X-Git-Url: http://sjero.net/git/?p=wget;a=blobdiff_plain;f=src%2Furl.c;h=2453157a617d0b4df79afe57a607ccf7644d4d04;hp=a84d5db3ee5a97dd3f8cfe92ace404e1ddecd183;hb=d5e283b1a75c5f8249300b465b4e7b55130bec49;hpb=7a54d852bfa09c828bbc4907adddee7e4beb1715 diff --git a/src/url.c b/src/url.c index a84d5db3..2453157a 100644 --- a/src/url.c +++ b/src/url.c @@ -85,7 +85,7 @@ static struct scheme_data supported_schemes[] = /* Forward declarations: */ -static bool path_simplify (char *); +static bool path_simplify (enum url_scheme, char *); /* Support for escaping and unescaping of URL strings. */ @@ -256,6 +256,15 @@ url_escape (const char *s) return url_escape_1 (s, urlchr_unsafe, false); } +/* URL-escape the unsafe and reserved characters (see urlchr_table) in + a given string, returning a freshly allocated string. */ + +char * +url_escape_unsafe_and_reserved (const char *s) +{ + return url_escape_1 (s, urlchr_unsafe|urlchr_reserved, false); +} + /* URL-escape the unsafe characters (see urlchr_table) in a given string. If no characters are unsafe, S is returned. */ @@ -623,18 +632,20 @@ static const char *parse_errors[] = { #define PE_NO_ERROR 0 N_("No error"), #define PE_UNSUPPORTED_SCHEME 1 - N_("Unsupported scheme"), -#define PE_INVALID_HOST_NAME 2 + N_("Unsupported scheme %s"), /* support for format token only here */ +#define PE_MISSING_SCHEME 2 + N_("Scheme missing"), +#define PE_INVALID_HOST_NAME 3 N_("Invalid host name"), -#define PE_BAD_PORT_NUMBER 3 +#define PE_BAD_PORT_NUMBER 4 N_("Bad port number"), -#define PE_INVALID_USER_NAME 4 +#define PE_INVALID_USER_NAME 5 N_("Invalid user name"), -#define PE_UNTERMINATED_IPV6_ADDRESS 5 +#define PE_UNTERMINATED_IPV6_ADDRESS 6 N_("Unterminated IPv6 numeric address"), -#define PE_IPV6_NOT_SUPPORTED 6 +#define PE_IPV6_NOT_SUPPORTED 7 N_("IPv6 addresses not supported"), -#define PE_INVALID_IPV6_ADDRESS 7 +#define PE_INVALID_IPV6_ADDRESS 8 N_("Invalid IPv6 numeric address") }; @@ -644,7 +655,7 @@ static const char *parse_errors[] = { error, and if ERROR is not NULL, also set *ERROR to the appropriate error code. */ struct url * -url_parse (const char *url, int *error) +url_parse (const char *url, int *error, struct iri *iri, bool percent_encode) { struct url *u; const char *p; @@ -663,20 +674,41 @@ url_parse (const char *url, int *error) int port; char *user = NULL, *passwd = NULL; - char *url_encoded = NULL; + const char *url_encoded = NULL; + char *new_url = NULL; int error_code; scheme = url_scheme (url); if (scheme == SCHEME_INVALID) { - error_code = PE_UNSUPPORTED_SCHEME; + if (url_has_scheme (url)) + error_code = PE_UNSUPPORTED_SCHEME; + else + error_code = PE_MISSING_SCHEME; goto error; } - url_encoded = reencode_escapes (url); + if (iri && iri->utf8_encode) + { + iri->utf8_encode = remote_to_utf8 (iri, iri->orig_url ? iri->orig_url : url, (const char **) &new_url); + if (!iri->utf8_encode) + new_url = NULL; + else + iri->orig_url = xstrdup (url); + } + + /* XXX XXX Could that change introduce (security) bugs ??? XXX XXX*/ + if (percent_encode) + url_encoded = reencode_escapes (new_url ? new_url : url); + else + url_encoded = new_url ? new_url : url; + p = url_encoded; + if (new_url && url_encoded != new_url) + xfree (new_url); + p += strlen (supported_schemes[scheme].leading_string); uname_b = p; p = url_skip_credentials (p); @@ -833,7 +865,7 @@ url_parse (const char *url, int *error) u->passwd = passwd; u->path = strdupdelim (path_b, path_e); - path_modified = path_simplify (u->path); + path_modified = path_simplify (scheme, u->path); split_path (u->path, &u->dir, &u->file); host_modified = lowercase_str (u->host); @@ -846,6 +878,18 @@ url_parse (const char *url, int *error) { url_unescape (u->host); host_modified = true; + + /* Apply IDNA regardless of iri->utf8_encode status */ + if (opt.enable_iri && iri) + { + char *new = idn_encode (iri, u->host); + if (new) + { + xfree (u->host); + u->host = new; + host_modified = true; + } + } } if (params_b) @@ -855,7 +899,7 @@ url_parse (const char *url, int *error) if (fragment_b) u->fragment = strdupdelim (fragment_b, fragment_e); - if (path_modified || u->fragment || host_modified || path_b == path_e) + if (opt.enable_iri || 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 @@ -870,7 +914,7 @@ url_parse (const char *url, int *error) if (url_encoded == url) u->url = xstrdup (url); else - u->url = url_encoded; + u->url = (char *) url_encoded; } return u; @@ -878,7 +922,7 @@ url_parse (const char *url, int *error) error: /* Cleanup in case of error: */ if (url_encoded && url_encoded != url) - xfree (url_encoded); + xfree ((char *) url_encoded); /* Transmit the error code to the caller, if the caller wants to know. */ @@ -890,11 +934,29 @@ url_parse (const char *url, int *error) /* Return the error message string from ERROR_CODE, which should have been retrieved from url_parse. The error message is translated. */ -const char * -url_error (int error_code) +char * +url_error (const char *url, int error_code) { - assert (error_code >= 0 && error_code < countof (parse_errors)); - return _(parse_errors[error_code]); + assert (error_code >= 0 && ((size_t) error_code) < countof (parse_errors)); + + if (error_code == PE_UNSUPPORTED_SCHEME) + { + char *error, *p; + char *scheme = xstrdup (url); + assert (url_has_scheme (url)); + + if ((p = strchr (scheme, ':'))) + *p = '\0'; + if (!strcasecmp (scheme, "https")) + error = aprintf (_("HTTPS support not compiled in")); + else + error = aprintf (_(parse_errors[error_code]), quote (scheme)); + xfree (scheme); + + return error; + } + else + return xstrdup (_(parse_errors[error_code])); } /* Split PATH into DIR and FILE. PATH comes from the URL and is @@ -1434,11 +1496,17 @@ url_file_name (const struct url *u) const char *u_file, *u_query; char *fname, *unique; + char *index_filename = "index.html"; /* The default index file is index.html */ fnres.base = NULL; fnres.size = 0; fnres.tail = 0; + /* If an alternative index file was defined, change index_filename */ + if (opt.default_page) + index_filename = opt.default_page; + + /* Start with the directory prefix, if specified. */ if (opt.dir_prefix) append_string (opt.dir_prefix, &fnres); @@ -1480,7 +1548,7 @@ url_file_name (const struct url *u) /* Add the file name. */ if (fnres.tail) append_char ('/', &fnres); - u_file = *u->file ? u->file : "index.html"; + u_file = *u->file ? u->file : index_filename; append_uri_pathel (u_file, u_file + strlen (u_file), false, &fnres); /* Append "?query" to the file name. */ @@ -1549,10 +1617,11 @@ url_file_name (const struct url *u) test case. */ static bool -path_simplify (char *path) +path_simplify (enum url_scheme scheme, char *path) { char *h = path; /* hare */ char *t = path; /* tortoise */ + char *beg = path; char *end = strchr (path, '\0'); while (h < end) @@ -1568,17 +1637,29 @@ path_simplify (char *path) { /* Handle "../" by retreating the tortoise by one path element -- but not past beggining. */ - if (t > path) + if (t > beg) { /* Move backwards until T hits the beginning of the previous path element or the beginning of path. */ - for (--t; t > path && t[-1] != '/'; t--) + for (--t; t > beg && t[-1] != '/'; t--) ; } + else if (scheme == SCHEME_FTP) + { + /* If we're at the beginning, copy the "../" literally + and move the beginning so a later ".." doesn't remove + it. This violates RFC 3986; but we do it for FTP + anyway because there is otherwise no way to get at a + parent directory, when the FTP server drops us in a + non-root directory (which is not uncommon). */ + beg = t + 3; + goto regular; + } 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. */ @@ -1955,12 +2036,12 @@ schemes_are_similar_p (enum url_scheme a, enum url_scheme b) 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])) @@ -2010,13 +2091,14 @@ are_urls_equal (const char *u1, const char *u2) p += pp; q += qq; } - + return (*p == 0 && *q == 0 ? true : false); } -#if 0 +#ifdef TESTING /* Debugging and testing support for path_simplify. */ +#if 0 /* Debug: run path_simplify on PATH and return the result in a new string. Useful for calling from the debugger. */ static char * @@ -2026,17 +2108,20 @@ ps (char *path) path_simplify (copy); return copy; } +#endif -static void -run_test (char *test, char *expected_result, bool expected_change) +static const char * +run_test (char *test, char *expected_result, enum url_scheme scheme, + bool expected_change) { char *test_copy = xstrdup (test); - bool modified = path_simplify (test_copy); + bool modified = path_simplify (scheme, test_copy); if (0 != strcmp (test_copy, expected_result)) { printf ("Failed path_simplify(\"%s\"): expected \"%s\", got \"%s\".\n", test, expected_result, test_copy); + mu_assert ("", 0); } if (modified != expected_change) { @@ -2048,51 +2133,60 @@ run_test (char *test, char *expected_result, bool expected_change) test); } xfree (test_copy); + mu_assert ("", modified == expected_change); + return NULL; } -static void +const char * test_path_simplify (void) { static struct { char *test, *result; + enum url_scheme scheme; bool should_modify; } tests[] = { - { "", "", 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 } + { "", "", SCHEME_HTTP, false }, + { ".", "", SCHEME_HTTP, true }, + { "./", "", SCHEME_HTTP, true }, + { "..", "", SCHEME_HTTP, true }, + { "../", "", SCHEME_HTTP, true }, + { "..", "..", SCHEME_FTP, false }, + { "../", "../", SCHEME_FTP, false }, + { "foo", "foo", SCHEME_HTTP, false }, + { "foo/bar", "foo/bar", SCHEME_HTTP, false }, + { "foo///bar", "foo///bar", SCHEME_HTTP, false }, + { "foo/.", "foo/", SCHEME_HTTP, true }, + { "foo/./", "foo/", SCHEME_HTTP, true }, + { "foo./", "foo./", SCHEME_HTTP, false }, + { "foo/../bar", "bar", SCHEME_HTTP, true }, + { "foo/../bar/", "bar/", SCHEME_HTTP, true }, + { "foo/bar/..", "foo/", SCHEME_HTTP, true }, + { "foo/bar/../x", "foo/x", SCHEME_HTTP, true }, + { "foo/bar/../x/", "foo/x/", SCHEME_HTTP, true }, + { "foo/..", "", SCHEME_HTTP, true }, + { "foo/../..", "", SCHEME_HTTP, true }, + { "foo/../../..", "", SCHEME_HTTP, true }, + { "foo/../../bar/../../baz", "baz", SCHEME_HTTP, true }, + { "foo/../..", "..", SCHEME_FTP, true }, + { "foo/../../..", "../..", SCHEME_FTP, true }, + { "foo/../../bar/../../baz", "../../baz", SCHEME_FTP, true }, + { "a/b/../../c", "c", SCHEME_HTTP, true }, + { "./a/../b", "b", SCHEME_HTTP, true } }; int i; for (i = 0; i < countof (tests); i++) { + const char *message; char *test = tests[i].test; char *expected_result = tests[i].result; + enum url_scheme scheme = tests[i].scheme; bool expected_change = tests[i].should_modify; - run_test (test, expected_result, expected_change); + message = run_test (test, expected_result, scheme, expected_change); + if (message) return message; } + return NULL; } -#endif - -#ifdef TESTING const char * test_append_uri_pathel() @@ -2106,7 +2200,7 @@ test_append_uri_pathel() } 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;