From 2ca1648ee260877c2a619ee45449cbf5b4aaa3a1 Mon Sep 17 00:00:00 2001
From: Romain de Laage <romain.delaage@rdelaage.ovh>
Date: Tue, 16 Feb 2021 19:38:50 +0100
Subject: [PATCH] Join url implemented

---
 lib/url.c | 167 +++++++++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 159 insertions(+), 8 deletions(-)

diff --git a/lib/url.c b/lib/url.c
index 5952c25..30055e7 100644
--- a/lib/url.c
+++ b/lib/url.c
@@ -16,6 +16,7 @@ Url *urllib_init (void);
 void urllib_free (Url *url);
 int urllib_parse (Url *url, const char *url_string);
 void urllib_tostring (Url *url, char *dest);
+int urllib_join (Url *url, const char *path);
 
 int
 main (int    argc,
@@ -26,11 +27,19 @@ main (int    argc,
 
 	url = urllib_init ();
 
-	if (argc >= 2)
+	if (argc == 2)
 	{
 		if (urllib_parse (url, argv[1]))
 			return 1;
 	}
+	if (argc >= 3)
+	{
+		if (urllib_parse (url, argv[1]))
+			return 1;
+
+		if (urllib_join (url, argv[2]))
+			return 1;
+	}
 	else
 	{
 		printf ("Enter a url : ");
@@ -38,6 +47,12 @@ main (int    argc,
 
 		if (urllib_parse (url, url_str))
 			return 1;
+
+		printf ("Enter a path : ");
+		scanf ("%s", url_str);
+
+		if (urllib_join (url, url_str))
+			return 1;
 	}
 
 	urllib_tostring (url, url_str);
@@ -74,7 +89,6 @@ urllib_free (Url *url)
 	free (url);
 }
 
-/* check this parser : http://draft.scyphus.co.jp/lang/c/url_parser.html not perfect but could give ideas */
 int
 urllib_parse (Url        *url,
               const char *url_string)
@@ -189,12 +203,11 @@ urllib_parse (Url        *url,
 	}
 
 	return 0;
-
-	//is there a path ? (test \0)
 }
 
 void
-urllib_tostring (Url *url, char *dest)
+urllib_tostring (Url  *url,
+                 char *dest)
 {
 	if (url->port == NULL)
 		if (url->path == NULL)
@@ -208,6 +221,144 @@ urllib_tostring (Url *url, char *dest)
 			sprintf (dest, "%s://%s:%s%s\r\n", url->scheme, url->host, url->port, url->path);
 }
 
-void
-urllib_join (Url *url, const char *path)
-{}
+int
+urllib_join (Url        *url,
+             const char *path)
+{// view RFC 3986, section 5.2.4
+	int len = strlen (path);
+	char *base, *result, *inputcursor;
+
+	if (len == 0)//nothing to merge
+		return 0;
+
+	if (url->path == NULL)//no current path, path = '/' 
+	{
+		url->path = malloc (sizeof (char) * 2);
+		if (url->path == NULL)
+			return 1;
+
+		url->path[0] = '/';
+		url->path[1] = '\0';
+	}
+
+	if (path[0] != '/') //relative path
+	{
+		int url_path_len = strlen (path);
+
+		if (url->path[url_path_len - 1] != '/')//we are on a file
+		{
+			char *tmpcursor = strrchr (url->path, '/');
+			if (tmpcursor == NULL)
+				return -1;
+
+			tmpcursor[1] = '\0';
+		}
+
+		len += url_path_len;
+		base = malloc (sizeof (char) * (len + 1));
+		if (base == NULL)
+			return 1;
+		result = malloc (sizeof (char) * (len + 1));
+		if (result == NULL)
+		{
+			free (base);
+			return 1;
+		}
+	
+		sprintf (base, "%s%s", url->path, path);
+		result[0] = '\0';
+	}
+	else //absolute path
+	{
+		base = malloc (sizeof (char) * (len + 1));
+		if (base == NULL)
+			return 1;
+		result = malloc (sizeof (char) * (len + 1));
+		if (result == NULL)
+		{
+			free (base);
+			return 1;
+		}
+
+		strncpy (base, path, len);
+		base[len] = '\0';
+		result[0] = '\0';
+	}
+
+	inputcursor = base;
+	len = strlen (inputcursor);
+	while (len > 0) //while input buffer is not empty
+	{
+		char *tmpcursor;
+
+		if (len >= 3 && inputcursor[0] == '.' && inputcursor[1] == '.' && inputcursor[2] == '/')
+		{
+			inputcursor += 3;
+			len = strlen (inputcursor);
+		}
+		else if (len >= 2 && inputcursor[0] == '.' && inputcursor[1] == '/')
+		{
+			inputcursor += 2;
+			len = strlen (inputcursor);
+		}
+		else if (len >= 3 && inputcursor[0] == '/' && inputcursor[1] == '.' && inputcursor[2] == '/')
+		{
+			inputcursor += 2;
+			len = strlen (inputcursor);
+		}
+		else if (len == 2 && inputcursor[0] == '/' && inputcursor[1] == '.')
+		{
+			inputcursor += 1;
+			inputcursor[0] = '/';
+			len = strlen (inputcursor);
+		}
+		else if (len >= 4 && inputcursor[0] == '/' && inputcursor[1] == '.' && inputcursor[2] == '.' && inputcursor[3] == '/')
+		{
+			inputcursor += 3;
+			len = strlen (inputcursor);
+			//remove result last segment
+			tmpcursor = strrchr (result, '/');
+			if (tmpcursor != NULL)
+				*tmpcursor = '\0';
+		}
+		else if (len == 3 && inputcursor[0] == '/' && inputcursor[1] == '.' && inputcursor[2] == '.')
+		{
+			inputcursor += 2;
+			inputcursor[0] = '/';
+			len = strlen (inputcursor);
+			//remove result last segment
+			tmpcursor = strrchr (result, '/');
+			if (tmpcursor != NULL)
+				*tmpcursor = '\0';
+		}
+		else if ((len == 2 && inputcursor[0] == '.' && inputcursor[1] == '.') || (len == 1 && inputcursor[0] == '.'))
+		{
+			inputcursor[0] = '\0';
+			len = strlen (inputcursor);
+		}
+		else
+		{
+			//get first segment of inputcursor then append it to result and delete it
+			tmpcursor = strchr (inputcursor + 1, '/');
+			if (tmpcursor == NULL)//last segment in input buffer
+			{
+				sprintf (result, "%s%s", result, inputcursor);
+				*inputcursor = '\0';
+			}
+			else
+			{
+				len = tmpcursor - inputcursor;
+				strncpy (result + strlen (result), inputcursor, len);
+				inputcursor = tmpcursor;
+			}
+
+			len = strlen (inputcursor);
+		}
+	}
+
+	free (url->path);
+	url->path = result;
+	free (base);
+
+	return 0;
+}