void get() {
outFile = ParamGetString("outputFile");
stopwords = ParamGetString("stopwords");
acronyms = ParamGetString("acronyms");
docFormat = ParamGetString("docFormat");
stemmer = ParamGetString("stemmer");
}
void get() {
// the string with quotes are the actual variable names to use for specifying the parameters
databaseIndex = ParamGetString("index");
queryStream = ParamGetString("query");
resultFile = ParamGetString("result","res");
weightScheme = ParamGetString("weightScheme","RawTF");
resultCount = ParamGetInt("resultCount", 100);
}
void get() {
index = ParamGetString("index");
indexType = ParamGetString("indexType");
manager = ParamGetString("manager");
mgrType = ParamGetString("managerType");
memory = ParamGetInt("memory", 128000000);
stopwords = ParamGetString("stopwords");
acronyms = ParamGetString("acronyms");
docFormat = ParamGetString("docFormat");
dataFiles = ParamGetString("dataFiles");
stemmer = ParamGetString("stemmer");
countStopWords = (ParamGetString("countStopWords", "false") == "true");
}
void get()
{
resultFile = ParamGetString("resultFile");
testFile = ParamGetString("testFile");
priorFile = ParamGetString("priorFile");
smoothingParam = ParamGetDouble("smoothingParam", 0.5);
dirParamNN = ParamGetDouble("dirParamNN", 100);
splitLimit = ParamGetInt("splitLimit", 5000);
minLongitudeStep = ParamGetDouble("minLongitudeStep", 0.01);
minLatitudeStep = ParamGetDouble("minLatitudeStep", 0.01);
std::string s = ParamGetString("excludeTestUsers","false");
excludeTestUsers = (s.compare("false")==0) ? false : true;
geoTermThreshold = ParamGetDouble("geoTermThreshold",-1.0);
generalUseTermFilter = ParamGetInt("generalUseTermFilter",-1);
defaultLatitude = ParamGetDouble("defaultLatitude", 0.0);
defaultLongitude = ParamGetDouble("defaultLongitude", 0.0);
fields = ParamGetString("fields");
alphas = ParamGetString("alphas");
geoTermThresholdFile = ParamGetString("geoTermThresholdFile");
takenInMonth = ParamGetInt("takenInMonth",-1);
modulosOp = ParamGetInt("modulosOp",1);//if fewer training docs should be used, modulosOp>1
termFile = ParamGetString("termFile","");
}
lemur::api::Stemmer* lemur::api::TextHandlerManager::createStemmer(string type, string datadir, string func) {
Stemmer *stemmer = NULL;
if (type.empty()) {
// didn't pass in type, try to get it from the paramstack
type = ParamGetString("stemmer");
}
// if it's still empty, return nothing
if (type.empty())
return NULL;
// make it all lowercase
for (int i=0;i<type.length();i++)
type[i] = tolower(type[i]);
try {
if (type == lemur::parse::KStemmer::identifier) {
stemmer = new lemur::parse::KStemmer();
// cerr << "created Kstemmer" << endl;
} else if (type == lemur::parse::PorterStemmer::identifier) {
stemmer = new lemur::parse::PorterStemmer();
//cerr << "created porter stemmer" << endl;
}
} catch (Exception &ex) {
LEMUR_RETHROW(ex, "Could not create Stemmer");
}
return stemmer;
}
void get()
{
resultFile = ParamGetString("resultFile");
splitLimit = ParamGetInt("splitLimit", 5000);
minLongitudeStep = ParamGetDouble("minLongitudeStep", 0.01);
minLatitudeStep = ParamGetDouble("minLatitudeStep", 0.01);
minAccuracy = ParamGetInt("minAccuracy", 11);
geoTermThreshold = ParamGetDouble("geoTermThreshold",-1.0);
generalUseTermFilter = ParamGetInt("generalUseTermFilter",-1);
}
lemur::api::Parser* lemur::api::TextHandlerManager::createParser(string type, string acros) {
// Create the appropriate parser.
Parser * parser = NULL;
if (type.empty()) {
// didn't pass in type, try to get it from the paramstack
type = ParamGetString("docFormat");
}
// if it's still empty, return nothing
if (type.empty())
return NULL;
// make it all lowercase
for (int i=0;i<type.length();i++)
type[i] = tolower(type[i]);
if (type == lemur::parse::WebParser::identifier) {
parser = new lemur::parse::WebParser();
} else if (type == lemur::parse::TrecParser::identifier) {
parser = new lemur::parse::TrecParser();
} else if (type == lemur::parse::IdentifinderParser::identifier) {
parser = new lemur::parse::IdentifinderParser();
}
if (parser) {
// tell the parser about the acronyms list
if (acros.empty()) {
acros = ParamGetString("acronyms");
}
if (!acros.empty()) {
parser->setAcroList(acros);
}
}
return parser;
}
lemur::api::Stopper* lemur::api::TextHandlerManager::createStopper(string filename) {
Stopper* stopper = NULL;
if (filename.empty())
filename = ParamGetString("stopwords");
if (!filename.empty()) {
try {
stopper = new Stopper(filename);
//cerr << "created stopper" << endl;
} catch (Exception &ex) {
LEMUR_RETHROW(ex, "Could not create Stopper using file name");
}
}
return stopper;
}
lemur::utility::String ParamGetString(const lemur::utility::String &s)
{
lemur::utility::String def = "";
return ParamGetString(s, def);
}
lemur::utility::String ParamGetString(const lemur::utility::String &s, lemur::utility::String &value, const lemur::utility::String &def)
{
value = ParamGetString(s, def);
return value;
}
void get() {
expandedQuery = ParamGetString("expandedQuery");
}
void get() {
index = ParamGetString("index");
L2File = ParamGetString("L2File");
}
