From d4807bb55065c40c6eb8f7e95b8dc03a00ffae16 Mon Sep 17 00:00:00 2001
From: Jim Hester <james.f.hester@gmail.com>
Date: Tue, 31 May 2016 10:28:30 -0400
Subject: [PATCH] Support aliased namespace prefixes

Fixes #35
---
 src/xml2_utils.h                       | 42 +++++++++-----------------
 tests/testthat/ns-multiple-aliases.xml |  4 +++
 tests/testthat/test-namespaces.R       |  9 ++++--
 3 files changed, 25 insertions(+), 30 deletions(-)
 create mode 100644 tests/testthat/ns-multiple-aliases.xml

diff --git a/src/xml2_utils.h b/src/xml2_utils.h
index 29cf955c..760079dd 100644
--- a/src/xml2_utils.h
+++ b/src/xml2_utils.h
@@ -59,10 +59,9 @@ class Xml2String {
 class NsMap {
 
   // We only store the index to avoid duplicating the data
-  typedef boost::unordered_map<std::string, std::string> url2prefix_t;
+  typedef std::multimap<std::string, std::string> prefix2url_t;
 
-  url2prefix_t url2prefix;
-  std::vector<std::string> order;
+  prefix2url_t prefix2url;
 
   public:
   NsMap() {
@@ -76,28 +75,28 @@ class NsMap {
     }
   }
 
-  bool hasUrl(std::string url) {
-    return url2prefix.find(url) != url2prefix.end();
+  bool hasPrefix(const std::string& prefix) {
+    return prefix2url.find(prefix) != prefix2url.end();
   }
 
-  std::string findPrefix(std::string url) {
-    url2prefix_t::const_iterator it = url2prefix.find(url);
-    if (it != url2prefix.end()) {
+  std::string findUrl(const std::string& prefix) {
+    prefix2url_t::const_iterator it = prefix2url.find(prefix);
+    if (it != prefix2url.end()) {
       return it->second;
     }
 
-    Rcpp::stop("Couldn't find prefix for url %s", url);
+    Rcpp::stop("Couldn't find url for prefix %s", prefix);
     return std::string();
   }
 
-  std::string findUrl(std::string prefix) {
-    for (url2prefix_t::const_iterator it = url2prefix.begin(); it != url2prefix.end(); ++it) {
-      if (it->second == prefix) {
+  std::string findPrefix(const std::string& url) {
+    for (prefix2url_t::const_iterator it = prefix2url.begin(); it != prefix2url.end(); ++it) {
+      if (it->second == url) {
         return it->first;
       }
     }
 
-    Rcpp::stop("Couldn't find url for prefix %s", prefix);
+    Rcpp::stop("Couldn't find prefix for url %s", url);
     return std::string();
   }
 
@@ -106,25 +105,12 @@ class NsMap {
   }
 
   bool add(std::string prefix, std::string url) {
-    if (hasUrl(url))
-      return false;
-
-    // Add the value to the vector and store the order.
-    order.push_back(url);
-    url2prefix.insert(url2prefix_t::value_type(url, prefix));
+    prefix2url.insert(prefix2url_t::value_type(prefix, url));
     return true;
   }
 
   Rcpp::CharacterVector out() {
-    Rcpp::CharacterVector out = Rcpp::wrap(order);
-    std::vector<std::string> name;
-    name.reserve(order.size());
-
-    for(std::vector<std::string>::const_iterator it = order.begin();it != order.end(); ++it) {
-      name.push_back(url2prefix.at(*it));
-    }
-    out.attr("names") = Rcpp::wrap(name);
-    return out;
+    return Rcpp::wrap(prefix2url);
   }
 };
 #endif
diff --git a/tests/testthat/ns-multiple-aliases.xml b/tests/testthat/ns-multiple-aliases.xml
new file mode 100644
index 00000000..bc14cf9b
--- /dev/null
+++ b/tests/testthat/ns-multiple-aliases.xml
@@ -0,0 +1,4 @@
+<root>
+  <doc1 xmlns:b="http://bar.com"><b:bar /></doc1>
+  <doc2 xmlns:c="http://bar.com"><c:bar /></doc2>
+</root>
diff --git a/tests/testthat/test-namespaces.R b/tests/testthat/test-namespaces.R
index 44904b80..24bb5fec 100644
--- a/tests/testthat/test-namespaces.R
+++ b/tests/testthat/test-namespaces.R
@@ -12,6 +12,11 @@ test_that("repeated prefixes given unique names", {
   expect_equal(ns, c(b = "http://baz.com", b1 = "http://bar.com"))
 })
 
+test_that("aliased prefixes retained", {
+  ns <- unclass(xml_ns(read_xml("ns-multiple-aliases.xml")))
+  expect_equal(ns, c(b = "http://bar.com", c = "http://bar.com"))
+})
+
 
 # Low-level character vector tests ---------------------------------------------
 
@@ -20,9 +25,9 @@ test_that("unique prefix-url combo unchanged", {
   expect_equal(unique_ns(x), x)
 })
 
-test_that("first prefix kept", {
+test_that("all prefixs kept", {
   x <- c(blah = "http://blah.com", rah = "http://blah.com")
-  expect_equal(names(unique_ns(x)), "blah")
+  expect_equal(names(unique_ns(x)), c("blah", "rah"))
 })
 
 test_that("multiple default namespaces can be stripped", {