miller/c/mapping/mapper_head.c
2015-05-03 16:11:45 -07:00

102 lines
3.4 KiB
C

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <math.h>
#include "lib/mlrutil.h"
#include "containers/sllv.h"
#include "containers/lhmslv.h"
#include "containers/lhmsv.h"
#include "containers/mixutil.h"
#include "mapping/mappers.h"
#include "cli/argparse.h"
// ================================================================
typedef struct _mapper_head_state_t {
slls_t* pgroup_by_field_names;
unsigned long long head_count;
lhmslv_t* precord_lists_by_group;
} mapper_head_state_t;
// ----------------------------------------------------------------
// xxx if empty key then make a way to communicate back to the reader that it
// can stop reading further records -- ?
sllv_t* mapper_head_func(lrec_t* pinrec, context_t* pctx, void* pvstate) {
mapper_head_state_t* pstate = pvstate;
if (pinrec != NULL) {
slls_t* pgroup_by_field_values = mlr_selected_values_from_record(pinrec, pstate->pgroup_by_field_names);
unsigned long long* pcount_for_group = lhmslv_get(pstate->precord_lists_by_group, pgroup_by_field_values);
if (pcount_for_group == NULL) {
pcount_for_group = mlr_malloc_or_die(sizeof(unsigned long long));
*pcount_for_group = 0LL;
lhmslv_put(pstate->precord_lists_by_group, slls_copy(pgroup_by_field_values), pcount_for_group);
}
(*pcount_for_group)++;
if (*pcount_for_group <= pstate->head_count) {
return sllv_single(pinrec);
} else {
lrec_free(pinrec);
return NULL;
}
}
else {
return sllv_single(NULL);
}
}
// ----------------------------------------------------------------
static void mapper_head_free(void* pvstate) {
mapper_head_state_t* pstate = (mapper_head_state_t*)pvstate;
if (pstate->pgroup_by_field_names != NULL)
slls_free(pstate->pgroup_by_field_names);
if (pstate->precord_lists_by_group != NULL)
// xxx recursively free void-stars ... here & elsewhere.
lhmslv_free(pstate->precord_lists_by_group);
}
mapper_t* mapper_head_alloc(slls_t* pgroup_by_field_names, unsigned long long head_count) {
mapper_t* pmapper = mlr_malloc_or_die(sizeof(mapper_t));
mapper_head_state_t* pstate = mlr_malloc_or_die(sizeof(mapper_head_state_t));
pstate->pgroup_by_field_names = pgroup_by_field_names;
pstate->head_count = head_count;
pstate->precord_lists_by_group = lhmslv_alloc();
pmapper->pvstate = pstate;
pmapper->pmapper_process_func = mapper_head_func;
pmapper->pmapper_free_func = mapper_head_free;
return pmapper;
}
// ----------------------------------------------------------------
void mapper_head_usage(char* argv0, char* verb) {
fprintf(stdout, "Usage: %s %s [options]\n", argv0, verb);
fprintf(stdout, "-n {count} Head count to print; default 10\n");
fprintf(stdout, "-g {a,b,c} Group-by-field names for head counts\n");
}
mapper_t* mapper_head_parse_cli(int* pargi, int argc, char** argv) {
int head_count = 10;
slls_t* pgroup_by_field_names = slls_alloc();
char* verb = argv[(*pargi)++];
ap_state_t* pstate = ap_alloc();
ap_define_int_flag(pstate, "-n", &head_count);
ap_define_string_list_flag(pstate, "-g", &pgroup_by_field_names);
if (!ap_parse(pstate, verb, pargi, argc, argv)) {
mapper_head_usage(argv[0], verb);
return NULL;
}
return mapper_head_alloc(pgroup_by_field_names, head_count);
}
// ----------------------------------------------------------------
mapper_setup_t mapper_head_setup = {
.verb = "head",
.pusage_func = mapper_head_usage,
.pparse_func = mapper_head_parse_cli,
};