/*
   crypt analyser
   - uses tetragrams in tetra.txt (created using stats prog)
   - crypt in file should begin with method to use
   analyse [file]
*/

#include <stdio.h>
#include <stdlib.h>
#include <process.h>
#include <string.h>
#include <time.h>
#include <math.h>

int tetfreq[26][26][26][26],ctype,bpt;
char buffer[100000];

#define CRYPT_SUBSTITUTION 1

void rndstr(char *d)
{ int i,j,k;
  char n;
  strcpy(d,"abcdefghijklmnopqrstuvwxyz");
  i=(rand()%1000)+100;
  while(i--)
  { j=rand()%26;
    k=rand()%26;
    if(j!=k)
    { n=d[j];
      d[j]=d[k];
      d[k]=n;
    }
  }
}

void smalladj(char *d)
{ int i,j,k;
  char n;
  i=(rand()%5)+5;
  while(i--)
  { j=rand()%26;
    k=rand()%26;
    if(j!=k)
    { n=d[j];
      d[j]=d[k];
      d[k]=n;
    }
  }
}

/* higher=better */
double score(char *d)
{ int i;
  double sc=0;
  for(i=0;i<bpt-3;i++)
    sc=sc+log(tetfreq[d[buffer[i]-'a']-'a'][d[buffer[i+1]-'a']-'a']
                 [d[buffer[i+2]-'a']-'a'][d[buffer[i+3]-'a']-'a']+1)*0.25;
  return sc;
}

void main(int argc, char *argv[])
{ FILE *tet,*crypt;
  int d,foundhint,i,j,besti,bestj,smallperturb;
  char s[20],chint[50],c,dec[26],testd[26];
  time_t t;
  double bestscore,sc,bestadj,bestoverall;
  srand(time(&t));
  tet=fopen("tetra.txt","rt");
  while(1)
  { if(fscanf(tet,"%s",s)!=1)
      break;
    fscanf(tet,"%d",&d);
    tetfreq[s[0]-'a'][s[1]-'a'][s[2]-'a'][s[3]-'a']=d;
  }
  fclose(tet);
  if(argc<2)
  { printf("No file specified ?\n");
    exit(0);
  }
  crypt=fopen(argv[1],"rt");
  fgets(buffer,10000,crypt);
  sscanf(buffer,"%s",chint);
  ctype=CRYPT_SUBSTITUTION;
  foundhint=0;
  if(!stricmp(chint,"substitution"))
  { foundhint=1;
    ctype=CRYPT_SUBSTITUTION;
  }
  bpt=0;
  if(!foundhint)
  { for(i=0;buffer[i];i++)
      if((buffer[i]>='a')&&(buffer[i]<='z'))
        buffer[bpt++]=buffer[i];
      else if((buffer[i]>='A')&&(buffer[i]<='Z'))
        buffer[bpt++]=buffer[i]-'A'+'a';
  }
  while(!feof(crypt))
  { if(fscanf(crypt,"%c",&c)!=1)
      break;
    if((c>='a')&&(c<='z'))
      buffer[bpt++]=c;
    else if((buffer[i]>='A')&&(buffer[i]<='Z'))
      buffer[bpt++]=c-'A'+'a';
  }
  buffer[bpt]=0; //strrev(buffer);
  printf("ho hum %s\n",buffer);
  printf("Analysing crypt:\n%.2lf %s\n",score("abcdefghijklmnopqrstuvwxyz"),buffer);
  rndstr(dec);
  bestscore=score(dec);
  bestoverall=bestscore-1;
  smallperturb=1;
  while(1)
  { bestadj=bestscore;
    for(i=0;i<26;i++)
      for(j=i+1;j<26;j++)
      { strcpy(testd,dec);
        c=testd[i];
        testd[i]=testd[j];
        testd[j]=c;
        sc=score(testd);
        if(sc>bestadj)
        { bestadj=sc;
          besti=i;
          bestj=j;
        }
      }
    if(bestadj>bestscore)
    { c=dec[besti];
      dec[besti]=dec[bestj];
      dec[bestj]=c;
      bestscore=bestadj;
      if(bestscore>bestoverall)
      { bestoverall=bestscore;
        smallperturb=1;
        printf("\n%.2lf %s\n",bestoverall,dec);
        for(i=0;i<bpt;i++)
          printf("%c",dec[buffer[i]-'a']);
        printf("\n");
      }
    }
    else
    { if(smallperturb)
      { smalladj(dec);
        smallperturb=0;
      }
      else
      { rndstr(dec);
        smallperturb=1;
      }
      bestscore=score(dec);
    }
  }
}