glubs

-module(glubs@webvtt).
-compile([no_auto_import, nowarn_unused_vars, nowarn_unused_function]).

-export([to_string/1, parse/1, tokenize/1]).
-export_type([item/0, web_vtt/0, token/0, tokenization_error/0]).

-type item() :: {note, binary()} |
    {cue, gleam@option:option(binary()), integer(), integer(), binary()}.

-type web_vtt() :: {web_vtt, gleam@option:option(binary()), list(item())}.

-type token() :: {start_tag,
        binary(),
        list(binary()),
        gleam@option:option(binary())} |
    {text, binary()} |
    {timestamp, integer()} |
    {end_tag, binary()}.

-type tokenization_error() :: invalid_start_token | invalid_end_token.

-spec header_to_string(gleam@option:option(binary())) -> gleam@string_builder:string_builder().
header_to_string(Comment) ->
    case Comment of
        {some, Comment@1} ->
            gleam@string_builder:from_strings([<<" "/utf8>>, Comment@1]);

        none ->
            gleam@string_builder:new()
    end.

-spec item_to_string(item()) -> gleam@string_builder:string_builder().
item_to_string(Item) ->
    case Item of
        {note, Content} ->
            case gleam_stdlib:contains_string(Content, <<"\n"/utf8>>) of
                true ->
                    gleam@string_builder:from_strings(
                        [<<"NOTE\n"/utf8>>, Content]
                    );

                false ->
                    gleam@string_builder:from_strings(
                        [<<"NOTE "/utf8>>, Content]
                    )
            end;

        {cue, Id, Start_time, End_time, Payload} ->
            Start_time@1 = glubs@timestamp:to_string(Start_time, <<"."/utf8>>),
            End_time@1 = glubs@timestamp:to_string(End_time, <<"."/utf8>>),
            Timestamp = begin
                _pipe = Start_time@1,
                _pipe@1 = gleam@string_builder:append(_pipe, <<" --> "/utf8>>),
                gleam@string_builder:append_builder(_pipe@1, End_time@1)
            end,
            _pipe@4 = case Id of
                {some, Id@1} ->
                    _pipe@2 = gleam@string_builder:from_string(Id@1),
                    _pipe@3 = gleam@string_builder:append(
                        _pipe@2,
                        <<"\n"/utf8>>
                    ),
                    gleam@string_builder:append_builder(_pipe@3, Timestamp);

                none ->
                    Timestamp
            end,
            _pipe@5 = gleam@string_builder:append(_pipe@4, <<"\n"/utf8>>),
            gleam@string_builder:append(_pipe@5, Payload)
    end.

-spec items_to_string(list(item())) -> gleam@string_builder:string_builder().
items_to_string(Items) ->
    _pipe = Items,
    _pipe@1 = gleam@list:map(_pipe, fun item_to_string/1),
    gleam@string_builder:join(_pipe@1, <<"\n\n"/utf8>>).

-spec to_string(web_vtt()) -> binary().
to_string(Webvtt) ->
    {web_vtt, Comment, Items} = case Webvtt of
        {web_vtt, _, _} -> Webvtt;
        _assert_fail ->
            erlang:error(#{gleam_error => let_assert,
                        message => <<"Assertion pattern match failed"/utf8>>,
                        value => _assert_fail,
                        module => <<"glubs/webvtt"/utf8>>,
                        function => <<"to_string"/utf8>>,
                        line => 38})
    end,
    _pipe = <<"WEBVTT"/utf8>>,
    _pipe@1 = gleam@string_builder:from_string(_pipe),
    _pipe@2 = gleam@string_builder:append_builder(
        _pipe@1,
        header_to_string(Comment)
    ),
    _pipe@3 = gleam@string_builder:append(_pipe@2, <<"\n\n"/utf8>>),
    _pipe@4 = gleam@string_builder:append_builder(
        _pipe@3,
        items_to_string(Items)
    ),
    _pipe@5 = gleam@string_builder:append(_pipe@4, <<"\n"/utf8>>),
    gleam@string_builder:to_string(_pipe@5).

-spec parse_comment(binary()) -> {ok, gleam@option:option(binary())} |
    {error, binary()}.
parse_comment(Header) ->
    case Header of
        <<"WEBVTT"/utf8>> ->
            {ok, none};

        <<"WEBVTT\t"/utf8, Comment/binary>> ->
            {ok, {some, Comment}};

        <<"WEBVTT "/utf8, Comment@1/binary>> ->
            {ok, {some, Comment@1}};

        <<"WEBVTT"/utf8, _/binary>> ->
            {error, <<"Header comment must start with space or tab"/utf8>>};

        _ ->
            {error, <<"Must start with \"WEBVTT\""/utf8>>}
    end.

-spec parse_note(binary()) -> {ok, item()} | {error, binary()}.
parse_note(Note) ->
    case Note of
        <<"NOTE\n"/utf8, Note@1/binary>> ->
            {ok, {note, Note@1}};

        <<"NOTE "/utf8, Note@2/binary>> ->
            {ok, {note, Note@2}};

        _ ->
            {error, <<"Invalid note"/utf8>>}
    end.

-spec parse_cue_id(binary()) -> {ok, {gleam@option:option(binary()), binary()}} |
    {error, binary()}.
parse_cue_id(Cue) ->
    case gleam@string:split_once(Cue, <<"\n"/utf8>>) of
        {ok, {Id, Rest}} ->
            case gleam_stdlib:contains_string(Id, <<"-->"/utf8>>) of
                true ->
                    {ok, {none, Cue}};

                false ->
                    {ok, {{some, Id}, Rest}}
            end;

        {error, nil} ->
            {error, <<"Invalid cue"/utf8>>}
    end.

-spec parse_cue(binary()) -> {ok, item()} | {error, binary()}.
parse_cue(Cue) ->
    gleam@result:'try'(
        parse_cue_id(Cue),
        fun(_use0) ->
            {Id, Rest} = _use0,
            case gleam@string:split_once(Rest, <<"\n"/utf8>>) of
                {ok, {Line, Payload}} ->
                    gleam@result:'try'(
                        glubs@timestamp:parse_range(Line, <<"."/utf8>>),
                        fun(_use0@1) ->
                            {Start, End} = _use0@1,
                            {ok, {cue, Id, Start, End, Payload}}
                        end
                    );

                {error, nil} ->
                    {error, <<"Invalid cue"/utf8>>}
            end
        end
    ).

-spec parse_item(binary()) -> {ok, item()} | {error, binary()}.
parse_item(Item) ->
    _pipe = Item,
    _pipe@1 = parse_note(_pipe),
    gleam@result:try_recover(_pipe@1, fun(_) -> parse_cue(Item) end).

-spec parse(binary()) -> {ok, web_vtt()} | {error, binary()}.
parse(Webvtt) ->
    [Header | Body] = begin
        _pipe = Webvtt,
        _pipe@1 = gleam@string:replace(_pipe, <<"\r\n"/utf8>>, <<"\n"/utf8>>),
        _pipe@2 = gleam@string:trim_right(_pipe@1),
        gleam@string:split(_pipe@2, <<"\n\n"/utf8>>)
    end,
    [Header@1 | _] = gleam@string:split(Header, <<"\n"/utf8>>),
    gleam@result:'try'(
        parse_comment(Header@1),
        fun(Comment) ->
            gleam@result:'try'(
                gleam@list:try_map(Body, fun parse_item/1),
                fun(Items) -> {ok, {web_vtt, Comment, Items}} end
            )
        end
    ).

-spec parse_tag_and_classes(binary()) -> {binary(), list(binary())}.
parse_tag_and_classes(Input) ->
    [Tag | Classes] = gleam@string:split(Input, <<"."/utf8>>),
    {Tag, Classes}.

-spec parse_start_tag(binary()) -> token().
parse_start_tag(Input) ->
    case gleam@string:split_once(Input, <<" "/utf8>>) of
        {ok, {Tag_and_classes, Annotation}} ->
            {Tag, Classes} = parse_tag_and_classes(Tag_and_classes),
            {start_tag, Tag, Classes, {some, Annotation}};

        {error, _} ->
            {Tag@1, Classes@1} = parse_tag_and_classes(Input),
            {start_tag, Tag@1, Classes@1, none}
    end.

-spec do_tokenize(binary(), list(token())) -> {ok, list(token())} |
    {error, tokenization_error()}.
do_tokenize(Payload, Acc) ->
    case Payload of
        <<""/utf8>> ->
            {ok, Acc};

        <<"</"/utf8, Rest/binary>> ->
            case gleam@string:split_once(Rest, <<">"/utf8>>) of
                {ok, {Tag, Rest@1}} ->
                    do_tokenize(Rest@1, [{end_tag, Tag} | Acc]);

                {error, nil} ->
                    {error, invalid_end_token}
            end;

        <<"<"/utf8, Rest@2/binary>> ->
            case gleam@string:split_once(Rest@2, <<">"/utf8>>) of
                {ok, {Tag@1, Rest@3}} ->
                    case glubs@timestamp:parse(Tag@1, <<"."/utf8>>) of
                        {ok, Ts} ->
                            do_tokenize(Rest@3, [{timestamp, Ts} | Acc]);

                        {error, _} ->
                            do_tokenize(Rest@3, [parse_start_tag(Tag@1) | Acc])
                    end;

                {error, nil} ->
                    {error, invalid_start_token}
            end;

        Text ->
            case gleam@string:split_once(Text, <<"<"/utf8>>) of
                {ok, {Content, Rest@4}} ->
                    do_tokenize(
                        <<"<"/utf8, Rest@4/binary>>,
                        [{text, Content} | Acc]
                    );

                {error, nil} ->
                    {ok, [{text, Text} | Acc]}
            end
    end.

-spec tokenize(binary()) -> {ok, list(token())} | {error, tokenization_error()}.
tokenize(Payload) ->
    _pipe = Payload,
    _pipe@1 = do_tokenize(_pipe, []),
    gleam@result:map(_pipe@1, fun gleam@list:reverse/1).