// Generic HTML handler/parser
//

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <shad.c>
#include "urlencode.c"
#include "htmtag.c"

#define MAXSTRLEN 4096
#define HTMLSTRLEN MAXSTRLEN
char htmlstr[HTMLSTRLEN];
char*htmls=htmlstr;
char*absURL;

char linestr[MAXSTRLEN];
int linestrlen=0;

FILE*fo;

int istag=0;
int pref=0;
int rowlen=0,lastchr=' ';

int lastchr;

int form=0;
int desiredform=32767;
char desiredformname[]="postcomment";
int usesubmit=0;

int textarea=0;

int formnum=0;
char formname[256]="";
char selectname[256]="";

int isheader=0;
long inputline=0;


void xfputs(char*s)
{ char*sp;
  sp=s;while(sp[0]){if(sp[0]==0x02)sp[0]=' ';sp++;}
  fprintf(fo,"%s",s);
}

void flushtag()
{ htmls[0]=0;
  if(allowoutput())xfputs(htmlstr);
  htmls=htmlstr;
}

void normalizetag(char*s)
{ char*s1;
  int t;
  char c;
  int quot=0;
  for(t=0;t<strlen(s)-1;t++)
   {if(s[t+1]=='\n')s[t+1]=' ';
    if(quot)if(s[t]==' ')s[t]=0x02;
    if((s[t]==' ')||(s[t]==0x01))if(s[t+1]==' ')s[t+1]=0x01;
    if(s[t]=='"')if(quot)quot=0;else quot=1;
    if(!quot)
     if(!strncmp(s+t," =",2))s[t]=0x01;else
     if(!strncmp(s+t,"= ",2))s[t+1]=0x01;
    }
  s1=s;while(s[0]){if(s[0]!=0x01){s1[0]=s[0];s1++;}s++;}s1[0]=0;
}

int xstrstri(char*s1,char*s2)
{ char s[1024],*sp;
  strncpy(s,s1,1024);s[1023]=0;
  strupr(s2);strupr(s);
  sp=strstr(s,s2);if(!sp)return 0;
  return sp-s;
}

int bold=0,ital=0,undl=0,sup=0,sub=0;

int table=0,tablecnt;
int tr[32],td[32];

int allowoutput()
{ if(!textarea)return 0;
  if(!desiredform)return 1;
  if(desiredform==form)return 1;
  return 0;
}
int allowoutputtag()
{ if(!desiredform)return 1;
  if(desiredform==form)return 1;
  return 0;
}

void printtdaddr()
{ int t;
  fprintf(fo,"\n(%i:%i)",tablecnt,table);
  for(t=1;t<=table;t++)
   {fprintf(fo,"-%i:%i",tr[t],td[t]);}
  fprintf(fo," ");
}

void printurlstr(char*s)
{ while(s[0]){printurlencodedchar(s[0]);s++;}}

int handleforminputtag(char*s,int lf)
{ char s1[4096],s2[4096],s3[4096],s4[4096];
  s1[0]=0;s2[0]=0;s3[0]=0;s4[0]=0;
  gethtmltagvalue(s,"TYPE",s1);
  gethtmltagvalue(s,"NAME",s2);
  gethtmltagvalue(s,"VALUE",s3);
  strupr(s);s4[0]=0;if(strstr(s,"CHECKED"))strcpy(s4,"|CHECKED");
  printf("FORM:%i:%s|INPUT:%s|NAME:%s|VALUE:%s",formnum,formname,s1,s2,s3);
  if(lf)printf("\n");
  return 1;
}

int handleformselecttag(char*s)
{ char s1[4096],s2[4096],s3[4096],s4[4096];
  s1[0]=0;s2[0]=0;s3[0]=0;s4[0]=0;
  gethtmltagvalue(s,"NAME",s1);
  strcpy(selectname,s1);
  printf("FORM:%i:%s|INPUT:SELECT|NAME:%s\n",formnum,formname,s1);
  return 1;
}

int handleformoptiontag(char*s)
{ char s1[4096],s2[4096],s3[4096],s4[4096];
  s1[0]=0;s2[0]=0;s3[0]=0;s4[0]=0;
  gethtmltagvalue(s,"VALUE",s1);
  strupr(s);s2[0]=0;if(strstr(s,"SELECTED"))strcpy(s2,"|SELECTED");
  printf("FORM:%i:%s|INPUT:OPTION|NAME:%s|VALUE:%s%s\n",formnum,formname,selectname,s1,s2);
  return 1;
}

void handleformformtag(char*s)
{ char s1[4096],s2[4096],s3[4096],s4[4096];
  s1[0]=0;s2[0]=0;s3[0]=0;s4[0]=0;
  formname[0]=0;formnum++;
  gethtmltagvalue(s,"ACTION",s1);
  gethtmltagvalue(s,"NAME",s2);
  gethtmltagvalue(s,"METHOD",s3);
  strncpy(formname,s2,255);
  if(formnum>1)printf("\n");
  printf("FORM:%i:%s|METHOD:%s|ACTION:%s\n",formnum,formname,s3,s1);
}

void handlebasetag(char*s)
{ char s1[4096],s2[4096],s3[4096],s4[4096];
  s1[0]=0;s2[0]=0;s3[0]=0;s4[0]=0;
  formname[0]=0;formnum++;
  gethtmltagvalue(s,"HREF",s1);
  printf("BASE|HREF:%s\n",s1);
}

void handleformtag(char*s)
{ int n=0;
  if(!strncmpi(s,"<FORM ",6))handleformformtag(s);else
  if(!strncmpi(s,"<INPUT ",7))n=handleforminputtag(s,1);else
  if(!strncmpi(s,"<TEXTAREA ",10))handleforminputtag(s,0);else
  if(!strncmpi(s,"</TEXTAREA ",10)){n=1;printf("\n");}else
  {}
//  if(n)printf("&");
}

void handletag()
{ htmls[0]=0;istag=0;
//printf("%s\n",htmlstr);
  if(!strncmpi(htmlstr,"<BASE ",6))      {handlebasetag(htmlstr);}else
  if(!strncmpi(htmlstr,"<INPUT ",7))     {handleformtag(htmlstr);}else
  if(!strncmpi(htmlstr,"<SELECT",7))     {handleformselecttag(htmlstr);}else
  if(!strncmpi(htmlstr,"<OPTION",7))     {handleformoptiontag(htmlstr);}else
  if(textarea)if(allowoutput)if(strncmpi(htmlstr,"</TEXTAREA",10))
     {printurlstr(htmlstr);htmlstr[0]=0;flushtag();return;}
  if(!strncmpi(htmlstr,"</TEXTAREA",10)) {handleformtag(htmlstr);textarea=0;}else
  if(!strncmpi(htmlstr,"<FORM ",6))      {form++;handleformtag(htmlstr);}else
  if(!strncmpi(htmlstr,"<TEXTAREA",9))   {handleformtag(htmlstr);textarea=1;}else
  {}
  htmlstr[0]=0;flushtag();
}

void handlehtmlchar(int c)
{ htmls[0]=c;htmls++;
  if(c=='<')if(htmls>htmlstr+1){flushtag();return;}
  if(c=='!')if(!strncmp(htmlstr,"<!",2))
    {//flushtag();
     htmls=htmlstr;htmls[0]=0;istag=0;return;}
  if(c=='>'){handletag();htmls=htmlstr;htmls[0]=0;}
}

void handlelinestr()
{ 
//  return;
  linestr[linestrlen]=0;
  if(inputline<2)if(strncmp(linestr,"HTTP/",5))isheader=1;
  if(isheader)if(!linestr[0])isheader=0;
  if(isheader)if(!strncasecmp(linestr,"Set-Cookie:",11))
     {char*sp;
      printf("FULLCOOKIE:%s\n",linestr+12);
      sp=strstr(linestr,"; ");if(sp)sp[0]=0;
      printf("COOKIE:%s\n",linestr+12);
     }
  //printf("LINE:%s\n",linestr);
  linestrlen=0;
  inputline++;
}

void handlechar(int c)
{ 
  if(linestrlen<MAXSTRLEN-1)if(c>=' '){linestr[linestrlen]=c;linestrlen++;}
  //
  if(c=='\r')if(!textarea)return;
  if(c=='\n'){if(!textarea)c=' ';handlelinestr();}
  if(c=='<')istag=1;
  if(istag){handlehtmlchar(c);return;}
  if(c=='\n')if(!textarea)c=' ';
  if(textarea)printurlencodedchar(c);
  lastchr=c;
}

void convfile(char*fn,char*fno)
{ char s[256];
  FILE*fi;
  unsigned int c;
  if(!fn)fn="-";
  if(!strcmp(fn,"-"))fi=stdin;else fi=fopen(fn,"r");if(!fi){fprintf(stderr,"Cannot open file '%s'.\n",fn);return;}
  if(!fno)fo=stdout;else fo=fopen(fno,"w");if(!fo){fprintf(stderr,"Cannot write to file '%s'.\n",fno);return;}
//  fo=stdout;
  for(;;){c=fgetc(fi);if(feof(fi)){break;}
          handlechar(c);}
//  fputc(0x1a,fo);
  fclose(fo);
  fclose(fi);
}

void help()
{ printf(
"handleform - for easy decoding of web forms.\n"
"Usage: eg. curl <URL> | handleform | ...other filters \n"
);
}

int main(int argc,char*argv[])
{ if(argc<1)return;
  if(argc==1)convfile(NULL,NULL);else
  if((!strcmp(argv[1],"-h"))||(!strcmp(argv[1],"--help"))){help();return 0;}else
  if(argc==2)convfile(argv[1],NULL);
        else convfile(argv[1],argv[2]);
  return 0;
}