refere~1.c

一百个病毒的源代码 包括熊猫烧香等 极其具有研究价值

// Copyright 1999 Jose M. Vidal
// Jose M. Vidal, [e-mail], http://jmvidal.ece.sc.edu
//
// This program is free software.  You can redistribute it and/or modify
// it under the terms of the GNU General Public License
//
// $Id: reference.C,v 1.76 2000/12/15 23:45:57 jmvidal Exp $
#ifdef HAVE_CONFIG_H
#include <config.h>
#endif

#include "reference.H"

//Table to convert mac chars to ISO-8859-1, starts at 128.
//
const int MACCHAR[128] = {
  196, 197, 199, 201, 209, 214, 220, 225, 224, 226, 228, 227, 229,
  231, 233, 232, 234, 235, 237, 236, 238, 239, 241, 243, 242, 244,
  246, 245, 250, 249, 251, 252, 43, 176, 162, 163, 167, 183, 182, 223,
  174, 169, 32, 180, 168, 32, 198, 216, 32, 177, 32, 32, 165, 181,
  240, 83, 80, 112, 83, 170, 186, 79, 230, 248, 191, 161, 172, 32,
  102, 126, 68, 171, 187, 32, 160, 192, 195, 213, 32, 32, 32, 32,
  32, 32, 96, 39, 247, 183, 255, 255, 47, 164, 60, 62, 32, 32, 32,
  183, 44, 32, 32, 194, 202, 193, 203, 200, 205, 206, 207, 204,
  211, 212, 183, 210, 218, 219, 217, 105, 94, 126, 175, 94, 183, 176,
  184, 32, 184, 94};

/*
string eliminateAccents(const string & s){
  string c;
  string::iterator ci = c.begin();
  for (string::const_iterator i = s.begin(); i != s.end(); ++ i, ++ci){
    char temp = *i;
    if (temp > 128)
      *ci = '_';
    else *ci = *i;
  };
  return c;
}
*/

/** Returns a copy of s where all the accented and other weird characters
    have been replaced with conterparts */
string replaceAccents(const string & s){
  string res = "";
  for (string::const_iterator i = s.begin(); i != s.end(); ++ i){
    char c = *i;
    unsigned short ascval = (unsigned short)c & (unsigned short)255;
    if ((ascval >= 32 ) && (ascval <= 255)){
      //      cout << ascval << "-" << c << " =" << EQUIVCHAR[ascval-32] << "=" << endl;
      res += EQUIVCHAR[ascval-32];
    }
    else{
      res += c;
      //      cout << "not in range=" << ascval << "-" << c << endl;
    }
  }
  return res;
}
	

//remove all HTML, that is, everything between < and >
void removeHTML(string & s){
  string::size_type start = 0;
  string::size_type end = 0;
  while ( (start = s.find("<")) != string::npos){
    end = s.find(">", start);
    s.replace(start, end-start + 1, ""); 
  }
}

//from Stroustrup book
int cmpNoCase(const string& s, const string &s2)
{
  string::const_iterator p = s.begin();
  string::const_iterator p2 = s2.begin();
  
  while (p != s.end() && p2 != s2.end()) {
    if (toupper(*p) != toupper(*p2))
      return (toupper(*p) < toupper(*p2)) ? -1 : 1;
    ++p;
    ++p2;
  }
  return (s2.size() == s.size()) ? 0 : (s.size() < s2.size()) ? -1 : 1;
}

//replace all instances of s1 with s2 on s
int replaceAll(string & s, string s1, string s2)
{
  int c = 0;
  string::size_type f1 = 0;
  while ((f1 = s.find(s1,f1)) != string::npos) {
    c++;
    s.replace(f1,s1.size(), s2);
    f1 += s2.size();
  }
  return c;
}

/** escape all funny characters */
string javaScriptEscape(const string & s){
  string res = "";
  for (unsigned int i = 0; i < s.size(); ++i){
    if (s[i] == '\\'){
      res += '\\';
      res += '\\';
    }
    else if (s[i] == '\'') {
      res += '\\';
      res += '\'';
    }
    else if (s[i] == '\"') {
      res += '\\';
      res += '\'';
    }
    else if (s[i] == '\n') {
      res += '\\';
      res += 'n';
    }
    else if (s[i] == '\t') {
      res += '\\';
      res += 't';
    }
    else if (s[i] == '\r') {
      res += '\\';
      res += 'n';
    }
    else
      res += s[i];
  }
  return res;
}

// get the month
string getMonth(int m)
{
  switch (m) {
  case 1: return "January"; break;
  case 2: return "February"; break;
  case 3: return "March"; break;
  case 4: return "April"; break;
  case 5: return "May"; break;
  case 6: return "June"; break;
  case 7: return "July"; break;
  case 8: return "August"; break;
  case 9: return "September"; break;
  case 10: return "October"; break;
  case 11: return "November"; break;
  case 12: return "December"; break;
  };
  return "Bad Month";
}


string::size_type findNC(const string & source, 
                         const string & search, 
                         const string::size_type startPos)
{
  string::size_type ssize = search.size();
  if (ssize > source.size())
    return string::npos; //because size_type cannot handle negatives!!
  for (string::size_type i = startPos; i <= source.size() - ssize; i++){
    string tu = source.substr(i, ssize);
    if (cmpNoCase(tu, search) == 0)
      return i;
  }
  return string::npos;
  
}

//replace all instances of s1 with s2 on s
int replaceAllNC(string & s, string s1, string s2)
{
  int c = 0;
  string::size_type f1 = 0;
  while ((f1 = findNC(s,s1,f1)) != string::npos) {
    c++;
    s.replace(f1,s1.size(), s2);
    f1 += s2.size();
  }
  return c;
}


string getArgumentValue(const string & str, const string & argument)
{
  int l=0, r=0;
  l = str.find(argument + "=");
  if (l>=0) {
    l = str.find("\"", l);
    l++;
    r = str.find("\"", l);
    string t = str.substr(l,r-l);
    replaceAll(t, "\n", "");
    return t;
  }
  return string("");
}

string name2filename(string const & n){
  string s = n;
  string::size_type p = 0;
  while ((p = s.find(" ")) != string::npos){
    s.replace(p++,1,"_");
  }
  p = 0;
  while ((p = s.find("/")) != string::npos){
    s.replace(p++,1,"-");
  }
  p = 0;
  while ((p = s.find("\"")) != string::npos){
    s.replace(p++,1,"-"); 
  }
  // Escape 8bit characters.
  string res = "";
  char hexbuf[4];
  for (unsigned int i = 0; i < s.size(); ++i){
    if (s[i] & 0x80) {
      ostrstream hexstr(hexbuf, 4);
      hexstr << 'x' << setw(2) << setfill('0') << hex << (int)(s[i] & 0xff);
      res += hexbuf[0];
      res += hexbuf[1];
      res += hexbuf[2];
    }
    else
      res += s[i];
  }
  return res;
}

//make a string usable as url
string name2url(string const & n){
  string s = name2filename(n);
  replaceAll(s, "?", "%3F");
  return s;
}


//RETURN a string that is like s but without any leading or
//  trailing whitespace.
// We also get rid of any \n
string trim(const string & s) 
{
  int l = s.length();
  int left =0, right =l-1;
  for (; left <= right; left++) 
    if (!(isspace(s[left]))) break;
  for (; right >= left; right--)
    if (!(isspace(s[right]))) break;
  string t(s.substr(left, right+1-left));
  replaceAll(t, "\n", "");
  return t;
}

string myItoa (int n)
{
  string         ret;
  strstream      tmp;
  tmp << n << ends;
  ret = tmp.str ();
  tmp.rdbuf()->freeze(0);
  return ret;
}


/** Our time zone*/
extern char *tzname[2];

string padWithZero(string s){
  if (s.length() == 1)
    return "0" + s;
  return s;
}

/** Returns the date and time (down to seconds) in ISO 8601 format as described by
    http://www.w3.org/TR/NOTE-datetime-970915.html */
string getISOTime (time_t li){
  struct tm * t = localtime(&li);
  int year = 1900 + t->tm_year;
  string res = myItoa(year) + "-" + padWithZero(myItoa(t->tm_mon)) + "-" + padWithZero(myItoa(t->tm_mday)) +
    "T" + padWithZero(myItoa(t->tm_hour)) + ":" + padWithZero(myItoa(t->tm_min)) + ":" + padWithZero(myItoa(t->tm_sec)) +
    "+05:00";
  return res;
}


//Autofill the comment, add \n's so that no line
// is bigger than n characters, break only at spaces.
//
void autoFill(string & s, int n = 72){
  int p = 0;
  int last = 0;
  bool set = false;
  replaceAll(s, "\n", " ");
  replaceAll(s, "\t", " ");
  replaceAll(s, "\r", " ");
  s = "    " + s; //indent parragraph
  for (unsigned int i=0; i < s.size(); i++, p++){
    if ((p >= n) && set) {
      s[last] = '\n';
      p = 0;
      set = false;
    };
    if (s[i] == ' '){
      last = i;
      set = true;
    };
  }
  //  cout << "O:" << s << endl;
}

reference::reference(const reference &r) : url(r.url), title(r.title), dirname(r.dirname), 
  comment(r.comment), creationTime(r.creationTime), 
  modifiedTime(r.modifiedTime), visitTime(r.visitTime), aliasDir(r.aliasDir),
  aliasID(r.aliasID), aliasOf(r.aliasOf), children(0), priv(r.priv),
  aliasof(r.aliasof), navBar(r.navBar), folder(r.folder), hits(r.hits)
{
  if (r.children)
    children = new referenceTree(*(r.children));
  else
    children = 0;
}

reference & reference::operator=(const reference & r)
{
  url = r.url;
  title = r.title;
  dirname = r.dirname;
  comment = r.comment;
  creationTime = r.creationTime;
  modifiedTime = r.modifiedTime;
  visitTime = r.modifiedTime;
  aliasDir = r.aliasDir;
  aliasID = r.aliasID;
  aliasOf = r.aliasOf;
  priv = r.priv;
  aliasof = r.aliasof;
  navBar = r.navBar;
  folder = r.folder;
  hits = r.hits;
  delete children;
  if (r.children)
    children = new referenceTree(*(r.children));
  else
    children = 0;
  return *this;
}

// a < b iff a was created more recently than b
int reference::operator<(const reference & r) const
{
  return creationTime > r.creationTime;
}


reference::~reference() 
{ 
  if (!isAliasof())
    delete children; 
}

//given an HTML <H?></H?> or <A></A> it extracts the appropiate values and 
//  sets the reference to these values.
//REQUIRES that value be a properly formed <H> or <A>
void reference::setValues (const string & str){
  int l, r;
  //  cout << "setValues children=" << children << endl;
  if (str[1] == 'H') { //folder
    string addDateValue;
    addDateValue = getArgumentValue(str, addDate);
    creationTime = atoi(addDateValue.c_str());
    l = str.find(">");
    r = str.find("<",l);
    title = str.substr(l+1, r-l-1);
    return;
  }
  else if (str[1] == 'A') { //leaf (URL)
    url = getArgumentValue(str, href);
    aliasID = getArgumentValue(str, ALIASID);
    aliasOf = getArgumentValue(str, ALIASOF);
    if (aliasOf != "")
      aliasof = true;
    string addDateValue, lastModifiedValue, lastVisistValue;
    addDateValue = getArgumentValue(str, addDate);
    creationTime = atoi(addDateValue.c_str());
    lastModifiedValue = getArgumentValue(str, lastModified);
    modifiedTime = atoi(lastModifiedValue.c_str());
    lastVisistValue = getArgumentValue(str, lastVisit);
    visitTime = atoi(lastVisistValue.c_str());
    l = str.find(">");
    r = str.find("<",l);
    title = str.substr(l+1, r-l-1);
    return;
  }
}

int reference::search(const string & s)
  // returns the number of times s appears in 
  //    title and comment
  // also adds <B> </B> around matches
{
  int c = 0;;
  c += replaceAllNC(title, s, "<B>" + s + "</B>");
  c += replaceAllNC(comment, s, "<B>" + s + "</B>");
  return c;
}

bool reference::isPrivate() const
{
  return priv;
}

bool reference::isAliasof() const
{
  return aliasof;
}

bool reference::isFolder() const
{
  return folder;
}


// return length of first sentence until first occurance of p.
// skip one letter abbreviations like in names, for example "O. Obst"
string::size_type reference::fullstop(const string &s, const string &p) const 
{
  string::size_type i, end;
  i = 0;
  end = 0;
  string t;
  
  while (end == 0) {
    t = s.substr(i,s.length());
    end = t.find(p); // search for an occurance of p
    if (end != string::npos) {
      i = i + end + 1;
      if (end <= 1 ||
          t[end - 2] == '.' ||
          t[end - 2] == '!' ||
          t[end - 2] == '?' ||
          t[end - 2] == ' ' ||
          t[end - 2] == '\t' ||
          t[end - 2] == '\r' ||
          t[end - 2] == '\n') 
        end = 0;
    }
    else i = s.length();
  }
  return i;
}

//this is not used anymore (except for debugging)
void reference::sendAsHTML(ostream & out)
{
  string t = title;
  if (isFolder()){
    out << navBar << " <A HREF=\"" << "../" << url << "\">" 
        << t << "</A><BR>" << endl;
  }
  else{
    //    out << navBar << "<BR>" << endl
    out << "<A HREF=\"" << url << "\">" << t << "</A>";
    if (comment != "")
      out << " - " << comment;
    out << endl;
  }
}

/**Format the reference in html and return it.
 Make sure all the %VARS are replaced by their values.
 templ is is template. */
string reference::sendAsHTML(const string & templ, 
                             const string varValues []) const
{
  string outputString = templ;
  time_t currentTime = time(0); //.tv_sec;
  static int timeCutoff = atoi(varValues[TIMECUTOFF].c_str()) * 60 * 60 * 24;