Add binary:part to erl_bif_binary.c

Change name of the 'scope' option for binary:match/matches.
Add split and replace to binary.erl.
Cleanup comments etc in binary.erl and atom.names
Add testcases for part, split, replace and scopes.

1 files changed, 150 insertions, 53 deletions

diff --git a/lib/stdlib/src/binary.erl b/lib/stdlib/src/binary.erl
index d649f974bb..3eaf011db4 100644
--- a/lib/stdlib/src/binary.erl
+++ b/lib/stdlib/src/binary.erl
@@ -20,61 +20,158 @@
 %%
 %% The following functions implemented as BIF's
 %%  binary:compile_pattern/1
-%%  binary:find/3
-%% XXX:PaN more to come...
-
--export([first/1,first/2,last/1,last/2,nth/2,extract/3]).
+%%  binary:match/{2,3}
+%%  binary:matches/{2,3}
+%% - Not yet:
+%%  binary:longest_common_prefix/1
+%%  binary:longest_common_suffix/1
+%%  binary:first/1
+%%  binary:last/1
+%%  binary:at/2
+%%  binary:part/{2,3}
+%%  binary:bin_to_list/{1,2,3}
+%%  binary:list_to_bin/1
+%%  binary:copy/{1,2}
+%%  binary:referenced_byte_size/1
+%%  binary:decode_unsigned/{1,2}
+%%
+%% Implemented in this module:
+-export([split/2,split/3,replace/3,replace/4]).
 
-first(<<F:1/binary,_/binary>>) ->
-    F;
-first(_) ->
-    erlang:error(badarg).
-first(N,Bin) when is_integer(N), N >= 0 ->
-    case Bin of
-	<<F:N/binary,_/binary>> ->
-	    F;
-	_ ->
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+%% split
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+split(H,N) ->
+    split(H,N,[]).
+split(Haystack,Needles,Options) ->
+    try
+	{Part,Global,Trim} = get_opts_split(Options,{no,false,false}),
+	Moptlist = case Part of
+		       no ->
+			   [];
+		       {A,B} ->
+			   [{scope,{A,B}}]
+		   end,
+	MList = if
+		    Global ->
+			binary:matches(Haystack,Needles,Moptlist);
+		    true ->
+			case binary:match(Haystack,Needles,Moptlist) of
+			    nomatch -> [];
+			    Match -> [Match]
+			end
+		end,
+	do_split(Haystack,MList,0,Trim)
+    catch
+	_:_ ->
 	    erlang:error(badarg)
-    end;
-first(_,_) ->
-    erlang:error(badarg).
-last(<<>>) ->
-    erlang:error(badarg);
-last(Bin) when is_binary(Bin) ->
-    Sz = byte_size(Bin) - 1,
-    <<_:Sz/binary,L:1/binary>> = Bin,
-    L;
-last(_) ->
-    erlang:error(badarg).
-last(N,Bin) when is_integer(N), N >= 0, is_binary(Bin) ->
-    Sz = byte_size(Bin) - N,
-    case Bin of
-	<<_:Sz/binary,L:N/binary>> ->
-	    L;
-	_ ->
-	    erlang:error(badarg)
-    end;
-last(_,_) ->
-    erlang:error(badarg).
+    end.
 
-nth(N, Bin) when is_integer(N), N > 0 ->
-    M = N - 1,
-    case Bin of
-	<<_:M/binary,V:1/binary,_/binary>> ->
-	    V;
-	_ ->
-	    erlang:error(badarg)
-    end;
-nth(_,_) ->
-    erlang:error(badarg).
+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) ->
+    case binary:part(H,{N,A-N}) of
+	<<>> ->
+	    Rest =  do_split(H,T,A+B,Trim),
+	    case {Trim, Rest} of
+		{true,[]} ->
+		    [];
+		_ ->
+		    [<<>> | Rest]
+	    end;
+	Oth ->
+	    [Oth | do_split(H,T,A+B,Trim)]
+    end.
 
-extract(N,Size,Bin) when is_integer(N), is_integer(Size), is_binary(Bin) ->
-    M = N - 1,
-    case Bin of
-	<<_:M/binary,V:Size/binary,_/binary>> ->
-	    V;
-	_ ->
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+%% replace
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+replace(H,N,R) ->
+    replace(H,N,R,[]).
+replace(Haystack,Needles,Replacement,Options) ->
+    try
+	true = is_binary(Replacement), % Make badarg instead of function clause
+	{Part,Global,Insert} = get_opts_replace(Options,{no,false,[]}),
+	Moptlist = case Part of
+		       no ->
+			   [];
+		       {A,B} ->
+			   [{scope,{A,B}}]
+		   end,
+	MList = if
+		    Global ->
+			binary:matches(Haystack,Needles,Moptlist);
+		    true ->
+			case binary:match(Haystack,Needles,Moptlist) of
+			    nomatch -> [];
+			    Match -> [Match]
+			end
+		end,
+	ReplList = case Insert of
+		       [] ->
+			   Replacement;
+		       Y when is_integer(Y) ->
+			   splitat(Replacement,0,[Y]);
+		       Li when is_list(Li) ->
+			   splitat(Replacement,0,lists:sort(Li))
+		   end,
+	erlang:iolist_to_binary(do_replace(Haystack,MList,ReplList,0))
+   catch
+       _:_ ->
 	    erlang:error(badarg)
-    end;
-extract(_,_,_) ->
-    erlang:error(badarg).
+   end.
+
+
+do_replace(H,[],_,N) ->
+    [binary:part(H,{N,byte_size(H)-N})];
+do_replace(H,[{A,B}|T],Replacement,N) ->
+    [binary:part(H,{N,A-N}),
+     if
+	 is_list(Replacement) ->
+	     do_insert(Replacement, binary:part(H,{A,B}));
+	 true ->
+	     Replacement
+     end
+     | do_replace(H,T,Replacement,A+B)].
+
+do_insert([X],_) ->
+    [X];
+do_insert([H|T],R) ->
+    [H,R|do_insert(T,R)].
+
+splitat(H,N,[]) ->
+    [binary:part(H,{N,byte_size(H)-N})];
+splitat(H,N,[I|T]) ->
+    [binary:part(H,{N,I-N})|splitat(H,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(_,_) ->
+    throw(badopt).
+
+get_opts_replace([],{Part,Global,Insert}) ->
+    {Part,Global,Insert};
+get_opts_replace([{scope,{A,B}} | T],{_Part,Global,Insert}) ->
+    get_opts_replace(T,{{A,B},Global,Insert});
+get_opts_replace([global | T],{Part,_Global,Insert}) ->
+    get_opts_replace(T,{Part,true,Insert});
+get_opts_replace([{insert_replaced,N} | T],{Part,Global,_Insert}) ->
+    get_opts_replace(T,{Part,Global,N});
+get_opts_replace(_,_) ->
+    throw(badopt).
+