-const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.51 2009/05/16 13:27:20 fabiankeil Exp $";
/*********************************************************************
*
* File : $Source: /cvsroot/ijbswa/current/urlmatch.c,v $
* Purpose : Declares functions to match URLs against URL
* patterns.
*
- * Copyright : Written by and Copyright (C) 2001-2009
- * the 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
#include <assert.h>
#include <string.h>
-#if !defined(_WIN32) && !defined(__OS2__)
+#if !defined(_WIN32)
#include <unistd.h>
#endif
#include "miscutil.h"
#include "errlog.h"
-const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
-
-enum regex_anchoring {NO_ANCHORING, LEFT_ANCHORED, RIGHT_ANCHORED};
-static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern);
+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
/*********************************************************************
*
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);
* 1 : http = pointer to the http structure to hold elements.
*
* Returns : JB_ERR_OK on success
- * JB_ERR_MEMORY on out of memory
* JB_ERR_PARSE on malformed command/URL
* or >100 domains deep.
*
size_t size;
char *p;
- http->dbuffer = strdup(http->host);
- if (NULL == http->dbuffer)
- {
- return JB_ERR_MEMORY;
- }
+ http->dbuffer = strdup_or_die(http->host);
/* map to lower case */
for (p = http->dbuffer; *p ; p++)
{
- *p = (char)tolower((int)(unsigned char)*p);
+ *p = (char)privoxy_tolower(*p);
}
/* split the domain name into components */
- http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec), 1, 1);
+ http->dcount = ssplit(http->dbuffer, ".", vec, SZ(vec));
if (http->dcount <= 0)
{
/* save a copy of the pointers in dvec */
size = (size_t)http->dcount * sizeof(*http->dvec);
- http->dvec = (char **)malloc(size);
- if (NULL == http->dvec)
- {
- return JB_ERR_MEMORY;
- }
+ http->dvec = malloc_or_die(size);
memcpy(http->dvec, vec, size);
}
+/*********************************************************************
+ *
+ * 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
* protocol are acceptable.
*
* Returns : JB_ERR_OK on success
- * JB_ERR_MEMORY on out of memory
* JB_ERR_PARSE on malformed command/URL
* or >100 domains deep.
*
/*
* Save our initial URL
*/
- http->url = strdup(url);
- if (http->url == NULL)
- {
- return JB_ERR_MEMORY;
- }
-
+ http->url = strdup_or_die(url);
/*
* Check for * URI. If found, we're done.
- */
+ */
if (*http->url == '*')
{
- if ( NULL == (http->path = strdup("*"))
- || NULL == (http->hostport = strdup("")) )
- {
- return JB_ERR_MEMORY;
- }
+ http->path = strdup_or_die("*");
+ http->hostport = strdup_or_die("");
if (http->url[1] != '\0')
{
return JB_ERR_PARSE;
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;
/*
* 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
{
* 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);
}
freez(buf);
-
- if ( (http->path == NULL)
- || (http->hostport == NULL))
- {
- return JB_ERR_MEMORY;
- }
}
if (!host_available)
char *host;
char *port;
- buf = strdup(http->hostport);
- if (buf == NULL)
- {
- return JB_ERR_MEMORY;
- }
+ buf = strdup_or_die(http->hostport);
/* check if url contains username and/or password */
host = strchr(buf, '@');
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
{
http->port = (http->ssl ? 443 : 80);
}
- http->host = strdup(host);
+ http->host = strdup_or_die(host);
freez(buf);
-
- if (http->host == NULL)
- {
- return JB_ERR_MEMORY;
- }
}
- /*
- * Split domain name so we can compare it against wildcards
- */
+ /* Split domain name so we can compare it against wildcards */
return init_domain_components(http);
}
*********************************************************************/
static int unknown_method(const char *method)
{
- static const char *known_http_methods[] = {
+ static const char * const known_http_methods[] = {
/* Basic HTTP request type */
"GET", "HEAD", "POST", "PUT", "DELETE", "OPTIONS", "TRACE", "CONNECT",
/* webDAV extensions (RFC2518) */
* 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
*/
"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;
}
+/*********************************************************************
+ *
+ * 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;
+
+ 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;
+
+}
+
+
/*********************************************************************
*
* Function : parse_http_request
* 2 : http = pointer to the http structure to hold elements
*
* 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)
{
char *buf;
- char *v[10]; /* XXX: Why 10? We should only need three. */
+ char *v[3];
int n;
jb_err err;
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)
{
freez(buf);
return JB_ERR_PARSE;
}
- if (strcmpic(v[2], "HTTP/1.1") && strcmpic(v[2], "HTTP/1.0"))
+ if (JB_ERR_OK != normalize_http_version(v[2]))
{
- log_error(LOG_LEVEL_ERROR, "The only supported HTTP "
- "versions are 1.0 and 1.1. This rules out: %s", v[2]);
freez(buf);
return JB_ERR_PARSE;
}
/*
* Copy the details into the structure
*/
- 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);
freez(buf);
- if ( (http->cmd == NULL)
- || (http->gpc == NULL)
- || (http->ver == NULL) )
- {
- return JB_ERR_MEMORY;
- }
-
return JB_ERR_OK;
}
*
* Parameters :
* 1 : pattern = The pattern to compile.
- * 2 : anchoring = How the regex should be anchored.
- * Can be either one of NO_ANCHORING,
- * LEFT_ANCHORED or RIGHT_ANCHORED.
+ * 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 : JB_ERR_OK - Success
- * JB_ERR_MEMORY - Out of memory
* JB_ERR_PARSE - Cannot parse regex
*
*********************************************************************/
static jb_err compile_pattern(const char *pattern, enum regex_anchoring anchoring,
- struct url_spec *url, regex_t **regex)
+ struct pattern_spec *url, regex_t **regex)
{
int errcode;
- char rebuf[BUFFER_SIZE];
const char *fmt = NULL;
+ char *rebuf;
+ size_t rebuf_size;
assert(pattern);
- assert(strlen(pattern) < sizeof(rebuf) - 2);
if (pattern[0] == '\0')
{
case RIGHT_ANCHORED:
fmt = "%s$";
break;
+ case RIGHT_ANCHORED_HOST:
+ fmt = "%s\\.?$";
+ break;
case LEFT_ANCHORED:
fmt = "^%s";
break;
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));
- *regex = zalloc(sizeof(**regex));
- if (NULL == *regex)
- {
- free_url_spec(url);
- return JB_ERR_MEMORY;
- }
-
- snprintf(rebuf, sizeof(rebuf), fmt, pattern);
+ 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, sizeof(rebuf));
- if (errlen > (sizeof(rebuf) - (size_t)1))
+ size_t errlen = regerror(errcode, *regex, rebuf, rebuf_size);
+ if (errlen > (rebuf_size - (size_t)1))
{
- errlen = sizeof(rebuf) - (size_t)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_url_spec(url);
+ free_pattern_spec(url);
+ freez(rebuf);
return JB_ERR_PARSE;
}
+ freez(rebuf);
return JB_ERR_OK;
* Description : Compiles the three parts of an URL pattern.
*
* Parameters :
- * 1 : url = Target url_spec to be filled in.
+ * 1 : url = Target pattern_spec to be filled in.
* 2 : buf = The url pattern to compile. Will be messed up.
*
* Returns : JB_ERR_OK - Success
* JB_ERR_PARSE - Cannot parse regex
*
*********************************************************************/
-static jb_err compile_url_pattern(struct url_spec *url, char *buf)
+static jb_err compile_url_pattern(struct pattern_spec *url, char *buf)
{
char *p;
+#ifdef FEATURE_PCRE_HOST_PATTERNS
+ const size_t prefix_length = 18;
+ if (strncmpic(buf, "PCRE-HOST-PATTERN:", prefix_length) == 0)
+ {
+ 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
+
p = strchr(buf, '/');
if (NULL != p)
{
/*
* XXX: does it make sense to compile the slash at the beginning?
*/
- jb_err err = compile_pattern(p, LEFT_ANCHORED, url, &url->preg);
+ jb_err err = compile_pattern(p, LEFT_ANCHORED, url, &url->pattern.url_spec.preg);
if (JB_ERR_OK != err)
{
if (NULL != p)
{
*p++ = '\0';
- url->port_list = strdup(p);
- if (NULL == url->port_list)
- {
- return JB_ERR_MEMORY;
- }
+ url->pattern.url_spec.port_list = strdup_or_die(p);
}
else
{
- url->port_list = NULL;
+ url->pattern.url_spec.port_list = NULL;
}
if (buf[0] != '\0')
{
- return compile_host_pattern(url, buf);
+#ifdef FEATURE_PCRE_HOST_PATTERNS
+ if (url->pattern.url_spec.host_regex_type == PCRE_HOST_PATTERN)
+ {
+ return compile_pcre_host_pattern(url, buf);
+ }
+ else
+#endif
+ {
+ return compile_vanilla_host_pattern(url, buf);
+ }
}
return JB_ERR_OK;
}
-#ifdef FEATURE_EXTENDED_HOST_PATTERNS
+#ifdef FEATURE_PCRE_HOST_PATTERNS
/*********************************************************************
*
- * Function : compile_host_pattern
+ * Function : compile_pcre_host_pattern
*
- * Description : Parses and compiles a host pattern..
+ * Description : Parses and compiles a pcre host pattern.
*
* Parameters :
- * 1 : url = Target url_spec to be filled in.
+ * 1 : url = Target pattern_spec to be filled in.
* 2 : host_pattern = Host pattern to compile.
*
* Returns : JB_ERR_OK - Success
* JB_ERR_PARSE - Cannot parse regex
*
*********************************************************************/
-static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
+static jb_err compile_pcre_host_pattern(struct pattern_spec *url, const char *host_pattern)
{
- return compile_pattern(host_pattern, RIGHT_ANCHORED, url, &url->host_regex);
+ return compile_pattern(host_pattern, RIGHT_ANCHORED_HOST, url, &url->pattern.url_spec.host_regex);
}
+#endif /* def FEATURE_PCRE_HOST_PATTERNS */
-#else
/*********************************************************************
*
- * Function : compile_host_pattern
+ * Function : compile_vanilla_host_pattern
*
* Description : Parses and "compiles" an old-school host pattern.
*
* Parameters :
- * 1 : url = Target url_spec to be filled in.
+ * 1 : url = Target pattern_spec to be filled in.
* 2 : host_pattern = Host pattern to parse.
*
* Returns : JB_ERR_OK - Success
- * JB_ERR_MEMORY - Out of memory
* JB_ERR_PARSE - Cannot parse regex
*
*********************************************************************/
-static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
+static jb_err compile_vanilla_host_pattern(struct pattern_spec *url, const char *host_pattern)
{
char *v[150];
size_t size;
*/
if (host_pattern[strlen(host_pattern) - 1] == '.')
{
- url->unanchored |= ANCHOR_RIGHT;
+ url->pattern.url_spec.unanchored |= ANCHOR_RIGHT;
}
if (host_pattern[0] == '.')
{
- url->unanchored |= ANCHOR_LEFT;
+ url->pattern.url_spec.unanchored |= ANCHOR_LEFT;
}
- /*
+ /*
* Split domain into components
*/
- url->dbuffer = strdup(host_pattern);
- if (NULL == url->dbuffer)
- {
- free_url_spec(url);
- return JB_ERR_MEMORY;
- }
+ url->pattern.url_spec.dbuffer = strdup_or_die(host_pattern);
- /*
+ /*
* Map to lower case
*/
- for (p = url->dbuffer; *p ; p++)
+ for (p = url->pattern.url_spec.dbuffer; *p ; p++)
{
- *p = (char)tolower((int)(unsigned char)*p);
+ *p = (char)privoxy_tolower(*p);
}
- /*
+ /*
* Split the domain name into components
*/
- url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
+ url->pattern.url_spec.dcount = ssplit(url->pattern.url_spec.dbuffer, ".", v, SZ(v));
- if (url->dcount < 0)
+ if (url->pattern.url_spec.dcount < 0)
{
- free_url_spec(url);
- return JB_ERR_MEMORY;
+ free_pattern_spec(url);
+ return JB_ERR_PARSE;
}
- else if (url->dcount != 0)
+ else if (url->pattern.url_spec.dcount != 0)
{
- /*
+ /*
* Save a copy of the pointers in dvec
*/
- size = (size_t)url->dcount * sizeof(*url->dvec);
-
- url->dvec = (char **)malloc(size);
- if (NULL == url->dvec)
- {
- free_url_spec(url);
- return JB_ERR_MEMORY;
- }
+ size = (size_t)url->pattern.url_spec.dcount * sizeof(*url->pattern.url_spec.dvec);
- memcpy(url->dvec, v, size);
+ url->pattern.url_spec.dvec = malloc_or_die(size);
+
+ memcpy(url->pattern.url_spec.dvec, v, size);
}
/*
* else dcount == 0 in which case we needn't do anything,
{
const unsigned char *pat = (const unsigned char *)pattern;
const unsigned char *txt = (const unsigned char *)text;
- const unsigned char *fallback = pat;
+ const unsigned char *fallback = pat;
int wildcard = 0;
-
+
unsigned char lastchar = 'a';
unsigned i;
unsigned char charmap[32];
-
+
while (*txt)
{
}
/* '*' in the pattern? */
- if (*pat == '*')
+ if (*pat == '*')
{
-
+
/* The pattern ends afterwards? Speed up the return. */
if (*++pat == '\0')
{
return 0;
}
-
+
/* Else, set wildcard mode and remember position after '*' */
wildcard = 1;
fallback = pat;
while (*++pat != ']')
{
if (!*pat)
- {
+ {
return 1;
}
else if (*pat == '-')
for (i = lastchar; i <= *pat; i++)
{
charmap[i / 8] |= (unsigned char)(1 << (i % 8));
- }
+ }
}
else
{
} /* -END- if Character range specification */
- /*
- * Char match, or char range match?
+ /*
+ * Char match, or char range match?
*/
- if ( (*pat == *txt)
- || (*pat == '?')
- || ((*pat == ']') && (charmap[*txt / 8] & (1 << (*txt % 8)))) )
+ if ((*pat == *txt)
+ || (*pat == '?')
+ || ((*pat == ']') && (charmap[*txt / 8] & (1 << (*txt % 8)))))
{
- /*
- * Sucess: Go ahead
+ /*
+ * Success: Go ahead
*/
pat++;
}
else if (!wildcard)
{
- /*
+ /*
* No match && no wildcard: No luck
*/
return 1;
}
/* Cut off extra '*'s */
- if(*pat == '*') pat++;
+ if (*pat == '*') pat++;
/* If this is the pattern's end, fine! */
return(*pat);
* Function : domain_match
*
* Description : Domain-wise Compare fqdn's. Governed by the bimap in
- * pattern->unachored, the comparison is un-, left-,
+ * p.pattern->unachored, the comparison is un-, left-,
* right-anchored, or both.
* The individual domain names are compared with
* simplematch().
*
* Parameters :
- * 1 : pattern = a domain that may contain a '*' as a wildcard.
+ * 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 url_spec *pattern, const struct http_request *fqdn)
+static int domain_match(const struct pattern_spec *p, const struct http_request *fqdn)
{
char **pv, **fv; /* vectors */
int plen, flen;
- int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
+ int unanchored = p->pattern.url_spec.unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
- plen = pattern->dcount;
+ plen = p->pattern.url_spec.dcount;
flen = fqdn->dcount;
if (flen < plen)
return 1;
}
- pv = pattern->dvec;
+ pv = p->pattern.url_spec.dvec;
fv = fqdn->dvec;
if (unanchored == ANCHOR_LEFT)
}
}
-#endif /* def FEATURE_EXTENDED_HOST_PATTERNS */
/*********************************************************************
*
- * Function : create_url_spec
+ * Function : create_pattern_spec
*
- * Description : Creates a "url_spec" structure from a string.
- * When finished, free with free_url_spec().
+ * Description : Creates a "pattern_spec" structure from a string.
+ * When finished, free with free_pattern_spec().
*
* Parameters :
- * 1 : url = Target url_spec to be filled in. Will be
- * zeroed before use.
+ * 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 url->spec. If this
+ * 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_MEMORY - Out of memory
* JB_ERR_PARSE - Cannot parse regex (Detailed message
* written to system log)
*
*********************************************************************/
-jb_err create_url_spec(struct url_spec *url, char *buf)
+jb_err create_pattern_spec(struct pattern_spec *pattern, char *buf)
{
- assert(url);
+ 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);
- memset(url, '\0', sizeof(*url));
+ memset(pattern, '\0', sizeof(*pattern));
/* Remember the original specification for the CGI pages. */
- url->spec = strdup(buf);
- if (NULL == url->spec)
- {
- return JB_ERR_MEMORY;
- }
+ pattern->spec = strdup_or_die(buf);
- /* Is it tag pattern? */
- if (0 == strncmpic("TAG:", url->spec, 4))
+ /* Check if it's a tag pattern */
+ for (i = 0; i < SZ(tag_pattern); i++)
{
- /* The pattern starts with the first character after "TAG:" */
- const char *tag_pattern = buf + 4;
- return compile_pattern(tag_pattern, NO_ANCHORING, url, &url->tag_regex);
+ 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;
+
+ pattern->flags |= tag_pattern[i].flag;
+
+ return compile_pattern(tag_regex, NO_ANCHORING, pattern,
+ &pattern->pattern.tag_regex);
+ }
}
- /* If it isn't a tag pattern it must be a URL pattern. */
- return compile_url_pattern(url, buf);
+ /* 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);
-#ifdef FEATURE_EXTENDED_HOST_PATTERNS
- if (url->host_regex)
+ freez(pattern->spec);
+#ifdef FEATURE_PCRE_HOST_PATTERNS
+ if (pattern->pattern.url_spec.host_regex)
{
- regfree(url->host_regex);
- freez(url->host_regex);
+ regfree(pattern->pattern.url_spec.host_regex);
+ freez(pattern->pattern.url_spec.host_regex);
}
-#else
- freez(url->dbuffer);
- freez(url->dvec);
- url->dcount = 0;
-#endif /* ndef FEATURE_EXTENDED_HOST_PATTERNS */
- freez(url->port_list);
- if (url->preg)
+#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(url->preg);
- freez(url->preg);
+ regfree(pattern->pattern.url_spec.preg);
+ freez(pattern->pattern.url_spec.preg);
}
- if (url->tag_regex)
+ 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)
{
- regfree(url->tag_regex);
- freez(url->tag_regex);
+ 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)));
}
* Returns : Nonzero if the URL matches the pattern, else 0.
*
*********************************************************************/
-int url_match(const struct url_spec *pattern,
+int url_match(const struct pattern_spec *pattern,
const struct http_request *http)
{
- /* XXX: these should probably be functions. */
-#define PORT_MATCHES ((NULL == pattern->port_list) || match_portlist(pattern->port_list, http->port))
-#ifdef FEATURE_EXTENDED_HOST_PATTERNS
-#define DOMAIN_MATCHES ((NULL == pattern->host_regex) || (0 == regexec(pattern->host_regex, http->host, 0, NULL, 0)))
-#else
-#define DOMAIN_MATCHES ((NULL == pattern->dbuffer) || (0 == domain_match(pattern, http)))
-#endif
-#define PATH_MATCHES ((NULL == pattern->preg) || (0 == regexec(pattern->preg, http->path, 0, NULL, 0)))
-
- if (pattern->tag_regex != NULL)
+ if (!(pattern->flags & PATTERN_SPEC_URL_PATTERN))
{
- /* It's a tag pattern and shouldn't be matched against URLs */
+ /* It's not an URL pattern and thus shouldn't be matched against URLs */
return 0;
- }
+ }
- return (PORT_MATCHES && DOMAIN_MATCHES && PATH_MATCHES);
+ return (port_matches(http->port, pattern->pattern.url_spec.port_list)
+ && host_matches(http, pattern) && path_matches(http->path, pattern));
}
{
char *min, *max, *next, *portlist_copy;
- min = portlist_copy = strdup(portlist);
+ min = portlist_copy = strdup_or_die(portlist);
/*
* Zero-terminate first item and remember offset for next
/*
* Loop through all items, checking for match
*/
- while(min)
+ while (NULL != min)
{
if (NULL == (max = strchr(min, (int) '-')))
{
* or, if max was omitted, between min and 65K
*/
*max++ = '\0';
- if(port >= atoi(min) && port <= (atoi(max) ? atoi(max) : 65535))
+ if (port >= atoi(min) && port <= (atoi(max) ? atoi(max) : 65535))
{
freez(portlist_copy);
return(1);
*
* Function : parse_forwarder_address
*
- * Description : Parse out the host and port from a 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)
+jb_err parse_forwarder_address(char *address, char **hostname, int *port,
+ char **username, char **password)
{
- char *p = address;
+ char *p;
+ char *tmp;
- if ((*address == '[') && (NULL == strchr(address, ']')))
+ tmp = *hostname = strdup_or_die(address);
+
+ /* Parse username and password */
+ if (username && password && (NULL != (p = strchr(*hostname, '@'))))
{
- /* XXX: Should do some more validity checks here. */
- return JB_ERR_PARSE;
+ *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);
}
- *hostname = strdup(address);
- if (NULL == *hostname)
+ /* Parse hostname and port */
+ p = *hostname;
+ if ((*p == '[') && (NULL == strchr(p, ']')))
{
- return JB_ERR_MEMORY;
+ /* XXX: Should do some more validity checks here. */
+ return JB_ERR_PARSE;
}
if ((**hostname == '[') && (NULL != (p = strchr(*hostname, ']'))))