Support https; deal with http servers closing immediately after accepting

5 files changed, 82 insertions, 23 deletions

diff --git a/src/lac.h b/src/lac.h
index 407b548..dc4f84e 100644
--- a/src/lac.h
+++ b/src/lac.h
@@ -355,9 +355,9 @@ LacUri * lac_uri_new_from_string (const LacUri *base,
                                   const gchar  *str);
 LacUri * lac_uri_copy            (const LacUri *uri);
 gchar  * lac_uri_string          (const LacUri *uri);
-void     lac_uri_free            (LacUri       *uri);
 gboolean lac_uri_equal           (const LacUri *uri1,
                                   const LacUri *uri2);
+void     lac_uri_free            (LacUri       *uri);
 
 /*
  * HTTP
diff --git a/src/lachttp.c b/src/lachttp.c
index 9e63556..343f4ad 100644
--- a/src/lachttp.c
+++ b/src/lachttp.c
@@ -73,7 +73,7 @@ struct _HttpHeader
     gchar *value;
 };
 
-#define junk_detected() g_print (G_STRLOC ": WARNING: junk detected after response")
+#define junk_detected() g_print (G_STRLOC ": WARNING: junk detected after response\n")
 
 GQuark
 lac_http_error_quark (void)
@@ -207,7 +207,8 @@ struct _HttpHost {
     
     GQueue *            unsent;
     gboolean            optimistic;             /* whether newly created transports are pipelines */
-
+    gboolean            use_tls;
+    
     /* handling broken servers */
     gboolean            broken_server;          /* if TRUE, _never_ try pipelining */
     gboolean            recovering;             /* if TRUE, don't create new connections. */
@@ -217,7 +218,8 @@ struct _HttpHost {
 };
 
 static HttpHost *http_host_new                            (const LacAddress *addr,
-                                                           gint              port);
+                                                           gint              port,
+                                                           gboolean          use_tls);
 static void      http_host_add_request                    (HttpHost         *host,
                                                            LacHttpRequest   *request);
 static void      http_host_cancel_request                 (HttpHost         *host,
@@ -1568,7 +1570,7 @@ address_callback (const LacAddress *new_addr,
             HttpHost *http_host;
             
             http_host = http_host_new (
-                new_addr, request->uri->u.http.port);
+                new_addr, request->uri->u.http.port, request->uri->u.http.is_https);
             
             request_emit_host_found (request, request->uri->u.http.host, new_addr);
             
@@ -1676,7 +1678,9 @@ host_dump_spam (HttpHost *host)
 #endif
 
 static HttpHost *
-http_host_new (const LacAddress *addr, gint port)
+http_host_new (const LacAddress *addr,
+               gint              port,
+               gboolean          use_tls)
 { 
     HttpHost *host;
     GList *list;
@@ -1684,8 +1688,12 @@ http_host_new (const LacAddress *addr, gint port)
     for (list = all_hosts; list != NULL; list = list->next)
     {
         host = list->data;
-        if (lac_address_equal (host->address, addr) && host->port == port)
+        if (lac_address_equal (host->address, addr)     &&
+            host->port == port                          &&
+            host->use_tls == use_tls)
+        {
             return host;
+        }
     }
     
     host = g_new0 (HttpHost, 1);
@@ -1697,6 +1705,7 @@ http_host_new (const LacAddress *addr, gint port)
     host->unsent = g_queue_new ();
     host->recovering = FALSE;
     host->timeout_id = 0;
+    host->use_tls = use_tls;
     
     if (USE_PIPELINING)
     {
@@ -2395,6 +2404,25 @@ http_transport_handle_close (HttpTransport *transport)
         lac_http_request_unref (transport->current);
         transport->current = NULL;
     }
+    else if (transport->successful_requests == 0)
+    {
+        /* If there is no current request, and we haven't succesfully
+         * processed any requests at all, then that means the server
+         * closed the connection before we even sent it anything.
+         *
+         * In that case, we should not return outstanding requests,
+         * as we would just spin forever. Instead we just fail
+         * all of them. Further conversation with this server is not
+         * going to be productive.
+         */
+        GError *err = g_error_new (
+            LAC_HTTP_ERROR, LAC_HTTP_ERROR_PREMATURE_CLOSE,
+            "Server closed connection before sending a complete response");
+        
+        http_transport_all_requests_failed (transport, err);
+
+        g_error_free (err);
+    }
     
     if (transport->pipelining && !g_queue_is_empty (transport->in_progress))
     {
@@ -2404,7 +2432,8 @@ http_transport_handle_close (HttpTransport *transport)
          * connection.
          *
 	 * This is known to happen with Netscape/Enterprise 3.6 SP.
-         * We should perhaps special case that particular server
+         * We should perhaps special case that particular server.
+         * news.google.com also does this.
          */
 
         g_print ("server indicated it supported pipelining, but it didn't\n");
@@ -2559,8 +2588,17 @@ http_transport_new            (HttpHost       *host,
 #endif
     
     transport->host = host;
-    transport->connection = lac_connection_new_tcp (
-        host->address, host->port, connection_callback, transport);
+
+    if (host->use_tls)
+    {
+        transport->connection = lac_connection_new_tls (
+            host->address, host->port, connection_callback, transport);
+    }
+    else
+    {
+        transport->connection = lac_connection_new_tcp (
+            host->address, host->port, connection_callback, transport);
+    }
     
     transport->unsent = g_queue_new ();
     transport->in_progress = g_queue_new ();
diff --git a/src/lacuri.c b/src/lacuri.c
index 7107815..ad01147 100644
--- a/src/lacuri.c
+++ b/src/lacuri.c
@@ -331,6 +331,7 @@ lac_uri_path_to_ftp_type (const gchar *path, LacFtpType *type)
 }
 
 enum {
+    LAC_HTTPS_DEFAULT_PORT = 443,
     LAC_HTTP_DEFAULT_PORT = 80,
     LAC_FTP_DEFAULT_PORT = 21
 };
@@ -350,7 +351,12 @@ lac_uri_unknown_to_http (LacUri *uri,
     lac_uri_authority_to_port_and_host (uri->u.unknown.authority,
 					&host, &port);      
     if (port == -1)
-	port = LAC_HTTP_DEFAULT_PORT;
+    {
+        if (is_https)
+            port = LAC_HTTPS_DEFAULT_PORT;
+        else
+            port = LAC_HTTP_DEFAULT_PORT;
+    }
     g_free (uri->u.unknown.authority);
     
     if (uri->u.unknown.path == NULL)
@@ -895,7 +901,7 @@ lac_uri_copy (const LacUri *uri)
 	copy->u.http.port = uri->u.http.port;
 	copy->u.http.path = g_strdup (uri->u.http.path);
 	copy->u.http.query = g_strdup (uri->u.http.query);
-        copy->u.http.is_https = uri->u.http.port;
+        copy->u.http.is_https = uri->u.http.is_https;
 	break;
     case LAC_SCHEME_FTP:
 	copy->u.ftp.host = g_strdup (uri->u.ftp.host);
@@ -931,22 +937,38 @@ lac_uri_string (const LacUri *uri)
     {
         gchar *port_str;
         gchar *query_str;
+        gchar *scheme_str;
+
+        if (uri->u.http.is_https)
+        {
+            if (uri->u.http.port != LAC_HTTPS_DEFAULT_PORT)
+                port_str = g_strdup_printf (":%d", uri->u.http.port);
+            else
+                port_str = g_strdup ("");
 
-        if (uri->u.http.port != LAC_HTTP_DEFAULT_PORT)
-            port_str = g_strdup_printf (":%d", uri->u.http.port);
+            scheme_str = g_strdup ("https://");
+        }
         else
-            port_str = g_strdup ("");
+        {
+            if (uri->u.http.port != LAC_HTTP_DEFAULT_PORT)
+                port_str = g_strdup_printf (":%d", uri->u.http.port);
+            else
+                port_str = g_strdup ("");
+
+            scheme_str = g_strdup ("http://");
+        }
 
         if (uri->u.http.query)
             query_str = g_strdup_printf ("?%s", uri->u.http.query);
         else
             query_str = g_strdup ("");
 
-        str = g_strconcat (uri->u.http.is_https? "https://" : "http://",
+        str = g_strconcat (scheme_str,
                            uri->u.http.host, port_str,
                            uri->u.http.path, query_str,
                            NULL);
 
+        g_free (scheme_str);
         g_free (port_str);
         g_free (query_str);
         break;
diff --git a/tests/lacwget.c b/tests/lacwget.c
index c7a182d..f992613 100644
--- a/tests/lacwget.c
+++ b/tests/lacwget.c
@@ -622,10 +622,10 @@ do_download (LacUri *base_uri, const gchar *uri_str, gboolean try_parse)
     if (already_downloaded (uri))
         return FALSE;
 
-#if 0
-    printf ("download: %s\n", uri_str);
-#endif
+    printf ("generated uri: %s\n", lac_uri_string (uri));
     
+    printf ("download: %s\n", uri_str);
+
     request = lac_http_request_new_get (
         uri, http_callback, page_info);
     
diff --git a/tests/uri-test.c b/tests/uri-test.c
index c040098..839aeab 100644
--- a/tests/uri-test.c
+++ b/tests/uri-test.c
@@ -41,11 +41,10 @@ print_uri (LacUri *uri, gchar *base_uri_str, gchar *uri_str)
 	g_print (" fragment:   %s\n", uri->fragment);
 	break;
     case LAC_SCHEME_HTTP:
-    case LAC_SCHEME_HTTPS:
-        if (uri->scheme == LAC_SCHEME_HTTP)
-            g_print ("HTTP\n");
-        else
+        if (uri->u.http.is_https)
             g_print ("HTTPS\n");
+        else
+            g_print ("HTTP\n");
 	g_print (" host:	  %s\n", uri->u.http.host);
 	g_print (" port:       %d\n", uri->u.http.port);
 	g_print (" path:       %s\n", uri->u.http.path);