Proper custom pkg index support, some tests

- The rebar package index files have been moved off the default path and
  will require a new `rebar3 update`
- Caching of downloaded packages automatically takes place in a path
  relative to the CDN used
- The cache path is not shared with hex as we now write and modify data
  in there arbitrarily
- Basic tests plus the working set for more of them is included

src/rebar.hrl

@@ -22,6 +22,7 @@
 -define(DEFAULT_TEST_DEPS_DIR, "test/lib").
 -define(DEFAULT_RELEASE_DIR, "rel").
 -define(DEFAULT_CONFIG_FILE, "rebar.config").
+-define(DEFAULT_CDN, "https://s3.amazonaws.com/s3.hex.pm/tarballs").
 -define(LOCK_FILE, "rebar.lock").
 
 -ifdef(namespaced_types).

src/rebar_packages.erl

@@ -2,6 +2,7 @@
 
 -export([get_packages/1
         ,registry/1
+        ,package_dir/1
         ,check_registry/3
         ,registry_checksum/2
         ,find_highest_matching/3]).
@@ -16,8 +17,7 @@
 
 -spec get_packages(rebar_state:t()) -> {rebar_dict(), rebar_digraph()}.
 get_packages(State) ->
-    RebarDir = rebar_dir:global_cache_dir(State),
-    RegistryDir = filename:join(RebarDir, "packages"),
+    RegistryDir = package_dir(State),
     DictFile = filename:join(RegistryDir, "dict"),
     Edges = filename:join(RegistryDir, "edges"),
     Vertices = filename:join(RegistryDir, "vertices"),
@@ -43,8 +43,7 @@ get_packages(State) ->
     end.
 
 registry(State) ->
-    Dir = rebar_dir:global_cache_dir(State),
-    RegistryDir = filename:join(Dir, "packages"),
+    RegistryDir = package_dir(State),
     HexFile = filename:join(RegistryDir, "registry"),
     case ets:file2tab(HexFile) of
         {ok, T} ->
@@ -54,6 +53,17 @@ registry(State) ->
             error
     end.
 
+package_dir(State) ->
+    CacheDir = rebar_dir:global_cache_dir(State),
+    CDN = rebar_state:get(State, rebar_packages_cdn, ?DEFAULT_CDN),
+    {ok, {_, _, Host, _, Path, _}} = http_uri:parse(CDN),
+    CDNHostPath = lists:reverse(string:tokens(Host, ".")),
+    CDNPath = tl(filename:split(Path)),
+    PackageDir = filename:join([CacheDir, "hex"] ++ CDNHostPath ++ CDNPath ++ ["packages"]),
+    ok = filelib:ensure_dir(filename:join(PackageDir, "placeholder")),
+    PackageDir.
+
+
 check_registry(Pkg, Vsn, State) ->
     case rebar_state:registry(State) of
         {ok, T} ->

src/rebar_pkg_resource.erl

@@ -11,8 +11,6 @@
 
 -include("rebar.hrl").
 
--define(DEFAULT_CDN, "https://s3.amazonaws.com/s3.hex.pm/tarballs").
-
 lock(_AppDir, Source) ->
     Source.
 
@@ -27,7 +25,7 @@ needs_update(Dir, {pkg, _Name, Vsn}) ->
 
 download(TmpDir, Pkg={pkg, Name, Vsn}, State) ->
     CDN = rebar_state:get(State, rebar_packages_cdn, ?DEFAULT_CDN),
-    PackageDir = hex_package_dir(CDN, State),
+    PackageDir = rebar_packages:package_dir(State),
     Package = binary_to_list(<<Name/binary, "-", Vsn/binary, ".tar">>),
     CachePath = filename:join(PackageDir, Package),
     Url = string:join([CDN, Package], "/"),
@@ -53,10 +51,13 @@ serve_from_cache(TmpDir, CachePath, Pkg, State) ->
             ok = erl_tar:extract({binary, Contents}, [{cwd, TmpDir}, compressed]),
             {ok, true};
         {_Bin, Chk, Chk} ->
+            ?DEBUG("Checksums: registry: ~p, pkg: ~p", [Chk, _Bin]),
             {failed_extract, CachePath};
         {Chk, _Reg, Chk} ->
+            ?DEBUG("Checksums: registry: ~p, pkg: ~p", [_Reg, Chk]),
             {bad_registry_checksum, CachePath};
         {_Bin, _Reg, _Tar} ->
+            ?DEBUG("Checksums: registry: ~p, pkg: ~p, meta: ~p", [_Reg, _Bin, _Tar]),
             {bad_checksum, CachePath}
     end.
 
@@ -67,7 +68,7 @@ serve_from_download(TmpDir, CachePath, Package, ETag, Binary, State) ->
         ETag ->
             serve_from_cache(TmpDir, CachePath, Package, State);
         FileETag ->
-            ?DEBUG("Download ETag ~s doesn't match returned ETag ~s", [ETag, FileETag]),
+            ?DEBUG("Downloaded file ~s ETag ~s doesn't match returned ETag ~s", [CachePath, ETag, FileETag]),
             {bad_download, CachePath}
     end.
 
@@ -84,28 +85,17 @@ checksums(Pkg, Files, Contents, Version, Meta, State) ->
     Blob = <<Version/binary, Meta/binary, Contents/binary>>,
     <<X:256/big-unsigned>> = crypto:hash(sha256, Blob),
     BinChecksum = list_to_binary(string:to_upper(lists:flatten(io_lib:format("~64.16.0b", [X])))),
-    RegistryChecksum = rebar_packages:registry_sum(Pkg, State),
+    RegistryChecksum = rebar_packages:registry_checksum(Pkg, State),
     {"CHECKSUM", TarChecksum} = lists:keyfind("CHECKSUM", 1, Files),
     {BinChecksum, RegistryChecksum, TarChecksum}.
 
 make_vsn(_) ->
     {error, "Replacing version of type pkg not supported."}.
 
-%% Use the shared hex package directory unless a non-default package repo is used
-hex_package_dir(?DEFAULT_CDN, _) ->
-    filename:join([rebar_dir:home_dir(), ".hex", "packages"]);
-hex_package_dir(CDN, State) ->
-    CacheDir = rebar_dir:global_cache_dir(State),
-    {ok, {_, _, Host, _, _, _}} = http_uri:parse(CDN),
-    CDNPath = filename:join(lists:reverse(string:tokens(Host, "."))),
-    PackageDir = filename:join([CacheDir, "hex", CDNPath, "packages"]),
-    ok = filelib:ensure_dir(filename:join(PackageDir, "placeholder")),
-    PackageDir.
-
 request(Url, ETag) ->
     case httpc:request(get, {Url, [{"if-none-match", ETag} || ETag =/= false]},
-                      [{relaxed, true}],
-                      [{body_format, binary}]) of
+                       [{relaxed, true}],
+                       [{body_format, binary}]) of
         {ok, {{_Version, 200, _Reason}, Headers, Body}} ->
             ?DEBUG("Successfully downloaded ~s", [Url]),
             {"etag", ETag1} = lists:keyfind("etag", 1, Headers),

src/rebar_prv_update.erl

@@ -35,8 +35,7 @@ init(State) ->
 do(State) ->
     ?INFO("Updating package index...", []),
     try
-        Dir = rebar_dir:global_cache_dir(State),
-        RegistryDir = filename:join(Dir, "packages"),
+        RegistryDir = rebar_packages:package_dir(State),
         filelib:ensure_dir(filename:join(RegistryDir, "dummy")),
         HexFile = filename:join(RegistryDir, "registry"),
         TmpDir = ec_file:insecure_mkdtemp(),
@@ -64,8 +63,7 @@ format_error(package_index_write) ->
     "Failed to write package index.".
 
 write_registry(Dict, {digraph, Edges, Vertices, Neighbors, _}, State) ->
-    Dir = rebar_dir:global_cache_dir(State),
-    RegistryDir = filename:join(Dir, "packages"),
+    RegistryDir = rebar_packages:package_dir(State),
     filelib:ensure_dir(filename:join(RegistryDir, "dummy")),
     ets:tab2file(Edges, filename:join(RegistryDir, "edges")),
     ets:tab2file(Vertices, filename:join(RegistryDir, "vertices")),

test/rebar_pkg_SUITE.erl

@@ -0,0 +1,100 @@
+%% Test suite for the rebar pkg index caching and decompression
+%% mechanisms.
+-module(rebar_pkg_SUITE).
+-compile(export_all).
+-include_lib("common_test/include/ct.hrl").
+-include_lib("eunit/include/eunit.hrl").
+
+-define(bad_etag, "abcdef").
+-define(good_etag, "22e1d7387c9085a462340088a2a8ba67").
+-define(bad_checksum, <<"D576B442A68C7B92BACDE1EFE9C6E54D8D6C74BDB71D8175B9D3C6EC8C7B62A7">>).
+-define(good_checksum, <<"1C6CE379D191FBAB41B7905075E0BF87CBBE23C77CECE775C5A0B786B2244C35">>).
+
+all() -> [good_uncached].
+
+
+init_per_suite(Config) ->
+    application:start(meck),
+    Config.
+
+end_per_suite(_Config) ->
+    application:stop(meck).
+
+init_per_testcase(good_uncached=Name, Config0) ->
+    Config = [{good_cache, false},
+              {pkg, {<<"goodpkg">>, <<"1.0.0">>}}
+             | Config0],
+    mock_config(Name, Config);
+init_per_testcase(good_cached=Name, Config0) ->
+    Pkg = {<<"goodpkg">>, <<"1.0.0">>},
+    Config1 = [{good_cache, true},
+               {pkg, Pkg}
+              | Config0],
+    Config = mock_config(Name, Config1),
+    copy_to_cache(Pkg, Config),
+    Config.
+
+
+end_per_testcase(_, Config) ->
+    unmock_config(Config),
+    Config.
+
+mock_config(Name, Config) ->
+    Priv = ?config(priv_dir, Config),
+    CacheRoot = filename:join([Priv, "cache", atom_to_list(Name)]),
+    TmpDir = filename:join([Priv, "tmp", atom_to_list(Name)]),
+    T = ets:new(fake_registry, [public]),
+    ets:insert_new(T, [
+        {{<<"badindexchk">>,<<"1.0.0">>}, [[], ?bad_checksum]},
+        {{<<"goodpkg">>,<<"1.0.0">>}, [[], ?good_checksum]},
+        {{<<"badpkg">>,<<"1.0.0">>}, [[], ?good_checksum]}
+    ]),
+    CacheDir = filename:join([CacheRoot, "hex", "com", "test", "packages"]),
+    filelib:ensure_dir(filename:join([CacheDir, "registry"])),
+    ok = ets:tab2file(T, filename:join([CacheDir, "registry"])),
+    %% The state returns us a fake registry
+    meck:new(rebar_state, [passthrough]),
+    meck:expect(rebar_state, registry,
+                fun(_State) -> {ok, fake_registry} end),
+    meck:expect(rebar_state, get,
+                fun(_State, rebar_packages_cdn, _Default) ->
+                    "http://test.com/"
+                end),
+    meck:new(rebar_dir, [passthrough]),
+    meck:expect(rebar_dir, global_cache_dir, fun(_) -> CacheRoot end),
+    %% Cache fetches are mocked -- we assume the server and clients are
+    %% correctly used.
+    GoodCache = ?config(good_cache, Config),
+    {Pkg,Vsn} = ?config(pkg, Config),
+    PkgFile = <<Pkg/binary, "-", Vsn/binary, ".tar">>,
+    {ok, PkgContents} = file:read_file(filename:join(?config(data_dir, Config), PkgFile)),
+    meck:new(httpc, [passthrough, unsticky]),
+    meck:expect(httpc, request,
+            fun(get, {_Url, _Opts}, _, _) when GoodCache ->
+                {ok, {{Vsn, 304, <<"Not Modified">>}, [{"etag", ?good_etag}], <<>>}};
+               (get, {_Url, _Opts}, _, _) ->
+                {ok, {{Vsn, 200, <<"OK">>}, [{"etag", ?good_etag}], PkgContents}}
+            end),
+    [{cache_root, CacheRoot},
+     {cache_dir, CacheDir},
+     {tmp_dir, TmpDir},
+     {mock_table, T} | Config].
+
+unmock_config(Config) ->
+    meck:unload(),
+    ets:delete(?config(mock_table, Config)).
+
+copy_to_cache({Pkg,Vsn}, Config) ->
+    Name = <<Pkg/binary, "-", Vsn/binary, ".tar">>,
+    Source = filename:join(?config(data_dir, Config), Name),
+    Dest = filename:join(?config(cache_dir, Config), Name),
+    ec_file:copy(Source, Dest).
+
+good_uncached(Config) ->
+    Tmp = ?config(tmp_dir, Config),
+    {Pkg,Vsn} = ?config(pkg, Config),
+    State = ?config(state, Config),
+    ?assertEqual({ok, true},
+                 rebar_pkg_resource:download(Tmp, {pkg, Pkg, Vsn}, State)),
+    Cache = ?config(cache_dir, Config),
+    ?assert(filelib:is_regular(filename:join(Cache, <<Pkg/binary, "-", Vsn/binary, ".tar">>))).