Merge pull request attractivechaos#173 from jmarshall/memchr

Apply seqtk PR to improve kseq.h parsing performance

Author: attractivechaos

kseq.h

@@ -107,12 +107,12 @@ typedef struct __kstring_t {
 					if (ks->end == -1) { ks->is_eof = 1; return -3; }	\
 				} else break;											\
 			}															\
-			if (delimiter == KS_SEP_LINE) { \
-				for (i = ks->begin; i < ks->end; ++i) \
-					if (ks->buf[i] == '\n') break; \
+			if (delimiter == KS_SEP_LINE) {								\
+				unsigned char *sep = memchr(ks->buf + ks->begin, '\n', ks->end - ks->begin); \
+				i = sep != NULL ? sep - ks->buf : ks->end;				\
 			} else if (delimiter > KS_SEP_MAX) {						\
-				for (i = ks->begin; i < ks->end; ++i)					\
-					if (ks->buf[i] == delimiter) break;					\
+				unsigned char *sep = memchr(ks->buf + ks->begin, delimiter, ks->end - ks->begin); \
+				i = sep != NULL ? sep - ks->buf : ks->end;				\
 			} else if (delimiter == KS_SEP_SPACE) {						\
 				for (i = ks->begin; i < ks->end; ++i)					\
 					if (isspace(ks->buf[i])) break;						\

test/kseq_bench2.c

@@ -3,6 +3,7 @@
 #include <stdint.h>
 #include <stdlib.h>
 #include <fcntl.h>
+#include <unistd.h>
 #include "kseq.h"
 KSTREAM_INIT(int, read, 4096)