path: root/tools/mail-scan.c

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

// Hint:
//     mail-scan +from * | iconv -c \...
//  | sed 's/.*@//;s/>$//' | sort | uniq -c | sort -nr > some-junk.from-count
//
// Also:
//   grep score=[34] /home/user/Maildir/new/* -l | xargs mail-scan +From | blacklist-update
// Then:
//   grep score=[34] /home/user/Maildir/new/* -l | xargs mv-to -i /home/user/Maildir/spam/

/////////////
// Another hint:  using the "-addr" feature:
//    mail-scan +x-spam.*:score=[234] /home/jean/Maildir/spam/* -l | \....
//      xargs mail-scan +from -addr | sort | uniq -c | sort -nr | head -20

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

#include <sys/types.h>          /*  for stat() */
#include <sys/stat.h>           /* stat() */

#include <stdio.h>              /* perror */
#include <boost/regex.hpp>
#include "utils.h"

using namespace std;

void usage(const int sts){
  (sts ? cerr : cout) <<
"Usage: mail-scan [options] filename [more filenames]\n"
"\n"
"  Options\n"
"    -help              print this msg (and exit immediately).\n"
"    -vert              invert:  print only if *no* match.\n"
"    -l                 print filename only, not matching text.\n"
"    -group             print a blank line after every match.\n"
"    -multi             print multi-line records on multiple lines\n"
"                       (as opposed to smashing them all onto one long line\n"
"    -addr              assume field contains somebody <foo@bar.com>;  print just foo@bar.com\n"
"\n"
"\n"
"  Messages containing the string '-please-bounce-this-' will be rejected.\n"
"  Messages with no date will be rejected.\n"
;
  exit(sts);
}

// exit codes, compatible with spamassassin (not with qmail-queue)
const int sa_good(0);
const int sa_spam(1);
const int sa_usage(64);

/////////////////////////////////////////////////////////
// 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;
}

void exeunt(const int sts){
  if (sts == sa_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 join(const string sep, list<string> stuff){
  string rslt;
  if (!stuff.size()) return rslt;
  list<string>::const_iterator ptr = stuff.begin();
  rslt = *ptr++;
  for (; ptr != stuff.end(); ptr++){
    rslt += sep;
    rslt += *ptr;
  }
  return rslt;
}

class watcher {
public:
  string key;
  string val;
  boost::regex valrx;
  boost::regex keyrx;

  watcher(const string init)
  : key(""), val("")
  {
    size_t where = init.find(":");
    if (where != string::npos) {
      key = init.substr(0, where);
      val = init.substr(1+where);
    } else {
      key = init;
    }
    keyrx = boost::regex(key, boost::regex_constants::icase);
    valrx = boost::regex(val, boost::regex_constants::icase);
    //xx cerr << "watcher key: " << key << "  val: " << val << endl;
  }
};

string noCR(const string bar){
  string foo(bar);
  int len = foo.length();
  if (len){
    if (foo[len-1] == '\r') {
      foo.erase(len-1);
    }
  }
  return foo;
}

////////////////////////////////////////////////////////////
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);
  string progname(*argv); argv++; argc--;

  int maxsize(1000000);
  list<watcher> watchword;
  list<string> dofile;
  int vflag(0);
  int group_flag(0);
  int multi(0);
  int fname_only(0);
  int addr_mode(0);
  boost::regex addr_filter(string("<.*@(.*)>"), boost::regex_constants::icase);

  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, "-vert" /* short */)){
      vflag++;
      continue;
    } if (prefix(arg, "-group" /* short */)){
      group_flag++;
      continue;
    } else if (prefix(arg, "-l" /* short */)){
      fname_only++;
      continue;
    } else if (prefix(arg, "-address" /* long */)){
      addr_mode++;
      continue;
    } else if (prefix(arg, "-multi" /* long */)){
      multi++;
      continue;
    } else if (arg.substr(0,1) == "-") {
      cerr << "Unrecognized option '" << arg << "'" << endl;
      cerr << "For help, try:  " << progname << " -help" << endl;
      exit(sa_usage);
    }
    if (arg.substr(0,1) == "+") {
      watchword.push_back(arg.substr(1));
    } else {
      dofile.push_back(arg);
    }
  }

// loop over all files
  for (list<string>::const_iterator file = dofile.begin();
       file != dofile.end(); file++) {
    int didprint(0);
    struct stat filestatus;
    stat(file->c_str(), &filestatus );
    if (S_ISDIR(filestatus.st_mode)) {
      cerr << "is directory: " << *file << endl;
      continue;
    }

    ifstream infile;
    infile.open(file->c_str());
    if (!infile.good()) {
      cerr << "Failed to open file: " << *file << endl;
      exit(1);
    }
    int inheads(1);
    string boundary("x-xx-x");
    int msgsize(0);
    int foundsome_infile(0);
    for (;;){           // loop over all records in this file
      if (inheads) {
        list<string> Header;
        string line;
        for (;;) {      // loop over all lines in this record
          if (infile.eof()) break;
          if (infile.bad()) {
            cerr << "mail-scan: read error on file '" << *file << "'" << endl;
            return 1;
          }
          if (getline(infile, line).fail()) continue;
          line = noCR(line);
          Header.push_back(line);
          msgsize += line.length()+1;
          if (msgsize > maxsize) {
            cerr << "skrewt rejection: bigger than " << maxsize << endl;
            exeunt(sa_spam);
          }
          char ch;
          if (infile.get(ch).fail()) continue;
          infile.putback(ch);
          if (ch != ' ' && ch != '\t') break;
        }
        if (Header.front().length() == 0) {
          inheads = 0;
          continue;     // blank line needs no further processing
        }
        string headword;      // the first thing on the line, e.g. "Subject"
        string rest;
        string header;
        if (!multi) header = join(" ", Header);
        else        header = join("\n", Header);
        size_t where = header.find(":");
        if (where != string::npos) {
          headword = header.substr(0, where);
          rest = ltrim(header.substr(1+where));
        }

        for (list<watcher>::const_iterator ptr = watchword.begin();
          ptr != watchword.end(); ptr++) {
// regex_match not regex_search ... keyrx must match *whole* headword
          if (boost::regex_match(headword, ptr->keyrx)){
            // here if match as to keyword;  check for match as to value
            if (ptr->val.length()==0
               || boost::regex_search(rest, ptr->valrx)){
              foundsome_infile++;
              if (!vflag) {
                if (!addr_mode){
                  // << foundsome_infile  << " " ;  (number of occurrences)
                  cout << *file;
                  if (!fname_only) {
                     cout << " :: " << header;
                  }
                  cout << endl;
                  didprint++;
                } else /* addr_mode */{
                     boost::smatch matches;
                     if (boost::regex_search(header, matches, addr_filter)){
                       cout << string(matches[1].first, matches[1].second) << endl;
                     } else {
                       cerr << "no match, ignoring: " << header << endl;
                     }
                    didprint++;
                }
              }
            }
          }
        }
// only show file once, even if there might have been multiple matches:
        if (fname_only && didprint) break;
      } else {
        // not in header
        break;
      }
    }   // end loop over matching records in this file

    if (vflag && !foundsome_infile) {
      cout << *file << endl;
      didprint++;
    }
    if (group_flag && didprint) cout << endl;
  }
}