Initial implementation of wget, from Chip Rosenthal <chip@laserlink.net>.

Very cool.  Still lacks "chunked" transfer-coding, so not totally RFC
compliant for HTTP1.1, but very nice nonethe less, and very small.
 -Erik

6 files changed, 537 insertions, 0 deletions

diff --git a/applets/busybox.c b/applets/busybox.c
index 30ea1121d..0ffa94e09 100644
--- a/applets/busybox.c
+++ b/applets/busybox.c
@@ -328,6 +328,9 @@ const struct BB_applet applets[] = {
 #ifdef BB_WC
 	{"wc", wc_main, _BB_DIR_USR_BIN, wc_usage},
 #endif
+#ifdef BB_WGET
+	{"wget", wget_main, _BB_DIR_USR_BIN, wget_usage},
+#endif
 #ifdef BB_WHICH
 	{"which", which_main, _BB_DIR_USR_BIN, which_usage},
 #endif
diff --git a/busybox.c b/busybox.c
index 30ea1121d..0ffa94e09 100644
--- a/busybox.c
+++ b/busybox.c
@@ -328,6 +328,9 @@ const struct BB_applet applets[] = {
 #ifdef BB_WC
 	{"wc", wc_main, _BB_DIR_USR_BIN, wc_usage},
 #endif
+#ifdef BB_WGET
+	{"wget", wget_main, _BB_DIR_USR_BIN, wget_usage},
+#endif
 #ifdef BB_WHICH
 	{"which", which_main, _BB_DIR_USR_BIN, which_usage},
 #endif
diff --git a/busybox.def.h b/busybox.def.h
index 7c5e0c0d0..3ad118dd6 100644
--- a/busybox.def.h
+++ b/busybox.def.h
@@ -100,6 +100,7 @@
 #define BB_UPTIME
 #define BB_USLEEP
 #define BB_WC
+#define BB_WGET
 #define BB_WHICH
 #define BB_WHOAMI
 #define BB_UUENCODE
diff --git a/internal.h b/internal.h
index f6edb1666..33f101e09 100644
--- a/internal.h
+++ b/internal.h
@@ -213,6 +213,7 @@ extern int uniq_main(int argc, char** argv);
 extern int update_main(int argc, char** argv);
 extern int usleep_main(int argc, char** argv);
 extern int wc_main(int argc, char** argv);
+extern int wget_main(int argc, char** argv);
 extern int which_main(int argc, char** argv);
 extern int whoami_main(int argc, char** argv);
 extern int yes_main(int argc, char** argv);
@@ -318,6 +319,7 @@ extern const char usleep_usage[];
 extern const char uudecode_usage[];
 extern const char uuencode_usage[];
 extern const char wc_usage[];
+extern const char wget_usage[];
 extern const char which_usage[];
 extern const char whoami_usage[];
 extern const char yes_usage[];
diff --git a/networking/wget.c b/networking/wget.c
new file mode 100644
index 000000000..4f894fcaa
--- /dev/null
+++ b/networking/wget.c
@@ -0,0 +1,264 @@
+/* vi: set sw=4 ts=4: */
+/*
+ * wget - retrieve a file using HTTP
+ *
+ * Chip Rosenthal
+ * Covad Communications
+ * <chip@laserlink.net>
+ */
+
+#include "internal.h"
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <ctype.h>
+#include <string.h>
+
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <sys/socket.h>
+#include <netinet/in.h>
+#include <arpa/inet.h>
+#include <netdb.h>
+
+
+void parse_url(char *url, char **uri_host, int *uri_port, char **uri_path);
+FILE *open_socket(char *host, int port);
+char *gethdr(char *buf, size_t bufsiz, FILE *fp, int *istrunc);
+
+
+int wget_main(int argc, char **argv)
+{
+	FILE *sfp;					/* socket to web server				*/
+	char *uri_host, *uri_path;	/* parsed from command line url		*/
+	int uri_port;
+	char *s, buf[512];
+	int n;
+
+	char *fname_out = NULL;		/* where to direct output (-O)		*/
+	int do_continue = 0;		/* continue a prev transfer (-c)	*/
+	long beg_range = 0L;		/*   range at which continue begins	*/
+	int got_clen = 0;			/* got content-length: from server	*/
+	long clen = 1L;				/*   the content length				*/
+
+	/*
+	 * Crack command line.
+	 */
+	while ((n = getopt(argc, argv, "cO:")) != EOF) {
+		switch (n) {
+		case 'c':
+			++do_continue;
+			break;
+		case 'O':
+			fname_out = optarg;
+			break;
+		default:
+			usage(wget_usage);
+		}
+	}
+	if (do_continue && !fname_out)
+		fatalError("wget: cannot specify continue (-c) without a filename (-O)\n");
+	if (argc - optind != 1)
+			usage(wget_usage);
+	/*
+	 * Parse url into components.
+	 */
+	parse_url(argv[optind], &uri_host, &uri_port, &uri_path);
+
+	/*
+	 * Open socket to server.
+	 */
+	sfp = open_socket(uri_host, uri_port);
+
+	/*
+	 * Open the output stream.
+	 */
+	if (fname_out != NULL) {
+		/* Check if the file is supposed to go to stdout */
+		if (!strcmp(fname_out, "-") == 0) {
+			/* Nope -- so open the output file */
+			if (freopen(fname_out, (do_continue ? "a" : "w"), stdout) == NULL)
+				fatalError("wget: freopen(%s): %s\n", fname_out, strerror(errno));
+		}
+	}
+
+	/*
+	 * Determine where to start transfer.
+	 */
+	if (do_continue) {
+		struct stat sbuf;
+		if (fstat(fileno(stdout), &sbuf) < 0)
+			fatalError("wget: fstat(): %s\n", strerror(errno));
+		if (sbuf.st_size > 0)
+			beg_range = sbuf.st_size;
+		else
+			do_continue = 0;
+	}
+
+	/*
+	 * Send HTTP request.
+	 */
+	fprintf(sfp, "GET %s HTTP/1.1\r\nHost: %s\r\n", uri_path, uri_host);
+	if (do_continue)
+		fprintf(sfp, "Range: bytes=%ld-\r\n", beg_range);
+	fputs("Connection: close\r\n\r\n", sfp);
+
+	/*
+	 * Retrieve HTTP response line and check for "200" status code.
+	 */
+	if (fgets(buf, sizeof(buf), sfp) == NULL)
+		fatalError("wget: no response from server\n");
+	for (s = buf ; *s != '\0' && !isspace(*s) ; ++s)
+		;
+	for ( ; isspace(*s) ; ++s)
+		;
+	switch (atoi(s)) {
+	case 200:
+		if (!do_continue)
+			break;
+		fatalError("wget: cannot continue - server does not properly support ranges\n");
+	case 206:
+		if (do_continue)
+			break;
+		/*FALLTHRU*/
+	default:
+		fatalError("wget: server returned error: %s", buf);
+	}
+
+	/*
+	 * Retrieve HTTP headers.
+	 */
+	while ((s = gethdr(buf, sizeof(buf), sfp, &n)) != NULL) {
+		if (strcmp(buf, "content-length") == 0) {
+			clen = atol(s);
+			got_clen = 1;
+			continue;
+		}
+		if (strcmp(buf, "transfer-encoding") == 0) {
+			fatalError("wget: i do not do transfer encodings, server wants to do: %s\n", s);
+			continue;
+		}
+	}
+
+	/*
+	 * Retrieve HTTP body.
+	 */
+	while (clen > 0 && (n = fread(buf, 1, sizeof(buf), sfp)) > 0) {
+		fwrite(buf, 1, n, stdout);
+		if (got_clen)
+			clen -= n;
+	}
+	if (n == 0 && ferror(sfp))
+		fatalError("wget: network read error: %s", strerror(errno));
+
+	exit(0);
+}
+
+
+void parse_url(char *url, char **uri_host, int *uri_port, char **uri_path)
+{
+	char *s, *h;
+
+	*uri_port = 80;
+
+	if (strncmp(url, "http://", 7) != 0)
+		fatalError("wget: not an http url: %s\n", url);
+
+	/* pull the host portion to the front of the buffer */
+	for (s = url, h = url+7 ; *h != '/' ; ++h) {
+		if (*h == '\0')
+			fatalError("wget: cannot parse url: %s\n", url);
+		if (*h == ':') {
+			*uri_port = atoi(h+1);
+			*h = '\0';
+		}
+		*s++ = *h;
+	}
+	*s = '\0';
+	*uri_host = url;
+	*uri_path = h;
+}
+
+
+FILE *open_socket(char *host, int port)
+{
+	struct sockaddr_in sin;
+	struct hostent *hp;
+	int fd;
+	FILE *fp;
+
+	memzero(&sin, sizeof(sin));
+	sin.sin_family = AF_INET;
+	if ((hp = (struct hostent *) gethostbyname(host)) == NULL)
+		fatalError("wget: cannot resolve %s\n", host);
+	memcpy(&sin.sin_addr, hp->h_addr_list[0], hp->h_length);
+	sin.sin_port = htons(port);
+
+	/*
+	 * Get the server onto a stdio stream.
+	 */
+	if ((fd = socket(AF_INET, SOCK_STREAM, 0)) < 0)
+		fatalError("wget: socket(): %s\n", strerror(errno));
+	if (connect(fd, (struct sockaddr *) &sin, sizeof(sin)) < 0)
+		fatalError("wget: connect(%s): %s\n", host, strerror(errno));
+	if ((fp = fdopen(fd, "r+")) == NULL)
+		fatalError("wget: fdopen(): %s\n", strerror(errno));
+
+	return fp;
+}
+
+
+char *gethdr(char *buf, size_t bufsiz, FILE *fp, int *istrunc)
+{
+	char *s, *hdrval;
+	int c;
+
+	*istrunc = 0;
+
+	/* retrieve header line */
+	if (fgets(buf, bufsiz, fp) == NULL)
+		return NULL;
+
+	/* see if we are at the end of the headers */
+	for (s = buf ; *s == '\r' ; ++s)
+		;
+	if (s[0] == '\n')
+		return NULL;
+
+	/* convert the header name to lower case */
+	for (s = buf ; isalnum(*s) || *s == '-' ; ++s)
+		*s = tolower(*s);
+
+	/* verify we are at the end of the header name */
+	if (*s != ':')
+		fatalError("wget: bad header line: %s\n", buf);
+
+	/* locate the start of the header value */
+	for (*s++ = '\0' ; *s == ' ' || *s == '\t' ; ++s)
+		;
+	hdrval = s;
+
+	/* locate the end of header */
+	while (*s != '\0' && *s != '\r' && *s != '\n')
+		++s;
+
+	/* end of header found */
+	if (*s != '\0') {
+		*s = '\0';
+		return hdrval;
+	}
+
+	/* Rat!  The buffer isn't big enough to hold the entire header value. */
+	while (c = getc(fp), c != EOF && c != '\n')
+		;
+	*istrunc = 1;
+	return hdrval;
+}
+
+/*
+Local Variables:
+c-file-style: "linux"
+c-basic-offset: 4
+tab-width: 4
+End:
+*/
diff --git a/wget.c b/wget.c
new file mode 100644
index 000000000..4f894fcaa
--- /dev/null
+++ b/wget.c
@@ -0,0 +1,264 @@
+/* vi: set sw=4 ts=4: */
+/*
+ * wget - retrieve a file using HTTP
+ *
+ * Chip Rosenthal
+ * Covad Communications
+ * <chip@laserlink.net>
+ */
+
+#include "internal.h"
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <ctype.h>
+#include <string.h>
+
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <sys/socket.h>
+#include <netinet/in.h>
+#include <arpa/inet.h>
+#include <netdb.h>
+
+
+void parse_url(char *url, char **uri_host, int *uri_port, char **uri_path);
+FILE *open_socket(char *host, int port);
+char *gethdr(char *buf, size_t bufsiz, FILE *fp, int *istrunc);
+
+
+int wget_main(int argc, char **argv)
+{
+	FILE *sfp;					/* socket to web server				*/
+	char *uri_host, *uri_path;	/* parsed from command line url		*/
+	int uri_port;
+	char *s, buf[512];
+	int n;
+
+	char *fname_out = NULL;		/* where to direct output (-O)		*/
+	int do_continue = 0;		/* continue a prev transfer (-c)	*/
+	long beg_range = 0L;		/*   range at which continue begins	*/
+	int got_clen = 0;			/* got content-length: from server	*/
+	long clen = 1L;				/*   the content length				*/
+
+	/*
+	 * Crack command line.
+	 */
+	while ((n = getopt(argc, argv, "cO:")) != EOF) {
+		switch (n) {
+		case 'c':
+			++do_continue;
+			break;
+		case 'O':
+			fname_out = optarg;
+			break;
+		default:
+			usage(wget_usage);
+		}
+	}
+	if (do_continue && !fname_out)
+		fatalError("wget: cannot specify continue (-c) without a filename (-O)\n");
+	if (argc - optind != 1)
+			usage(wget_usage);
+	/*
+	 * Parse url into components.
+	 */
+	parse_url(argv[optind], &uri_host, &uri_port, &uri_path);
+
+	/*
+	 * Open socket to server.
+	 */
+	sfp = open_socket(uri_host, uri_port);
+
+	/*
+	 * Open the output stream.
+	 */
+	if (fname_out != NULL) {
+		/* Check if the file is supposed to go to stdout */
+		if (!strcmp(fname_out, "-") == 0) {
+			/* Nope -- so open the output file */
+			if (freopen(fname_out, (do_continue ? "a" : "w"), stdout) == NULL)
+				fatalError("wget: freopen(%s): %s\n", fname_out, strerror(errno));
+		}
+	}
+
+	/*
+	 * Determine where to start transfer.
+	 */
+	if (do_continue) {
+		struct stat sbuf;
+		if (fstat(fileno(stdout), &sbuf) < 0)
+			fatalError("wget: fstat(): %s\n", strerror(errno));
+		if (sbuf.st_size > 0)
+			beg_range = sbuf.st_size;
+		else
+			do_continue = 0;
+	}
+
+	/*
+	 * Send HTTP request.
+	 */
+	fprintf(sfp, "GET %s HTTP/1.1\r\nHost: %s\r\n", uri_path, uri_host);
+	if (do_continue)
+		fprintf(sfp, "Range: bytes=%ld-\r\n", beg_range);
+	fputs("Connection: close\r\n\r\n", sfp);
+
+	/*
+	 * Retrieve HTTP response line and check for "200" status code.
+	 */
+	if (fgets(buf, sizeof(buf), sfp) == NULL)
+		fatalError("wget: no response from server\n");
+	for (s = buf ; *s != '\0' && !isspace(*s) ; ++s)
+		;
+	for ( ; isspace(*s) ; ++s)
+		;
+	switch (atoi(s)) {
+	case 200:
+		if (!do_continue)
+			break;
+		fatalError("wget: cannot continue - server does not properly support ranges\n");
+	case 206:
+		if (do_continue)
+			break;
+		/*FALLTHRU*/
+	default:
+		fatalError("wget: server returned error: %s", buf);
+	}
+
+	/*
+	 * Retrieve HTTP headers.
+	 */
+	while ((s = gethdr(buf, sizeof(buf), sfp, &n)) != NULL) {
+		if (strcmp(buf, "content-length") == 0) {
+			clen = atol(s);
+			got_clen = 1;
+			continue;
+		}
+		if (strcmp(buf, "transfer-encoding") == 0) {
+			fatalError("wget: i do not do transfer encodings, server wants to do: %s\n", s);
+			continue;
+		}
+	}
+
+	/*
+	 * Retrieve HTTP body.
+	 */
+	while (clen > 0 && (n = fread(buf, 1, sizeof(buf), sfp)) > 0) {
+		fwrite(buf, 1, n, stdout);
+		if (got_clen)
+			clen -= n;
+	}
+	if (n == 0 && ferror(sfp))
+		fatalError("wget: network read error: %s", strerror(errno));
+
+	exit(0);
+}
+
+
+void parse_url(char *url, char **uri_host, int *uri_port, char **uri_path)
+{
+	char *s, *h;
+
+	*uri_port = 80;
+
+	if (strncmp(url, "http://", 7) != 0)
+		fatalError("wget: not an http url: %s\n", url);
+
+	/* pull the host portion to the front of the buffer */
+	for (s = url, h = url+7 ; *h != '/' ; ++h) {
+		if (*h == '\0')
+			fatalError("wget: cannot parse url: %s\n", url);
+		if (*h == ':') {
+			*uri_port = atoi(h+1);
+			*h = '\0';
+		}
+		*s++ = *h;
+	}
+	*s = '\0';
+	*uri_host = url;
+	*uri_path = h;
+}
+
+
+FILE *open_socket(char *host, int port)
+{
+	struct sockaddr_in sin;
+	struct hostent *hp;
+	int fd;
+	FILE *fp;
+
+	memzero(&sin, sizeof(sin));
+	sin.sin_family = AF_INET;
+	if ((hp = (struct hostent *) gethostbyname(host)) == NULL)
+		fatalError("wget: cannot resolve %s\n", host);
+	memcpy(&sin.sin_addr, hp->h_addr_list[0], hp->h_length);
+	sin.sin_port = htons(port);
+
+	/*
+	 * Get the server onto a stdio stream.
+	 */
+	if ((fd = socket(AF_INET, SOCK_STREAM, 0)) < 0)
+		fatalError("wget: socket(): %s\n", strerror(errno));
+	if (connect(fd, (struct sockaddr *) &sin, sizeof(sin)) < 0)
+		fatalError("wget: connect(%s): %s\n", host, strerror(errno));
+	if ((fp = fdopen(fd, "r+")) == NULL)
+		fatalError("wget: fdopen(): %s\n", strerror(errno));
+
+	return fp;
+}
+
+
+char *gethdr(char *buf, size_t bufsiz, FILE *fp, int *istrunc)
+{
+	char *s, *hdrval;
+	int c;
+
+	*istrunc = 0;
+
+	/* retrieve header line */
+	if (fgets(buf, bufsiz, fp) == NULL)
+		return NULL;
+
+	/* see if we are at the end of the headers */
+	for (s = buf ; *s == '\r' ; ++s)
+		;
+	if (s[0] == '\n')
+		return NULL;
+
+	/* convert the header name to lower case */
+	for (s = buf ; isalnum(*s) || *s == '-' ; ++s)
+		*s = tolower(*s);
+
+	/* verify we are at the end of the header name */
+	if (*s != ':')
+		fatalError("wget: bad header line: %s\n", buf);
+
+	/* locate the start of the header value */
+	for (*s++ = '\0' ; *s == ' ' || *s == '\t' ; ++s)
+		;
+	hdrval = s;
+
+	/* locate the end of header */
+	while (*s != '\0' && *s != '\r' && *s != '\n')
+		++s;
+
+	/* end of header found */
+	if (*s != '\0') {
+		*s = '\0';
+		return hdrval;
+	}
+
+	/* Rat!  The buffer isn't big enough to hold the entire header value. */
+	while (c = getc(fp), c != EOF && c != '\n')
+		;
+	*istrunc = 1;
+	return hdrval;
+}
+
+/*
+Local Variables:
+c-file-style: "linux"
+c-basic-offset: 4
+tab-width: 4
+End:
+*/