Patch wine & iris dataset requests (#40)

* change ssl verify option for iris, wine dataset requests

* compare wine, iris dataset hashes

lib/scidata/iris.ex

@@ -5,6 +5,8 @@ defmodule Scidata.Iris do
 
   @base_url "https://archive.ics.uci.edu/static/public/53/iris.zip"
   @dataset_file "iris.data"
+  @data_hash <<111, 96, 139, 113, 167, 49, 114, 22, 49, 155, 77, 39, 180, 217, 188, 132, 230,
+  171, 215, 52, 237, 167, 135, 43, 113, 164, 88, 86, 158, 38, 86, 192>>
 
   alias Scidata.Utils
 
@@ -46,6 +48,9 @@ defmodule Scidata.Iris do
     base_url = opts[:base_url] || @base_url
     dataset_file = opts[:dataset_file] || @dataset_file
 
+    # Temporary fix to cope with bad cert on source site
+    opts = Keyword.put(opts, :ssl_verify, :verify_none)
+
     [{_, data}] =
       Utils.get!(base_url, opts).body
       |> Enum.filter(fn {fname, _} ->
@@ -55,6 +60,10 @@ defmodule Scidata.Iris do
         )
       end)
 
+    if :crypto.hash(:sha256, data) != @data_hash do
+      raise RuntimeError, "Dataset hashed to unexpected value"
+    end
+
     data
     |> String.split()
     |> Enum.reverse()

lib/scidata/utils.ex

@@ -9,7 +9,7 @@ defmodule Scidata.Utils do
 
     request
     |> if_modified_since(opts)
-    |> run!()
+    |> run!(opts)
     |> raise_errors()
     |> handle_cache(opts)
     |> decode()
@@ -31,21 +31,23 @@ defmodule Scidata.Utils do
     Calendar.strftime(datetime, "%a, %d %b %Y %H:%m:%S GMT")
   end
 
-  defp run!(request) do
+  defp run!(request, opts) do
+    verify = opts[:ssl_verify] || :verify_peer
+
     http_opts = [
       ssl: [
-        verify: :verify_peer,
+        verify: verify,
         cacertfile: CAStore.file_path(),
         customize_hostname_check: [
           match_fun: :public_key.pkix_verify_hostname_match_fun(:https)
         ]
       ]
     ]
 
-    opts = [body_format: :binary]
+    request_opts = [body_format: :binary]
     arg = {request.url, request.headers}
 
-    case :httpc.request(:get, arg, http_opts, opts) do
+    case :httpc.request(:get, arg, http_opts, request_opts) do
       {:ok, {{_, status, _}, headers, body}} ->
         response = %{status: status, headers: headers, body: body}
         {request, response}

lib/scidata/wine.ex

@@ -5,6 +5,8 @@ defmodule Scidata.Wine do
 
   @base_url "https://archive.ics.uci.edu/static/public/109/wine.zip"
   @dataset_file "wine.data"
+  @data_hash <<107, 230, 177, 32, 63, 61, 81, 223, 11, 85, 58, 112, 229, 123, 138, 114, 60,
+  212, 5, 104, 57, 88, 32, 79, 150, 210, 61, 124, 214, 174, 166, 89>>
 
   alias Scidata.Utils
 
@@ -55,6 +57,9 @@ defmodule Scidata.Wine do
     base_url = opts[:base_url] || @base_url
     dataset_file = opts[:dataset_file] || @dataset_file
 
+    # Temporary fix to cope with bad cert on source site
+    opts = Keyword.put(opts, :ssl_verify, :verify_none)
+
     [{_, data}] =
       Utils.get!(base_url, opts).body
       |> Enum.filter(fn {fname, _} ->
@@ -64,6 +69,10 @@ defmodule Scidata.Wine do
         )
       end)
 
+    if :crypto.hash(:sha256, data) != @data_hash do
+      raise RuntimeError, "Dataset hashed to unexpected value"
+    end
+
     label_attr =
       data
       |> String.split()