* samtools-0.1.4-7 (r337)

 * bgzf.c: support mode string "wu": uncompressed output
 * "samtools view" support "-u" command-line option

diff --git a/bamtk.c b/bamtk.c
index f762a136e45e455390744a9806ea4b60ad6a4db7..878562cca946bb417779456d329230eb2c1d9da4 100644 (file)
--- a/bamtk.c
+++ b/bamtk.c
@@ -3,7 +3,7 @@
 #include "bam.h"
 
 #ifndef PACKAGE_VERSION
-#define PACKAGE_VERSION "0.1.4-6 (r334)"
+#define PACKAGE_VERSION "0.1.4-7 (r337)"
 #endif
 
 int bam_taf2baf(int argc, char *argv[]);

diff --git a/bgzf.c b/bgzf.c
index bffe8b54b4253f9aa6b3703d0cbc6cbd1d422b91..49a78f13e1b053a6da30c37be76d5a1769a0a530 100644 (file)
--- a/bgzf.c
+++ b/bgzf.c
@@ -9,6 +9,8 @@
  * or functionality.
  */
 
+/* 2009-06-12 by lh3: support a mode string like "wu" where 'u' for uncompressed output */
+
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
@@ -91,7 +93,7 @@ open_read(int fd)
        fp = malloc(sizeof(BGZF));
     fp->file_descriptor = fd;
     fp->open_mode = 'r';
-    fp->owned_file = 0;
+    fp->owned_file = 0; fp->is_uncompressed = 0;
     fp->file = file;
     fp->uncompressed_block_size = MAX_BLOCK_SIZE;
     fp->uncompressed_block = malloc(MAX_BLOCK_SIZE);
@@ -106,7 +108,7 @@ open_read(int fd)
 
 static
 BGZF*
-open_write(int fd)
+open_write(int fd, bool is_uncompressed)
 {
     FILE* file = fdopen(fd, "w");
     BGZF* fp;
@@ -114,7 +116,7 @@ open_write(int fd)
        fp = malloc(sizeof(BGZF));
     fp->file_descriptor = fd;
     fp->open_mode = 'w';
-    fp->owned_file = 0;
+    fp->owned_file = 0; fp->is_uncompressed = is_uncompressed;
     fp->file = file;
     fp->uncompressed_block_size = DEFAULT_BLOCK_SIZE;
     fp->uncompressed_block = NULL;
@@ -131,16 +133,16 @@ BGZF*
 bgzf_open(const char* __restrict path, const char* __restrict mode)
 {
     BGZF* fp = NULL;
-    if (strcasecmp(mode, "r") == 0) {
+    if (mode[0] == 'r' || mode[0] == 'R') { /* The reading mode is preferred. */
                int oflag = O_RDONLY;
                int fd = open(path, oflag);
                if (fd == -1) return 0;
         fp = open_read(fd);
-    } else if (strcasecmp(mode, "w") == 0) {
+    } else if (mode[0] == 'w' || mode[0] == 'W') {
                int oflag = O_WRONLY | O_CREAT | O_TRUNC;
                int fd = open(path, oflag, 0644);
                if (fd == -1) return 0;
-        fp = open_write(fd);
+        fp = open_write(fd, strstr(mode, "u")? 1 : 0);
     }
     if (fp != NULL) {
         fp->owned_file = 1;
@@ -152,10 +154,10 @@ BGZF*
 bgzf_fdopen(int fd, const char * __restrict mode)
 {
        if (fd == -1) return 0;
-    if (strcasecmp(mode, "r") == 0) {
+    if (mode[0] == 'r' || mode[0] == 'R') {
         return open_read(fd);
-    } else if (strcasecmp(mode, "w") == 0) {
-        return open_write(fd);
+    } else if (mode[0] == 'w' || mode[0] == 'W') {
+        return open_write(fd, strstr(mode, "u")? 1 : 0);
     } else {
         return NULL;
     }
@@ -195,7 +197,7 @@ deflate_block(BGZF* fp, int block_length)
     int input_length = block_length;
     int compressed_length = 0;
     while (1) {
-
+               int compress_level = fp->is_uncompressed? 0 : Z_DEFAULT_COMPRESSION;
         z_stream zs;
         zs.zalloc = NULL;
         zs.zfree = NULL;
@@ -204,7 +206,7 @@ deflate_block(BGZF* fp, int block_length)
         zs.next_out = (void*)&buffer[BLOCK_HEADER_LENGTH];
         zs.avail_out = buffer_size - BLOCK_HEADER_LENGTH - BLOCK_FOOTER_LENGTH;
 
-        int status = deflateInit2(&zs, Z_DEFAULT_COMPRESSION, Z_DEFLATED,
+        int status = deflateInit2(&zs, compress_level, Z_DEFLATED,
                                   GZIP_WINDOW_BITS, Z_DEFAULT_MEM_LEVEL, Z_DEFAULT_STRATEGY);
         if (status != Z_OK) {
             report_error(fp, "deflate init failed");

diff --git a/bgzf.h b/bgzf.h
index 7e76bcb9d7eeb128ba3b7047cb1211bdca582a91..ca5e7ab08b53723491b18534a48e03b7188fbba7 100644 (file)
--- a/bgzf.h
+++ b/bgzf.h
@@ -14,16 +14,15 @@
 
 #include <stdint.h>
 #include <stdio.h>
-#include "zlib.h"
 #include <stdbool.h>
-//#include "zutil.h"
+#include <zlib.h>
 
 //typedef int8_t bool;
 
 typedef struct {
     int file_descriptor;
     char open_mode;  // 'r' or 'w'
-    bool owned_file;
+    bool owned_file, is_uncompressed;
     FILE* file;
     int uncompressed_block_size;
     int compressed_block_size;

diff --git a/sam.c b/sam.c
index 9a6e201d75e24d366903ff4e937f80d71e45303c..04be5c57745a8036dce8b2e3ede563f38aa8cdbb 100644 (file)
--- a/sam.c
+++ b/sam.c
@@ -49,8 +49,10 @@ samfile_t *samopen(const char *fn, const char *mode, const void *aux)
        } else if (mode[0] == 'w') { // write
                fp->header = bam_header_dup((const bam_header_t*)aux);
                if (mode[1] == 'b') { // binary
+                       char bmode[3];
+                       bmode[0] = 'w'; bmode[1] = strstr(mode, "u")? 'u' : 0; bmode[2] = 0;
                        fp->type |= TYPE_BAM;
-                       fp->x.bam = strcmp(fn, "-")? bam_open(fn, "w") : bam_dopen(fileno(stdout), "w");
+                       fp->x.bam = strcmp(fn, "-")? bam_open(fn, bmode) : bam_dopen(fileno(stdout), bmode);
                        if (fp->x.bam == 0) goto open_err_ret;
                        bam_header_write(fp->x.bam, fp->header);
                } else { // text

diff --git a/sam_view.c b/sam_view.c
index 5bd679c5cc65eae574865dacb71082582e053d4c..48586106aabe3fb9bdad2f7c48d7acba8e3f36ea 100644 (file)
--- a/sam_view.c
+++ b/sam_view.c
@@ -20,16 +20,16 @@ static int usage(void);
 
 int main_samview(int argc, char *argv[])
 {
-       int c, is_header = 0, is_header_only = 0, is_bamin = 1, ret = 0;
+       int c, is_header = 0, is_header_only = 0, is_bamin = 1, ret = 0, is_uncompressed = 0, is_bamout = 0;
        samfile_t *in = 0, *out = 0;
-       char in_mode[4], out_mode[4], *fn_out = 0, *fn_list = 0;
+       char in_mode[5], out_mode[5], *fn_out = 0, *fn_list = 0;
 
        /* parse command-line options */
        strcpy(in_mode, "r"); strcpy(out_mode, "w");
-       while ((c = getopt(argc, argv, "Sbt:hHo:q:f:F:")) >= 0) {
+       while ((c = getopt(argc, argv, "Sbt:hHo:q:f:F:u")) >= 0) {
                switch (c) {
                case 'S': is_bamin = 0; break;
-               case 'b': strcat(out_mode, "b"); break;
+               case 'b': is_bamout = 1; break;
                case 't': fn_list = strdup(optarg); is_bamin = 0; break;
                case 'h': is_header = 1; break;
                case 'H': is_header_only = 1; break;
@@ -37,12 +37,16 @@ int main_samview(int argc, char *argv[])
                case 'f': g_flag_on = strtol(optarg, 0, 0); break;
                case 'F': g_flag_off = strtol(optarg, 0, 0); break;
                case 'q': g_min_mapQ = atoi(optarg); break;
+               case 'u': is_uncompressed = 1; break;
                default: return usage();
                }
        }
+       if (is_uncompressed) is_bamout = 1;
        if (is_header_only) is_header = 1;
+       if (is_bamout) strcat(out_mode, "b");
        if (is_bamin) strcat(in_mode, "b");
        if (is_header) strcat(out_mode, "h");
+       if (is_uncompressed) strcat(out_mode, "u");
        if (argc == optind) return usage();
 
        // open file handlers
@@ -101,6 +105,7 @@ static int usage()
        fprintf(stderr, "         -h       print header for the SAM output\n");
        fprintf(stderr, "         -H       print header only (no alignments)\n");
        fprintf(stderr, "         -S       input is SAM\n");
+       fprintf(stderr, "         -u       uncompressed BAM output (force -b)\n");
        fprintf(stderr, "         -t FILE  list of reference names and lengths (force -S) [null]\n");
        fprintf(stderr, "         -o FILE  output file name [stdout]\n");
        fprintf(stderr, "         -f INT   required flag, 0 for unset [0]\n");