mirror of
https://github.com/philomena-dev/philomena.git
synced 2024-11-27 21:47:59 +01:00
101 lines
No EOL
1.9 KiB
Elixir
101 lines
No EOL
1.9 KiB
Elixir
defmodule Search.Lexer do
|
|
import NimbleParsec
|
|
import Search.Helpers
|
|
|
|
float =
|
|
optional(ascii_char('-+'))
|
|
|> ascii_string([?0..?9], min: 1)
|
|
|> optional(ascii_char('.') |> ascii_string([?0..?9], min: 1))
|
|
|> reduce({List, :to_string, []})
|
|
|> reduce(:to_number)
|
|
|
|
l_and =
|
|
choice([string("AND"), string("&&"), string(",")])
|
|
|> unwrap_and_tag(:and)
|
|
|
|
l_or =
|
|
choice([string("OR"), string("||")])
|
|
|> unwrap_and_tag(:or)
|
|
|
|
l_not =
|
|
choice([string("NOT"), string("!"), string("-")])
|
|
|> unwrap_and_tag(:not)
|
|
|
|
lparen = string("(") |> unwrap_and_tag(:lparen)
|
|
rparen = string(")") |> unwrap_and_tag(:rparen)
|
|
|
|
space =
|
|
choice([string(" "), string("\t"), string("\n"), string("\r"), string("\v"), string("\f")])
|
|
|> ignore()
|
|
|
|
quot = string("\"")
|
|
backslash = string("\\")
|
|
|
|
boost =
|
|
ignore(string("^"))
|
|
|> concat(float)
|
|
|> unwrap_and_tag(:boost)
|
|
|
|
stop_words =
|
|
repeat(space)
|
|
|> choice([
|
|
backslash |> eos(),
|
|
l_and,
|
|
l_or,
|
|
rparen,
|
|
boost
|
|
])
|
|
|
|
defcombinatorp(
|
|
:dirty_text,
|
|
lookahead_not(stop_words)
|
|
|> choice([
|
|
string("\\") |> utf8_char([]),
|
|
string("(") |> parsec(:dirty_text) |> string(")"),
|
|
utf8_char(not: ?(..?))
|
|
])
|
|
|> times(min: 1)
|
|
)
|
|
|
|
text =
|
|
parsec(:dirty_text)
|
|
|> reduce({List, :to_string, []})
|
|
|> unwrap_and_tag(:text)
|
|
|
|
quoted_text =
|
|
ignore(quot)
|
|
|> choice([
|
|
ignore(string("\\")) |> string("\""),
|
|
ignore(string("\\")) |> string("\\"),
|
|
string("\\") |> utf8_char([]),
|
|
utf8_char(not: ?")
|
|
])
|
|
|> repeat()
|
|
|> ignore(quot)
|
|
|> reduce({List, :to_string, []})
|
|
|> unwrap_and_tag(:text)
|
|
|
|
term =
|
|
choice([
|
|
quoted_text,
|
|
text
|
|
])
|
|
|
|
outer =
|
|
choice([
|
|
l_and,
|
|
l_or,
|
|
l_not,
|
|
lparen,
|
|
rparen,
|
|
boost,
|
|
space,
|
|
term
|
|
])
|
|
|
|
search =
|
|
repeat(outer)
|
|
|> eos()
|
|
|
|
defparsec :lex, search
|
|
end |