void FileScanner::Run(int thread_index)
{
// Set the name of the thread.
std::stringstream temp_ss;
temp_ss << "FILESCAN_";
temp_ss << thread_index;
set_thread_name(temp_ss.str());
if(m_manually_assign_cores)
{
// Spread the scanner threads across cores. Linux at least doesn't seem to want to do that by default.
AssignToNextCore();
}
// Create a reusable, resizable buffer for the File() reads.
auto file_data_storage = std::make_shared<ResizableArray<char>>();
// Pull new filenames off the input queue until it's closed.
std::string next_string;
while(m_in_queue.wait_pull(std::move(next_string)) != queue_op_status::closed)
{
MatchList ml(next_string);
try
{
// Try to open and read the file. This could throw.
LOG(INFO) << "Attempting to scan file \'" << next_string << "\'";
File f(next_string, file_data_storage);
if(f.size() == 0)
{
LOG(INFO) << "WARNING: Filesize of \'" << next_string << "\' is 0, skipping.";
continue;
}
const char *file_data = f.data();
size_t file_size = f.size();
// Scan the file data for occurrences of the regex, sending matches to the MatchList ml.
ScanFile(file_data, file_size, ml);
if(!ml.empty())
{
// Force move semantics here.
m_output_queue.wait_push(std::move(ml));
}
}
catch(const FileException &error)
{
// The File constructor threw an exception.
ERROR() << error.what();
}
catch(const std::system_error& error)
{
// A system error. Currently should only be errors from File.
ERROR() << error.code() << " - " << error.code().message();
}
catch(...)
{
// Rethrow whatever it was.
throw;
}
}
}
void FileScanner::Run()
{
if(m_manually_assign_cores)
{
// Spread the scanner threads across cores. Linux at least doesn't seem to want to do that by default.
AssignToNextCore();
}
#ifndef HAVE_LIBPCRE
// Create the std::regex we're looking for, possibly ignoring case, possibly with match-whole-word.
auto stack_regex = m_regex;
if(m_word_regexp)
{
// Surround the regex with \b (word boundary) assertions.
stack_regex = "\\b(?:" + m_regex + ")\\b";
}
std::regex expression(stack_regex,
std::regex_constants::ECMAScript |
std::regex_constants::optimize |
m_ignore_case ? std::regex_constants::icase : static_cast<typeof(std::regex_constants::icase)>(0));
#endif
// Pull new filenames off the input queue until it's closed.
std::string next_string;
while(m_in_queue.wait_pull(std::move(next_string)) != queue_op_status::closed)
{
MatchList ml(next_string);
try
{
// Try to open and read the file. This could throw.
File f(next_string);
if(f.size() == 0)
{
//std::clog << "WARNING: Filesize of \"" << next_string << "\" is 0" << std::endl;
continue;
}
const char *file_data = f.data();
size_t file_size = f.size();
// Scan the file data for the regex.
#if HAVE_LIBPCRE
ScanFileLibPCRE(file_data, file_size, ml);
#else
ScanFileCpp11(expression, file_data, file_size, ml);
#endif
if(!ml.empty())
{
// Force move semantics here.
m_output_queue.wait_push(std::move(ml));
}
}
catch(const FileException &error)
{
// The File constructor threw an exception.
std::cerr << "ucg: ERROR: " << error.what() << std::endl;
}
catch(const std::system_error& error)
{
// A system error. Currently should only be errors from File.
std::cerr << "ucg: ERROR: " << error.code() << " - " << error.code().message() << std::endl;
}
catch(...)
{
// Rethrow whatever it was.
throw;
}
}
}