int inithtk(int argc, char *argv[],const char * app_version, bool noGraphics)
{
InitThreads(HT_MSGMON); // enable msg driven monitoring
if(InitShell(argc,argv,app_version)<SUCCESS) return FAIL;
// CommonInit() expanded with incompatible things commented out:
InitMem(); InitLabel();
InitMath(); InitSigP(); InitUtil();
InitWave(); /*InitAudio();*/ InitModel();
if(InitParm()<SUCCESS) return FAIL;
InitGraf((Boolean)noGraphics);
InitDict(); InitNet(); InitLM();
InitRec(); InitLat(); /*InitAdapt();*/
EnableBTrees(); /* allows unseen triphones to be synthesised */
/*
Register(abuffer_version);
Register(acode_version);
Register(acomponent_version);
Register(adict_version);
Register(agram_version);
Register(angram_version);
Register(ahmms_version);
Register(ahtk_version);
Register(amonitor_version);
Register(apacket_version);
Register(arec_version);
Register(aresource_version);
Register(arman_version);
Register(asource_version);
*/
InfoPrinted();
return SUCCESS;
}
int main(int argc, char *argv[])
{
char * labFn, *listfn, *s;
int i,fidx;
MLFEntry *me = NULL;
Transcription *t;
void InitStats(char *listfn);
void GatherStats(Transcription *t);
void OutputStats(void);
if(InitShell(argc,argv,hlstats_version,hlstats_vc_id)<SUCCESS)
HError(1300,"HLStats: InitShell failed");
InitMem(); InitMath();
InitWave(); InitLabel();
InitLM();
if (!InfoPrinted() && NumArgs() == 0)
ReportUsage();
if (NumArgs() == 0) Exit(0);
SetConfParms();
enterId=GetLabId("!ENTER",TRUE); /* All sentences should or are coerced */
exitId=GetLabId("!EXIT",TRUE); /* to start enterId and end exitId */
nullId=GetLabId("!NULL",TRUE); /* Name for words not in list */
while (NextArg() == SWITCHARG) {
s = GetSwtArg();
if (strlen(s)!=1)
HError(1319,"HLStats: Bad switch %s; must be single letter",s);
switch(s[0]){
case 'b':
doBigram = TRUE;
if (NextArg() != STRINGARG)
HError(1319,"HLStats: Ngram output file name expected");
bigFile = GetStrArg();
break;
case 'c':
doLCount = TRUE;
lCountLimit = GetChkedInt(0,100000,s);
break;
case 'd':
doDurs = TRUE; break;
case 'f':
bigFloor = GetChkedFlt(0.0,1000.0,s);
break;
case 'h':
hSize = GetChkedInt(1,2,s);
break;
case 'l':
doList = TRUE;
if (NextArg() != STRINGARG)
HError(1319,"HLStats: Output label list file name expected");
listFile = GetStrArg();
break;
case 'o':
doBOff = TRUE;
break;
case 'p':
doPCount = TRUE;
pCountLimit = GetChkedInt(0,100000,s);
break;
case 's':
if (NextArg() != STRINGARG)
HError(1319,"HLStats: ENTER label name expected");
enterId=GetLabId(GetStrArg(),TRUE);
if (NextArg() != STRINGARG)
HError(1319,"HLStats: EXIT label name expected");
exitId=GetLabId(GetStrArg(),TRUE);
break;
case 't':
bigThresh = GetChkedInt(0,100,s);
break;
case 'u':
uniFloor = GetChkedFlt(0.0,1000.0,s);
break;
case 'G':
if (NextArg() != STRINGARG)
HError(1319,"HLStats: Input label File format expected");
if((ff = Str2Format(GetStrArg())) == ALIEN)
HError(-1389,"HLStats: Warning ALIEN Label file format set");
break;
case 'I':
if (NextArg() != STRINGARG)
HError(1319,"HLStats: Input MLF file name expected");
LoadMasterFile(GetStrArg());
break;
case 'T':
if (NextArg() != INTARG)
HError(1319,"HLStats: Trace value expected");
trace = GetChkedInt(0,017,s); break;
default:
HError(1319,"HLStats: Unknown switch %s",s);
}
}
if (NextArg()!=STRINGARG)
HError(1319,"HLStats: Label list file name expected");
listfn = GetStrArg();
if (!(doDurs || doBigram || doList || doLCount || doPCount))
HError(1330,"HLStats: Nothing to do!");
InitStats(listfn);
i=0;
while (NumArgs()>0) {
if (NextArg()!=STRINGARG)
HError(1319,"HLStats: Input label file name expected");
labFn = GetStrArg();
if (IsMLFFile(labFn)) {
fidx = NumMLFFiles();
if ((me=GetMLFTable()) != NULL) {
while(me->next != NULL) me=me->next;
LoadMasterFile(labFn);
me=me->next;
}
else {
LoadMasterFile(labFn);
me=GetMLFTable();
}
while (me != NULL) {
if (me->type == MLF_IMMEDIATE && me->def.immed.fidx == fidx) {
if (trace&T_FIL) {
printf(" Processing file %s\n",me->pattern); fflush(stdout);
}
t = LOpen(&tmpHeap,me->pattern,ff);
if (t->numLists<1)
HError(-1330,"HLStats: Empty file %s",me->pattern);
else
GatherStats(t),i++;
Dispose(&tmpHeap,t);
}
me = me->next;
if ((trace&T_BAS) && !(trace&T_FIL) &&
NumMLFEntries()>5000 && i%1000==0)
printf(". "),fflush(stdout);
}
if ((trace&T_BAS) && !(trace&T_FIL) && NumMLFEntries()>5000)
printf("\n");
} else {
if (trace&T_FIL) {
printf(" Processing file %s\n",labFn); fflush(stdout);
}
t = LOpen(&tmpHeap,labFn,ff);
if (t->numLists<1)
HError(-1330,"HLStats: Empty file %s",me->pattern);
else
GatherStats(t),i++;
Dispose(&tmpHeap,t);
}
}
if (trace&T_MEM)
PrintAllHeapStats();
OutputStats();
if (trace&T_MEM)
PrintAllHeapStats();
Exit(0);
return (0); /* never reached -- make compiler happy */
}
#include "HLM.h"
typedef enum {unknown, wordLoop, boBiGram, matBiGram, multiLat, wordPair} BuildType;
static int trace = 0; /* Trace flags */
static LabId enterId; /* id of !ENTRY label in ngram */
static LabId exitId; /* id of !EXIT label in ngram */
static LabId bStartId=NULL; /* id of start bracket */
static LabId bEndId=NULL; /* id of end bracket */
static LabId unknownId; /* id of unknown label in ngram */
static Boolean zapUnknown = FALSE; /* zap unknown symbols from bigram */
MemHeap buildStack;
/* ---------------- Configuration Parameters --------------------- */
static ConfParam *cParm[MAXGLOBS];
static int nParm = 0; /* total num params */
/* ---------------- Process Command Line ------------------------- */
/* SetConfParms: set conf parms relevant to this tool */
void SetConfParms(void)
{
int i;
nParm = GetConfig("HBUILD", TRUE, cParm, MAXGLOBS);
if (nParm>0){
if (GetConfInt(cParm,nParm,"TRACE",&i)) trace = i;
}
}
void ReportUsage(void)
{
printf("\nUSAGE: HBuild [options] wordList latFile\n\n");
printf(" Option Default\n\n");
printf(" -b binary lattice output ASCII\n");
printf(" -m s load matrix bigram from s off\n");
printf(" -n s load back-off bigram from s off\n");
printf(" -s s1 s2 s1/s2 are bigram start/end labels !ENTER !EXIT\n");
printf(" -t s1 s2 bracket word-loop/pair with s1 s2 off\n");
printf(" -u s set unknown symbol to s !NULL\n");
printf(" -w s load word-pair grammar from s off\n");
printf(" -x s load multi-level lattice from s off\n");
printf(" -z ignore ngrams with unknown symbol off\n");
PrintStdOpts("");
printf("\n\n");
}
int main(int argc, char *argv[])
{
char *wordListFn,*latFn,*ipFn=NULL;
LModel *bigramLm;
BuildType bType = unknown;
Boolean saveLatBin = FALSE;
LatFormat format = HLAT_LMLIKE;
Lattice *lat,*ipLat;
Vocab voc;
char *s;
Lattice *ProcessWordLoop(MemHeap *latHeap, Vocab *voc);
Lattice *ProcessBiGram(MemHeap *latHeap, Vocab *voc, LModel *biLM);
void SaveLattice(Lattice *lat, char *latFn, LatFormat format);
Lattice *LoadLattice(MemHeap *latHeap, char *latFn, Vocab *voc,
Boolean shortArc);
Lattice *ProcessWordPair(MemHeap *latHeap, Vocab *voc, char *fn);
if(InitShell(argc,argv,hbuild_version,hbuild_vc_id)<SUCCESS)
HError(3000,"HBuild: InitShell failed");
InitMem(); InitLabel();
InitMath();
InitDict(); InitNet();
InitLM();
CreateHeap(&buildStack, "HBuild Stack", MSTAK, 1, 0.0, 100000, LONG_MAX );
if (!InfoPrinted() && NumArgs() == 0)
ReportUsage();
if (NumArgs() == 0) Exit(0);
SetConfParms();
