Use wasmfs memfs backend instead for speedup and 1 less extra thread. Reorder function argument

This commit is contained in:
msqr1
2024-03-20 23:30:30 -07:00
parent 37e526cbc1
commit 638de69e67
12 changed files with 417 additions and 1108 deletions

View File

@@ -12,10 +12,6 @@
- All related files (pthread worker, worklet processor,...) are merged - All related files (pthread worker, worklet processor,...) are merged
- Shorter from-scratch build time - Shorter from-scratch build time
# Disclaimer:
- Will log empty statement into JS console, see *global.cc* line 34
- Model loading can be slow due due to slow OPFS reading speed
# Basic usage # Basic usage
- Use the prebuilt binary through JsDelivr - Use the prebuilt binary through JsDelivr

7
devel/Vosklet.d.ts vendored
View File

@@ -13,8 +13,7 @@ interface WasmModule {
} }
export interface genericModel { export interface genericModel {
check(): void; extractAndLoad(_0: number, _1: number): void;
afterFetch(): void;
delete(): void; delete(): void;
} }
@@ -30,8 +29,8 @@ export interface recognizer {
} }
interface EmbindModule { interface EmbindModule {
genericModel: {new(_0: ArrayBuffer|Uint8Array|Uint8ClampedArray|Int8Array|string, _1: ArrayBuffer|Uint8Array|Uint8ClampedArray|Int8Array|string, _2: number, _3: boolean): genericModel}; genericModel: {new(_0: number, _1: boolean, _2: ArrayBuffer|Uint8Array|Uint8ClampedArray|Int8Array|string, _3: ArrayBuffer|Uint8Array|Uint8ClampedArray|Int8Array|string): genericModel};
recognizer: {new(_0: genericModel, _1: number, _2: number): recognizer; new(_0: genericModel, _1: genericModel, _2: number, _3: number): recognizer; new(_0: genericModel, _1: ArrayBuffer|Uint8Array|Uint8ClampedArray|Int8Array|string, _2: number, _3: number, _4: number): recognizer}; recognizer: {new(_0: number, _1: number, _2: genericModel): recognizer; new(_0: number, _1: number, _2: genericModel, _3: genericModel): recognizer; new(_0: number, _1: number, _2: genericModel, _3: ArrayBuffer|Uint8Array|Uint8ClampedArray|Int8Array|string, _4: number): recognizer};
setLogLevel(_0: number): void; setLogLevel(_0: number): void;
} }
export type MainModule = WasmModule & EmbindModule; export type MainModule = WasmModule & EmbindModule;

File diff suppressed because one or more lines are too long

View File

@@ -31,9 +31,9 @@ cd $SRC &&
MODE=0 && # 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" && echo "Mode = $MODE" &&
if [ $MODE = 0 ]; then if [ $MODE = 0 ]; then
em++ -O0 JSBridge.cc genericModel.cc recognizer.cc bindings.cc -sWASMFS -sWASM_BIGINT -sSINGLE_FILE -sGZIP_EMBEDDINGS -sMODULARIZE -sEMBIND_STD_STRING_IS_UTF8 -sPTHREAD_POOL_DELAY_LOAD -sRUNTIME_DEBUG -sSTACK_OVERFLOW_CHECK=2 -sTEXTDECODER=2 -sPTHREAD_POOL_SIZE_STRICT=2 -sASSERTIONS=2 -sINITIAL_MEMORY=$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 em++ -O0 proxier.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 -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 elif [ $MODE = 1 ]; then
em++ -O3 JSBridge.cc genericModel.cc recognizer.cc bindings.cc -sWASMFS -sWASM_BIGINT -sSINGLE_FILE -sGZIP_EMBEDDINGS -sMODULARIZE -sEMBIND_STD_STRING_IS_UTF8 -sPTHREAD_POOL_DELAY_LOAD -sTEXTDECODER=2 -sPTHREAD_POOL_SIZE_STRICT=2 -sINITIAL_MEMORY=$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 em++ -O3 proxier.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$VOSK/src -l:vosk.a -lembind -pthread -flto -msimd128 --pre-js pre.js -o ../devel/Vosklet.js
else else
em++ -O0 JSBridge.cc genericModel.cc recognizer.cc bindings.cc -sWASMFS -sWASM_BIGINT -sSINGLE_FILE -sGZIP_EMBEDDINGS -sMODULARIZE -sEMBIND_STD_STRING_IS_UTF8 -sPTHREAD_POOL_DELAY_LOAD -sTEXTDECODER=2 -sPTHREAD_POOL_SIZE_STRICT=2 -sERROR_ON_UNDEFINED_SYMBOLS=0 -sINITIAL_MEMORY=$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 em++ -O0 proxier.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 -lembind -pthread -flto -msimd128 --pre-js pre.js -o ../devel/Vosklet.js
fi fi

View File

@@ -7,14 +7,13 @@ using namespace emscripten;
EMSCRIPTEN_BINDINGS() { EMSCRIPTEN_BINDINGS() {
function("setLogLevel", &vosk_set_log_level, allow_raw_pointers()); function("setLogLevel", &vosk_set_log_level, allow_raw_pointers());
class_<genericModel>("genericModel") class_<genericModel>("genericModel")
.constructor<std::string, std::string, int, bool>(allow_raw_pointers()) .constructor<int, bool, std::string, std::string>(allow_raw_pointers())
.function("check", &genericModel::check, allow_raw_pointers()) .function("extractAndLoad", &genericModel::extractAndLoad, allow_raw_pointers());
.function("afterFetch", &genericModel::afterFetch, allow_raw_pointers());
class_<recognizer>("recognizer") class_<recognizer>("recognizer")
.constructor<genericModel*, float, int>(allow_raw_pointers()) .constructor<int, float, genericModel*>(allow_raw_pointers())
.constructor<genericModel*, genericModel*, float, int>(allow_raw_pointers()) .constructor<int, float, genericModel*, genericModel*>(allow_raw_pointers())
.constructor<genericModel*, std::string, float, int, int>(allow_raw_pointers()) .constructor<int, float, genericModel*, std::string, int>(allow_raw_pointers())
.function("setWords", &recognizer::setWords, allow_raw_pointers()) .function("setWords", &recognizer::setWords, allow_raw_pointers())
.function("setPartialWords", &recognizer::setPartialWords, allow_raw_pointers()) .function("setPartialWords", &recognizer::setPartialWords, allow_raw_pointers())
.function("setGrm", &recognizer::setGrm, allow_raw_pointers()) .function("setGrm", &recognizer::setGrm, allow_raw_pointers())

View File

@@ -1,156 +1,57 @@
#include "genericModel.h" #include "genericModel.h"
genericModel::genericModel(std::string storepath, std::string id, int index, bool normalMdl) : index(index), normalMdl(normalMdl), storepath("/opfs/" + std::move(storepath)), id(std::move(id)), entry(archive_entry_new()) {} genericModel::genericModel(int index, bool normalMdl, std::string storepath, std::string id) : index(index), normalMdl(normalMdl), storepath(std::move(storepath)), id(std::move(id)), entry(archive_entry_new()) {}
void genericModel::load() { void genericModel::extractAndLoad(int tarStart, int tarSize) {
thrd.addTask([this](){
if(normalMdl) mdl = vosk_model_new(storepath.c_str());
else vosk_spk_model_new(storepath.c_str());
emscripten_console_log("Loading finished!");
if(normalMdl ? std::get<0>(mdl) == nullptr : std::get<1>(mdl) == nullptr) fireEv(index, "Unable to load model for recognition");
else fireEv(index, "0");
});
}
bool genericModel::checkFiles() {
if(normalMdl) {
return fs::exists(storepath + "/am/final.mdl") &&
fs::exists(storepath + "/conf/mfcc.conf") &&
fs::exists(storepath + "/conf/model.conf") &&
fs::exists(storepath + "/graph/phones/word_boundary.int") &&
fs::exists(storepath + "/graph/Gr.fst") &&
fs::exists(storepath + "/graph/HCLr.fst") &&
fs::exists(storepath + "/graph/disambig_tid.int") &&
fs::exists(storepath + "/ivector/final.dubm") &&
fs::exists(storepath + "/ivector/final.ie") &&
fs::exists(storepath + "/ivector/final.mat") &&
fs::exists(storepath + "/ivector/global_cmvn.stats") &&
fs::exists(storepath + "/ivector/online_cmvn.conf") &&
fs::exists(storepath + "/ivector/splice.conf");
}
return fs::exists(storepath + "/mfcc.conf") &&
fs::exists(storepath + "/final.ext.raw") &&
fs::exists(storepath + "/mean.vec") &&
fs::exists(storepath + "/transform.mat");
}
genericModel::~genericModel() {
archive_entry_free(entry);
if(normalMdl) vosk_model_free(std::get<0>(mdl));
else vosk_spk_model_free(std::get<1>(mdl));
}
void genericModel::check() {
thrd.addTask([this](){
if(OPFSTried && !OPFSOk) {
emscripten_console_log("OPFS isn't available");
fireEv(index, "OPFS isn't available");
return;
}
if(!OPFSTried){
emscripten_console_log("Trying to acquire OPFS");
OPFSTried = true;
OPFSOk = wasmfs_create_directory("/opfs", 0777, wasmfs_create_opfs_backend()) == 0;
};
emscripten_console_log("Initializing OPFS");
if(!OPFSOk) {
emscripten_console_log("OPFS initialization failed");
fireEv(index, "OPFS initialization failed");
return;
}
if(!checkFiles() && !fs::exists(storepath + "/id")) {
emscripten_console_log("Model is not available or incorrect file, fetching...");
fireEv(index, "1");
return;
}
emscripten_console_log("Model is available, verifying ID");
std::ifstream idFile(storepath + "/id");
idFile.seekg(0, std::ios::end);
size_t oldSize = idFile.tellg();
std::string oldID(oldSize, ' ');
idFile.seekg(0);
idFile.read(&oldID[0], oldSize);
if(id.compare(oldID.c_str()) != 0) {
emscripten_console_log("ID doesn't match, fetching...");
fireEv(index, "1");
}
else {
emscripten_console_log("ID matches, loading...");
load();
}
emscripten_console_log("Success! Model is ready!");
});
}
void genericModel::afterFetch() {
thrd.addTask([this](){
emscripten_console_log("Trying to extract...");
if(!extract()) {
fs::remove(storepath + ".tar");
fs::remove_all(storepath);
emscripten_console_log("Unable to extract model");
fireEv(index, "Unable to extract model");
return;
}
emscripten_console_log("Extract done, checking files...");
if(!checkFiles()) {
emscripten_console_log("URL points to invalid model files");
fireEv(index, "URL points to invalid model files");
return;
}
emscripten_console_log("Creating ID file...");
int idFd {creat((storepath + "/id").c_str(),0777)};
if(idFd == -1) {
emscripten_console_log("Unable to create ID file");
fireEv(index, "Unable to create ID file");
fs::remove(storepath + ".tar");
fs::remove_all(storepath);
return;
}
emscripten_console_log("Writing ID...");
if(write(idFd,id.data(), id.size()) == -1) {
emscripten_console_log("Unable to write new ID");
fireEv(index, "Unable to write new ID");
fs::remove(storepath + ".tar");
fs::remove_all(storepath);
close(idFd);
return;
};
close(idFd);
emscripten_console_log("Success! Model is ready!");
});
load();
}
bool genericModel::extract() {
static fs::path path{}; static fs::path path{};
static int fd{}; static int fd{};
thrd.addTask([this, tarStart, tarSize](){
archive* src {archive_read_new()}; archive* src {archive_read_new()};
archive_read_support_format_tar(src); archive_read_support_format_tar(src);
archive_read_open_filename(src, (storepath + ".tar").c_str(), 10240); archive_read_open_memory(src, (void*)tarStart, tarSize);
free((void*)tarStart);
if(archive_errno(src) != 0) { if(archive_errno(src) != 0) {
emscripten_console_log(archive_error_string(src)); emscripten_console_logf("Unable to open tar in WASM memory: %s", archive_error_string(src));
return false; fireEv(index, "Unable to open tar in WASM memory");
return;
} }
while(archive_read_next_header2(src, entry) == ARCHIVE_OK) { while(1) {
if(archive_errno(src) != 0) { if(archive_read_next_header2(src, entry) != ARCHIVE_OK) {
emscripten_console_log(archive_error_string(src)); emscripten_console_logf("Tar header read failed: %s", archive_error_string(src));
return false; fireEv(index, "Tar header read failed");
return;
} }
path = archive_entry_pathname(entry); path = archive_entry_pathname(entry);
path = storepath + path.generic_string().substr(path.generic_string().find("/")); path = storepath + path.generic_string().substr(path.generic_string().find("/"));
emscripten_console_logf("Creating file/dir %s", path.c_str()); emscripten_console_logf("Creating file or directory %s", path.c_str());
if(!path.has_extension()) { if(!path.has_extension()) {
fs::create_directory(path); fs::create_directory(path);
continue; continue;
} }
fd = creat(path.c_str(),0777); fd = creat(path.c_str(),0777);
if(fd == -1) { if(fd == -1) {
return false; emscripten_console_log("Unable to create model files");
fireEv(index, "Unable to create model files");
return;
} }
archive_read_data_into_fd(src, fd); archive_read_data_into_fd(src, fd);
close(fd); close(fd);
if(archive_errno(src) != 0) { if(archive_errno(src) != 0) {
emscripten_console_log(archive_error_string(src)); emscripten_console_logf("Cannot write into model files: %s", archive_error_string(src));
return false; fireEv(index, "Cannot write into model files");
return;
} }
} }
fs::remove(storepath + "/README"); fs::remove(storepath + "/README");
fs::remove(storepath + ".tar" );
archive_read_free(src); archive_read_free(src);
return true; /*if(normalMdl) mdl = vosk_model_new(storepath.c_str());
else vosk_spk_model_new(storepath.c_str());
emscripten_console_log("Loading finished!");
if(normalMdl ? std::get<0>(mdl) == nullptr : std::get<1>(mdl) == nullptr) fireEv(index, "Unable to load model for recognition");
else fireEv(index, "0");*/
});
}
genericModel::~genericModel() {
archive_entry_free(entry);
if(normalMdl) vosk_model_free(std::get<0>(mdl));
else vosk_spk_model_free(std::get<1>(mdl));
} }

View File

@@ -1,7 +1,6 @@
#pragma once #pragma once
#include "jsBridge.h" #include "proxier.h"
#include <string>
#include <filesystem> #include <filesystem>
#include <fstream> #include <fstream>
#include <fcntl.h> #include <fcntl.h>
@@ -20,12 +19,8 @@ struct genericModel {
std::variant<VoskModel*, VoskSpkModel*> mdl; std::variant<VoskModel*, VoskSpkModel*> mdl;
reusableThrd thrd; reusableThrd thrd;
archive_entry* entry; archive_entry* entry;
bool extract(); void extractAndLoad(int tarStart, int tarSize);
void load(); genericModel(int index, bool normalMdl, std::string storepath, std::string id);
void check();
bool checkFiles();
void afterFetch();
genericModel(std::string storepath, std::string id, int index, bool normalMdl);
~genericModel(); ~genericModel();
}; };

View File

@@ -8,6 +8,14 @@ Module.locateFile = (path, scriptDir) => {
if(path === "Vosklet.js") return pthreadUrl if(path === "Vosklet.js") return pthreadUrl
return scriptDir+path return scriptDir+path
} }
async function getFileHandle(path, create = false) {
let components = path.split("/")
let prevDir = await navigator.storage.getDirectory()
for(let component of components.slice(0, -1)) {
prevDir = await prevDir.getDirectoryHandle(component, { create : create })
}
return prevDir.getFileHandle(components[components.length - 1], { create : create })
}
class genericModel extends EventTarget { class genericModel extends EventTarget {
constructor(url, storepath, id, normalMdl) { constructor(url, storepath, id, normalMdl) {
super() super()
@@ -18,39 +26,46 @@ class genericModel extends EventTarget {
this.normalMdl = normalMdl this.normalMdl = normalMdl
} }
static async _init(url, storepath, id, normalMdl) { static async _init(url, storepath, id, normalMdl) {
return new Promise((resolve, reject) => {
let mdl = new genericModel(url, storepath, id, normalMdl) let mdl = new genericModel(url, storepath, id, normalMdl)
mdl.addEventListener("0", async function listener(ev) { let tar
mdl.addEventListener("0", async (ev) => {
switch(ev.detail) { switch(ev.detail) {
case "0": case "0":
mdl.removeEventListener("0", listener) return mdl
return resolve(mdl)
case "1":
let res = await fetch(url)
if(!res.ok) {
return reject("Unable to download model")
}
let wStream = await (await (await navigator.storage.getDirectory()).getFileHandle(storepath + ".tar", {create : true})).createWritable()
let tarReader = res.body.pipeThrough(new DecompressionStream("gzip")).getReader()
while(true) {
let readRes = await tarReader.read()
if(!readRes.done) await wStream.write(readRes.value)
else break
}
tarReader.releaseLock()
await wStream.close()
console.log("Fetching and writing tar done!")
mdl.obj.afterFetch()
break
default: default:
mdl.delete() mdl.delete()
mdl.removeEventListener("0", listener) throw ev.detail
reject(ev.detail)
} }
}) }, { once : true })
mdl.obj = new Module.genericModel(storepath, id, objs.length-1, normalMdl) mdl.obj = new Module.genericModel(objs.length - 1, normalMdl, "/" + storepath, id)
mdl.obj.check() try {
}) let dataFile = await (await getFileHandle(storepath + "/model.tgz")).getFile()
let idFile = await (await getFileHandle(storepath + "/id")).getFile()
if(await idFile.text() !== id) throw ""
tar = await new Response(dataFile.stream().pipeThrough(new DecompressionStream("gzip"))).arrayBuffer()
}
catch {
try {
let res = await fetch(url)
if(!res.ok) {
throw "Unable to download model"
}
let newDataFile = await (await getFileHandle(storepath + "/model.tgz", true)).createWritable()
tar = await new Response(res.body.pipeThrough(new DecompressionStream("gzip"))).arrayBuffer()
await newDataFile.write(tar)
await newDataFile.close()
let newIDFile = await (await getFileHandle(storepath + "/id", true)).createWritable()
await newIDFile.write(id)
await newIDFile.close()
}
catch(e) {
mdl.obj.delete()
throw e
}
}
let tarStart = Module._malloc(tar.byteLength)
Module.HEAPU8.set(new Uint8Array(tar), tarStart)
mdl.obj.extractAndLoad(tarStart, tar.byteLength)
} }
delete() { delete() {
if (this.obj) this.obj.delete() if (this.obj) this.obj.delete()
@@ -68,15 +83,14 @@ class Recognizer extends EventTarget {
objs.push(this) objs.push(this)
} }
static async _init(model, sampleRate, mode, grammar, spkModel) { static async _init(model, sampleRate, mode, grammar, spkModel) {
return new Promise((resolve, reject) => {
let rec = new Recognizer() let rec = new Recognizer()
rec.addEventListener("0", (ev) => { rec.addEventListener("0", (ev) => {
if(ev.detail === "0") { if(ev.detail === "0") {
rec.ptr = Module._malloc(512) rec.ptr = Module._malloc(512)
return resolve(rec) return rec
} }
rec.delete() rec.delete()
reject(ev.detail) throw ev.detail
}, { once : true }) }, { once : true })
switch(mode) { switch(mode) {
case 1: case 1:
@@ -88,7 +102,6 @@ class Recognizer extends EventTarget {
default: default:
rec.obj = new Module.recognizer(model, grammar, sampleRate, objs.length-1, 0) rec.obj = new Module.recognizer(model, grammar, sampleRate, objs.length-1, 0)
} }
})
} }
async getNode(ctx, channelIndex = 0) { async getNode(ctx, channelIndex = 0) {
if(typeof this.node === "undefined") { if(typeof this.node === "undefined") {

View File

@@ -1,8 +1,6 @@
#include "jsBridge.h" #include "proxier.h"
pthread_t dstThrd{pthread_self()}; pthread_t dstThrd{pthread_self()};
bool OPFSOk{};
bool OPFSTried{};
ProxyingQueue glbQ{}; ProxyingQueue glbQ{};
void fireEv(int index, const char* content, const char* type) { void fireEv(int index, const char* content, const char* type) {
auto proxy{[index, content, type](){ auto proxy{[index, content, type](){

View File

@@ -2,7 +2,6 @@
#include <thread> #include <thread>
#include <mutex> #include <mutex>
#include <queue> #include <queue>
#include <condition_variable>
#include <emscripten/wasmfs.h> #include <emscripten/wasmfs.h>
#include <emscripten/console.h> #include <emscripten/console.h>
@@ -10,9 +9,6 @@
#include <emscripten/proxying.h> #include <emscripten/proxying.h>
using namespace emscripten; using namespace emscripten;
extern bool OPFSOk;
extern bool OPFSTried;
extern std::error_code tank;
extern pthread_t dstThrd; extern pthread_t dstThrd;
extern ProxyingQueue glbQ; extern ProxyingQueue glbQ;
@@ -20,7 +16,6 @@ void fireEv(int index, const char* content, const char* type = nullptr);
struct reusableThrd { struct reusableThrd {
std::queue<std::function<void()>> queue{}; std::queue<std::function<void()>> queue{};
bool done{}; bool done{};
void addTask(std::function<void()>&& task); void addTask(std::function<void()>&& task);
reusableThrd(); reusableThrd();
~reusableThrd(); ~reusableThrd();

View File

@@ -1,16 +1,16 @@
#include "recognizer.h" #include "recognizer.h"
recognizer::recognizer(genericModel* model, float sampleRate, int index) : index(index) { recognizer::recognizer(int index, float sampleRate, genericModel* model) : index(index) {
emscripten_console_log("Recognizer constructor called..."); emscripten_console_log("Recognizer constructor called...");
rec = vosk_recognizer_new(std::get<0>(model->mdl),sampleRate); rec = vosk_recognizer_new(std::get<0>(model->mdl),sampleRate);
finishConstruction(model, nullptr); finishConstruction(model, nullptr);
} }
recognizer::recognizer(genericModel* model, genericModel* spkMdl, float sampleRate, int index) : index(index) { recognizer::recognizer(int index, float sampleRate, genericModel* model, genericModel* spkMdl) : index(index) {
emscripten_console_log("Recognizer constructor called..."); emscripten_console_log("Recognizer constructor called...");
rec = vosk_recognizer_new_spk(std::get<0>(model->mdl), sampleRate, std::get<1>(spkMdl->mdl)); rec = vosk_recognizer_new_spk(std::get<0>(model->mdl), sampleRate, std::get<1>(spkMdl->mdl));
finishConstruction(model, spkMdl); finishConstruction(model, spkMdl);
} }
recognizer::recognizer(genericModel* model, const std::string& grm, float sampleRate, int index, int dummy) : index(index) { recognizer::recognizer(int index, float sampleRate, genericModel* model, const std::string& grm, int dummy) : index(index) {
emscripten_console_log("Recognizer constructor called..."); emscripten_console_log("Recognizer constructor called...");
rec = vosk_recognizer_new_grm(std::get<0>(model->mdl), sampleRate, grm.c_str()); rec = vosk_recognizer_new_grm(std::get<0>(model->mdl), sampleRate, grm.c_str());
finishConstruction(model, nullptr); finishConstruction(model, nullptr);

View File

@@ -1,6 +1,5 @@
#pragma once #pragma once
#include "genericModel.h" #include "genericModel.h"
#include "jsBridge.h"
struct recognizer { struct recognizer {
std::atomic_flag done{}; std::atomic_flag done{};
@@ -8,9 +7,9 @@ struct recognizer {
float* dataPtr; float* dataPtr;
int index; int index;
VoskRecognizer* rec; VoskRecognizer* rec;
recognizer(genericModel* model, float sampleRate, int index); recognizer(int index, float sampleRate, genericModel* model);
recognizer(genericModel* model, genericModel* spkModel, float sampleRate, int index); recognizer(int index, float sampleRate, genericModel* model, genericModel* spkModel);
recognizer(genericModel* model, const std::string& grm, float sampleRate, int index, int dummy); recognizer(int index, float sampleRate, genericModel* model, const std::string& grm, int dummy);
~recognizer(); ~recognizer();
void finishConstruction(genericModel* model, genericModel* spkModel); void finishConstruction(genericModel* model, genericModel* spkModel);
void acceptWaveForm(); void acceptWaveForm();