126 lines
3.7 KiB
Elixir
126 lines
3.7 KiB
Elixir
defmodule Outlook.InternalTree.RawInternalBasic do
|
|
@moduledoc """
|
|
Function used for the raw_internal_tree which is a transitory state after importing
|
|
Html and before splitting textnodes into %TranslationUnit{}s.
|
|
"""
|
|
|
|
alias Ecto.UUID
|
|
alias Outlook.InternalTree.InternalNode
|
|
alias Outlook.InternalTree.TranslationUnit
|
|
alias Outlook.InternalTree.Html
|
|
|
|
@splitmarker "@@translationunit@@"
|
|
@nonperiodmarker "@@nonperiod@@"
|
|
|
|
def set_split_markers([ %InternalNode{type: :text} = textnode | rest ]) do
|
|
[ %InternalNode{textnode |
|
|
content: textnode.content
|
|
|> String.replace(~r/([[:upper:]\d])\.(\d)?/u, "\\1#{@nonperiodmarker}\\2")
|
|
|> String.replace(~r|([.?!]["'”]?\s*)|u, "\\1#{@splitmarker}")
|
|
|> String.replace(@nonperiodmarker, ".")
|
|
} | set_split_markers(rest) ]
|
|
end
|
|
|
|
def set_split_markers([ %InternalNode{type: :element} = node | rest ]) do
|
|
[ %InternalNode{node | content: set_split_markers(node.content)}
|
|
| set_split_markers(rest) ]
|
|
end
|
|
|
|
def set_split_markers([ node | rest ]) do
|
|
[ node | set_split_markers(rest) ]
|
|
end
|
|
|
|
def set_split_markers([]), do: []
|
|
|
|
def partition_to_tunits(raw_tree) do
|
|
partition_blocklevel(raw_tree)
|
|
end
|
|
|
|
def partition_blocklevel([ %InternalNode{type: :element} = node | rest ]) do
|
|
[ %InternalNode{node | content: case get_sibling_collocation(node.content) do
|
|
:block -> partition_blocklevel(node.content)
|
|
:inline -> inline_to_translation_units(node.content)
|
|
nil -> node.content
|
|
end
|
|
} | partition_blocklevel(rest) ]
|
|
end
|
|
|
|
def partition_blocklevel([ node | rest ]) do
|
|
[ node | partition_blocklevel(rest) ]
|
|
end
|
|
|
|
def partition_blocklevel([]), do: []
|
|
|
|
defp inline_to_translation_units(inline_tree) do
|
|
partition_inlinelevel(inline_tree)
|
|
|> chunk_with_list()
|
|
|> Html.strip_attributes
|
|
|> Enum.map(fn sentence ->
|
|
%TranslationUnit{
|
|
content: Html.to_html(sentence),
|
|
status: :untranslated,
|
|
uuid: UUID.generate()
|
|
}
|
|
end
|
|
)
|
|
end
|
|
|
|
def partition_inlinelevel([ %InternalNode{type: :element} = node | rest ]) do
|
|
[ partition_inlinelevel(node.content)
|
|
|> chunk_with_list()
|
|
|> Enum.map(fn nodelist -> %InternalNode{node | content: nodelist} end)
|
|
| partition_inlinelevel(rest) ]
|
|
end
|
|
|
|
def partition_inlinelevel([ %InternalNode{type: :text} = textnode | rest ]) do
|
|
content = if String.contains?(textnode.content, @splitmarker) do
|
|
String.split(textnode.content, @splitmarker, trim: true)
|
|
|> Enum.map(fn cont -> %InternalNode{textnode | content: cont} end)
|
|
else
|
|
textnode
|
|
end
|
|
[ content
|
|
| partition_inlinelevel(rest) ]
|
|
end
|
|
|
|
def partition_inlinelevel([ node | rest ]) do
|
|
[ node | partition_inlinelevel(rest) ]
|
|
end
|
|
|
|
def partition_inlinelevel([]), do: []
|
|
|
|
|
|
@doc """
|
|
iex> chunk_with_list([1, 1, [2, 2], 3, 3, [4, 4, 4], 5, 5])
|
|
[[1, 1, 2], [2, 3, 3, 4], [4], [4, 5, 5]]
|
|
iex> chunk_with_list([1, 1, [1, 2], 2, 2, [2, 3, 4], 4, 4])
|
|
[[1, 1, 1], [2, 2, 2, 2], [3], [4, 4, 4]]
|
|
"""
|
|
def chunk_with_list(list) do
|
|
chunk_fun = fn el, acc ->
|
|
if el do
|
|
{:cont, [el | acc]}
|
|
else
|
|
{:cont, Enum.reverse(acc), []}
|
|
end
|
|
end
|
|
after_fun = fn
|
|
[] -> {:cont, []}
|
|
acc -> {:cont, Enum.reverse(acc), []}
|
|
end
|
|
Enum.map(list, fn el -> is_list(el) && Enum.intersperse(el, nil) || el end)
|
|
|> List.flatten()
|
|
|> Enum.chunk_while([], chunk_fun, after_fun)
|
|
end
|
|
|
|
|
|
@doc "Returns just either :block, :inline or nil. Assumes that it doesn't contain both."
|
|
def get_sibling_collocation(content) do
|
|
content
|
|
|> Enum.map(fn node -> node.eph.sibling_with end)
|
|
|> Enum.uniq()
|
|
|> List.delete(:both)
|
|
|> List.first
|
|
end
|
|
end
|