edoc: Fix EDoc regarding Unicode atoms

Also extended the scanner to handle the \x{...} syntax.

Capitalizing now works with a few more characters.

3 files changed, 99 insertions, 2 deletions

diff --git a/lib/edoc/test/edoc_SUITE.erl b/lib/edoc/test/edoc_SUITE.erl
index 4d846ad63d..29ca9d1203 100644
--- a/lib/edoc/test/edoc_SUITE.erl
+++ b/lib/edoc/test/edoc_SUITE.erl
@@ -23,12 +23,13 @@
 	 init_per_group/2,end_per_group/2]).
 
 %% Test cases
--export([app/1,appup/1,build_std/1,build_map_module/1,otp_12008/1, build_app/1]).
+-export([app/1,appup/1,build_std/1,build_map_module/1,otp_12008/1,
+         build_app/1, otp_14285/1]).
 
 suite() -> [{ct_hooks,[ts_install_cth]}].
 
 all() -> 
-    [app,appup,build_std,build_map_module,otp_12008, build_app].
+    [app,appup,build_std,build_map_module,otp_12008, build_app, otp_14285].
 
 groups() -> 
     [].
@@ -113,3 +114,18 @@ build_app(Config) ->
 	true = filelib:is_regular(filename:join(OutDir, "a.html")),
 	true = filelib:is_regular(filename:join(OutDir, "b.html")),
 	ok.
+
+otp_14285(Config) ->
+    DataDir  = ?config(data_dir, Config),
+    PrivDir  = ?config(priv_dir, Config),
+    Un1 = filename:join(DataDir, "un_atom1.erl"),
+    Un2 = filename:join(DataDir, "un_atom2.erl"),
+    %% epp_dodger
+    Opts1 = [{dir, PrivDir}],
+    ok = edoc:files([Un1], Opts1),
+    ok = edoc:files([Un2], Opts1),
+    %% epp
+    Opts2 = [{preprocess, true}, {dir, PrivDir}],
+    ok = edoc:files([Un1], Opts2),
+    ok = edoc:files([Un2], Opts2),
+    ok.
diff --git a/lib/edoc/test/edoc_SUITE_data/un_atom1.erl b/lib/edoc/test/edoc_SUITE_data/un_atom1.erl
new file mode 100644
index 0000000000..20ca50d5d2
--- /dev/null
+++ b/lib/edoc/test/edoc_SUITE_data/un_atom1.erl
@@ -0,0 +1,41 @@
+%% coding:latin-1
+
+-module(un_atom1).
+
+-export(['\x{aaa}memory'/0, 'func-\x{400}'/1, func/1, �func/1]).
+
+-record('rec-\x{400}', {'field-\x{400}'}).
+
+-type cs() :: $\x{a}
+            | $\x{aa}
+            | $\x{aaa}
+            | $\xaa.
+
+-callback 'callback-\x{400}'() -> 'apa'.
+
+-type 'type-\x{400}'() :: 'atom-\x{400}'
+                        | cs()
+                        | #'rec-\x{400}'{'field-\x{400}' :: 'type-\x{400}'()}.
+
+-spec '\x{aaa}memory'() -> 'type-\x{400}'().
+
+'\x{aaa}memory'() ->
+    apa:foo().
+
+%% @deprecated Please use {@link m:f/1}.
+-spec 'func-\x{400}'(#'rec-\x{400}'{}) -> #'rec-\x{400}'{}.
+
+'func-\x{400}'(_T) ->
+    foo:bar(#'rec-\x{400}'{}).
+
+-record(rec, {}).
+
+-spec func(#rec{}) -> #rec{}.
+
+func(#rec{}) -> #rec{}.
+
+-record(�rec, {}).
+
+-spec �func(#�rec{}) -> #�rec{}.
+
+�func(#�rec{}) -> #�rec{}.
diff --git a/lib/edoc/test/edoc_SUITE_data/un_atom2.erl b/lib/edoc/test/edoc_SUITE_data/un_atom2.erl
new file mode 100644
index 0000000000..66c83e30e0
--- /dev/null
+++ b/lib/edoc/test/edoc_SUITE_data/un_atom2.erl
@@ -0,0 +1,40 @@
+%% coding:utf-8
+-module(un_atom2).
+
+-export(['\x{aaa}memory'/0, 'func-\x{400}'/1, func/1, äfunc/1]).
+
+-record('rec-\x{400}', {'field-\x{400}'}).
+
+-type cs() :: $\x{a}
+            | $\x{aa}
+            | $\x{aaa}
+            | $\xaa.
+
+-callback 'callback-\x{400}'() -> 'apa'.
+
+-type 'type-\x{400}'() :: 'atom-\x{400}'
+                        | cs()
+                        | #'rec-\x{400}'{'field-\x{400}' :: 'type-\x{400}'()}.
+
+-spec '\x{aaa}memory'() -> 'type-\x{400}'().
+
+'\x{aaa}memory'() ->
+    apa:foo().
+
+%% @deprecated Please use {@link m:f/1}.
+-spec 'func-\x{400}'(#'rec-\x{400}'{}) -> #'rec-\x{400}'{}.
+
+'func-\x{400}'(_T) ->
+    foo:bar(#'rec-\x{400}'{}).
+
+-record(rec, {}).
+
+-spec func(#rec{}) -> #rec{}.
+
+func(#rec{}) -> #rec{}.
+
+-record(ärec, {}).
+
+-spec äfunc(#ärec{}) -> #ärec{}.
+
+äfunc(#ärec{}) -> #ärec{}.