From f388d2a57ad742c872f4d7e368738660b9768292 Mon Sep 17 00:00:00 2001
From: prx <prx@ybad.name>
Date: Thu, 18 Aug 2022 10:53:08 +0200
Subject: [PATCH] improve regex to handle :1234 in url

---
 vger.c |  3 +++
 vger.h | 11 +++++++----
 2 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/vger.c b/vger.c
index ab14d5e..e4b537d 100644
--- a/vger.c
+++ b/vger.c
@@ -527,6 +527,9 @@ split_request(const char *request, char *hostname, char *path, char *query)
 	getsubexp(request, match[1], hostname);
 	getsubexp(request, match[2], path);
 	getsubexp(request, match[3], query);
+	syslog(LOG_DAEMON, "hostname:%s", hostname);
+	syslog(LOG_DAEMON, "path:%s", path);
+	syslog(LOG_DAEMON, "query:%s", query);
 
 	regfree(&greg);
 }
diff --git a/vger.h b/vger.h
index 04b0ade..3f0f926 100644
--- a/vger.h
+++ b/vger.h
@@ -18,9 +18,11 @@
  * =============
  * ^gemini://+ : in case of gemini:///
  * 1: hostname
- *	([^/|^\?]*) :
- *		catch everything, stop when / or ? is found
- *		then skip multiple /
+ *	([^/|^\?|^:]*) :
+ *		catch everything, stop when /, ? or : is found
+ *		don't catch :port
+ *	[:[0-9]*]? : skip :1234 (port number) if any
+ *	/ * : skip "/" if any
  * 2: path
  *	([^\?]*) :
  *		catch everything and stop at ? if any
@@ -28,7 +30,8 @@
  *	[\?]?(.*)$:
  *		catch everything after ? if any
  */
-static const char *_gemini_regex = "^gemini://+([^/|^\?]*)/*([^\?]*)[\?]?(.*)$";
+static const char *_gemini_regex =
+	"^gemini://+([^/|^\?|^:]*)[:[0-9]*]?/*([^\?]*)[\?]?(.*)$";
 
 /* global vars */
 static int	_retcode = 0;