Big update

This commit is contained in:
Krzosa Karol
2024-07-12 08:27:19 +02:00
parent 37982e0448
commit 9891a302ac
5 changed files with 261 additions and 162 deletions

View File

@@ -1,68 +1,31 @@
struct TimeString {
uint16_t hour;
uint16_t minute;
uint16_t second;
String string;
};
Array<TimeString> ParseSrtFile(Arena *arena, String filename) {
String content = ReadFile(*arena, filename);
Array<String> lines = Split(*arena, content, "\n");
IterRemove(lines) {
IterRemovePrepare(lines);
it = Trim(it);
if (it.len == 0) remove_item = true;
}
long section_number = 1;
Array<TimeString> time_strings = {*arena};
for (int i = 0; i < lines.len;) {
String it0 = lines[i++];
long num = strtol(it0.data, NULL, 10);
Assert(section_number == num);
section_number += 1;
TimeString item = {};
String it1 = lines[i++];
item.hour = (uint16_t)strtol(it1.data, NULL, 10);
item.minute = (uint16_t)strtol(it1.data + 3, NULL, 10);
item.second = (uint16_t)strtol(it1.data + 6, NULL, 10);
String next_section_number = Format(*arena, "%d", section_number);
while (i < lines.len && lines[i] != next_section_number) {
String it = lines[i];
item.string = lines[i];
time_strings.add(item);
i += 1;
}
}
IterRemove(time_strings) {
IterRemovePrepare(time_strings);
if (i > 0 && AreEqual(time_strings[i - 1].string, time_strings[i].string, true)) {
remove_item = true;
}
}
return time_strings;
}
struct TimeFile {
Array<TimeString> time_strings;
String file;
};
struct ParseThreadIO {
Array<String> input_files;
};
struct XToTimeString {
String string; // String inside transcript arena
uint16_t hour;
uint16_t minute;
uint16_t second;
String filepath;
enum SourceKind {
SourceKind_Invalid,
SourceKind_SRT,
SourceKind_PDF,
SourceKind_TXT,
};
struct XToSource {
SourceKind kind;
String string; // String inside x arena
String filepath;
union {
struct {
uint16_t hour;
uint16_t minute;
uint16_t second;
} srt;
struct {
int page;
} pdf;
struct {
int64_t row;
} txt;
};
};
struct FileLoadResult {
@@ -72,7 +35,7 @@ struct FileLoadResult {
Arena XArena;
std::mutex XArenaAddMutex;
Array<XToTimeString> XToTime;
Array<XToSource> XToSourceArray;
Array<FileLoadResult> XFileLoadResults;
int64_t XLoadThreadComplete;
@@ -80,24 +43,49 @@ void XInitLoading() {
InitArena(&XArena);
XArena.align = 0;
XToTime.reserve(1000000);
XToSourceArray.reserve(1000000);
XFileLoadResults.reserve(10000);
}
WORK_FUNCTION(ParseFilesWork) {
ParseThreadIO *io = (ParseThreadIO *)data;
ForItem(it_time_file, io->input_files) {
Scratch scratch;
Array<TimeString> time_strings = ParseSrtFile(scratch, it_time_file);
Scratch scratch;
if (EndsWith(it_time_file, ".srt", true)) {
Array<TimeString> time_strings = ParseSrtFile(scratch, it_time_file);
XArenaAddMutex.lock();
For(time_strings) {
String s = Copy(XArena, it.string);
s.data[s.len] = ' ';
XToTime.add({s, it.hour, it.minute, it.second, it_time_file});
XArenaAddMutex.lock();
For(time_strings) {
String s = Copy(XArena, it.string);
s.data[s.len] = ' ';
XToSource t = {SourceKind_SRT, s, it_time_file};
t.srt = {it.hour, it.minute, it.second};
XToSourceArray.add(t);
}
XArenaAddMutex.unlock();
XFileLoadResults.bounded_add({it_time_file});
} else if (EndsWith(it_time_file, ".txt", true) || EndsWith(it_time_file, ".html", true)) {
String string = ReadFile(scratch, it_time_file);
if (string.data) {
XArenaAddMutex.lock();
Array<String> lines = Split(scratch, string, "\n");
For(lines) {
String s = Copy(XArena, it);
s.data[s.len] = ' ';
XToSource t = {SourceKind_TXT, s, it_time_file};
t.txt = {lines.get_index(it)};
XToSourceArray.add(t);
}
XArenaAddMutex.unlock();
XFileLoadResults.bounded_add({it_time_file});
} else {
XFileLoadResults.bounded_add({it_time_file, "failed to read the file"});
}
} else {
XFileLoadResults.bounded_add({it_time_file, "internal error: extension is not supported but got propagated to parse stage"});
}
XArenaAddMutex.unlock();
XFileLoadResults.bounded_add({it_time_file});
}
AtomicIncrement(&XLoadThreadComplete);
}
@@ -118,31 +106,33 @@ void XUnlockFileResults() {
void XAddFolder(String folder, Array<String> *filenames) {
Scratch scratch;
Array<String> srt_files = {scratch};
Array<String> files_to_parse = {scratch};
for (FileIter iter = IterateFiles(scratch, folder); IsValid(iter); Advance(&iter)) {
String file = Copy(Perm, iter.absolute_path);
filenames->add(file);
if (EndsWith(iter.filename, ".srt")) {
srt_files.add(file);
if (EndsWith(iter.filename, ".srt", true)) {
files_to_parse.add(file);
} else if (EndsWith(iter.filename, ".txt", true) || EndsWith(iter.filename, ".html", true)) {
files_to_parse.add(file);
}
}
if (srt_files.len == 0) {
if (files_to_parse.len == 0) {
XFileLoadResults.add({Copy(Perm, folder), "no files found"});
return;
}
int64_t thread_count = MainWorkQueue.thread_count;
int64_t files_per_thread = srt_files.len / thread_count;
int64_t remainder = srt_files.len % thread_count;
int64_t files_per_thread = files_to_parse.len / thread_count;
int64_t remainder = files_to_parse.len % thread_count;
int64_t fi = 0;
Array<ParseThreadIO> io = {Perm};
io.reserve(thread_count);
for (int ti = 0; ti < thread_count; ti += 1) {
Array<String> files = {Perm};
for (int i = 0; fi < srt_files.len && i < files_per_thread + remainder; fi += 1, i += 1) {
files.add(srt_files[fi]);
for (int i = 0; fi < files_to_parse.len && i < files_per_thread + remainder; fi += 1, i += 1) {
files.add(files_to_parse[fi]);
}
if (remainder) remainder = 0;
@@ -152,11 +142,11 @@ void XAddFolder(String folder, Array<String> *filenames) {
}
}
XToTimeString *XFindItem(String string) {
XToTimeString *result = NULL;
XToSource *XFindSource(String string) {
XToSource *result = NULL;
XArenaAddMutex.lock();
For(XToTime) {
For(XToSourceArray) {
uintptr_t begin = (uintptr_t)(it.string.data);
uintptr_t end = (uintptr_t)(it.string.data + it.string.len);
uintptr_t needle = (uintptr_t)string.data;