Merge pull request #9 from Kazakbay/without_namespaces

Add the ability to parse xml without namespaces/key prefix
nhu313 · Nov 27, 2017 · 9683414 · 9683414
2 parents 1db1758 + 97a59d5
commit 9683414
Show file tree

Hide file tree

Showing 5 changed files with 85 additions and 20 deletions.
diff --git a/README.md b/README.md
@@ -18,6 +18,38 @@ Calling parse on the xml will produce
    value: [%{attr: [short_name: "yah"], name: :title, value: ["Yahoo"]},
            %{attr: [], name: :"title:content", value: ["Bing"]}]}]
 ```
+### Parsing without namespaces(key prefix)
+```elixir
+xml = "<m:return xsi:type="d4p1:Answer">
+      <d4p1:Title> Title </d4p1:Title>
+      <d4p1:Description> Description </d4p1:Description>
+     </m:return>"
+
+Quinn.parse(xml, %{strip_namespaces: true})
+```
+
+Calling parse on the xml will produce
+```elixir
+[%{attr: ["xsi:type": "d4p1:Answer"],
+   name: :return,
+   value: [%{attr: [], name: :title, value: ["Title"]},
+%{attr: [], name: :description, value: ["Description"]}]}]
+```
+
+### Parsing comments
+```elixir
+xml = ~s(<head><title short_name = "yah">Yahoo</title><!--- <test pattern="SECAM" /><test pattern="NTSC" /> --></head>)
+result = Quinn.parse(xml, %{comments: true})
+```
+The xml above will give you this. Note the name is `comments`.
+
+```elixir
+[%{attr: [],
+   name: :head,
+   value: [%{attr: [short_name: "yah"], name: :title, value: ["Yahoo"]},
+           %{attr: [], name: :comments, value: ~s(- <test pattern="SECAM" /><test pattern="NTSC" />)}]}]
+```
+
 # Finding nodes
 
 Suppose you want to find all the body nodes from this structure:

diff --git a/lib/quinn.ex b/lib/quinn.ex
@@ -1,7 +1,7 @@
 defmodule Quinn do
 
-  def parse(xml) do
-    Quinn.XmlParser.parse(xml)
+  def parse(xml, strip_namespaces \\ nil) do
+    Quinn.XmlParser.parse(xml, strip_namespaces)
   end
 
   def find(node, node_names) do

diff --git a/lib/xml_parser/xml_parser.ex b/lib/xml_parser/xml_parser.ex
@@ -1,10 +1,10 @@
 defmodule Quinn.XmlParser do
 
-  def parse(xml) do
+  def parse(xml, options \\ %{}) do
     :erlang.bitstring_to_list(xml)
     |> :xmerl_scan.string
     |> elem(0)
-    |> parse_record
+    |> parse_record(options)
   end
 
   defp combine_values([]), do: []
@@ -17,12 +17,13 @@ defmodule Quinn.XmlParser do
     end
   end
 
-  defp parse_record({:xmlElement, name, _, _, _, _, _, attributes, elements, _, _, _}) do
-    value = combine_values(parse_record(elements))
+  defp parse_record({:xmlElement, name, _, _, _, _, _, attributes, elements, _, _, _}, options) do
+    value = combine_values(parse_record(elements, options))
+    name = parse_name(name, options)
     [%{name: name, attr: parse_attribute(attributes), value: value}]
   end
 
-  defp parse_record({:xmlText, _, _, _, value, _}) do
+  defp parse_record({:xmlText, _, _, _, value, _}, _) do
     string_value = String.strip(to_string(value))
     if (String.length(string_value) > 0) do
       [string_value]
@@ -31,16 +32,18 @@ defmodule Quinn.XmlParser do
     end
   end
 
-  defp parse_record({:xmlComment, _, _, _, _value}) do
-    []
+  defp parse_record({:xmlComment, _, _, _, value}, options) do
+    if options[:comments] do
+      [%{name: :comments, attr: [], value: String.strip(to_string(value))}]
+    else
+      []
+    end
   end
 
-  defp parse_record([]), do: []
+  defp parse_record([], _), do: []
 
-  defp parse_record([element]), do: parse_record(element)
-
-  defp parse_record([head | tail]) do
-    parse_record(head) ++ parse_record(tail)
+  defp parse_record([head | tail], options) do
+    parse_record(head, options) ++ parse_record(tail, options)
   end
 
   defp parse_attribute([]), do: []
@@ -49,11 +52,18 @@ defmodule Quinn.XmlParser do
     [{name, to_string(value)}]
   end
 
-  defp parse_attribute([attribute]) do
-    parse_attribute(attribute)
-  end
-
   defp parse_attribute([head | tail]) do
     parse_attribute(head) ++ parse_attribute(tail)
   end
+
+  defp parse_name(name, %{strip_namespaces: true}) do
+    name
+    |> to_string
+    |> String.split(":")
+    |> List.last
+    |> Macro.underscore
+    |> String.to_atom
+  end
+
+  defp parse_name(name, _), do: name
 end
diff --git a/mix.exs b/mix.exs
@@ -4,7 +4,7 @@ defmodule Quinn.Mixfile do
   def project do
     [
       app: :quinn,
-      version: "1.0.2",
+      version: "1.1.0",
       elixir: "~> 1.3",
       deps: deps(),
       description: description(),

diff --git a/test/xml_parser/xml_parser_test.exs b/test/xml_parser/xml_parser_test.exs
@@ -52,7 +52,7 @@ defmodule Quinn.XmlParserTest do
     assert expected == Quinn.parse(xml)
   end
 
-  test "comments" do
+  test "ignore comments" do
     xml = "<head><title short_name = \"yah\">Yahoo</title><title:content>Bing</title:content><!-- foo --></head>"
     expected = [%{attr: [],
                   name: :head,
@@ -61,6 +61,16 @@ defmodule Quinn.XmlParserTest do
     assert expected == Quinn.parse(xml)
   end
 
+  test "parse comments" do
+    xml = ~s(<head><title short_name = "yah">Yahoo</title><!--- <test pattern="SECAM" /><test pattern="NTSC" /> --></head>)
+    comments = ~s(- <test pattern="SECAM" /><test pattern="NTSC" />)
+    expected = [%{attr: [],
+                  name: :head,
+                  value: [%{attr: [short_name: "yah"], name: :title, value: ["Yahoo"]},
+                          %{attr: [], name: :comments, value: comments}]}]
+    assert expected == Quinn.parse(xml, %{comments: true})
+  end
+
   test "parse small rss feed" do
     [title] = File.read!("test/xml_parser/fixture/rss_small.xml")
               |> Quinn.parse
@@ -81,4 +91,17 @@ defmodule Quinn.XmlParserTest do
                   |> Quinn.find([:entry, :title])
     assert hd(title.value) =~ "Wearing The Pants"
   end
+
+  test "parse without namespace" do
+    xml = ~s(<m:return xsi:type="d4p1:Answer"><d4p1:Title> Title </d4p1:Title><d4p1:Description> Description </d4p1:Description></m:return>)
+
+    expected = [%{attr: ["xsi:type": "d4p1:Answer"],
+                  name: :return,
+                  value: [%{attr: [], name: :title, value: ["Title"]},
+                %{attr: [],
+                  name: :description,
+                  value: ["Description"]}]}]
+
+    assert expected == Quinn.parse(xml, %{strip_namespaces: true})
+  end
 end