diff options
Diffstat (limited to 'lib/stdlib')
| -rw-r--r-- | lib/stdlib/doc/src/io.xml | 12 | ||||
| -rw-r--r-- | lib/stdlib/doc/src/io_lib.xml | 14 | ||||
| -rw-r--r-- | lib/stdlib/src/io.erl | 7 | ||||
| -rw-r--r-- | lib/stdlib/src/io_lib.erl | 48 | ||||
| -rw-r--r-- | lib/stdlib/src/io_lib_pretty.erl | 29 | ||||
| -rw-r--r-- | lib/stdlib/test/io_SUITE.erl | 96 | ||||
| -rw-r--r-- | lib/stdlib/test/io_proto_SUITE.erl | 6 |
7 files changed, 177 insertions, 35 deletions
diff --git a/lib/stdlib/doc/src/io.xml b/lib/stdlib/doc/src/io.xml index 63f814ad2e..90f24c4cbc 100644 --- a/lib/stdlib/doc/src/io.xml +++ b/lib/stdlib/doc/src/io.xml @@ -211,6 +211,18 @@ </desc> </func> <func> + <name name="printable_range" arity="0"/> + <fsummary>Get user requested printable character range</fsummary> + <desc> + <p>Return the user requested range of printable Unicode characters.</p> + <p>The user can request a range of characters that are to be considered printable in heuristic detection of strings by the shell and by the formatting functions. This is done by supplying <c>+pc <range></c> when starting Erlang.</p> + <p>Currently the only valid values for <c><range></c> are <c>latin1</c> and <c>unicode</c>. <c>latin1</c> means that only code points below 256 (with the exception of control characters etc) will be considered printable. <c>unicode</c> means that all printable characters in all unicode character ranges are considered printable by the io functions.</p> + <p>By default, Erlang is started so that only the <c>latin1</c> range of characters will indicate that a list of integers is a string.</p> + <p>The simplest way to utilize the setting is to call <seealso marker="io_lib#printable_list/1">io_lib:printable_list/1</seealso>, which will use the return value of this function to decide if a list is a string of printable characters or not.</p> + <note><p>In the future, this function may return more values and ranges. It is recommended to use the io_lib:printable_list/1 function to avoid compatibility problems.</p></note> + </desc> + </func> + <func> <name name="setopts" arity="1"/> <name name="setopts" arity="2"/> <fsummary>Set options</fsummary> diff --git a/lib/stdlib/doc/src/io_lib.xml b/lib/stdlib/doc/src/io_lib.xml index 001d34a7c2..3dac259477 100644 --- a/lib/stdlib/doc/src/io_lib.xml +++ b/lib/stdlib/doc/src/io_lib.xml @@ -301,7 +301,11 @@ <fsummary>Test for a list of printable characters</fsummary> <desc> <p>Returns <c>true</c> if <c><anno>Term</anno></c> is a flat list of - printable Unicode characters, otherwise it returns <c>false</c>.</p> + printable characters, otherwise it returns <c>false</c>.</p> + <p>What is a printable character in this case is determined by the + <c>+pc</c> start up flag to the Erlang VM. See + <seealso marker="io#printable_range/0">io:printable_range/0</seealso> + and <seealso marker="erts:erl#erl">erl(1)</seealso>.</p> </desc> </func> <func> @@ -312,6 +316,14 @@ printable ISO-latin-1 characters, otherwise it returns <c>false</c>.</p> </desc> </func> + <func> + <name name="printable_unicode_list" arity="1"/> + <fsummary>Test for a list of printable Unicode characters</fsummary> + <desc> + <p>Returns <c>true</c> if <c><anno>Term</anno></c> is a flat list of + printable Unicode characters, otherwise it returns <c>false</c>.</p> + </desc> + </func> </funcs> </erlref> diff --git a/lib/stdlib/src/io.erl b/lib/stdlib/src/io.erl index 3dddb0d6e7..c92e9e3ade 100644 --- a/lib/stdlib/src/io.erl +++ b/lib/stdlib/src/io.erl @@ -32,6 +32,8 @@ parse_erl_exprs/4,parse_erl_form/1,parse_erl_form/2, parse_erl_form/3,parse_erl_form/4]). -export([request/1,request/2,requests/1,requests/2]). +%% Implemented in native code +-export([printable_range/0]). -export_type([device/0, format/0, server_no_data/0]). @@ -66,6 +68,11 @@ o_request(Io, Request, Func) -> Other end. +%% Request what the user considers printable characters +-spec printable_range() -> 'unicode' | 'latin1'. +printable_range() -> + erlang:nif_error(undefined). + %% Put chars takes mixed *unicode* list from R13 onwards. -spec put_chars(CharData) -> 'ok' when CharData :: unicode:chardata(). diff --git a/lib/stdlib/src/io_lib.erl b/lib/stdlib/src/io_lib.erl index b7ec848e1e..ff389cbe22 100644 --- a/lib/stdlib/src/io_lib.erl +++ b/lib/stdlib/src/io_lib.erl @@ -72,7 +72,7 @@ -export([quote_atom/2, char_list/1, latin1_char_list/1, deep_char_list/1, deep_latin1_char_list/1, - printable_list/1, printable_latin1_list/1]). + printable_list/1, printable_latin1_list/1, printable_unicode_list/1]). %% Utilities for collecting characters. -export([collect_chars/3, collect_chars/4, @@ -533,27 +533,45 @@ printable_latin1_list(_) -> false. %Everything else is false %% Return true if CharList is a list of printable characters, else %% false. The notion of printable in Unicode terms is somewhat floating. %% Everything that is not a control character and not invalid unicode -%% will be considered printable. +%% will be considered printable. +%% What the user has noted as printable characters is what actually +%% specifies when this function will return true. If the VM is started +%% with +pc latin1, only the latin1 range will be deemed as printable +%% if on the other hand +pc unicode is given, all characters in the Unicode +%% character set are deemed printable. latin1 is default. -spec printable_list(Term) -> boolean() when Term :: term(). -printable_list([C|Cs]) when is_integer(C), C >= $\040, C =< $\176 -> - printable_list(Cs); -printable_list([C|Cs]) +printable_list(L) -> + %% There will be more alternatives returns from io:printable range + %% in the future. To not have a catch-all caluse is deliberate. + case io:printable_range() of + latin1 -> + printable_latin1_list(L); + unicode -> + printable_unicode_list(L) + end. + +-spec printable_unicode_list(Term) -> boolean() when + Term :: term(). + +printable_unicode_list([C|Cs]) when is_integer(C), C >= $\040, C =< $\176 -> + printable_unicode_list(Cs); +printable_unicode_list([C|Cs]) when is_integer(C), C >= 16#A0, C < 16#D800; is_integer(C), C > 16#DFFF, C < 16#FFFE; is_integer(C), C > 16#FFFF, C =< 16#10FFFF -> - printable_list(Cs); -printable_list([$\n|Cs]) -> printable_list(Cs); -printable_list([$\r|Cs]) -> printable_list(Cs); -printable_list([$\t|Cs]) -> printable_list(Cs); -printable_list([$\v|Cs]) -> printable_list(Cs); -printable_list([$\b|Cs]) -> printable_list(Cs); -printable_list([$\f|Cs]) -> printable_list(Cs); -printable_list([$\e|Cs]) -> printable_list(Cs); -printable_list([]) -> true; -printable_list(_) -> false. %Everything else is false + printable_unicode_list(Cs); +printable_unicode_list([$\n|Cs]) -> printable_unicode_list(Cs); +printable_unicode_list([$\r|Cs]) -> printable_unicode_list(Cs); +printable_unicode_list([$\t|Cs]) -> printable_unicode_list(Cs); +printable_unicode_list([$\v|Cs]) -> printable_unicode_list(Cs); +printable_unicode_list([$\b|Cs]) -> printable_unicode_list(Cs); +printable_unicode_list([$\f|Cs]) -> printable_unicode_list(Cs); +printable_unicode_list([$\e|Cs]) -> printable_unicode_list(Cs); +printable_unicode_list([]) -> true; +printable_unicode_list(_) -> false. %Everything else is false %% List = nl() %% Return a list of characters to generate a newline. diff --git a/lib/stdlib/src/io_lib_pretty.erl b/lib/stdlib/src/io_lib_pretty.erl index 525b534249..4a61b033b0 100644 --- a/lib/stdlib/src/io_lib_pretty.erl +++ b/lib/stdlib/src/io_lib_pretty.erl @@ -485,7 +485,7 @@ printable_bin(Bin, Len, D, latin1) -> false end; printable_bin(Bin, Len, D, _Uni) -> - case printable_unicode(Bin, Len, []) of + case printable_unicode(Bin, Len, [], io:printable_range()) of {_, <<>>, L} -> {byte_size(Bin) =:= length(L), L}; {NC, Bin1, L} when D > 0, Len - NC >= D -> @@ -522,24 +522,27 @@ printable_latin1_list([$\e | Cs], N) -> printable_latin1_list(Cs, N - 1); printable_latin1_list([], _) -> all; printable_latin1_list(_, N) -> N. -printable_unicode(<<C/utf8, R/binary>>=Bin, I, L) when I > 0 -> - case printable_char(C) of +printable_unicode(<<C/utf8, R/binary>>=Bin, I, L, Range) when I > 0 -> + case printable_char(C,Range) of true -> - printable_unicode(R, I - 1, [C | L]); + printable_unicode(R, I - 1, [C | L],Range); false -> {I, Bin, lists:reverse(L)} end; -printable_unicode(Bin, I, L) -> +printable_unicode(Bin, I, L,_) -> {I, Bin, lists:reverse(L)}. -printable_char($\n) -> true; -printable_char($\r) -> true; -printable_char($\t) -> true; -printable_char($\v) -> true; -printable_char($\b) -> true; -printable_char($\f) -> true; -printable_char($\e) -> true; -printable_char(C) -> +printable_char($\n,_) -> true; +printable_char($\r,_) -> true; +printable_char($\t,_) -> true; +printable_char($\v,_) -> true; +printable_char($\b,_) -> true; +printable_char($\f,_) -> true; +printable_char($\e,_) -> true; +printable_char(C,latin1) -> + C >= $\s andalso C =< $~ orelse + C >= 16#A0 andalso C =< 16#FF; +printable_char(C,unicode) -> C >= $\s andalso C =< $~ orelse C >= 16#A0 andalso C < 16#D800 orelse C > 16#DFFF andalso C < 16#FFFE orelse diff --git a/lib/stdlib/test/io_SUITE.erl b/lib/stdlib/test/io_SUITE.erl index aa698ecaa2..3d4945f30e 100644 --- a/lib/stdlib/test/io_SUITE.erl +++ b/lib/stdlib/test/io_SUITE.erl @@ -29,9 +29,12 @@ manpage/1, otp_6708/1, otp_7084/1, otp_7421/1, io_lib_collect_line_3_wb/1, cr_whitespace_in_string/1, io_fread_newlines/1, otp_8989/1, io_lib_fread_literal/1, + printable_range/1, io_lib_print_binary_depth_one/1, otp_10302/1, otp_10755/1, otp_10836/1]). +-export([pretty/2]). + %-define(debug, true). -ifdef(debug). @@ -66,6 +69,7 @@ all() -> manpage, otp_6708, otp_7084, otp_7421, io_lib_collect_line_3_wb, cr_whitespace_in_string, io_fread_newlines, otp_8989, io_lib_fread_literal, + printable_range, io_lib_print_binary_depth_one, otp_10302, otp_10755, otp_10836]. groups() -> @@ -2026,6 +2030,80 @@ io_lib_fread_literal(Suite) when is_list(Suite) -> ?line {done,{ok,[]},[]} = io_lib:fread(C2, "d\n", " d"), ok. + +printable_range(doc) -> + "Check that the printable range set by the user actually works"; +printable_range(Suite) when is_list(Suite) -> + Pa = filename:dirname(code:which(?MODULE)), + {ok, UNode} = test_server:start_node(printable_range_unicode, slave, + [{args, " +pc unicode -pa " ++ Pa}]), + {ok, LNode} = test_server:start_node(printable_range_latin1, slave, + [{args, " +pc latin1 -pa " ++ Pa}]), + {ok, DNode} = test_server:start_node(printable_range_default, slave, + [{args, " -pa " ++ Pa}]), + unicode = rpc:call(UNode,io,printable_range,[]), + latin1 = rpc:call(LNode,io,printable_range,[]), + latin1 = rpc:call(DNode,io,printable_range,[]), + test_server:stop_node(UNode), + test_server:stop_node(LNode), + {ok, UNode} = test_server:start_node(printable_range_unicode, slave, + [{args, " +pcunicode -pa " ++ Pa}]), + {ok, LNode} = test_server:start_node(printable_range_latin1, slave, + [{args, " +pclatin1 -pa " ++ Pa}]), + unicode = rpc:call(UNode,io,printable_range,[]), + latin1 = rpc:call(LNode,io,printable_range,[]), + {error, _} = test_server:start_node(printable_range_unnicode, slave, + [{args, " +pcunnicode -pa " ++ Pa}]), + PrettyOptions = [{column,1}, + {line_length,109}, + {depth,30}, + {max_chars,60}, + {record_print_fun, + fun(_,_) -> no end}, + {encoding,unicode}], + 1025 = lists:max(lists:flatten(rpc:call(UNode,io_lib_pretty,print, + [{hello, [1024,1025]}, + PrettyOptions]))), + 125 = lists:max(lists:flatten(rpc:call(LNode,io_lib_pretty,print, + [{hello, [1024,1025]}, + PrettyOptions]))), + 125 = lists:max(lists:flatten(rpc:call(DNode,io_lib_pretty,print, + [{hello, [1024,1025]}, + PrettyOptions]))), + 1025 = lists:max(lists:flatten(rpc:call(UNode,io_lib_pretty,print, + [{hello, <<1024/utf8,1025/utf8>>}, + PrettyOptions]))), + 125 = lists:max(lists:flatten(rpc:call(LNode,io_lib_pretty,print, + [{hello, <<1024/utf8,1025/utf8>>}, + PrettyOptions]))), + 125 = lists:max(lists:flatten(rpc:call(DNode,io_lib_pretty,print, + [{hello, <<1024/utf8,1025/utf8>>}, + PrettyOptions]))), + + 1025 = lists:max(lists:flatten(rpc:call(UNode,io_lib,format, + ["~tp",[{hello, [1024,1025]}]]))), + 125 = lists:max(lists:flatten(rpc:call(LNode,io_lib,format, + ["~tp",[{hello, [1024,1025]}]]))), + 125 = lists:max(lists:flatten(rpc:call(DNode,io_lib,format, + ["~tp",[{hello, [1024,1025]}]]))), + 1025 = lists:max(lists:flatten(rpc:call(UNode,io_lib,format, + ["~tp", + [{hello, + <<1024/utf8,1025/utf8>>}]]))), + 125 = lists:max(lists:flatten(rpc:call(LNode,io_lib,format, + ["~tp", + [{hello, + <<1024/utf8,1025/utf8>>}]]))), + 125 = lists:max(lists:flatten(rpc:call(DNode,io_lib,format, + ["~tp", + [{hello, + <<1024/utf8,1025/utf8>>}]]))), + test_server:stop_node(UNode), + test_server:stop_node(LNode), + test_server:stop_node(DNode), + ok. + + io_lib_print_binary_depth_one(doc) -> "Test binaries printed with a depth of one behave correctly"; io_lib_print_binary_depth_one(Suite) when is_list(Suite) -> @@ -2040,10 +2118,22 @@ io_lib_print_binary_depth_one(Suite) when is_list(Suite) -> otp_10302(doc) -> "OTP-10302. Unicode"; otp_10302(Suite) when is_list(Suite) -> - "\"\x{400}\"" = pretty("\x{400}", -1), - "<<\"\x{400}\"/utf8>>" = pretty(<<"\x{400}"/utf8>>, -1), + Pa = filename:dirname(code:which(?MODULE)), + {ok, UNode} = test_server:start_node(printable_range_unicode, slave, + [{args, " +pc unicode -pa " ++ Pa}]), + {ok, LNode} = test_server:start_node(printable_range_latin1, slave, + [{args, " +pc latin1 -pa " ++ Pa}]), + "\"\x{400}\"" = rpc:call(UNode,?MODULE,pretty,["\x{400}", -1]), + "<<\"\x{400}\"/utf8>>" = rpc:call(UNode,?MODULE,pretty,[<<"\x{400}"/utf8>>, -1]), + + "<<\"\x{400}foo\"/utf8>>" = rpc:call(UNode,?MODULE,pretty,[<<"\x{400}foo"/utf8>>, 2]), + "[1024]" = rpc:call(LNode,?MODULE,pretty,["\x{400}", -1]), + "<<208,128>>" = rpc:call(LNode,?MODULE,pretty,[<<"\x{400}"/utf8>>, -1]), + + "<<208,...>>" = rpc:call(LNode,?MODULE,pretty,[<<"\x{400}foo"/utf8>>, 2]), + test_server:stop_node(UNode), + test_server:stop_node(LNode), - "<<\"\x{400}foo\"/utf8>>" = pretty(<<"\x{400}foo"/utf8>>, 2), "<<\"äppl\"/utf8>>" = pretty(<<"äppl"/utf8>>, 2), "<<\"äppl\"/utf8...>>" = pretty(<<"äpple"/utf8>>, 2), "<<\"apel\">>" = pretty(<<"apel">>, 2), diff --git a/lib/stdlib/test/io_proto_SUITE.erl b/lib/stdlib/test/io_proto_SUITE.erl index 299daf0e42..ddcc8dfdab 100644 --- a/lib/stdlib/test/io_proto_SUITE.erl +++ b/lib/stdlib/test/io_proto_SUITE.erl @@ -716,7 +716,7 @@ binary_options(Config) when is_list(Config) -> {getline, "<<\"hej\\n\">>"}, {putline, "io:get_line('')."}, {putline, binary_to_list(<<"\345\344\366"/utf8>>)}, - {getline, "<<\""++binary_to_list(unicode:characters_to_binary(<<"\345\344\366"/utf8>>,latin1,utf8))++"\\n\">>"} + {getline, "<<\""++binary_to_list(<<"\345\344\366"/utf8>>)++"\\n\"/utf8>>"} ],[]) end, %% And one with oldshell @@ -1784,8 +1784,8 @@ get_default_shell() -> {putline, "whereis(user_drv)."}, {getline, "undefined"}],[]), old - catch E:R -> - ?dbg({E,R}), + catch _E:_R -> + ?dbg({_E,_R}), new end. |