level0

main.c

#include "bloom.h"
#include <stdio.h>
#include <stdlib.h>
#include <assert.h>
#ifndef DEBUG
#include "filter.h"
#include "filter_bytes.h"
#endif

#define BUFFER_SIZE 8192
#define LINE_BUFFER_SIZE 1024
#define WORD_BUFFER_SIZE 1024

struct bloom * constructBloomFilter(char * file){
  FILE * f = fopen(file, "r");
  char buffer[BUFFER_SIZE];

  size_t amt;
  int last_position = 0;
  int i = 0;
  struct bloom * bfilter = malloc(sizeof(struct bloom));
  // we'll just assume this is acceptable chance of error
  if(bloom_init(bfilter, 234937, 0.00001) != 0){
    printf("Couldn't initialize filter\n");
    exit(1);
  }
  unsigned int count = 0;
  while(!feof(f)){
    fseek(f, last_position, SEEK_SET);
    amt = fread(buffer, 1, BUFFER_SIZE, f);
    if(amt > 0){
      int last_word = 0;
      for(i = 0; i < amt; i++){
        if(buffer[i] == '\n'){
          count ++;
          if(bloom_add(bfilter, &buffer[last_word], i - last_word) == -1){
            printf("Couldn't create bloom filter\n");
            exit(1);
          }
          last_position += (i - last_word) + 1;
          last_word = i + 1;
        }
      }
    }
  }
  fclose(f);
  return bfilter;
}

int main(int argc, const char ** argv){
  struct bloom * filter = NULL;
  #ifdef DEBUG
  if(argc == 2){        // file specified
    filter = constructBloomFilter(argv[1]);
  } else if(argc == 3){ // file and bloom filter output specified
    filter = constructBloomFilter(argv[1]);
  } else {              // use default
    printf("Default\n");
  }
  #else
  filter = (struct bloom *) __struct_dat;
  filter->bf = __struct_bytes_dat;
  if(filter->bytes != __struct_bytes_dat_len){
    printf("Precompiled filter not same size");
    exit(1);
  }
  #endif
  char buffer[LINE_BUFFER_SIZE];
  char wordBuffer[WORD_BUFFER_SIZE];
  char lowerWordBuffer[WORD_BUFFER_SIZE];
  size_t readAmt = 0;
  int wordLen = 0;
  int i = 0;
  int res = 0;
  while( (readAmt = fread(buffer, 1, WORD_BUFFER_SIZE, stdin)) > 0){
    for(i = 0; i < readAmt; i++){
      if(buffer[i] == '\n' || buffer[i] == ' '){
        if(wordLen > 0){
          res = bloom_check(filter, lowerWordBuffer, wordLen);
          if(res == 0){
            fwrite("<", 1, 1, stdout);
            fwrite(wordBuffer, 1, wordLen, stdout);
            fwrite(">", 1, 1, stdout);
          } else if(res == 1){
            fwrite(wordBuffer, 1, wordLen, stdout);
          } else {
            printf("Bloom not initialized\n");
            exit(1);
          }
        }
        fwrite(&buffer[i], 1, 1, stdout);
        wordLen = 0;
      } else {
        wordBuffer[wordLen] = buffer[i];
        lowerWordBuffer[wordLen] = tolower(buffer[i]);
        wordLen ++;
      }
    }
  }
  return 0;
}

filter and filter_bytes are header files w/ struct memory inline