From a051b36ae73266e8d2a91bb8e86fcdad6ad01677 Mon Sep 17 00:00:00 2001
From: John Mark Bell <jmb@netsurf-browser.org>
Date: Sat, 23 Apr 2005 22:26:05 +0000
Subject: [project @ 2005-04-23 22:26:05 by jmb] Make url_join match the spec -
 see http://www.ics.uci.edu/~fielding/url/test1.html for testcases.

svn path=/import/netsurf/; revision=1682
---
 utils/url.c | 33 +++++++++++++++++++++++++++++++--
 1 file changed, 31 insertions(+), 2 deletions(-)

diff --git a/utils/url.c b/utils/url.c
index 006b71f1e..1c587b86b 100644
--- a/utils/url.c
+++ b/utils/url.c
@@ -236,6 +236,7 @@ url_func_result url_join(const char *rel, const char *base, char **result)
 	path = base + base_match[5].rm_so;
 	path_len = base_match[5].rm_eo - base_match[5].rm_so;
 
+
 	/* 1) */
 	m = regexec(&url_re, rel, 10, rel_match, 0);
 	if (m) {
@@ -245,7 +246,6 @@ url_func_result url_join(const char *rel, const char *base, char **result)
 
 	/* 2) */
 	/* base + "#s" = (current document)#s (see Appendix C.1) */
-	/** \todo does (current document) include the query? */
 	if (rel_match[9].rm_so != -1) {
 		fragment = rel + rel_match[9].rm_so;
 		fragment_len = rel_match[9].rm_eo - rel_match[9].rm_so;
@@ -254,6 +254,11 @@ url_func_result url_join(const char *rel, const char *base, char **result)
 			rel_match[2].rm_so == -1 &&
 			rel_match[4].rm_so == -1 &&
 			rel_match[6].rm_so == -1) {
+		if (base_match[7].rm_so != -1) {
+			query = base + base_match[7].rm_so;
+			query_len = base_match[7].rm_eo -
+						base_match[7].rm_so;
+		}
 		goto step7;
 	}
 	if (rel_match[7].rm_so != -1) {
@@ -261,6 +266,14 @@ url_func_result url_join(const char *rel, const char *base, char **result)
 		query_len = rel_match[7].rm_eo - rel_match[7].rm_so;
 	}
 
+	/* base + "?y" = (base - query)?y
+	 * e.g http://a/b/c/d;p?q + ?y = http://a/b/c/d;p?y */
+	if (rel_match[5].rm_so == rel_match[5].rm_eo &&
+			rel_match[2].rm_so == -1 &&
+			rel_match[4].rm_so == -1 &&
+			rel_match[6].rm_so != -1)
+		goto step7;
+
 	/* 3) */
 	if (rel_match[2].rm_so != -1) {
 		scheme = rel + rel_match[2].rm_so;
@@ -334,8 +347,24 @@ url_func_result url_join(const char *rel, const char *base, char **result)
 		} else
 			path_len -= up_match[1].rm_eo - up_match[1].rm_so + 3;
 	}
+
+	/* and strip any remaining ../ | ./ pairs */
+	for (path = buf; path - buf < path_len; ) {
+		if (*path == '.' && path[1] == '.' && path[2] == '/') {
+			memmove(buf + (path - buf), path + 3,
+				((buf + path_len) - path) - 3);
+			path_len -= 3;
+		}
+		else if (*path == '.' && path[1] == '/') {
+			memmove(buf + (path - buf), path + 2,
+				((buf + path_len) - path) - 2);
+			path_len -= 2;
+		}
+		else
+			path++;
+	}
 	buf[path_len] = 0;
-        path = buf;
+	path = buf;
 
 step7:	/* 7) */
 	(*result) = malloc(scheme_len + 1 + 2 + authority_len + path_len + 1 + 1 +
-- 
cgit v1.2.3