Imported Upstream version 0.1.5

[tabix.git] / index.c

diff --git a/index.c b/index.c
index 704b037d2add76bf0c9d02624c3f449ac2bd314b..cfb2938ee4ee9d490cd70f5635608b8f21abe291 100644 (file)
--- a/index.c
+++ b/index.c
@@ -1,5 +1,6 @@
 #include <ctype.h>
 #include <assert.h>
+#include <sys/stat.h>
 #include "khash.h"
 #include "ksort.h"
 #include "kstring.h"
@@ -161,6 +162,9 @@ static int get_intv(ti_index_t *idx, kstring_t *str, ti_intv_t *intv)
                                // here ->beg is 0-based.
                                intv->beg = intv->end = strtol(str->s + b, &s, 0);
                                if (!(idx->conf.preset&TI_FLAG_UCSC)) --intv->beg;
+                               else ++intv->end;
+                               if (intv->beg < 0) intv->beg = 0;
+                               if (intv->end < 1) intv->end = 1;
                        } else {
                                if ((idx->conf.preset&0xffff) == TI_PRESET_GENERIC) {
                                        if (id == idx->conf.ec) intv->end = strtol(str->s + b, &s, 0);
@@ -174,6 +178,7 @@ static int get_intv(ti_index_t *idx, kstring_t *str, ti_intv_t *intv)
                                                        if (op == 'M' || op == 'D' || op == 'N') l += x;
                                                        s = t + 1;
                                                }
+                                               if (l == 0) l = 1;
                                                intv->end = intv->beg + l;
                                        }
                                } else if ((idx->conf.preset&0xffff) == TI_PRESET_VCF) {
@@ -262,6 +267,17 @@ static void merge_chunks(ti_index_t *idx)
        } // ~for(i)
 }
 
+static void fill_missing(ti_index_t *idx)
+{
+       int i, j;
+       for (i = 0; i < idx->n; ++i) {
+               ti_lidx_t *idx2 = &idx->index2[i];
+               for (j = 1; j < idx2->n; ++j)
+                       if (idx2->offset[j] == 0)
+                               idx2->offset[j] = idx2->offset[j-1];
+       }
+}
+
 ti_index_t *ti_index_core(BGZF *fp, const ti_conf_t *conf)
 {
        int ret;
@@ -316,6 +332,7 @@ ti_index_t *ti_index_core(BGZF *fp, const ti_conf_t *conf)
        }
        if (save_tid >= 0) insert_offset(idx->index[save_tid], save_bin, save_off, bgzf_tell(fp));
        merge_chunks(idx);
+       fill_missing(idx);
 
        free(str->s); free(str);
        return idx;
@@ -523,22 +540,10 @@ static ti_index_t *ti_index_load_core(BGZF *fp)
        return idx;
 }
 
-ti_index_t *ti_index_load_local(const char *_fn)
+ti_index_t *ti_index_load_local(const char *fnidx)
 {
        BGZF *fp;
-       char *fnidx, *fn;
-
-       if (strstr(_fn, "ftp://") == _fn || strstr(_fn, "http://") == _fn) {
-               const char *p;
-               int l = strlen(_fn);
-               for (p = _fn + l - 1; p >= _fn; --p)
-                       if (*p == '/') break;
-               fn = strdup(p + 1);
-       } else fn = strdup(_fn);
-       fnidx = (char*)calloc(strlen(fn) + 5, 1);
-       strcpy(fnidx, fn); strcat(fnidx, ".tbi");
        fp = bgzf_open(fnidx, "r");
-       free(fnidx); free(fn);
        if (fp) {
                ti_index_t *idx = ti_index_load_core(fp);
                bgzf_close(fp);
@@ -584,17 +589,53 @@ static void download_from_remote(const char *url)
 }
 #endif
 
+static char *get_local_version(const char *fn)
+{
+    struct stat sbuf;
+       char *fnidx = (char*)calloc(strlen(fn) + 5, 1);
+       strcat(strcpy(fnidx, fn), ".tbi");
+       if ((strstr(fnidx, "ftp://") == fnidx || strstr(fnidx, "http://") == fnidx)) {
+               char *p, *url;
+               int l = strlen(fnidx);
+               for (p = fnidx + l - 1; p >= fnidx; --p)
+                       if (*p == '/') break;
+               url = fnidx; fnidx = strdup(p + 1);
+               if (stat(fnidx, &sbuf) == 0) {
+                       free(url);
+                       return fnidx;
+               }
+               fprintf(stderr, "[%s] downloading the index file...\n", __func__);
+               download_from_remote(url);
+               free(url);
+       }
+    if (stat(fnidx, &sbuf) == 0) return fnidx;
+       free(fnidx); return 0;
+}
+
+int ti_list_chromosomes(const char *fn)
+{
+       ti_index_t *idx;
+       char **names;
+       int i;
+       khint_t k;
+       idx = ti_index_load(fn);
+       names = calloc(idx->n, sizeof(void*));
+       for (k = kh_begin(idx->tname); k < kh_end(idx->tname); ++k)
+               if (kh_exist(idx->tname, k))
+                       names[kh_val(idx->tname, k)] = (char*)kh_key(idx->tname, k);
+       for (i = 0; i < idx->n; ++i) printf("%s\n", names[i]);
+       free(names);
+       ti_index_destroy(idx);
+       return 0;
+}
+
 ti_index_t *ti_index_load(const char *fn)
 {
        ti_index_t *idx;
-       idx = ti_index_load_local(fn);
-       if (idx == 0 && (strstr(fn, "ftp://") == fn || strstr(fn, "http://") == fn)) {
-               char *fnidx = calloc(strlen(fn) + 5, 1);
-               strcat(strcpy(fnidx, fn), ".tbi");
-               fprintf(stderr, "[ti_index_load] attempting to download the remote index file.\n");
-               download_from_remote(fnidx);
-               idx = ti_index_load_local(fn);
-       }
+    char *fname = get_local_version(fn);
+       if (fname == 0) return 0;
+       idx = ti_index_load_local(fname);
+    free(fname);
        if (idx == 0) fprintf(stderr, "[ti_index_load] fail to load BAM index.\n");
        return idx;
 }
@@ -709,7 +750,8 @@ pair64_t *get_chunk_coordinates(const ti_index_t *idx, int tid, int beg, int end
        bins = (uint16_t*)calloc(MAX_BIN, 2);
        n_bins = reg2bins(beg, end, bins);
        index = idx->index[tid];
-       min_off = (beg>>TAD_LIDX_SHIFT >= idx->index2[tid].n)? 0 : idx->index2[tid].offset[beg>>TAD_LIDX_SHIFT];
+       min_off = (beg>>TAD_LIDX_SHIFT >= idx->index2[tid].n)? idx->index2[tid].offset[idx->index2[tid].n-1]
+               : idx->index2[tid].offset[beg>>TAD_LIDX_SHIFT];
        for (i = n_off = 0; i < n_bins; ++i) {
                if ((k = kh_get(i, index, bins[i])) != kh_end(index))
                        n_off += kh_value(index, k).n;