| 1 |
|
-module(klsn_binstr). |
| 2 |
|
|
| 3 |
|
-export([ |
| 4 |
|
urlencode/1 |
| 5 |
|
, urldecode/1 |
| 6 |
|
, from_any/1 |
| 7 |
|
, replace/2 |
| 8 |
|
, hash/1 |
| 9 |
|
]). |
| 10 |
|
|
| 11 |
|
%% UTF-8 binary string used as the canonical textual representation across |
| 12 |
|
%% the klsn_* helper modules. |
| 13 |
|
-export_type([ |
| 14 |
|
binstr/0 |
| 15 |
|
]). |
| 16 |
|
|
| 17 |
|
-type binstr() :: unicode:unicode_binary(). |
| 18 |
|
|
| 19 |
|
|
| 20 |
|
%% @doc |
| 21 |
|
%% Return a lowercase hexadecimal SHA-256 digest of Binary. |
| 22 |
|
-spec hash(binstr()) -> binstr(). |
| 23 |
|
hash(Binary) -> |
| 24 |
3 |
list_to_binary( |
| 25 |
|
string:lowercase( |
| 26 |
|
binary_to_list( |
| 27 |
|
binary:encode_hex( |
| 28 |
|
crypto:hash(sha256, Binary) |
| 29 |
|
) |
| 30 |
|
) |
| 31 |
|
) |
| 32 |
|
). |
| 33 |
|
|
| 34 |
|
%% @doc |
| 35 |
|
%% Apply each replacement rule {Before,After} to Sub using |
| 36 |
|
%% binary:replace/4 with the global option. Rules are processed from |
| 37 |
|
%% left to right so earlier replacements feed into later ones. |
| 38 |
|
%% |
| 39 |
|
%% Returns the transformed binary. |
| 40 |
|
-spec replace([{binstr(), binstr()}], binstr()) -> binstr(). |
| 41 |
|
replace(Rule, Sub) -> |
| 42 |
7 |
lists:foldl(fun({Before, After}, Acc) -> |
| 43 |
8 |
binary:replace(Acc, Before, After, [global]) |
| 44 |
|
end, Sub, Rule). |
| 45 |
|
|
| 46 |
|
%% @doc |
| 47 |
|
%% Convert an integer, float, atom or arbitrary iolist to a UTF-8 binary. |
| 48 |
|
%% |
| 49 |
|
%% If the value is already a binary it is returned unchanged (through the |
| 50 |
|
%% iolist clause). |
| 51 |
|
-spec from_any(any()) -> binstr(). |
| 52 |
|
from_any(Integer) when is_integer(Integer) -> |
| 53 |
1 |
integer_to_binary(Integer); |
| 54 |
|
from_any(Float) when is_float(Float) -> |
| 55 |
1 |
float_to_binary(Float); |
| 56 |
|
from_any(Atom) when is_atom(Atom) -> |
| 57 |
1 |
atom_to_binary(Atom); |
| 58 |
|
from_any(IOList) -> |
| 59 |
1 |
iolist_to_binary(IOList). |
| 60 |
|
|
| 61 |
|
%% @doc |
| 62 |
|
%% Percent-encode Bin so that it becomes safe to embed inside a URL. |
| 63 |
|
%% The implementation keeps alphanumerics and the unreserved characters |
| 64 |
|
%% - . _ ~ intact and encodes everything else as %XX. |
| 65 |
|
-spec urlencode(binstr()) -> binstr(). |
| 66 |
|
urlencode(Bin) -> |
| 67 |
8 |
IsSafe = fun |
| 68 |
2 |
($-) -> true; |
| 69 |
2 |
($.) -> true; |
| 70 |
2 |
($_) -> true; |
| 71 |
2 |
($~) -> true; |
| 72 |
30 |
(C) when $A =< C, C =< $Z -> true; |
| 73 |
54 |
(C) when $a =< C, C =< $z -> true; |
| 74 |
13 |
(C) when $0 =< C, C =< $9 -> true; |
| 75 |
76 |
(_) -> false |
| 76 |
|
end, |
| 77 |
8 |
urlencode(Bin, IsSafe, <<>>). |
| 78 |
|
|
| 79 |
|
urlencode(<<>>, _, Acc) -> |
| 80 |
7 |
Acc; |
| 81 |
|
urlencode(<<Char/utf8, Rest/binary>>, IsSafe, Acc) -> |
| 82 |
181 |
Encoded = case IsSafe(Char) of |
| 83 |
|
false -> |
| 84 |
76 |
percent_encode(<<Char/utf8>>, <<>>); |
| 85 |
|
true -> |
| 86 |
105 |
<<Char/utf8>> |
| 87 |
|
end, |
| 88 |
181 |
urlencode(Rest, IsSafe, <<Acc/binary, Encoded/binary>>); |
| 89 |
|
urlencode(<<Char:1/binary, Rest/binary>>, IsSafe, Acc) -> |
| 90 |
128 |
Encoded = percent_encode(<<Char/binary>>, <<>>), |
| 91 |
128 |
urlencode(Rest, IsSafe, <<Acc/binary, Encoded/binary>>); |
| 92 |
|
urlencode(_, _, _) -> |
| 93 |
1 |
error(badarg). |
| 94 |
|
|
| 95 |
|
percent_encode(<<>>, Acc) -> |
| 96 |
204 |
Acc; |
| 97 |
|
percent_encode(<<A:4, B:4, Rest/binary>>, Acc) -> |
| 98 |
214 |
EncodedLeft = encode_half_char(A), |
| 99 |
214 |
EncodedRight = encode_half_char(B), |
| 100 |
214 |
percent_encode(Rest, << |
| 101 |
|
Acc/binary |
| 102 |
|
, $% |
| 103 |
|
, EncodedLeft |
| 104 |
|
, EncodedRight |
| 105 |
|
>>). |
| 106 |
|
|
| 107 |
|
encode_half_char(HalfChar) -> |
| 108 |
428 |
if |
| 109 |
239 |
((HalfChar) >= 0) andalso ((HalfChar) =< 9) -> (HalfChar) + $0; |
| 110 |
189 |
((HalfChar) >= 10) andalso ((HalfChar) =< 15) -> (HalfChar) + $A - 10 |
| 111 |
|
end. |
| 112 |
|
|
| 113 |
|
-spec urldecode(binstr()) -> binstr(). |
| 114 |
|
urldecode(Bin) when is_binary(Bin) -> |
| 115 |
8 |
urldecode(Bin, <<>>); |
| 116 |
|
urldecode(_) -> |
| 117 |
1 |
error(badarg). |
| 118 |
|
|
| 119 |
|
urldecode(<<>>, Acc) -> |
| 120 |
8 |
Acc; |
| 121 |
|
urldecode(<<"%", H, L, Rest/binary>>, Acc) -> |
| 122 |
217 |
V0 = hex_val(H), |
| 123 |
217 |
V1 = hex_val(L), |
| 124 |
217 |
Byte = V0 * 16 + V1, |
| 125 |
217 |
urldecode(Rest, <<Acc/binary, Byte>>); |
| 126 |
|
urldecode(<<Byte, Rest/binary>>, Acc) -> |
| 127 |
117 |
urldecode(Rest, <<Acc/binary, Byte>>). |
| 128 |
|
|
| 129 |
|
-spec hex_val(integer()) -> integer(). |
| 130 |
242 |
hex_val(H) when $0 =< H, H =< $9 -> H - $0; |
| 131 |
189 |
hex_val(H) when $A =< H, H =< $F -> H - $A + 10; |
| 132 |
3 |
hex_val(H) when $a =< H, H =< $f -> H - $a + 10. |
| 133 |
|
|
| 134 |
|
|
| 135 |
|
|