Changed *_parse_region functions to use 64-bit coords.

Author: jkbonfield

faidx.c

@@ -730,10 +730,12 @@ static char *fai_retrieve(const faidx_t *fai, const faidx1_t *val,
     return s;
 }
 
-static int fai_get_val(const faidx_t *fai, const char *str, int *len, faidx1_t *val, long *fbeg, long *fend) {
+static int fai_get_val(const faidx_t *fai, const char *str,
+                       int *len, faidx1_t *val, int64_t *fbeg, int64_t *fend) {
     khiter_t iter;
     khash_t(s) *h;
-    int id, beg, end;
+    int id;
+    int64_t beg, end;
 
     if (!fai_parse_region(fai, str, &id, &beg, &end, 0)) {
         hts_log_warning("Reference %s not found in FASTA file, returning empty sequence", str);
@@ -765,7 +767,7 @@ static int fai_get_val(const faidx_t *fai, const char *str, int *len, faidx1_t *
 char *fai_fetch(const faidx_t *fai, const char *str, int *len)
 {
     faidx1_t val;
-    long beg, end;
+    int64_t beg, end;
 
     if (fai_get_val(fai, str, len, &val, &beg, &end)) {
         return NULL;
@@ -778,7 +780,7 @@ char *fai_fetch(const faidx_t *fai, const char *str, int *len)
 
 char *fai_fetchqual(const faidx_t *fai, const char *str, int *len) {
     faidx1_t val;
-    long beg, end;
+    int64_t beg, end;
 
     if (fai_get_val(fai, str, len, &val, &beg, &end)) {
         return NULL;
@@ -878,7 +880,8 @@ int faidx_has_seq(const faidx_t *fai, const char *seq)
     return 1;
 }
 
-const char *fai_parse_region(const faidx_t *fai, const char *s, int *tid, int *beg, int *end, int flags)
+const char *fai_parse_region(const faidx_t *fai, const char *s,
+                             int *tid, int64_t *beg, int64_t *end, int flags)
 {
     return hts_parse_region(s, tid, beg, end, (hts_name2id_f)fai_name2id, (void *)fai, flags);
 }

hts.c

@@ -2686,7 +2686,7 @@ static void *hts_memrchr(const void *s, int c, size_t n) {
  *            beg/end will be set, plus tid if getid has been supplied.
  * On failure NULL is returned.
  */
-const char *hts_parse_region(const char *s, int *tid, int *beg, int *end,
+const char *hts_parse_region(const char *s, int *tid, int64_t *beg, int64_t *end,
                              hts_name2id_f getid, void *hdr, int flags)
 {
     if (!s || !tid || !beg || !end || !getid)
@@ -2861,7 +2861,8 @@ const char *hts_parse_reg(const char *s, int *beg, int *end)
 
 hts_itr_t *hts_itr_querys(const hts_idx_t *idx, const char *reg, hts_name2id_f getid, void *hdr, hts_itr_query_func *itr_query, hts_readrec_func *readrec)
 {
-    int tid, beg, end;
+    int tid;
+    int64_t beg, end;
 
     if (strcmp(reg, ".") == 0)
         return itr_query(idx, HTS_IDX_START, 0, 0, readrec);

htslib/faidx.h

@@ -29,6 +29,7 @@
 #ifndef HTSLIB_FAIDX_H
 #define HTSLIB_FAIDX_H
 
+#include <inttypes.h>
 #include "hts_defs.h"
 
 #ifdef __cplusplus
@@ -234,7 +235,7 @@ int faidx_seq_len(const faidx_t *fai, const char *seq);
     @param  flags Parsing method, see HTS_PARSE_* in hts.h.
     @return      pointer to end of parsed s if successs, NULL if not.
 */
-const char *fai_parse_region(const faidx_t *fai, const char *s, int *tid, int *beg, int *end, int flags);
+const char *fai_parse_region(const faidx_t *fai, const char *s, int *tid, int64_t *beg, int64_t *end, int flags);
 
 #ifdef __cplusplus
 }

htslib/hts.h

@@ -819,7 +819,7 @@ const char *hts_parse_reg(const char *str, int *beg, int *end);
     @return  Pointer to the colon or '\0' after the reference sequence name,
              or NULL if @a str could not be parsed.
 */
-const char *hts_parse_region(const char *str, int *tid, int *beg, int *end,
+const char *hts_parse_region(const char *str, int *tid, int64_t *beg, int64_t *end,
                              hts_name2id_f getid, void *hdr, int flags);
 
 

htslib/sam.h

@@ -281,7 +281,7 @@ typedef struct {
     int bam_hdr_write(BGZF *fp, const bam_hdr_t *h) HTS_RESULT_USED;
     void bam_hdr_destroy(bam_hdr_t *h);
     int bam_name2id(bam_hdr_t *h, const char *ref);
-    const char *sam_parse_region(bam_hdr_t *h, const char *s, int *tid, int *beg, int *end, int flags);
+    const char *sam_parse_region(bam_hdr_t *h, const char *s, int *tid, int64_t *beg, int64_t *end, int flags);
     bam_hdr_t* bam_hdr_dup(const bam_hdr_t *h0);
 
     bam1_t *bam_init1(void);

sam.c

@@ -298,7 +298,7 @@ int bam_name2id(bam_hdr_t *h, const char *ref)
     return k == kh_end(d)? -1 : kh_val(d, k);
 }
 
-const char *sam_parse_region(bam_hdr_t *h, const char *s, int *tid, int *beg, int *end, int flags) {
+const char *sam_parse_region(bam_hdr_t *h, const char *s, int *tid, int64_t *beg, int64_t *end, int flags) {
     return hts_parse_region(s, tid, beg, end, (hts_name2id_f)bam_name2id, h, flags);
 }
 

test/test-parse-reg.c

@@ -41,14 +41,17 @@
 #include <string.h>
 #include <stdlib.h>
 #include <limits.h>
+#include <stdint.h>
+#include <inttypes.h>
 
 #include <htslib/hts.h>
 #include <htslib/sam.h>
 
 void reg_expected(bam_hdr_t *hdr, const char *reg, int flags,
-                 char *reg_exp, int tid_exp, int beg_exp, int end_exp) {
+                 char *reg_exp, int tid_exp, int64_t beg_exp, int64_t end_exp) {
     const char *reg_out;
-    int tid_out = -1, beg_out = -1, end_out = -1;
+    int tid_out = -1;
+    int64_t beg_out = -1, end_out = -1;
 
     reg_out = sam_parse_region(hdr, reg, &tid_out, &beg_out, &end_out, flags);
 
@@ -57,16 +60,12 @@ void reg_expected(bam_hdr_t *hdr, const char *reg, int flags,
         (reg_exp && tid_out != tid_exp) ||
         (reg_exp && beg_out != beg_exp) ||
         (reg_exp && end_out != end_exp)) {
-        fprintf(stderr, "Parsing \"%s\" expected return \"%s\", %d:%d-%d, "
-                "but got \"%s\", %d:%d-%d\n",
+        fprintf(stderr, "Parsing \"%s\" expected return \"%s\", %d:%"PRId64"-%"PRId64", "
+                "but got \"%s\", %d:%"PRId64"-%"PRId64"\n",
                 reg,
                 reg_exp?reg_exp:"(null)", tid_exp, beg_exp, end_exp,
                 reg_out?reg_out:"(null)", tid_out, beg_out, end_out);
         exit(1);
-//    } else {
-//        fprintf(stderr, "%s parsed as expected result: \"%s\" %d:%d-%d\n",
-//                reg,
-//                reg_out?reg_out:"(null)", tid_out, beg_out, end_out);
     }
 }
 
@@ -88,6 +87,7 @@ int reg_test(char *fn) {
     // 5 chr1,chr3
 
     // Check range extensions.
+    // NB: fix INT_MAX once we merge the 64-bit pos code.
     reg_expected(hdr, "chr1", 0, "",  0, 0, INT_MAX);
     reg_expected(hdr, "chr1:50", 0, "",  0, 49, INT_MAX);
     reg_expected(hdr, "chr1:50", HTS_PARSE_ONE_COORD, "",  0, 49, 50);
@@ -186,13 +186,14 @@ int main(int argc, char **argv) {
 
     const char *reg = argv[2];
     while (*reg) {
-        int tid, beg, end;
+        int tid;
+        int64_t beg, end;
         reg = sam_parse_region(hdr, reg, &tid, &beg, &end, flags);
         if (!reg) {
             fprintf(stderr, "Failed to parse region\n");
             exit(1);
         }
-        printf("%-20s %12d %12d\n",
+        printf("%-20s %12"PRId64" %12"PRId64"\n",
                tid == -1 ? "*" : hdr->target_name[tid],
                beg, end);
     }