From adcd4d3c1ef2ba35730e0f9a837ed44cf6d74d51 Mon Sep 17 00:00:00 2001
From: Calvin Morrison <mutantturkey@gmail.com>
Date: Thu, 27 Mar 2014 17:19:22 -0400
Subject: add strstream

---
 Makefile    |   4 ++-
 strstream.c | 111 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 114 insertions(+), 1 deletion(-)
 create mode 100644 strstream.c

diff --git a/Makefile b/Makefile
index 8193e1d..9826622 100644
--- a/Makefile
+++ b/Makefile
@@ -1,7 +1,7 @@
 CC = gcc
 CFLAGS= -O3 -s -mtune=native -Wextra -Wall 
 
-all: count_nucleobases sequence_lengths
+all: count_nucleobases sequence_lengths strstream
 
 count_nucleobases: count_nucleobases.c
 	$(CC) $(CFLAGS) count_nucleobases.c -o count_nucleobases
@@ -9,6 +9,8 @@ sequence_lengths: sequence_lengths.c
 	$(CC) $(CFLAGS) sequence_lengths.c -o sequence_lengths
 sequence_end_points: sequence_end_points.c
 	$(CC) $(CFLAGS) sequence_end_points.c -o sequence_end_points
+strstream: strstream.c
+	$(CC) $(CFLAGS) strstream.c -o strstream
 
 clean:
 	rm -vf count_nucleobases sequence_lengths sequence_end_points
diff --git a/strstream.c b/strstream.c
new file mode 100644
index 0000000..19c7614
--- /dev/null
+++ b/strstream.c
@@ -0,0 +1,111 @@
+// find the positions of any strings in the file pointed by argv[1] in stdin
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <unistd.h>
+
+char **load_mers_from_file(FILE *fh, ssize_t *len) {
+	char line[4096];
+
+	size_t realloc_size = 1;
+	*len = 0;
+
+	char **mers = NULL;
+
+	while((fgets(line, 4096, fh)) != NULL) { 
+		size_t line_len = strlen(line);
+		if(line_len == 0) 
+			continue;
+
+		line[line_len - 1] = '\0';
+
+		mers = realloc(mers, sizeof(char *) * realloc_size);
+		if(mers == NULL) {
+			fprintf(stderr, "could not realloc mers\n");
+			exit(EXIT_FAILURE);
+		}
+
+		char *cpy = malloc(line_len + 1);
+		if(cpy == NULL) {
+			fprintf(stderr, "could not alloc mers\n");
+			exit(EXIT_FAILURE);
+		}
+		strncpy(cpy, line, line_len);
+		mers[*len] = cpy;
+		*len += 1;
+
+		realloc_size++;
+	}
+
+	if(*len != 0)
+		return mers;
+
+	*len = -1;
+	return NULL;
+}
+
+int main(int argc, char **argv){
+
+	char buffer[BUFSIZ] = { 0 };
+	char *buf, *start;
+	ssize_t len = 0;
+	ssize_t mer_len = 0;
+
+	int save_size = 0;
+	int cpy = 0;
+
+	unsigned long long pos = 0;
+	unsigned long long cpy_size = 0; 
+
+	int i = 0;
+
+	if(argc != 2)  {
+		fprintf(stderr, "usage: strstream merlist.txt\n");
+		exit(EXIT_FAILURE);
+	}
+
+	// load mers
+	FILE *fh = fopen(argv[1], "r");
+	if(fh == NULL) {
+		fprintf(stderr, "could not open %s\n", argv[1]);
+		exit(EXIT_FAILURE);
+	}
+	char **mers = load_mers_from_file(fh, &mer_len);
+	if(mers == NULL) {
+		fprintf(stderr, "could not load mers from %s\n", argv[1]);
+		exit(EXIT_FAILURE);
+	}
+
+	// get max argument length
+	for(i = 0; i < mer_len; i++) {
+		int current_len = strlen(mers[i]);
+		if( current_len > save_size)
+			save_size = current_len;
+	}
+
+	cpy = save_size - 1;
+	cpy_size = BUFSIZ - cpy;
+
+	buf = buffer;
+	start = buf + cpy;
+
+	// copy our first cpy length into the first part of our buffer
+	len = fread(buffer, 1, cpy, stdin);
+	if(len == 0)
+		exit(EXIT_FAILURE);
+
+	// read into "start" (buf + cpy) from tdin
+	while((len = fread(start, 1, cpy_size, stdin)) != 0) {
+		for(i = 0; i < mer_len; i++) {
+			char *p = buffer;
+				while((p = strstr(p, mers[i])) != NULL) {
+				printf("%d %llu\n", i, pos + (p - buffer));
+				p++;
+			}
+		}
+		memcpy(buffer, buffer + len, cpy);
+		pos = pos + len;
+	}
+
+	return 0;
+}
-- 
cgit v1.2.1