virtual void Visit(const WCHAR *name, const WCHAR *url, int level) {
if (!url || IsExternalUrl(url))
return;
ScopedMem<WCHAR> plainUrl(str::ToPlainUrl(url));
if (added.FindI(plainUrl) != -1)
return;
ScopedMem<char> urlUtf8(str::conv::ToUtf8(plainUrl));
size_t pageHtmlLen;
ScopedMem<unsigned char> pageHtml(doc->GetData(urlUtf8, &pageHtmlLen));
if (!pageHtml)
return;
html.AppendFmt("<pagebreak page_path=\"%s\" page_marker />", urlUtf8);
html.AppendAndFree(doc->ToUtf8(pageHtml, ExtractHttpCharset((const char *)pageHtml.Get(), pageHtmlLen)));
added.Append(plainUrl.StealData());
}
bool DjVuEngineImpl::ExtractPageText(miniexp_t item, const WCHAR *lineSep, str::Str<WCHAR>& extracted, Vec<RectI>& coords)
{
miniexp_t type = miniexp_car(item);
if (!miniexp_symbolp(type))
return false;
item = miniexp_cdr(item);
if (!miniexp_numberp(miniexp_car(item))) return false;
int x0 = miniexp_to_int(miniexp_car(item)); item = miniexp_cdr(item);
if (!miniexp_numberp(miniexp_car(item))) return false;
int y0 = miniexp_to_int(miniexp_car(item)); item = miniexp_cdr(item);
if (!miniexp_numberp(miniexp_car(item))) return false;
int x1 = miniexp_to_int(miniexp_car(item)); item = miniexp_cdr(item);
if (!miniexp_numberp(miniexp_car(item))) return false;
int y1 = miniexp_to_int(miniexp_car(item)); item = miniexp_cdr(item);
RectI rect = RectI::FromXY(x0, y0, x1, y1);
miniexp_t str = miniexp_car(item);
if (miniexp_stringp(str) && !miniexp_cdr(item)) {
if (type != miniexp_symbol("char") && type != miniexp_symbol("word") ||
coords.Count() > 0 && rect.y < coords.Last().y - coords.Last().dy * 0.8) {
AppendNewline(extracted, coords, lineSep);
}
const char *content = miniexp_to_str(str);
WCHAR *value = str::conv::FromUtf8(content);
if (value) {
size_t len = str::Len(value);
// TODO: split the rectangle into individual parts per glyph
for (size_t i = 0; i < len; i++)
coords.Append(RectI(rect.x, rect.y, rect.dx, rect.dy));
extracted.AppendAndFree(value);
}
if (miniexp_symbol("word") == type) {
extracted.Append(' ');
coords.Append(RectI(rect.x + rect.dx, rect.y, 2, rect.dy));
}
item = miniexp_cdr(item);
}
while (miniexp_consp(str)) {
ExtractPageText(str, lineSep, extracted, coords);
item = miniexp_cdr(item);
str = miniexp_car(item);
}
return !item;
}