From 77f06521f497b70c6bcd0d26618f65375e012aaf Mon Sep 17 00:00:00 2001
From: victor felder <victor@draft.li>
Date: Tue, 19 May 2026 09:13:38 +0200
Subject: [PATCH] fix(zarith): raise ArgumentError on malformed input

Now consume/1 validates upfront (non-empty, even-length hex) and
find_int + read_next raise on truncation (continuation bit set with
no following byte). read/1 now branches on the halt bit explicitly
so single-byte zarith inputs still terminate cleanly.
---
 lib/zarith.ex        | 24 +++++++++++++++---------
 test/zarith_test.exs | 24 +++++++++++++++++++++++-
 2 files changed, 38 insertions(+), 10 deletions(-)

diff --git a/lib/zarith.ex b/lib/zarith.ex
index 38882ed..4a80a67 100644
--- a/lib/zarith.ex
+++ b/lib/zarith.ex
@@ -164,9 +164,13 @@ defmodule Tezex.Zarith do
   """
   @spec consume(zarith_hex()) :: {%{int: String.t()}, pos_integer()}
   def consume(binary_input) when is_binary(binary_input) do
+    if binary_input == "" or rem(byte_size(binary_input), 2) != 0 do
+      raise ArgumentError, "Zarith input must be a non-empty even-length hex string"
+    end
+
     {carved_int, rest} = find_int(binary_input)
 
-    consumed = String.length(binary_input) - String.length(rest)
+    consumed = byte_size(binary_input) - byte_size(rest)
 
     binary =
       carved_int
@@ -194,16 +198,18 @@ defmodule Tezex.Zarith do
     end
   end
 
-  defp find_int("", acc) do
-    {Enum.reverse(acc) |> Enum.join(""), ""}
+  defp find_int("", _acc) do
+    raise ArgumentError, "Zarith input truncated: continuation bit set with no following byte"
   end
 
-  defp read([<<_halt::1, sign::1, tail::integer-size(6)>> | rest]) do
-    bits =
-      for bits_part <- read_next(rest, [<<tail::integer-size(6)>>]), into: <<>> do
-        bits_part
+  defp read([<<halt::1, sign::1, tail::integer-size(6)>> | rest]) do
+    parts =
+      case halt do
+        0 -> [<<tail::integer-size(6)>>]
+        1 -> read_next(rest, [<<tail::integer-size(6)>>])
       end
 
+    bits = for p <- parts, into: <<>>, do: p
     bits_count = bit_size(bits)
     <<integer::integer-size(bits_count)>> = bits
 
@@ -221,8 +227,8 @@ defmodule Tezex.Zarith do
     [<<tail::integer-size(7)>> | acc]
   end
 
-  defp read_next(_, acc) do
-    acc
+  defp read_next([], _acc) do
+    raise ArgumentError, "Zarith input truncated: continuation bit set with no following byte"
   end
 
   defp hex_to_dec(hex) do
diff --git a/test/zarith_test.exs b/test/zarith_test.exs
index eec5b1e..9c0f852 100644
--- a/test/zarith_test.exs
+++ b/test/zarith_test.exs
@@ -6,7 +6,8 @@ defmodule Tezex.ZarithTest do
 
   test "encode/decode" do
     n = 1_000_000_000_000
-    random = fn -> trunc(:rand.uniform(n) - n / 2) end
+    h = n / 2
+    random = fn -> trunc(:rand.uniform(n) - h) end
 
     Enum.each(1..5000, fn _ ->
       number = random.()
@@ -72,4 +73,25 @@ defmodule Tezex.ZarithTest do
       assert {%{int: "-610913435200"}, byte_size(packed)} == Zarith.consume(packed)
     end
   end
+
+  describe "malformed input" do
+    test "raises on truncated input with continuation bit set" do
+      # 0x80: continuation=1, sign=0, tail=0, claims more bytes follow but none do
+      assert_raise ArgumentError, fn -> Zarith.decode("80") end
+      # 0xff: continuation=1, sign=1, tail=63, same truncation
+      assert_raise ArgumentError, fn -> Zarith.decode("ff") end
+      # Multi-byte truncation: first byte continues, second byte continues, then ends
+      assert_raise ArgumentError, fn -> Zarith.decode("8080") end
+    end
+
+    test "raises on empty input" do
+      assert_raise ArgumentError, fn -> Zarith.consume("") end
+      assert_raise ArgumentError, fn -> Zarith.decode("") end
+    end
+
+    test "raises on odd-length hex input" do
+      assert_raise ArgumentError, fn -> Zarith.consume("a1d22") end
+      assert_raise ArgumentError, fn -> Zarith.decode("a1d22") end
+    end
+  end
 end