '0');
d *= 10;
break;
default:
fz_unread_byte(ctx, f);
/* Fallthrough */
case EOF:
v = (float)i + ((float)n / (float)d);
if (neg)
v = -v;
buf->f = v;
return PDF_TOK_REAL;
}
}
underflow:
fd = 1 / (float)d;
v = (float)i + ((float)n * fd);
while (1)
{
c = fz_read_byte(ctx, f);
switch (c)
{
case RANGE_0_9:
fd /= 10;
v += (c - '0') * fd;
break;
default:
fz_unread_byte(ctx, f);
/* Fallthrough */
case EOF:
if (neg)
v = -v;
buf->f = v;
return PDF_TOK_REAL;
}
}
}
static void
lex_name(fz_context *ctx, fz_stream *f, pdf_lexbuf *buf)
{
char *s = buf->scratch;
int n = buf->size;
while (n > 1)
{
int c = fz_read_byte(ctx, f);
switch (c)
{
case IS_WHITE:
case IS_DELIM:
fz_unread_byte(ctx, f);
goto end;
case EOF:
goto end;
case '#':
{
int d;
c = fz_read_byte(ctx, 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_unread_byte(ctx, f);
/* fallthrough */
case EOF:
goto end;
}
c = fz_read_byte(ctx, 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_unread_byte(ctx, f);
/* fallthrough */
case EOF:
*s++ = d;
n--;
goto end;
}
*s++ = d + c;
n--;
break;
}
default:
*s++ = c;
n--;
break;
}
}
end:
*s = '\0';
buf->len = s - buf->scratch;
}
static int
lex_number(fz_stream *f, char *s, int n, int *tok)
{
char *buf = s;
*tok = PDF_TOK_INT;
/* Initially we might have +, -, . or a digit */
if (n > 1)
{
int c = fz_read_byte(f);
switch (c)
{
case '.':
*tok = PDF_TOK_REAL;
*s++ = c;
n--;
goto loop_after_dot;
case '+':
case '-':
case RANGE_0_9:
*s++ = c;
n--;
goto loop_after_sign;
default:
fz_unread_byte(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_read_byte(f);
switch (c)
{
case '.':
*tok = PDF_TOK_REAL;
*s++ = c;
n--;
goto loop_after_dot;
case RANGE_0_9:
*s++ = c;
break;
default:
fz_unread_byte(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_read_byte(f);
switch (c)
{
case RANGE_0_9:
*s++ = c;
break;
default:
fz_unread_byte(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_read_byte(f);
switch (c)
{
case EOF:
*tok = PDF_TOK_EOF;
return fz_okay;
case IS_WHITE:
lex_white(f);
break;
case '%':
lex_comment(f);
break;
case '/':
lex_name(f, buf, n);
*sl = strlen(buf);
*tok = PDF_TOK_NAME;
return fz_okay;
case '(':
*sl = lex_string(f, buf, n);
*tok = PDF_TOK_STRING;
return fz_okay;
case ')':
*tok = PDF_TOK_ERROR;
goto cleanuperror;
case '<':
c = fz_read_byte(f);
if (c == '<')
{
*tok = PDF_TOK_OPEN_DICT;
}
else
{
fz_unread_byte(f);
*sl = lex_hex_string(f, buf, n);
*tok = PDF_TOK_STRING;
}
return fz_okay;
case '>':
c = fz_read_byte(f);
if (c == '>')
{
*tok = PDF_TOK_CLOSE_DICT;
return fz_okay;
}
*tok = PDF_TOK_ERROR;
goto cleanuperror;
case '[':
*tok = PDF_TOK_OPEN_ARRAY;
return fz_okay;
case ']':
*tok = PDF_TOK_CLOSE_ARRAY;
return fz_okay;
case '{':
*tok = PDF_TOK_OPEN_BRACE;
return fz_okay;
case '}':
*tok = PDF_TOK_CLOSE_BRACE;
return fz_okay;
case IS_NUMBER:
fz_unread_byte(f);
*sl = lex_number(f, buf, n, tok);
return fz_okay;
default: /* isregular: !isdelim && !iswhite && c != EOF */
fz_unread_byte(f);
lex_name(f, buf, n);
*sl = strlen(buf);
*tok = pdf_token_from_keyword(buf);
return fz_okay;
}
}
cleanuperror:
*tok = PDF_TOK_ERROR;
return fz_throw("lexical error");
}
static int
lex_string(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_read_byte(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_read_byte(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_read_byte(f);
if (c >= '0' && c <= '9')
{
oct = oct * 8 + (c - '0');
c = fz_read_byte(f);
if (c >= '0' && c <= '9')
oct = oct * 8 + (c - '0');
else if (c != EOF)
fz_unread_byte(f);
}
else if (c != EOF)
fz_unread_byte(f);
*s++ = oct;
break;
case '\n':
break;
case '\r':
c = fz_read_byte(f);
if ((c != '\n') && (c != EOF))
fz_unread_byte(f);
break;
default:
*s++ = c;
}
break;
default:
*s++ = c;
break;
}
}
end:
return s - buf;
}
static void
lex_name(fz_stream *f, char *s, int n)
{
while (n > 1)
{
int c = fz_read_byte(f);
switch (c)
{
case IS_WHITE:
case IS_DELIM:
fz_unread_byte(f);
goto end;
case EOF:
goto end;
case '#':
{
int d;
c = fz_read_byte(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_unread_byte(f);
/* fallthrough */
case EOF:
goto end;
}
c = fz_read_byte(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_unread_byte(f);
/* fallthrough */
case EOF:
*s++ = d;
n--;
goto end;
}
*s++ = d + c;
n--;
break;
}
default:
*s++ = c;
n--;
break;
}
}
end:
*s = '\0';
}
