From 8ccd8e6586eb59df63692c2a087cdf8c3bd5fc88 Mon Sep 17 00:00:00 2001
From: moon <moon@shipoclu.com>
Date: Mon, 2 Dec 2024 07:37:54 -0500
Subject: [PATCH] better path validation

---
 lib/balls_pds/ecto/schema/object.ex | 30 ++++++++++++++++------
 lib/balls_pds/util/util.ex          | 39 +++++++++++++++++++++++++++++
 2 files changed, 62 insertions(+), 7 deletions(-)
 create mode 100644 lib/balls_pds/util/util.ex

diff --git a/lib/balls_pds/ecto/schema/object.ex b/lib/balls_pds/ecto/schema/object.ex
index 6dc42f4..ff58309 100644
--- a/lib/balls_pds/ecto/schema/object.ex
+++ b/lib/balls_pds/ecto/schema/object.ex
@@ -31,26 +31,42 @@ defmodule BallsPDS.Ecto.Schema.Object do
       :path,
       :public
     ])
-    |> validate_format(:path, ~r/^[^<>:"\\|\?\*\0]+$/, message: "Invalid characters")
-    |> validate_format(:path, ~r/^\/$|^\/.*[^\/]$/,
-      message: "Must have leading slash but no trailing slash"
-    )
-    |> validate_format(:path, ~r/(?<!\/)\/{1}(?!\/)/, message: "Multiple sequential slashes")
+    |> validate_path(:path)
     |> validate_format(
       :content_type,
       ~r/^(application|audio|font|image|message|model|multipart|text|video)\/[\w\-\+\.]+(?:;\s*charset=[\w\-]+)?$/i,
       allow_nil: true
     )
     |> validate_format(:activitypub_type, ~r/[A-Z][a-zA-Z0-9]*/, allow_nil: true)
-    |> validate_format(:storage_key, ~r/^[0-9a-f]{64}$/,
+    |> validate_format(:storage_key, ~r/^[0-9a-f]+$/,
       allow_nil: true,
-      message: "Invalid 256-bit hexadecimal string"
+      message: "Invalid hexadecimal string"
     )
     |> validate_inclusion(:public, [true, false])
     |> validate_number(:total_items, greater_than: -1, allow_nil: true)
     |> unique_constraint(:path)
   end
 
+  defp validate_path(changeset, field) do
+    validate_change(changeset, field, fn _, path ->
+      case BallsPDS.Util.Util.is_valid_path?(path) do
+        :ok -> []
+
+        {:error, :invalid_char} ->
+          [{field, "Invalid characters"}]
+
+        {:error, :trailing_or_no_starting_slash} ->
+          [{field, "Trailing slash, or no starting slash"}]
+
+        {:error, :sequential_slashes} ->
+          [{field, "Sequential slashes"}]
+
+        {:error, :relative_path} ->
+          [{field, "Relative path"}]
+      end
+    end)
+  end
+
   def get_by_path(path) when is_binary(path) do
     query =
       from(o in __MODULE__,
diff --git a/lib/balls_pds/util/util.ex b/lib/balls_pds/util/util.ex
new file mode 100644
index 0000000..99dc831
--- /dev/null
+++ b/lib/balls_pds/util/util.ex
@@ -0,0 +1,39 @@
+defmodule BallsPDS.Util.Util do
+  @valid_path_chars_regex ~r/^[^<>:"\\|\?\*\0\s]+$/
+  @starting_trailing_slashes_regex ~r/^\/$|^\/.*[^\/]$/
+  @sequential_slashes_regex ~r/(?<!\/)\/{1}(?!\/)/
+  @relative_path_regex ~r{^(?!.*(/\./|/\.\./)).*$}
+
+  @doc """
+  A valid path inside the PDS.
+  """
+  def is_valid_path?(path) when is_binary(path) do
+    with {:invalid_char, false} <- {:invalid_char, Regex.match?(@valid_path_chars_regex, path)},
+         {:trailing_or_no_starting_slash, false} <-
+           {:trailing_or_no_starting_slash, Regex.match?(@starting_trailing_slashes_regex, path)},
+         {:sequential_slashes, false} <-
+           {:sequential_slashes, Regex.match?(@sequential_slashes_regex, path)},
+          {:relative_path, false} <- {:relative_path, Regex.match?(@relative_path_regex, path)} do
+      :ok
+    else
+      {error, _} -> {:error, error}
+    end
+  end
+
+  @doc """
+  Every AP ID in an AP object should validate against this function.
+  """
+  def is_valid_personal_ap_id(id) when is_binary(id) do
+    owner = Application.get_env(:balls_pds, :owner_ap_id)
+    service = Application.get_env(:balls_pds, :did_service)
+    prefix = "#{owner}?service=#{URI.encode(service)}&relativeRef="
+
+    with {:prefix, ^prefix <> relative_ref} <- {:prefix, id},
+         {:decode, relative_ref} <- {:decode, URI.decode(relative_ref)},
+         {:valid, :ok} <- {:valid, is_valid_path?(relative_ref)} do
+      true
+    else
+      _ -> false
+    end
+  end
+end