//////////////////////////////////////////////////////////////////////
// IP traffic analyzer
// Written by Jonathan A. Foster <ChipMaster@YeOlPiShack.net>
// Started April 23rd, 2021
//
// The idea is to analyze iptables LOG entries in combination with
// DNSmasq's query log entries and combine them to list the hosts
// that were accessed. The main reasons for not just inspecting HTTP
// packets through a netfilter socket is due to HTTPS hiding the
// "host" field. So I'm deducing based on DNS query timing.
//
// NOTE: its assumed that the log being processed is in chronological
// order. This is the usual way things get logged. ;-)
//
// 2021-05-14 <ChipMaster@YeOlPiShack.net>
//   Dumbed down for < C++11.
//   Split into modules
//
// 2021-06-18 <ChipMaster@YeOlPiShack.net>
//   Wrapped in application object and added command line handling.
//   This includes:
//     - Getting config file name from CLI args
//     - Getting input and output filenams from CLI args
//     - Reading and writing from STDIN & STDOUT
//     - Send all non-data output to stderr
//////////////////////////////////////////////////////////////////////

// TODO: map names according to time and requesting host. time is probably automatic

#include <string.h>
#include <string>
#include <iostream>
#include <fstream>
#include <stdexcept>
#include <vector>
#include <map>
#include "cli.h"
#include "strutil.h"
#include "data.h"
#include "config.h"
using namespace std;



//////////////////////////////////////////////////////////////////////
// Application class to process files.
//////////////////////////////////////////////////////////////////////
//#define DEBUG

struct IPtraffic: public cBaseApp {
  Config      config;
  StringList  dns_ignore, dns_del;
  NameVal     rdns; // Reverse DNS lookup cache
  istream    *log;
  ostream    *out;
  LiveBug     bug;
  int         line_no;



  IPtraffic(): out(&cout), log(0)
  { // I'd rather this initialization be static...
    dns_ignore.push_back("v=spf1");
    dns_ignore.push_back("https:");
    dns_del.push_back("NODATA-");
    dns_del.push_back("NXDOMAIN-");
  }



  ~IPtraffic() {  if(log && log!=&cin) delete(log); }



  void dlog(const string msg) {
#ifdef DEBUG
    cerr << line_no << ": " << msg << endl;
#endif
  }


  // TODO: elaborate
  void help() {
    cerr <<
      "\n"
      "iptraffic -c {config file} [-o {output file}] [{input file} [...]]\n";
    ExitCode = 1;
  }



  unsigned do_switch(const char *sw) {
    if((sw[0]=='c' || sw[0]=='o') && sw[1]==0) return 1;
    throw CLIerror("Unrecognized Switch");
  }



  void do_switch_arg(const char *sw, const std::string &val) {
    switch(*sw) {
    case 'c':
      config.load(val);
      if(!config.us.size()) throw CLIerror(
        "The configuration files MUST contain an [us] section with "
        "appropriate values"
      );
      break;

    case 'o':
      if(out!=&cout)
        throw CLIerror("Output file has already been specified");
      out = new ofstream(val.c_str()); // c_str(), really?!?!
    }
  }



  void do_arg(const char *fname) {
    if(log && log!=&cin) delete(log);
    log = 0;
    log = new ifstream(fname);
    ExitCode = do_log();
  }



  // NOTE: the return values isn't really used yet but the channel is here if
  // it can be of use.
  int do_log() {
    if(!config.us.size()) throw CLIerror(
      "A configuration file must be specified before input files."
    );

    Splits ln;
    int ict=0;
    NameVal::iterator nvp;
    string name, address, s;
    Conn conn;
    bool match;

    /// parse log file ///

    line_no=0;
    while((*log >> ln)) {
      line_no++;
      cerr << bug << ' ' << line_no << '\r' << flush;

      ///  DNS query result ///

      // TODO: need to get more specific on tying us + them + time to DNS
      // TODO: doesn't seem that CNAMEs are getting attached to requests properly.
      //       the logs are cryptic on this front.
      if(ln.count>8 && strncmp(ln.fields[4], "dnsmasq[", 8)==0) {
        if(ln[5]=="reply" || ln[5]=="cached") {
          name = ln[6];
          address = ln[8];
          // Hmm... is this reply an address?
          if(pre_match(dns_ignore, address)) continue; // nope
          if(pre_match(dns_del,    address)) continue; // does not exist reply
          if((nvp=rdns.find(address))!=rdns.end()) {
            if(nvp->second==name) continue;
            dlog("WARN: DNS address overlap "+address+": "+nvp->second+" : "+name);
          }
          rdns[address] = name;
          dlog("Added "+address+" = "+name);
          continue;
        }
      }

      /// process connections ///

      if((ln.count>5  // old style
      && ln[4]=="kernel:"
      && ln[5]=="ACCEPT"
      ) || (ln.count>6  // new style
      && ln[4]=="vmunix:"
      && ln[6]=="ACCEPT")
      ) {
        conn = ln;
        conn.compact();
        if(!pre_match(config.us, conn.us)) conn.swap();
        if((nvp=rdns.find(conn.them))!=rdns.end())
          conn.name = nvp->second;
        if(config.ignores.find(conn)<0)
          *out << ln[0] << " " << ln[1] << " " << ln[2] << " " << conn << "\n";
        else
          ict++;
      }
    }
    *out << flush; // make sure all data gets written.
    cerr << "\nIgnored: " << ict << endl;
    cerr << "Total rDNS: " << rdns.size() << "\n";
    return 0;
  }



  int main() {
    try {
      cBaseApp::main();
      if(!log) {
        // no inputs were specified run stdin.
        log = &cin;
        ExitCode = do_log();
      }
    } catch(const CLIerror &e) {
      cerr << "ERROR: " << e.what() << "\n";
      help();
    }
    return ExitCode;
  }



};



//////////////////////////////////////////////////////////////////////
// Run it
//////////////////////////////////////////////////////////////////////

MAIN(IPtraffic)