void TInputFile::initializeScriptAfterSimulation(my_string gotScript, my_string gotParameter){
vector<my_string> scriptParam;
while(!gotParameter.empty()){
scriptParam.push_back(gotParameter.extract_sub_str(*"#"));
gotParameter.remove(0,1);
}
valuesToPassToScriptAfterSimulations=new my_string[scriptParam.size()];
numValuesToPassToScriptAfterSimulations=scriptParam.size();
for(int i=0;i<scriptParam.size();++i){
if(scriptParam[i].contains("SIMINPUTNAME")){
my_string a=scriptParam[i].extract_after(11);
if(a.empty()) valuesToPassToScriptAfterSimulations[i]=newSimulationProgramInputFilenames[0];
else valuesToPassToScriptAfterSimulations[i]=newSimulationProgramInputFilenames[a.toInt()];
} else {
if(scriptParam[i]=="SIMNUM") checkParametersPassedToScriptAfterSimulations=true;
else {
//check priors
if(priors->isPrior(scriptParam[i])) checkParametersPassedToScriptAfterSimulations=true;
}
valuesToPassToScriptAfterSimulations[i]=scriptParam[i];
}
}
if(checkParametersPassedToScriptAfterSimulations) scriptAfterSimulations=new TExecuteProgram(gotScript.c_str());
else scriptAfterSimulations=new TExecuteProgram(gotScript.c_str(), valuesToPassToScriptAfterSimulations, numValuesToPassToScriptAfterSimulations);
}
void my_regex::NFA::create_nfa_node(const my_string ®)
{
// 第一遍遍历的时候主要处理\开头引出的特殊符号
std::vector<p_element> element_vec;
element_vec.reserve(reg.length());
p_element elem;
for (int i = 0; i < reg.length(); i++)
{
switch (reg[i])
{
case my_char('+'):
case my_char('?'):
case my_char('*'):
elem = p_element(new num_symbol(reg[i]));
element_vec.push_back(elem);
break;
case my_char('\\'):
elem = p_element(new num_symbol(reg[++i]));// 注意，这里现在是把所有跟在反斜杠后面的字符都当作普通字符来处理
element_vec.push_back(elem);
break;
case my_char('.'):
elem = p_element(new point_symbol);
element_vec.push_back(elem);
break;
case my_char('('):
elem = p_element(new open_parentheses);
element_vec.push_back(elem);
break;
case my_char(')'):
elem = p_element(new close_parentheses);
element_vec.push_back(elem);
break;
case my_char('|'):
elem = p_element(new or_symbol);
element_vec.push_back(elem);
break;
default:
elem = p_element(new character(reg[i]));
element_vec.push_back(elem);
break;
}
}
// 将基础元素转化为前缀表达式
element_vec_2_prefix_exp(element_vec);
//assert(element_vec.size() == 1);//最后只剩下一个大的前缀表达式//已经在函数里判断过了
p_nfa_node begin(new nfa_node), end(new nfa_node);
prefix_exp_2_nfa_node(element_vec, begin, end);
end->accept = true;
// todo:加入清理内存的操作
start = begin;
}
void add_text(tuple* t, int id, my_string key_value) {
++t->key_num;
t->aid.add(id);
t->offset.add(t->len);
t->data = t->data + key_value;
t->len += key_value.size();
return;
}
//---------------------------------------------------------------------------
TInputFile::TInputFile(my_string siminputname, TPriorVector* gotPriors){
//siminputname may contain several files....
my_string buf, extension;
while(!siminputname.empty()){
buf=siminputname.extract_sub_str('#');
siminputname.remove(0,1);
buf.trim_blanks();
if(!buf.empty()){
simulationProgramInputFilenames.push_back(buf);
extension=buf.get_file_extension();
buf.remove_file_extension();
buf+="-temp";
buf+=extension;
newSimulationProgramInputFilenames.push_back(buf);
}
}
priors=gotPriors;
checkParametersPassedToSimulationProgram=false;
checkParametersPassedToScriptAfterSimulations=false;
}
size_t my_regex::NFA::try_match(const my_string &str)
{
size_t farthest = 0;//记录匹配到的最远距离
std::vector<cp_nfa_node> current_pnodes, next_pnodes;
current_pnodes.push_back(start);
get_epson_closure(current_pnodes);
bool matched = start->accept;//当前字符匹配后是否满足正则表达式
for (auto &cur_node : current_pnodes)
{
matched |= cur_node->accept;
}
for (size_t i = 0; i < str.length(); i++)
{
matched = false;
next_pnodes.clear();
for (auto &cur_node : current_pnodes)
{
// 先沿着正常边走
for (auto &n_edg : cur_node->norm_edgs)
{
if (n_edg.c == str[i])
{
next_pnodes.push_back(n_edg.p_next_node);
}
}
// 小数点通配符
for (auto &p_node : cur_node->point_edgs)
{
next_pnodes.push_back(p_node.p_next_node);
}
}
if (next_pnodes.empty())
{
// 匹配不下去了，停止
return farthest;
}
else
{
current_pnodes.swap(next_pnodes);
get_epson_closure(current_pnodes);
for (auto &cur_node : current_pnodes)
{
matched |= cur_node->accept;
}
if (matched)
{
farthest = i+1;
}
}
}
return farthest;
}
//------------------------------------------------------------------------------
void TInputFile::initializeSimulationProgramm(my_string gotSimulationprogram, my_string gotParameter){
vector<my_string> simParam;
while(!gotParameter.empty()){
simParam.push_back(gotParameter.extract_sub_str(*"#"));
gotParameter.remove(0,1);
}
valuesToPassToSimulationProgramm=new my_string[simParam.size()];
numValuesToPassToSimulationProgramm=simParam.size();
for(int i=0;i<(int)simParam.size();++i){
//check tags
if(simParam[i].contains("SIMINPUTNAME")){
my_string a=simParam[i].extract_after(11);
if(a.empty()) valuesToPassToSimulationProgramm[i]=newSimulationProgramInputFilenames[0];
else valuesToPassToSimulationProgramm[i]=newSimulationProgramInputFilenames[a.toInt()];
} else {
if(simParam[i]=="SIMNUM"){
checkParametersPassedToSimulationProgram=true;
break;
}
else {
//check priors
if(priors->isPrior(simParam[i])){
checkParametersPassedToSimulationProgram=true;
}
}
valuesToPassToSimulationProgramm[i]=simParam[i];
}
}
if(gotSimulationprogram=="INTERNALGLM"){
if(!checkParametersPassedToSimulationProgram) throw TException("The INTERNALGLM program requires model parameter tag as arguments!", _FATAL_ERROR);
simulationProgram=new TGLM(valuesToPassToSimulationProgramm, numValuesToPassToSimulationProgramm);
} else {
if(checkParametersPassedToSimulationProgram) simulationProgram=new TExecuteProgram(gotSimulationprogram.c_str());
else simulationProgram=new TExecuteProgram(gotSimulationprogram.c_str(), valuesToPassToSimulationProgramm, numValuesToPassToSimulationProgramm);
}
}
//---------------------------------------------------------------------------
TParameters::TParameters(my_string fileName, int numCommandLineParams, my_string* commandLineParams){
ifstream is (fileName.c_str());
if(!is) throw TException("Input file '" + fileName + "' could not be opened!", _FATAL_ERROR);
my_string buf, my_name;
my_string my_value;
while(is.good() && !is.eof()){
strstream isLine; // create the new stream
buf.read_line(is); // read the line
buf=buf.extract_before_doubleSlash(); // remove the commentated part
if(!buf.empty()){
// allocate the line to a new stream for easy reading
isLine << buf;
my_name.read_to_delim(isLine);
isLine >> ws;
my_value.read_to_delim(isLine);
if(!my_name.empty() && !my_value.empty()){
mapParameter[my_name]= my_value;
}
}
}
bool my_string::operator < (const my_string &a) const
{
return strcmp(c_str(), a.c_str()) < 0;
}
//---------------------------------------------------------------------------
int TExecuteProgram::execute(my_string passedParam){
// parameters to pass have to be char*!!!!!!
if(myPassedParamIdentifier>0 && passedParam!="") myParameter[myPassedParamIdentifier]=passedParam.c_str();
return execute();
}
my_string::my_string(my_string const &str) : data_(new char[str.length() + 1])
{
strcpy(data_, str.c_str());
}
