From a1c5ccd3b5df9bcf7d6388937bb4bdc4767e4a4b Mon Sep 17 00:00:00 2001 From: Hans Bolinder Date: Tue, 12 Feb 2019 14:51:08 +0100 Subject: stdlib: Optimize formatted printing of terms Try calling iolist_size() before calling string:length(). The reason is that calls to string:length/1 are slow when the argument is not a list of integers (or contains UNICODE). --- lib/stdlib/src/io_lib.erl | 18 +++++++++++++++++- lib/stdlib/src/io_lib_format.erl | 27 ++++++++++++++++----------- lib/stdlib/src/io_lib_pretty.erl | 20 ++++++++++---------- 3 files changed, 43 insertions(+), 22 deletions(-) (limited to 'lib') diff --git a/lib/stdlib/src/io_lib.erl b/lib/stdlib/src/io_lib.erl index 8223a52873..2b5a374cf2 100644 --- a/lib/stdlib/src/io_lib.erl +++ b/lib/stdlib/src/io_lib.erl @@ -1,7 +1,7 @@ %% %% %CopyrightBegin% %% -%% Copyright Ericsson AB 1996-2018. All Rights Reserved. +%% Copyright Ericsson AB 1996-2019. All Rights Reserved. %% %% Licensed under the Apache License, Version 2.0 (the "License"); %% you may not use this file except in compliance with the License. @@ -87,6 +87,8 @@ -export([limit_term/2]). +-export([chars_length/1]). + -export_type([chars/0, latin1_string/0, continuation/0, fread_error/0, fread_item/0, format_spec/0, chars_limit/0]). @@ -1131,3 +1133,17 @@ test_limit_map_assoc(K, V, D) -> test_limit(V, D - 1). test_limit_bitstring(_, _) -> ok. + +-spec chars_length(chars()) -> non_neg_integer(). +%% Optimized for deep lists S such that deep_latin1_char_list(S) is +%% true. No binaries allowed! It is assumed that $\r is never followed +%% by $\n if S is an iolist() (string:length() assigns such a +%% sub-sequence length 1). +chars_length(S) -> + try + %% true = deep_latin1_char_list(S), + iolist_size(S) + catch + _:_ -> + string:length(S) + end. diff --git a/lib/stdlib/src/io_lib_format.erl b/lib/stdlib/src/io_lib_format.erl index ab9031573b..d1aa4cd157 100644 --- a/lib/stdlib/src/io_lib_format.erl +++ b/lib/stdlib/src/io_lib_format.erl @@ -1,7 +1,7 @@ %% %% %CopyrightBegin% %% -%% Copyright Ericsson AB 1996-2018. All Rights Reserved. +%% Copyright Ericsson AB 1996-2019. All Rights Reserved. %% %% Licensed under the Apache License, Version 2.0 (the "License"); %% you may not use this file except in compliance with the License. @@ -248,7 +248,7 @@ count_small([#{control_char := $s}|Cs], #{w := W} = Cnts) -> count_small(Cs, Cnts#{w := W + 1}); count_small([S|Cs], #{other := Other} = Cnts) when is_list(S); is_binary(S) -> - count_small(Cs, Cnts#{other := Other + string:length(S)}); + count_small(Cs, Cnts#{other := Other + io_lib:chars_length(S)}); count_small([C|Cs], #{other := Other} = Cnts) when is_integer(C) -> count_small(Cs, Cnts#{other := Other + 1}); count_small([], #{p := P, s := S, w := W, other := Other}) -> @@ -280,10 +280,15 @@ build_limited([#{control_char := C, args := As, width := F, adjust := Ad, true -> MaxLen0 div Count0 end, S = control_limited(C, As, F, Ad, P, Pad, Enc, Str, MaxChars, I), - Len = string:length(S), NumOfPs = decr_pc(C, NumOfPs0), Count = Count0 - 1, - MaxLen = sub(MaxLen0, Len), + MaxLen = if + MaxLen0 < 0 -> % optimization + MaxLen0; + true -> + Len = io_lib:chars_length(S), + sub(MaxLen0, Len) + end, if NumOfPs > 0 -> [S|build_limited(Cs, NumOfPs, Count, MaxLen, indentation(S, I))]; @@ -406,7 +411,7 @@ base(B) when is_integer(B) -> term(T, none, _Adj, none, _Pad) -> T; term(T, none, Adj, P, Pad) -> term(T, P, Adj, P, Pad); term(T, F, Adj, P0, Pad) -> - L = string:length(T), + L = io_lib:chars_length(T), P = erlang:min(L, case P0 of none -> F; _ -> min(P0, F) end), if L > P -> @@ -713,7 +718,7 @@ fwrite_g(Fl, F, Adj, P, Pad) when P >= 1 -> end. -%% iolist_to_chars(iolist()) -> deep_char_list() +%% iolist_to_chars(iolist()) -> io_lib:chars() iolist_to_chars([C|Cs]) when is_integer(C), C >= $\000, C =< $\377 -> [C | iolist_to_chars(Cs)]; @@ -729,7 +734,7 @@ iolist_to_chars(B) when is_binary(B) -> %% cbinary() | nil()) %% cbinary() :: unicode:unicode_binary() | unicode:latin1_binary() -%% cdata_to_chars(cdata()) -> io_lib:deep_char_list() +%% cdata_to_chars(cdata()) -> io_lib:chars() cdata_to_chars([C|Cs]) when is_integer(C), C >= $\000 -> [C | cdata_to_chars(Cs)]; @@ -745,7 +750,7 @@ cdata_to_chars(B) when is_binary(B) -> limit_string(S, F, CharsLimit) when CharsLimit < 0; CharsLimit >= F -> S; limit_string(S, _F, CharsLimit) -> - case string:length(S) =< CharsLimit of + case io_lib:chars_length(S) =< CharsLimit of true -> S; false -> [string:slice(S, 0, sub(CharsLimit, 3)), "..."] end. @@ -759,11 +764,11 @@ limit_field(F, CharsLimit) -> string(S, none, _Adj, none, _Pad) -> S; string(S, F, Adj, none, Pad) -> - string_field(S, F, Adj, string:length(S), Pad); + string_field(S, F, Adj, io_lib:chars_length(S), Pad); string(S, none, _Adj, P, Pad) -> - string_field(S, P, left, string:length(S), Pad); + string_field(S, P, left, io_lib:chars_length(S), Pad); string(S, F, Adj, P, Pad) when F >= P -> - N = string:length(S), + N = io_lib:chars_length(S), if F > P -> if N > P -> adjust(flat_trunc(S, P), chars(Pad, F-P), Adj); diff --git a/lib/stdlib/src/io_lib_pretty.erl b/lib/stdlib/src/io_lib_pretty.erl index ba9d9e8434..5483ea87b5 100644 --- a/lib/stdlib/src/io_lib_pretty.erl +++ b/lib/stdlib/src/io_lib_pretty.erl @@ -1,7 +1,7 @@ %% %% %CopyrightBegin% %% -%% Copyright Ericsson AB 1996-2018. All Rights Reserved. +%% Copyright Ericsson AB 1996-2019. All Rights Reserved. %% %% Licensed under the Apache License, Version 2.0 (the "License"); %% you may not use this file except in compliance with the License. @@ -507,20 +507,20 @@ print_length(#{}=M, _D, _T, _RF, _Enc, _Str) when map_size(M) =:= 0 -> {"#{}", 3, 0, no_more}; print_length(Atom, _D, _T, _RF, Enc, _Str) when is_atom(Atom) -> S = write_atom(Atom, Enc), - {S, string:length(S), 0, no_more}; + {S, io_lib:chars_length(S), 0, no_more}; print_length(List, D, T, RF, Enc, Str) when is_list(List) -> %% only flat lists are "printable" case Str andalso printable_list(List, D, T, Enc) of true -> %% print as string, escaping double-quotes in the list S = write_string(List, Enc), - {S, string:length(S), 0, no_more}; + {S, io_lib:chars_length(S), 0, no_more}; {true, Prefix} -> %% Truncated lists when T < 0 could break some existing code. S = write_string(Prefix, Enc), %% NumOfDots = 0 to avoid looping--increasing the depth %% does not make Prefix longer. - {[S | "..."], 3 + string:length(S), 0, no_more}; + {[S | "..."], 3 + io_lib:chars_length(S), 0, no_more}; false -> case print_length_list(List, D, T, RF, Enc, Str) of {What, Len, Dots, _More} when Dots > 0 -> @@ -564,7 +564,7 @@ print_length(<<_/bitstring>> = Bin, D, T, RF, Enc, Str) -> {[$<,$<,S,$>,$>], 4 + length(S), 0, no_more}; {false, List} when is_list(List) -> S = io_lib:write_string(List, $"), %" - {[$<,$<,S,"/utf8>>"], 9 + string:length(S), 0, no_more}; + {[$<,$<,S,"/utf8>>"], 9 + io_lib:chars_length(S), 0, no_more}; {true, true, Prefix} -> S = io_lib:write_string(Prefix, $"), %" More = fun(T1, Dd) -> @@ -576,7 +576,7 @@ print_length(<<_/bitstring>> = Bin, D, T, RF, Enc, Str) -> More = fun(T1, Dd) -> ?FUNCTION_NAME(Bin, D+Dd, T1, RF, Enc, Str) end, - {[$<,$<,S|"/utf8...>>"], 12 + string:length(S), 3, More}; + {[$<,$<,S|"/utf8...>>"], 12 + io_lib:chars_length(S), 3, More}; false -> case io_lib:write_binary(Bin, D, T) of {S, <<>>} -> @@ -591,7 +591,7 @@ print_length(<<_/bitstring>> = Bin, D, T, RF, Enc, Str) -> print_length(Term, _D, _T, _RF, _Enc, _Str) -> S = io_lib:write(Term), %% S can contain unicode, so iolist_size(S) cannot be used here - {S, string:length(S), 0, no_more}. + {S, io_lib:chars_length(S), 0, no_more}. print_length_map(Map, 1, _T, RF, Enc, Str) -> More = fun(T1, Dd) -> ?FUNCTION_NAME(Map, 1+Dd, T1, RF, Enc, Str) end, @@ -651,7 +651,7 @@ print_length_record(Tuple, 1, _T, RF, RDefs, Enc, Str) -> {"{...}", 5, 3, More}; print_length_record(Tuple, D, T, RF, RDefs, Enc, Str) -> Name = [$# | write_atom(element(1, Tuple), Enc)], - NameL = string:length(Name), + NameL = io_lib:chars_length(Name), T1 = tsub(T, NameL+2), L = print_length_fields(RDefs, D - 1, T1, Tuple, 2, RF, Enc, Str), {Len, Dots} = list_length(L, NameL + 2, 0), @@ -677,7 +677,7 @@ print_length_fields([Def | Defs], D, T, Tuple, I, RF, Enc, Str) -> print_length_field(Def, D, T, E, RF, Enc, Str) -> Name = write_atom(Def, Enc), - NameL = string:length(Name) + 3, + NameL = io_lib:chars_length(Name) + 3, {_, Len, Dots, _} = Field = print_length(E, D, tsub(T, NameL), RF, Enc, Str), {{field, Name, NameL, Field}, NameL + Len, Dots, no_more}. @@ -738,7 +738,7 @@ printable_list(L, _D, T, _Uni) when T < 0-> io_lib:printable_list(L). slice(L, N) -> - try string:length(L) =< N of + try io_lib:chars_length(L) =< N of true -> all; false -> -- cgit v1.2.3