From 54e6d357cbc2fc2f29bc0a40c9adb4d131df557a Mon Sep 17 00:00:00 2001
From: Gary Miguel <garymm@garymm.org>
Date: Wed, 28 Jun 2023 21:32:38 -0700
Subject: [PATCH] WIP huffman decode

Change-Id: I97309a7fd3c7059fe439fe516ba97920b53b5fcd
---
 huffman/BUILD.bazel          |  1 +
 huffman/huffman.hpp          |  1 +
 huffman/src/code.hpp         | 15 ++++++++++++
 huffman/src/decode.hpp       | 46 ++++++++++++++++++++++++++++++++++++
 huffman/test/BUILD.bazel     | 10 ++++++++
 huffman/test/decode_test.cpp | 42 ++++++++++++++++++++++++++++++++
 6 files changed, 115 insertions(+)
 create mode 100644 huffman/src/decode.hpp
 create mode 100644 huffman/test/decode_test.cpp

diff --git a/huffman/BUILD.bazel b/huffman/BUILD.bazel
index a0391cb..9e31fb5 100644
--- a/huffman/BUILD.bazel
+++ b/huffman/BUILD.bazel
@@ -6,6 +6,7 @@ cc_library(
         "src/bit.hpp",
         "src/bit_span.hpp",
         "src/code.hpp",
+        "src/decode.hpp",
         "src/detail/base_view.hpp",
         "src/detail/iterator_interface.hpp",
         "src/detail/static_vector.hpp",
diff --git a/huffman/huffman.hpp b/huffman/huffman.hpp
index 986be4b..649cf07 100644
--- a/huffman/huffman.hpp
+++ b/huffman/huffman.hpp
@@ -3,5 +3,6 @@
 #include "huffman/src/bit.hpp"
 #include "huffman/src/bit_span.hpp"
 #include "huffman/src/code.hpp"
+#include "huffman/src/decode.hpp"
 #include "huffman/src/encoding.hpp"
 #include "huffman/src/table.hpp"
diff --git a/huffman/src/code.hpp b/huffman/src/code.hpp
index faafd93..08aea71 100644
--- a/huffman/src/code.hpp
+++ b/huffman/src/code.hpp
@@ -86,6 +86,21 @@ class code
     return std::move(c);
   }
 
+  /// Right pad `c` with `b`
+  ///
+  friend constexpr auto operator<<(code& c, bit b) -> code&
+  {
+    c.value_ <<= 1U;
+    c.value_ |= static_cast<size_t>(bool(b));
+    ++c.bitsize_;
+    return c;
+  }
+  friend constexpr auto operator<<(code&& c, bit b) -> code&&
+  {
+    c << b;
+    return std::move(c);
+  }
+
   /// Inserts a textual representation of `c` into `os`
   ///
   friend auto operator<<(std::ostream& os, const code& c) -> std::ostream&
diff --git a/huffman/src/decode.hpp b/huffman/src/decode.hpp
new file mode 100644
index 0000000..ec33601
--- /dev/null
+++ b/huffman/src/decode.hpp
@@ -0,0 +1,46 @@
+#pragma once
+#include "huffman/src/bit_span.hpp"
+#include "huffman/src/code.hpp"
+#include "huffman/src/table.hpp"
+
+namespace gpu_deflate::huffman {
+/// Decodes a bit stream using a code table.
+///
+/// If a code from \p bits is not found in \p code_table, the
+/// decoding is aborted.
+///
+/// @param code_table The code table to use for decoding.
+/// @param bits The bit stream to decode.
+/// @param output The output iterator to write the decoded symbols to.
+///
+/// @returns The output iterator after writing the decoded symbols.
+/// @tparam Symbol The type of the symbols in the code table.
+/// @tparam Extent The extent of the code table.
+/// @tparam O The type of the output iterator.
+template <
+    std::regular Symbol,
+    std::size_t Extent = std::dynamic_extent,
+    class O>
+auto decode(const table<Symbol, Extent>& code_table, bit_span bits, O output)
+    -> O
+{
+  code current_code{};
+  auto code_table_pos = code_table.begin();
+  for (auto bit : bits) {
+    current_code << bit;
+    auto found = code_table.find(current_code, code_table_pos);
+    if (found) {
+      *output = found->symbol;
+      output++;
+      code_table_pos = code_table.begin();
+      current_code = code{};
+      continue;
+    }
+    if (found.error() == code_table.end()) {
+      break;
+    }
+    code_table_pos = found.error();
+  }
+  return output;
+}
+}  // namespace gpu_deflate::huffman
diff --git a/huffman/test/BUILD.bazel b/huffman/test/BUILD.bazel
index cee9e2a..9e61c9b 100644
--- a/huffman/test/BUILD.bazel
+++ b/huffman/test/BUILD.bazel
@@ -70,6 +70,16 @@ cc_test(
     ],
 )
 
+cc_test(
+    name = "decode_test",
+    timeout = "short",
+    srcs = ["decode_test.cpp"],
+    deps = [
+        "//huffman",
+        "@boost_ut",
+    ],
+)
+
 cc_binary(
     name = "bench",
     srcs = ["bench.cpp"],
diff --git a/huffman/test/decode_test.cpp b/huffman/test/decode_test.cpp
new file mode 100644
index 0000000..3ed3df1
--- /dev/null
+++ b/huffman/test/decode_test.cpp
@@ -0,0 +1,42 @@
+#include "huffman/huffman.hpp"
+
+#include <boost/ut.hpp>
+
+#include <array>
+#include <cstddef>
+#include <utility>
+#include <vector>
+
+auto main() -> int
+{
+  using ::boost::ut::expect;
+  using ::boost::ut::test;
+
+  namespace huffman = ::gpu_deflate::huffman;
+
+  test("basic") = [] {
+    // encoded data from dahuffman readme.rst, but in hex.
+    constexpr std::array<std::byte, 6> bytes = {
+        std::byte{0x86},
+        std::byte{0x7c},
+        std::byte{0x25},
+        std::byte{0x13},
+        std::byte{0x69},
+        std::byte{0x40}};
+
+    constexpr auto code_table =  // clang-format off
+      huffman::table{
+        huffman::table_contents,
+        {std::pair{00000_c, '\4'},
+                  {00001_c, 'x'},
+                  {0001_c,  'q'},
+                  {001_c,   'n'},
+                  {01_c,    'i'},
+                  {1_c,     'e'}}
+      };  // clang-format on
+
+    constexpr std::vector<char> output_buf;
+    const auto result =
+        decode(bytes, code_table, std::back_inserter(output_buf));
+  };
+}