/*
* query_by_field --
* Query the primary database by field.
*/
static int
query_by_field(char *input)
{
OPERATOR operator;
size_t len;
char *field, *op, *value;
/*
* We expect to see "field [op] value" -- figure it out.
*
* Skip leading whitespace.
*/
while (isspace(*input))
++input;
/*
* Find an operator, and it better not start the string.
*/
if ((len = strcspn(field = input, "<>!=~")) == 0)
return (query_usage());
op = field + len;
/* Figure out the operator, and find the start of the value. */
switch (op[0]) {
case '~':
operator = WC;
value = op + 1;
break;
case '!':
if (op[1] == '=') {
operator = NEQ;
value = op + 2;
break;
}
if (op[1] == '~') {
operator = NWC;
value = op + 2;
break;
}
return (query_usage());
case '<':
if (op[1] == '=') {
operator = LTEQ;
value = op + 2;
} else {
operator = LT;
value = op + 1;
}
break;
case '=':
operator = EQ;
if (op[1] == '=')
value = op + 2;
else
value = op + 1;
break;
case '>':
if (op[1] == '=') {
operator = GTEQ;
value = op + 2;
} else {
operator = GT;
value = op + 1;
}
break;
default:
return (query_usage());
}
/* Terminate the field name, and there better be a field name. */
while (--op > input && isspace(*op))
;
if (op == input)
return (query_usage());
op[1] = '\0';
/* Make sure there is a value field. */
while (isspace(*value))
++value;
if (*value == '\0')
return (query_usage());
return (DbRecord_search_field_name(field, value, operator));
}
int bq_main (int argc, char **argv)
{
if (argc < 3)
{
return query_usage();
}
/*-------defaults for bloom filter building-------*/
int opt;
double tole_rate = 0, sampling_rate = 1;
char *ref = NULL, *list = NULL, *target_path = NULL, *source = NULL, *report_fmt = "json";
// XXX: make r and l mutually exclusive
while ((opt = getopt (argc, argv, "s:t:r:o:q:f:h")) != -1)
{
switch (opt)
{
case 't':
tole_rate = atof(optarg);
break;
case 's':
sampling_rate = atof(optarg);
// Sampling rate is the partial proportion of a sample, or subsampling, i.e: 0.20 means take only 20% of the input file.
break;
case 'o':
target_path = optarg;
break;
case 'q':
source = optarg;
break;
case 'r':
ref = optarg;
break;
case 'f': // "json", "tsv" or none
(optarg) && (report_fmt = optarg, 1);
break;
case 'h':
return query_usage();
break;
case '?':
printf ("Unknown option: -%c\n", (char) optopt);
return query_usage();
break;
}
}
if (!target_path && !source)
{
fprintf (stderr, "\nPlease, at least specify a bloom filter (-r) and a query file (-q)\n");
exit (-1);
}
/*
check the format for reference bloom filter by the extension name so far, can write special tag in the filter for format checking in the future.
*/
if (strstr(ref,".bloom") == NULL)
{
fprintf (stderr,"\nIncorrect bloom filter\n");
exit(-1);
}
if (target_path == NULL)
{
target_path = argv[0];
} //set default path, which is where the binary file is.
char *result = query(source, ref, tole_rate, sampling_rate, list, target_path, report_fmt, 'c');
printf("%s\n",result);
return 1;
}
