///////////////////
// skrewt.c
//
// scrutinize email
//

#include <iostream>
#include <stdlib.h>             /* for exit() */
#include <string>               /* for strcmp() */
#include <ctype.h>              /* toupper */
#include <signal.h>

#include <stdio.h>              /* perror */
#include <sstream>

using namespace std;

void usage(const int sts){
  (sts ? cerr : cout) <<
"Usage: skrewt [options]\n"
"\n"
"  Scrutinizes email.  Reads stdin, copies it to stdout.\n"
"  Exit result 0 means good, 21 means rejection (spam).\n"
"  Writes reason for rejection to stderr.\n"
"\n"
"  Typically used as a filter in a pipeline, along with spamc -E\n"
"  Options\n"
"    -help              print this msg (and exit immediately).\n"
"    -maxsize ii        msg size in bytes; anything bigger will be rejected.\n"
"\n"
"  Messages containing the string '-please-bounce-this-' will be rejected.\n"
"  Messages with no date will be rejected.\n"
;
  exit(sts);
}

//  error exit codes, mostly as stated in qmail.c
#define bar \
foo(good, 0) ;\
foo(spam, 21) ;\
foo(permerr, 31) ;\
foo(usage, 39) ;\
foo(greylisting, 70) ;\
foo(syserr, 71) ;\
foo(comerr, 74) ;

#define foo(name, num) const int ex_ ## name = num
bar
#undef foo


/////////////////////////////////////////////////////////
// Case insensitive comparison of strings

class lessthan_foldcase{
public:
  bool operator() (const std::string& a, const std::string& b) const {
    size_t a_len = a.length();
    size_t b_len = b.length();

    size_t lim = a_len < b_len ? a_len : b_len;

    for (size_t i=0; i<lim; ++i)
    {
            char cha = toupper(a[i]);
            char chb = toupper(b[i]);

            if (cha < chb) return true;
            if (cha > chb) return false;
    }
    // here if one is an extension of the other
    if ( a_len < b_len ) return true;
    return false;
  }
};


// Returns negative if a is less than b in alphabetical order
// returns 0 if they are the same, or positive if a is greater.
// Like perl cmp operator, but ignores case.
int cmp_casefold(const std::string& a, const std::string& b) {
  string::const_iterator aa, bb;
  aa = a.begin();
  bb = b.begin();
  while (aa != a.end() && bb != b.end()){
    char ca = tolower(*aa++);
    char cb = tolower(*bb++);
    if (ca != cb) return ca < cb ? -2 : 2;
  }
  if (aa != a.end()) return 1;          // a is longer
  if (bb != b.end()) return -1;         // b is longer
  return 0;
}


string toLower(const std::string& a){
  string rslt = a;
  string::iterator rr;
  for (rr = rslt.begin(); rr != rslt.end(); rr++){
    *rr = tolower(*rr);
  }
  return rslt;
}

////////////////
string ltrim(string foo){
  size_t where = foo.find_first_not_of(" \t\r\n");
  if (where == foo.npos) return foo;
  return foo.substr(where);
}

////////////////
// little utility to help with argument parsing:
//
int prefix(const string shorter, const string longer){
  return shorter == longer.substr(0, shorter.length());
}

void exeunt(const int sts){
  if (sts == ex_good) exit(sts);

  const char* foo = getenv("HI_Q_GROUP");
  if (!foo) exit(sts);

// No point in signalling ourself:
  sighandler_t rslt = signal(SIGUSR1, SIG_IGN);
  if (rslt == SIG_ERR) {
    cerr << "error setting signal" << endl;
  }
  int k = kill(-atoi(foo), SIGUSR1);
  if (k) {
    cerr << "kill failed on group " << atoi(foo) << " ... ";
    perror(0);
  }
  exit(sts);
}

string basename(const string path){
  size_t where = path.rfind("/");
  if (where != string::npos) return path.substr(1+where);
  return path;
}

string progname, progid;
int mypid;

////////////////////////////////////////////////////////////
int main(int _argc, const char** _argv){
////  pid_t pid = getpid();
////  cout << pid << endl;
////  cout << getpgid(pid) << endl;
  int argc(_argc);
  const char **argv(_argv);
  {
    progname = *argv++; argc--;
    mypid = getpid();
    stringstream binder;
    binder << basename(progname) << "[" << mypid << "]";
    progid = binder.str();
  }


  int maxsize(1000000);

  while (argc) {
    string arg(*argv); argv++; argc--;
    if (arg.substr(0,2) == "--") arg = arg.substr(1);
    if (prefix(arg, "-help")) {
      usage(0);
    }
    if (prefix(arg, "-maxsize")) {
      if (!argc) {
        cerr << "Option -maxsize requires an argument" << endl;
        exit(ex_usage);
      }
      maxsize = atoi(*argv); argv++; argc--;
    }
    if (arg.substr(0,1) == "-") {
      cerr << "Unrecognized option '" << arg << "'" << endl;
      cerr << "For help, try:  " << progname << " -help" << endl;
      exit(ex_usage);
    } else {
      cerr << "Extraneous verbiage '" << arg << "'" << endl;
      cerr << "For help, try:  " << progname << " -help" << endl;
      exit(ex_usage);
    }
  }

  int inheads(1);
  string boundary("x-xx-x");
  int text_type(1);
  int textlines(0);
  int gotdate(0);
  int msgsize(0);
  for (;;){
    if (cin.eof()) break;
    if (cin.bad()) return 1;
    if (inheads) {
      string header;
      if (getline(cin, header).fail()) continue;
      msgsize += header.length()+1;
      if (msgsize > maxsize) {
        cerr << progid << " rejection: bigger than " << maxsize << endl;
        exeunt(ex_spam);
      }
      for (;;) {
        if (cin.eof()) break;
        if (cin.bad()) return 1;
        char ch;
        if (cin.get(ch).fail()) continue;
        cin.putback(ch);
        if (ch != ' ' && ch != '\t') break;
        string line;
        if (getline(cin, line).fail()) continue;
        msgsize += line.length()+1;
        if (msgsize > maxsize) {
          cerr << progid << " rejection: bigger than " << maxsize << endl;
          exeunt(ex_spam);
        }
        header += "\n" + line;
      }
      int len = header.length();
      if (len && header[len-1] == '\r') len--;  // reduced length, not counting <cr>
      if (len == 0) {
        if (!gotdate) {
          cerr << progid << " rejection: no date" << endl;
          exeunt(ex_spam);              // disallow mail with no date
        }
        inheads = 0;
        //cerr << "end of headers" << endl;
      }
      else {
// here if it's a header line
        string headword;
        string rest;
        size_t where = header.find(":");
        if (where != string::npos) {
          headword = header.substr(0, where);
          rest = ltrim(header.substr(1+where));
        }
        headword = toLower(headword);
        if (headword == "content-type") {
          string the_type = rest;
          size_t where = the_type.find_first_of(" \t;\n");
          if (where == string::npos) {
            /* do nothing */
          }
          else {
            the_type = the_type.substr(0,where);
          }
          the_type = toLower(the_type);
          text_type = (the_type.find("text/") == 0);
          if (0) cerr << "type (" << the_type << ")  "
                << (text_type?"text":"nope") << endl;
          string srch = "boundary=";
          where = header.find(srch);
          if (where != string::npos) {
            where += srch.length();
            boundary = header.substr(where);
            if (boundary[0] == '"') {
              boundary = boundary.substr(1);
              where = boundary.find_first_of("\"");
            } else {
              where = boundary.find_first_of(" \t;\n");
            }
            if (where == string::npos) {
              /* do nothing, boundary=boundary as a whole */
            } else {
              boundary = boundary.substr(0, where);
            }
          }
        } else if (headword == "date") {
          gotdate++;
        } else if (headword == "subject") {
          if (rest.find("-please-bounce-this-") != string::npos) {
            cerr << progid << " rejection: by request" << endl;
            exeunt(ex_spam);
          }
        }
      }
      //xxxx  cout << header.length() << " ... ";
      cout << header << endl;
    } else {
      string line;
      if (!getline(cin, line).fail()) {
        msgsize += line.length()+1;
        if (msgsize > maxsize) {
          cerr << progid << " rejection: bigger than " << maxsize << endl;
          exeunt(ex_spam);
        }
        if (line == "--" + boundary) {
          inheads = 1;
        } else {
          if (text_type) {
            if (ltrim(line).length()) textlines++;
          }
        }
        cout << line << endl;
      }
    }
  }
  if (0) cerr << "textlines: " << textlines << endl;
  if (!textlines) {
    cerr << progid << " rejection: no text" << endl;
    exeunt(ex_spam);
  }
  cerr << progid << " normal completion" << endl;
  exit(ex_good);
}