stdlib: Add classifying functions

When making scanners there might be a need to classify codepoints if
the specification follows the Unicode standard.

The following:
 is_whitespace/1,
 is_ID_start/1,
 is_ID_continue/1,
 category/1
is (at least) needed to scan and parse for example markdown.

This commit is backward incompatible if the *undocumented*
unicode_util:[is_]whitespace(..) functions have been used.

unicode_util:whitespace() have been changed to
unicode_util:pattern_whitespace() string module uses it and documents
that is uses the `pattern_whitespace` property.

So the name of the function have been changed, this is because
`pattern_whitespace` is not a subset of `whitespace` characters,
as was discovered during testing.

is_whitespace(Char) have been changed to use the Unicode `whitespace`
property.

See Unicode Standard Annex #31 and #44.

Author: dgud

lib/stdlib/src/erl_stdlib_errors.erl

@@ -428,7 +428,15 @@ format_unicode_error(characters_to_nfkc_list, [_]) ->
 format_unicode_error(characters_to_nfkd_binary, [_]) ->
     [bad_char_data];
 format_unicode_error(characters_to_nfkd_list, [_]) ->
-    [bad_char_data].
+    [bad_char_data];
+format_unicode_error(category, [_]) ->
+    [bad_char];
+format_unicode_error(is_whitespace, [_]) ->
+    [bad_char];
+format_unicode_error(is_ID_start, [_]) ->
+    [bad_char];
+format_unicode_error(is_ID_continue, [_]) ->
+    [bad_char].
 
 unicode_char_data(Chars) ->
     try unicode:characters_to_binary(Chars) of
@@ -1121,6 +1129,8 @@ expand_error(bad_boolean) ->
     <<"not a boolean value">>;
 expand_error(bad_binary_list) ->
     <<"not a flat list of binaries">>;
+expand_error(bad_char) ->
+    <<"not a valid character">>;
 expand_error(bad_char_data) ->
     <<"not valid character data (an iodata term)">>;
 expand_error(bad_binary_pattern) ->

lib/stdlib/src/string.erl

@@ -477,7 +477,7 @@ pad(CD, Length, both, Char) when is_integer(Length) ->
 -spec trim(String) -> unicode:chardata() when
       String :: unicode:chardata().
 trim(Str) ->
-    trim(Str, both, unicode_util:whitespace()).
+    trim(Str, both, unicode_util:pattern_whitespace()).
 
 -doc """
 Equivalent to [`trim(String, Dir, Whitespace})`](`trim/3`) where 
@@ -490,7 +490,7 @@ as Pattern_White_Space in
       String :: unicode:chardata(),
       Dir :: direction() | 'both'.
 trim(Str, Dir) ->
-    trim(Str, Dir, unicode_util:whitespace()).
+    trim(Str, Dir, unicode_util:pattern_whitespace()).
 
 -doc """
 Returns a string, where leading or trailing, or both, `Characters` have been

lib/stdlib/src/unicode.erl

@@ -21,7 +21,7 @@
 %%
 -module(unicode).
 -moduledoc """
-Functions for converting Unicode characters.
+Functions for converting and classifying Unicode characters.
 
 This module contains functions for converting between different character
 representations. It converts between ISO Latin-1 characters and Unicode
@@ -71,9 +71,12 @@ normalization can be found in the
          characters_to_nfkc_list/1, characters_to_nfkc_binary/1
         ]).
 
+-export([is_whitespace/1, is_ID_start/1, is_ID_continue/1, category/1]).
+
 -export_type([chardata/0, charlist/0, encoding/0, external_chardata/0,
               external_charlist/0, latin1_char/0, latin1_chardata/0,
-              latin1_charlist/0, latin1_binary/0, unicode_binary/0]).
+              latin1_charlist/0, latin1_binary/0, unicode_binary/0,
+              category/0]).
 
 -type encoding()  :: 'latin1' | 'unicode' | 'utf8'
                    | 'utf16' | {'utf16', endian()}
@@ -108,6 +111,38 @@ than UTF-8 (that is, UTF-16 or UTF-32).
                               latin1_binary() |
                               latin1_charlist(),
                             latin1_binary() | nil()).
+-doc "Character category".
+-type category() ::
+     {letter,uppercase} |
+     {letter,lowercase} |
+     {letter,titlecase} |
+     {letter,modifier} |
+     {letter,other} |
+     {mark,non_spacing} |
+     {mark,spacing_combining} |
+     {mark,enclosing} |
+     {number,decimal} |
+     {number,letter} |
+     {number,other} |
+     {separator,space} |
+     {separator,line} |
+     {separator,paragraph} |
+     {other,control} |
+     {other,format} |
+     {other,surrogate} |
+     {other,private} |
+     {other,not_assigned} |
+     {punctuation,connector} |
+     {punctuation,dash} |
+     {punctuation,open} |
+     {punctuation,close} |
+     {punctuation,initial} |
+     {punctuation,final} |
+     {punctuation,other} |
+     {symbol,math} |
+     {symbol,currency} |
+     {symbol,modifier} |
+     {symbol,other}.
 
 %% We must inline these functions so that the stacktrace points to
 %% the correct function.
@@ -122,6 +157,8 @@ than UTF-8 (that is, UTF-16 or UTF-32).
 
 -export([bin_is_7bit/1, characters_to_binary/2, characters_to_list/2]).
 
+-define(IS_CP(CP), is_integer(CP, 0, 16#10FFFF)).
+
 -doc false.
 -spec bin_is_7bit(Binary) -> boolean() when
       Binary :: binary().
@@ -681,13 +718,141 @@ characters_to_nfkc_binary(CD, N, Row, Acc) when N > 0 ->
 characters_to_nfkc_binary(CD, _, Row, Acc) ->
     characters_to_nfkc_binary(CD, ?GC_N, [], prepend_row_to_acc(Row, Acc)).
 
+-doc """
+Returns true if `Char` is a whitespace.
+
+Whitespace is defined in
+[Unicode Standard Annex #44](http://unicode.org/reports/tr44/).
+
+```erlang
+1> unicode:is_whitespace($\s).
+true
+2> unicode:is_whitespace($😊).
+false
+```
+""".
+-doc(#{since => ~"@OTP-19858@"}).
+-spec is_whitespace(char()) -> boolean().
+is_whitespace(Char) when is_integer(Char, 9, 13) ->
+    true;
+is_whitespace(32) ->
+    true;
+is_whitespace(133) ->
+    true;
+is_whitespace(160) ->
+    true;
+is_whitespace(Char) when is_integer(Char, 0, 5000) ->
+    false;
+is_whitespace(Char) when ?IS_CP(Char) ->
+    unicode_util:is_whitespace(Char);
+is_whitespace(Term) ->
+    badarg_with_info([Term]).
+
+
+-doc """
+Returns true if `Char` is an identifier start.
+
+Identifier start is defined by the ID_Start property in
+[Unicode Standard Annex #31](http://unicode.org/reports/tr31/).
+
+```erlang
+1> unicode:is_ID_start($a).
+true
+2> unicode:is_ID_start($_).
+false
+3> unicode:is_ID_start($-).
+false
+```
+""".
+-doc(#{since => ~"@OTP-19858@"}).
+-spec is_ID_start(char()) -> boolean().
+is_ID_start(Char) %% ASCII optimizations
+  when is_integer(Char, 16#41, 16#5A);
+       is_integer(Char, 16#61, 16#7A) ->
+    true;
+is_ID_start(Char) when is_integer(Char, 0, 127) ->
+    false;
+is_ID_start(Char) when ?IS_CP(Char) ->
+    case unicode_util:category(Char) of
+        {number,letter} -> true;
+        {letter,modifier} -> unicode_util:is_letter_not_pattern_syntax(Char);
+        {letter,_} -> true;
+        {_,_} -> unicode_util:is_other_id_start(Char)
+    end;
+is_ID_start(Term) ->
+    badarg_with_info([Term]).
+
+
+-doc """
+Returns true if `Char` is an identifier continuation.
+
+Identifier continuation is defined by the ID_Continue property in
+[Unicode Standard Annex #31](http://unicode.org/reports/tr31/).
+
+```erlang
+1> unicode:is_ID_continue($a).
+true
+2> unicode:is_ID_continue($_).
+true
+3> unicode:is_ID_continue($-).
+false
+```
+""".
+-doc(#{since => ~"@OTP-19858@"}).
+-spec is_ID_continue(char()) -> boolean().
+is_ID_continue(Char) %% ASCII optimizations
+  when is_integer(Char, 16#30, 16#39);
+       is_integer(Char, 16#41, 16#5A);
+       (Char =:= 16#5F);
+       is_integer(Char, 16#61, 16#7A) ->
+    true;
+is_ID_continue(Char) when is_integer(Char, 0, 127) ->
+    false;
+is_ID_continue(Char) when ?IS_CP(Char) ->
+    case unicode_util:category(Char) of
+        {punctuation, connector} -> true;
+        {mark,non_spacing} -> true;
+        {mark,spacing_combining} -> true;
+        {number,other} -> unicode_util:is_other_id_continue(Char);
+        {number,_} -> true;
+        {letter,modifier} -> unicode_util:is_letter_not_pattern_syntax(Char);
+        {letter,_} -> true;
+        {_,_} -> unicode_util:is_other_id_start(Char) orelse
+                     unicode_util:is_other_id_continue(Char)
+    end;
+is_ID_continue(Term) ->
+    badarg_with_info([Term]).
+
+-doc """
+Returns the `Char` category.
+
+```erlang
+1> unicode:category($a).
+{letter,lowercase}
+2> unicode:category($Ä).
+{letter,uppercase}
+3> unicode:category($😊).
+{symbol,other}
+3> unicode:category($€).
+{symbol,currency}
+3> unicode:category($[).
+{punctuation,open}
+```
+""".
+-doc(#{since => ~"@OTP-19858@"}).
+-spec category(char()) -> category().
+category(Char) when ?IS_CP(Char) ->
+    unicode_util:category(Char);
+category(Term) ->
+    badarg_with_info([Term]).
+
+%% internals
+
 acc_to_binary(Acc) ->
     list_to_binary(lists:reverse(Acc)).
 prepend_row_to_acc(Row, Acc) ->
     [characters_to_binary(lists:reverse(Row))|Acc].
 
-%% internals
-
 -doc false.
 characters_to_list_int(ML, Encoding) ->
     try

lib/stdlib/test/Makefile

@@ -197,6 +197,9 @@ release_tests_spec: make_emakefile
 		$(ERL_FILES) $(COVERFILE) $(EXTRA_FILES) "$(RELSYSDIR)"
 	chmod -R u+w "$(RELSYSDIR)"
 	@tar cf - *_SUITE_data property_test | (cd "$(RELSYSDIR)"; tar xf -)
+	$(INSTALL_DIR) "$(RELSYSDIR)/unicode_SUITE_data"
+	$(INSTALL_DATA) ../uc_spec/PropList.txt ../uc_spec/DerivedCoreProperties.txt \
+	       "$(RELSYSDIR)/unicode_SUITE_data"
 	$(INSTALL_DIR) "$(RELSYSDIR)/stdlib_SUITE_data"
 	$(INSTALL_DATA) $(ERL_TOP)/make/otp_version_tickets "$(RELSYSDIR)/stdlib_SUITE_data"