require "lpeg"

local C, Cs, Ct, P, R, S, V, match = lpeg.C, lpeg.Cs, lpeg.Ct, lpeg.P, lpeg.R, lpeg.S, lpeg.V, lpeg.match

local iso_parser

rules = P{
    [1] = "weight_table",

    -- Define collation tables as sequences of lines

    weight_table = V"common_template_table" + V"tailored_table",
    common_template_table = V"simple_line"^0,
    tailored_table = V"table_line"^0,

    -- Define the line types

    simple_line = (V"symbol_definition" + V"collating_element" +
                   V"weight_assignment" + V"order_end")^-1 * V"line_completion" --/ function (first) io.write("simple: "..first) end
                   ,
    --table_line = V"simple_line" + V"tailoring_line",
    table_line = V"tailoring_line" + V"simple_line",
    tailoring_line = (V"reorder_after" + V"order_start" + V"reorder_end" +
                      V"section_definition" + V"reorder_section_after") *
                      V"line_completion" --/ function (first) io.write("tailoring: "..first) end
                      ,

    -- Define the basic syntax for collation weighting

    symbol_definition = P"collating-symbol" * V"space"^1 * V"symbol_element",
    symbol_element = V"symbol"-V"symbol_range" + V"symbol_range",
    symbol_range = V"symbol" * P".." * V"symbol",
    symbol = V"simple_symbol" + V"ucs_symbol",
    ucs_symbol = (P"<U"  * V"one_to_eight_digit_hex_string" * P">") +
                 (P"<U-" * V"one_to_eight_digit_hex_string" * P">"),
    simple_symbol = P"<" * V"identifier" * P">",
    collating_element = P"collating-element" * V"space"^1 * V"symbol" * V"space"^1 *
                        P"from" * V"space"^1 * V"quoted_symbol_sequence",
    quoted_symbol_sequence = P'"' * V"simple_weight"^1 * P'"',
    --weight_assignment = V"simple_weight" + V"symbol_weight",
    weight_assignment = V"symbol_weight" + V"simple_weight",
    simple_weight = V"symbol_element" + P"UNDEFINED",
    symbol_weight = V"symbol_element" * V"space"^1 * V"weight_list",
    weight_list = V"level_token" * (V"semicolon" * V"level_token")^0,
    level_token = V"symbol_group" + P"IGNORE",
    symbol_group = V"symbol_element" + V"quoted_symbol_sequence",
    order_end = P"order_end",

    -- Define the tailoring syntax

    reorder_after = P"reorder-after" * V"space"^1 * V"target_symbol",
    target_symbol = V"symbol",
    order_start = P"order_start" * V"space"^1 * V"multiple_level_direction",
    multiple_level_direction = V"direction" * (V"semicolon" * V"direction")^0 * P",position"^-1,
    direction = P"forward" + P"backward",
    reorder_end = P"reorder-end",
    section_definition = V"section_definition_simple" + V"section_definition_list",
    section_definition_simple = P"section" * V"space"^1 * V"section_identifier",
    section_identifier = V"identifier",
    section_definition_list = P"section" * V"space"^1 * V"section_identifier" * V"space"^1 * V"symbol_list",
    symbol_list = V"symbol_element" * (V"semicolon" * V"symbol_element")^0,
    reorder_section_after = P"reorder-section-after" * V"space"^1 * V"section_identifier" * V"space"^1 * V"target_symbol",

    -- Define low-level tokens used by the rest of the syntax

    identifier = (V"letter" + V"digit") * V"id_part"^0,
    id_part = V"letter" + V"digit" + S"-_",
    line_completion = V"space"^0 * V"comment"^-1 * V"EOL",
    comment = V"comment_char" * V"character"^0,
    one_to_eight_digit_hex_string = V"hex_upper"^-8,
    hex_numeric_string = V"hex_upper"^1,
    space = S" \t",
    semicolon = P";",
    comment_char = P"%",
    digit = R"09",
    hex_upper = V"digit" + S"ABCDEF",
    letter = R"az" + R"AZ",
    EOL = P"\n",
    character = 1-V"EOL",
}

f = io.open("iso14651.txt", "r")
tab = f:read("*all")
f:close()

--rules:print()
print(rules:match(tab))