X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=faidx.c;h=38292a13d73f33e75b12ad02baa3af2f66242bab;hb=7251efa37992752a8cf62ff363da0ad5099937bd;hp=44e7f5710ce06fdd350d265ae4887eb2326eea51;hpb=f93dae0d03856955f9424e8b2aaf261304ca647e;p=samtools.git

diff --git a/faidx.c b/faidx.c
index 44e7f57..38292a1 100644
--- a/faidx.c
+++ b/faidx.c
@@ -1,8 +1,8 @@
 #include <ctype.h>
-#include <assert.h>
 #include <string.h>
 #include <stdlib.h>
 #include <stdio.h>
+#include <stdint.h>
 #include "faidx.h"
 #include "khash.h"
 
@@ -12,11 +12,16 @@ typedef struct {
 } faidx1_t;
 KHASH_MAP_INIT_STR(s, faidx1_t)
 
-#ifdef HAVE_RAZF
+#ifndef _NO_RAZF
 #include "razf.h"
 #else
+#ifdef _WIN32
+#define ftello(fp) ftell(fp)
+#define fseeko(fp, offset, whence) fseek(fp, offset, whence)
+#else
 extern off_t ftello(FILE *stream);
 extern int fseeko(FILE *stream, off_t offset, int whence);
+#endif
 #define RAZF FILE
 #define razf_read(fp, buf, size) fread(buf, 1, size, fp)
 #define razf_open(fn, mode) fopen(fn, mode)
@@ -24,6 +29,9 @@ extern int fseeko(FILE *stream, off_t offset, int whence);
 #define razf_seek(fp, offset, whence) fseeko(fp, offset, whence)
 #define razf_tell(fp) ftello(fp)
 #endif
+#ifdef _USE_KNETFILE
+#include "knetfile.h"
+#endif
 
 struct __faidx_t {
 	RAZF *rz;
@@ -66,6 +74,12 @@ faidx_t *fai_build_core(RAZF *rz)
 	name = 0; l_name = m_name = 0;
 	len = line_len = line_blen = -1; state = 0; l1 = l2 = -1; offset = 0;
 	while (razf_read(rz, &c, 1)) {
+		if (c == '\n') { // an empty line
+			if (state == 1) {
+				offset = razf_tell(rz);
+				continue;
+			} else if ((state == 0 && len < 0) || state == 2) continue;
+		}
 		if (c == '>') { // fasta header
 			if (len >= 0)
 				fai_insert_index(idx, name, len, line_len, line_blen, offset);
@@ -79,14 +93,19 @@ faidx_t *fai_build_core(RAZF *rz)
 				name[l_name++] = c;
 			}
 			name[l_name] = '\0';
-			assert(ret);
+			if (ret == 0) {
+				fprintf(stderr, "[fai_build_core] the last entry has no sequence\n");
+				free(name); fai_destroy(idx);
+				return 0;
+			}
 			if (c != '\n') while (razf_read(rz, &c, 1) && c != '\n');
 			state = 1; len = 0;
 			offset = razf_tell(rz);
 		} else {
 			if (state == 3) {
-				fprintf(stderr, "[fai_build_core] inlined empty line is not allowed in sequence '%s'. Abort!\n", name);
-				exit(1);
+				fprintf(stderr, "[fai_build_core] inlined empty line is not allowed in sequence '%s'.\n", name);
+				free(name); fai_destroy(idx);
+				return 0;
 			}
 			if (state == 2) state = 3;
 			l1 = l2 = 0;
@@ -95,13 +114,15 @@ faidx_t *fai_build_core(RAZF *rz)
 				if (isgraph(c)) ++l2;
 			} while ((ret = razf_read(rz, &c, 1)) && c != '\n');
 			if (state == 3 && l2) {
-				fprintf(stderr, "[fai_build_core] different line length in sequence '%s'. Abort!\n", name);
-				exit(1);
+				fprintf(stderr, "[fai_build_core] different line length in sequence '%s'.\n", name);
+				free(name); fai_destroy(idx);
+				return 0;
 			}
 			++l1; len += l2;
 			if (l2 >= 0x10000) {
-				fprintf(stderr, "[fai_build_core] line length exceeds 65535 in sequence '%s'. Abort!\n", name);
-				exit(1);
+				fprintf(stderr, "[fai_build_core] line length exceeds 65535 in sequence '%s'.\n", name);
+				free(name); fai_destroy(idx);
+				return 0;
 			}
 			if (state == 1) line_len = l1, line_blen = l2, state = 0;
 			else if (state == 0) {
@@ -122,7 +143,11 @@ void fai_save(const faidx_t *fai, FILE *fp)
 		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
 		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
 	}
 }
 
@@ -131,14 +156,22 @@ faidx_t *fai_read(FILE *fp)
 	faidx_t *fai;
 	char *buf, *p;
 	int len, line_len, line_blen;
+#ifdef _WIN32
+	long offset;
+#else
 	long long offset;
+#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
 		sscanf(p, "%d%lld%d%d", &len, &offset, &line_blen, &line_len);
+#endif
 		fai_insert_index(fai, buf, len, line_len, line_blen, offset);
 	}
 	free(buf);
@@ -155,7 +188,7 @@ void fai_destroy(faidx_t *fai)
 	free(fai);
 }
 
-void fai_build(const char *fn)
+int fai_build(const char *fn)
 {
 	char *str;
 	RAZF *rz;
@@ -164,17 +197,67 @@ void fai_build(const char *fn)
 	str = (char*)calloc(strlen(fn) + 5, 1);
 	sprintf(str, "%s.fai", fn);
 	rz = razf_open(fn, "r");
-	assert(rz);
+	if (rz == 0) {
+		fprintf(stderr, "[fai_build] fail to open the FASTA file %s\n",fn);
+		free(str);
+		return -1;
+	}
 	fai = fai_build_core(rz);
 	razf_close(rz);
-	fp = fopen(str, "w");
-	assert(fp);
+	fp = fopen(str, "wb");
+	if (fp == 0) {
+		fprintf(stderr, "[fai_build] fail to write FASTA index %s\n",str);
+		fai_destroy(fai); free(str);
+		return -1;
+	}
 	fai_save(fai, fp);
 	fclose(fp);
 	free(str);
 	fai_destroy(fai);
+	return 0;
 }
 
+#ifdef _USE_KNETFILE
+FILE *download_and_open(const char *fn)
+{
+    const int buf_size = 1 * 1024 * 1024;
+    uint8_t *buf;
+    FILE *fp;
+    knetFile *fp_remote;
+    const char *url = fn;
+    const char *p;
+    int l = strlen(fn);
+    for (p = fn + l - 1; p >= fn; --p)
+        if (*p == '/') break;
+    fn = p + 1;
+
+    // First try to open a local copy
+    fp = fopen(fn, "r");
+    if (fp)
+        return fp;
+
+    // If failed, download from remote and open
+    fp_remote = knet_open(url, "rb");
+    if (fp_remote == 0) {
+        fprintf(stderr, "[download_from_remote] fail to open remote file %s\n",url);
+        return NULL;
+    }
+    if ((fp = fopen(fn, "wb")) == 0) {
+        fprintf(stderr, "[download_from_remote] fail to create file in the working directory %s\n",fn);
+        knet_close(fp_remote);
+        return NULL;
+    }
+    buf = (uint8_t*)calloc(buf_size, 1);
+    while ((l = knet_read(fp_remote, buf, buf_size)) != 0)
+        fwrite(buf, 1, l, fp);
+    free(buf);
+    fclose(fp);
+    knet_close(fp_remote);
+
+    return fopen(fn, "r");
+}
+#endif
+
 faidx_t *fai_load(const char *fn)
 {
 	char *str;
@@ -182,22 +265,41 @@ faidx_t *fai_load(const char *fn)
 	faidx_t *fai;
 	str = (char*)calloc(strlen(fn) + 5, 1);
 	sprintf(str, "%s.fai", fn);
-	fp = fopen(str, "r");
+
+#ifdef _USE_KNETFILE
+    if (strstr(fn, "ftp://") == fn || strstr(fn, "http://") == fn)
+    {
+        fp = download_and_open(str);
+        if ( !fp )
+        {
+            fprintf(stderr, "[fai_load] failed to open remote FASTA index %s\n", str);
+            free(str);
+            return 0;
+        }
+    }
+    else
+#endif
+        fp = fopen(str, "rb");
 	if (fp == 0) {
 		fprintf(stderr, "[fai_load] build FASTA index.\n");
 		fai_build(fn);
-		fp = fopen(str, "r");
+		fp = fopen(str, "rb");
 		if (fp == 0) {
+			fprintf(stderr, "[fai_load] fail to open FASTA index.\n");
 			free(str);
 			return 0;
 		}
 	}
+
 	fai = fai_read(fp);
 	fclose(fp);
-	fai->rz = razf_open(fn, "r");
-	if (fai->rz == 0) return 0;
-	assert(fai->rz);
+
+	fai->rz = razf_open(fn, "rb");
 	free(str);
+	if (fai->rz == 0) {
+		fprintf(stderr, "[fai_load] fail to open FASTA file.\n");
+		return 0;
+	}
 	return fai;
 }
 
@@ -246,7 +348,7 @@ char *fai_fetch(const faidx_t *fai, const char *str, int *len)
 	l = 0;
 	s = (char*)malloc(end - beg + 2);
 	razf_seek(fai->rz, val.offset + beg / val.line_blen * val.line_len + beg % val.line_blen, SEEK_SET);
-	while (razf_read(fai->rz, &c, 1) == 1 && l < end - beg)
+	while (razf_read(fai->rz, &c, 1) == 1 && l < end - beg && !fai->rz->z_err)
 		if (isgraph(c)) s[l++] = c;
 	s[l] = '\0';
 	*len = l;
@@ -265,7 +367,7 @@ int faidx_main(int argc, char *argv[])
 			char *s;
 			faidx_t *fai;
 			fai = fai_load(argv[1]);
-			assert(fai);
+			if (fai == 0) return 1;
 			for (i = 2; i != argc; ++i) {
 				printf(">%s\n", argv[i]);
 				s = fai_fetch(fai, argv[i], &l);
@@ -282,6 +384,40 @@ int faidx_main(int argc, char *argv[])
 	return 0;
 }
 
+int faidx_fetch_nseq(const faidx_t *fai) 
+{
+	return fai->n;
+}
+
+char *faidx_fetch_seq(const faidx_t *fai, char *c_name, int p_beg_i, int p_end_i, int *len)
+{
+	int l;
+	char c;
+    khiter_t iter;
+    faidx1_t val;
+	char *seq=NULL;
+
+    // Adjust position
+    iter = kh_get(s, fai->hash, c_name);
+    if(iter == kh_end(fai->hash)) return 0;
+    val = kh_value(fai->hash, iter);
+	if(p_end_i < p_beg_i) p_beg_i = p_end_i;
+    if(p_beg_i < 0) p_beg_i = 0;
+    else if(val.len <= p_beg_i) p_beg_i = val.len - 1;
+    if(p_end_i < 0) p_end_i = 0;
+    else if(val.len <= p_end_i) p_end_i = val.len - 1;
+
+    // Now retrieve the sequence 
+	l = 0;
+	seq = (char*)malloc(p_end_i - p_beg_i + 2);
+	razf_seek(fai->rz, val.offset + p_beg_i / val.line_blen * val.line_len + p_beg_i % val.line_blen, SEEK_SET);
+	while (razf_read(fai->rz, &c, 1) == 1 && l < p_end_i - p_beg_i + 1)
+		if (isgraph(c)) seq[l++] = c;
+	seq[l] = '\0';
+	*len = l;
+	return seq;
+}
+
 #ifdef FAIDX_MAIN
 int main(int argc, char *argv[]) { return faidx_main(argc, argv); }
 #endif