added urs.prerewrite()

qse/cmd/http/httpd.c

@@ -195,7 +195,7 @@ struct loccfg_t
 		int dns_retries;
 		int urs_timeout;
 		int urs_retries;
-		const qse_mchar_t pseudonym[128]; /* TODO: good size? */
+		qse_mchar_t pseudonym[128]; /* TODO: good size? */
 	} proxy;
 
 	loccfg_t* next;
@@ -224,6 +224,7 @@ struct server_xtn_t
 	qse_httpd_serverstd_freersrc_t orgfreersrc;
 	qse_httpd_serverstd_query_t orgquery;
 
+	
 	qse_htb_t* cfgtab;
 };
 
@@ -234,9 +235,10 @@ struct httpd_xtn_t
 {
 	const  qse_char_t* cfgfile;
 	qse_xli_t* xli;
-	qse_httpd_impede_t orgimpede;
+	qse_httpd_impede_t org_impede;
 	int impede_code;
 
+	qse_httpd_urs_prerewrite_t org_urs_prerewrite;
 	loccfg_t dflcfg;
 };
 
@@ -538,6 +540,10 @@ static int get_server_root (
 			if (loccfg->proxy.pseudonym[0]) 
 				root->u.proxy.pseudonym = loccfg->proxy.pseudonym;
 
+			/* move the host name part backward by 1 byte to make a room for
+			 * terminating null. An orginal input of http://www.yahoo.com/ab/cd
+			 * becomes http:/www.yahoo.com\0ab/cd. host gets to point to the 
+			 * www.yahoo.com. qpath(qinfo->req.u.q.path) is updated to ab/cd. */
 			qse_memmove (host - 1, host, slash - host); 
 			slash[-1] = QSE_MT('\0');
 			host = host - 1;
@@ -589,7 +595,6 @@ static int get_server_root (
 	root->u.path.rpl = loccfg->locname.len;
 	return 0;
 
-
 proxy_ok:
 	if (loccfg->proxy.dns_enabled)
 	{
@@ -2125,15 +2130,26 @@ static void impede_httpd (qse_httpd_t* httpd)
 	}
 
 	/* chain-call the orignal impedence function */
-	if (httpd_xtn->orgimpede) httpd_xtn->orgimpede (httpd);
+	if (httpd_xtn->org_impede) httpd_xtn->org_impede (httpd);
 }
 
+static int prerewrite_url (qse_httpd_t* httpd, qse_httpd_client_t* client, qse_htre_t* req, const qse_mchar_t* host, qse_mchar_t** url)
+{
+	httpd_xtn_t* httpd_xtn;
+
+	httpd_xtn = qse_httpd_getxtnstd (httpd);
+
+/* TODO: override prerewrite url */
+	return httpd_xtn->org_urs_prerewrite (httpd, client, req, host, url);
+}
+
+
 static void logact_httpd (qse_httpd_t* httpd, const qse_httpd_act_t* act)
 {
-	httpd_xtn_t* httpd_xtn;
+	/*httpd_xtn_t* httpd_xtn;*/
 	qse_char_t tmp[128], tmp2[128], tmp3[128];
 
-	httpd_xtn = qse_httpd_getxtnstd (httpd);
+	/*httpd_xtn = qse_httpd_getxtnstd (httpd);*/
 
 	switch (act->code)
 	{
@@ -2302,6 +2318,7 @@ static int httpd_main (int argc, qse_char_t* argv[])
 	int trait, ret;
 	qse_httpd_rcb_t rcb;
 	qse_httpd_ecb_t ecb;
+	qse_httpd_scb_t scb;
 
 	ret = handle_args (argc, argv);
 	if (ret <= -1) return -1;
@@ -2346,8 +2363,13 @@ static int httpd_main (int argc, qse_char_t* argv[])
 	tmout.nsec = 0;
 	qse_httpd_setopt (httpd, QSE_HTTPD_IDLELIMIT, &tmout);
 
+	qse_httpd_getopt (httpd, QSE_HTTPD_SCB, &scb);
+	httpd_xtn->org_urs_prerewrite = scb.urs.prerewrite;
+	scb.urs.prerewrite = prerewrite_url;
+	qse_httpd_setopt (httpd, QSE_HTTPD_SCB, &scb);
+
 	qse_httpd_getopt (httpd, QSE_HTTPD_RCB, &rcb);
-	httpd_xtn->orgimpede = rcb.impede;
+	httpd_xtn->org_impede = rcb.impede;
 	rcb.impede = impede_httpd; /* executed when qse_httpd_impede() is called */
 	if (g_debug) rcb.logact = logact_httpd; /* i don't remember old logging handler */
 	qse_httpd_setopt (httpd, QSE_HTTPD_RCB, &rcb);

qse/include/qse/http/httpd.h

@@ -145,11 +145,59 @@ typedef void (*qse_httpd_resol_t) (
 
 typedef void (*qse_httpd_rewrite_t) (
 	qse_httpd_t*       httpd,
+	/* optional original URL. if the original URL is not 
+	 * avaialble,it can be set to #QSE_NULL */
 	const qse_mchar_t* url,     
+	/* rewritten URL */
 	const qse_mchar_t* new_url,
+	/* content data pointer */
 	void*              ctx
 );
 
+
+
+
+typedef int (*qse_httpd_urs_open_t) (
+	qse_httpd_t*     httpd, 
+	qse_httpd_urs_t* urs
+);
+
+typedef void (*qse_httpd_urs_close_t) (
+	qse_httpd_t*     httpd,
+	qse_httpd_urs_t* urs
+);
+
+typedef int (*qse_httpd_urs_recv_t) (
+	qse_httpd_t*     httpd,
+	qse_httpd_urs_t* urs,
+	qse_ubi_t        handle
+);
+
+typedef int (*qse_httpd_urs_send_t) (
+	qse_httpd_t*            httpd,
+	qse_httpd_urs_t*        urs, 
+	const qse_mchar_t*      url,
+	qse_httpd_rewrite_t     rewrite,
+	const qse_httpd_natr_t* urs_server,
+	void*                   ctx
+);
+
+/* on success, url must point to a null-teminated string which 
+ * can be freed with qse_httpd_freemem() when not needed anymore.
+ * the return value of 0 indicates that the string is the final 
+ * rewriting result and no sending to urs-server is required.
+ * if the return value is greater than 0, the string sent to the
+ * urs-server for the actual rewriting. a negative return value
+ * indicates failure. 
+ */
+typedef int (*qse_httpd_urs_prerewrite_t) (
+	qse_httpd_t*        httpd, 
+	qse_httpd_client_t* client,
+	qse_htre_t*         req,
+	const qse_mchar_t*  host,
+	qse_mchar_t**       url
+);
+
 typedef struct qse_httpd_scb_t qse_httpd_scb_t;
 struct qse_httpd_scb_t
 {
@@ -278,12 +326,11 @@ struct qse_httpd_scb_t
 
 	struct
 	{
-		int (*open) (qse_httpd_t* httpd, qse_httpd_urs_t* urs);
+		qse_httpd_urs_open_t open;
-		void (*close) (qse_httpd_t* httpd, qse_httpd_urs_t* urs);
+		qse_httpd_urs_close_t close;
-		int (*recv) (qse_httpd_t* httpd, qse_httpd_urs_t* urs, qse_ubi_t handle);
+		qse_httpd_urs_recv_t recv;
-		int (*send) (qse_httpd_t* httpd, qse_httpd_urs_t* urs, 
+		qse_httpd_urs_send_t send;
-		             const qse_mchar_t* url, qse_httpd_rewrite_t rewrite,
+		qse_httpd_urs_prerewrite_t prerewrite;
-		             const qse_httpd_natr_t* urs_server, void* ctx);
 	} urs;
 };
 
@@ -598,7 +645,6 @@ struct qse_httpd_rsrc_proxy_t
 {
 	int flags; /* bitwise-ORed of qse_httpd_rsrc_proxy_flag_t enumerators */
 
-	const qse_mchar_t* host; /* host name part that were in the original request url */
 	union
 	{
 		qse_nwad_t nwad; 
@@ -606,13 +652,22 @@ struct qse_httpd_rsrc_proxy_t
 	union
 	{
 		qse_nwad_t nwad;
+
+		/* turn QSE_HTTPD_RSRC_PROXY_DST_STR on in flags and set the
+		 * destination host name to the str field. if the bit is not set
+		 * nwad field is used. */
 		const qse_mchar_t* str; 
 	} dst; /* remote destination address to connect to */
 
 	qse_httpd_natr_t dns_server;
 	qse_httpd_natr_t urs_server;
 
-	const qse_mchar_t* pseudonym; /* pseudonym to use in Via: */
+	/* optional pseudonym to use for Via: */
+	const qse_mchar_t* pseudonym;
+
+	/* optional host name. it's preferred over the Host header in some 
+	 * contexts. */
+	const qse_mchar_t* host;
 };
 
 typedef struct qse_httpd_rsrc_dir_t qse_httpd_rsrc_dir_t;

qse/lib/awk/std.c

@@ -277,6 +277,9 @@ static void* custom_awk_modopen (qse_awk_t* awk, const qse_awk_mod_spec_t* spec)
 
 #elif defined(__DOS__)
 
+	/* the DOS code here is not generic enough. it's for a specific
+	 * dos-extender only. the best is to enable QSE_ENABLE_STATIC_MODULE
+	 * when building for DOS. */
 	void* h;
 	qse_mchar_t* modpath;
 	const qse_char_t* tmp[4];

qse/lib/http/httpd-proxy.c

@@ -48,11 +48,12 @@ struct task_proxy_t
 #define PROXY_PEER_NAME_RESOLVED   (1 << 7)
 #define PROXY_PEER_NAME_UNRESOLVED (1 << 8)
 #define PROXY_REWRITE_URL          (1 << 9)
-#define PROXY_URL_REWRITTEN        (1 << 10)
+#define PROXY_URL_PREREWRITTEN     (1 << 10) /* URL has been prerewritten in prerewrite(). */
-#define PROXY_URL_REDIRECTED       (1 << 11)
+#define PROXY_URL_REWRITTEN        (1 << 11)
-#define PROXY_X_FORWARDED_FOR      (1 << 12) /* X-Forwarded-For added */
+#define PROXY_URL_REDIRECTED       (1 << 12)
-#define PROXY_VIA                  (1 << 13) /* Via added to the request */
+#define PROXY_X_FORWARDED_FOR      (1 << 13) /* X-Forwarded-For added */
-#define PROXY_VIA_RETURNING        (1 << 14) /* Via added to the response */
+#define PROXY_VIA                  (1 << 14) /* Via added to the request */
+#define PROXY_VIA_RETURNING        (1 << 15) /* Via added to the response */
 	int flags;
 	qse_httpd_t* httpd;
 	qse_httpd_client_t* client;
@@ -84,7 +85,6 @@ struct task_proxy_t
 	int          reqflags;
 	qse_htre_t*  req; /* set to original client request associated with this if necessary */
 	qse_mbs_t*   reqfwdbuf; /* content from the request */
-	const qse_mchar_t* host;
 
 	qse_mbs_t*   res;
 	qse_size_t   res_consumed;
@@ -894,12 +894,11 @@ static void adjust_peer_name_and_port (task_proxy_t* proxy)
 	colon = qse_mbschr (proxy->peer_name, QSE_MT(':'));
 	if (colon) 
 	{
-		qse_mchar_t* endptr;
 		/* handle a port number after the colon sign */
-
 		*colon = QSE_MT('\0');
-		QSE_MBSTONUM (proxy->peer_port, colon + 1, &endptr, 10);
+		proxy->peer_port = qse_mbstoui (colon + 1, 10);
-		/* TODO: check if *endptr is QSE_T('\0')? */
+		/* TODO: check if there is a garbage after the port number.
+		 *       check if the port number has overflown */
 	}
 	else
 	{
@@ -948,7 +947,6 @@ static int task_init_proxy (
 	{
 		if (arg->rsrc->flags & QSE_HTTPD_RSRC_PROXY_ENABLE_DNS)
 		{
-/* TODO: get pseudonym from parameter... */
 			proxy->peer_name = proxy->pseudonym + len + 1;
 
 			qse_mbscpy (proxy->peer_name, arg->rsrc->dst.str);
@@ -977,50 +975,18 @@ static int task_init_proxy (
 
 	if (arg->rsrc->flags & QSE_HTTPD_RSRC_PROXY_ENABLE_URS)
 	{
-		const qse_mchar_t* qpath;
+		int x = httpd->opt.scb.urs.prerewrite (httpd, client, arg->req, arg->rsrc->host, &proxy->url_to_rewrite);
-		const qse_mchar_t* metnam;
+		if (x <= -1) goto oops;
-		const qse_mchar_t* host_ptr = QSE_NULL;
-		qse_mchar_t cliaddrbuf[128];
-		qse_size_t total_len;
-
-		qpath = qse_htre_getqpath(arg->req);
-		metnam = qse_httpmethodtombs(proxy->method);
-
-		total_len = qse_mbslen(qpath) + qse_mbslen(metnam);
-
-		if (arg->rsrc->host)
-		{
-			host_ptr = arg->rsrc->host;
-			total_len += qse_mbslen(host_ptr);
-		}
-		else 
-		{
-			const qse_htre_hdrval_t* hosthv;
-			hosthv = qse_htre_getheaderval(arg->req, QSE_MT("Host"));
-			if (hosthv)
-			{
-				/* the first host header only */
-				host_ptr = hosthv->ptr;
-				total_len += hosthv->len;
-			}
-		}
-
-		total_len += qse_nwadtombs (&client->remote_addr, cliaddrbuf, QSE_COUNTOF(cliaddrbuf), QSE_NWADTOMBS_ADDR);
-
-		total_len += 128; /* extra space */
-
-		proxy->url_to_rewrite = qse_httpd_allocmem (httpd, total_len);
-		if (proxy->url_to_rewrite == QSE_NULL) goto oops;
-
-		/* URL client-ip/client-fqdn ident method  */
-		if (proxy->method != QSE_HTTP_CONNECT && host_ptr)
-			qse_mbsxfmt (proxy->url_to_rewrite, total_len, QSE_MT("http://%s%s %s/- - %s"), host_ptr, qpath, cliaddrbuf, metnam);
-		else
-			qse_mbsxfmt (proxy->url_to_rewrite, total_len, QSE_MT("%s %s/- - %s"), qpath, cliaddrbuf, metnam);
 
 printf (">>>>>>>>>>>>>>>>>>>>>>>> [%s] <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<\n", proxy->url_to_rewrite);
 		/* enable url rewriting */
 		proxy->flags |= PROXY_REWRITE_URL;
+		if (x == 0) 
+		{
+			/* prerewrite() indicates that proxy->url_to_rewrite is the final
+			 * rewriting result and no futher rewriting is required */
+			proxy->flags |= PROXY_URL_PREREWRITTEN;
+		}
 
 		if (arg->rsrc->flags & QSE_HTTPD_RSRC_PROXY_URS_SERVER)
 		{
@@ -1042,7 +1008,6 @@ printf (">>>>>>>>>>>>>>>>>>>>>>>> [%s] <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<\n", proxy
 	proxy->reqfwdbuf = qse_mbs_open (httpd->mmgr, 0, (len < 512? 512: len));
 	if (proxy->reqfwdbuf == QSE_NULL) goto nomem_oops;
 
-	proxy->host = arg->rsrc->host;
 	if (proxy->flags & PROXY_RAW)
 	{
 /* TODO: when connect is attempted, no keep-alive must be hornored. 
@@ -2090,6 +2055,15 @@ static int task_main_proxy (
 	if (proxy->flags & PROXY_URL_REDIRECTED) return 0; /* URL redirected. task finished */ 
 
 	if (proxy->flags & PROXY_REWRITE_URL)
+	{
+		if (proxy->flags & PROXY_URL_PREREWRITTEN)
+		{
+			/* proxy->url_to_rewrite is the final rewritten URL by prerewrite(). 
+			 * pass QSE_NULL as the second parameter to on_url_rewritten() to
+			 * indicate that the original URL is not known */
+			on_url_rewritten (httpd, QSE_NULL, proxy->url_to_rewrite, task);
+		}
+		else
 		{
 			/* note that url_to_rewrite is URL + extra information. */
 			if (qse_httpd_rewriteurl (httpd, proxy->url_to_rewrite, on_url_rewritten, 
@@ -2099,6 +2073,7 @@ static int task_main_proxy (
 			
 			if ((proxy->flags & PROXY_REWRITE_URL) && 
 			    qse_httpd_inactivatetasktrigger (httpd, client, task) <= -1) goto oops;
+		}
 
 		return 1;
 	}

qse/lib/http/httpd-std-urs.h

@@ -476,3 +476,55 @@ oops:
 	return -1;
 }
 
+static int urs_prerewrite (qse_httpd_t* httpd, qse_httpd_client_t* client, qse_htre_t* req, const qse_mchar_t* host, qse_mchar_t** url)
+{
+	const qse_mchar_t* qpath;
+	int mtype;
+	const qse_mchar_t* mname;
+
+	const qse_mchar_t* host_ptr = QSE_NULL;
+	qse_mchar_t cliaddrbuf[128];
+	qse_size_t total_len;
+	qse_mchar_t* url_to_rewrite;
+
+	qpath = qse_htre_getqpath(req);
+	mtype = qse_htre_getqmethodtype(req);
+	mname = qse_htre_getqmethodname(req);
+
+	total_len = qse_mbslen(qpath) + qse_mbslen(mname);
+
+	if (host)
+	{
+		/* use the host name set explicitly by the caller */
+		host_ptr = host;
+		total_len += qse_mbslen(host_ptr);
+	}
+	else 
+	{
+		/* find the host name in the http header */
+		const qse_htre_hdrval_t* hosthv;
+		hosthv = qse_htre_getheaderval(req, QSE_MT("Host"));
+		if (hosthv)
+		{
+			/* the first host header only */
+			host_ptr = hosthv->ptr;
+			total_len += hosthv->len;
+		}
+	}
+
+	total_len += qse_nwadtombs (&client->remote_addr, cliaddrbuf, QSE_COUNTOF(cliaddrbuf), QSE_NWADTOMBS_ADDR);
+
+	total_len += 128; /* extra space */
+
+	url_to_rewrite = qse_httpd_allocmem (httpd, total_len);
+	if (url_to_rewrite == QSE_NULL) return -1;
+
+	/* URL client-ip/client-fqdn ident method  */
+	if (mtype != QSE_HTTP_CONNECT && host_ptr)
+		qse_mbsxfmt (url_to_rewrite, total_len, QSE_MT("http://%s%s %s/- - %s"), host_ptr, qpath, cliaddrbuf, mname);
+	else
+		qse_mbsxfmt (url_to_rewrite, total_len, QSE_MT("%s %s/- - %s"), qpath, cliaddrbuf, mname);
+
+	*url = url_to_rewrite;
+	return 1;
+}

qse/lib/http/httpd-std.c

@@ -2317,6 +2317,7 @@ static qse_size_t hash_string (const qse_mchar_t *str)
 
 #include "httpd-std-dns.h"
 #include "httpd-std-urs.h"
+
 /* ------------------------------------------------------------------- */
 
 static qse_httpd_scb_t httpd_system_callbacks =
@@ -2324,13 +2325,15 @@ static qse_httpd_scb_t httpd_system_callbacks =
 	/* server */
 	{ server_open,
 	  server_close,
-	  server_accept },
+	  server_accept
+	},
 
 	{ peer_open,
 	  peer_close,
 	  peer_connected,
 	  peer_recv,
-	  peer_send },
+	  peer_send
+	},
 
 	/* multiplexer */
 	{ mux_open,
@@ -2369,19 +2372,23 @@ static qse_httpd_scb_t httpd_system_callbacks =
 	  client_send,
 	  client_sendfile,
 	  client_accepted,
-	  client_closed },
+	  client_closed
+	},
 
 	/* dns */
 	{ dns_open,
 	  dns_close,
 	  dns_recv,
-	  dns_send },
+	  dns_send
+	},
 
 	/* urs */
 	{ urs_open,
 	  urs_close,
 	  urs_recv,
-	  urs_send }
+	  urs_send,
+	  urs_prerewrite
+	}
 };
 
 static qse_httpd_rcb_t httpd_request_callbacks =
@@ -3282,3 +3289,4 @@ int qse_httpd_loopstd (qse_httpd_t* httpd, const qse_httpd_dnsstd_t* dns, const
 	/* main loop */
 	return qse_httpd_loop (httpd);
 }
+