void WebPage::parse () {
ifstream fin;
fin.open(infile.c_str());
if(!fin){
throw infile;
}
string temp;
while(getline(fin,temp)) //while there are lines to get
lfile.push_back(temp); //add it to the list
fin.close();
string line;
for(unsigned int i =0; i < lfile.size(); i++){
line = lfile.at(i); //for each item i the list, set to line
while(!line.empty()){ //while each line still has characters
string temp;
int k=0;
for( ;isalnum(line[k]) || line[k] == '['; k++){ //checks if alphanumeric
if(line[k] == '['){ //appropriately parses anchor text
k++;
}
temp.push_back(line[k]); //pushes back character until next word
}
if(line[k] == ']' && line[k+1] == '('){
string link;
k+=2;
while(line[k] != ')'){
link.push_back(line[k]);
k++;
}
//find matching strings,
//if found just add the mapped webpage as outgoing link
//if not create a dummy link, add that to the map with the filename
//then add the dummy link to the outgoing links
map<string,WebPage*>::iterator it = allLinks->find(link);
if(it!=allLinks->end()){
addOutgoingLink(it->second);
}
else{
WebPage *dummy = new WebPage(link,allLinks);
(*allLinks)[link] = dummy;
addOutgoingLink(dummy);
}
}
for(int j = 0; temp[j]; j++){ //converts to lower case
temp[j] = tolower(temp[j]);
}
if(words.find(temp)==words.end() && !temp.empty()) //makes sure there is no duplicate word
words.insert(temp);
line.erase(0,k+1); //moves to the next word in the string
}
}
}
WebPage::WebPage (std::string filename) {
std::ifstream file;
name = filename;
file.open(filename);
if (file.fail()) {
throw(std::runtime_error(std::string("Failed to open file.")));
} else {
fileGiven = true;
}
while(!file.fail()) {
std::string buffer = "";
file >> buffer;
if(!file.fail()) {
data += (buffer + " ");
}
}
for(unsigned int i = 0; i < data.length(); i++) {
data[i] = tolower(data[i]);
}
file.close();
std::stringstream ss;
ss.str(data);
while(!ss.fail()) {
std::string word = "";
ss >> word;
unsigned int start, end = 0;
for(unsigned int i = 0; i < word.length(); i++) {
if(word[i] == '[') {
start = i+1;
}
if(word[i] == ']') {
end = i;
}
}
if(!(start == 0 && end == 0) && start < word.length() && end < word.length()) {
string otherFileName = word.substr(start, end - start);
addOutgoingLink(otherFileName);
}
}
}
void WebPage::parse (string total, string temp)
{
string linkName = "";
string linkFile = "";
int start = 0;
int end = 0;
for (unsigned int j = 0; j < 2; j++)
{
for (unsigned int i = 0; i < total.length(); i++)
{
if (total[i] == '[')
{
start = i;
i++;
while (total[i] != ']')
{
linkName += total[i];
i++;
}
i++;
if (total[i] == '(')
{
i++;
while (total[i] != ')')
{
linkFile += total[i];
i++;
}
i++;
end = i;
total.erase(total.begin()+start, total.begin()+end);
addOutgoingLink(linkFile);
linkName = "";
linkFile = "";
}
}
}
}
totalWebPage = total;
replace(total.begin(), total.end(), '!', ' ');
replace(total.begin(), total.end(), '\\', ' ');
replace(total.begin(), total.end(), '#', ' ');
replace(total.begin(), total.end(), '$', ' ');
replace(total.begin(), total.end(), '%', ' ');
replace(total.begin(), total.end(), '&', ' ');
replace(total.begin(), total.end(), '\'', ' ');
replace(total.begin(), total.end(), '(', ' ');
replace(total.begin(), total.end(), ')', ' ');
replace(total.begin(), total.end(), '*', ' ');
replace(total.begin(), total.end(), '+', ' ');
replace(total.begin(), total.end(), ',', ' ');
replace(total.begin(), total.end(), '-', ' ');
replace(total.begin(), total.end(), '.', ' ');
replace(total.begin(), total.end(), '/', ' ');
replace(total.begin(), total.end(), ':', ' ');
replace(total.begin(), total.end(), ';', ' ');
replace(total.begin(), total.end(), '<', ' ');
replace(total.begin(), total.end(), '>', ' ');
replace(total.begin(), total.end(), '=', ' ');
replace(total.begin(), total.end(), '?', ' ');
replace(total.begin(), total.end(), '@', ' ');
replace(total.begin(), total.end(), '[', ' ');
replace(total.begin(), total.end(), ']', ' ');
replace(total.begin(), total.end(), '\"', ' ');
replace(total.begin(), total.end(), '/', ' ');
replace(total.begin(), total.end(), '^', ' ');
replace(total.begin(), total.end(), '_', ' ');
replace(total.begin(), total.end(), '\'', ' ');
replace(total.begin(), total.end(), '{', ' ');
replace(total.begin(), total.end(), '}', ' ');
replace(total.begin(), total.end(), '|', ' ');
replace(total.begin(), total.end(), '~', ' ');
stringstream buffer(total);
while (buffer >> temp)
{
locale loc;
string lowerTemp;
for(unsigned int j = 0; j < temp.length(); j++)
{
lowerTemp = lowerTemp + tolower(temp[j], loc);
}
words.insert(lowerTemp);
}
}