fix unintended (hah) bug

Author: ironholds

DESCRIPTION

@@ -1,8 +1,8 @@
 Package: urltools
 Type: Package
 Title: Vectorised Tools for URL Handling and Parsing
-Version: 1.3.1
-Date: 2015-10-07
+Version: 1.3.2
+Date: 2015-10-09
 Author: Oliver Keyes [aut, cre], Jay Jacobs [aut, cre], Mark Greenaway [ctb], 
         Bob Rudis [ctb]
 Maintainer: Oliver Keyes <[email protected]>

NEWS

@@ -1,4 +1,11 @@
+Version 1.3.2
+-------------------------------------------------------------------------
+
+BUG FIXES
+* Fixed a critical bug impacting URLs with colons in the path
+
 Version 1.3.1
+-------------------------------------------------------------------------
 
 CHANGES
 * suffix_refresh has been removed, since LazyData's parameters prevented it from functioning; thanks to

src/parsing.cpp

@@ -33,7 +33,7 @@ std::vector < std::string > parsing::domain_and_port(std::string& url){
   // url goes into the holding string.
   std::size_t port = url.find(":");
 
-  if(port != std::string::npos){
+  if(port != std::string::npos && url.find("/") >= port){
     output[0] = url.substr(0,port);
     holding = url.substr(port+1);
     output_offset++;
@@ -47,6 +47,7 @@ std::vector < std::string > parsing::domain_and_port(std::string& url){
   // If there is one, that's when everything ends
   if(trailing_slash != std::string::npos){
     output[output_offset] = holding.substr(0, trailing_slash);
+    output_offset++;
     url = holding.substr(trailing_slash+1);
     return output;
   }

tests/testthat/test_parsing.R

@@ -51,4 +51,10 @@ test_that("Port handling works", {
   expect_that(port(url), equals("4000"))
   expect_that(path(url), equals(""))
   expect_that(parameters(url), equals("foo=bar"))
+})
+
+test_that("Port handling does not break path handling", {
+  url <- "https://en.wikipedia.org/wiki/File:Vice_City_Public_Radio_(logo).jpg"
+  expect_that(port(url), equals(""))
+  expect_that(path(url), equals("wiki/file:vice_city_public_radio_(logo).jpg"))
 })