Imported Upstream version 0.1.7~dfsg

[samtools.git] / bam_index.c

diff --git a/bam_index.c b/bam_index.c
index 72ef270f9c7afd143759e1f98d681ce69147bb13..a6278843583c605799360affb0b9c3c21fbd02c4 100644 (file)
--- a/bam_index.c
+++ b/bam_index.c
@@ -4,7 +4,9 @@
 #include "khash.h"
 #include "ksort.h"
 #include "bam_endian.h"
 #include "khash.h"
 #include "ksort.h"
 #include "bam_endian.h"

+#ifdef _USE_KNETFILE

 #include "knetfile.h"
 #include "knetfile.h"

+#endif

 
 /*!
   @header
 
 /*!
   @header


@@ -328,7 +330,7 @@ bam_index_t *bam_index_load_local(const char *_fn)
        FILE *fp;
        char *fnidx, *fn;
 
        FILE *fp;
        char *fnidx, *fn;
 

-       if (strstr(_fn, "ftp://") == _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)
                const char *p;
                int l = strlen(_fn);
                for (p = _fn + l - 1; p >= _fn; --p)


@@ -354,6 +356,7 @@ bam_index_t *bam_index_load_local(const char *_fn)
        } else return 0;
 }
 
        } else return 0;
 }
 

+#ifdef _USE_KNETFILE

 static void download_from_remote(const char *url)
 {
        const int buf_size = 1 * 1024 * 1024;
 static void download_from_remote(const char *url)
 {
        const int buf_size = 1 * 1024 * 1024;


@@ -362,7 +365,7 @@ static void download_from_remote(const char *url)
        uint8_t *buf;
        knetFile *fp_remote;
        int l;
        uint8_t *buf;
        knetFile *fp_remote;
        int l;

-       if (strstr(url, "ftp://") != url) return;
+       if (strstr(url, "ftp://") != url && strstr(url, "http://") != url) return;

        l = strlen(url);
        for (fn = (char*)url + l - 1; fn >= url; --fn)
                if (*fn == '/') break;
        l = strlen(url);
        for (fn = (char*)url + l - 1; fn >= url; --fn)
                if (*fn == '/') break;


@@ -384,12 +387,18 @@ static void download_from_remote(const char *url)
        fclose(fp);
        knet_close(fp_remote);
 }
        fclose(fp);
        knet_close(fp_remote);
 }

+#else
+static void download_from_remote(const char *url)
+{
+       return;
+}
+#endif

 
 bam_index_t *bam_index_load(const char *fn)
 {
        bam_index_t *idx;
        idx = bam_index_load_local(fn);
 
 bam_index_t *bam_index_load(const char *fn)
 {
        bam_index_t *idx;
        idx = bam_index_load_local(fn);

-       if (idx == 0 && strstr(fn, "ftp://") == fn) {
+       if (idx == 0 && (strstr(fn, "ftp://") == fn || strstr(fn, "http://") == fn)) {

                char *fnidx = calloc(strlen(fn) + 5, 1);
                strcat(strcpy(fnidx, fn), ".bai");
                fprintf(stderr, "[bam_index_load] attempting to download the remote index file.\n");
                char *fnidx = calloc(strlen(fn) + 5, 1);
                strcat(strcpy(fnidx, fn), ".bai");
                fprintf(stderr, "[bam_index_load] attempting to download the remote index file.\n");


@@ -467,7 +476,8 @@ static inline int is_overlap(uint32_t beg, uint32_t end, const bam1_t *b)
        return (rend > beg && rbeg < end);
 }
 
        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;
 {
        uint16_t *bins;
        int i, n_bins, n_off;


@@ -498,10 +508,8 @@ int bam_fetch(bamFile fp, const bam_index_t *idx, int tid, int beg, int end, voi
        }
        free(bins);
        {
        }
        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)
                ks_introsort(off, n_off, off);
                // resolve completely contained adjacent blocks
                for (i = 1, l = 0; i < n_off; ++i)


@@ -524,8 +532,23 @@ int bam_fetch(bamFile fp, const bam_index_t *idx, int tid, int beg, int end, voi
                        n_off = l + 1;
 #endif
                }
                        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
                // retrive alignments

+               uint64_t curr_off;
+               int i, ret, n_seeks;

                n_seeks = 0; i = -1; curr_off = 0;
                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
                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