int VMPI_getchar_unlocked()
{
return _getchar_nolock();
}
int main(int argc, char *argv[])
{
// system("chcp 65001");
system("chcp 1251");
// Read HTML file contents
std::string htmlFileName = "C:\\Projects\\__DATA\\test_page_30.html";
std::ifstream htmlFileStream(htmlFileName, std::ios::in | std::ios::binary);
if (!htmlFileStream)
{
std::cout << "File " << htmlFileName << " not found!";
return( EXIT_FAILURE );
}
std::string htmlFileContents;
htmlFileStream.seekg(0, std::ios::end);
htmlFileContents.resize(htmlFileStream.tellg());
htmlFileStream.seekg(0, std::ios::beg);
htmlFileStream.read(&htmlFileContents[0], htmlFileContents.size());
htmlFileStream.close();
// Convert it to UTF-8
// NOTE: Gumbo works ONLY with UTF-8 documents
char* bufUtf8 = new char[htmlFileContents.length() * 2];
memset(bufUtf8, 0, htmlFileContents.length() * 2);
cp1251ToUtf8(bufUtf8, htmlFileContents.c_str());
// delete[] bufUtf8;
// return 0;
// Parse web page contents
GumboOutput* output = gumbo_parse(/*htmlFileContents.c_str()*/ bufUtf8 );
// search_for_links(output->root);
searchForDivBlocks(output->root);
gumbo_destroy_output(&kGumboDefaultOptions, output);
delete[] bufUtf8;
_getchar_nolock();
return 0;
}
