/* ************************************************************************* */
Ordering Ordering::ColamdConstrainedFirst(const VariableIndex& variableIndex,
const std::vector<Key>& constrainFirst, bool forceOrder) {
gttic(Ordering_COLAMDConstrainedFirst);
const int none = -1;
size_t n = variableIndex.size();
std::vector<int> cmember(n, none);
// Build a mapping to look up sorted Key indices by Key
FastMap<Key, size_t> keyIndices;
size_t j = 0;
for (auto key_factors: variableIndex)
keyIndices.insert(keyIndices.end(), make_pair(key_factors.first, j++));
// If at least some variables are not constrained to be last, constrain the
// ones that should be constrained.
int group = 0;
for (Key key: constrainFirst) {
cmember[keyIndices.at(key)] = group;
if (forceOrder)
++group;
}
if (!forceOrder && !constrainFirst.empty())
++group;
for(int& c: cmember)
if (c == none)
c = group;
return Ordering::ColamdConstrained(variableIndex, cmember);
}
/* ************************************************************************* */
Ordering Ordering::ColamdConstrainedLast(const VariableIndex& variableIndex,
const std::vector<Key>& constrainLast, bool forceOrder) {
gttic(Ordering_COLAMDConstrainedLast);
size_t n = variableIndex.size();
std::vector<int> cmember(n, 0);
// Build a mapping to look up sorted Key indices by Key
// TODO(frank): think of a way to not build this
FastMap<Key, size_t> keyIndices;
size_t j = 0;
for (auto key_factors: variableIndex)
keyIndices.insert(keyIndices.end(), make_pair(key_factors.first, j++));
// If at least some variables are not constrained to be last, constrain the
// ones that should be constrained.
int group = (constrainLast.size() != n ? 1 : 0);
for (Key key: constrainLast) {
cmember[keyIndices.at(key)] = group;
if (forceOrder)
++group;
}
return Ordering::ColamdConstrained(variableIndex, cmember);
}
bool is_terminal_label(const Label label) {
assert(open_label_list);
assert(label != NO_LABEL);
assert(label < label_list.size());
assert(constituent_set.exists(label) == !terminal_set.exists(label));
return terminal_set.exists(label);
}
/* ************************************************************************* */
JointMarginal Marginals::jointMarginalInformation(const std::vector<Key>& variables) const {
// If 2 variables, we can use the BayesTree::joint function, otherwise we
// have to use sequential elimination.
if(variables.size() == 1) {
Matrix info = marginalInformation(variables.front());
std::vector<size_t> dims;
dims.push_back(info.rows());
Ordering indices;
indices.insert(variables.front(), 0);
return JointMarginal(info, dims, indices);
} else {
// Obtain requested variables as ordered indices
vector<Index> indices(variables.size());
for(size_t i=0; i<variables.size(); ++i) { indices[i] = ordering_[variables[i]]; }
// Compute joint marginal factor graph.
GaussianFactorGraph jointFG;
if(variables.size() == 2) {
if(factorization_ == CHOLESKY)
jointFG = *bayesTree_.joint(indices[0], indices[1], EliminatePreferCholesky);
else if(factorization_ == QR)
jointFG = *bayesTree_.joint(indices[0], indices[1], EliminateQR);
} else {
if(factorization_ == CHOLESKY)
jointFG = *GaussianSequentialSolver(graph_, false).jointFactorGraph(indices);
else if(factorization_ == QR)
jointFG = *GaussianSequentialSolver(graph_, true).jointFactorGraph(indices);
}
// Build map from variable keys to position in factor graph variables,
// which are sorted in index order.
Ordering variableConversion;
{
// First build map from index to key
FastMap<Index,Key> usedIndices;
for(size_t i=0; i<variables.size(); ++i)
usedIndices.insert(make_pair(indices[i], variables[i]));
// Next run over indices in sorted order
size_t slot = 0;
typedef pair<Index,Key> Index_Key;
BOOST_FOREACH(const Index_Key& index_key, usedIndices) {
variableConversion.insert(index_key.second, slot);
++ slot;
}
}
// Get dimensions from factor graph
std::vector<size_t> dims(indices.size(), 0);
BOOST_FOREACH(Key key, variables) {
dims[variableConversion[key]] = values_.at(key).dim();
}
/* ************************************************************************* */
Ordering Ordering::ColamdConstrained(const VariableIndex& variableIndex,
const FastMap<Key, int>& groups) {
gttic(Ordering_COLAMDConstrained);
size_t n = variableIndex.size();
std::vector<int> cmember(n, 0);
// Build a mapping to look up sorted Key indices by Key
FastMap<Key, size_t> keyIndices;
size_t j = 0;
for (auto key_factors: variableIndex)
keyIndices.insert(keyIndices.end(), make_pair(key_factors.first, j++));
// Assign groups
typedef FastMap<Key, int>::value_type key_group;
for(const key_group& p: groups) {
// FIXME: check that no groups are skipped
cmember[keyIndices.at(p.first)] = p.second;
}
return Ordering::ColamdConstrained(variableIndex, cmember);
}
/* ************************************************************************* */
Ordering Ordering::ColamdConstrainedLast(const VariableIndex& variableIndex,
const std::vector<Key>& constrainLast, bool forceOrder) {
gttic(Ordering_COLAMDConstrainedLast);
size_t n = variableIndex.size();
std::vector<int> cmember(n, 0);
// Build a mapping to look up sorted Key indices by Key
FastMap<Key, size_t> keyIndices;
size_t j = 0;
BOOST_FOREACH(const VariableIndex::value_type key_factors, variableIndex)
keyIndices.insert(keyIndices.end(), make_pair(key_factors.first, j++));
// If at least some variables are not constrained to be last, constrain the
// ones that should be constrained.
int group = (constrainLast.size() != n ? 1 : 0);
BOOST_FOREACH(Key key, constrainLast) {
cmember[keyIndices.at(key)] = group;
if (forceOrder)
++group;
}
static void posclass_init() {
assert(!posclass_is_init);
posclass_map.clear();
_all_posclass.clear();
unsigned skipcnt = 0;
for (unsigned i = 0; i < tag_to_class_cnt; i++) {
// Skip labels not in the vocabulary.
if (!is_label_string(tag_to_class[i][0])) {
Debug::log(1) << "Skipping unknown label " << tag_to_class[i][0] << " in posclass_init()\n";
skipcnt++;
continue;
}
Label tag = string_to_label(tag_to_class[i][0]);
assert(is_terminal_label(tag));
Posclass c = string_to_posclass(tag_to_class[i][1]);
assert(posclass_list[c] == tag_to_class[i][1]);
assert(posclass_list[c] != "");
posclass_map.insert(tag, c);
}
posclass_map.lock();
for (unsigned i = 0; i < posclass_cnt; i++) {
if (posclass_list[i] != "")
_all_posclass.push_back(i);
if (posclass_list[i] == "::N")
_Posclass_N = i;
else if (posclass_list[i] == "::NP")
_Posclass_NP = i;
}
posclass_is_init = true;
}
int main( int argc, char * argv[] )
{
Process process;
Session session1, session2;
FastMapEditor * editor;
FastMap * hashmap;
string fname = "/cpp_fastmap_get";
string hname = fname + "/test";
const char * key = "My Key";
const char * data = "My Data";
uint32_t length;
char buffer[50];
psoObjectDefinition mapDef = { PSO_FAST_MAP, 0, 0, 0 };
psoKeyFieldDefinition keys = { "MyKey", PSO_KEY_VARBINARY, 20 };
psoFieldDefinition fields[1] = {
{ "Field_1", PSO_VARCHAR, {10} }
};
try {
if ( argc > 1 ) {
process.Init( argv[1], argv[0] );
}
else {
process.Init( "10701", argv[0] );
}
}
catch( pso::Exception exc ) {
cerr << "Test failed in init phase, error = " << exc.Message() << endl;
cerr << "Is the server running?" << endl;
return 1;
}
try {
session1.Init();
session2.Init();
session2.CreateFolder( fname );
DataDefinition dataDefObj( session2,
"cpp_fastmap_get",
PSO_DEF_PHOTON_ODBC_SIMPLE,
(unsigned char *)fields,
sizeof(psoFieldDefinition) );
KeyDefinition keyDefObj( session2,
"cpp_fastmap_get",
PSO_DEF_PHOTON_ODBC_SIMPLE,
(unsigned char *)&keys,
sizeof(psoKeyFieldDefinition) );
session2.CreateMap( hname, mapDef, dataDefObj, keyDefObj );
session2.Commit();
hashmap = new FastMap( session1, hname );
editor = new FastMapEditor( session2, hname );
editor->Insert( key, strlen(key), data, strlen(data) );
}
catch( pso::Exception exc ) {
cerr << "Test failed - line " << __LINE__ << ", error = " << exc.Message() << endl;
return 1;
}
// Invalid arguments to tested function.
try {
// No commit yet.
hashmap->Get( key, strlen(key), buffer, 50, length );
// Should never come here
cerr << "Test failed - line " << __LINE__ << endl;
return 1;
}
catch( pso::Exception exc ) {
if ( exc.ErrorCode() != PSO_NO_SUCH_ITEM ) {
cerr << "Test failed - line " << __LINE__ << ", error = " << exc.Message() << endl;
return 1;
}
}
try {
editor->Close();
session2.Commit(); // Commit the editions
session1.Commit(); // Refresh session1 (and hashmap)
}
catch( pso::Exception exc ) {
cerr << "Test failed - line " << __LINE__ << ", error = " << exc.Message() << endl;
return 1;
}
try {
hashmap->Get( NULL, strlen(key), buffer, 50, length );
// Should never come here
cerr << "Test failed - line " << __LINE__ << endl;
return 1;
}
catch( pso::Exception exc ) {
if ( exc.ErrorCode() != PSO_NULL_POINTER ) {
cerr << "Test failed - line " << __LINE__ << ", error = " << exc.Message() << endl;
return 1;
}
}
try {
hashmap->Get( key, 0, buffer, 50, length );
// Should never come here
cerr << "Test failed - line " << __LINE__ << endl;
return 1;
}
catch( pso::Exception exc ) {
if ( exc.ErrorCode() != PSO_INVALID_LENGTH ) {
cerr << "Test failed - line " << __LINE__ << ", error = " << exc.Message() << endl;
return 1;
}
}
try {
hashmap->Get( key, strlen(key), NULL, 50, length );
// Should never come here
cerr << "Test failed - line " << __LINE__ << endl;
return 1;
}
catch( pso::Exception exc ) {
if ( exc.ErrorCode() != PSO_NULL_POINTER ) {
cerr << "Test failed - line " << __LINE__ << ", error = " << exc.Message() << endl;
return 1;
}
}
try {
hashmap->Get( key, strlen(key), buffer, 2, length );
// Should never come here
cerr << "Test failed - line " << __LINE__ << endl;
return 1;
}
catch( pso::Exception exc ) {
if ( exc.ErrorCode() != PSO_INVALID_LENGTH ) {
cerr << "Test failed - line " << __LINE__ << ", error = " << exc.Message() << endl;
return 1;
}
}
// End of invalid args. This call should succeed.
try {
hashmap->Get( key, strlen(key), buffer, 50, length );
}
catch( pso::Exception exc ) {
cerr << "Test failed - line " << __LINE__ << ", error = " << exc.Message() << endl;
return 1;
}
if ( memcmp( buffer, data, 7 ) != 0 ) {
cerr << "Test failed - line " << __LINE__ << endl;
return 1;
}
return 0;
}
void label_frequency_init() {
_label_freq_map.insert(Label_ADJP(), 0.021609);
_label_freq_map.insert(Label_ADVP(), 0.0364083);
_label_freq_map.insert(Label_CONJP(), 0.000107241);
_label_freq_map.insert(Label_FRAG(), 0.0020054);
_label_freq_map.insert(Label_INTJ(), 0.000557653);
_label_freq_map.insert(Label_LST(), 0.000160861);
_label_freq_map.insert(Label_NAC(), 0.000139413);
/// \bug This will be broken depending on whether we're using NPBs or not
// _label_freq_map.insert(Label_NP(), 0.357198); // w/o NPB
// _label_freq_map.insert(Label_NP(), 0.0775364); // w/ NPB
// _label_freq_map.insert(Label_NPB(), 0.279904);
_label_freq_map.insert(Label_NX(), 0.000589825);
_label_freq_map.insert(Label_PP(), 0.0881842);
_label_freq_map.insert(Label_PRN(), 0.00154427);
_label_freq_map.insert(Label_QP(), 0.0127724);
_label_freq_map.insert(Label_RRC(), 2.14482e-05);
_label_freq_map.insert(Label_S(), 0.140453);
_label_freq_map.insert(Label_SBAR(), 0.0185205);
_label_freq_map.insert(Label_SBARQ(), 0.00102951);
_label_freq_map.insert(Label_SINV(), 0.00170513);
_label_freq_map.insert(Label_SQ(), 0.00164079);
_label_freq_map.insert(Label_TOP(), 0.109107);
_label_freq_map.insert(Label_UCP(), 0.000353895);
_label_freq_map.insert(Label_VP(), 0.199125);
_label_freq_map.insert(Label_WHADJP(), 9.65168e-05);
_label_freq_map.insert(Label_WHADVP(), 0.00163006);
_label_freq_map.insert(Label_WHNP(), 0.00412877);
_label_freq_map.insert(Label_WHPP(), 8.57927e-05);
_label_freq_map.insert(Label_X(), 0.000825755);
_label_freq_map.lock();
_label_freq_is_init = true;
}
int main( int argc, char * argv[] )
{
Process process;
Session session;
FastMap * map;
string fname = "/cpp_fastmap_fielddefODBC";
string hname = fname + "/test";
size_t len;
psoObjectDefinition mapDef = { PSO_FAST_MAP, 0, 0, 0 };
DataDefBuilderODBC fieldDef( 5 );
KeyDefBuilderODBC keyDef( 3 );
DataDefinition * returnedDataDef;
KeyDefinition * returnedKeyDef;
unsigned char * buffer;
try {
fieldDef.AddField( "field1", 6, PSO_TINYINT, 0, 0, 0 );
fieldDef.AddField( "field2", 6, PSO_INTEGER, 0, 0, 0 );
fieldDef.AddField( "field3", 6, PSO_CHAR, 30, 0, 0 );
fieldDef.AddField( "field4", 6, PSO_SMALLINT, 0, 0, 0 );
fieldDef.AddField( "field5", 6, PSO_LONGVARBINARY, 0, 0, 0 );
keyDef.AddKeyField( "keyfield1", 9, PSO_KEY_INTEGER, 0 );
keyDef.AddKeyField( "keyfield2", 9, PSO_KEY_CHAR, 30 );
keyDef.AddKeyField( "keyfield3", 9, PSO_KEY_LONGVARBINARY, 0 );
}
catch( pso::Exception exc ) {
cerr << "Test failed - line " << __LINE__ << ", error = " << exc.Message() << endl;
return 1;
}
try {
if ( argc > 1 ) {
process.Init( argv[1], argv[0] );
}
else {
process.Init( "10701", argv[0] );
}
}
catch( pso::Exception exc ) {
cerr << "Test failed in init phase, error = " << exc.Message() << endl;
cerr << "Is the server running?" << endl;
return 1;
}
try {
session.Init();
session.CreateFolder( fname );
DataDefinition dataDefObj( session,
"cpp_fastmap_datadef_odbc",
PSO_DEF_PHOTON_ODBC_SIMPLE,
fieldDef.GetDefinition(),
fieldDef.GetDefLength() );
KeyDefinition keyDefObj( session,
"cpp_fastmap_datadef_odbc",
PSO_DEF_PHOTON_ODBC_SIMPLE,
keyDef.GetDefinition(),
keyDef.GetDefLength() );
session.CreateMap( hname, mapDef, dataDefObj, keyDefObj );
map = new FastMap( session, hname );
}
catch( pso::Exception exc ) {
cerr << "Test failed - line " << __LINE__ << ", error = " << exc.Message() << endl;
return 1;
}
try {
returnedDataDef = map->GetDataDefinition();
returnedKeyDef = map->GetKeyDefinition();
}
catch( pso::Exception exc ) {
cerr << "Test failed - line " << __LINE__ << ", error = " << exc.Message() << endl;
return 1;
}
try {
len = 5 * sizeof(psoFieldDefinition);
buffer = new unsigned char [len];
if ( len != returnedDataDef->GetLength() ) {
cerr << "Test failed - line " << __LINE__ << endl;
return 1;
}
returnedDataDef->GetDefinition( buffer, len );
if ( memcmp( fieldDef.GetDefinition(), buffer, len ) != 0 ) {
cerr << "Test failed - line " << __LINE__ << endl;
return 1;
}
delete [] buffer;
len = 3 * sizeof(psoKeyFieldDefinition);
buffer = new unsigned char [len];
if ( len != returnedKeyDef->GetLength() ) {
cerr << "Test failed - line " << __LINE__ << endl;
return 1;
}
returnedKeyDef->GetDefinition( buffer, len );
if ( memcmp( keyDef.GetDefinition(), buffer, len ) != 0 ) {
cerr << "Test failed - line " << __LINE__ << endl;
return 1;
}
}
catch( pso::Exception exc ) {
cerr << "Test failed - line " << __LINE__ << ", error = " << exc.Message() << endl;
return 1;
}
return 0;
}
/* ************************************************************************* */
FastMap<Key, size_t> Ordering::invert() const {
FastMap<Key, size_t> inverted;
for (size_t pos = 0; pos < this->size(); ++pos)
inverted.insert(make_pair((*this)[pos], pos));
return inverted;
}
/// \todo Make some assertion about # of constit. labels, and/or that
/// they are the lowest numbered ones?
void read_labels(const string filename) {
assert(!open_label_list);
ifstream fin(filename.c_str());
assert(fin.is_open());
_all_labels.clear();
_all_constituent_labels.clear();
_all_terminal_labels.clear();
_max_label = 0;
label_list.clear();
label_map.clear();
terminal_set.clear();
constituent_set.clear();
unsigned i, is_terminal, cnt;
string label;
while(!fin.eof()) {
fin >> i >> ws >> is_terminal >> ws >> cnt >> label >> ws;
if (i == NO_LABEL)
label = "";
label_list.push_back(label);
assert(label_list.at(i) == label);
label_map[label] = i; // FIXME: Don't use [] operator
if (i != NO_LABEL) {
_all_labels.push_back(i);
if (is_terminal) {
terminal_set.insert(i, true);
_all_terminal_labels.push_back(i);
} else {
constituent_set.insert(i, true);
_all_constituent_labels.push_back(i);
}
if (i > _max_label) _max_label = i+1;
}
}
assert(label_list.at(NO_LABEL) == "");
assert(label_map[""] == NO_LABEL); // FIXME: Don't use [] operator
assert(label_list.size() == label_map.size());
// assert(label_map.size() == terminal_set.size() + constituent_set.size() + 1);
terminal_set.lock();
constituent_set.lock();
/*
Debug::log(1) << "Read " << constituent_set.size() << " constituents, " << \
terminal_set.size() << " terminals from '" << filename << "'\n";
*/
fin.close();
open_label_list = true;
if (is_label_string("ADJP")) _Label_ADJP = string_to_label("ADJP");
if (is_label_string("ADVP")) _Label_ADVP = string_to_label("ADVP");
if (is_label_string("AUX")) _Label_AUX = string_to_label("AUX");
if (is_label_string("AUXG")) _Label_AUXG = string_to_label("AUXG");
if (is_label_string("CC")) _Label_CC = string_to_label("CC");
if (is_label_string("CD")) _Label_CD = string_to_label("CD");
if (is_label_string("COLON")) _Label_COLON = string_to_label(":");
if (is_label_string("COMMA")) _Label_COMMA = string_to_label(",");
if (is_label_string("CONJP")) _Label_CONJP = string_to_label("CONJP");
if (is_label_string("DOLLAR")) _Label_DOLLAR = string_to_label("$");
if (is_label_string("DT")) _Label_DT = string_to_label("DT");
if (is_label_string("EX")) _Label_EX = string_to_label("EX");
if (is_label_string("FRAG")) _Label_FRAG = string_to_label("FRAG");
if (is_label_string("FW")) _Label_FW = string_to_label("FW");
if (is_label_string("HASH")) _Label_HASH = string_to_label("#");
if (is_label_string("IN")) _Label_IN = string_to_label("IN");
if (is_label_string("INTJ")) _Label_INTJ = string_to_label("INTJ");
if (is_label_string("JJ")) _Label_JJ = string_to_label("JJ");
if (is_label_string("JJR")) _Label_JJR = string_to_label("JJR");
if (is_label_string("JJS")) _Label_JJS = string_to_label("JJS");
if (is_label_string("LS")) _Label_LS = string_to_label("LS");
if (is_label_string("LST")) _Label_LST = string_to_label("LST");
if (is_label_string("MD")) _Label_MD = string_to_label("MD");
if (is_label_string("NAC")) _Label_NAC = string_to_label("NAC");
if (is_label_string("NN")) _Label_NN = string_to_label("NN");
if (is_label_string("NNP")) _Label_NNP = string_to_label("NNP");
if (is_label_string("NNPS")) _Label_NNPS = string_to_label("NNPS");
if (is_label_string("NNS")) _Label_NNS = string_to_label("NNS");
if (is_label_string("NP")) _Label_NP = string_to_label("NP");
if (is_label_string("NPB")) _Label_NPB = string_to_label("NPB");
if (is_label_string("NX")) _Label_NX = string_to_label("NX");
if (is_label_string("POS")) _Label_POS = string_to_label("POS");
if (is_label_string("PP")) _Label_PP = string_to_label("PP");
if (is_label_string("PRN")) _Label_PRN = string_to_label("PRN");
if (is_label_string("PRP")) _Label_PRP = string_to_label("PRP");
if (is_label_string("PRPP")) _Label_PRPP = string_to_label("PRP$");
if (is_label_string("PRT")) _Label_PRT = string_to_label("PRT");
if (is_label_string("QP")) _Label_QP = string_to_label("QP");
if (is_label_string("RB")) _Label_RB = string_to_label("RB");
if (is_label_string("RBR")) _Label_RBR = string_to_label("RBR");
if (is_label_string("RBS")) _Label_RBS = string_to_label("RBS");
if (is_label_string("RP")) _Label_RP = string_to_label("RP");
if (is_label_string("RRC")) _Label_RRC = string_to_label("RRC");
if (is_label_string("S")) _Label_S = string_to_label("S");
if (is_label_string("SBAR")) _Label_SBAR = string_to_label("SBAR");
if (is_label_string("SBARQ")) _Label_SBARQ = string_to_label("SBARQ");
if (is_label_string("SINV")) _Label_SINV = string_to_label("SINV");
if (is_label_string("SQ")) _Label_SQ = string_to_label("SQ");
if (is_label_string("SYM")) _Label_SYM = string_to_label("SYM");
if (is_label_string("TO")) _Label_TO = string_to_label("TO");
if (is_label_string("TOP")) _Label_TOP = string_to_label("TOP");
if (is_label_string("UCP")) _Label_UCP = string_to_label("UCP");
if (is_label_string("UH")) _Label_UH = string_to_label("UH");
if (is_label_string("VB")) _Label_VB = string_to_label("VB");
if (is_label_string("VBD")) _Label_VBD = string_to_label("VBD");
if (is_label_string("VBG")) _Label_VBG = string_to_label("VBG");
if (is_label_string("VBN")) _Label_VBN = string_to_label("VBN");
if (is_label_string("VBP")) _Label_VBP = string_to_label("VBP");
if (is_label_string("VBZ")) _Label_VBZ = string_to_label("VBZ");
if (is_label_string("VP")) _Label_VP = string_to_label("VP");
if (is_label_string("WDT")) _Label_WDT = string_to_label("WDT");
if (is_label_string("WHADJP")) _Label_WHADJP = string_to_label("WHADJP");
if (is_label_string("WHADVP")) _Label_WHADVP = string_to_label("WHADVP");
if (is_label_string("WHNP")) _Label_WHNP = string_to_label("WHNP");
if (is_label_string("WHPP")) _Label_WHPP = string_to_label("WHPP");
if (is_label_string("WP")) _Label_WP = string_to_label("WP");
if (is_label_string("X")) _Label_X = string_to_label("X");
}
