defmodule Floki do
alias Floki.Finder
alias Floki.Parser
@moduledoc """
A HTML parser and seeker.
This is a simple HTML parser that enables searching using CSS like selectors.
You can search elements by class, tag name and id.
## Example
Assuming that you have the following HTML:
```html
```
You can perform the following queries:
* Floki.find(html, "#content") : returns the section with all children;
* Floki.find(html, ".headline") : returns a list with the `p` element;
* Floki.find(html, "a") : returns a list with the `a` element;
* Floki.find(html, "[data-model=user]") : returns a list with elements that match that data attribute;
* Floki.find(html, "#content a") # returns all links inside content section;
* Floki.find(html, ".headline, a") # returns the .headline elements and links.
Each HTML node is represented by a tuple like:
{tag_name, attributes, children_nodes}
Example of node:
{"p", [{"class", "headline"}], ["Floki"]}
So even if the only child node is the element text, it is represented
inside a list.
You can write a simple HTML crawler (with support of [HTTPoison](https://github.com/edgurgel/httpoison)) with a few lines of code:
html
|> Floki.find(".pages a")
|> Floki.attribute("href")
|> Enum.map(fn(url) -> HTTPoison.get!(url) end)
It is simple as that!
"""
@type html_tree :: tuple | list
@doc """
Parses a HTML string.
## Examples
iex> Floki.parse("hello world
")
{"div", [{"class", "js-action"}], ["hello world"]}
iex> Floki.parse("first
second
")
[{"div", [], ["first"]}, {"div", [], ["second"]}]
"""
@spec parse(binary) :: html_tree
def parse(html) do
Parser.parse(html)
end
@self_closing_tags ["area", "base", "br", "col", "command", "embed", "hr", "img", "input", "keygen", "link", "mete", "param", "source", "track", "wbr"]
@doc """
Converts HTML tree to raw HTML.
Note that the resultant HTML may be different from the original one.
Spaces after tags and doctypes are ignored.
## Examples
iex> Floki.parse(~s(my content
)) |> Floki.raw_html
~s(my content
)
"""
def raw_html(html_tree), do: raw_html(html_tree, "")
defp raw_html([], html), do: html
defp raw_html(tuple, html) when is_tuple(tuple), do: raw_html([tuple], html)
defp raw_html([string|tail], html) when is_binary(string), do: raw_html(tail, html <> string)
defp raw_html([{:comment, comment}|tail], html), do: raw_html(tail, html <> "")
defp raw_html([{type, attrs, children}|tail], html) do
raw_html(tail, html <> tag_for(type, tag_attrs(attrs), children))
end
defp tag_attrs(attr_list) do
attr_list
|> Enum.reduce("", fn({attr, value}, attrs) -> ~s(#{attrs} #{attr}="#{value}") end)
|> String.strip
end
defp tag_for(type, attrs, _children) when type in @self_closing_tags do
case attrs do
"" -> "<#{type}/>"
_ -> "<#{type} #{attrs}/>"
end
end
defp tag_for(type, attrs, children) do
case attrs do
"" -> "<#{type}>#{raw_html(children)}"
_ -> "<#{type} #{attrs}>#{raw_html(children)}"
end
end
@doc """
Find elements inside a HTML tree or string.
## Examples
iex> Floki.find("hello
", ".hint")
[{"span", [{"class", "hint"}], ["hello"]}]
iex> Floki.find("", "#important")
[{"div", [{"id", "important"}], [{"div", [], ["Content"]}]}]
iex> Floki.find("Google
", "a")
[{"a", [{"href", "https://google.com"}], ["Google"]}]
"""
@spec find(binary | html_tree, binary) :: html_tree
def find(html, selector) when is_binary(html) do
parse(html) |> Finder.find(selector)
end
def find(html_tree, selector) do
Finder.find(html_tree, selector)
end
@doc """
Returns the text nodes from a HTML tree.
By default, it will perform a deep search through the HTML tree.
You can disable deep search with the option `deep` assigned to false.
## Examples
iex> Floki.text("hello world
")
"hello world"
iex> Floki.text("hello world
", deep: false)
" world"
"""
@spec text(html_tree | binary) :: binary
def text(html, opts \\ [deep: true]) do
html_tree =
case is_binary(html) do
true -> parse(html)
false -> html
end
search_strategy =
case opts[:deep] do
true -> Floki.DeepText
false -> Floki.FlatText
end
search_strategy.get(html_tree)
end
@doc """
Returns a list with attribute values for a given selector.
## Examples
iex> Floki.attribute("Google", "a", "href")
["https://google.com"]
"""
@spec attribute(binary | html_tree, binary, binary) :: list
def attribute(html, selector, attribute_name) do
html
|> find(selector)
|> attribute_values(attribute_name)
end
@doc """
Returns a list with attribute values from elements.
## Examples
iex> Floki.attribute("Google", "href")
["https://google.com"]
"""
@spec attribute(binary | html_tree, binary) :: list
def attribute(html_tree, attribute_name) when is_binary(html_tree) do
html_tree
|> parse
|> attribute_values(attribute_name)
end
def attribute(elements, attribute_name) do
attribute_values(elements, attribute_name)
end
defp attribute_values(element, attr_name) when is_tuple(element) do
attribute_values([element], attr_name)
end
defp attribute_values(elements, attr_name) do
values = Enum.reduce elements, [], fn({_, attributes, _}, acc) ->
case attribute_match?(attributes, attr_name) do
{_attr_name, value} ->
[value|acc]
_ ->
acc
end
end
Enum.reverse(values)
end
defp attribute_match?(attributes, attribute_name) do
Enum.find attributes, fn({attr_name, _}) ->
attr_name == attribute_name
end
end
end