X-Git-Url: http://www.privoxy.org/gitweb/?p=privoxy.git;a=blobdiff_plain;f=urlmatch.c;h=a028361e0c17dbb192db78f31fb8180d71606ec5;hp=1b2aee86f89733797c03cb35383e3daeceee1291;hb=283e98e3f6ca776c1a15cd076e6ae1e999d8a6b3;hpb=d0c1f6261a9773bd2271117cdde20d734d320a93 diff --git a/urlmatch.c b/urlmatch.c index 1b2aee86..a028361e 100644 --- a/urlmatch.c +++ b/urlmatch.c @@ -1,4 +1,3 @@ -const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.9 2002/04/04 00:36:36 gliptak Exp $"; /********************************************************************* * * File : $Source: /cvsroot/ijbswa/current/urlmatch.c,v $ @@ -6,8 +5,8 @@ const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.9 2002/04/04 00:36:36 gliptak E * Purpose : Declares functions to match URLs against URL * patterns. * - * Copyright : Written by and Copyright (C) 2001 the SourceForge - * Privoxy team. http://www.privoxy.org/ + * Copyright : Written by and Copyright (C) 2001-2020 + * the Privoxy team. https://www.privoxy.org/ * * Based on the Internet Junkbuster originally written * by and Copyright (C) 1997 Anonymous Coders and @@ -31,53 +30,8 @@ const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.9 2002/04/04 00:36:36 gliptak E * or write to the Free Software Foundation, Inc., 59 * Temple Place - Suite 330, Boston, MA 02111-1307, USA. * - * Revisions : - * $Log: urlmatch.c,v $ - * Revision 1.9 2002/04/04 00:36:36 gliptak - * always use pcre for matching - * - * Revision 1.8 2002/04/03 23:32:47 jongfoster - * Fixing memory leak on error - * - * Revision 1.7 2002/03/26 22:29:55 swa - * we have a new homepage! - * - * Revision 1.6 2002/03/24 13:25:43 swa - * name change related issues - * - * Revision 1.5 2002/03/13 00:27:05 jongfoster - * Killing warnings - * - * Revision 1.4 2002/03/07 03:46:17 oes - * Fixed compiler warnings - * - * Revision 1.3 2002/03/03 14:51:11 oes - * Fixed CLF logging: Added ocmd member for client's request to struct http_request - * - * Revision 1.2 2002/01/21 00:14:09 jongfoster - * Correcting comment style - * Fixing an uninitialized memory bug in create_url_spec() - * - * Revision 1.1 2002/01/17 20:53:46 jongfoster - * Moving all our URL and URL pattern parsing code to the same file - it - * was scattered around in filters.c, loaders.c and parsers.c. - * - * Providing a single, simple url_match(pattern,url) function - rather than - * the 3-line match routine which was repeated all over the place. - * - * Renaming free_url to free_url_spec, since it frees a struct url_spec. - * - * Providing parse_http_url() so that URLs can be parsed without faking a - * HTTP request line for parse_http_request() or repeating the parsing - * code (both of which were techniques that were actually in use). - * - * Standardizing that struct http_request is used to represent a URL, and - * struct url_spec is used to represent a URL pattern. (Before, URLs were - * represented as seperate variables and a partially-filled-in url_spec). - * - * *********************************************************************/ - + #include "config.h" @@ -101,8 +55,17 @@ const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.9 2002/04/04 00:36:36 gliptak E #include "miscutil.h" #include "errlog.h" -const char urlmatch_h_rcs[] = URLMATCH_H_VERSION; - +enum regex_anchoring +{ + NO_ANCHORING, + LEFT_ANCHORED, + RIGHT_ANCHORED, + RIGHT_ANCHORED_HOST +}; +static jb_err compile_vanilla_host_pattern(struct pattern_spec *url, const char *host_pattern); +#ifdef FEATURE_PCRE_HOST_PATTERNS +static jb_err compile_pcre_host_pattern(struct pattern_spec *url, const char *host_pattern); +#endif /********************************************************************* * @@ -127,7 +90,7 @@ void free_http_request(struct http_request *http) freez(http->url); freez(http->hostport); freez(http->path); - freez(http->ver); + freez(http->version); freez(http->host_ip_addr_str); freez(http->dbuffer); freez(http->dvec); @@ -135,6 +98,109 @@ void free_http_request(struct http_request *http) } +/********************************************************************* + * + * Function : init_domain_components + * + * Description : Splits the domain name so we can compare it + * against wildcards. It used to be part of + * parse_http_url, but was separated because the + * same code is required in chat in case of + * intercepted requests. + * + * Parameters : + * 1 : http = pointer to the http structure to hold elements. + * + * Returns : JB_ERR_OK on success + * JB_ERR_PARSE on malformed command/URL + * or >100 domains deep. + * + *********************************************************************/ +jb_err init_domain_components(struct http_request *http) +{ + char *vec[BUFFER_SIZE]; + size_t size; + char *p; + + http->dbuffer = strdup_or_die(http->host); + + /* map to lower case */ + for (p = http->dbuffer; *p ; p++) + { + *p = (char)privoxy_tolower(*p); + } + + /* split the domain name into components */ + http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec)); + + if (http->dcount <= 0) + { + /* + * Error: More than SZ(vec) components in domain + * or: no components in domain + */ + log_error(LOG_LEVEL_ERROR, "More than SZ(vec) components in domain or none at all."); + return JB_ERR_PARSE; + } + + /* save a copy of the pointers in dvec */ + size = (size_t)http->dcount * sizeof(*http->dvec); + + http->dvec = malloc_or_die(size); + + memcpy(http->dvec, vec, size); + + return JB_ERR_OK; +} + + +/********************************************************************* + * + * Function : url_requires_percent_encoding + * + * Description : Checks if an URL contains invalid characters + * according to RFC 3986 that should be percent-encoded. + * Does not verify whether or not the passed string + * actually is a valid URL. + * + * Parameters : + * 1 : url = URL to check + * + * Returns : True in case of valid URLs, false otherwise + * + *********************************************************************/ +int url_requires_percent_encoding(const char *url) +{ + static const char allowed_characters[128] = { + '\0', '\0', '\0', '\0', '\0', '\0', '\0', '\0', '\0', '\0', + '\0', '\0', '\0', '\0', '\0', '\0', '\0', '\0', '\0', '\0', + '\0', '\0', '\0', '\0', '\0', '\0', '\0', '\0', '\0', '\0', + '\0', '\0', '\0', '!', '\0', '#', '$', '%', '&', '\'', + '(', ')', '*', '+', ',', '-', '.', '/', '0', '1', + '2', '3', '4', '5', '6', '7', '8', '9', ':', ';', + '\0', '=', '\0', '?', '@', 'A', 'B', 'C', 'D', 'E', + 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', + 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', + 'Z', '[', '\0', ']', '\0', '_', '\0', 'a', 'b', 'c', + 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm', + 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', + 'x', 'y', 'z', '\0', '\0', '\0', '~', '\0' + }; + + while (*url != '\0') + { + const unsigned int i = (unsigned char)*url++; + if (i >= sizeof(allowed_characters) || '\0' == allowed_characters[i]) + { + return TRUE; + } + } + + return FALSE; + +} + + /********************************************************************* * * Function : parse_http_url @@ -145,34 +211,36 @@ void free_http_request(struct http_request *http) * Parameters : * 1 : url = URL (or is it URI?) to break down * 2 : http = pointer to the http structure to hold elements. - * Will be zeroed before use. Note that this - * function sets the http->gpc and http->ver - * members to NULL. - * 3 : csp = Current client state (buffers, headers, etc...) + * Must be initialized with valid values (like NULLs). + * 3 : require_protocol = Whether or not URLs without + * protocol are acceptable. * * Returns : JB_ERR_OK on success - * JB_ERR_MEMORY on out of memory - * JB_ERR_CGI_PARAMS on malformed command/URL - * or >100 domains deep. + * JB_ERR_PARSE on malformed command/URL + * or >100 domains deep. * *********************************************************************/ -jb_err parse_http_url(const char * url, - struct http_request *http, - struct client_state *csp) +jb_err parse_http_url(const char *url, struct http_request *http, int require_protocol) { + int host_available = 1; /* A proxy can dream. */ + /* - * Zero out the results structure + * Save our initial URL */ - memset(http, '\0', sizeof(*http)); - + http->url = strdup_or_die(url); /* - * Save our initial URL + * Check for * URI. If found, we're done. */ - http->url = strdup(url); - if (http->url == NULL) + if (*http->url == '*') { - return JB_ERR_MEMORY; + http->path = strdup_or_die("*"); + http->hostport = strdup_or_die(""); + if (http->url[1] != '\0') + { + return JB_ERR_PARSE; + } + return JB_ERR_OK; } @@ -184,27 +252,36 @@ jb_err parse_http_url(const char * url, char *url_noproto; char *url_path; - buf = strdup(url); - if (buf == NULL) - { - return JB_ERR_MEMORY; - } + buf = strdup_or_die(url); /* Find the start of the URL in our scratch space */ url_noproto = buf; if (strncmpic(url_noproto, "http://", 7) == 0) { url_noproto += 7; - http->ssl = 0; } else if (strncmpic(url_noproto, "https://", 8) == 0) { + /* + * Should only happen when called from cgi_show_url_info(). + */ url_noproto += 8; http->ssl = 1; } - else + else if (*url_noproto == '/') + { + /* + * Short request line without protocol and host. + * Most likely because the client's request + * was intercepted and redirected into Privoxy. + */ + http->host = NULL; + host_available = 0; + } + else if (require_protocol) { - http->ssl = 0; + freez(buf); + return JB_ERR_PARSE; } url_path = strchr(url_noproto, '/'); @@ -213,14 +290,19 @@ jb_err parse_http_url(const char * url, /* * Got a path. * - * NOTE: The following line ignores the path for HTTPS URLS. - * This means that you get consistent behaviour if you type a - * https URL in and it's parsed by the function. (When the - * URL is actually retrieved, SSL hides the path part). + * If FEATURE_HTTPS_INSPECTION isn't available, ignore the + * path for https URLs so that we get consistent behaviour + * if a https URL is parsed. When the URL is actually + * retrieved, https hides the path part. */ - http->path = strdup(http->ssl ? "/" : url_path); + http->path = strdup_or_die( +#ifndef FEATURE_HTTPS_INSPECTION + http->ssl ? "/" : +#endif + url_path + ); *url_path = '\0'; - http->hostport = strdup(url_noproto); + http->hostport = strdup_or_die(url_noproto); } else { @@ -228,21 +310,18 @@ jb_err parse_http_url(const char * url, * Repair broken HTTP requests that don't contain a path, * or CONNECT requests */ - http->path = strdup("/"); - http->hostport = strdup(url_noproto); + http->path = strdup_or_die("/"); + http->hostport = strdup_or_die(url_noproto); } - free(buf); - - if ( (http->path == NULL) - || (http->hostport == NULL)) - { - free(buf); - free_http_request(http); - return JB_ERR_MEMORY; - } + freez(buf); } + if (!host_available) + { + /* Without host, there is nothing left to do here */ + return JB_ERR_OK; + } /* * Split hostport into user/password (ignored), host, port. @@ -252,12 +331,7 @@ jb_err parse_http_url(const char * url, char *host; char *port; - buf = strdup(http->hostport); - if (buf == NULL) - { - free_http_request(http); - return JB_ERR_MEMORY; - } + buf = strdup_or_die(http->hostport); /* check if url contains username and/or password */ host = strchr(buf, '@'); @@ -272,82 +346,174 @@ jb_err parse_http_url(const char * url, host = buf; } + /* Move after hostname before port number */ + if (*host == '[') + { + /* Numeric IPv6 address delimited by brackets */ + host++; + port = strchr(host, ']'); + + if (port == NULL) + { + /* Missing closing bracket */ + freez(buf); + return JB_ERR_PARSE; + } + + *port++ = '\0'; + + if (*port == '\0') + { + port = NULL; + } + else if (*port != ':') + { + /* Garbage after closing bracket */ + freez(buf); + return JB_ERR_PARSE; + } + } + else + { + /* Plain non-escaped hostname */ + port = strchr(host, ':'); + } + /* check if url contains port */ - port = strchr(host, ':'); if (port != NULL) { /* Contains port */ + char *endptr; + long parsed_port; /* Terminate hostname and point to start of port string */ *port++ = '\0'; - http->port = atoi(port); + parsed_port = strtol(port, &endptr, 10); + if ((parsed_port <= 0) || (parsed_port > 65535) || (*endptr != '\0')) + { + log_error(LOG_LEVEL_ERROR, "Invalid port in URL: %s.", url); + freez(buf); + return JB_ERR_PARSE; + } + http->port = (int)parsed_port; } else { /* No port specified. */ - http->port = (http->ssl ? 143 : 80); + http->port = (http->ssl ? 443 : 80); } - http->host = strdup(host); + http->host = strdup_or_die(host); - free(buf); - - if (http->host == NULL) - { - free_http_request(http); - return JB_ERR_MEMORY; - } + freez(buf); } + /* Split domain name so we can compare it against wildcards */ + return init_domain_components(http); - /* - * Split domain name so we can compare it against wildcards - */ - { - char *vec[BUFFER_SIZE]; - size_t size; - char *p; +} - http->dbuffer = strdup(http->host); - if (NULL == http->dbuffer) - { - free_http_request(http); - return JB_ERR_MEMORY; - } - /* map to lower case */ - for (p = http->dbuffer; *p ; p++) +/********************************************************************* + * + * Function : unknown_method + * + * Description : Checks whether a method is unknown. + * + * Parameters : + * 1 : method = points to a http method + * + * Returns : TRUE if it's unknown, FALSE otherwise. + * + *********************************************************************/ +static int unknown_method(const char *method) +{ + static const char * const known_http_methods[] = { + /* Basic HTTP request type */ + "GET", "HEAD", "POST", "PUT", "DELETE", "OPTIONS", "TRACE", "CONNECT", + /* webDAV extensions (RFC2518) */ + "PROPFIND", "PROPPATCH", "MOVE", "COPY", "MKCOL", "LOCK", "UNLOCK", + /* + * Microsoft webDAV extension for Exchange 2000. See: + * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html + * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp + */ + "BCOPY", "BMOVE", "BDELETE", "BPROPFIND", "BPROPPATCH", + /* + * Another Microsoft webDAV extension for Exchange 2000. See: + * http://systems.cs.colorado.edu/grunwald/MobileComputing/Papers/draft-cohen-gena-p-base-00.txt + * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html + * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp + */ + "SUBSCRIBE", "UNSUBSCRIBE", "NOTIFY", "POLL", + /* + * Yet another WebDAV extension, this time for + * Web Distributed Authoring and Versioning (RFC3253) + */ + "VERSION-CONTROL", "REPORT", "CHECKOUT", "CHECKIN", "UNCHECKOUT", + "MKWORKSPACE", "UPDATE", "LABEL", "MERGE", "BASELINE-CONTROL", "MKACTIVITY", + /* + * The PATCH method is defined by RFC5789, the format of the + * actual patch in the body depends on the application, but from + * Privoxy's point of view it doesn't matter. + */ + "PATCH", + }; + int i; + + for (i = 0; i < SZ(known_http_methods); i++) + { + if (0 == strcmpic(method, known_http_methods[i])) { - *p = tolower((int)(unsigned char)*p); + return FALSE; } + } - /* split the domain name into components */ - http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1); + return TRUE; - if (http->dcount <= 0) - { - /* - * Error: More than SZ(vec) components in domain - * or: no components in domain - */ - free_http_request(http); - return JB_ERR_PARSE; - } +} - /* save a copy of the pointers in dvec */ - size = http->dcount * sizeof(*http->dvec); - http->dvec = (char **)malloc(size); - if (NULL == http->dvec) - { - free_http_request(http); - return JB_ERR_MEMORY; - } +/********************************************************************* + * + * Function : normalize_http_version + * + * Description : Take a supported HTTP version string and remove + * leading zeroes etc., reject unsupported versions. + * + * This is an explicit RFC 2616 (3.1) MUST and + * RFC 7230 mandates that intermediaries send their + * own HTTP-version in forwarded messages. + * + * Parameters : + * 1 : http_version = HTTP version string + * + * Returns : JB_ERR_OK on success + * JB_ERR_PARSE if the HTTP version is unsupported + * + *********************************************************************/ +static jb_err normalize_http_version(char *http_version) +{ + unsigned int major_version; + unsigned int minor_version; - memcpy(http->dvec, vec, size); + if (2 != sscanf(http_version, "HTTP/%u.%u", &major_version, &minor_version)) + { + log_error(LOG_LEVEL_ERROR, "Unsupported HTTP version: %s", http_version); + return JB_ERR_PARSE; + } + + if (major_version != 1 || (minor_version != 0 && minor_version != 1)) + { + log_error(LOG_LEVEL_ERROR, "The only supported HTTP " + "versions are 1.0 and 1.1. This rules out: %s", http_version); + return JB_ERR_PARSE; } + assert(strlen(http_version) >= 8); + snprintf(http_version, 9, "HTTP/%u.%u", major_version, minor_version); return JB_ERR_OK; + } @@ -361,420 +527,808 @@ jb_err parse_http_url(const char * url, * Parameters : * 1 : req = HTTP request line to break down * 2 : http = pointer to the http structure to hold elements - * 3 : csp = Current client state (buffers, headers, etc...) * * Returns : JB_ERR_OK on success - * JB_ERR_MEMORY on out of memory * JB_ERR_CGI_PARAMS on malformed command/URL * or >100 domains deep. * *********************************************************************/ -jb_err parse_http_request(const char *req, - struct http_request *http, - struct client_state *csp) +jb_err parse_http_request(const char *req, struct http_request *http) { char *buf; - char *v[10]; + char *v[3]; int n; jb_err err; - int is_connect = 0; memset(http, '\0', sizeof(*http)); - buf = strdup(req); - if (buf == NULL) - { - return JB_ERR_MEMORY; - } + buf = strdup_or_die(req); - n = ssplit(buf, " \r\n", v, SZ(v), 1, 1); + n = ssplit(buf, " \r\n", v, SZ(v)); if (n != 3) { - free(buf); + freez(buf); return JB_ERR_PARSE; } - /* this could be a CONNECT request */ - if (strcmpic(v[0], "connect") == 0) - { - /* Secure */ - is_connect = 1; - } - /* or it could be any other basic HTTP request type */ - else if ((0 == strcmpic(v[0], "get")) - || (0 == strcmpic(v[0], "head")) - || (0 == strcmpic(v[0], "post")) - || (0 == strcmpic(v[0], "put")) - || (0 == strcmpic(v[0], "delete")) - - /* or a webDAV extension (RFC2518) */ - || (0 == strcmpic(v[0], "propfind")) - || (0 == strcmpic(v[0], "proppatch")) - || (0 == strcmpic(v[0], "move")) - || (0 == strcmpic(v[0], "copy")) - || (0 == strcmpic(v[0], "mkcol")) - || (0 == strcmpic(v[0], "lock")) - || (0 == strcmpic(v[0], "unlock")) - ) - { - /* Normal */ - is_connect = 0; + /* + * Fail in case of unknown methods + * which we might not handle correctly. + * + * XXX: There should be a config option + * to forward requests with unknown methods + * anyway. Most of them don't need special + * steps. + */ + if (unknown_method(v[0])) + { + log_error(LOG_LEVEL_ERROR, "Unknown HTTP method detected: %s", v[0]); + freez(buf); + return JB_ERR_PARSE; } - else + + if (JB_ERR_OK != normalize_http_version(v[2])) { - /* Unknown HTTP method */ - free(buf); + freez(buf); return JB_ERR_PARSE; } - err = parse_http_url(v[1], http, csp); + http->ssl = !strcmpic(v[0], "CONNECT"); + + err = parse_http_url(v[1], http, !http->ssl); if (err) { - free(buf); + freez(buf); return err; } /* * Copy the details into the structure */ - http->ssl = is_connect; - http->cmd = strdup(req); - http->gpc = strdup(v[0]); - http->ver = strdup(v[2]); + http->cmd = strdup_or_die(req); + http->gpc = strdup_or_die(v[0]); + http->version = strdup_or_die(v[2]); + http->ocmd = strdup_or_die(http->cmd); - if ( (http->cmd == NULL) - || (http->gpc == NULL) - || (http->ver == NULL) ) - { - free(buf); - free_http_request(http); - return JB_ERR_MEMORY; - } + freez(buf); return JB_ERR_OK; + } /********************************************************************* * - * Function : simple_domaincmp + * Function : compile_pattern * - * Description : Domain-wise Compare fqdn's. The comparison is - * both left- and right-anchored. The individual - * domain names are compared with simplematch(). - * This is only used by domain_match. + * Description : Compiles a host, domain or TAG pattern. * * Parameters : - * 1 : pv = array of patterns to compare - * 2 : fv = array of domain components to compare - * 3 : len = length of the arrays (both arrays are the - * same length - if they weren't, it couldn't - * possibly be a match). + * 1 : pattern = The pattern to compile. + * 2 : anchoring = How the regex should be modified + * before compilation. Can be either + * one of NO_ANCHORING, LEFT_ANCHORED, + * RIGHT_ANCHORED or RIGHT_ANCHORED_HOST. + * 3 : url = In case of failures, the spec member is + * logged and the structure freed. + * 4 : regex = Where the compiled regex should be stored. * - * Returns : 0 => domains are equivalent, else no match. + * Returns : JB_ERR_OK - Success + * JB_ERR_PARSE - Cannot parse regex * *********************************************************************/ -static int simple_domaincmp(char **pv, char **fv, int len) +static jb_err compile_pattern(const char *pattern, enum regex_anchoring anchoring, + struct pattern_spec *url, regex_t **regex) { - int n; + int errcode; + const char *fmt = NULL; + char *rebuf; + size_t rebuf_size; - for (n = 0; n < len; n++) + assert(pattern); + + if (pattern[0] == '\0') { - if (simplematch(pv[n], fv[n])) + *regex = NULL; + return JB_ERR_OK; + } + + switch (anchoring) + { + case NO_ANCHORING: + fmt = "%s"; + break; + case RIGHT_ANCHORED: + fmt = "%s$"; + break; + case RIGHT_ANCHORED_HOST: + fmt = "%s\\.?$"; + break; + case LEFT_ANCHORED: + fmt = "^%s"; + break; + default: + log_error(LOG_LEVEL_FATAL, + "Invalid anchoring in compile_pattern %d", anchoring); + } + rebuf_size = strlen(pattern) + strlen(fmt); + rebuf = malloc_or_die(rebuf_size); + *regex = zalloc_or_die(sizeof(**regex)); + + snprintf(rebuf, rebuf_size, fmt, pattern); + + errcode = regcomp(*regex, rebuf, (REG_EXTENDED|REG_NOSUB|REG_ICASE)); + + if (errcode) + { + size_t errlen = regerror(errcode, *regex, rebuf, rebuf_size); + if (errlen > (rebuf_size - (size_t)1)) { - return 1; + errlen = rebuf_size - (size_t)1; } + rebuf[errlen] = '\0'; + log_error(LOG_LEVEL_ERROR, "error compiling %s from %s: %s", + pattern, url->spec, rebuf); + free_pattern_spec(url); + freez(rebuf); + + return JB_ERR_PARSE; } + freez(rebuf); - return 0; + return JB_ERR_OK; } /********************************************************************* * - * Function : domain_match + * Function : compile_url_pattern * - * Description : Domain-wise Compare fqdn's. Governed by the bimap in - * pattern->unachored, the comparison is un-, left-, - * right-anchored, or both. - * The individual domain names are compared with - * simplematch(). + * Description : Compiles the three parts of an URL pattern. * * Parameters : - * 1 : pattern = a domain that may contain a '*' as a wildcard. - * 2 : fqdn = domain name against which the patterns are compared. + * 1 : url = Target pattern_spec to be filled in. + * 2 : buf = The url pattern to compile. Will be messed up. * - * Returns : 0 => domains are equivalent, else no match. + * Returns : JB_ERR_OK - Success + * JB_ERR_MEMORY - Out of memory + * JB_ERR_PARSE - Cannot parse regex * *********************************************************************/ -static int domain_match(const struct url_spec *pattern, const struct http_request *fqdn) +static jb_err compile_url_pattern(struct pattern_spec *url, char *buf) { - char **pv, **fv; /* vectors */ - int plen, flen; - int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT); - - plen = pattern->dcount; - flen = fqdn->dcount; + char *p; - if (flen < plen) +#ifdef FEATURE_PCRE_HOST_PATTERNS + const size_t prefix_length = 18; + if (strncmpic(buf, "PCRE-HOST-PATTERN:", prefix_length) == 0) { - /* fqdn is too short to match this pattern */ - return 1; + url->pattern.url_spec.host_regex_type = PCRE_HOST_PATTERN; + /* Overwrite the "PCRE-HOST-PATTERN:" prefix */ + memmove(buf, buf+prefix_length, strlen(buf+prefix_length)+1); } + else + { + url->pattern.url_spec.host_regex_type = VANILLA_HOST_PATTERN; + } +#endif - pv = pattern->dvec; - fv = fqdn->dvec; - - if (unanchored == ANCHOR_LEFT) + p = strchr(buf, '/'); + if (NULL != p) { /* - * Right anchored. - * - * Convert this into a fully anchored pattern with - * the fqdn and pattern the same length + * Only compile the regex if it consists of more than + * a single slash, otherwise it wouldn't affect the result. */ - fv += (flen - plen); /* flen - plen >= 0 due to check above */ - return simple_domaincmp(pv, fv, plen); - } - else if (unanchored == 0) - { - /* Fully anchored, check length */ - if (flen != plen) + if (p[1] != '\0') { - return 1; + /* + * XXX: does it make sense to compile the slash at the beginning? + */ + jb_err err = compile_pattern(p, LEFT_ANCHORED, url, &url->pattern.url_spec.preg); + + if (JB_ERR_OK != err) + { + return err; + } } - return simple_domaincmp(pv, fv, plen); + *p = '\0'; } - else if (unanchored == ANCHOR_RIGHT) - { - /* Left anchored, ignore all extra in fqdn */ - return simple_domaincmp(pv, fv, plen); + + /* + * IPv6 numeric hostnames can contain colons, thus we need + * to delimit the hostname before the real port separator. + * As brackets are already used in the hostname pattern, + * we use angle brackets ('<', '>') instead. + */ + if ((buf[0] == '<') && (NULL != (p = strchr(buf + 1, '>')))) + { + *p++ = '\0'; + buf++; + + if (*p == '\0') + { + /* IPv6 address without port number */ + p = NULL; + } + else if (*p != ':') + { + /* Garbage after address delimiter */ + return JB_ERR_PARSE; + } } else { - /* Unanchored */ - int n; - int maxn = flen - plen; - for (n = 0; n <= maxn; n++) + p = strchr(buf, ':'); + } + + if (NULL != p) + { + *p++ = '\0'; + url->pattern.url_spec.port_list = strdup_or_die(p); + } + else + { + url->pattern.url_spec.port_list = NULL; + } + + if (buf[0] != '\0') + { +#ifdef FEATURE_PCRE_HOST_PATTERNS + if (url->pattern.url_spec.host_regex_type == PCRE_HOST_PATTERN) { - if (!simple_domaincmp(pv, fv, plen)) - { - return 0; - } - /* - * Doesn't match from start of fqdn - * Try skipping first part of fqdn - */ - fv++; + return compile_pcre_host_pattern(url, buf); + } + else +#endif + { + return compile_vanilla_host_pattern(url, buf); } - return 1; } + return JB_ERR_OK; + } +#ifdef FEATURE_PCRE_HOST_PATTERNS /********************************************************************* * - * Function : create_url_spec + * Function : compile_pcre_host_pattern * - * Description : Creates a "url_spec" structure from a string. - * When finished, free with unload_url(). + * Description : Parses and compiles a pcre host pattern. * * Parameters : - * 1 : url = Target url_spec to be filled in. Will be - * zeroed before use. - * 2 : buf = Source pattern, null terminated. NOTE: The - * contents of this buffer are destroyed by this - * function. If this function succeeds, the - * buffer is copied to url->spec. If this - * function fails, the contents of the buffer - * are lost forever. + * 1 : url = Target pattern_spec to be filled in. + * 2 : host_pattern = Host pattern to compile. * * Returns : JB_ERR_OK - Success * JB_ERR_MEMORY - Out of memory - * JB_ERR_PARSE - Cannot parse regex (Detailed message - * written to system log) + * JB_ERR_PARSE - Cannot parse regex * *********************************************************************/ -jb_err create_url_spec(struct url_spec * url, const char * buf) +static jb_err compile_pcre_host_pattern(struct pattern_spec *url, const char *host_pattern) { + return compile_pattern(host_pattern, RIGHT_ANCHORED_HOST, url, &url->pattern.url_spec.host_regex); +} +#endif /* def FEATURE_PCRE_HOST_PATTERNS */ + + +/********************************************************************* + * + * Function : compile_vanilla_host_pattern + * + * Description : Parses and "compiles" an old-school host pattern. + * + * Parameters : + * 1 : url = Target pattern_spec to be filled in. + * 2 : host_pattern = Host pattern to parse. + * + * Returns : JB_ERR_OK - Success + * JB_ERR_PARSE - Cannot parse regex + * + *********************************************************************/ +static jb_err compile_vanilla_host_pattern(struct pattern_spec *url, const char *host_pattern) +{ + char *v[150]; + size_t size; char *p; - assert(url); - assert(buf); + /* + * Parse domain part + */ + if (host_pattern[strlen(host_pattern) - 1] == '.') + { + url->pattern.url_spec.unanchored |= ANCHOR_RIGHT; + } + if (host_pattern[0] == '.') + { + url->pattern.url_spec.unanchored |= ANCHOR_LEFT; + } - /* Zero memory */ - memset(url, '\0', sizeof(*url)); + /* + * Split domain into components + */ + url->pattern.url_spec.dbuffer = strdup_or_die(host_pattern); - /* save a copy of the orignal specification */ - if ((url->spec = strdup(buf)) == NULL) + /* + * Map to lower case + */ + for (p = url->pattern.url_spec.dbuffer; *p ; p++) { - return JB_ERR_MEMORY; + *p = (char)privoxy_tolower(*p); } - if ((p = strchr(buf, '/')) != NULL) + /* + * Split the domain name into components + */ + url->pattern.url_spec.dcount = ssplit(url->pattern.url_spec.dbuffer, ".", v, SZ(v)); + + if (url->pattern.url_spec.dcount < 0) { - if (NULL == (url->path = strdup(p))) - { - freez(url->spec); - return JB_ERR_MEMORY; - } - url->pathlen = strlen(url->path); - *p = '\0'; + free_pattern_spec(url); + return JB_ERR_PARSE; } - else + else if (url->pattern.url_spec.dcount != 0) { - url->path = NULL; - url->pathlen = 0; + /* + * Save a copy of the pointers in dvec + */ + size = (size_t)url->pattern.url_spec.dcount * sizeof(*url->pattern.url_spec.dvec); + + url->pattern.url_spec.dvec = malloc_or_die(size); + + memcpy(url->pattern.url_spec.dvec, v, size); } - if (url->path) + /* + * else dcount == 0 in which case we needn't do anything, + * since dvec will never be accessed and the pattern will + * match all domains. + */ + return JB_ERR_OK; +} + + +/********************************************************************* + * + * Function : simplematch + * + * Description : String matching, with a (greedy) '*' wildcard that + * stands for zero or more arbitrary characters and + * character classes in [], which take both enumerations + * and ranges. + * + * Parameters : + * 1 : pattern = pattern for matching + * 2 : text = text to be matched + * + * Returns : 0 if match, else nonzero + * + *********************************************************************/ +static int simplematch(const char *pattern, const char *text) +{ + const unsigned char *pat = (const unsigned char *)pattern; + const unsigned char *txt = (const unsigned char *)text; + const unsigned char *fallback = pat; + int wildcard = 0; + + unsigned char lastchar = 'a'; + unsigned i; + unsigned char charmap[32]; + + while (*txt) { - int errcode; - char rebuf[BUFFER_SIZE]; - if (NULL == (url->preg = zalloc(sizeof(*url->preg)))) + /* EOF pattern but !EOF text? */ + if (*pat == '\0') { - freez(url->spec); - freez(url->path); - return JB_ERR_MEMORY; + if (wildcard) + { + pat = fallback; + } + else + { + return 1; + } } - sprintf(rebuf, "^(%s)", url->path); - - errcode = regcomp(url->preg, rebuf, - (REG_EXTENDED|REG_NOSUB|REG_ICASE)); - if (errcode) + /* '*' in the pattern? */ + if (*pat == '*') { - size_t errlen = regerror(errcode, - url->preg, rebuf, sizeof(rebuf)); - if (errlen > (sizeof(rebuf) - (size_t)1)) + /* The pattern ends afterwards? Speed up the return. */ + if (*++pat == '\0') { - errlen = sizeof(rebuf) - (size_t)1; + return 0; } - rebuf[errlen] = '\0'; - log_error(LOG_LEVEL_ERROR, "error compiling %s: %s", - url->spec, rebuf); + /* Else, set wildcard mode and remember position after '*' */ + wildcard = 1; + fallback = pat; + } - freez(url->spec); - freez(url->path); - regfree(url->preg); - freez(url->preg); + /* Character range specification? */ + if (*pat == '[') + { + memset(charmap, '\0', sizeof(charmap)); - return JB_ERR_PARSE; + while (*++pat != ']') + { + if (!*pat) + { + return 1; + } + else if (*pat == '-') + { + if ((*++pat == ']') || *pat == '\0') + { + return(1); + } + for (i = lastchar; i <= *pat; i++) + { + charmap[i / 8] |= (unsigned char)(1 << (i % 8)); + } + } + else + { + charmap[*pat / 8] |= (unsigned char)(1 << (*pat % 8)); + lastchar = *pat; + } + } + } /* -END- if Character range specification */ + + + /* + * Char match, or char range match? + */ + if ((*pat == *txt) + || (*pat == '?') + || ((*pat == ']') && (charmap[*txt / 8] & (1 << (*txt % 8))))) + { + /* + * Success: Go ahead + */ + pat++; + } + else if (!wildcard) + { + /* + * No match && no wildcard: No luck + */ + return 1; + } + else if (pat != fallback) + { + /* + * Increment text pointer if in char range matching + */ + if (*pat == ']') + { + txt++; + } + /* + * Wildcard mode && nonmatch beyond fallback: Rewind pattern + */ + pat = fallback; + /* + * Restart matching from current text pointer + */ + continue; } + txt++; } - if ((p = strchr(buf, ':')) == NULL) + + /* Cut off extra '*'s */ + if (*pat == '*') pat++; + + /* If this is the pattern's end, fine! */ + return(*pat); + +} + + +/********************************************************************* + * + * Function : simple_domaincmp + * + * Description : Domain-wise Compare fqdn's. The comparison is + * both left- and right-anchored. The individual + * domain names are compared with simplematch(). + * This is only used by domain_match. + * + * Parameters : + * 1 : pv = array of patterns to compare + * 2 : fv = array of domain components to compare + * 3 : len = length of the arrays (both arrays are the + * same length - if they weren't, it couldn't + * possibly be a match). + * + * Returns : 0 => domains are equivalent, else no match. + * + *********************************************************************/ +static int simple_domaincmp(char **pv, char **fv, int len) +{ + int n; + + for (n = 0; n < len; n++) { - url->port = 0; + if (simplematch(pv[n], fv[n])) + { + return 1; + } } - else + + return 0; + +} + + +/********************************************************************* + * + * Function : domain_match + * + * Description : Domain-wise Compare fqdn's. Governed by the bimap in + * p.pattern->unachored, the comparison is un-, left-, + * right-anchored, or both. + * The individual domain names are compared with + * simplematch(). + * + * Parameters : + * 1 : p = a domain that may contain a '*' as a wildcard. + * 2 : fqdn = domain name against which the patterns are compared. + * + * Returns : 0 => domains are equivalent, else no match. + * + *********************************************************************/ +static int domain_match(const struct pattern_spec *p, const struct http_request *fqdn) +{ + char **pv, **fv; /* vectors */ + int plen, flen; + int unanchored = p->pattern.url_spec.unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT); + + plen = p->pattern.url_spec.dcount; + flen = fqdn->dcount; + + if (flen < plen) { - *p++ = '\0'; - url->port = atoi(p); + /* fqdn is too short to match this pattern */ + return 1; } - if (buf[0] != '\0') - { - char *v[150]; - size_t size; + pv = p->pattern.url_spec.dvec; + fv = fqdn->dvec; - /* Parse domain part */ - if (buf[strlen(buf) - 1] == '.') + if (unanchored == ANCHOR_LEFT) + { + /* + * Right anchored. + * + * Convert this into a fully anchored pattern with + * the fqdn and pattern the same length + */ + fv += (flen - plen); /* flen - plen >= 0 due to check above */ + return simple_domaincmp(pv, fv, plen); + } + else if (unanchored == 0) + { + /* Fully anchored, check length */ + if (flen != plen) { - url->unanchored |= ANCHOR_RIGHT; + return 1; } - if (buf[0] == '.') + return simple_domaincmp(pv, fv, plen); + } + else if (unanchored == ANCHOR_RIGHT) + { + /* Left anchored, ignore all extra in fqdn */ + return simple_domaincmp(pv, fv, plen); + } + else + { + /* Unanchored */ + int n; + int maxn = flen - plen; + for (n = 0; n <= maxn; n++) { - url->unanchored |= ANCHOR_LEFT; + if (!simple_domaincmp(pv, fv, plen)) + { + return 0; + } + /* + * Doesn't match from start of fqdn + * Try skipping first part of fqdn + */ + fv++; } + return 1; + } - /* split domain into components */ +} - url->dbuffer = strdup(buf); - if (NULL == url->dbuffer) - { - freez(url->spec); - freez(url->path); - regfree(url->preg); - freez(url->preg); - return JB_ERR_MEMORY; - } - /* map to lower case */ - for (p = url->dbuffer; *p ; p++) - { - *p = tolower((int)(unsigned char)*p); - } +/********************************************************************* + * + * Function : create_pattern_spec + * + * Description : Creates a "pattern_spec" structure from a string. + * When finished, free with free_pattern_spec(). + * + * Parameters : + * 1 : pattern = Target pattern_spec to be filled in. + * Will be zeroed before use. + * 2 : buf = Source pattern, null terminated. NOTE: The + * contents of this buffer are destroyed by this + * function. If this function succeeds, the + * buffer is copied to pattern->spec. If this + * function fails, the contents of the buffer + * are lost forever. + * + * Returns : JB_ERR_OK - Success + * JB_ERR_PARSE - Cannot parse regex (Detailed message + * written to system log) + * + *********************************************************************/ +jb_err create_pattern_spec(struct pattern_spec *pattern, char *buf) +{ + static const struct + { + /** The tag pattern prefix to match */ + const char *prefix; + + /** The length of the prefix to match */ + const size_t prefix_length; + + /** The pattern flag */ + const unsigned flag; + } tag_pattern[] = { + { "TAG:", 4, PATTERN_SPEC_TAG_PATTERN}, + #ifdef FEATURE_CLIENT_TAGS + { "CLIENT-TAG:", 11, PATTERN_SPEC_CLIENT_TAG_PATTERN}, + #endif + { "NO-REQUEST-TAG:", 15, PATTERN_SPEC_NO_REQUEST_TAG_PATTERN}, + { "NO-RESPONSE-TAG:", 16, PATTERN_SPEC_NO_RESPONSE_TAG_PATTERN} + }; + int i; + + assert(pattern); + assert(buf); - /* split the domain name into components */ - url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1); + memset(pattern, '\0', sizeof(*pattern)); - if (url->dcount < 0) - { - freez(url->spec); - freez(url->path); - regfree(url->preg); - freez(url->preg); - freez(url->dbuffer); - url->dcount = 0; - return JB_ERR_MEMORY; - } - else if (url->dcount != 0) + /* Remember the original specification for the CGI pages. */ + pattern->spec = strdup_or_die(buf); + + /* Check if it's a tag pattern */ + for (i = 0; i < SZ(tag_pattern); i++) + { + if (0 == strncmpic(pattern->spec, tag_pattern[i].prefix, tag_pattern[i].prefix_length)) { + /* The regex starts after the prefix */ + const char *tag_regex = buf + tag_pattern[i].prefix_length; - /* save a copy of the pointers in dvec */ - size = url->dcount * sizeof(*url->dvec); + pattern->flags |= tag_pattern[i].flag; - url->dvec = (char **)malloc(size); - if (NULL == url->dvec) - { - freez(url->spec); - freez(url->path); - regfree(url->preg); - freez(url->preg); - freez(url->dbuffer); - url->dcount = 0; - return JB_ERR_MEMORY; - } - - memcpy(url->dvec, v, size); + return compile_pattern(tag_regex, NO_ANCHORING, pattern, + &pattern->pattern.tag_regex); } } - return JB_ERR_OK; + /* If it isn't a tag pattern it must be an URL pattern. */ + pattern->flags |= PATTERN_SPEC_URL_PATTERN; + + return compile_url_pattern(pattern, buf); } /********************************************************************* * - * Function : free_url_spec + * Function : free_pattern_spec * - * Description : Called from the "unloaders". Freez the url + * Description : Called from the "unloaders". Freez the pattern * structure elements. * * Parameters : - * 1 : url = pointer to a url_spec structure. + * 1 : pattern = pointer to a pattern_spec structure. * * Returns : N/A * *********************************************************************/ -void free_url_spec(struct url_spec *url) +void free_pattern_spec(struct pattern_spec *pattern) { - if (url == NULL) return; + if (pattern == NULL) return; - freez(url->spec); - freez(url->dbuffer); - freez(url->dvec); - freez(url->path); - if (url->preg) + freez(pattern->spec); +#ifdef FEATURE_PCRE_HOST_PATTERNS + if (pattern->pattern.url_spec.host_regex) { - regfree(url->preg); - freez(url->preg); + regfree(pattern->pattern.url_spec.host_regex); + freez(pattern->pattern.url_spec.host_regex); } +#endif /* def FEATURE_PCRE_HOST_PATTERNS */ + freez(pattern->pattern.url_spec.dbuffer); + freez(pattern->pattern.url_spec.dvec); + pattern->pattern.url_spec.dcount = 0; + freez(pattern->pattern.url_spec.port_list); + if (pattern->pattern.url_spec.preg) + { + regfree(pattern->pattern.url_spec.preg); + freez(pattern->pattern.url_spec.preg); + } + if (pattern->pattern.tag_regex) + { + regfree(pattern->pattern.tag_regex); + freez(pattern->pattern.tag_regex); + } +} + + +/********************************************************************* + * + * Function : port_matches + * + * Description : Compares a port against a port list. + * + * Parameters : + * 1 : port = The port to check. + * 2 : port_list = The list of port to compare with. + * + * Returns : TRUE for yes, FALSE otherwise. + * + *********************************************************************/ +static int port_matches(const int port, const char *port_list) +{ + return ((NULL == port_list) || match_portlist(port_list, port)); +} + + +/********************************************************************* + * + * Function : host_matches + * + * Description : Compares a host against a host pattern. + * + * Parameters : + * 1 : url = The URL to match + * 2 : pattern = The URL pattern + * + * Returns : TRUE for yes, FALSE otherwise. + * + *********************************************************************/ +static int host_matches(const struct http_request *http, + const struct pattern_spec *pattern) +{ + assert(http->host != NULL); +#ifdef FEATURE_PCRE_HOST_PATTERNS + if (pattern->pattern.url_spec.host_regex_type == PCRE_HOST_PATTERN) + { + return ((NULL == pattern->pattern.url_spec.host_regex) + || (0 == regexec(pattern->pattern.url_spec.host_regex, + http->host, 0, NULL, 0))); + } +#endif + return ((NULL == pattern->pattern.url_spec.dbuffer) || (0 == domain_match(pattern, http))); +} + + +/********************************************************************* + * + * Function : path_matches + * + * Description : Compares a path against a path pattern. + * + * Parameters : + * 1 : path = The path to match + * 2 : pattern = The URL pattern + * + * Returns : TRUE for yes, FALSE otherwise. + * + *********************************************************************/ +static int path_matches(const char *path, const struct pattern_spec *pattern) +{ + return ((NULL == pattern->pattern.url_spec.preg) + || (0 == regexec(pattern->pattern.url_spec.preg, path, 0, NULL, 0))); } @@ -788,17 +1342,172 @@ void free_url_spec(struct url_spec *url) * 1 : pattern = a URL pattern * 2 : url = URL to match * - * Returns : 0 iff the URL matches the pattern, else nonzero. + * Returns : Nonzero if the URL matches the pattern, else 0. + * + *********************************************************************/ +int url_match(const struct pattern_spec *pattern, + const struct http_request *http) +{ + if (!(pattern->flags & PATTERN_SPEC_URL_PATTERN)) + { + /* It's not an URL pattern and thus shouldn't be matched against URLs */ + return 0; + } + + return (port_matches(http->port, pattern->pattern.url_spec.port_list) + && host_matches(http, pattern) && path_matches(http->path, pattern)); + +} + + +/********************************************************************* + * + * Function : match_portlist + * + * Description : Check if a given number is covered by a comma + * separated list of numbers and ranges (a,b-c,d,..) + * + * Parameters : + * 1 : portlist = String with list + * 2 : port = port to check + * + * Returns : 0 => no match + * 1 => match * *********************************************************************/ -int url_match(const struct url_spec *pattern, - const struct http_request *url) +int match_portlist(const char *portlist, int port) { - return ((pattern->port == 0) || (pattern->port == url->port)) - && ((pattern->dbuffer == NULL) || (domain_match(pattern, url) == 0)) - && ((pattern->path == NULL) || - (regexec(pattern->preg, url->path, 0, NULL, 0) == 0) - ); + char *min, *max, *next, *portlist_copy; + + min = portlist_copy = strdup_or_die(portlist); + + /* + * Zero-terminate first item and remember offset for next + */ + if (NULL != (next = strchr(portlist_copy, (int) ','))) + { + *next++ = '\0'; + } + + /* + * Loop through all items, checking for match + */ + while (NULL != min) + { + if (NULL == (max = strchr(min, (int) '-'))) + { + /* + * No dash, check for equality + */ + if (port == atoi(min)) + { + freez(portlist_copy); + return(1); + } + } + else + { + /* + * This is a range, so check if between min and max, + * or, if max was omitted, between min and 65K + */ + *max++ = '\0'; + if (port >= atoi(min) && port <= (atoi(max) ? atoi(max) : 65535)) + { + freez(portlist_copy); + return(1); + } + + } + + /* + * Jump to next item + */ + min = next; + + /* + * Zero-terminate next item and remember offset for n+1 + */ + if ((NULL != next) && (NULL != (next = strchr(next, (int) ',')))) + { + *next++ = '\0'; + } + } + + freez(portlist_copy); + return 0; + +} + + +/********************************************************************* + * + * Function : parse_forwarder_address + * + * Description : Parse out the username, password, host and port from + * a forwarder address. + * + * Parameters : + * 1 : address = The forwarder address to parse. + * 2 : hostname = Used to return the hostname. NULL on error. + * 3 : port = Used to return the port. Untouched if no port + * is specified. + * 4 : username = Used to return the username if any. + * 5 : password = Used to return the password if any. + * + * Returns : JB_ERR_OK on success + * JB_ERR_MEMORY on out of memory + * JB_ERR_PARSE on malformed address. + * + *********************************************************************/ +jb_err parse_forwarder_address(char *address, char **hostname, int *port, + char **username, char **password) +{ + char *p; + char *tmp; + + tmp = *hostname = strdup_or_die(address); + + /* Parse username and password */ + if (username && password && (NULL != (p = strchr(*hostname, '@')))) + { + *p++ = '\0'; + *username = strdup_or_die(*hostname); + *hostname = strdup_or_die(p); + + if (NULL != (p = strchr(*username, ':'))) + { + *p++ = '\0'; + *password = strdup_or_die(p); + } + freez(tmp); + } + + /* Parse hostname and port */ + p = *hostname; + if ((*p == '[') && (NULL == strchr(p, ']'))) + { + /* XXX: Should do some more validity checks here. */ + return JB_ERR_PARSE; + } + + if ((**hostname == '[') && (NULL != (p = strchr(*hostname, ']')))) + { + *p++ = '\0'; + memmove(*hostname, (*hostname + 1), (size_t)(p - *hostname)); + if (*p == ':') + { + *port = (int)strtol(++p, NULL, 0); + } + } + else if (NULL != (p = strchr(*hostname, ':'))) + { + *p++ = '\0'; + *port = (int)strtol(p, NULL, 0); + } + + return JB_ERR_OK; + }