greg/pcompress
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Use Libbsc for DNA Sequence data instead of PPMD. Faster, better compression.

Browse source 
Fix pz extension handling for real.
This commit is contained in:
Moinak Ghosh 2013-11-30 09:58:21 +05:30
parent dfeea8c19b
commit c4c4b47138
4 changed files with 12 additions and 8 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
adaptive_compress.c
View file

@ -288,7 +288,7 @@ adapt_compress(void *src, uint64_t srclen, void *dst,
} else {
#ifdef ENABLE_PC_LIBBSC
if (adat->bsc_data && (PC_SUBTYPE(btype) == TYPE_MARKUP ||
PC_SUBTYPE(btype) == TYPE_BMP)) {
PC_SUBTYPE(btype) == TYPE_BMP || PC_SUBTYPE(btype) == TYPE_DNA_SEQ)) {
rv = libbsc_compress(src, srclen, dst, dstlen, level, chdr, btype, adat->bsc_data);
if (rv < 0)
return (rv);

8
pcompress.c
View file

@ -2050,7 +2050,12 @@ start_compress(pc_ctx_t *pctx, const char *filename, uint64_t chunksize, int lev
}
add_fname(tmpfile1);
} else {
snprintf(to_filename, sizeof (to_filename), "%s" COMP_EXTN, pctx->to_filename);
if (!endswith(pctx->to_filename, COMP_EXTN))
snprintf(to_filename, sizeof (to_filename),
"%s" COMP_EXTN, pctx->to_filename);
else
snprintf(to_filename, sizeof (to_filename),
"%s", pctx->to_filename);
if ((compfd = open(to_filename, O_CREAT|O_RDWR, S_IRUSR|S_IWUSR)) == -1) {
log_msg(LOG_ERR, 1, "open ");
COMP_BAIL;
@ -3131,7 +3136,6 @@ init_pc_context(pc_ctx_t *pctx, int argc, char *argv[])
}
} else {
strcpy(apath, pctx->filename);
if (!endswith(apath, COMP_EXTN))
strcat(apath, COMP_EXTN);
pctx->to_filename = realpath(apath, NULL);

4
utils/phash/extensions.h
View file

@ -69,11 +69,11 @@ struct ext_entry {
{"lua" , TYPE_TEXT, 3},
{"qml" , TYPE_TEXT|TYPE_MARKUP, 3},
{"fa" , TYPE_TEXT|TYPE_DNA_SEQ, 2},
{"faa" , TYPE_TEXT, 3},
{"faa" , TYPE_TEXT|TYPE_DNA_SEQ, 3},
{"asn" , TYPE_TEXT|TYPE_MARKUP, 3},
{"ffn" , TYPE_TEXT|TYPE_DNA_SEQ, 3},
{"fna" , TYPE_TEXT|TYPE_DNA_SEQ, 3},
{"frn" , TYPE_TEXT, 3},
{"frn" , TYPE_TEXT|TYPE_DNA_SEQ, 3},
{"gbk" , TYPE_TEXT, 3},
{"gff" , TYPE_TEXT, 3},
{"ptt" , TYPE_TEXT, 3},

4
utils/phash/extensions.txt
View file

@ -58,11 +58,11 @@ qml,TYPE_TEXT|TYPE_MARKUP
# These are all genomic data file extensions
fa,TYPE_TEXT|TYPE_DNA_SEQ
faa,TYPE_TEXT
faa,TYPE_TEXT|TYPE_DNA_SEQ
asn,TYPE_TEXT|TYPE_MARKUP
ffn,TYPE_TEXT|TYPE_DNA_SEQ
fna,TYPE_TEXT|TYPE_DNA_SEQ
frn,TYPE_TEXT
frn,TYPE_TEXT|TYPE_DNA_SEQ
gbk,TYPE_TEXT
gff,TYPE_TEXT
ptt,TYPE_TEXT