/* HTTP support.
- Copyright (C) 2003 Free Software Foundation, Inc.
+ Copyright (C) 2005 Free Software Foundation, Inc.
This file is part of GNU Wget.
\f
static int cookies_loaded_p;
-struct cookie_jar *wget_cookie_jar;
+static struct cookie_jar *wget_cookie_jar;
#define TEXTHTML_S "text/html"
#define TEXTXHTML_S "application/xhtml+xml"
}
\f
static const char *
-head_terminator (const char *hunk, int oldlen, int peeklen)
+response_head_terminator (const char *hunk, int oldlen, int peeklen)
{
const char *start, *end;
return NULL;
}
+/* The maximum size of a single HTTP response we care to read. This
+ is not meant to impose an arbitrary limit, but to protect the user
+ from Wget slurping up available memory upon encountering malicious
+ or buggy server output. Define it to 0 to remove the limit. */
+
+#define HTTP_RESPONSE_MAX_SIZE 65536
+
/* Read the HTTP request head from FD and return it. The error
conditions are the same as with fd_read_hunk.
data can be treated as body. */
static char *
-fd_read_http_head (int fd)
+read_http_response_head (int fd)
{
- return fd_read_hunk (fd, head_terminator, 512);
+ return fd_read_hunk (fd, response_head_terminator, 512,
+ HTTP_RESPONSE_MAX_SIZE);
}
struct response {
/* Create a new response object from the text of the HTTP response,
available in HEAD. That text is automatically split into
constituent header lines for fast retrieval using
- response_header_*. */
+ resp_header_*. */
static struct response *
-response_new (const char *head)
+resp_new (const char *head)
{
const char *hdr;
int count, size;
return resp;
}
- /* Split HEAD into header lines, so that response_header_* functions
+ /* Split HEAD into header lines, so that resp_header_* functions
don't need to do this over and over again. */
size = count = 0;
return resp;
}
-/* Locate the header named NAME in the request data. If found, set
- *BEGPTR to its starting, and *ENDPTR to its ending position, and
- return 1. Otherwise return 0.
+/* Locate the header named NAME in the request data, starting with
+ position START. This allows the code to loop through the request
+ data, filtering for all requests of a given name. Returns the
+ found position, or -1 for failure. The code that uses this
+ function typically looks like this:
+
+ for (pos = 0; (pos = resp_header_locate (...)) != -1; pos++)
+ ... do something with header ...
- This function is used as a building block for response_header_copy
- and response_header_strdup. */
+ If you only care about one header, use resp_header_get instead of
+ this function. */
static int
-response_header_bounds (const struct response *resp, const char *name,
- const char **begptr, const char **endptr)
+resp_header_locate (const struct response *resp, const char *name, int start,
+ const char **begptr, const char **endptr)
{
int i;
const char **headers = resp->headers;
int name_len;
if (!headers || !headers[1])
- return 0;
+ return -1;
name_len = strlen (name);
+ if (start > 0)
+ i = start;
+ else
+ i = 1;
- for (i = 1; headers[i + 1]; i++)
+ for (; headers[i + 1]; i++)
{
const char *b = headers[i];
const char *e = headers[i + 1];
--e;
*begptr = b;
*endptr = e;
- return 1;
+ return i;
}
}
- return 0;
+ return -1;
+}
+
+/* Find and retrieve the header named NAME in the request data. If
+ found, set *BEGPTR to its starting, and *ENDPTR to its ending
+ position, and return 1. Otherwise return 0.
+
+ This function is used as a building block for resp_header_copy
+ and resp_header_strdup. */
+
+static int
+resp_header_get (const struct response *resp, const char *name,
+ const char **begptr, const char **endptr)
+{
+ int pos = resp_header_locate (resp, name, 0, begptr, endptr);
+ return pos != -1;
}
/* Copy the response header named NAME to buffer BUF, no longer than
BUFSIZE (BUFSIZE includes the terminating 0). If the header
exists, 1 is returned, otherwise 0. If there should be no limit on
- the size of the header, use response_header_strdup instead.
+ the size of the header, use resp_header_strdup instead.
If BUFSIZE is 0, no data is copied, but the boolean indication of
whether the header is present is still returned. */
static int
-response_header_copy (const struct response *resp, const char *name,
- char *buf, int bufsize)
+resp_header_copy (const struct response *resp, const char *name,
+ char *buf, int bufsize)
{
const char *b, *e;
- if (!response_header_bounds (resp, name, &b, &e))
+ if (!resp_header_get (resp, name, &b, &e))
return 0;
if (bufsize)
{
malloc. If such a header does not exist in RESP, return NULL. */
static char *
-response_header_strdup (const struct response *resp, const char *name)
+resp_header_strdup (const struct response *resp, const char *name)
{
const char *b, *e;
- if (!response_header_bounds (resp, name, &b, &e))
+ if (!resp_header_get (resp, name, &b, &e))
return NULL;
return strdupdelim (b, e);
}
returned in *MESSAGE. */
static int
-response_status (const struct response *resp, char **message)
+resp_status (const struct response *resp, char **message)
{
int status;
const char *p, *end;
/* Release the resources used by RESP. */
static void
-response_free (struct response *resp)
+resp_free (struct response *resp)
{
xfree_null (resp->headers);
xfree (resp);
is done. */
int keep_alive;
- /* Whether keep-alive should be inhibited. */
- int inhibit_keep_alive = !opt.http_keep_alive || opt.ignore_length;
+ /* Whether keep-alive should be inhibited.
+
+ RFC 2068 requests that 1.0 clients not send keep-alive requests
+ to proxies. This is because many 1.0 proxies do not interpret
+ the Connection header and transfer it to the remote server,
+ causing it to not close the connection and leave both the proxy
+ and the client hanging. */
+ int inhibit_keep_alive =
+ !opt.http_keep_alive || opt.ignore_length /*|| proxy != NULL*/;
/* Headers sent when using POST. */
wgint post_data_size = 0;
look up conn->host in some cases. If that lookup failed, we
don't need to bother with connect_to_host. */
if (host_lookup_failed)
- return HOSTERR;
+ {
+ request_free (req);
+ return HOSTERR;
+ }
sock = connect_to_host (conn->host, conn->port);
if (sock == E_HOST)
- return HOSTERR;
+ {
+ request_free (req);
+ return HOSTERR;
+ }
else if (sock < 0)
- return (retryable_socket_connect_error (errno)
- ? CONERROR : CONIMPOSSIBLE);
+ {
+ request_free (req);
+ return (retryable_socket_connect_error (errno)
+ ? CONERROR : CONIMPOSSIBLE);
+ }
#ifdef HAVE_SSL
if (proxy && u->scheme == SCHEME_HTTPS)
return WRITEFAILED;
}
- head = fd_read_http_head (sock);
+ head = read_http_response_head (sock);
if (!head)
{
logprintf (LOG_VERBOSE, _("Failed reading proxy response: %s\n"),
}
DEBUGP (("proxy responded with: [%s]\n", head));
- resp = response_new (head);
- statcode = response_status (resp, &message);
- response_free (resp);
+ resp = resp_new (head);
+ statcode = resp_status (resp, &message);
+ resp_free (resp);
+ xfree (head);
if (statcode != 200)
{
failed_tunnel:
contrange = 0;
*dt &= ~RETROKF;
- head = fd_read_http_head (sock);
+ head = read_http_response_head (sock);
if (!head)
{
if (errno == 0)
}
DEBUGP (("\n---response begin---\n%s---response end---\n", head));
- resp = response_new (head);
+ resp = resp_new (head);
/* Check for status line. */
message = NULL;
- statcode = response_status (resp, &message);
+ statcode = resp_status (resp, &message);
if (!opt.server_response)
logprintf (LOG_VERBOSE, "%2d %s\n", statcode,
message ? escnonprint (message) : "");
}
if (!opt.ignore_length
- && response_header_copy (resp, "Content-Length", hdrval, sizeof (hdrval)))
+ && resp_header_copy (resp, "Content-Length", hdrval, sizeof (hdrval)))
{
wgint parsed;
errno = 0;
/* Check for keep-alive related responses. */
if (!inhibit_keep_alive && contlen != -1)
{
- if (response_header_copy (resp, "Keep-Alive", NULL, 0))
+ if (resp_header_copy (resp, "Keep-Alive", NULL, 0))
keep_alive = 1;
- else if (response_header_copy (resp, "Connection", hdrval,
- sizeof (hdrval)))
+ else if (resp_header_copy (resp, "Connection", hdrval, sizeof (hdrval)))
{
if (0 == strcasecmp (hdrval, "Keep-Alive"))
keep_alive = 1;
}
else
{
- char *www_authenticate = response_header_strdup (resp,
- "WWW-Authenticate");
+ char *www_authenticate = resp_header_strdup (resp,
+ "WWW-Authenticate");
/* If the authentication scheme is unknown or if it's the
"Basic" authentication (which we try by default), there's
no sense in retrying. */
hs->error = xstrdup (_("(no description)"));
else
hs->error = xstrdup (message);
+ xfree (message);
- type = response_header_strdup (resp, "Content-Type");
+ type = resp_header_strdup (resp, "Content-Type");
if (type)
{
char *tmp = strchr (type, ';');
*tmp = '\0';
}
}
- hs->newloc = response_header_strdup (resp, "Location");
- hs->remote_time = response_header_strdup (resp, "Last-Modified");
+ hs->newloc = resp_header_strdup (resp, "Location");
+ hs->remote_time = resp_header_strdup (resp, "Last-Modified");
+
+ /* Handle (possibly multiple instances of) the Set-Cookie header. */
{
- char *set_cookie = response_header_strdup (resp, "Set-Cookie");
- if (set_cookie)
+ int scpos;
+ const char *scbeg, *scend;
+ /* The jar should have been created by now. */
+ assert (wget_cookie_jar != NULL);
+ for (scpos = 0;
+ (scpos = resp_header_locate (resp, "Set-Cookie", scpos,
+ &scbeg, &scend)) != -1;
+ ++scpos)
{
- /* The jar should have been created by now. */
- assert (wget_cookie_jar != NULL);
+ char *set_cookie = strdupdelim (scbeg, scend);
cookie_handle_set_cookie (wget_cookie_jar, u->host, u->port, u->path,
set_cookie);
xfree (set_cookie);
}
}
- if (response_header_copy (resp, "Content-Range", hdrval, sizeof (hdrval)))
+
+ if (resp_header_copy (resp, "Content-Range", hdrval, sizeof (hdrval)))
{
wgint first_byte_pos, last_byte_pos, entity_length;
if (parse_content_range (hdrval, &first_byte_pos, &last_byte_pos,
&entity_length))
contrange = first_byte_pos;
}
- response_free (resp);
+ resp_free (resp);
+ xfree (head);
/* 20x responses are counted among successful by default. */
if (H_20X (statcode))
logputs (LOG_VERBOSE, _("Length: "));
if (contlen != -1)
{
- logputs (LOG_VERBOSE, legible (contlen + contrange));
+ logputs (LOG_VERBOSE, with_thousand_seps (contlen + contrange));
+ if (contlen + contrange >= 1024)
+ logprintf (LOG_VERBOSE, " (%s)",
+ human_readable (contlen + contrange));
if (contrange)
- logprintf (LOG_VERBOSE, _(" (%s to go)"), legible (contlen));
+ {
+ if (contlen >= 1024)
+ logprintf (LOG_VERBOSE, _(", %s (%s) remaining"),
+ with_thousand_seps (contlen),
+ human_readable (contlen));
+ else
+ logprintf (LOG_VERBOSE, _(", %s remaining"),
+ with_thousand_seps (contlen));
+ }
}
else
logputs (LOG_VERBOSE,
static char *
basic_authentication_encode (const char *user, const char *passwd)
{
- char *t1, *t2, *res;
+ char *t1, *t2;
int len1 = strlen (user) + 1 + strlen (passwd);
int len2 = BASE64_LENGTH (len1);
t2 = (char *)alloca (len2 + 1);
base64_encode (t1, t2, len1);
- res = (char *)xmalloc (6 + len2 + 1);
- sprintf (res, "Basic %s", t2);
-
- return res;
+ return concat_strings ("Basic ", t2, (char *) 0);
}
#define SKIP_WS(x) do { \
return NULL;
}
\f
+void
+save_cookies (void)
+{
+ if (wget_cookie_jar)
+ cookie_jar_save (wget_cookie_jar, opt.cookies_output);
+}
+
void
http_cleanup (void)
{
+ xfree_null (pconn.host);
+ if (wget_cookie_jar)
+ cookie_jar_delete (wget_cookie_jar);
}
projects / wget / blobdiff