File samtools-0.1.19-R-fixes.patch of Package samtools-legacy
diff -up samtools-0.1.19/faidx.c.Rfixes samtools-0.1.19/faidx.c
--- samtools-0.1.19/faidx.c.Rfixes 2015-05-29 23:25:42.646798526 -0400
+++ samtools-0.1.19/faidx.c 2015-05-29 23:28:55.628504008 -0400
@@ -94,6 +94,11 @@ faidx_t *fai_build_core(RAZF *rz)
}
name[l_name++] = c;
}
+ if (m_name < l_name + 2) { /* MTM: 0-length id */
+ m_name = l_name + 2;
+ kroundup32(m_name);
+ name = (char*)realloc(name, m_name);
+ }
name[l_name] = '\0';
if (ret == 0) {
fprintf(stderr, "[fai_build_core] the last entry has no sequence\n");
@@ -127,11 +132,24 @@ faidx_t *fai_build_core(RAZF *rz)
}
}
}
- fai_insert_index(idx, name, len, line_len, line_blen, offset);
+ if (len < 0) { /* MTM; should also check state */
+ fprintf(stderr, "[fai_build_core] no entries in file\n");
+ free(name); fai_destroy(idx);
+ return 0;
+ }
+ fai_insert_index(idx, name, len, line_len, line_blen, offset);
free(name);
return idx;
}
+// HP - Jan 13, 2014: I've no idea why the original authors of the fai_save()
+// and fai_read() functions below decided to use the (long) type instead of
+// (long long) for the sequence offsets on Windows. Problem with this is that
+// these functions then break if the FASTA file contains sequences with offsets
+// > LONG_MAX which turns out to be 2^31-1 on Windows, hence not big enough if
+// the FASTA file contains the full genome sequences for Human and other
+// mammals. So I modified fai_save() and fai_read() to always use (long long).
+
void fai_save(const faidx_t *fai, FILE *fp)
{
khint_t k;
@@ -140,11 +158,12 @@ void fai_save(const faidx_t *fai, FILE *
faidx1_t x;
k = kh_get(s, fai->hash, fai->name[i]);
x = kh_value(fai->hash, k);
-#ifdef _WIN32
- fprintf(fp, "%s\t%d\t%ld\t%d\t%d\n", fai->name[i], (int)x.len, (long)x.offset, (int)x.line_blen, (int)x.line_len);
-#else
+// HP - Jan 13, 2014: See above note.
+//#ifdef _WIN32
+// fprintf(fp, "%s\t%d\t%ld\t%d\t%d\n", fai->name[i], (int)x.len, (long)x.offset, (int)x.line_blen, (int)x.line_len);
+//#else
fprintf(fp, "%s\t%d\t%lld\t%d\t%d\n", fai->name[i], (int)x.len, (long long)x.offset, (int)x.line_blen, (int)x.line_len);
-#endif
+//#endif
}
}
@@ -153,22 +172,24 @@ faidx_t *fai_read(FILE *fp)
faidx_t *fai;
char *buf, *p;
int len, line_len, line_blen;
-#ifdef _WIN32
- long offset;
-#else
+// HP - Jan 13, 2014: See above note.
+//#ifdef _WIN32
+// long offset;
+//#else
long long offset;
-#endif
+//#endif
fai = (faidx_t*)calloc(1, sizeof(faidx_t));
fai->hash = kh_init(s);
buf = (char*)calloc(0x10000, 1);
while (!feof(fp) && fgets(buf, 0x10000, fp)) {
for (p = buf; *p && isgraph(*p); ++p);
*p = 0; ++p;
-#ifdef _WIN32
- sscanf(p, "%d%ld%d%d", &len, &offset, &line_blen, &line_len);
-#else
+// HP - Jan 13, 2014: See above note.
+//#ifdef _WIN32
+// sscanf(p, "%d%ld%d%d", &len, &offset, &line_blen, &line_len);
+//#else
sscanf(p, "%d%lld%d%d", &len, &offset, &line_blen, &line_len);
-#endif
+//#endif
fai_insert_index(fai, buf, len, line_len, line_blen, offset);
}
free(buf);
@@ -200,6 +221,10 @@ int fai_build(const char *fn)
return -1;
}
fai = fai_build_core(rz);
+ if (fai == NULL) { /* MTM */
+ free(str);
+ return -1;
+ }
razf_close(rz);
fp = fopen(str, "wb");
if (fp == 0) {
@@ -258,10 +283,18 @@ FILE *download_and_open(const char *fn)
faidx_t *fai_load(const char *fn)
{
char *str;
- FILE *fp;
faidx_t *fai;
str = (char*)calloc(strlen(fn) + 5, 1);
sprintf(str, "%s.fai", fn);
+ fai = fai_load0(fn, str);
+ free(str);
+ return fai;
+}
+
+faidx_t *fai_load0(const char *fn, const char *str)
+{
+ FILE *fp;
+ faidx_t *fai;
#ifdef _USE_KNETFILE
if (strstr(fn, "ftp://") == fn || strstr(fn, "http://") == fn)
@@ -270,7 +303,6 @@ faidx_t *fai_load(const char *fn)
if ( !fp )
{
fprintf(stderr, "[fai_load] failed to open remote FASTA index %s\n", str);
- free(str);
return 0;
}
}
@@ -283,7 +315,6 @@ faidx_t *fai_load(const char *fn)
fp = fopen(str, "rb");
if (fp == 0) {
fprintf(stderr, "[fai_load] fail to open FASTA index.\n");
- free(str);
return 0;
}
}
@@ -292,7 +323,6 @@ faidx_t *fai_load(const char *fn)
fclose(fp);
fai->rz = razf_open(fn, "rb");
- free(str);
if (fai->rz == 0) {
fprintf(stderr, "[fai_load] fail to open FASTA file.\n");
return 0;
diff -up samtools-0.1.19/faidx.h.Rfixes samtools-0.1.19/faidx.h
--- samtools-0.1.19/faidx.h.Rfixes 2013-03-19 03:48:09.000000000 -0400
+++ samtools-0.1.19/faidx.h 2015-05-29 23:29:33.916247174 -0400
@@ -64,6 +64,13 @@ extern "C" {
faidx_t *fai_load(const char *fn);
/*!
+ @abstract Load index from "fn.fai".
+ @param fn File name of the FASTA file
+ @param index File name of the FASTA index
+ */
+ faidx_t *fai_load0(const char *fn, const char *index);
+
+ /*!
@abstract Fetch the sequence in a region.
@param fai Pointer to the faidx_t struct
@param reg Region in the format "chr2:20,000-30,000"
@@ -96,6 +103,22 @@ extern "C" {
*/
char *faidx_fetch_seq(const faidx_t *fai, char *c_name, int p_beg_i, int p_end_i, int *len);
+ /*!
+ @abstract Alternative to faidx_fetch_seq().
+ @param fai Pointer to the faidx_t struct
+ @param c_name Region name
+ @param p_beg_i Beginning position number (zero-based)
+ @param p_end_i End position number (zero-based)
+ @param out User-supplied output buffer
+ @return Number of bytes written; -1 on failure
+
+ @discussion Differences with faidx_fetch_seq(): (1) writes the
+ incoming sequence to user-supplied output buffer, (2) doesn't write
+ the terminating null byte ('\0'), (3) properly handles 0-length
+ sequences, (4) returns the number of bytes written; -1 on failure.
+ */
+ int faidx_fetch_seq2(const faidx_t *fai, const char *c_name, int p_beg_i, int p_end_i, char *out);
+
#ifdef __cplusplus
}
#endif
