add coalescer

This commit is contained in:
byte[] 2019-11-03 19:35:04 -05:00
parent 9cfc426103
commit 43217607cd
2 changed files with 24 additions and 0 deletions

View file

@ -5,6 +5,13 @@ defmodule Textile.MarkupLexer do
# Markup tags # Markup tags
def markup_ending_in(ending_sequence) do def markup_ending_in(ending_sequence) do
double_newline =
string("\n\n")
|> unwrap_and_tag(:double_newline)
newline =
string("\n")
|> unwrap_and_tag(:newline)
# The literal tag is special, because # The literal tag is special, because
# 1. It needs to capture everything inside it as a distinct token. # 1. It needs to capture everything inside it as a distinct token.
@ -147,6 +154,8 @@ defmodule Textile.MarkupLexer do
bracketed_markup_opening_tags |> lookahead_not(space()), bracketed_markup_opening_tags |> lookahead_not(space()),
special_characters() |> concat(markup_opening_tags), special_characters() |> concat(markup_opening_tags),
markup_closing_tags |> choice([special_characters(), ending_sequence]), markup_closing_tags |> choice([special_characters(), ending_sequence]),
double_newline,
newline,
utf8_char([]) utf8_char([])
]) ])

View file

@ -0,0 +1,15 @@
defmodule Textile.TokenCoalescer do
# The lexer, as a practical concern, does not coalesce runs of
# character tokens. This fixes that.
def coalesce(tokens) do
tokens
|> Enum.chunk_by(&is_number(&1))
|> Enum.flat_map(fn
[t | _rest] = str when is_number(t) ->
[text: List.to_string(str)]
t ->
t
)
end
end