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");
}
int reheader_file(const char *header, const char *file, int meta)
{
BGZF *fp = bgzf_open(file,"r");
if (bgzf_read_block(fp) != 0 || !fp->block_length)
return -1;
char *buffer = fp->uncompressed_block;
int skip_until = 0;
if ( buffer[0]==meta )
{
skip_until = 1;
// Skip the header
while (1)
{
if ( buffer[skip_until]=='\n' )
{
skip_until++;
if ( skip_until>=fp->block_length )
{
if (bgzf_read_block(fp) != 0 || !fp->block_length)
error("no body?\n");
skip_until = 0;
}
// The header has finished
if ( buffer[skip_until]!=meta ) break;
}
skip_until++;
if ( skip_until>=fp->block_length )
{
if (bgzf_read_block(fp) != 0 || !fp->block_length)
error("no body?\n");
skip_until = 0;
}
}
}
FILE *fh = fopen(header,"r");
if ( !fh )
error("%s: %s", header,strerror(errno));
int page_size = getpagesize();
char *buf = valloc(page_size);
BGZF *bgzf_out = bgzf_fdopen(fileno(stdout), "w");
ssize_t nread;
while ( (nread=fread(buf,1,page_size-1,fh))>0 )
{
if ( nread<page_size-1 && buf[nread-1]!='\n' )
buf[nread++] = '\n';
if (bgzf_write(bgzf_out, buf, nread) < 0) error("Error: %s\n",bgzf_out->error);
}
fclose(fh);
if ( fp->block_length - skip_until > 0 )
{
if (bgzf_write(bgzf_out, buffer+skip_until, fp->block_length-skip_until) < 0)
error("Error: %s\n",fp->error);
}
if (bgzf_flush(bgzf_out) < 0)
error("Error: %s\n",bgzf_out->error);
while (1)
{
#ifdef _USE_KNETFILE
nread = knet_read(fp->x.fpr, buf, page_size);
#else
nread = fread(buf, 1, page_size, fp->file);
#endif
if ( nread<=0 )
break;
#ifdef _USE_KNETFILE
int count = fwrite(buf, 1, nread, bgzf_out->x.fpw);
#else
int count = fwrite(buf, 1, nread, bgzf_out->file);
#endif
if (count != nread)
error("Write failed, wrote %d instead of %d bytes.\n", count,(int)nread);
}
if (bgzf_close(bgzf_out) < 0)
error("Error: %s\n",bgzf_out->error);
return 0;
}
int bam_cat(int nfn, char * const *fn, const bam_header_t *h, const char* outbam)
{
BGZF *fp;
FILE* fp_file;
uint8_t *buf;
uint8_t ebuf[BGZF_EMPTY_BLOCK_SIZE];
const int es=BGZF_EMPTY_BLOCK_SIZE;
int i;
fp = strcmp(outbam, "-")? bgzf_open(outbam, "w") : bgzf_fdopen(_fileno(stdout), "w");
if (fp == 0) {
fprintf(stderr, "[%s] ERROR: fail to open output file '%s'.\n", __FUNCTION__, outbam);
return 1;
}
if (h) bam_header_write(fp, h);
buf = (uint8_t*) malloc(BUF_SIZE);
for(i = 0; i < nfn; ++i){
BGZF *in;
bam_header_t *old;
int len,j;
in = strcmp(fn[i], "-")? bam_open(fn[i], "r") : bam_dopen(_fileno(stdin), "r");
if (in == 0) {
fprintf(stderr, "[%s] ERROR: fail to open file '%s'.\n", __FUNCTION__, fn[i]);
return -1;
}
if (in->open_mode != 'r') return -1;
old = bam_header_read(in);
if (h == 0 && i == 0) bam_header_write(fp, old);
if (in->block_offset < in->block_length) {
bgzf_write(fp, (uint8_t*)in->uncompressed_block + in->block_offset, in->block_length - in->block_offset);
bgzf_flush(fp);
}
j=0;
#ifdef _USE_KNETFILE
fp_file=fp->x.fpw;
while ((len = knet_read(in->x.fpr, buf, BUF_SIZE)) > 0) {
#else
fp_file=fp->file;
while (!feof(in->file) && (len = fread(buf, 1, BUF_SIZE, in->file)) > 0) {
#endif
if(len<es){
int diff=es-len;
if(j==0) {
fprintf(stderr, "[%s] ERROR: truncated file?: '%s'.\n", __FUNCTION__, fn[i]);
return -1;
}
fwrite(ebuf, 1, len, fp_file);
memcpy(ebuf,ebuf+len,diff);
memcpy(ebuf+diff,buf,len);
} else {
if(j!=0) fwrite(ebuf, 1, es, fp_file);
len-= es;
memcpy(ebuf,buf+len,es);
fwrite(buf, 1, len, fp_file);
}
j=1;
}
/* check final gzip block */
{
const uint8_t gzip1=ebuf[0];
const uint8_t gzip2=ebuf[1];
const uint32_t isize=*((uint32_t*)(ebuf+es-4));
if(((gzip1!=GZIPID1) || (gzip2!=GZIPID2)) || (isize!=0)) {
fprintf(stderr, "[%s] WARNING: Unexpected block structure in file '%s'.", __FUNCTION__, fn[i]);
fprintf(stderr, " Possible output corruption.\n");
fwrite(ebuf, 1, es, fp_file);
}
}
bam_header_destroy(old);
bgzf_close(in);
}
free(buf);
bgzf_close(fp);
return 0;
}
int main_cat(int argc, char *argv[])
{
bam_header_t *h = 0;
char *outfn = 0;
int c, ret;
while ((c = getopt(argc, argv, "h:o:")) >= 0) {
switch (c) {
case 'h': {
tamFile fph = sam_open(optarg);
if (fph == 0) {
fprintf(stderr, "[%s] ERROR: fail to read the header from '%s'.\n", __FUNCTION__, argv[1]);
return 1;
}
h = sam_header_read(fph);
sam_close(fph);
break;
}
case 'o': outfn = strdup(optarg); break;
}
}
if (argc - optind < 2) {
fprintf(stderr, "Usage: samtools cat [-h header.sam] [-o out.bam] <in1.bam> <in2.bam> [...]\n");
return 1;
}
ret = bam_cat(argc - optind, argv + optind, h, outfn? outfn : "-");
free(outfn);
return ret;
}
static ssize_t net_read(hFILE *fpv, void *buffer, size_t nbytes)
{
hFILE_net *fp = (hFILE_net *) fpv;
return knet_read(fp->netfp, buffer, nbytes);
}
