APG v2.1.0

bloom.c

@@ -1,5 +1,5 @@
 /*
-** Copyright (c) 2001
+** Copyright (c) 2001, 2002
 ** Adel I. Mirzazhanov. All rights reserved
 **
 ** Redistribution and use in source and binary forms, with or without
@@ -38,8 +38,8 @@
 **   open_filter    - open APG Bloom filter file
 **   get_filtersize - get APG Bloom filter size
 **   count_words    - count words in plain dictionary file
-***************************************************************************
-**   hash2bit       - generates 4 values (should be 4 values of independent
+**=============================================================
+**   hash2bit       - generates 5 values (should be 5 values of independent
 **                    hash functions) from input string.
 **   getbit         - get the bit value from file.
 **   putbit         - put the bit in the file.
@@ -57,7 +57,7 @@ int putbit(FILE * f, h_val bitnum);
 **    char *word        - word to incert in the filter
 **    FILE *file        - filter file descriptor
 **    h_val filter_size - filter size in bits
-** RETURN:
+** OUTPUT:
 **    int
 **      0 - everything OK
 **     -1 - something wrong
@@ -65,14 +65,14 @@ int putbit(FILE * f, h_val bitnum);
 int
 insert_word(char *word, FILE *file, h_val filter_size)
 {
-h_val h[4];
-int i = 0;
+ h_val h[H_NUM];
+ int i = 0;
 
-hash2bit (word, &h[0]);
-for(i = 0; i < 4; i++)
-  if (putbit (file, h[i] % filter_size)== -1)
-     return (-1);
-return(0);
+ hash2bit (word, &h[0]);
+ for(i = 0; i < H_NUM; i++)
+   if (putbit (file, h[i] % filter_size)== -1)
+      return (-1);
+ return(0);
 }
 
 /*
@@ -81,7 +81,7 @@ return(0);
 **    char *word        - word to check
 **    FILE *file        - filter file descriptor
 **    h_val filter_size - filter size in bits
-** RETURN:
+** OUTPUT:
 **    int
 **      0 - word is not in dictionary
 **      1 - word is in dictionary
@@ -90,11 +90,11 @@ return(0);
 int
 check_word(char *word, FILE *file,  h_val filter_size)
 {
- h_val h[4];
+ h_val h[H_NUM];
  int i = 0;
 
  hash2bit (word, &h[0]);
- for(i = 0; i < 4; i++)
+ for(i = 0; i < H_NUM; i++)
   {
    switch(getbit(file, h[i] % filter_size))
      {
@@ -116,24 +116,25 @@ check_word(char *word, FILE *file,  h_val filter_size)
 ** open filter file and check is this the real bloom filter file
 ** INPUT:
 **    char * f_name - filter filename
-** RETURN:
+**    const char *mode - "r" or "r+"
+** OUTPUT:
 **    FILE * - file pointer
 **    NULL   - something wrong. 
 */
 FILE *
-open_filter(char * f_name)
+open_filter(char * f_name, const char *mode)
 {
  FILE *f;
  struct apg_bf_hdr bf_hdr;
- if ((f = fopen (f_name, "r+")) == NULL)
+ if ((f = fopen (f_name, mode)) == NULL)
    return(NULL);
- if(fread ( (void *)&bf_hdr, APGBFHDRSIZE, 1, f) < APGBFHDRSIZE)
+ if (fread ( (void *)&bf_hdr, APGBFHDRSIZE, 1, f) < APGBFHDRSIZE)
     if (ferror (f) != 0)
        return(NULL);
- if( (bf_hdr.id[0] != 'A') || (bf_hdr.id[1] != 'P') ||
+ if ((bf_hdr.id[0] != 'A') || (bf_hdr.id[1] != 'P') ||
      (bf_hdr.id[2] != 'G') || (bf_hdr.id[3] != 'B') ||
      (bf_hdr.id[4] != 'F') || (bf_hdr.id[5] != '1') ||
-     (bf_hdr.id[6] != '0') || (bf_hdr.id[7] != '0') ) return (NULL);
+     (bf_hdr.id[6] != '0') || (bf_hdr.id[7] != '1') ) return (NULL);
  else
   {
    if (fseek (f, 0, SEEK_SET) == -1)
@@ -142,47 +143,64 @@ open_filter(char * f_name)
   }
 }
 
+/*
+** close_filter - close APG Bloom filter file
+** close filter file
+** INPUT:
+**    FILE * f_dsk - filter file pointer
+** OUTPUT:
+**    int - same as fclose() return value
+*/
+int
+close_filter(FILE *f_dsk)
+{
+ return(fclose(f_dsk));
+}
 
 /*
 ** get_filtersize - get APG Bloom filter size
 ** INPUT:
 **    FILE *f - filter file descriptor
-** RETURN:
-**    h_val - size of APG Bloom filter. 
+** OUTPUT:
+**    h_val - size of APG Bloom filter.
+**    0     - something wrong 
 */
 h_val
 get_filtersize(FILE * f)
 {
  struct apg_bf_hdr bf_hdr;
- fread ( (void *)&bf_hdr, APGBFHDRSIZE, 1, f);
- fseek (f, 0, SEEK_SET);
+ if (fread ( (void *)&bf_hdr, APGBFHDRSIZE, 1, f) < APGBFHDRSIZE)
+    if (ferror (f) != 0)
+       return(0);
+ if (fseek (f, 0, SEEK_SET) == -1)
+    return(0);
  return( (h_val)bf_hdr.fs);
 } 
 
 /*
 ** create_filter - create initial(empty) filter file 
-** 4 - number of hash functions
+** 5 - number of hash functions
 ** 0.0001 (0.01%) - probability of false positives
 ** INPUT:
 **    char * f_name - filter filename
 **    unsigned long int n_words - number of words in filter
-** RETURN:
+** OUTPUT:
 **    FILE * - filter file descriptor
 **    NULL   - something wrong
-***********************************************************
-** n - number of words in the filter
-** N - size of filter(?)
+** NOTES:
+**   n - number of words in the filter
+**   N - size of filter(?)
 **
-** a=(1-(4/N))^n
-** 0.0001=(1-a)^4 ==> 1-a=0.1 ==> a=0.9 ==>
-** 0.9=(1-(4/N))^n ==> 0.9^(1/n)=1-(4/N) ==>
+**   a=(1-(4/N))^n
+**   0.0001=(1-a)^5 ==> 1-a=0.15849... ==> a=0.84151068 ==>
+**   0.84151068=(1-(5/N))^n ==> 0.84151068^(1/n)=1-(5/N) ==>
 **
-** N=4/(1-[0.9^(1/n)])
+**   N=5/(1-[0.84151068^(1/n)])
 **
-**         4
-** N = ----------
-**            1/n
-**     1 - 0.9
+**               5
+**   N = -----------------
+**                     1/n
+**       1 - 0.84151068
 */
 FILE *
 create_filter(char * f_name, unsigned long int n_words) 
@@ -199,7 +217,7 @@ create_filter(char * f_name, unsigned long int n_words)
  bf_hdr.id[4] = 'F';
  bf_hdr.id[5] = '1';
  bf_hdr.id[6] = '0';
- bf_hdr.id[7] = '0';
+ bf_hdr.id[7] = '1';
  bf_hdr.fs = FSIZE_BIT(n_words);
 
  if ((f = fopen (f_name, "w+")) == NULL)
@@ -223,18 +241,25 @@ create_filter(char * f_name, unsigned long int n_words)
 ** count_words - count words in plain dictionary file
 ** INPUT:
 **    FILE *dict_file -plain dicionary file descriptor
-** RETURN:
+** OUTPUT:
 **    h_val - amount of words in dictionary file
+**    0     - something wrong
 */
 h_val
 count_words(FILE *dict_file)
 {
  h_val i = 0L; /* word counter */
  char *string; /* temp string holder */
- string = (char *) calloc(1,MAX_DICT_STRLEN);
+ char *tmp;    /* just tmp char pointer and nothing more it has no memory assigned */
+ if ((string = (char *) calloc(1,MAX_DICT_STRLEN)) == NULL)
+    return(0);
  while ((fgets(string, MAX_DICT_STRLEN, dict_file) != NULL))
-   i++;
- fseek (dict_file, 0, SEEK_SET);
+  {
+   tmp = (char *)strtok (string," \t\n\0");
+   if (tmp != NULL) i++;
+  }
+ if (fseek (dict_file, 0, SEEK_SET) == -1)
+    return (0);
  free ((void *) string);
  return (i);
 }
@@ -245,19 +270,19 @@ count_words(FILE *dict_file)
 ** INPUT:
 **    char *word - word to hash
 **    h_val *b   - pointer to bitnumber array
-** RETURN
+** OUTPUT:
 **    h_val * - pointer to bitnumber array
 */
 h_val *
 hash2bit(char * word, h_val *b)
 {
- struct apg_MD5Context context;
- unsigned char cs[16];
+ apg_SHA_INFO context;
+ BYTE cs[SHA_DIGESTSIZE];
 
- apg_MD5Init (&context);
- apg_MD5Update (&context, word, strlen(word));
- apg_MD5Final (cs, &context);
- return ( (h_val *)memcpy( (void *)b, (void *)&cs[0], 16));
+ apg_shaInit (&context);
+ apg_shaUpdate (&context, (BYTE *)word, strlen(word));
+ apg_shaFinal (&context, cs);
+ return ( (h_val *)memcpy( (void *)b, (void *)&cs[0], SHA_DIGESTSIZE));
 }
 
 /*
@@ -265,7 +290,7 @@ hash2bit(char * word, h_val *b)
 ** INPUT:
 **    FILE *f - file descriptor
 **    h_val bitnum - bit number
-** RETURN:
+** OUTPUT:
 **    int
 **      0,1 - bit value
 **     -1   - something wrong
@@ -297,7 +322,7 @@ getbit(FILE * f, h_val bitnum)
 ** INPUT:
 **    FILE *f - file descriptor
 **    h_val bitnum - bit number
-** RETURN:
+** OUTPUT:
 **    int
 **      0 - everything OK
 **     -1 - something wrong
@@ -327,4 +352,4 @@ putbit(FILE * f, h_val bitnum)
        return (-1);
  return (0);
 }
-/* END OF bloom.c file */
+/* END OF bloom.c file */