From de3774d93b061a54d33c7c58f07c69cbbb36cadf Mon Sep 17 00:00:00 2001 From: Hans Bolinder Date: Wed, 25 Jun 2014 15:26:45 +0200 Subject: edoc, syntax_tools: Don't fail on invalid UTF-8 As a temporary measure to ease the transition to default UTF-8 encoding, automatically fall back to the Latin-1 encoding (without any warnings; the Erlang Compiler will emit a proper warning). The intention is to remove this workaround in OTP 18 or 19. --- lib/edoc/src/edoc.erl | 35 ++++++++++++++++++++++++++++++++--- lib/edoc/test/edoc_SUITE.erl | 22 ++++++++++++++++++++-- lib/edoc/test/edoc_SUITE_data/un1.erl | 7 +++++++ lib/edoc/test/edoc_SUITE_data/un2.erl | 8 ++++++++ lib/edoc/test/edoc_SUITE_data/un3.erl | 8 ++++++++ 5 files changed, 75 insertions(+), 5 deletions(-) create mode 100644 lib/edoc/test/edoc_SUITE_data/un1.erl create mode 100644 lib/edoc/test/edoc_SUITE_data/un2.erl create mode 100644 lib/edoc/test/edoc_SUITE_data/un3.erl (limited to 'lib/edoc') diff --git a/lib/edoc/src/edoc.erl b/lib/edoc/src/edoc.erl index a87a8471e3..983f04e8b6 100644 --- a/lib/edoc/src/edoc.erl +++ b/lib/edoc/src/edoc.erl @@ -696,15 +696,44 @@ read_source_2(Name, Opts) -> %% The line of the dot token will be copied to the integer token. parse_file(Name, Includes, Macros) -> - case epp:open(Name, Includes, Macros) of - {ok, Epp} -> - try {ok, parse_file(Epp)} + case parse_file(utf8, Name, Includes, Macros) of + invalid_unicode -> + parse_file(latin1, Name, Includes, Macros); + Ret -> + Ret + end. + +parse_file(DefEncoding, Name, Includes, Macros) -> + Options = [{name, Name}, + {includes, Includes}, + {macros, Macros}, + {default_encoding, DefEncoding}], + case epp:open([extra | Options]) of + {ok, Epp, Extra} -> + try parse_file(Epp) of + Forms -> + Encoding = proplists:get_value(encoding, Extra), + case find_invalid_unicode(Forms) of + invalid_unicode when Encoding =/= utf8 -> + invalid_unicode; + _ -> + {ok, Forms} + end after _ = epp:close(Epp) end; Error -> Error end. +find_invalid_unicode([H|T]) -> + case H of + {error,{_Line,file_io_server,invalid_unicode}} -> + invalid_unicode; + _Other -> + find_invalid_unicode(T) + end; +find_invalid_unicode([]) -> none. + parse_file(Epp) -> case scan_and_parse(Epp) of {ok, Form} -> diff --git a/lib/edoc/test/edoc_SUITE.erl b/lib/edoc/test/edoc_SUITE.erl index c9c7811afb..c63660c8c0 100644 --- a/lib/edoc/test/edoc_SUITE.erl +++ b/lib/edoc/test/edoc_SUITE.erl @@ -22,12 +22,12 @@ init_per_group/2,end_per_group/2]). %% Test cases --export([app/1,appup/1,build_std/1,build_map_module/1]). +-export([app/1,appup/1,build_std/1,build_map_module/1,otp_12008/1]). suite() -> [{ct_hooks,[ts_install_cth]}]. all() -> - [app,appup,build_std,build_map_module]. + [app,appup,build_std,build_map_module,otp_12008]. groups() -> []. @@ -77,3 +77,21 @@ build_map_module(Config) when is_list(Config) -> Filename = filename:join(DataDir, "map_module.erl"), ok = edoc:file(Filename, [{dir, PrivDir}]), ok. + +otp_12008(Config) when is_list(Config) -> + DataDir = ?config(data_dir, Config), + PrivDir = ?config(priv_dir, Config), + Un1 = filename:join(DataDir, "un1.erl"), + Un2 = filename:join(DataDir, "un2.erl"), + Un3 = filename:join(DataDir, "un3.erl"), + %% epp_dodger + Opts1 = [{dir, PrivDir}], + ok = edoc:files([Un1], Opts1), + ok = edoc:files([Un2], Opts1), + {'EXIT', error} = (catch edoc:files([Un3], Opts1)), + %% epp + Opts2 = [{preprocess, true}, {dir, PrivDir}], + ok = edoc:files([Un1], Opts2), + ok = edoc:files([Un2], Opts2), + {'EXIT', error} = (catch edoc:files([Un3], Opts2)), + ok. diff --git a/lib/edoc/test/edoc_SUITE_data/un1.erl b/lib/edoc/test/edoc_SUITE_data/un1.erl new file mode 100644 index 0000000000..0c48e7f940 --- /dev/null +++ b/lib/edoc/test/edoc_SUITE_data/un1.erl @@ -0,0 +1,7 @@ +-module(un1). + +-export([t/0]). + +%% @doc Fäpp +t() -> + ärlig. diff --git a/lib/edoc/test/edoc_SUITE_data/un2.erl b/lib/edoc/test/edoc_SUITE_data/un2.erl new file mode 100644 index 0000000000..a6d13f4723 --- /dev/null +++ b/lib/edoc/test/edoc_SUITE_data/un2.erl @@ -0,0 +1,8 @@ +-module(un2). +%% coding: latin-1 + +-export([t/0]). + +%% @doc Fäpp +t() -> + ärlig. diff --git a/lib/edoc/test/edoc_SUITE_data/un3.erl b/lib/edoc/test/edoc_SUITE_data/un3.erl new file mode 100644 index 0000000000..fbe9591dce --- /dev/null +++ b/lib/edoc/test/edoc_SUITE_data/un3.erl @@ -0,0 +1,8 @@ +-module(un3). +%% coding: utf-8 + +-export([t/0]). + +%% @doc Fäpp +t() -> + ärlig. -- cgit v1.2.3 From dba1d881f3232a939c6620f5fd6b4a97ff454bee Mon Sep 17 00:00:00 2001 From: Erlang/OTP Date: Tue, 1 Jul 2014 13:58:47 +0200 Subject: Prepare release --- lib/edoc/doc/src/notes.xml | 19 +++++++++++++++++++ lib/edoc/vsn.mk | 2 +- 2 files changed, 20 insertions(+), 1 deletion(-) (limited to 'lib/edoc') diff --git a/lib/edoc/doc/src/notes.xml b/lib/edoc/doc/src/notes.xml index d7cbfa1fdc..b3440ce6e1 100644 --- a/lib/edoc/doc/src/notes.xml +++ b/lib/edoc/doc/src/notes.xml @@ -31,6 +31,25 @@

This document describes the changes made to the EDoc application.

+
Edoc 0.7.14 + +
Fixed Bugs and Malfunctions + + +

The default encoding for Erlang source files is now + UTF-8. As a temporary measure to ease the transition from + the old default of Latin-1, if EDoc encounters byte + sequences that are not valid UTF-8 sequences, EDoc will + re-try in Latin-1 mode. This workaround will be removed + in a future release.

+

+ Own Id: OTP-12008

+ + +
+ +
+
Edoc 0.7.13
Fixed Bugs and Malfunctions diff --git a/lib/edoc/vsn.mk b/lib/edoc/vsn.mk index 0172aac48b..281a792118 100644 --- a/lib/edoc/vsn.mk +++ b/lib/edoc/vsn.mk @@ -1 +1 @@ -EDOC_VSN = 0.7.13 +EDOC_VSN = 0.7.14 -- cgit v1.2.3