From 8c3378d73e2ed23f9250aaf3dae95db44a664e75 Mon Sep 17 00:00:00 2001 From: Christopher Faulet Date: Thu, 23 Oct 2014 16:42:57 +0200 Subject: Add 'trim_all' option to binary:split/3 This option can be set to remove _ALL_ empty parts of the result of a call to binary:split/3. --- lib/stdlib/doc/src/binary.xml | 4 ++++ lib/stdlib/src/binary.erl | 39 +++++++++++++++++++-------------- lib/stdlib/test/binary_module_SUITE.erl | 28 +++++++++++++++++++++++ lib/stdlib/test/binref.erl | 37 ++++++++++++++++++------------- 4 files changed, 77 insertions(+), 31 deletions(-) (limited to 'lib/stdlib') diff --git a/lib/stdlib/doc/src/binary.xml b/lib/stdlib/doc/src/binary.xml index 2410f1f9b8..cda4b974b2 100644 --- a/lib/stdlib/doc/src/binary.xml +++ b/lib/stdlib/doc/src/binary.xml @@ -578,6 +578,10 @@ store(Binary, GBSet) ->

Removes trailing empty parts of the result (as does trim in re:split/3)

 + trim_all + +

Removes all empty parts of the result.

 + global

Repeats the split until the Subject is diff --git a/lib/stdlib/src/binary.erl b/lib/stdlib/src/binary.erl index 4850a59eb6..b94829892d 100644 --- a/lib/stdlib/src/binary.erl +++ b/lib/stdlib/src/binary.erl @@ -215,12 +215,13 @@ split(H,N) -> Subject :: binary(), Pattern :: binary() | [binary()] | cp(), Options :: [Option], - Option :: {scope, part()} | trim | global, + Option :: {scope, part()} | trim | global | trim_all, Parts :: [binary()]. split(Haystack,Needles,Options) -> try - {Part,Global,Trim} = get_opts_split(Options,{no,false,false}), + {Part,Global,Trim,TrimAll} = + get_opts_split(Options,{no,false,false,false}), Moptlist = case Part of no -> []; @@ -236,20 +237,24 @@ split(Haystack,Needles,Options) -> Match -> [Match] end end, - do_split(Haystack,MList,0,Trim) + do_split(Haystack,MList,0,Trim,TrimAll) catch _:_ -> erlang:error(badarg) end. -do_split(H,[],N,true) when N >= byte_size(H) -> +do_split(H,[],N,true,_) when N >= byte_size(H) -> []; -do_split(H,[],N,_) -> +do_split(H,[],N,_,true) when N >= byte_size(H) -> + []; +do_split(H,[],N,_,_) -> [binary:part(H,{N,byte_size(H)-N})]; -do_split(H,[{A,B}|T],N,Trim) -> +do_split(H,[{A,B}|T],N,Trim,TrimAll) -> case binary:part(H,{N,A-N}) of + <<>> when TrimAll == true -> + do_split(H,T,A+B,Trim,TrimAll); <<>> -> - Rest = do_split(H,T,A+B,Trim), + Rest = do_split(H,T,A+B,Trim,TrimAll), case {Trim, Rest} of {true,[]} -> []; @@ -257,7 +262,7 @@ do_split(H,[{A,B}|T],N,Trim) -> [<<>> | Rest] end; Oth -> - [Oth | do_split(H,T,A+B,Trim)] + [Oth | do_split(H,T,A+B,Trim,TrimAll)] end. @@ -346,14 +351,16 @@ splitat(H,N,[I|T]) -> %% Simple helper functions %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% -get_opts_split([],{Part,Global,Trim}) -> - {Part,Global,Trim}; -get_opts_split([{scope,{A,B}} | T],{_Part,Global,Trim}) -> - get_opts_split(T,{{A,B},Global,Trim}); -get_opts_split([global | T],{Part,_Global,Trim}) -> - get_opts_split(T,{Part,true,Trim}); -get_opts_split([trim | T],{Part,Global,_Trim}) -> - get_opts_split(T,{Part,Global,true}); +get_opts_split([],{Part,Global,Trim,TrimAll}) -> + {Part,Global,Trim,TrimAll}; +get_opts_split([{scope,{A,B}} | T],{_Part,Global,Trim,TrimAll}) -> + get_opts_split(T,{{A,B},Global,Trim,TrimAll}); +get_opts_split([global | T],{Part,_Global,Trim,TrimAll}) -> + get_opts_split(T,{Part,true,Trim,TrimAll}); +get_opts_split([trim | T],{Part,Global,_Trim,TrimAll}) -> + get_opts_split(T,{Part,Global,true,TrimAll}); +get_opts_split([trim_all | T],{Part,Global,Trim,_TrimAll}) -> + get_opts_split(T,{Part,Global,Trim,true}); get_opts_split(_,_) -> throw(badopt). diff --git a/lib/stdlib/test/binary_module_SUITE.erl b/lib/stdlib/test/binary_module_SUITE.erl index 32cec0db6f..f828c70b63 100644 --- a/lib/stdlib/test/binary_module_SUITE.erl +++ b/lib/stdlib/test/binary_module_SUITE.erl @@ -506,12 +506,35 @@ do_interesting(Module) -> ?line [<<1,2,3>>,<<6>>] = Module:split(<<1,2,3,4,5,6,7,8>>, [<<4,5>>,<<7>>,<<8>>], [global,trim]), + ?line [<<1,2,3>>,<<6>>] = Module:split(<<1,2,3,4,5,6,7,8>>, + [<<4,5>>,<<7>>,<<8>>], + [global,trim_all]), ?line [<<1,2,3,4,5,6,7,8>>] = Module:split(<<1,2,3,4,5,6,7,8>>, [<<4,5>>,<<7>>,<<8>>], [global,trim,{scope,{0,4}}]), ?line [<<1,2,3>>,<<6,7,8>>] = Module:split(<<1,2,3,4,5,6,7,8>>, [<<4,5>>,<<7>>,<<8>>], [global,trim,{scope,{0,5}}]), + + ?line [<<>>,<<>>,<<3>>,<<6,7,8>>] = Module:split(<<1,2,3,4,5,6,7,8>>, + [<<1>>,<<2>>,<<4,5>>], + [global,trim]), + ?line [<<3>>,<<6,7,8>>] = Module:split(<<1,2,3,4,5,6,7,8>>, + [<<1>>,<<2>>,<<4,5>>], + [global,trim_all]), + + ?line [<<1,2,3>>,<<>>,<<7,8>>] = Module:split(<<1,2,3,4,5,6,7,8>>, + [<<4,5>>,<<6>>], + [global,trim]), + ?line [<<1,2,3>>,<<7,8>>] = Module:split(<<1,2,3,4,5,6,7,8>>, + [<<4,5>>,<<6>>], + [global,trim_all]), + ?line [<<>>,<<>>,<<3>>,<<>>,<<6>>] = Module:split(<<1,2,3,4,5,6,7,8>>, + [<<1>>,<<2>>,<<4>>,<<5>>,<<7>>,<<8>>], + [global,trim]), + ?line [<<3>>,<<6>>] = Module:split(<<1,2,3,4,5,6,7,8>>, + [<<1>>,<<2>>,<<4>>,<<5>>,<<7>>,<<8>>], + [global,trim_all]), ?line badarg = ?MASK_ERROR( Module:replace(<<1,2,3,4,5,6,7,8>>, [<<4,5>>,<<7>>,<<8>>],<<99>>, @@ -1247,6 +1270,8 @@ do_random_split_comp(N,NeedleRange,HaystackRange) -> true = do_split_comp(Needle,Haystack,[]), true = do_split_comp(Needle,Haystack,[global]), true = do_split_comp(Needle,Haystack,[global,trim]), + true = do_split_comp(Needle,Haystack,[global,trim_all]), + true = do_split_comp(Needle,Haystack,[global,trim,trim_all]), do_random_split_comp(N-1,NeedleRange,HaystackRange). do_random_split_comp2(0,_,_) -> ok; @@ -1257,6 +1282,9 @@ do_random_split_comp2(N,NeedleRange,HaystackRange) -> _ <- lists:duplicate(NumNeedles,a)], true = do_split_comp(Needles,Haystack,[]), true = do_split_comp(Needles,Haystack,[global]), + true = do_split_comp(Needles,Haystack,[global,trim]), + true = do_split_comp(Needles,Haystack,[global,trim_all]), + true = do_split_comp(Needles,Haystack,[global,trim,trim_all]), do_random_split_comp2(N-1,NeedleRange,HaystackRange). do_split_comp(N,H,Opts) -> diff --git a/lib/stdlib/test/binref.erl b/lib/stdlib/test/binref.erl index 6d96736ef3..a52ea98e5a 100644 --- a/lib/stdlib/test/binref.erl +++ b/lib/stdlib/test/binref.erl @@ -155,7 +155,8 @@ split(Haystack,Needles0,Options) -> true -> exit(badtype) end, - {Part,Global,Trim} = get_opts_split(Options,{nomatch,false,false}), + {Part,Global,Trim,TrimAll} = + get_opts_split(Options,{nomatch,false,false,false}), {Start,End,NewStack} = case Part of nomatch -> @@ -180,20 +181,24 @@ split(Haystack,Needles0,Options) -> [X] end end, - do_split(Haystack,MList,0,Trim) + do_split(Haystack,MList,0,Trim,TrimAll) catch _:_ -> erlang:error(badarg) end. -do_split(H,[],N,true) when N >= byte_size(H) -> +do_split(H,[],N,true,_) when N >= byte_size(H) -> []; -do_split(H,[],N,_) -> +do_split(H,[],N,_,true) when N >= byte_size(H) -> + []; +do_split(H,[],N,_,_) -> [part(H,{N,byte_size(H)-N})]; -do_split(H,[{A,B}|T],N,Trim) -> +do_split(H,[{A,B}|T],N,Trim,TrimAll) -> case part(H,{N,A-N}) of + <<>> when TrimAll == true -> + do_split(H,T,A+B,Trim,TrimAll); <<>> -> - Rest = do_split(H,T,A+B,Trim), + Rest = do_split(H,T,A+B,Trim,TrimAll), case {Trim, Rest} of {true,[]} -> []; @@ -201,7 +206,7 @@ do_split(H,[{A,B}|T],N,Trim) -> [<<>> | Rest] end; Oth -> - [Oth | do_split(H,T,A+B,Trim)] + [Oth | do_split(H,T,A+B,Trim,TrimAll)] end. @@ -565,14 +570,16 @@ get_opts_match([{scope,{A,B}} | T],_Part) -> get_opts_match(_,_) -> throw(badopt). -get_opts_split([],{Part,Global,Trim}) -> - {Part,Global,Trim}; -get_opts_split([{scope,{A,B}} | T],{_Part,Global,Trim}) -> - get_opts_split(T,{{A,B},Global,Trim}); -get_opts_split([global | T],{Part,_Global,Trim}) -> - get_opts_split(T,{Part,true,Trim}); -get_opts_split([trim | T],{Part,Global,_Trim}) -> - get_opts_split(T,{Part,Global,true}); +get_opts_split([],{Part,Global,Trim,TrimAll}) -> + {Part,Global,Trim,TrimAll}; +get_opts_split([{scope,{A,B}} | T],{_Part,Global,Trim,TrimAll}) -> + get_opts_split(T,{{A,B},Global,Trim,TrimAll}); +get_opts_split([global | T],{Part,_Global,Trim,TrimAll}) -> + get_opts_split(T,{Part,true,Trim,TrimAll}); +get_opts_split([trim | T],{Part,Global,_Trim,TrimAll}) -> + get_opts_split(T,{Part,Global,true,TrimAll}); +get_opts_split([trim_all | T],{Part,Global,Trim,_TrimAll}) -> + get_opts_split(T,{Part,Global,Trim,true}); get_opts_split(_,_) -> throw(badopt). -- cgit v1.2.3