Refactor to be more input/output, remove globals from parsing

This commit is contained in:
Krzosa Karol
2024-06-19 07:11:18 +02:00
parent a75c8a2e4f
commit 055be9b058

View File

@@ -8,14 +8,6 @@
#include <semaphore>
#include <mutex>
struct XToTimeString {
String string; // String inside transcript arena
uint16_t hour;
uint16_t minute;
uint16_t second;
String filepath;
};
struct TimeString {
uint16_t hour;
uint16_t minute;
@@ -23,15 +15,6 @@ struct TimeString {
String string;
};
struct TimeFile {
Array<TimeString> time_strings;
String file;
};
bool AppInitializedWithFolder;
Arena XArena;
Arena Perm;
Array<TimeString> ParseSrtFile(Arena *arena, String filename) {
String content = ReadFile(*arena, filename);
Array<String> lines = Split(*arena, content, "\n");
@@ -75,31 +58,46 @@ Array<TimeString> ParseSrtFile(Arena *arena, String filename) {
return time_strings;
}
Array<Arena *> ArenasToFree;
Array<TimeFile> TimeFiles;
std::mutex ParseThreadMutex;
void ParseThreadEntry(Array<String> files) {
Arena *arena = AllocArena();
ParseThreadMutex.lock();
ArenasToFree.add(arena);
ParseThreadMutex.unlock();
For(files) {
Array<TimeString> time_strings = ParseSrtFile(arena, it);
ParseThreadMutex.lock();
TimeFiles.add({time_strings, it});
ParseThreadMutex.unlock();
struct TimeFile {
Array<TimeString> time_strings;
String file;
};
struct ParseThreadIO {
Array<String> input_files;
// output
Arena *arena;
Array<TimeFile> time_files;
};
void ParseThreadEntry(ParseThreadIO *io) {
io->arena = AllocArena();
io->time_files.allocator = *io->arena;
For(io->input_files) {
Array<TimeString> time_strings = ParseSrtFile(io->arena, it);
io->time_files.add({time_strings, it});
}
}
Array<XToTimeString> XToTimeStringArray = {Perm};
Array<String> InitForFolder(String folder) {
Scratch scratch;
ArenasToFree.allocator = scratch;
TimeFiles.allocator = scratch;
struct XToTimeString {
String string; // String inside transcript arena
uint16_t hour;
uint16_t minute;
uint16_t second;
String filepath;
};
int thread_count = 16;
Array<String> filenames = {Perm};
Array<String> srt_files = {scratch};
Arena Perm;
Array<XToTimeString> XToTimeStringArray = {Perm};
bool AppInitializedWithFolder;
Arena XArena;
Array<String> InitForFolder(String folder) {
Scratch scratch;
Array<String> filenames = {Perm};
Array<String> srt_files = {scratch};
for (FileIter iter = IterateFiles(scratch, folder); IsValid(iter); Advance(&iter)) {
filenames.add(Copy(Perm, iter.absolute_path));
if (EndsWith(iter.filename, ".srt")) {
@@ -107,17 +105,24 @@ Array<String> InitForFolder(String folder) {
}
}
int thread_count = 16;
Array<std::thread *> threads = {scratch};
int files_per_thread = srt_files.len / thread_count;
int remainder = srt_files.len % thread_count;
int fi = 0;
Array<ParseThreadIO> io = {scratch};
io.reserve(thread_count);
for (int ti = 0; ti < thread_count; ti += 1) {
Array<String> files = {scratch};
for (int i = 0; fi < srt_files.len && i < files_per_thread + remainder; fi += 1, i += 1) {
files.add(srt_files[fi]);
}
if (remainder) remainder = 0;
threads.add(new std::thread(ParseThreadEntry, files));
ParseThreadIO *i = io.alloc();
i->input_files = files;
threads.add(new std::thread(ParseThreadEntry, i));
}
For(threads) {
@@ -125,14 +130,16 @@ Array<String> InitForFolder(String folder) {
delete it;
}
ForItem(file, TimeFiles) {
For(file.time_strings) {
String s = Copy(XArena, it.string);
s.data[s.len] = ' ';
XToTimeStringArray.add({s, it.hour, it.minute, it.second, file.file});
ForItem(it_io, io) {
ForItem(it_time_file, it_io.time_files) {
For(it_time_file.time_strings) {
String s = Copy(XArena, it.string);
s.data[s.len] = ' ';
XToTimeStringArray.add({s, it.hour, it.minute, it.second, it_time_file.file});
}
}
Release(it_io.arena);
}
For(ArenasToFree) Release(it);
AppInitializedWithFolder = true;
return filenames;