Project

General

Profile

Download (15.8 KB) Statistics
| Branch: | Tag: | Revision:

root/software/photo_compress_archiver/PhotoCompressArchiver.cc @ 0b45b84e

b216a5b0 David Sorber
#include <chrono>
328d0fa5 David Sorber
#include <cstdio>
f8e600bc David Sorber
#include <iomanip>
b216a5b0 David Sorber
#include <iostream>
e0b7f8b9 David Sorber
#include <limits.h>
b216a5b0 David Sorber
#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\
}

3eeb1485 David Sorber
const uint32_t PTR_LEN = sizeof(char*);
const uint32_t BAILOUT_MAX = 600; // 30s in 50ms increments
const uint32_t FILES_PER_WORK_UNIT = 40;

e0b7f8b9 David Sorber
const char *PROG_NAME = "./packjpg";
b216a5b0 David Sorber
const char *PROG_OPT1 = "-np";
const char *PROG_OPT2 = "-o";
const char *PROG_OPT3 = "-p";

e0b7f8b9 David Sorber
const uint32_t ARGV_MAX_SIZE(sysconf(_SC_ARG_MAX));
57e6c22f David Sorber
const uint32_t ARGV_EXTRA(4096);
e0b7f8b9 David Sorber
b216a5b0 David Sorber
PhotoCompressArchiver::PhotoCompressArchiver(
328d0fa5 David Sorber
const std::string& path,
uint32_t num_cores,
bool decompress,
e0b7f8b9 David Sorber
bool keep_orig,
boost::regex& filter_regex)
b216a5b0 David Sorber
: m_path(path),
m_num_cores(num_cores),
328d0fa5 David Sorber
m_decompress(decompress),
m_keep_orig(keep_orig),
f8e600bc David Sorber
m_files_processed(0),
b216a5b0 David Sorber
m_num_input_files(0),
m_total_uncompressed_size(0),
m_total_compressed_size(0),
e0b7f8b9 David Sorber
p_finder_thread(nullptr),
m_filter_empty(true),
m_filter_regex(filter_regex)
{
// This is a somewhat crude way to determining if the filter regex is empty
m_filter_empty = std::string("").compare(filter_regex.str()) == 0;
57e6c22f David Sorber
// Remove this...
e0b7f8b9 David Sorber
std::cout << "argv size: " << sysconf(_SC_ARG_MAX) << std::endl;
}
b216a5b0 David Sorber
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,
328d0fa5 David Sorber
m_path,
(m_decompress ? PJG_REGEX : JPEG_REGEX));
b216a5b0 David Sorber
OUT(std::cout << "DONE" << std::endl;)
3eeb1485 David Sorber
// Start worker threads
OUT(std::cout << "Starting worker threads..." << std::flush;)
uint32_t num_threads = m_num_cores;
for (uint32_t idx = 0; idx < num_threads; ++idx)
{
// Spawn worker thread and add to list for bookkeeping
std::thread* worker = new std::thread(&PhotoCompressArchiver::fork_worker,
this, idx);
m_worker_threads.push_back(worker);
}
OUT(std::cout << "DONE\n" << std::endl;)
// Join the file finder thread
b216a5b0 David Sorber
p_finder_thread->join();
OUT(std::cout << "File finder completed:" << std::endl;)
57e6c22f David Sorber
// Error out of no input files were found
if (m_num_input_files == 0)
{
OUT(std::cerr << BOLD << RED << "ERROR: " << ENDC << "No matching "
<< "input files found. Please check the patch and try "
<< "again. " << std::endl;)
return -1;
}
3eeb1485 David Sorber
OUT(std::cout << " Found " << m_num_input_files << " files" << std::endl;)
328d0fa5 David Sorber
if (! m_decompress)
{
OUT(std::cout << " Total uncompressed bytes: "
<< m_total_uncompressed_size << "\n" << std::endl;)
}
b216a5b0 David Sorber
3eeb1485 David Sorber
// Build work unit queue
uint32_t input_list_idx = 0;
uint32_t queue_size = 0;
while (input_list_idx < m_file_list.size())
b216a5b0 David Sorber
{
3eeb1485 David Sorber
// Safely read the size of the queue
b216a5b0 David Sorber
{
3eeb1485 David Sorber
std::lock_guard<std::mutex> lock(m_work_unit_mtx);
queue_size = m_work_unit_queue.size();
b216a5b0 David Sorber
}
3eeb1485 David Sorber
// Wait until the queue has decreased in the size before proceeding,
// otherwise we could chew up a bunch of memory
if (queue_size > num_threads)
b216a5b0 David Sorber
{
3eeb1485 David Sorber
std::this_thread::sleep_for(std::chrono::milliseconds(250));
continue;
b216a5b0 David Sorber
}
3eeb1485 David Sorber
WorkUnit* work_unit = new WorkUnit();
work_unit->exec_argv.push_back(&PROG_NAME[2]);
work_unit->exec_argv.push_back(PROG_OPT1);
work_unit->exec_argv.push_back(PROG_OPT2);
work_unit->exec_argv.push_back(PROG_OPT3);
uint32_t argv_size = sizeof(PROG_NAME) + sizeof(PROG_OPT1) +
sizeof(PROG_OPT2) + sizeof(PROG_OPT3) + ARGV_EXTRA;
b216a5b0 David Sorber
3eeb1485 David Sorber
uint32_t local_count = 0;
while ((input_list_idx < m_file_list.size()) &&
(local_count < FILES_PER_WORK_UNIT))
{
bfs::path& file_path = m_file_list[input_list_idx];
uint32_t path_len = file_path.string().size();
// Check if adding this next file would make the argv too big; if
// so bail out
if (argv_size + path_len + PTR_LEN > ARGV_MAX_SIZE)
{
break;
}
// Add the file to the argv array
work_unit->exec_argv.push_back(file_path.string().c_str());
argv_size += path_len + PTR_LEN;
// Create new filename for expected
uint32_t end_pos = file_path.string().rfind(".");
std::string* filename = new std::string(file_path.string().begin(),
file_path.string().begin()
+ end_pos);
// Add extension depending on if decompress mode is enabled or not
(*filename) += (m_decompress ? ".jpg" : ".pjg");
work_unit->output_filenames.push_back(filename);
// Increment our index to the next file
++input_list_idx;
++local_count;
}
work_unit->exec_argv.push_back(nullptr);
// Add the create work unit to the queue; and notify the worker threads
// that it has been handed off
{
std::lock_guard<std::mutex> lock(m_work_unit_mtx);
m_work_unit_queue.push_back(work_unit);
}
m_work_unit_cv.notify_all();
b216a5b0 David Sorber
}
3eeb1485 David Sorber
// Add null terminate indicator for each thread
{
std::lock_guard<std::mutex> lock(m_work_unit_mtx);
for (uint32_t idx = 0; idx < num_threads; ++idx)
{

m_work_unit_queue.push_back(nullptr);
}
}
m_work_unit_cv.notify_all();
b216a5b0 David Sorber
// 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;
}
cec188e4 David Sorber
// Print out compression information if compressing
328d0fa5 David Sorber
if (! m_decompress)
{
OUT(std::cout << "\n Total compressed bytes: "
<< m_total_compressed_size
<< std::endl;)
f6c21781 David Sorber
328d0fa5 David Sorber
// 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;)
}
b216a5b0 David Sorber
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
e0b7f8b9 David Sorber
// If the filter is not empty, then check against it before proceeding
if (! m_filter_empty)
{
if (boost::regex_match(dir_iter->path().filename().string(),
m_filter_regex))
{
// The filter regex matched, skip this file
continue;
}
}
b216a5b0 David Sorber
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)
{
328d0fa5 David Sorber
OUT(std::cerr << BOLD << RED << "ERROR: " << ENDC
<< "unable to stat file " << dir_iter->path()
<< std::endl;)
b216a5b0 David Sorber
}
m_total_uncompressed_size += statbuf.st_size;
}
}
}

void PhotoCompressArchiver::fork_worker(
3eeb1485 David Sorber
uint32_t tid)
b216a5b0 David Sorber
{
3eeb1485 David Sorber
// Loop vars
57e6c22f David Sorber
int status;
3eeb1485 David Sorber
uint32_t work_unit_items = 0;
uint32_t bailout_ctr = 0;
uint32_t output_idx = 1;
int stat_rc = 0;
struct stat statbuf;
uint32_t files_processed = 0;
double complete_percent = 0.0;
e0b7f8b9 David Sorber
3eeb1485 David Sorber
while (true)
b216a5b0 David Sorber
{
57e6c22f David Sorber
3eeb1485 David Sorber
std::unique_lock<std::mutex> mux_lock(m_work_unit_mtx);
while (m_work_unit_queue.empty())
{
m_work_unit_cv.wait_for(mux_lock, std::chrono::milliseconds(20));
}

// While holding the mutex grab the next chunk off of the add chunk
// queue
WorkUnit* work_unit = m_work_unit_queue.front();
m_work_unit_queue.pop_front();
mux_lock.unlock();
b216a5b0 David Sorber
3eeb1485 David Sorber
if (work_unit == nullptr)
e0b7f8b9 David Sorber
{
3eeb1485 David Sorber
break;
e0b7f8b9 David Sorber
}
57e6c22f David Sorber
3eeb1485 David Sorber
work_unit_items = work_unit->output_filenames.size();
43c9a095 David Sorber
#if 0
c8352650 David Sorber
// DEBUGGING
43c9a095 David Sorber
uint32_t idx = 4;
2423ae90 David Sorber
OUT(
3eeb1485 David Sorber
for (auto outfile : work_unit->output_filenames)
c8352650 David Sorber
{
2423ae90 David Sorber
std::cout << "T[" << tid << "] output file: " << *outfile
43c9a095 David Sorber
<< "\n input file: "
3eeb1485 David Sorber
<< static_cast<const char*>(work_unit->exec_argv[idx++])
<< std::endl;
c8352650 David Sorber
}
2423ae90 David Sorber
)
3eeb1485 David Sorber
continue;
c8352650 David Sorber
#endif
57e6c22f David Sorber
// Create a pipe to hold stdout from child process
int filedes[2];
if (pipe(filedes) == -1)
b216a5b0 David Sorber
{
57e6c22f David Sorber
perror("pipe");
exit(1);
f8e600bc David Sorber
}
57e6c22f David Sorber
// Fork off the childprocess
//~ pid_t parent = getpid();
pid_t pid = vfork();
if (pid == -1)
f8e600bc David Sorber
{
57e6c22f David Sorber
// 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
2423ae90 David Sorber
while ((dup2(filedes[1], STDOUT_FILENO) == -1) && (errno == EINTR)) {}
close(filedes[0]);
close(filedes[1]);
cec188e4 David Sorber
57e6c22f David Sorber
// Child after fork
3eeb1485 David Sorber
int rc = execv(PROG_NAME, (char **)work_unit->exec_argv.data());
57e6c22f David Sorber
if (rc)
{
std::cerr << "execv failed: " << errno << std::endl;
}
_exit(EXIT_FAILURE); // exec never returns
f8e600bc David Sorber
}
57e6c22f David Sorber
// 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.
3eeb1485 David Sorber
bailout_ctr = 0;
output_idx = 1;
stat_rc = 0;
files_processed = 0;
complete_percent = 0.0;
while (output_idx < work_unit_items)
57e6c22f David Sorber
{
3eeb1485 David Sorber
// Poll, waiting for output file X + 1 (aka "z") to exist
57e6c22f David Sorber
bailout_ctr = 0;
3eeb1485 David Sorber
const char* filename_z = work_unit->output_filenames[output_idx]->c_str();
while ((stat(filename_z, &statbuf) != 0) && (++bailout_ctr < BAILOUT_MAX))
57e6c22f David Sorber
{
std::this_thread::sleep_for(std::chrono::milliseconds(50));
}
// Check if timeout has occurred
if (bailout_ctr == BAILOUT_MAX)
{
OUT(std::cerr << "T[" << tid << "] " << BOLD << RED << "ERROR: "
<< ENDC << "timed out while waiting "
<< "for output file: "
3eeb1485 David Sorber
<< *work_unit->output_filenames[output_idx]
<< "; aborting" << std::endl;)
57e6c22f David Sorber
// The subprocess timed out... get its status
waitpid(pid, &status, 0);
OUT(std::cerr << "T[" << tid << "] " << BOLD << RED << "ERROR: "
<< ENDC << PURPLE << " RC => " << status << ENDC
<< std::endl;)
f8e600bc David Sorber
57e6c22f David Sorber
return;
}
3eeb1485 David Sorber
// Now stat output file X (aka "y") which should exist
const char* filename_y = work_unit->output_filenames[output_idx - 1]->c_str();
stat_rc = stat(filename_y, &statbuf);
57e6c22f David Sorber
if (stat_rc)
{
OUT(std::cerr << "T[" << tid << "] " << BOLD << RED << "ERROR "
3eeb1485 David Sorber
<< ENDC << "while stating: " << filename_y
<< std::endl;)
57e6c22f David Sorber
}
// Increment counts
m_total_compressed_size += statbuf.st_size;
++m_files_processed;
++files_processed;
// Unless keeping original files, remove the source file
if (! m_keep_orig)
{
3eeb1485 David Sorber
std::remove(work_unit->exec_argv[4 + output_idx - 1]);
57e6c22f David Sorber
}
// Calculate the local percent done
3eeb1485 David Sorber
complete_percent = (double)files_processed / work_unit_items;
57e6c22f David Sorber
complete_percent *= 100;
OUT(std::cout << "T[" << tid << "] " << std::setw(6)
<< std::fixed << std::setprecision(2)
<< get_global_percent_done()
3eeb1485 David Sorber
<< "% -- (" << std::setw(2) << files_processed << "/"
<< work_unit_items << " -- " << std::setw(6)
57e6c22f David Sorber
<< complete_percent
3eeb1485 David Sorber
<< "%) file: " << filename_y
57e6c22f David Sorber
<< " -- " << statbuf.st_size << std::endl;)
++output_idx;
}
// Wait for forked child process to terminate
waitpid(pid, &status, 0);
2423ae90 David Sorber
close(filedes[0]);
close(filedes[1]);
57e6c22f David Sorber
// Now that the forked process has completed, we can handle the last
// output file
3eeb1485 David Sorber
const char* filename_y = work_unit->output_filenames[output_idx - 1]->c_str();
stat_rc = stat(filename_y, &statbuf);
f8e600bc David Sorber
if (stat_rc)
{
57e6c22f David Sorber
OUT(std::cerr << "T[" << tid << "] " << BOLD << RED << "ERROR "
3eeb1485 David Sorber
<< ENDC << "while stating: " << filename_y << std::endl;)
f8e600bc David Sorber
}
// Increment counts
b216a5b0 David Sorber
m_total_compressed_size += statbuf.st_size;
f8e600bc David Sorber
++m_files_processed;
++files_processed;
328d0fa5 David Sorber
// Unless keeping original files, remove the source file
if (! m_keep_orig)
{
3eeb1485 David Sorber
std::remove(work_unit->exec_argv[4 + output_idx - 1]);
328d0fa5 David Sorber
}
f8e600bc David Sorber
// Calculate the local percent done
3eeb1485 David Sorber
complete_percent = (double)files_processed / work_unit_items;
f8e600bc David Sorber
complete_percent *= 100;
57e6c22f David Sorber
OUT(std::cout << "T[" << tid << "] " << std::setw(6) << std::fixed
<< std::setprecision(2) << get_global_percent_done()
3eeb1485 David Sorber
<< "% -- (" << std::setw(2) << files_processed
<< "/" << work_unit_items << " -- " << std::setw(5)
<< complete_percent << "%) file: " << filename_y
f8e600bc David Sorber
<< " -- " << statbuf.st_size << std::endl;)
57e6c22f David Sorber
// Free up the output filename list
3eeb1485 David Sorber
for (auto output_filename : work_unit->output_filenames)
57e6c22f David Sorber
{
delete output_filename;
}
3eeb1485 David Sorber
delete work_unit;
328d0fa5 David Sorber
}
57e6c22f David Sorber
OUT(std::cout << "T[" << tid << "] " << PURPLE << "complete ==> RC: "
328d0fa5 David Sorber
<< status << ENDC << std::endl;)
f8e600bc David Sorber
}

double PhotoCompressArchiver::get_global_percent_done()
{
return ((double)m_files_processed / m_num_input_files) * 100;
b216a5b0 David Sorber
}