util_uri.c revision 2d71630471d1c23f0137309e3c3957c633ecbfd6
* 3. All advertising materials mentioning features or use of this * software must display the following acknowledgment: * "This product includes software developed by the Apache Group * 4. The names "Apache Server" and "Apache Group" must not be used to * endorse or promote products derived from this software without * prior written permission. * 5. Products derived from this software may not be called "Apache" * nor may "Apache" appear in their names without prior written * permission of the Apache Group. * 6. Redistributions of any form whatsoever must retain the following * "This product includes software developed by the Apache Group * THIS SOFTWARE IS PROVIDED BY THE APACHE GROUP ``AS IS'' AND ANY * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE APACHE GROUP OR * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED * OF THE POSSIBILITY OF SUCH DAMAGE. * ==================================================================== * This software consists of voluntary contributions made by many * individuals on behalf of the Apache Group and was originally based * on public domain software written at the National Center for * Supercomputing Applications, University of Illinois, Urbana-Champaign. * For more information on the Apache Group and the Apache HTTP server /* Some WWW schemes and their default ports; this is basically /etc/services */ /* This will become global when the protocol abstraction comes */ /* As the schemes are searched by a linear search, */ /* they are sorted by their expected frequency */ {
NULL,
0xFFFF }
/* unknown port *//* Create a copy of a "struct hostent" record; it was presumably returned * from a call to gethostbyname() and lives in static storage. * By creating a copy we can tuck it away for later use. /* Count number of alias entries */ /* Count number of in_addr entries */ /* Allocate hostent structure, alias ptrs, addr ptrs, addrs */ /* Copy address entries */ /* pgethostbyname(): resolve hostname, if successful return an ALLOCATED * COPY OF the hostent structure, intended to be stored and used later. * (gethostbyname() uses static storage that would be overwritten on each call) /* Unparse a uri_components structure to an URI string. * Optionally suppress the password for security reasons. /* If suppressing the site part, omit both user name & scheme://hostname */ /* Construct a "user:password@" string, honoring the passed UNP_ flags: */ /* Construct scheme://site string */ /* Should we suppress all path info? */ /* Append path, query and fragment strings: */ /* The regex version of parse_uri_components has the advantage that it is * relatively easy to understand and extend. But it has the disadvantage * that the regexes are complex enough that regex libraries really * don't do a great job with them performancewise. * The default is a hand coded scanner that is two orders of magnitude /* This is a modified version of the regex that appeared in * draft-fielding-uri-syntax-01. It doesnt allow the uri to contain a * scheme but no hostinfo or vice versa. * Although the BNF defines what is allowed in each component, it is * ambiguous in terms of differentiating between a site component and * a path component that begins with two slash characters. * RFC2068 disambiguates this for the Request-URI, which may only ever be * the "abs_path" portion of the URI. So a request "GET //foo/bar * HTTP/1.1" is really referring to the path //foo/bar, not the host foo, * path /bar. Nowhere in RFC2068 is it possible to have a scheme but no * hostinfo or a hostinfo but no scheme. (Unless you're proxying a * protocol other than HTTP, but this parsing engine probably won't work re_str =
"^(([^:/?#]+)://([^/?#]*))?([^?#]*)(\\?([^#]*))?(#(.*))?$";
/* ^scheme--^ ^site---^ ^path--^ ^query^ ^frag */ /* Make a readable error message */ "Internal error: regcomp(\"%s\") returned non-zero (%s) - " "possibly due to broken regex lib! " "Did you define WANTHSREGEX=yes?",
/* This is a sub-RE which will break down the hostinfo part, * i.e., user, password, hostname and port. re_str =
"^(([^:]*)(:(.*))?@)?([^@:]*)(:([0-9]*))?$";
/* ^^user^ :pw ^host^ ^:[port]^ */ /* Make a readable error message */ "Internal error: regcomp(\"%s\") returned non-zero (%s) - " "possibly due to broken regex lib! " "Did you define WANTHSREGEX=yes?",
/* parse_uri_components(): * Parse a given URI, fill in all supplied fields of a uri_components * structure. This eliminates the necessity of extracting host, port, * path, query info repeatedly in the modules. * - fills in fields of uri_components *uptr * - none on any of the r->* fields * as there are braces in the re_strings */ /* Initialize the structure. parse_uri() and parse_uri_components() * can be called more than once per request. "ap_regexec() could not parse uri (\"%s\")",
/* empty hostinfo is valid, that's why we test $1 but use $3 */ /* empty query string is valid, that's why we test $5 but use $6 */ /* empty fragment is valid, test $7 use $8 */ /* Parse the hostinfo part to extract user, password, host, and port */ "ap_regexec() could not parse (\"%s\") as host part",
/* "^(([^:]*)(:(.*))?@)?([^@:]*)(:([0-9]*))?$" */ /* ^^user^ :pw ^host^ ^:[port]^ */ /* empty user is valid, that's why we test $1 but use $2 */ /* empty password is valid, test $3 but use $4 */ /* empty hostname is valid, and implied by the existence of hostinfo */ /* Note that the port string can be empty. * If it is, we use the default port associated with the scheme /* Invalid characters after ':' found */ /* Here is the hand-optimized parse_uri_components(). There are some wild * tricks we could pull in assembly language that we don't pull here... like we * can do word-at-time scans for delimiter characters using the same technique * that fast memchr()s use. But that would be way non-portable. -djg /* We have a ap_table_t that we can index by character and it tells us if the * character is one of the interesting delimiters. Note that we even get * compares for NUL for free -- it's just another delimiter. #
define T_NUL 0x80 /* '\0' */ if (uri_delims[ch] & NOTEND_foobar) { then we're not at a delimiter for foobar /* Note that we optimize the scheme scanning here, we cheat and let the * compiler know that it doesn't have to do the & masking. /* parse_uri_components(): * Parse a given URI, fill in all supplied fields of a uri_components * structure. This eliminates the necessity of extracting host, port, * path, query info repeatedly in the modules. * - fills in fields of uri_components *uptr * - none on any of the r->* fields /* Initialize the structure. parse_uri() and parse_uri_components() * can be called more than once per request. /* We assume the processor has a branch predictor like most -- * it assumes forward branches are untaken and backwards are taken. That's * the reason for the gotos. -djg /* we expect uri to point to first character of path ... remember /* otherwise it's a fragment */ /* scheme must be non-empty and followed by :// */ if (s ==
uri || s[0] !=
':' || s[
1] !=
'/' || s[
2] !=
'/') {
uri = s;
/* whatever follows hostinfo is start of uri */ /* If there's a username:password@host:port, the @ we want is the last @... * too bad there's no memrchr()... For the C purists, note that hostinfo * is definately not the first character of the original uri so therefore * &hostinfo[-1] < &hostinfo[0] ... and this loop is valid C. /* again we want the common case to be fall through */ /* We expect hostinfo to point to the first character of * the hostname. If there's a port it is the first colon. /* we expect the common case to have no port */ /* Invalid characters after ':' found */ /* first colon delimits username:password */ /* Special case for CONNECT parsing: it comes with the hostinfo part only */ /* See the INTERNET-DRAFT document "Tunneling SSL Through a WWW Proxy" * for the format of the "CONNECT host:port HTTP/1.0" request /* Initialize the structure. parse_uri() and parse_uri_components() * can be called more than once per request. /* We expect hostinfo to point to the first character of * the hostname. There must be a port, separated by a colon /* Invalid characters after ':' found */