Allow tabs in the CO header field

[samtools.git] / bam_index.c

diff --git a/bam_index.c b/bam_index.c
index 5126f00c8d8da0c41e0ba2179ca23cc8d436e53f..853ac705fc9bef5a653ab9aae9eb9100ec551e12 100644 (file)
--- a/bam_index.c
+++ b/bam_index.c
@@ -4,7 +4,9 @@
 #include "khash.h"
 #include "ksort.h"
 #include "bam_endian.h"
+#ifdef _USE_KNETFILE
 #include "knetfile.h"
+#endif
 
 /*!
   @header
@@ -337,13 +339,13 @@ bam_index_t *bam_index_load_local(const char *_fn)
        } else fn = strdup(_fn);
        fnidx = (char*)calloc(strlen(fn) + 5, 1);
        strcpy(fnidx, fn); strcat(fnidx, ".bai");
-       fp = fopen(fnidx, "r");
+       fp = fopen(fnidx, "rb");
        if (fp == 0) { // try "{base}.bai"
                char *s = strstr(fn, "bam");
                if (s == fn + strlen(fn) - 3) {
                        strcpy(fnidx, fn);
                        fnidx[strlen(fn)-1] = 'i';
-                       fp = fopen(fnidx, "r");
+                       fp = fopen(fnidx, "rb");
                }
        }
        free(fnidx); free(fn);
@@ -373,7 +375,7 @@ static void download_from_remote(const char *url)
                fprintf(stderr, "[download_from_remote] fail to open remote file.\n");
                return;
        }
-       if ((fp = fopen(fn, "w")) == 0) {
+       if ((fp = fopen(fn, "wb")) == 0) {
                fprintf(stderr, "[download_from_remote] fail to create file in the working directory.\n");
                knet_close(fp_remote);
                return;
@@ -423,7 +425,7 @@ int bam_index_build2(const char *fn, const char *_fnidx)
                fnidx = (char*)calloc(strlen(fn) + 5, 1);
                strcpy(fnidx, fn); strcat(fnidx, ".bai");
        } else fnidx = strdup(_fnidx);
-       fpidx = fopen(fnidx, "w");
+       fpidx = fopen(fnidx, "wb");
        if (fpidx == 0) {
                fprintf(stderr, "[bam_index_build2] fail to create the index file.\n");
                free(fnidx);
@@ -474,7 +476,8 @@ static inline int is_overlap(uint32_t beg, uint32_t end, const bam1_t *b)
        return (rend > beg && rbeg < end);
 }
 
-int bam_fetch(bamFile fp, const bam_index_t *idx, int tid, int beg, int end, void *data, bam_fetch_f func)
+// bam_fetch helper function retrieves 
+pair64_t * get_chunk_coordinates(const bam_index_t *idx, int tid, int beg, int end, int* cnt_off)
 {
        uint16_t *bins;
        int i, n_bins, n_off;
@@ -505,10 +508,8 @@ int bam_fetch(bamFile fp, const bam_index_t *idx, int tid, int beg, int end, voi
        }
        free(bins);
        {
-               bam1_t *b;
-               int l, ret, n_seeks;
-               uint64_t curr_off;
-               b = (bam1_t*)calloc(1, sizeof(bam1_t));
+               bam1_t *b = (bam1_t*)calloc(1, sizeof(bam1_t));
+               int l;
                ks_introsort(off, n_off, off);
                // resolve completely contained adjacent blocks
                for (i = 1, l = 0; i < n_off; ++i)
@@ -531,8 +532,23 @@ int bam_fetch(bamFile fp, const bam_index_t *idx, int tid, int beg, int end, voi
                        n_off = l + 1;
 #endif
                }
+               bam_destroy1(b);
+       }
+       *cnt_off = n_off;
+       return off;
+}
+
+int bam_fetch(bamFile fp, const bam_index_t *idx, int tid, int beg, int end, void *data, bam_fetch_f func)
+{
+       int n_off;
+       pair64_t *off = get_chunk_coordinates(idx, tid, beg, end, &n_off);
+       if (off == 0) return 0;
+       {
                // retrive alignments
+               uint64_t curr_off;
+               int i, ret, n_seeks;
                n_seeks = 0; i = -1; curr_off = 0;
+               bam1_t *b = (bam1_t*)calloc(1, sizeof(bam1_t));
                for (;;) {
                        if (curr_off == 0 || curr_off >= off[i].v) { // then jump to the next chunk
                                if (i == n_off - 1) break; // no more chunks