SweetXml

SweetXml is a thin wrapper around :xmerl. It allows you to convert a char_list or xmlElement record as defined in :xmerl to an elixir value such as map, list, string, integer, float or any combination of these.

Examples

Given a xml document such as below

<?xml version="1.05" encoding="UTF-8"?>
<game>
  <matchups>
    <matchup winner-id="1">
      <name>Match One</name>
      <teams>
        <team>
          <id>1</id>
          <name>Team One</name>
        </team>
        <team>
          <id>2</id>
          <name>Team Two</name>
        </team>
      </teams>
    </matchup>
    <matchup winner-id="2">
      <name>Match Two</name>
      <teams>
        <team>
          <id>2</id>
          <name>Team Two</name>
        </team>
        <team>
          <id>3</id>
          <name>Team Three</name>
        </team>
      </teams>
    </matchup>
    <matchup winner-id="1">
      <name>Match Three</name>
      <teams>
        <team>
          <id>1</id>
          <name>Team One</name>
        </team>
        <team>
          <id>3</id>
          <name>Team Three</name>
        </team>
      </teams>
    </matchup>
  </matchups>
</game>

We can do the following

import SweetXml
doc = "..." # as above

get the name of the first match

result = doc |> xpath(~x"//matchup/name/text()") # `sigil_x` for (x)path
assert result == 'Match One'

get the xml record of the name of the first match

result = doc |> xpath(~x"//matchup/name"e) # `e` is the modifier for (e)ntity
assert result == {:xmlElement, :name, :name, [], {:xmlNamespace, [], []},
        [matchup: 2, matchups: 2, game: 1], 2, [],
        [{:xmlText, [name: 2, matchup: 2, matchups: 2, game: 1], 1, [],
          'Match One', :text}], [],
        ...}

get the full list of matchup name

result = doc |> xpath(~x"//matchup/name/text()"l) # `l` stands for (l)ist
assert result == ['Match One', 'Match Two', 'Match Three']

get a list of winner-id by attributes

result = doc |> xpath(~x"//matchup/@winner-id"l)
assert result == ['1', '2', '1']

get a list of matchups with different map structure

result = doc |> xpath(
  ~x"//matchups/matchup"l,
  name: ~x"./name/text()",
  winner: [
    ~x".//team/id[.=ancestor::matchup/@winner-id]/..",
    name: ~x"./name/text()"
  ]
)
assert result == [
  %{name: 'Match One', winner: %{name: 'Team One'}},
  %{name: 'Match Two', winner: %{name: 'Team Two'}},
  %{name: 'Match Three', winner: %{name: 'Team One'}}
]

Or directly return a mapping of your liking

result = doc |> xmap(
  matchups: [
    ~x"//matchups/matchup"l,
    name: ~x"./name/text()",
    winner: [
      ~x".//team/id[.=ancestor::matchup/@winner-id]/..",
      name: ~x"./name/text()"
    ]
  ],
  last_matchup: [
    ~x"//matchups/matchup[last()]",
    name: ~x"./name/text()",
    winner: [
      ~x".//team/id[.=ancestor::matchup/@winner-id]/..",
      name: ~x"./name/text()"
    ]
  ]
)
assert result == %{
  matchups: [
    %{name: 'Match One', winner: %{name: 'Team One'}},
    %{name: 'Match Two', winner: %{name: 'Team Two'}},
    %{name: 'Match Three', winner: %{name: 'Team One'}}
  ],
  last_matchup: %{name: 'Match Three', winner: %{name: 'Team One'}}
}

The ~x Sigil

In the above examples, we used the expression ~x"//some/path" to define the path. The reason is it allows us to more precisely specify what is being returned.

• ~x"//some/path"

  without any modifiers, xpath/2 will return the value of the entity if the entity is of type xmlText, xmlAttribute, xmlPI, xmlComment as defined in :xmerl

• ~x"//some/path"e

  e stands for (e)ntity. This forces xpath/2 to return the entity with which you can further chain your xpath/2 call

• ~x"//some/path"l

  'l' stands for (l)ist. This forces xpath/2 to return a list. Without l, xpath/2 will only return the first element of the match

• ~x"//some/path"k

  'k' stands for (k)eyword. This forces xpath/2 to return a Keyword instead of a Map.

• ~x"//some/path"el - mix of the above

• ~x"//some/path"s

  's' stands for (s)tring. This forces xpath/2 to return the value as string instead of a char list.

• ~x"//some/path"S

  'S' stands for soft (S)tring. This forces xpath/2 to return the value as string instead of a char list, but if node content is incompatible with a string, set "".

• x"//some/path"o

  'o' stands for (o)ptional. This allows the path to not exist, and will return nil.

• ~x"//some/path"sl - string list.

• ~x"//some/path"i

  'i' stands for (i)nteger. This forces xpath/2 to return the value as integer instead of a char list.

• ~x//some/path"I

  'I' stands for soft (I)nteger. This forces xpath/2 to return the value as integer instead of a char list, but if node content is incompatible with an integer, set 0.

• ~x"//some/path"f

  'f' stands for (f)loat. This forces xpath/2 to return the value as float instead of a char list.

• ~x//some/path"F

  'F' stands for soft (F)loat. This forces xpath/2 to return the value as float instead of a char list, but if node content is incompatible with a float, set 0.0.

• ~x"//some/path"il - integer list.

If you use the optional modifier o together with a soft cast modifier (uppercase), then the value is set to nil when the value is not compatible for instance ~x//some/path/text()"Fo return nil if the text is not a number.

Also in the examples section, we always import SweetXml first. This makes x_sigil available in the current scope. Without it, instead of using ~x, you can use the %SweetXpath struct

assert ~x"//some/path"e == %SweetXpath{path: '//some/path', is_value: false, is_list: false, cast_to: false}

Note the use of char_list in the path definition.

Namespace support

Given a xml document such as below

<?xml version="1.05" encoding="UTF-8"?>
<game xmlns="http://example.com/fantasy-league" xmlns:ns1="http://example.com/baseball-stats">
  <matchups>
    <matchup winner-id="1">
      <name>Match One</name>
      <teams>
        <team>
          <id>1</id>
          <name>Team One</name>
          <ns1:runs>5</ns1:runs>
        </team>
        <team>
          <id>2</id>
          <name>Team Two</name>
          <ns1:runs>2</ns1:runs>
        </team>
      </teams>
    </matchup>
  </matchups>
</game>

We can do the following

import SweetXml
xml_str = "..." # as above
doc = parse(xml_str, namespace_conformant: true)

Note the fact that we explicitly parse the XML with the namespace_conformant: true option. This is needed to allow nodes to be identified in a prefix independent way.

We can use namespace prefixes of our preference, regardless of what prefix is used in the document:

result = doc
  |> xpath(~x"//ff:matchup/ff:name/text()"
           |> add_namespace("ff", "http://example.com/fantasy-league"))

assert result == 'Match One'

We can specify multiple namespace prefixes:

result = doc
  |> xpath(~x"//ff:matchup//bb:runs/text()"
           |> add_namespace("ff", "http://example.com/fantasy-league")
           |> add_namespace("bb", "http://example.com/baseball-stats"))

assert result == '5'

From Chaining to Nesting

Here's a brief explanation to how nesting came about.

Chaining

Both xpath and xmap can take an :xmerl xml record as the first argment. Therefore you can chain calls to these functions like below:

doc
|> xpath(~x"//li"l)
|> Enum.map fn (li_node) ->
  %{
    name: li_node |> xpath(~x"./name/text()"),
    age: li_node |> xpath(~x"./age/text()")
  }
end

Mapping to a structure

Since the previous example is such a common use case, SweetXml allows you just simply do the following

doc
|> xpath(
  ~x"//li"l,
  name: ~x"./name/text()",
  age: ~x"./age/text()"
)

Nesting

But what you want is sometimes more complex than just that, SweetXml thus also allows nesting

doc
|> xpath(
  ~x"//li"l,
  name: [
    ~x"./name",
    first: ~x"./first/text()",
    last: ~x"./last/text()"
  ],
  age: ~x"./age/text()"
)

Transform By

Sometimes we need to transform the value to what we need, SweetXml supports that via transform_by/2

doc = "<li><name><first>john</first><last>doe</last></name><age>30</age></li>"

result = doc |> xpath(
  ~x"//li"l,
  name: [
    ~x"./name",
    first: ~x"./first/text()"s |> transform_by(&String.capitalize/1),
    last: ~x"./last/text()"s |> transform_by(&String.capitalize/1)
  ],
  age: ~x"./age/text()"i
)

^result = [%{age: 30, name: %{first: "John", last: "Doe"}}]

The same can be used to break parsing code into reusable functions that can be used in nesting

doc = "<li><name><first>john</first><last>doe</last></name><age>30</age></li>"

parse_name = fn xpath_node ->
  xpath_node |> xmap(
    first: ~x"./first/text()"s |> transform_by(&String.capitalize/1),
    last: ~x"./last/text()"s |> transform_by(&String.capitalize/1)
  )
end

result = doc |> xpath(
  ~x"//li"l,
  name: ~x"./name" |> transform_by(parse_name),
  age: ~x"./age/text()"i
)

^result = [%{age: 30, name: %{first: "John", last: "Doe"}}]

For more examples, please take a look at the tests and help.

Streaming

SweetXml now also supports streaming in various forms. Here's a sample xml doc. Notice the certain lines have xml tags that span multiple lines.

<?xml version="1.05" encoding="UTF-8"?>
<html>
  <head>
    <title>XML Parsing</title>
    <head><title>Nested Head</title></head>
  </head>
  <body>
    <p>Neato €</p><ul>
      <li class="first star" data-index="1">
        First</li><li class="second">Second
      </li><li
            class="third">Third</li>
    </ul>
    <div>
      <ul>
        <li>Forth</li>
      </ul>
    </div>
    <special_match_key>first star</special_match_key>
  </body>
</html>

Working with File.stream!

Working with streams is exactly the same as working with binaries.

File.stream!("file_above.xml") |> xpath(...)

SweetXml element streaming

Once you have a file stream, you may not want to work with the entire document to save memory.

file_stream = File.stream!("file_above.xml")

result = file_stream
|> stream_tags([:li, :special_match_key])
|> Stream.map(fn
    {_, doc} ->
      xpath(doc, ~x"./text()")
  end)
|> Enum.to_list

assert result == ['\n        First', 'Second\n      ', 'Third', 'Forth', 'first star']

⚠️ In case of large document, you may want to use the discard option to avoid memory leak.

result = file_stream
|> stream_tags([:li, :special_match_key], discard: [:li, :special_match_key])