/* URL handling.
- Copyright (C) 2005 Free Software Foundation, Inc.
+ Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003,
+ 2004, 2005, 2006, 2007 Free Software Foundation, Inc.
This file is part of GNU Wget.
GNU Wget is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
-the Free Software Foundation; either version 2 of the License, or (at
+the Free Software Foundation; either version 3 of the License, or (at
your option) any later version.
GNU Wget is distributed in the hope that it will be useful,
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
-along with Wget; if not, write to the Free Software
-Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+along with Wget. If not, see <http://www.gnu.org/licenses/>.
In addition, as a special exception, the Free Software Foundation
gives permission to link the code of its release of Wget with the
file, but you are not obligated to do so. If you do not wish to do
so, delete this exception statement from your version. */
-#include <config.h>
+#include "wget.h"
#include <stdio.h>
#include <stdlib.h>
#include <errno.h>
#include <assert.h>
-#include "wget.h"
#include "utils.h"
#include "url.h"
#include "host.h" /* for is_valid_ipv6_address */
+#ifdef TESTING
+#include "test.h"
+#endif
+
+enum {
+ scm_disabled = 1, /* for https when OpenSSL fails to init. */
+ scm_has_params = 2, /* whether scheme has ;params */
+ scm_has_query = 4, /* whether scheme has ?query */
+ scm_has_fragment = 8 /* whether scheme has #fragment */
+};
+
struct scheme_data
{
+ /* Short name of the scheme, such as "http" or "ftp". */
const char *name;
+ /* Leading string that identifies the scheme, such as "https://". */
const char *leading_string;
+ /* Default port of the scheme when none is specified. */
int default_port;
- bool enabled;
+ /* Various flags. */
+ int flags;
};
/* Supported schemes: */
static struct scheme_data supported_schemes[] =
{
- { "http", "http://", DEFAULT_HTTP_PORT, 1 },
+ { "http", "http://", DEFAULT_HTTP_PORT, scm_has_query|scm_has_fragment },
#ifdef HAVE_SSL
- { "https", "https://", DEFAULT_HTTPS_PORT, 1 },
+ { "https", "https://", DEFAULT_HTTPS_PORT, scm_has_query|scm_has_fragment },
#endif
- { "ftp", "ftp://", DEFAULT_FTP_PORT, 1 },
+ { "ftp", "ftp://", DEFAULT_FTP_PORT, scm_has_params|scm_has_fragment },
/* SCHEME_INVALID */
- { NULL, NULL, -1, 0 }
+ { NULL, NULL, -1, 0 }
};
/* Forward declarations: */
static void
url_unescape (char *s)
{
- char *t = s; /* t - tortoise */
- char *h = s; /* h - hare */
+ char *t = s; /* t - tortoise */
+ char *h = s; /* h - hare */
for (; *h; h++, t++)
{
if (*h != '%')
- {
- copychar:
- *t = *h;
- }
+ {
+ copychar:
+ *t = *h;
+ }
else
- {
- char c;
- /* Do nothing if '%' is not followed by two hex digits. */
- if (!h[1] || !h[2] || !(ISXDIGIT (h[1]) && ISXDIGIT (h[2])))
- goto copychar;
- c = X2DIGITS_TO_NUM (h[1], h[2]);
- /* Don't unescape %00 because there is no way to insert it
- into a C string without effectively truncating it. */
- if (c == '\0')
- goto copychar;
- *t = c;
- h += 2;
- }
+ {
+ char c;
+ /* Do nothing if '%' is not followed by two hex digits. */
+ if (!h[1] || !h[2] || !(c_isxdigit (h[1]) && c_isxdigit (h[2])))
+ goto copychar;
+ c = X2DIGITS_TO_NUM (h[1], h[2]);
+ /* Don't unescape %00 because there is no way to insert it
+ into a C string without effectively truncating it. */
+ if (c == '\0')
+ goto copychar;
+ *t = c;
+ h += 2;
+ }
}
*t = '\0';
}
for (p1 = s; *p1; p1++)
if (urlchr_test (*p1, mask))
- addition += 2; /* Two more characters (hex digits) */
+ addition += 2; /* Two more characters (hex digits) */
if (!addition)
return allow_passthrough ? (char *)s : xstrdup (s);
{
/* Quote the characters that match the test mask. */
if (urlchr_test (*p1, mask))
- {
- unsigned char c = *p1++;
- *p2++ = '%';
- *p2++ = XNUM_TO_DIGIT (c >> 4);
- *p2++ = XNUM_TO_DIGIT (c & 0xf);
- }
+ {
+ unsigned char c = *p1++;
+ *p2++ = '%';
+ *p2++ = XNUM_TO_DIGIT (c >> 4);
+ *p2++ = XNUM_TO_DIGIT (c & 0xf);
+ }
else
- *p2++ = *p1++;
+ *p2++ = *p1++;
}
assert (p2 - newstr == newlen);
*p2 = '\0';
{
if (*p == '%')
{
- if (ISXDIGIT (*(p + 1)) && ISXDIGIT (*(p + 2)))
- return false;
+ if (c_isxdigit (*(p + 1)) && c_isxdigit (*(p + 2)))
+ return false;
else
- /* Garbled %.. sequence: encode `%'. */
- return true;
+ /* Garbled %.. sequence: encode `%'. */
+ return true;
}
else if (URL_UNSAFE_CHAR (*p) && !URL_RESERVED_CHAR (*p))
return true;
if (!encode_count)
/* The string is good as it is. */
- return (char *) s; /* C const model sucks. */
+ return (char *) s; /* C const model sucks. */
oldlen = p1 - s;
/* Each encoding adds two characters (hex digits). */
while (*p1)
if (char_needs_escaping (p1))
{
- unsigned char c = *p1++;
- *p2++ = '%';
- *p2++ = XNUM_TO_DIGIT (c >> 4);
- *p2++ = XNUM_TO_DIGIT (c & 0xf);
+ unsigned char c = *p1++;
+ *p2++ = '%';
+ *p2++ = XNUM_TO_DIGIT (c >> 4);
+ *p2++ = XNUM_TO_DIGIT (c & 0xf);
}
else
*p2++ = *p1++;
for (i = 0; supported_schemes[i].leading_string; i++)
if (0 == strncasecmp (url, supported_schemes[i].leading_string,
- strlen (supported_schemes[i].leading_string)))
+ strlen (supported_schemes[i].leading_string)))
{
- if (supported_schemes[i].enabled)
- return (enum url_scheme) i;
- else
- return SCHEME_INVALID;
+ if (!(supported_schemes[i].flags & scm_disabled))
+ return (enum url_scheme) i;
+ else
+ return SCHEME_INVALID;
}
return SCHEME_INVALID;
}
-#define SCHEME_CHAR(ch) (ISALNUM (ch) || (ch) == '-' || (ch) == '+')
+#define SCHEME_CHAR(ch) (c_isalnum (ch) || (ch) == '-' || (ch) == '+')
/* Return 1 if the URL begins with any "scheme", 0 otherwise. As
currently implemented, it returns true if URL begins with
void
scheme_disable (enum url_scheme scheme)
{
- supported_schemes[scheme].enabled = false;
+ supported_schemes[scheme].flags |= scm_disabled;
}
/* Skip the username and password, if present in the URL. The
const char *userend;
if (beg == end)
- return false; /* empty user name */
+ return false; /* empty user name */
colon = memchr (beg, ':', end - beg);
if (colon == beg)
- return false; /* again empty user name */
+ return false; /* again empty user name */
if (colon)
{
}
/* Used by main.c: detect URLs written using the "shorthand" URL forms
- popularized by Netscape and NcFTP. HTTP shorthands look like this:
+ originally popularized by Netscape and NcFTP. HTTP shorthands look
+ like this:
www.foo.com[:port]/dir/file -> http://www.foo.com[:port]/dir/file
www.foo.com[:port] -> http://www.foo.com[:port]
rewrite_shorthand_url (const char *url)
{
const char *p;
+ char *ret;
if (url_scheme (url) != SCHEME_INVALID)
return NULL;
/* Look for a ':' or '/'. The former signifies NcFTP syntax, the
latter Netscape. */
- for (p = url; *p && *p != ':' && *p != '/'; p++)
- ;
-
+ p = strpbrk (url, ":/");
if (p == url)
return NULL;
/* If we're looking at "://", it means the URL uses a scheme we
don't support, which may include "https" when compiled without
SSL support. Don't bogusly rewrite such URLs. */
- if (p[0] == ':' && p[1] == '/' && p[2] == '/')
+ if (p && p[0] == ':' && p[1] == '/' && p[2] == '/')
return NULL;
- if (*p == ':')
+ if (p && *p == ':')
{
- const char *pp;
- char *res;
- /* If the characters after the colon and before the next slash
- or end of string are all digits, it's HTTP. */
- int digits = 0;
- for (pp = p + 1; ISDIGIT (*pp); pp++)
- ++digits;
- if (digits > 0 && (*pp == '/' || *pp == '\0'))
- goto http;
-
- /* Prepend "ftp://" to the entire URL... */
- res = xmalloc (6 + strlen (url) + 1);
- sprintf (res, "ftp://%s", url);
- /* ...and replace ':' with '/'. */
- res[6 + (p - url)] = '/';
- return res;
+ /* Colon indicates ftp, as in foo.bar.com:path. Check for
+ special case of http port number ("localhost:10000"). */
+ int digits = strspn (p + 1, "0123456789");
+ if (digits && (p[1 + digits] == '/' || p[1 + digits] == '\0'))
+ goto http;
+
+ /* Turn "foo.bar.com:path" to "ftp://foo.bar.com/path". */
+ ret = aprintf ("ftp://%s", url);
+ ret[6 + (p - url)] = '/';
}
else
{
- char *res;
http:
- /* Just prepend "http://" to what we have. */
- res = xmalloc (7 + strlen (url) + 1);
- sprintf (res, "http://%s", url);
- return res;
+ /* Just prepend "http://" to URL. */
+ ret = aprintf ("http://%s", url);
}
+ return ret;
}
\f
static void split_path (const char *, char **, char **);
/* Like strpbrk, with the exception that it returns the pointer to the
terminating zero (end-of-string aka "eos") if no matching character
- is found.
-
- Although I normally balk at Gcc-specific optimizations, it probably
- makes sense here: glibc has optimizations that detect strpbrk being
- called with literal string as ACCEPT and inline the search. That
- optimization is defeated if strpbrk is hidden within the call to
- another function. (And no, making strpbrk_or_eos inline doesn't
- help because the check for literal accept is in the
- preprocessor.) */
-
-#if defined(__GNUC__) && __GNUC__ >= 3
-
-#define strpbrk_or_eos(s, accept) ({ \
- char *SOE_p = strpbrk (s, accept); \
- if (!SOE_p) \
- SOE_p = strchr (s, '\0'); \
- SOE_p; \
-})
-
-#else /* not __GNUC__ or old gcc */
+ is found. */
static inline char *
strpbrk_or_eos (const char *s, const char *accept)
p = strchr (s, '\0');
return p;
}
-#endif /* not __GNUC__ or old gcc */
/* Turn STR into lowercase; return true if a character was actually
changed. */
{
bool changed = false;
for (; *str; str++)
- if (ISUPPER (*str))
+ if (c_isupper (*str))
{
- changed = true;
- *str = TOLOWER (*str);
+ changed = true;
+ *str = c_tolower (*str);
}
return changed;
}
+static const char *
+init_seps (enum url_scheme scheme)
+{
+ static char seps[8] = ":/";
+ char *p = seps + 2;
+ int flags = supported_schemes[scheme].flags;
+
+ if (flags & scm_has_params)
+ *p++ = ';';
+ if (flags & scm_has_query)
+ *p++ = '?';
+ if (flags & scm_has_fragment)
+ *p++ = '#';
+ *p++ = '\0';
+ return seps;
+}
+
static const char *parse_errors[] = {
-#define PE_NO_ERROR 0
+#define PE_NO_ERROR 0
N_("No error"),
-#define PE_UNSUPPORTED_SCHEME 1
+#define PE_UNSUPPORTED_SCHEME 1
N_("Unsupported scheme"),
-#define PE_EMPTY_HOST 2
- N_("Empty host"),
-#define PE_BAD_PORT_NUMBER 3
+#define PE_INVALID_HOST_NAME 2
+ N_("Invalid host name"),
+#define PE_BAD_PORT_NUMBER 3
N_("Bad port number"),
-#define PE_INVALID_USER_NAME 4
+#define PE_INVALID_USER_NAME 4
N_("Invalid user name"),
-#define PE_UNTERMINATED_IPV6_ADDRESS 5
+#define PE_UNTERMINATED_IPV6_ADDRESS 5
N_("Unterminated IPv6 numeric address"),
-#define PE_IPV6_NOT_SUPPORTED 6
+#define PE_IPV6_NOT_SUPPORTED 6
N_("IPv6 addresses not supported"),
-#define PE_INVALID_IPV6_ADDRESS 7
+#define PE_INVALID_IPV6_ADDRESS 7
N_("Invalid IPv6 numeric address")
};
bool path_modified, host_modified;
enum url_scheme scheme;
+ const char *seps;
const char *uname_b, *uname_e;
const char *host_b, *host_e;
scheme://host[:port][/path][;params][?query][#fragment] */
+ path_b = path_e = NULL;
params_b = params_e = NULL;
query_b = query_e = NULL;
fragment_b = fragment_e = NULL;
+ /* Initialize separators for optional parts of URL, depending on the
+ scheme. For example, FTP has params, and HTTP and HTTPS have
+ query string and fragment. */
+ seps = init_seps (scheme);
+
host_b = p;
if (*p == '[')
{
/* Handle IPv6 address inside square brackets. Ideally we'd
- just look for the terminating ']', but rfc2732 mandates
- rejecting invalid IPv6 addresses. */
+ just look for the terminating ']', but rfc2732 mandates
+ rejecting invalid IPv6 addresses. */
/* The address begins after '['. */
host_b = p + 1;
host_e = strchr (host_b, ']');
if (!host_e)
- {
- error_code = PE_UNTERMINATED_IPV6_ADDRESS;
- goto error;
- }
+ {
+ error_code = PE_UNTERMINATED_IPV6_ADDRESS;
+ goto error;
+ }
#ifdef ENABLE_IPV6
/* Check if the IPv6 address is valid. */
if (!is_valid_ipv6_address(host_b, host_e))
- {
- error_code = PE_INVALID_IPV6_ADDRESS;
- goto error;
- }
+ {
+ error_code = PE_INVALID_IPV6_ADDRESS;
+ goto error;
+ }
/* Continue parsing after the closing ']'. */
p = host_e + 1;
error_code = PE_IPV6_NOT_SUPPORTED;
goto error;
#endif
+
+ /* The closing bracket must be followed by a separator or by the
+ null char. */
+ /* http://[::1]... */
+ /* ^ */
+ if (!strchr (seps, *p))
+ {
+ /* Trailing garbage after []-delimited IPv6 address. */
+ error_code = PE_INVALID_HOST_NAME;
+ goto error;
+ }
}
else
{
- p = strpbrk_or_eos (p, ":/;?#");
+ p = strpbrk_or_eos (p, seps);
host_e = p;
}
+ ++seps; /* advance to '/' */
if (host_b == host_e)
{
- error_code = PE_EMPTY_HOST;
+ error_code = PE_INVALID_HOST_NAME;
goto error;
}
/* ^ */
++p;
port_b = p;
- p = strpbrk_or_eos (p, "/;?#");
+ p = strpbrk_or_eos (p, seps);
port_e = p;
/* Allow empty port, as per rfc2396. */
if (port_b != port_e)
- {
- for (port = 0, pp = port_b; pp < port_e; pp++)
- {
- if (!ISDIGIT (*pp))
- {
- /* http://host:12randomgarbage/blah */
- /* ^ */
- error_code = PE_BAD_PORT_NUMBER;
- goto error;
- }
- port = 10 * port + (*pp - '0');
- /* Check for too large port numbers here, before we have
- a chance to overflow on bogus port values. */
- if (port > 65535)
- {
- error_code = PE_BAD_PORT_NUMBER;
- goto error;
- }
- }
- }
+ for (port = 0, pp = port_b; pp < port_e; pp++)
+ {
+ if (!c_isdigit (*pp))
+ {
+ /* http://host:12randomgarbage/blah */
+ /* ^ */
+ error_code = PE_BAD_PORT_NUMBER;
+ goto error;
+ }
+ port = 10 * port + (*pp - '0');
+ /* Check for too large port numbers here, before we have
+ a chance to overflow on bogus port values. */
+ if (port > 0xffff)
+ {
+ error_code = PE_BAD_PORT_NUMBER;
+ goto error;
+ }
+ }
}
+ /* Advance to the first separator *after* '/' (either ';' or '?',
+ depending on the scheme). */
+ ++seps;
+
+ /* Get the optional parts of URL, each part being delimited by
+ current location and the position of the next separator. */
+#define GET_URL_PART(sepchar, var) do { \
+ if (*p == sepchar) \
+ var##_b = ++p, var##_e = p = strpbrk_or_eos (p, seps); \
+ ++seps; \
+} while (0)
- if (*p == '/')
- {
- ++p;
- path_b = p;
- p = strpbrk_or_eos (p, ";?#");
- path_e = p;
- }
- else
- {
- /* Path is not allowed not to exist. */
- path_b = path_e = p;
- }
+ GET_URL_PART ('/', path);
+ if (supported_schemes[scheme].flags & scm_has_params)
+ GET_URL_PART (';', params);
+ if (supported_schemes[scheme].flags & scm_has_query)
+ GET_URL_PART ('?', query);
+ if (supported_schemes[scheme].flags & scm_has_fragment)
+ GET_URL_PART ('#', fragment);
- if (*p == ';')
- {
- ++p;
- params_b = p;
- p = strpbrk_or_eos (p, "?#");
- params_e = p;
- }
- if (*p == '?')
- {
- ++p;
- query_b = p;
- p = strpbrk_or_eos (p, "#");
- query_e = p;
-
- /* Hack that allows users to use '?' (a wildcard character) in
- FTP URLs without it being interpreted as a query string
- delimiter. */
- if (scheme == SCHEME_FTP)
- {
- query_b = query_e = NULL;
- path_e = p;
- }
- }
- if (*p == '#')
- {
- ++p;
- fragment_b = p;
- p += strlen (p);
- fragment_e = p;
- }
+#undef GET_URL_PART
assert (*p == 0);
if (uname_b != uname_e)
/* ^ ^ */
/* uname_b uname_e */
if (!parse_credentials (uname_b, uname_e - 1, &user, &passwd))
- {
- error_code = PE_INVALID_USER_NAME;
- goto error;
- }
+ {
+ error_code = PE_INVALID_USER_NAME;
+ goto error;
+ }
}
u = xnew0 (struct url);
if (path_modified || u->fragment || host_modified || path_b == path_e)
{
/* If we suspect that a transformation has rendered what
- url_string might return different from URL_ENCODED, rebuild
- u->url using url_string. */
- u->url = url_string (u, false);
+ url_string might return different from URL_ENCODED, rebuild
+ u->url using url_string. */
+ u->url = url_string (u, URL_AUTH_SHOW);
if (url_encoded != url)
- xfree ((char *) url_encoded);
+ xfree ((char *) url_encoded);
}
else
{
if (url_encoded == url)
- u->url = xstrdup (url);
+ u->url = xstrdup (url);
else
- u->url = url_encoded;
+ u->url = url_encoded;
}
return u;
static void
full_path_write (const struct url *url, char *where)
{
-#define FROB(el, chr) do { \
- char *f_el = url->el; \
- if (f_el) { \
- int l = strlen (f_el); \
- *where++ = chr; \
- memcpy (where, f_el, l); \
- where += l; \
- } \
+#define FROB(el, chr) do { \
+ char *f_el = url->el; \
+ if (f_el) { \
+ int l = strlen (f_el); \
+ *where++ = chr; \
+ memcpy (where, f_el, l); \
+ where += l; \
+ } \
} while (0)
FROB (path, '/');
{
const char c1 = XNUM_TO_DIGIT (chr >> 4);
const char c2 = XNUM_TO_DIGIT (chr & 0xf);
- char *h = str; /* hare */
- char *t = str; /* tortoise */
+ char *h = str; /* hare */
+ char *t = str; /* tortoise */
for (; *h; h++, t++)
{
if (h[0] == '%' && h[1] == c1 && h[2] == c2)
- {
- *t = chr;
- h += 2;
- }
+ {
+ *t = chr;
+ h += 2;
+ }
else
- *t = *h;
+ *t = *h;
}
*t = '\0';
}
/* Regenerate u->url as well. */
xfree (u->url);
- u->url = url_string (u, false);
+ u->url = url_string (u, URL_AUTH_SHOW);
}
/* Mutators. Code in ftp.c insists on changing u->dir and u->file.
if ((stat (t, &st) == 0))
{
if (S_ISDIR (st.st_mode))
- {
- xfree (t);
- return 0;
- }
+ {
+ xfree (t);
+ return 0;
+ }
else
- {
- /* If the dir exists as a file name, remove it first. This
- is *only* for Wget to work with buggy old CERN http
- servers. Here is the scenario: When Wget tries to
- retrieve a directory without a slash, e.g.
- http://foo/bar (bar being a directory), CERN server will
- not redirect it too http://foo/bar/ -- it will generate a
- directory listing containing links to bar/file1,
- bar/file2, etc. Wget will lose because it saves this
- HTML listing to a file `bar', so it cannot create the
- directory. To work around this, if the file of the same
- name exists, we just remove it and create the directory
- anyway. */
- DEBUGP (("Removing %s because of directory danger!\n", t));
- unlink (t);
- }
+ {
+ /* If the dir exists as a file name, remove it first. This
+ is *only* for Wget to work with buggy old CERN http
+ servers. Here is the scenario: When Wget tries to
+ retrieve a directory without a slash, e.g.
+ http://foo/bar (bar being a directory), CERN server will
+ not redirect it too http://foo/bar/ -- it will generate a
+ directory listing containing links to bar/file1,
+ bar/file2, etc. Wget will lose because it saves this
+ HTML listing to a file `bar', so it cannot create the
+ directory. To work around this, if the file of the same
+ name exists, we just remove it and create the directory
+ anyway. */
+ DEBUGP (("Removing %s because of directory danger!\n", t));
+ unlink (t);
+ }
}
res = make_directory (t);
if (res != 0)
the current TAIL position. If necessary, this will grow the string
and update its allocated size. If the string is already large
enough to take TAIL+APPEND_COUNT characters, this does nothing. */
-#define GROW(g, append_size) do { \
- struct growable *G_ = g; \
- DO_REALLOC (G_->base, G_->size, G_->tail + append_size, char); \
+#define GROW(g, append_size) do { \
+ struct growable *G_ = g; \
+ DO_REALLOC (G_->base, G_->size, G_->tail + append_size, char); \
} while (0)
/* Return the tail position of the string. */
}
enum {
- filechr_not_unix = 1, /* unusable on Unix, / and \0 */
- filechr_not_windows = 2, /* unusable on Windows, one of \|/<>?:*" */
- filechr_control = 4 /* a control character, e.g. 0-31 */
+ filechr_not_unix = 1, /* unusable on Unix, / and \0 */
+ filechr_not_windows = 2, /* unusable on Windows, one of \|/<>?:*" */
+ filechr_control = 4 /* a control character, e.g. 0-31 */
};
#define FILE_CHAR_TEST(c, mask) (filechr_table[(unsigned char)(c)] & (mask))
0, 0, 0, 0, 0, 0, 0, 0, /* ` a b c d e f g */
0, 0, 0, 0, 0, 0, 0, 0, /* h i j k l m n o */
0, 0, 0, 0, 0, 0, 0, 0, /* p q r s t u v w */
- 0, 0, 0, 0, 0, 0, 0, 0, /* x y z { | } ~ DEL */
+ 0, 0, 0, 0, W, 0, 0, C, /* x y z { | } ~ DEL */
C, C, C, C, C, C, C, C, C, C, C, C, C, C, C, C, /* 128-143 */
C, C, C, C, C, C, C, C, C, C, C, C, C, C, C, C, /* 144-159 */
static void
append_uri_pathel (const char *b, const char *e, bool escaped,
- struct growable *dest)
+ struct growable *dest)
{
const char *p;
int quoted, outlen;
if (!quoted)
{
/* If there's nothing to quote, we can simply append the string
- without processing it again. */
+ without processing it again. */
memcpy (TAIL (dest), b, outlen);
}
else
{
char *q = TAIL (dest);
for (p = b; p < e; p++)
- {
- if (!FILE_CHAR_TEST (*p, mask))
- *q++ = *p;
- else
- {
- unsigned char ch = *p;
- *q++ = '%';
- *q++ = XNUM_TO_DIGIT (ch >> 4);
- *q++ = XNUM_TO_DIGIT (ch & 0xf);
- }
- }
+ {
+ if (!FILE_CHAR_TEST (*p, mask))
+ *q++ = *p;
+ else
+ {
+ unsigned char ch = *p;
+ *q++ = '%';
+ *q++ = XNUM_TO_DIGIT (ch >> 4);
+ *q++ = XNUM_TO_DIGIT (ch & 0xf);
+ }
+ }
assert (q - TAIL (dest) == outlen);
}
+
+ /* Perform inline case transformation if required. */
+ if (opt.restrict_files_case == restrict_lowercase
+ || opt.restrict_files_case == restrict_uppercase)
+ {
+ char *q;
+ for (q = TAIL (dest); q < TAIL (dest) + outlen; ++q)
+ {
+ if (opt.restrict_files_case == restrict_lowercase)
+ *q = c_tolower (*q);
+ else
+ *q = c_toupper (*q);
+ }
+ }
+
TAIL_INCR (dest, outlen);
}
for (; (next = strchr (pathel, '/')) != NULL; pathel = next + 1)
{
if (cut-- > 0)
- continue;
+ continue;
if (pathel == next)
- /* Ignore empty pathels. */
- continue;
+ /* Ignore empty pathels. */
+ continue;
if (dest->tail)
- append_char ('/', dest);
+ append_char ('/', dest);
append_uri_pathel (pathel, next, true, dest);
}
}
char *
url_file_name (const struct url *u)
{
- struct growable fnres; /* stands for "file name result" */
+ struct growable fnres; /* stands for "file name result" */
const char *u_file, *u_query;
char *fname, *unique;
if (opt.dirstruct)
{
if (opt.protocol_directories)
- {
- if (fnres.tail)
- append_char ('/', &fnres);
- append_string (supported_schemes[u->scheme].name, &fnres);
- }
+ {
+ if (fnres.tail)
+ append_char ('/', &fnres);
+ append_string (supported_schemes[u->scheme].name, &fnres);
+ }
if (opt.add_hostdir)
- {
- if (fnres.tail)
- append_char ('/', &fnres);
- if (0 != strcmp (u->host, ".."))
- append_string (u->host, &fnres);
- else
- /* Host name can come from the network; malicious DNS may
- allow ".." to be resolved, causing us to write to
- "../<file>". Defang such host names. */
- append_string ("%2E%2E", &fnres);
- if (u->port != scheme_default_port (u->scheme))
- {
- char portstr[24];
- number_to_string (portstr, u->port);
- append_char (FN_PORT_SEP, &fnres);
- append_string (portstr, &fnres);
- }
- }
+ {
+ if (fnres.tail)
+ append_char ('/', &fnres);
+ if (0 != strcmp (u->host, ".."))
+ append_string (u->host, &fnres);
+ else
+ /* Host name can come from the network; malicious DNS may
+ allow ".." to be resolved, causing us to write to
+ "../<file>". Defang such host names. */
+ append_string ("%2E%2E", &fnres);
+ if (u->port != scheme_default_port (u->scheme))
+ {
+ char portstr[24];
+ number_to_string (portstr, u->port);
+ append_char (FN_PORT_SEP, &fnres);
+ append_string (portstr, &fnres);
+ }
+ }
append_dir_structure (u, &fnres);
}
static bool
path_simplify (char *path)
{
- char *h = path; /* hare */
- char *t = path; /* tortoise */
- char *beg = path; /* boundary for backing the tortoise */
- char *end = path + strlen (path);
+ char *h = path; /* hare */
+ char *t = path; /* tortoise */
+ char *end = strchr (path, '\0');
while (h < end)
{
/* Hare should be at the beginning of a path element. */
if (h[0] == '.' && (h[1] == '/' || h[1] == '\0'))
- {
- /* Ignore "./". */
- h += 2;
- }
+ {
+ /* Ignore "./". */
+ h += 2;
+ }
else if (h[0] == '.' && h[1] == '.' && (h[2] == '/' || h[2] == '\0'))
- {
- /* Handle "../" by retreating the tortoise by one path
- element -- but not past beggining. */
- if (t > beg)
- {
- /* Move backwards until T hits the beginning of the
- previous path element or the beginning of path. */
- for (--t; t > beg && t[-1] != '/'; t--)
- ;
- }
- else
- {
- /* If we're at the beginning, copy the "../" literally
- move the beginning so a later ".." doesn't remove
- it. */
- beg = t + 3;
- goto regular;
- }
- h += 3;
- }
+ {
+ /* Handle "../" by retreating the tortoise by one path
+ element -- but not past beggining. */
+ if (t > path)
+ {
+ /* Move backwards until T hits the beginning of the
+ previous path element or the beginning of path. */
+ for (--t; t > path && t[-1] != '/'; t--)
+ ;
+ }
+ h += 3;
+ }
else
- {
- regular:
- /* A regular path element. If H hasn't advanced past T,
- simply skip to the next path element. Otherwise, copy
- the path element until the next slash. */
- if (t == h)
- {
- /* Skip the path element, including the slash. */
- while (h < end && *h != '/')
- t++, h++;
- if (h < end)
- t++, h++;
- }
- else
- {
- /* Copy the path element, including the final slash. */
- while (h < end && *h != '/')
- *t++ = *h++;
- if (h < end)
- *t++ = *h++;
- }
- }
+ {
+ /* A regular path element. If H hasn't advanced past T,
+ simply skip to the next path element. Otherwise, copy
+ the path element until the next slash. */
+ if (t == h)
+ {
+ /* Skip the path element, including the slash. */
+ while (h < end && *h != '/')
+ t++, h++;
+ if (h < end)
+ t++, h++;
+ }
+ else
+ {
+ /* Copy the path element, including the final slash. */
+ while (h < end && *h != '/')
+ *t++ = *h++;
+ if (h < end)
+ *t++ = *h++;
+ }
+ }
}
if (t != h)
}
\f
/* Return the length of URL's path. Path is considered to be
- terminated by one of '?', ';', '#', or by the end of the
- string. */
+ terminated by one or more of the ?query or ;params or #fragment,
+ depending on the scheme. */
-static int
-path_length (const char *url)
+static const char *
+path_end (const char *url)
{
- const char *q = strpbrk_or_eos (url, "?;#");
- return q - url;
+ enum url_scheme scheme = url_scheme (url);
+ const char *seps;
+ if (scheme == SCHEME_INVALID)
+ scheme = SCHEME_HTTP; /* use http semantics for rel links */
+ /* +2 to ignore the first two separators ':' and '/' */
+ seps = init_seps (scheme) + 2;
+ return strpbrk_or_eos (url, seps);
}
/* Find the last occurrence of character C in the range [b, e), or
- NULL, if none are present. We might want to use memrchr (a GNU
- extension) under GNU libc. */
-
-static const char *
-find_last_char (const char *b, const char *e, char c)
-{
- for (; e > b; e--)
- if (*e == c)
- return e;
- return NULL;
-}
+ NULL, if none are present. */
+#define find_last_char(b, e, c) memrchr ((b), (c), (e) - (b))
/* Merge BASE with LINK and return the resulting URI.
return xstrdup (link);
/* We may not examine BASE past END. */
- end = base + path_length (base);
+ end = path_end (base);
linklength = strlen (link);
if (!*link)
else if (*link == '?')
{
/* LINK points to the same location, but changes the query
- string. Examples: */
+ string. Examples: */
/* uri_merge("path", "?new") -> "path?new" */
/* uri_merge("path?foo", "?new") -> "path?new" */
/* uri_merge("path?foo#bar", "?new") -> "path?new" */
int baselength;
const char *end1 = strchr (base, '#');
if (!end1)
- end1 = base + strlen (base);
+ end1 = base + strlen (base);
baselength = end1 - base;
merge = xmalloc (baselength + linklength + 1);
memcpy (merge, base, baselength);
else if (*link == '/' && *(link + 1) == '/')
{
/* LINK begins with "//" and so is a net path: we need to
- replace everything after (and including) the double slash
- with LINK. */
+ replace everything after (and including) the double slash
+ with LINK. */
/* uri_merge("foo", "//new/bar") -> "//new/bar" */
/* uri_merge("//old/foo", "//new/bar") -> "//new/bar" */
/* Look for first slash. */
slash = memchr (base, '/', end - base);
/* If found slash and it is a double slash, then replace
- from this point, else default to replacing from the
- beginning. */
+ from this point, else default to replacing from the
+ beginning. */
if (slash && *(slash + 1) == '/')
- start_insert = slash;
+ start_insert = slash;
else
- start_insert = base;
+ start_insert = base;
span = start_insert - base;
merge = xmalloc (span + linklength + 1);
if (span)
- memcpy (merge, base, span);
+ memcpy (merge, base, span);
memcpy (merge + span, link, linklength);
merge[span + linklength] = '\0';
}
else if (*link == '/')
{
/* LINK is an absolute path: we need to replace everything
- after (and including) the FIRST slash with LINK.
+ after (and including) the FIRST slash with LINK.
- So, if BASE is "http://host/whatever/foo/bar", and LINK is
- "/qux/xyzzy", our result should be
- "http://host/qux/xyzzy". */
+ So, if BASE is "http://host/whatever/foo/bar", and LINK is
+ "/qux/xyzzy", our result should be
+ "http://host/qux/xyzzy". */
int span;
const char *slash;
const char *start_insert = NULL; /* for gcc to shut up. */
const char *pos = base;
bool seen_slash_slash = false;
/* We're looking for the first slash, but want to ignore
- double slash. */
+ double slash. */
again:
slash = memchr (pos, '/', end - pos);
if (slash && !seen_slash_slash)
- if (*(slash + 1) == '/')
- {
- pos = slash + 2;
- seen_slash_slash = true;
- goto again;
- }
+ if (*(slash + 1) == '/')
+ {
+ pos = slash + 2;
+ seen_slash_slash = true;
+ goto again;
+ }
/* At this point, SLASH is the location of the first / after
- "//", or the first slash altogether. START_INSERT is the
- pointer to the location where LINK will be inserted. When
- examining the last two examples, keep in mind that LINK
- begins with '/'. */
+ "//", or the first slash altogether. START_INSERT is the
+ pointer to the location where LINK will be inserted. When
+ examining the last two examples, keep in mind that LINK
+ begins with '/'. */
if (!slash && !seen_slash_slash)
- /* example: "foo" */
- /* ^ */
- start_insert = base;
+ /* example: "foo" */
+ /* ^ */
+ start_insert = base;
else if (!slash && seen_slash_slash)
- /* example: "http://foo" */
- /* ^ */
- start_insert = end;
+ /* example: "http://foo" */
+ /* ^ */
+ start_insert = end;
else if (slash && !seen_slash_slash)
- /* example: "foo/bar" */
- /* ^ */
- start_insert = base;
+ /* example: "foo/bar" */
+ /* ^ */
+ start_insert = base;
else if (slash && seen_slash_slash)
- /* example: "http://something/" */
- /* ^ */
- start_insert = slash;
+ /* example: "http://something/" */
+ /* ^ */
+ start_insert = slash;
span = start_insert - base;
merge = xmalloc (span + linklength + 1);
if (span)
- memcpy (merge, base, span);
+ memcpy (merge, base, span);
memcpy (merge + span, link, linklength);
merge[span + linklength] = '\0';
}
else
{
/* LINK is a relative URL: we need to replace everything
- after last slash (possibly empty) with LINK.
+ after last slash (possibly empty) with LINK.
- So, if BASE is "whatever/foo/bar", and LINK is "qux/xyzzy",
- our result should be "whatever/foo/qux/xyzzy". */
+ So, if BASE is "whatever/foo/bar", and LINK is "qux/xyzzy",
+ our result should be "whatever/foo/qux/xyzzy". */
bool need_explicit_slash = false;
int span;
const char *start_insert;
const char *last_slash = find_last_char (base, end, '/');
if (!last_slash)
- {
- /* No slash found at all. Replace what we have with LINK. */
- start_insert = base;
- }
+ {
+ /* No slash found at all. Replace what we have with LINK. */
+ start_insert = base;
+ }
else if (last_slash && last_slash >= base + 2
- && last_slash[-2] == ':' && last_slash[-1] == '/')
- {
- /* example: http://host" */
- /* ^ */
- start_insert = end + 1;
- need_explicit_slash = true;
- }
+ && last_slash[-2] == ':' && last_slash[-1] == '/')
+ {
+ /* example: http://host" */
+ /* ^ */
+ start_insert = end + 1;
+ need_explicit_slash = true;
+ }
else
- {
- /* example: "whatever/foo/bar" */
- /* ^ */
- start_insert = last_slash + 1;
- }
+ {
+ /* example: "whatever/foo/bar" */
+ /* ^ */
+ start_insert = last_slash + 1;
+ }
span = start_insert - base;
merge = xmalloc (span + linklength + 1);
if (span)
- memcpy (merge, base, span);
+ memcpy (merge, base, span);
if (need_explicit_slash)
- merge[span - 1] = '/';
+ merge[span - 1] = '/';
memcpy (merge + span, link, linklength);
merge[span + linklength] = '\0';
}
return merge;
}
\f
-#define APPEND(p, s) do { \
- int len = strlen (s); \
- memcpy (p, s, len); \
- p += len; \
+#define APPEND(p, s) do { \
+ int len = strlen (s); \
+ memcpy (p, s, len); \
+ p += len; \
} while (0)
/* Use this instead of password when the actual password is supposed
the URL will be quoted. */
char *
-url_string (const struct url *url, bool hide_password)
+url_string (const struct url *url, enum url_auth_mode auth_mode)
{
int size;
char *result, *p;
/* Make sure the user name and password are quoted. */
if (url->user)
{
- quoted_user = url_escape_allow_passthrough (url->user);
- if (url->passwd)
- {
- if (hide_password)
- quoted_passwd = HIDDEN_PASSWORD;
- else
- quoted_passwd = url_escape_allow_passthrough (url->passwd);
- }
+ if (auth_mode != URL_AUTH_HIDE)
+ {
+ quoted_user = url_escape_allow_passthrough (url->user);
+ if (url->passwd)
+ {
+ if (auth_mode == URL_AUTH_HIDE_PASSWD)
+ quoted_passwd = HIDDEN_PASSWORD;
+ else
+ quoted_passwd = url_escape_allow_passthrough (url->passwd);
+ }
+ }
}
/* In the unlikely event that the host name contains non-printable
brackets_around_host = strchr (quoted_host, ':') != NULL;
size = (strlen (scheme_str)
- + strlen (quoted_host)
- + (brackets_around_host ? 2 : 0)
- + fplen
- + 1);
+ + strlen (quoted_host)
+ + (brackets_around_host ? 2 : 0)
+ + fplen
+ + 1);
if (url->port != scheme_port)
size += 1 + numdigit (url->port);
if (quoted_user)
{
size += 1 + strlen (quoted_user);
if (quoted_passwd)
- size += 1 + strlen (quoted_passwd);
+ size += 1 + strlen (quoted_passwd);
}
p = result = xmalloc (size);
{
APPEND (p, quoted_user);
if (quoted_passwd)
- {
- *p++ = ':';
- APPEND (p, quoted_passwd);
- }
+ {
+ *p++ = ':';
+ APPEND (p, quoted_passwd);
+ }
*p++ = '@';
}
if (quoted_user && quoted_user != url->user)
xfree (quoted_user);
- if (quoted_passwd && !hide_password && quoted_passwd != url->passwd)
+ if (quoted_passwd && auth_mode == URL_AUTH_SHOW
+ && quoted_passwd != url->passwd)
xfree (quoted_passwd);
if (quoted_host != url->host)
xfree (quoted_host);
return false;
}
\f
+static int
+getchar_from_escaped_string (const char *str, char *c)
+{
+ const char *p = str;
+
+ assert (str && *str);
+ assert (c);
+
+ if (p[0] == '%')
+ {
+ if (!c_isxdigit(p[1]) || !c_isxdigit(p[2]))
+ {
+ *c = '%';
+ return 1;
+ }
+ else
+ {
+ if (p[2] == 0)
+ return 0; /* error: invalid string */
+
+ *c = X2DIGITS_TO_NUM (p[1], p[2]);
+ if (URL_RESERVED_CHAR(*c))
+ {
+ *c = '%';
+ return 1;
+ }
+ else
+ return 3;
+ }
+ }
+ else
+ {
+ *c = p[0];
+ }
+
+ return 1;
+}
+
+bool
+are_urls_equal (const char *u1, const char *u2)
+{
+ const char *p, *q;
+ int pp, qq;
+ char ch1, ch2;
+ assert(u1 && u2);
+
+ p = u1;
+ q = u2;
+
+ while (*p && *q
+ && (pp = getchar_from_escaped_string (p, &ch1))
+ && (qq = getchar_from_escaped_string (q, &ch2))
+ && (c_tolower(ch1) == c_tolower(ch2)))
+ {
+ p += pp;
+ q += qq;
+ }
+
+ return (*p == 0 && *q == 0 ? true : false);
+}
+\f
#if 0
/* Debugging and testing support for path_simplify. */
if (0 != strcmp (test_copy, expected_result))
{
printf ("Failed path_simplify(\"%s\"): expected \"%s\", got \"%s\".\n",
- test, expected_result, test_copy);
+ test, expected_result, test_copy);
}
if (modified != expected_change)
{
if (expected_change)
- printf ("Expected modification with path_simplify(\"%s\").\n",
- test);
+ printf ("Expected modification with path_simplify(\"%s\").\n",
+ test);
else
- printf ("Expected no modification with path_simplify(\"%s\").\n",
- test);
+ printf ("Expected no modification with path_simplify(\"%s\").\n",
+ test);
}
xfree (test_copy);
}
char *test, *result;
bool should_modify;
} tests[] = {
- { "", "", false },
- { ".", "", true },
- { "./", "", true },
- { "..", "..", false },
- { "../", "../", false },
- { "foo", "foo", false },
- { "foo/bar", "foo/bar", false },
- { "foo///bar", "foo///bar", false },
- { "foo/.", "foo/", true },
- { "foo/./", "foo/", true },
- { "foo./", "foo./", false },
- { "foo/../bar", "bar", true },
- { "foo/../bar/", "bar/", true },
- { "foo/bar/..", "foo/", true },
- { "foo/bar/../x", "foo/x", true },
- { "foo/bar/../x/", "foo/x/", true },
- { "foo/..", "", true },
- { "foo/../..", "..", true },
- { "foo/../../..", "../..", true },
- { "foo/../../bar/../../baz", "../../baz", true },
- { "a/b/../../c", "c", true },
- { "./a/../b", "b", true }
+ { "", "", false },
+ { ".", "", true },
+ { "./", "", true },
+ { "..", "", true },
+ { "../", "", true },
+ { "foo", "foo", false },
+ { "foo/bar", "foo/bar", false },
+ { "foo///bar", "foo///bar", false },
+ { "foo/.", "foo/", true },
+ { "foo/./", "foo/", true },
+ { "foo./", "foo./", false },
+ { "foo/../bar", "bar", true },
+ { "foo/../bar/", "bar/", true },
+ { "foo/bar/..", "foo/", true },
+ { "foo/bar/../x", "foo/x", true },
+ { "foo/bar/../x/", "foo/x/", true },
+ { "foo/..", "", true },
+ { "foo/../..", "", true },
+ { "foo/../../..", "", true },
+ { "foo/../../bar/../../baz", "baz", true },
+ { "a/b/../../c", "c", true },
+ { "./a/../b", "b", true }
};
int i;
}
}
#endif
+\f
+#ifdef TESTING
+
+const char *
+test_append_uri_pathel()
+{
+ int i;
+ struct {
+ char *original_url;
+ char *input;
+ bool escaped;
+ char *expected_result;
+ } test_array[] = {
+ { "http://www.yoyodyne.com/path/", "somepage.html", false, "http://www.yoyodyne.com/path/somepage.html" },
+ };
+
+ for (i = 0; i < sizeof(test_array)/sizeof(test_array[0]); ++i)
+ {
+ struct growable dest;
+ const char *p = test_array[i].input;
+
+ memset (&dest, 0, sizeof (dest));
+
+ append_string (test_array[i].original_url, &dest);
+ append_uri_pathel (p, p + strlen(p), test_array[i].escaped, &dest);
+ append_char ('\0', &dest);
+
+ mu_assert ("test_append_uri_pathel: wrong result",
+ strcmp (dest.base, test_array[i].expected_result) == 0);
+ }
+
+ return NULL;
+}
+
+const char*
+test_are_urls_equal()
+{
+ int i;
+ struct {
+ char *url1;
+ char *url2;
+ bool expected_result;
+ } test_array[] = {
+ { "http://www.adomain.com/apath/", "http://www.adomain.com/apath/", true },
+ { "http://www.adomain.com/apath/", "http://www.adomain.com/anotherpath/", false },
+ { "http://www.adomain.com/apath/", "http://www.anotherdomain.com/path/", false },
+ { "http://www.adomain.com/~path/", "http://www.adomain.com/%7epath/", true },
+ { "http://www.adomain.com/longer-path/", "http://www.adomain.com/path/", false },
+ { "http://www.adomain.com/path%2f", "http://www.adomain.com/path/", false },
+ };
+
+ for (i = 0; i < sizeof(test_array)/sizeof(test_array[0]); ++i)
+ {
+ mu_assert ("test_are_urls_equal: wrong result",
+ are_urls_equal (test_array[i].url1, test_array[i].url2) == test_array[i].expected_result);
+ }
+
+ return NULL;
+}
+
+#endif /* TESTING */
+
+/*
+ * vim: et ts=2 sw=2
+ */
+
projects / wget / blobdiff