static void GetLeftRightCounts(DocTocItem *node, int& l2r, int& r2l)
{
if (!node)
return;
if (node->title) {
for (const WCHAR *c = node->title; *c; c++) {
if (ISLEFTTORIGHTCHAR(*c))
l2r++;
else if (ISRIGHTTOLEFTCHAR(*c))
r2l++;
}
}
GetLeftRightCounts(node->child, l2r, r2l);
GetLeftRightCounts(node->next, l2r, r2l);
}
static void
fixup_text_span(fz_text_span *span)
{
int i;
for (i = 0; i < span->len; i++)
{
switch (span->text[i].c)
{
/* recombine characters and their accents */
case 0x00A8: /* diaeresis/umlaut */
case 0x00B4: /* accute accent */
case 0x0060: /* grave accent */
case 0x005E: case 0x02C6: /* circumflex accent */
case 0x02DA: /* ring above */
if (i + 1 < span->len)
{
int newC = 0;
if (span->text[i + 1].c != 32 || i + 2 == span->len)
newC = ornate_character(&span->text[i], &span->text[i + 1]);
else if ((newC = ornate_character(&span->text[i], &span->text[i + 2])))
delete_character(span, i + 1);
if (newC)
{
delete_character(span, i);
span->text[i].c = newC;
}
}
break;
default:
/* cf. http://code.google.com/p/sumatrapdf/issues/detail?id=733 */
/* reverse words written in RTL languages */
if (ISRIGHTTOLEFTCHAR(span->text[i].c))
{
int j = i + 1;
while (j < span->len && span->text[j - 1].bbox.x0 <= span->text[j].bbox.x0 && !ISLEFTTORIGHTCHAR(span->text[i].c))
j++;
reverse_characters(span, i, j - 1);
i = j;
}
}
}
}
static void
fixuptextspan(fz_text_span *head)
{
fz_text_span *span;
int i;
for (span = head; span; span = span->next)
{
for (i = 0; i < span->len; i++)
{
switch (span->text[i].c)
{
/* recombine characters and their accents */
case 0x00A8: /* ?*/
case 0x00B4: /* ?*/
case 0x0060: /* ` */
case 0x005E: /* ^ */
case 0x02DA: /* ?*/
if (span->next && span->next->len > 0 && (i + 1 == span->len || i + 2 == span->len && span->text[i + 1].c == 32))
{
mergetwospans(span);
}
if (i + 1 < span->len)
{
int newC = 0;
if (span->text[i + 1].c != 32 || i + 2 >= span->len)
newC = ornatecharacter(span->text[i].c, span->text[i + 1].c);
else if ((newC = ornatecharacter(span->text[i].c, span->text[i + 2].c)))
deletecharacter(span, i + 1);
if (newC)
{
deletecharacter(span, i);
span->text[i].c = newC;
}
}
break;
default:
/* cf. http://code.google.com/p/sumatrapdf/issues/detail?id=733 */
/* reverse words written in RTL languages */
if (ISRIGHTTOLEFTCHAR(span->text[i].c))
{
int j = i + 1;
while (j < span->len && span->text[j - 1].bbox.x0 <= span->text[j].bbox.x0 && !ISLEFTTORIGHTCHAR(span->text[i].c))
j++;
reversecharacters(span, i, j - 1);
i = j;
}
}
}
}
/* cf. http://code.google.com/p/sumatrapdf/issues/detail?id=734 */
/* remove duplicate character sequences in (almost) the same spot */
for (span = head; span; span = span->next)
{
if (span->size < 5) /* doglyphsoverlap fails too often for small fonts */
continue;
for (i = 0; i < span->len; i++)
{
fz_text_span *span2;
int newlines, j;
for (span2 = span, j = i + 1, newlines = 0; span2 && newlines < 2; newlines += span2->eol, span2 = span2->next, j = 0)
for (; j < span2->len; j++)
if (span->text[i].c != 32 && doglyphsoverlap(span, i, span2, j))
goto fixup_delete_duplicates;
continue;
fixup_delete_duplicates:
do
deletecharacter(span, i);
while (doglyphsoverlap(span, i, span2, ++j));
if (i < span->len && span->text[i].c == 32)
deletecharacter(span, i);
else if (i == span->len && span->eol)
span->eol = 0;
i--;
}
}
}