|
#include <chrono>
|
|
#include <iomanip>
|
|
#include <iostream>
|
|
#include <sys/stat.h>
|
|
#include <sys/wait.h>
|
|
#include <unistd.h>
|
|
|
|
#include "PhotoCompressArchiver.hh"
|
|
|
|
#define OUT(x) {\
|
|
std::lock_guard<std::mutex> lock(m_output_mutex);\
|
|
x\
|
|
}
|
|
|
|
const char *PROG_NAME = "./packJPG";
|
|
const char *PROG_OPT1 = "-np";
|
|
const char *PROG_OPT2 = "-o";
|
|
const char *PROG_OPT3 = "-p";
|
|
|
|
PhotoCompressArchiver::PhotoCompressArchiver(
|
|
const char* path,
|
|
uint32_t num_cores)
|
|
: m_path(path),
|
|
m_num_cores(num_cores),
|
|
m_files_processed(0),
|
|
m_num_input_files(0),
|
|
m_total_uncompressed_size(0),
|
|
m_total_compressed_size(0),
|
|
p_finder_thread(nullptr)
|
|
{}
|
|
|
|
PhotoCompressArchiver::~PhotoCompressArchiver()
|
|
{}
|
|
|
|
int PhotoCompressArchiver::execute()
|
|
{
|
|
// Start the file finder thread
|
|
OUT(std::cout << "Starting file finder..." << std::flush;);
|
|
p_finder_thread = new std::thread(&PhotoCompressArchiver::find_files, this,
|
|
m_path, JPEG_REGEX);
|
|
OUT(std::cout << "DONE" << std::endl;)
|
|
p_finder_thread->join();
|
|
OUT(std::cout << "File finder completed:" << std::endl;)
|
|
|
|
OUT(std::cout << " Found " << m_num_input_files << " images" << std::endl;)
|
|
OUT(std::cout << " Total uncompressed bytes: " << m_total_uncompressed_size
|
|
<< "\n" << std::endl;)
|
|
|
|
// Start worker threads
|
|
OUT(std::cout << "Starting worker threads..." << std::endl;)
|
|
|
|
// TODO: need to handle the cause were number of input files is less than
|
|
// the number of threads
|
|
uint32_t num_threads = m_num_cores;
|
|
uint32_t sublist_len = m_num_input_files / num_threads;
|
|
|
|
for (uint32_t idx = 0; idx < num_threads; ++idx)
|
|
{
|
|
// Subdivide the input file list (using the vector copy constructor
|
|
// and start/end iterators)
|
|
decltype(m_file_list.begin()) start_iter;
|
|
decltype(m_file_list.begin()) end_iter;
|
|
if (idx == (num_threads - 1))
|
|
{
|
|
// Make sure the last sublist goes all the way to the end of the
|
|
// input file list (if number of images is not an even multiple of
|
|
// number of threads)
|
|
start_iter = m_file_list.begin() + (idx * sublist_len);
|
|
end_iter = m_file_list.end();
|
|
}
|
|
else
|
|
{
|
|
start_iter = m_file_list.begin() + (idx * sublist_len);
|
|
end_iter = m_file_list.begin() + ((idx + 1) * sublist_len);
|
|
}
|
|
|
|
auto sublist = new std::vector<bfs::path>(start_iter, end_iter);
|
|
|
|
// Spawn worker thread and add to list for bookkeeping
|
|
std::thread* worker = new std::thread(&PhotoCompressArchiver::fork_worker,
|
|
this, idx, sublist);
|
|
m_worker_threads.push_back(worker);
|
|
}
|
|
OUT(std::cout << "All worker threads started" << std::endl;)
|
|
|
|
// Clean up the finder thread
|
|
delete p_finder_thread;
|
|
p_finder_thread = nullptr;
|
|
|
|
// Now join the worker threads
|
|
for (auto worker : m_worker_threads)
|
|
{
|
|
worker->join();
|
|
delete worker;
|
|
}
|
|
|
|
OUT(std::cout << "\n Total compressed bytes: " << m_total_compressed_size
|
|
<< std::endl;)
|
|
|
|
// Calculate and display ratio
|
|
double ratio = (double)m_total_compressed_size / m_total_uncompressed_size;
|
|
OUT(std::cout << " Compression ratio: " << std::fixed
|
|
<< std::setprecision(4) << (ratio * 100) << "%\n" << std::endl;)
|
|
|
|
return 0;
|
|
}
|
|
|
|
void PhotoCompressArchiver::find_files(
|
|
const bfs::path& dir_path,
|
|
const boost::regex& file_regex)
|
|
{
|
|
bfs::directory_iterator end_iter;
|
|
for (bfs::directory_iterator dir_iter(dir_path); dir_iter != end_iter; ++dir_iter)
|
|
{
|
|
if (bfs::is_directory(dir_iter->status()))
|
|
{
|
|
// Found directory recurse
|
|
find_files(dir_iter->path(), file_regex);
|
|
}
|
|
else if (boost::regex_match(dir_iter->path().filename().string(), file_regex))
|
|
{
|
|
// Found file match
|
|
m_file_list.push_back(dir_iter->path());
|
|
++m_num_input_files;
|
|
|
|
// Stat the file to get its size
|
|
struct stat statbuf;
|
|
int rc = stat(dir_iter->path().string().c_str(), &statbuf);
|
|
if (rc)
|
|
{
|
|
OUT(std::cerr << "ERROR: unable to stat file "
|
|
<< dir_iter->path() << std::endl;)
|
|
}
|
|
m_total_uncompressed_size += statbuf.st_size;
|
|
}
|
|
}
|
|
}
|
|
|
|
void PhotoCompressArchiver::fork_worker(
|
|
uint32_t tid,
|
|
std::vector<bfs::path>* file_sublist)
|
|
{
|
|
uint32_t sublist_file_count = file_sublist->size();
|
|
uint32_t files_processed = 0;
|
|
|
|
OUT(std::cout << "T[" << tid << "] top of the morning to ya: "
|
|
<< sublist_file_count << std::endl;)
|
|
|
|
// Print out the input files given to this worker
|
|
#if 0
|
|
for (auto& filepath : *file_sublist)
|
|
{
|
|
OUT(std::cout << "T[" << tid << "] file: " << filepath.string()
|
|
<< std::endl;)
|
|
}
|
|
#endif
|
|
|
|
uint32_t argv_size = file_sublist->size() + 5;
|
|
const char **exec_argv = new const char* [argv_size];
|
|
|
|
// Build the argv array; pass in default program options
|
|
exec_argv[0] = &PROG_NAME[2]; // discard leading "./"
|
|
exec_argv[1] = PROG_OPT1;
|
|
exec_argv[2] = PROG_OPT2;
|
|
exec_argv[3] = PROG_OPT3;
|
|
|
|
// Add the sublist files to the argv array
|
|
uint32_t idx = 4;
|
|
for (auto& image_path : *file_sublist)
|
|
{
|
|
exec_argv[idx++] = image_path.string().c_str();
|
|
}
|
|
exec_argv[argv_size - 1] = nullptr;
|
|
|
|
|
|
// Build a list of the output files names for use in output file monitoring
|
|
// below. Output file names have ".jpg" or ".jpeg" replaced with ".pjg"
|
|
std::vector<std::string*> output_filenames;
|
|
for (auto& filepath : *file_sublist)
|
|
{
|
|
// Create new filename that contains the ".jpg" replaced with ".pjg"
|
|
uint32_t end_pos = filepath.string().find(".");
|
|
std::string* filename = new std::string(filepath.string().begin(),
|
|
filepath.string().begin() + end_pos);
|
|
(*filename) += ".pjg";
|
|
|
|
output_filenames.push_back(filename);
|
|
}
|
|
|
|
// Create a pipe to hold stdout from child process
|
|
int filedes[2];
|
|
if (pipe(filedes) == -1)
|
|
{
|
|
perror("pipe");
|
|
exit(1);
|
|
}
|
|
|
|
// Fork off the childprocess
|
|
//~ pid_t parent = getpid();
|
|
pid_t pid = vfork();
|
|
|
|
if (pid == -1)
|
|
{
|
|
// error, failed to fork()
|
|
OUT(std::cout << "T[" << tid << "] fork failed!" << std::endl;)
|
|
}
|
|
else if (pid == 0)
|
|
{
|
|
// This is the child process...
|
|
// Set child process's stdout to the pipe entry
|
|
while ((dup2(filedes[1], STDOUT_FILENO) == -1) && (errno == EINTR)) {}
|
|
close(filedes[1]);
|
|
close(filedes[0]);
|
|
|
|
// Child after fork
|
|
execv(PROG_NAME, (char **)exec_argv);
|
|
_exit(EXIT_FAILURE); // exec never returns
|
|
}
|
|
|
|
// This is the parent process...
|
|
|
|
// This is a little silly... but it works. We know the order in which the
|
|
// output files will be created so we wait until output files X + 1 exists
|
|
// (i.e. we can stat() it) and then process file X.
|
|
const uint32_t BAILOUT_MAX = 600; // 30s in 50ms increments
|
|
uint32_t bailout_ctr = 0;
|
|
uint32_t output_idx = 1;
|
|
int stat_rc = 0;
|
|
struct stat statbuf;
|
|
double complete_percent = 0.0;
|
|
while (output_idx < output_filenames.size())
|
|
{
|
|
// Poll, waiting for output file X + 1 to exist
|
|
bailout_ctr = 0;
|
|
while ((stat(output_filenames[output_idx]->c_str(), &statbuf) != 0) &&
|
|
(++bailout_ctr < BAILOUT_MAX))
|
|
{
|
|
std::this_thread::sleep_for(std::chrono::milliseconds(50));
|
|
}
|
|
|
|
// Check if timeout has occurred
|
|
if (bailout_ctr == BAILOUT_MAX)
|
|
{
|
|
OUT(std::cerr << "T[" << tid << "] ERROR: timed out while waiting "
|
|
<< "for output file: "
|
|
<< *output_filenames[output_idx] << "; aborting"
|
|
<< std::endl;)
|
|
return;
|
|
}
|
|
|
|
// Now stat output file X which should exist
|
|
stat_rc = stat(output_filenames[output_idx - 1]->c_str(), &statbuf);
|
|
if (stat_rc)
|
|
{
|
|
OUT(std::cerr << "ERROR while stating: "
|
|
<< *output_filenames[output_idx - 1] << std::endl;)
|
|
}
|
|
|
|
// Increment counts
|
|
m_total_compressed_size += statbuf.st_size;
|
|
++m_files_processed;
|
|
++files_processed;
|
|
|
|
// Calculate the local percent done
|
|
complete_percent = (double)files_processed / sublist_file_count;
|
|
complete_percent *= 100;
|
|
|
|
OUT(std::cout << "T[" << tid << "] " << std::setw(6)
|
|
<< std::fixed << std::setprecision(2)
|
|
<< get_global_percent_done()
|
|
<< "% -- (" << files_processed << "/"
|
|
<< sublist_file_count << " -- " << std::setw(6)
|
|
<< complete_percent
|
|
<< "%) file: " << *output_filenames[output_idx - 1]
|
|
<< " -- " << statbuf.st_size << std::endl;)
|
|
|
|
++output_idx;
|
|
}
|
|
|
|
// Wait for forked child process to terminate
|
|
int status;
|
|
waitpid(pid, &status, 0);
|
|
|
|
// Now that the forked process has completed, we can handle the last output
|
|
// file
|
|
stat_rc = stat(output_filenames[output_idx - 1]->c_str(), &statbuf);
|
|
if (stat_rc)
|
|
{
|
|
OUT(std::cerr << "ERROR while stating: "
|
|
<< *output_filenames[output_idx - 1] << std::endl;)
|
|
}
|
|
|
|
// Increment counts
|
|
m_total_compressed_size += statbuf.st_size;
|
|
++m_files_processed;
|
|
++files_processed;
|
|
|
|
// Calculate the local percent done
|
|
complete_percent = (double)files_processed / sublist_file_count;
|
|
complete_percent *= 100;
|
|
|
|
OUT(std::cout << "T[" << tid << "] " << std::setw(6) << std::fixed
|
|
<< std::setprecision(2) << get_global_percent_done()
|
|
<< "% -- (" << files_processed << "/" << sublist_file_count
|
|
<< " -- " << std::setw(5) << complete_percent << "%) file: "
|
|
<< *output_filenames[output_idx - 1] << " -- "
|
|
<< statbuf.st_size << std::endl;)
|
|
|
|
OUT(std::cout << "T[" << tid << "] complete ==> RC: " << status << std::endl;)
|
|
|
|
// Free up the argv array and output filename list
|
|
delete[] exec_argv;
|
|
for (auto output_filename : output_filenames)
|
|
{
|
|
delete output_filename;
|
|
}
|
|
}
|
|
|
|
double PhotoCompressArchiver::get_global_percent_done()
|
|
{
|
|
return ((double)m_files_processed / m_num_input_files) * 100;
|
|
}
|