//  Program:  striphtm.c
//  Purpose:  remove tags from .htm and .html files
//  Author:   Scott Wenger, 2000
//            Box 802, Stevens Point, WI 54481
//            panther@wctc.net
//            http://www.private-files.com/sware.html


//  The program accepts the html (or htm) file name on the command line.
//  After removing the tags, the results are stored in file "clean.txt"

//  This source code is profusely commented.
//  The malloc function is used to store input file in memory.
//  Use whatever include file your compliler requires for malloc.  


#include <stdio.h>
#include <stdlib.h>
 

// function prototype
void striphtm(char *buffer);


void striphtm(char *buffer)
{
  static int intag = 0;                // flag is 1 if inside html tag
  char *source, *dest;                 // declare two pointers to char  
  source = dest = buffer;              // point to start of buffer
    
  while (*source != '\0') {            // while html in memory

    if (intag) {                       // if inside an html tag
      if (*source == '>') intag = 0;   // if end of tag, turn off flag
      source++;                        // advance ptr to ignore tag contents
    }
    else {                             
      if (*source == '<') intag = 1;   // if start of tag, set flag to 1
      else *dest++ = *source;          // else keep text outside tag
      source++;                        // point to next byte in memory
    }
  }                                    // continue processing html file
  
  *dest = '\0';                        // null-terminate the results
}



main()
{
  FILE *fp, *ofp;
  long flen = 0L;      
  int ch;
  char *mem, *start;
  char fname[500];

  printf("\nstriphtm (c) 2000 Scott Wenger");
  printf("\nremoves tags from html files.\n");
  printf("\nName of html file: ");
  fflush(stdout);  
  scanf("%s", fname);    
  fp = fopen(fname, "r");

  if ((fp = fopen(fname, "r")) == NULL) {
    fprintf(stderr, "\nFile not found !");
    fprintf(stderr, "\nCheck spelling and add file path, if necessary.\n\n");
    fflush(stdout);
    exit(1);
  }

  if ((ofp = fopen("clean.txt", "w")) == NULL) {
    fprintf(stderr, "\nUnable to open output file for writing.\n\n");
    fflush(stdout);
    fclose(fp);
    exit(1);
  }
  
  while (fgetc(fp) != EOF) flen++; // primitive yet portable file length   
  rewind(fp);                    
  
  if ( (mem = (char *) malloc(flen + 1L)) == NULL) {
    fprintf(stderr, "\nUnable to allocate required memory.\n\n");
    fclose(fp);
    fclose(ofp);
    exit(1);
  }	          
  start = mem;                 // start points to beginning of memory
  
  while ( (*mem = fgetc(fp)) != EOF) mem++;  // put htm file into memory
  mem = start;	                            
  striphtm(mem);                             // remove the tags

  while (ch = *mem++) fputc(ch, ofp);        // write to output file

  printf("Done.  The results were written to file clean.txt\n\n");
  fclose(fp);
  fclose(ofp);
  free(mem);
  exit(0);
}