Rename, ready for a super big change about OPFS

This commit is contained in:
msqr1
2024-03-17 21:41:49 -07:00
parent ecab9aa5d5
commit 37e526cbc1
9 changed files with 210 additions and 82 deletions

File diff suppressed because one or more lines are too long

View File

@@ -28,12 +28,12 @@ fi
. $EMSDK/emsdk_env.sh &&
cd $SRC &&
MODE=2 && # 0: Ultra debug info, 1: Extremely optimized release, else custom
MODE=0 && # 0: Ultra debug info, 1: Extremely optimized release, else custom
echo "Mode = $MODE" &&
if [ $MODE = 0 ]; then
em++ -O0 global.cc genericModel.cc recognizer.cc bindings.cc -sWASMFS -sWASM_BIGINT -sSINGLE_FILE -sGZIP_EMBEDDINGS -sMODULARIZE -sEMBIND_STD_STRING_IS_UTF8 -sPTHREAD_POOL_DELAY_LOAD -sRUNTIME_DEBUG -sSTACK_OVERFLOW_CHECK=2 -sTEXTDECODER=2 -sPTHREAD_POOL_SIZE_STRICT=2 -sASSERTIONS=2 -sINITIAL_MEMORY=$MAX_MEMORY -sPTHREAD_POOL_SIZE=$MAX_THREADS -sDISABLE_EXCEPTION_CATCHING=0 -sEXIT_RUNTIME=0 -sINVOKE_RUN=0 -sPOLYFILL=0 -sEXPORTED_FUNCTIONS=_malloc -sEXPORT_NAME=loadVosklet -sMALLOC=emmalloc -sEXPORTED_RUNTIME_METHODS=UTF8ToString,stringToUTF8OnStack -sENVIRONMENT=web,worker -I. -I$LIBARCHIVE/include -I$VOSK/src -L$LIBARCHIVE/lib -larchive -L$KALDI/src -l:online2/kaldi-online2.a -l:decoder/kaldi-decoder.a -l:ivector/kaldi-ivector.a -l:gmm/kaldi-gmm.a -l:tree/kaldi-tree.a -l:feat/kaldi-feat.a -l:cudamatrix/kaldi-cudamatrix.a -l:lat/kaldi-lat.a -l:lm/kaldi-lm.a -l:rnnlm/kaldi-rnnlm.a -l:hmm/kaldi-hmm.a -l:nnet3/kaldi-nnet3.a -l:transform/kaldi-transform.a -l:matrix/kaldi-matrix.a -l:fstext/kaldi-fstext.a -l:util/kaldi-util.a -l:base/kaldi-base.a -L$OPENFST/lib -l:libfst.a -l:libfstngram.a -L$CLAPACK_WASM -l:CBLAS/lib/cblas.a -l:CLAPACK-3.2.1/lapack.a -l:CLAPACK-3.2.1/libcblaswr.a -l:f2c_BLAS-3.8.0/blas.a -l:libf2c/libf2c.a -L$VOSK/src -l:vosk.a -lopfs.js -lembind -pthread -flto -msimd128 --embind-emit-tsd Vosklet.d.ts -fsanitize=undefined -fsanitize=address -fsanitize=leak -g3 --pre-js pre.js -o ../devel/Vosklet.js
em++ -O0 JSBridge.cc genericModel.cc recognizer.cc bindings.cc -sWASMFS -sWASM_BIGINT -sSINGLE_FILE -sGZIP_EMBEDDINGS -sMODULARIZE -sEMBIND_STD_STRING_IS_UTF8 -sPTHREAD_POOL_DELAY_LOAD -sRUNTIME_DEBUG -sSTACK_OVERFLOW_CHECK=2 -sTEXTDECODER=2 -sPTHREAD_POOL_SIZE_STRICT=2 -sASSERTIONS=2 -sINITIAL_MEMORY=$MAX_MEMORY -sPTHREAD_POOL_SIZE=$MAX_THREADS -sDISABLE_EXCEPTION_CATCHING=0 -sEXIT_RUNTIME=0 -sINVOKE_RUN=0 -sPOLYFILL=0 -sEXPORTED_FUNCTIONS=_malloc -sEXPORT_NAME=loadVosklet -sMALLOC=emmalloc -sEXPORTED_RUNTIME_METHODS=UTF8ToString,stringToUTF8OnStack -sENVIRONMENT=web,worker -I. -I$LIBARCHIVE/include -I$VOSK/src -L$LIBARCHIVE/lib -larchive -L$KALDI/src -l:online2/kaldi-online2.a -l:decoder/kaldi-decoder.a -l:ivector/kaldi-ivector.a -l:gmm/kaldi-gmm.a -l:tree/kaldi-tree.a -l:feat/kaldi-feat.a -l:cudamatrix/kaldi-cudamatrix.a -l:lat/kaldi-lat.a -l:lm/kaldi-lm.a -l:rnnlm/kaldi-rnnlm.a -l:hmm/kaldi-hmm.a -l:nnet3/kaldi-nnet3.a -l:transform/kaldi-transform.a -l:matrix/kaldi-matrix.a -l:fstext/kaldi-fstext.a -l:util/kaldi-util.a -l:base/kaldi-base.a -L$OPENFST/lib -l:libfst.a -l:libfstngram.a -L$CLAPACK_WASM -l:CBLAS/lib/cblas.a -l:CLAPACK-3.2.1/lapack.a -l:CLAPACK-3.2.1/libcblaswr.a -l:f2c_BLAS-3.8.0/blas.a -l:libf2c/libf2c.a -L$VOSK/src -l:vosk.a -lopfs.js -lembind -pthread -flto -msimd128 --embind-emit-tsd Vosklet.d.ts -fsanitize=undefined -fsanitize=address -fsanitize=leak -g3 --pre-js pre.js -o ../devel/Vosklet.js
elif [ $MODE = 1 ]; then
em++ -O3 global.cc genericModel.cc recognizer.cc bindings.cc -sWASMFS -sWASM_BIGINT -sSINGLE_FILE -sGZIP_EMBEDDINGS -sMODULARIZE -sEMBIND_STD_STRING_IS_UTF8 -sPTHREAD_POOL_DELAY_LOAD -sTEXTDECODER=2 -sPTHREAD_POOL_SIZE_STRICT=2 -sINITIAL_MEMORY=$MAX_MEMORY-sPTHREAD_POOL_SIZE=$MAX_THREADS -sPOLYFILL=0 -sEXIT_RUNTIME=0 -sINVOKE_RUN=0 -sSUPPORT_LONGJMP=0 -sEXPORTED_FUNCTIONS=_malloc -sEXPORT_NAME=loadVosklet -sMALLOC=emmalloc -sEXPORTED_RUNTIME_METHODS=UTF8ToString,stringToUTF8OnStack -sENVIRONMENT=web,worker -I. -I$LIBARCHIVE/include -I$VOSK/src -L$LIBARCHIVE/lib -larchive -L$KALDI/src -l:online2/kaldi-online2.a -l:decoder/kaldi-decoder.a -l:ivector/kaldi-ivector.a -l:gmm/kaldi-gmm.a -l:tree/kaldi-tree.a -l:feat/kaldi-feat.a -l:cudamatrix/kaldi-cudamatrix.a -l:lat/kaldi-lat.a -l:lm/kaldi-lm.a -l:rnnlm/kaldi-rnnlm.a -l:hmm/kaldi-hmm.a -l:nnet3/kaldi-nnet3.a -l:transform/kaldi-transform.a -l:matrix/kaldi-matrix.a -l:fstext/kaldi-fstext.a -l:util/kaldi-util.a -l:base/kaldi-base.a -L$OPENFST/lib -l:libfst.a -l:libfstngram.a -L$CLAPACK_WASM -l:CBLAS/lib/cblas.a -l:CLAPACK-3.2.1/lapack.a -l:CLAPACK-3.2.1/libcblaswr.a -l:f2c_BLAS-3.8.0/blas.a -l:libf2c/libf2c.a -L$VOSK/src -l:vosk.a -lopfs.js -lembind -pthread -flto -msimd128 --pre-js pre.js -o ../devel/Vosklet.js
em++ -O3 JSBridge.cc genericModel.cc recognizer.cc bindings.cc -sWASMFS -sWASM_BIGINT -sSINGLE_FILE -sGZIP_EMBEDDINGS -sMODULARIZE -sEMBIND_STD_STRING_IS_UTF8 -sPTHREAD_POOL_DELAY_LOAD -sTEXTDECODER=2 -sPTHREAD_POOL_SIZE_STRICT=2 -sINITIAL_MEMORY=$MAX_MEMORY-sPTHREAD_POOL_SIZE=$MAX_THREADS -sPOLYFILL=0 -sEXIT_RUNTIME=0 -sINVOKE_RUN=0 -sSUPPORT_LONGJMP=0 -sEXPORTED_FUNCTIONS=_malloc -sEXPORT_NAME=loadVosklet -sMALLOC=emmalloc -sEXPORTED_RUNTIME_METHODS=UTF8ToString,stringToUTF8OnStack -sENVIRONMENT=web,worker -I. -I$LIBARCHIVE/include -I$VOSK/src -L$LIBARCHIVE/lib -larchive -L$KALDI/src -l:online2/kaldi-online2.a -l:decoder/kaldi-decoder.a -l:ivector/kaldi-ivector.a -l:gmm/kaldi-gmm.a -l:tree/kaldi-tree.a -l:feat/kaldi-feat.a -l:cudamatrix/kaldi-cudamatrix.a -l:lat/kaldi-lat.a -l:lm/kaldi-lm.a -l:rnnlm/kaldi-rnnlm.a -l:hmm/kaldi-hmm.a -l:nnet3/kaldi-nnet3.a -l:transform/kaldi-transform.a -l:matrix/kaldi-matrix.a -l:fstext/kaldi-fstext.a -l:util/kaldi-util.a -l:base/kaldi-base.a -L$OPENFST/lib -l:libfst.a -l:libfstngram.a -L$CLAPACK_WASM -l:CBLAS/lib/cblas.a -l:CLAPACK-3.2.1/lapack.a -l:CLAPACK-3.2.1/libcblaswr.a -l:f2c_BLAS-3.8.0/blas.a -l:libf2c/libf2c.a -L$VOSK/src -l:vosk.a -lopfs.js -lembind -pthread -flto -msimd128 --pre-js pre.js -o ../devel/Vosklet.js
else
em++ -O0 global.cc genericModel.cc recognizer.cc bindings.cc -sWASMFS -sWASM_BIGINT -sSINGLE_FILE -sGZIP_EMBEDDINGS -sMODULARIZE -sEMBIND_STD_STRING_IS_UTF8 -sPTHREAD_POOL_DELAY_LOAD -sRUNTIME_DEBUG -sSTACK_OVERFLOW_CHECK=2 -sTEXTDECODER=2 -sPTHREAD_POOL_SIZE_STRICT=2 -sASSERTIONS=2 -sINITIAL_MEMORY=$MAX_MEMORY -sPTHREAD_POOL_SIZE=$MAX_THREADS -sDISABLE_EXCEPTION_CATCHING=0 -sEXIT_RUNTIME=0 -sINVOKE_RUN=0 -sPOLYFILL=0 -sEXPORTED_FUNCTIONS=_malloc -sEXPORT_NAME=loadVosklet -sMALLOC=emmalloc -sEXPORTED_RUNTIME_METHODS=UTF8ToString,stringToUTF8OnStack -sENVIRONMENT=web,worker -I. -I$LIBARCHIVE/include -I$VOSK/src -L$LIBARCHIVE/lib -larchive -L$KALDI/src -l:online2/kaldi-online2.a -l:decoder/kaldi-decoder.a -l:ivector/kaldi-ivector.a -l:gmm/kaldi-gmm.a -l:tree/kaldi-tree.a -l:feat/kaldi-feat.a -l:cudamatrix/kaldi-cudamatrix.a -l:lat/kaldi-lat.a -l:lm/kaldi-lm.a -l:rnnlm/kaldi-rnnlm.a -l:hmm/kaldi-hmm.a -l:nnet3/kaldi-nnet3.a -l:transform/kaldi-transform.a -l:matrix/kaldi-matrix.a -l:fstext/kaldi-fstext.a -l:util/kaldi-util.a -l:base/kaldi-base.a -L$OPENFST/lib -l:libfst.a -l:libfstngram.a -L$CLAPACK_WASM -l:CBLAS/lib/cblas.a -l:CLAPACK-3.2.1/lapack.a -l:CLAPACK-3.2.1/libcblaswr.a -l:f2c_BLAS-3.8.0/blas.a -l:libf2c/libf2c.a -L$VOSK/src -l:vosk.a -lopfs.js -lembind -pthread -flto -msimd128 --embind-emit-tsd Vosklet.d.ts -fsanitize=undefined -fsanitize=address -fsanitize=leak -g3 --pre-js pre.js -o ../devel/Vosklet.js
em++ -O0 JSBridge.cc genericModel.cc recognizer.cc bindings.cc -sWASMFS -sWASM_BIGINT -sSINGLE_FILE -sGZIP_EMBEDDINGS -sMODULARIZE -sEMBIND_STD_STRING_IS_UTF8 -sPTHREAD_POOL_DELAY_LOAD -sTEXTDECODER=2 -sPTHREAD_POOL_SIZE_STRICT=2 -sERROR_ON_UNDEFINED_SYMBOLS=0 -sINITIAL_MEMORY=$MAX_MEMORY -sPTHREAD_POOL_SIZE=$MAX_THREADS -sPOLYFILL=0 -sEXIT_RUNTIME=0 -sINVOKE_RUN=0 -sSUPPORT_LONGJMP=0 -sEXPORTED_FUNCTIONS=_malloc -sEXPORT_NAME=loadVosklet -sMALLOC=emmalloc -sEXPORTED_RUNTIME_METHODS=UTF8ToString,stringToUTF8OnStack -sENVIRONMENT=web,worker -I. -I$LIBARCHIVE/include -I$VOSK/src -L$LIBARCHIVE/lib -larchive -L$KALDI/src -l:online2/kaldi-online2.a -l:decoder/kaldi-decoder.a -l:ivector/kaldi-ivector.a -l:gmm/kaldi-gmm.a -l:tree/kaldi-tree.a -l:feat/kaldi-feat.a -l:cudamatrix/kaldi-cudamatrix.a -l:lat/kaldi-lat.a -l:lm/kaldi-lm.a -l:rnnlm/kaldi-rnnlm.a -l:hmm/kaldi-hmm.a -l:nnet3/kaldi-nnet3.a -l:transform/kaldi-transform.a -l:matrix/kaldi-matrix.a -l:fstext/kaldi-fstext.a -l:util/kaldi-util.a -l:base/kaldi-base.a -L$OPENFST/lib -l:libfst.a -l:libfstngram.a -L$CLAPACK_WASM -l:CBLAS/lib/cblas.a -l:CLAPACK-3.2.1/lapack.a -l:CLAPACK-3.2.1/libcblaswr.a -l:f2c_BLAS-3.8.0/blas.a -l:libf2c/libf2c.a -L$VOSK/src -l:vosk.a -lopfs.js -lembind -pthread -flto -msimd128 --pre-js pre.js -o ../devel/Vosklet.js
fi

View File

@@ -1,4 +1,4 @@
#include "global.h"
#include "jsBridge.h"
pthread_t dstThrd{pthread_self()};
bool OPFSOk{};
@@ -17,9 +17,7 @@ void fireEv(int index, const char* content, const char* type) {
}
reusableThrd::reusableThrd() {
std::thread thrd{[this](){
while(!done.test(std::memory_order_relaxed)) {
blocker.wait(queue.empty() || done.test(std::memory_order_relaxed), std::memory_order_relaxed);
blocker.clear(std::memory_order_relaxed);
while(!done) {
while(!queue.empty()) {
emscripten_console_log("==========Executing task===========");
queue.front()();
@@ -31,12 +29,7 @@ reusableThrd::reusableThrd() {
}
void reusableThrd::addTask(std::function<void()>&& task) {
queue.emplace(task);
// The following line will make -O3 magically work. Yes, I know, it is a god dang PRINT STATEMENT.
//emscripten_console_log("");
blocker.test_and_set(std::memory_order_relaxed);
blocker.notify_one();
}
reusableThrd::~reusableThrd() {
done.test_and_set(std::memory_order_relaxed);
done.notify_one();
done = true;
}

View File

@@ -1,5 +1,5 @@
#pragma once
#include "global.h"
#include "jsBridge.h"
#include <string>
#include <filesystem>
@@ -11,7 +11,6 @@
#include <archive.h>
#include <archive_entry.h>
namespace fs = std::filesystem;
struct genericModel {
bool normalMdl;
bool recognizerUsedThrd{};

View File

@@ -1,7 +1,8 @@
#pragma once
#include <thread>
#include <atomic>
#include <mutex>
#include <queue>
#include <condition_variable>
#include <emscripten/wasmfs.h>
#include <emscripten/console.h>
@@ -18,8 +19,8 @@ extern ProxyingQueue glbQ;
void fireEv(int index, const char* content, const char* type = nullptr);
struct reusableThrd {
std::queue<std::function<void()>> queue{};
std::atomic_flag blocker{};
std::atomic_flag done{};
bool done{};
void addTask(std::function<void()>&& task);
reusableThrd();
~reusableThrd();

View File

@@ -1,6 +1,6 @@
#pragma once
#include "genericModel.h"
#include "global.h"
#include "jsBridge.h"
struct recognizer {
std::atomic_flag done{};

135
src/untar.cc Normal file
View File

@@ -0,0 +1,135 @@
// Taken from https://github.com/libarchive/libarchive/blob/master/contrib/untar.c
// Stripped all the way down for smallest size
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <filesystem>
#include <emscripten/console.h>
namespace fs = std::filesystem;
#define TO_STRING(x) #x
#define STR(x) TO_STRING(x)
/* Parse an octal number, ignoring leading and trailing nonsense. */
int parseoct(const char *p, size_t n)
{
int i = 0;
while ((*p < '0' || *p > '7') && n > 0) {
++p;
--n;
}
while (*p >= '0' && *p <= '7' && n > 0) {
i *= 8;
i += *p - '0';
++p;
--n;
}
return i;
}
/* Returns true if this is 512 zero bytes. */
bool is_end_of_archive(const char *p)
{
int n;
for (n = 511; n >= 0; --n)
if (p[n] != '\0')
return false;
return true;
}
/* Create a file, including parent directory as necessary. */
FILE * create_file(char *pathname) {
FILE *f;
if (f == NULL) {
/* Try creating parent dir and then creating file. */
char *p = strrchr(pathname, '/');
if (p != NULL) {
*p = '\0';
fs::create_directories(pathname);
*p = '/';
f = fopen(pathname, "wb+");
}
}
return f;
}
/* Verify the tar checksum. */
int verify_checksum(const char *p)
{
int n, u = 0;
for (n = 0; n < 512; ++n) {
if (n < 148 || n > 155)
/* Standard tar checksum adds unsigned bytes. */
u += ((unsigned char *)p)[n];
else
u += 0x20;
}
return (u == parseoct(p + 148, 8));
}
/* Extract a tar archive. */
bool untar(FILE *a, const char *path)
{
char buff[512];
FILE *f = NULL;
size_t bytes_read;
off_t filesize;
emscripten_console_logf("Extracting from %s", path);
for (;;) {
bytes_read = fread(buff, 1, 512, a);
if (bytes_read < 512) {
emscripten_console_logf("Short read on %s", path);
return false;
}
if (is_end_of_archive(buff)) {
emscripten_console_logf("End of %s", path);
return true;
}
if (!verify_checksum(buff)) {
emscripten_console_log("Checksum failure");
return false;
}
filesize = parseoct(buff + 124, 12);
switch (buff[156]) {
// Directory
case '5':
emscripten_console_logf(" Extracting dir %s", buff);
fs::create_directories(buff);
filesize = 0;
break;
// Normal file
case '0':
case '\0':
emscripten_console_logf(" Extracting file %s", buff);
f = create_file(buff);
}
while (filesize > 0) {
bytes_read = fread(buff, 1, 512, a);
if (bytes_read < 512) {
emscripten_console_logf("Short read on %s", path);
return false;
}
if (filesize < 512)
bytes_read = filesize;
if (f != NULL) {
if (fwrite(buff, 1, bytes_read, f)
!= bytes_read)
{
emscripten_console_log("Failed write");
fclose(f);
f = NULL;
}
}
filesize -= (off_t)bytes_read;
}
if (f != NULL) {
fclose(f);
f = NULL;
}
}
return true;
}

View File

@@ -43,7 +43,7 @@ Vosklet require SharedArrayBuffer to share thread's data, so these response head
If you can't set them, you may use a HACKY workaround at *src/addCOI.js*.
## Origin Private Filesystem (OPFS)
Vosklet needs the Emscripten WASMFS' OPFS to store its model, IDBFS was considered, but dropped because there is no direct way to read from IDBFS to C++ without copying to MEMFS (RAM). For safety with this, always:
Vosklet needs the Emscripten WASMFS' OPFS to store its model, IDBFS was considered. For safety with this, always:
- Try catch ```window.loadVosklet()``` to to check for OPFS availability.
- Check if there is enough space via ```navigator.storage.estimate()``` for TWICE THE MODEL SIZE before calling ```Module.makeModel```