fix: correctly encode and sign block ids

dkataskin · May 20, 2024 · 4495fba · 4495fba
1 parent 4b072a0
commit 4495fba
Show file tree

Hide file tree

Showing 3 changed files with 50 additions and 2 deletions.
diff --git a/src/erlazure.erl b/src/erlazure.erl
@@ -667,7 +667,7 @@ handle_call({delete_blob, Container, Blob, Options}, _From, State) ->
 handle_call({put_block, Container, Blob, BlockId, Content, Options}, _From, State) ->
         ServiceContext = new_service_context(?blob_service, State),
         Params = [{comp, block},
-                  {blob_block_id, base64:encode_to_string(BlockId)}],
+                  {blob_block_id, uri_string:quote(base64:encode_to_string(BlockId))}],
         ReqOptions = [{method, put},
                       {path, lists:concat([Container, "/", Blob])},
                       {body, Content},
@@ -938,6 +938,11 @@ combine_canonical_param({Param, Value}, _PreviousParam, Acc, ParamList) ->
         [H | T] = ParamList,
         combine_canonical_param(H, Param, add_param_value(Param, Value, Acc), T).
 
+add_param_value(Param, Value, Acc) ->
+        %% special case: `blockid' must be URL-encoded when sending the request, but not
+        %% when signing it.  At this point, we've already encoded it.
+        Acc ++ "\n" ++ string:to_lower(Param) ++ ":" ++ uri_string:unquote(Value);
+
 add_param_value(Param, Value, Acc) ->
         Acc ++ "\n" ++ string:to_lower(Param) ++ ":" ++ Value.
 

diff --git a/src/erlazure_blob.erl b/src/erlazure_blob.erl
@@ -188,7 +188,11 @@ get_request_body(BlockRefs) ->
         FoldFun = fun(BlockRef=#blob_block{}, Acc) ->
                       [{block_type_to_node(BlockRef#blob_block.type),
                         [],
-                        [base64:encode_to_string(BlockRef#blob_block.id)]} | Acc]
+                        [base64:encode_to_string(BlockRef#blob_block.id)]} | Acc];
+                     ({BlockId, BlockType}, Acc) ->
+                      [{block_type_to_node(BlockType),
+                        [],
+                        [base64:encode_to_string(BlockId)]} | Acc]
                   end,
         Data = {'BlockList', [], lists:reverse(lists:foldl(FoldFun, [], BlockRefs))},
         lists:flatten(xmerl:export_simple([Data], xmerl_xml)).

diff --git a/test/erlazure_SUITE.erl b/test/erlazure_SUITE.erl
@@ -166,3 +166,42 @@ t_blob_failure_to_connect(_Config) ->
     ?assertMatch({error, {failed_connect, _}}, erlazure:append_block(Pid, "c", "b1", <<"a">>)),
     ?assertMatch({error, {failed_connect, _}}, erlazure:get_blob(Pid, "c", "b1")),
     ok.
+
+%% Basic smoke test for block blob storage operations.
+t_put_block(Config) ->
+    Endpoint = ?config(endpoint, Config),
+    {ok, Pid} = erlazure:start(#{account => ?ACCOUNT, key => ?KEY, endpoint => Endpoint}),
+    %% Create a container
+    Container = container_name(?FUNCTION_NAME),
+    ?assertMatch({[], _}, erlazure:list_containers(Pid)),
+    ?assertMatch({ok, created}, erlazure:create_container(Pid, Container)),
+    %% Upload some blocks
+    Opts = [{content_type, "text/csv"}],
+    BlobName = "blob1",
+    ?assertMatch({ok, created}, erlazure:put_block_blob(Pid, Container, BlobName, <<>>, Opts)),
+    %% Note: this short name is important for this test.  It'll produce a base64 string
+    %% that's padded.  That padding must be URL-encoded when sending the request, but not
+    %% when generating the string to sign.
+    BlockId1 = <<"blo1">>,
+    ?assertMatch({ok, created}, erlazure:put_block(Pid, Container, BlobName, BlockId1, <<"a">>)),
+    %% Testing iolists
+    BlockId2 = <<"blo2">>,
+    ?assertMatch({ok, created}, erlazure:put_block(Pid, Container, BlobName, BlockId2, [<<"\n">>, ["b", [$\n]]])),
+    %% Not yet committed.
+    ?assertMatch({ok, <<"">>}, erlazure:get_blob(Pid, Container, BlobName)),
+    %% Committing
+    BlockList1 = [{BlockId1, latest}],
+    ?assertMatch({ok, created}, erlazure:put_block_list(Pid, Container, BlobName, BlockList1)),
+    %% Committed only first block.
+    ?assertMatch({ok, <<"a">>}, erlazure:get_blob(Pid, Container, BlobName)),
+    %% Block 2 was dropped after committing.
+    ?assertMatch({[#blob_block{id = "blo1"}], _}, erlazure:get_block_list(Pid, Container, BlobName)),
+    BlockId3 = <<"blo3">>,
+    ?assertMatch({ok, created}, erlazure:put_block(Pid, Container, BlobName, BlockId3, [<<"\n">>, ["b", [$\n]]])),
+    %% Commit both blocks
+    BlockList2 = [{BlockId1, committed}, {BlockId3, uncommitted}],
+    ?assertMatch({ok, created}, erlazure:put_block_list(Pid, Container, BlobName, BlockList2)),
+    ?assertMatch({ok, <<"a\nb\n">>}, erlazure:get_blob(Pid, Container, BlobName)),
+    %% Delete container
+    ?assertMatch({ok, deleted}, erlazure:delete_container(Pid, Container)),
+    ok.