-module(polyline).
-export([decode/1, eight_bit_chunks/1, six_bit_chunks/1, six_bit_chunk/1, split_up_six_bits/1, five_bit_chunks/1, bin_flip/1]).
%See https://developers.google.com/maps/documentation/utilities/polylinealgorithm
decode(Encoded_polyline) ->
%Steps 11 back to 8
Six_bit_chunks = six_bit_chunks(Encoded_polyline),
%Step 8
Groups_of_chunks_list = split_up_six_bits(Six_bit_chunks),
%Step 8 back to 6
Five_bit_chunks = five_bit_chunks(Groups_of_chunks_list),
%---TODO
%Maybe some more of the below need splitting out into different functions or nesting in a map?
%Which option to go for, a function that maps as per five_bit_chunks, or mapping functions as per below?
%I don't think I can map all functions in one go because ultimately need to change number of members in groups.
%I.e. following will go from groups of five to eight.
%---
%Step 5
Eight_bit_chunks = lists:map(
fun(Group_of_chunks) ->
eight_bit_chunks(Group_of_chunks)
end,
Five_bit_chunks),
Results = lists:map(
fun(Group_of_chunks) ->
%TODO These should probably be separate functions, rather than one long mess inside a map
Last_bit = [hd(lists:reverse(hd(lists:reverse(Group_of_chunks))))],
Flipped_chunks = lists:map(
fun(Chunk) ->
bin_flip_(Chunk, [])
end,
Group_of_chunks),
%Step 5
Chunks = if Last_bit =:= "1" ->
Flipped_chunks;
true ->
Group_of_chunks
end,
{ok, [Flattened_binary], []} = io_lib:fread("~2u", lists:flatten(Chunks)),
%Step 4
Shifted_binary = Flattened_binary bsr 1,
%Since bin_flip returns a string need to then change back to a number
{ok, [Shifted_binary_], []} = io_lib:fread("~2u", bin_flip(Shifted_binary - 1)),
%Step 3
Final_binary = if Last_bit =:= "1" ->
Shifted_binary_;
true ->
Shifted_binary
end,
%Step 2 back to 1
Decoded = if Last_bit =:= "1" ->
-1 * Final_binary/100000;
true ->
Final_binary/100000
end,
Decoded
end,
Eight_bit_chunks),
Results.
%Step 8 - Split up six bit chunks, per the 0x20 bit
split_up_six_bits(Bit_chunks_list) ->
split_up_six_bits_(Bit_chunks_list, [], []).
split_up_six_bits_([Head | Tail], Group_of_bit_chunks, Groups_of_bit_chunks_list) when [hd(Head)] == "1" ->
split_up_six_bits_(Tail, [Head]++Group_of_bit_chunks, Groups_of_bit_chunks_list);
split_up_six_bits_([Head | Tail], Group_of_bit_chunks, Groups_of_bit_chunks_list) when [hd(Head)] == "0" ->
%Then need to start a new list, but after this 0 one!
split_up_six_bits_(Tail, [], [lists:reverse([Head]++Group_of_bit_chunks)]++Groups_of_bit_chunks_list);
split_up_six_bits_([], Group_of_bit_chunks, Groups_of_bit_chunks_list) when length(Group_of_bit_chunks) > 0 ->
split_up_six_bits_([], [], [lists:reverse(Group_of_bit_chunks)]++Groups_of_bit_chunks_list);
split_up_six_bits_([], [], Groups_of_bit_chunks_list) ->
%TODO Might be neater to map lists:reverse over the list instead of doing above and here.
lists:reverse(Groups_of_bit_chunks_list).
%Step 5
%TODO See if better way of doing this
eight_bit_chunks(Five_bit_chunks_list) ->
Five_bit_chunk_string = lists:reverse(lists:flatten(Five_bit_chunks_list)),
eight_bit_chunks_(Five_bit_chunk_string, []).
eight_bit_chunks_(Five_bit_chunk_string, Eight_bit_chunks_list) when length(Five_bit_chunk_string) > 8 ->
Eight_bit_chunk = lists:reverse(lists:sublist(Five_bit_chunk_string,1,8)),
Rest_of_five_bit_chunk_string = lists:nthtail(8,Five_bit_chunk_string),
eight_bit_chunks_(Rest_of_five_bit_chunk_string, [Eight_bit_chunk]++Eight_bit_chunks_list);
eight_bit_chunks_(Five_bit_chunk_string, Eight_bit_chunks_list) when length(Five_bit_chunk_string) =< 8, Five_bit_chunk_string /= [] ->
Padded_bit_string = pad_to(8, lists:reverse(Five_bit_chunk_string)),
eight_bit_chunks_([], [Padded_bit_string]++Eight_bit_chunks_list);
eight_bit_chunks_([], Eight_bit_chunks_list) ->
Eight_bit_chunks_list.
six_bit_chunks(Encoded_polyline) ->
six_bit_chunks_(Encoded_polyline, []).
six_bit_chunks_([Head | Rest], Chunks_list) ->
Six_bit_chunk = six_bit_chunk(Head),
%Add to Reversed_chunks
six_bit_chunks_(Rest, [Six_bit_chunk]++Chunks_list);
six_bit_chunks_([], Chunks_list) ->
lists:reverse(Chunks_list).
six_bit_chunk(Ascii_bit) ->
%Step 10
Shifted_bit = Ascii_bit - 63,
%Step 9
%From http://erlangcentral.org/wiki/index.php/Converting_Between_Binary_and_Decimal
Binary_chunk = hd(io_lib:format("~.2B", [Shifted_bit])),
%---TODO
%What if Binary_chunk is shorter than 6?
%Well in that case, I guess that means we'd want to split, but for now pad to six and check for 0x20 elsewhere.
%---
pad_to(6, Binary_chunk).
five_bit_chunks(Groups_of_chunks_list) ->
lists:map(
fun(Group_of_chunks) ->
%Step 7 - Un-reverse the five bit chunks
lists:reverse(lists:map(
fun(Chunk) ->
%Step 8 - "Un-or" the 0x20 bit
lists:sublist(Chunk,2,6)
end,
Group_of_chunks))
end,
Groups_of_chunks_list).
%I can't figure out padding with io:format etc when printing binary numbers
pad_to(Length, Binary_string) when length(Binary_string) < Length ->
Padded_binary_string = "0"++Binary_string,
pad_to(Length, Padded_binary_string);
pad_to(Length, Binary_string) when length(Binary_string) == Length ->
Binary_string.
%bnot doesn't seem to work as I thought it would so do it very inelegantly by switching each "bit" in a string.
bin_flip(Binary_number) ->
Binary_string = hd(io_lib:format("~.2B", [Binary_number])),
bin_flip_(Binary_string, []).
bin_flip_([Head | Rest], Flipped_string) ->
Head_bit = hd(io_lib:format("~c",[Head])),
Flipped_bit = if Head_bit =:= "0" ->
"1";
true ->
"0"
end,
bin_flip_(Rest, Flipped_bit++Flipped_string);
bin_flip_([], Flipped_string) ->
lists:reverse(Flipped_string).
%Notes
%To enter numbers of a different base 2#101010
%0x1f is 31, 0x20 is 32
%Need to ensure functions are tail recursive
%To convert to integer, just need to do $<char>
%just do hd(String) will come as a number
%Last = hd(io_lib:format("~c", [hd(lists:reverse(hd(io_lib:format("~.2B", [Binary_number]))))])),
%{ok, Almost_flipped, _} = io_lib:fread("~2u",Almost_flipped_string),