/* Read a single item. Syntax is:
*
* element :
* whitespace* item whitespace* [EOF|EOL|separator]
*
* item :
* double |
* "anything" |
* empty
*
* the anything in quotes can contain " escaped with \
*
* Return the char that caused failure on fail (EOF or \n).
*/
static int
read_double( FILE *fp, const char whitemap[256], const char sepmap[256],
int lineno, int colno, double *out )
{
int ch;
/* The fscanf() may change this ... but all other cases need a zero.
*/
*out = 0;
ch = skip_white( fp, whitemap );
if( ch == EOF ||
ch == '\n' )
return( ch );
if( ch == '"' ) {
(void) fgetc( fp );
(void) skip_to_quote( fp );
ch = fgetc( fp );
}
else if( !sepmap[ch] &&
fscanf( fp, "%lf", out ) != 1 ) {
/* Only a warning, since (for example) exported spreadsheets
* will often have text or date fields.
*/
vips_warn( "csv2vips",
_( "error parsing number, line %d, column %d" ),
lineno, colno );
/* Step over the bad data to the next separator.
*/
ch = skip_to_sep( fp, sepmap );
}
/* Don't need to check result, we have read a field successfully.
*/
ch = skip_white( fp, whitemap );
/* If it's a separator, we have to step over it.
*/
if( ch != EOF &&
sepmap[ch] )
(void) fgetc( fp );
return( 0 );
}
bool to_internal_SU( char * * scanp, su * converted )
{
bool converterror = true;
char * p;
char * ps;
ldiv_t div;
su * s;
long wh;
long wd;
char * pp; // ptr to decimal point or embedded unit
char * pu; // ptr to trailing unit
char * pd1; // ptr to 0.1 decimal
char * pdn; // ptr to last digit +1
char unit[4];
char quote;
long k;
char sign;
unit[3] = '\0';
unit[2] = '\0';
unit[1] = '\0';
unit[0] = '\0';
s = converted;
p = *scanp;
ps = s->su_txt;
*ps = '\0';
wh = 0;
wd = 0;
pp = NULL;
quote = '\0';
s->su_u = SU_undefined;
if( *p == '\'' || *p == '"' ) { // ignore but remember quote
quote = *p++;
}
if( *p == '+' || *p == '-' ) {
sign = *p;
*ps++ = *p++;
s->su_relative = true; // value is added / subtracted from old value
} else {
sign = '+';
s->su_relative = false; // value replaces old value
}
/***********************************************************************/
/* Special for layout :BANREGION */
/***********************************************************************/
if( quote == '\0' && isalpha( *p ) ) {
converterror = su_layout_special( scanp, converted );
if( !converterror ) {
return( converterror ); // layout special ok
}
}
while( *p >= '0' && *p <= '9' ) { // whole part
wh = 10 * wh + *p - '0';
*ps++ = *p++;
if( ps >= s->su_txt + sizeof( s->su_txt ) ) {
if( quote ) {
p = skip_to_quote( p, quote );
}
*scanp = p;
return( converterror );
}
}
pp = p;
k = 0;
while( *p && isalpha( *p ) ) {
unit[k++] = tolower( *p ); // save Unit
*ps++ = *p++;
if( ps >= s->su_txt + sizeof( s->su_txt ) || k > 2 ) {
if( quote ) {
p = skip_to_quote( p, quote );
}
*scanp = p;
return( converterror );
}
}
if( p > pp + 2 ) {
if( quote ) {
p = skip_to_quote( p, quote );
}
*scanp = p;
return( converterror ); // no unit has more than 2 chars
}
pd1 = NULL;
pdn = NULL;
if( p == pp && *p == '.' ) { // no unit found, try dec point
*ps++ = *p++;
pd1 = p; // remember start of decimals
if( ps >= s->su_txt + sizeof( s->su_txt ) ) {
if( quote ) {
p = skip_to_quote( p, quote );
}
*scanp = p;
return( converterror );
}
while( *p >= '0' && *p <= '9' ) { // try decimal part
wd = 10 * wd + *p - '0';
*ps++ = *p++;
if( ps >= s->su_txt + sizeof( s->su_txt ) ) {
if( quote ) {
p = skip_to_quote( p, quote );
}
*scanp = p;
return( converterror );
}
if( p > pd1 + 2 ) { // more than two digits
if( quote ) {
p = skip_to_quote( p, quote );
}
*scanp = p;
return( converterror );
}
}
pdn = p;
} else {
if( k ) { // unit found
pd1 = p;
if( ps >= s->su_txt + sizeof( s->su_txt ) ) {
if( quote ) {
p = skip_to_quote( p, quote );
}
*scanp = p;
return( converterror );
}
while( *p >= '0' && *p <= '9' ) { // try decimal part
wd = 10 * wd + *p - '0';
*ps++ = *p++;
if( ps >= s->su_txt + sizeof( s->su_txt ) ) {
if( quote ) {
p = skip_to_quote( p, quote );
}
*scanp = p;
return( converterror );
}
if( p > pd1 + 2 ) { // more than two digits
if( quote ) {
p = skip_to_quote( p, quote );
}
*scanp = p;
return( converterror );
}
}
pdn = p;
}
}
k = 0;
pu = p;
if( *p ) { // not yet at end
while( *p && isalpha( *p ) ) { // try trailing unit
unit[k++] = tolower( *p );
*ps++ = *p++;
if( ps >= s->su_txt + sizeof( s->su_txt ) || k > 2 ) {
if( quote ) {
p = skip_to_quote( p, quote );
}
*scanp = p;
return( converterror );
}
}
}
*ps = '\0';
s->su_whole = wh;
s->su_dec = wd;
if( (quote && *p != quote ) || (!quote && *p == '\'') ) {
if( quote ) {
p = skip_to_quote( p, quote );
}
*scanp = p;
return( converterror );
}
if( quote ) {
p++; // over quote
}
*scanp = p; // report back scan position
if( k == 0 ) { // no trailing unit
pu = NULL;
} else {
if( pu == pp ) { // no decimals, no unit
pu = NULL;
}
}
if( *pp == '.' ) { // dec point found
if( pu == NULL ) { // need trailing unit
return( converterror );
}
} else { // no decimals
if( pu != NULL ) { // but unit specified twice?
return( converterror );
}
}
/***********************************************************************/
/* check for valid unit */
/***********************************************************************/
if( unit[1] == '\0' ) { // single letter unit
switch( unit[0] ) {
case 'i' :
s->su_u = SU_inch;
break;
case 'm' :
s->su_u = SU_ems;
break;
case 'c' :
s->su_u = SU_cicero;
break;
case 'p' :
s->su_u = SU_pica;
break;
case '\0' : // no unit is characters or lines
s->su_u = SU_chars_lines;
break;
default:
return( converterror );
break;
}
} else { // two letter unit
if( unit[1] == 'm' ) { // cm, mm ?
if( unit[0] == 'c' ) {
s->su_u = SU_cm;
} else if( unit[0] == 'm' ) {
s->su_u = SU_mm;
} else { // invalid unit
return( converterror );
}
} else if( unit[0] == 'd' ) { // dv ?
if( unit[1] == 'v' ) {
s->su_u = SU_dv;
}
} else { // invalid unit
return( converterror );
}
}
s->su_inch = 0;
s->su_mm = 0;
k = 1;
if( pd1 != NULL ) {
if( pdn - pd1 == 1 ) {
k = 10; // only 0.1 digit
}
}
switch( s->su_u ) {
// the relative units are only stored, not converted
case SU_chars_lines :
case SU_ems :
if( wd != 0 ) { // no decimals allowed
return( converterror );
}
break;
case SU_dv : // decimals are allowed for dv
break;
case SU_inch : // inch, cm, mm valid with decimals
s->su_mm = (wh * 100L + wd * k) * 2540L;
s->su_inch = (wh * 100L + wd * k) * 100L;
break;
case SU_cm :
s->su_mm = (wh * 100L + wd * k) * 1000L;
s->su_inch = s->su_mm * 10L / 254L;
break;
case SU_mm :
s->su_mm = (wh * 100L + wd * k) * 100L;
s->su_inch = s->su_mm * 10L / 254L;
break;
case SU_cicero : // cicero
if( wd > 11 ) {
div = ldiv( wd, 12L);
wh += div.quot;
wd = div.rem;
}
s->su_inch = wh * 10000L / 6L + wd * 10000L / 72L;
s->su_inch = (int64_t)s->su_inch * 10656L / 10000L;
s->su_mm = s->su_inch * 254L / 10L;
break;
case SU_pica : // pica
if( wd > 11 ) {
div = ldiv( wd, 12L);
wh += div.quot;
wd = div.rem;
}
s->su_inch = wh * 10000L / 6L + wd * 10000L / 72L;
s->su_inch = (int64_t)s->su_inch * 9978L / 10000L;
s->su_mm = s->su_inch * 254L / 10L;
break;
default:
break;
}
if( sign == '-' ) {
s->su_inch = -s->su_inch;
s->su_mm = -s->su_mm;
s->su_whole = -s->su_whole;
}
converterror = false;
return( converterror );
}
