X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=knetfile.c;h=9029ee4c6f93fecd353866ad6a3cbc6cb723c0e5;hb=e7c6e7a7c181d8ee9d1a206d264a24b54b4911f4;hp=d30fcbbb0617f0e86d255ed2261f1891c8426b0f;hpb=cd86e7be490dcde2d2166caaf3b61de38417ba18;p=samtools.git

diff --git a/knetfile.c b/knetfile.c
index d30fcbb..9029ee4 100644
--- a/knetfile.c
+++ b/knetfile.c
@@ -15,14 +15,15 @@ static int kftp_get_response(knetFile *ftp)
 	int n = 0;
 	char *p;
 	while (read(ftp->ctrl_fd, &c, 1)) { // FIXME: this is *VERY BAD* for unbuffered I/O
-//		fputc(c, stderr);
+		//fputc(c, stderr);
 		if (n >= ftp->max_response) {
 			ftp->max_response = ftp->max_response? ftp->max_response<<1 : 256;
 			ftp->response = realloc(ftp->response, ftp->max_response);
 		}
 		ftp->response[n++] = c;
 		if (c == '\n') {
-			if (n >= 4 && isdigit(ftp->response[0]) && isdigit(ftp->response[1]) && isdigit(ftp->response[2]) && ftp->response[3] != '-') break;
+			if (n >= 4 && isdigit(ftp->response[0]) && isdigit(ftp->response[1]) && isdigit(ftp->response[2])
+				&& ftp->response[3] != '-') break;
 			n = 0;
 			continue;
 		}
@@ -58,7 +59,7 @@ static int kftp_pasv_connect(knetFile *ftp)
 
 	struct addrinfo hints, *res;
 	struct linger lng = { 0, 0 };
-	int on;
+	int on = 1;
 	char host[80], port[10];
 
 	if (ftp->pasv_port == 0) {
@@ -83,7 +84,7 @@ int kftp_connect(knetFile *ftp)
 {
 #define __err_connect(func) do { perror(func); return -1; } while (0)
 
-	int on;
+	int on = 1;
 	{ // open socket
 		struct addrinfo hints, *res;
 		memset(&hints, 0, sizeof(struct addrinfo));
@@ -106,16 +107,16 @@ int kftp_connect(knetFile *ftp)
 
 int kftp_reconnect(knetFile *ftp)
 {
-	if (ftp->ctrl_fd) {
+	if (ftp->ctrl_fd >= 0) {
 		close(ftp->ctrl_fd);
-		ftp->ctrl_fd = 0;
+		ftp->ctrl_fd = -1;
 	}
 	close(ftp->fd);
 	return kftp_connect(ftp);
 }
 
 // initialize ->type, ->host and ->retr
-knetFile *kftp_prep(const char *fn, const char *mode)
+knetFile *kftp_parse_url(const char *fn, const char *mode)
 {
 	knetFile *fp;
 	char *p;
@@ -126,30 +127,39 @@ knetFile *kftp_prep(const char *fn, const char *mode)
 	l = p - fn - 6;
 	fp = calloc(1, sizeof(knetFile));
 	fp->type = KNF_TYPE_FTP;
+	fp->fd = -1;
 	fp->host = calloc(l + 1, 1);
 	if (strchr(mode, 'c')) fp->no_reconnect = 1;
 	strncpy(fp->host, fn + 6, l);
-	fp->retr = calloc(strlen(p) + 7, 1);
+	fp->retr = calloc(strlen(p) + 8, 1);
 	sprintf(fp->retr, "RETR %s\r\n", p);
+	fp->seek_offset = -1;
 	return fp;
 }
 // place ->fd at offset off
-int kftp_connect_file(knetFile *fp, off_t off)
+int kftp_connect_file(knetFile *fp)
 {
-	if (fp->fd) {
+	int ret;
+	if (fp->fd >= 0) {
 		close(fp->fd);
 		if (fp->no_reconnect) kftp_get_response(fp);
 	}
 	kftp_pasv_prep(fp);
-	if (off) {
+	if (fp->offset) {
 		char tmp[32];
-		sprintf(tmp, "REST %lld\r\n", (long long)off);
+		sprintf(tmp, "REST %lld\r\n", (long long)fp->offset);
 		kftp_send_cmd(fp, tmp, 1);
 	}
 	kftp_send_cmd(fp, fp->retr, 0);
 	kftp_pasv_connect(fp);
-	kftp_get_response(fp);
-	fp->offset = off;
+	ret = kftp_get_response(fp);
+	if (ret != 150) {
+		fprintf(stderr, "[kftp_connect_file] %s\n", fp->response);
+		close(fp->fd);
+		fp->fd = -1;
+		return -1;
+	}
+	fp->is_ready = 1;
 	return 0;
 }
 
@@ -161,13 +171,17 @@ knetFile *knet_open(const char *fn, const char *mode)
 		return 0;
 	}
 	if (strstr(fn, "ftp://") == fn) {
-		fp = kftp_prep(fn, mode);
+		fp = kftp_parse_url(fn, mode);
 		if (fp == 0) return 0;
 		if (kftp_connect(fp) == -1) {
 			knet_close(fp);
 			return 0;
 		}
-		kftp_connect_file(fp, 0);
+		kftp_connect_file(fp);
+		if (fp->fd < 0) {
+			knet_close(fp);
+			return 0;
+		}
 	} else {
 		int fd = open(fn, O_RDONLY);
 		if (fd == -1) {
@@ -181,36 +195,65 @@ knetFile *knet_open(const char *fn, const char *mode)
 	return fp;
 }
 
+knetFile *knet_dopen(int fd, const char *mode)
+{
+	knetFile *fp = (knetFile*)calloc(1, sizeof(knetFile));
+	fp->type = KNF_TYPE_LOCAL;
+	fp->fd = fd;
+	return fp;
+}
+
 off_t knet_read(knetFile *fp, void *buf, off_t len)
 {
-	off_t l = read(fp->fd, buf, len);
-	fp->offset += l;
+	off_t l = 0;
+	if (fp->fd < 0) return 0;
+	if (fp->type == KNF_TYPE_LOCAL) {
+		l = read(fp->fd, buf, len);
+		fp->offset += l;
+	} else {
+		off_t rest = len, curr;
+		if (fp->is_ready == 0) {
+			if (!fp->no_reconnect) kftp_reconnect(fp);
+			kftp_connect_file(fp);
+			fp->is_ready = 1;
+		}
+		while (rest) {
+			curr = read(fp->fd, buf + l, rest);
+			if (curr == 0) break; // FIXME: end of file or bad network? I do not know...
+			l += curr; rest -= curr;
+		}
+		fp->offset += l;
+	}
 	return l;
 }
 
-off_t knet_seek(knetFile *fp, off_t off, int whence)
+int knet_seek(knetFile *fp, off_t off, int whence)
 {
 	if (fp->type == KNF_TYPE_LOCAL) {
-		fp->offset = lseek(fp->fd, off, whence);
-		return fp->offset;
+		if (lseek(fp->fd, off, whence) == -1) {
+			perror("lseek");
+			return -1;
+		}
+		fp->offset = off;
+		return 0;
 	}
 	if (fp->type == KNF_TYPE_FTP) {
-		if (whence != SEEK_SET) {
+		if (whence != SEEK_SET) { // FIXME: we can surely allow SEEK_CUR and SEEK_END in future
 			fprintf(stderr, "[knet_seek] only SEEK_SET is supported for FTP. Offset is unchanged.\n");
 			return -1;
 		}
-		if (!fp->no_reconnect) kftp_reconnect(fp);
-		kftp_connect_file(fp, off);
-		return fp->offset;
+		fp->offset = off;
+		fp->is_ready = 0;
+		return 0;
 	}
-	return 0;
+	return -1;
 }
 
 int knet_close(knetFile *fp)
 {
 	if (fp == 0) return 0;
-	if (fp->ctrl_fd > 0) close(fp->ctrl_fd);
-	if (fp->fd > 0) close(fp->fd);
+	if (fp->ctrl_fd >= 0) close(fp->ctrl_fd);
+	if (fp->fd >= 0) close(fp->fd);
 	free(fp->response); free(fp->retr); free(fp->host);
 	free(fp);
 	return 0;
@@ -222,8 +265,8 @@ int main(void)
 	char buf[256];
 	knetFile *fp;
 //	fp = knet_open("ftp://ftp.ncbi.nih.gov/1000genomes/ftp/data/NA12878/alignment/NA12878.chrom6.SLX.SRP000032.2009_06.bam", "r"); knet_seek(fp, 2500000000ll, SEEK_SET);
-//	fp = knet_open("ftp://ftp.sanger.ac.uk/pub4/treefam/tmp/index.shtml", "r"); knet_seek(fp, 2000, SEEK_SET);
-	fp = knet_open("knetfile.c", "r"); knet_seek(fp, 2000, SEEK_SET);
+	fp = knet_open("ftp://ftp.sanger.ac.uk/pub4/treefam/tmp/index.shtml", "r"); knet_seek(fp, 2000, SEEK_SET);
+//	fp = knet_open("knetfile.c", "r"); knet_seek(fp, 2000, SEEK_SET);
 	knet_read(fp, buf, 255);
 	buf[255] = 0;
 	printf("%s\n", buf);