Merge pull request #13685 from obsidiansystems/hash-sentinal-encapsulation

Encapsulate `invalidBase32`, avoid 0xFF magic number

src/libstore-tests/ref-scan-bench.cc

@@ -1,5 +1,6 @@
 #include "nix/util/references.hh"
 #include "nix/store/path.hh"
+#include "nix/util/base-nix-32.hh"
 
 #include <benchmark/benchmark.h>
 
@@ -10,9 +11,9 @@ using namespace nix;
 template<typename OIt>
 static void randomReference(std::mt19937 & urng, OIt outIter)
 {
-    auto dist = std::uniform_int_distribution<std::size_t>(0, nix32Chars.size() - 1);
+    auto dist = std::uniform_int_distribution<std::size_t>(0, BaseNix32::characters.size() - 1);
     dist(urng);
-    std::generate_n(outIter, StorePath::HashLen, [&]() { return nix32Chars[dist(urng)]; });
+    std::generate_n(outIter, StorePath::HashLen, [&]() { return BaseNix32::characters[dist(urng)]; });
 }
 
 /**

src/libutil/base-nix-32.cc

@@ -0,0 +1,42 @@
+#include <cassert>
+
+#include "nix/util/base-nix-32.hh"
+
+namespace nix {
+
+constexpr const std::array<unsigned char, 256> BaseNix32::reverseMap = [] {
+    std::array<unsigned char, 256> map{};
+
+    for (size_t i = 0; i < map.size(); ++i)
+        map[i] = invalid; // invalid
+
+    for (unsigned char i = 0; i < 32; ++i)
+        map[static_cast<unsigned char>(characters[i])] = i;
+
+    return map;
+}();
+
+std::string BaseNix32::encode(std::span<const uint8_t> originalData)
+{
+    if (originalData.size() == 0)
+        return {};
+
+    size_t len = encodedLength(originalData.size());
+    assert(len);
+
+    std::string s;
+    s.reserve(len);
+
+    for (int n = (int) len - 1; n >= 0; n--) {
+        unsigned int b = n * 5;
+        unsigned int i = b / 8;
+        unsigned int j = b % 8;
+        unsigned char c =
+            (originalData.data()[i] >> j) | (i >= originalData.size() - 1 ? 0 : originalData.data()[i + 1] << (8 - j));
+        s.push_back(characters[c & 0x1f]);
+    }
+
+    return s;
+}
+
+} // namespace nix

src/libutil/hash.cc

@@ -11,6 +11,7 @@
 #include "nix/util/archive.hh"
 #include "nix/util/configuration.hh"
 #include "nix/util/split.hh"
+#include "nix/util/base-nix-32.hh"
 
 #include <sys/types.h>
 #include <sys/stat.h>
@@ -71,39 +72,10 @@ static std::string printHash16(const Hash & hash)
     return buf;
 }
 
-// omitted: E O U T
-constexpr char nix32Chars[] = "0123456789abcdfghijklmnpqrsvwxyz";
-
-constexpr const std::array<unsigned char, 256> reverseNix32Map = [] {
-    std::array<unsigned char, 256> map{};
-
-    for (size_t i = 0; i < map.size(); ++i)
-        map[i] = 0xFF; // invalid
-
-    for (unsigned char i = 0; i < 32; ++i)
-        map[static_cast<unsigned char>(nix32Chars[i])] = i;
-
-    return map;
-}();
-
 static std::string printHash32(const Hash & hash)
 {
     assert(hash.hashSize);
-    size_t len = hash.base32Len();
-    assert(len);
-
-    std::string s;
-    s.reserve(len);
-
-    for (int n = (int) len - 1; n >= 0; n--) {
-        unsigned int b = n * 5;
-        unsigned int i = b / 8;
-        unsigned int j = b % 8;
-        unsigned char c = (hash.hash[i] >> j) | (i >= hash.hashSize - 1 ? 0 : hash.hash[i + 1] << (8 - j));
-        s.push_back(nix32Chars[c & 0x1f]);
-    }
-
-    return s;
+    return BaseNix32::encode({&hash.hash[0], hash.hashSize});
 }
 
 std::string printHash16or32(const Hash & hash)
@@ -229,11 +201,13 @@ Hash::Hash(std::string_view rest, HashAlgorithm algo, bool isSRI)
 
         for (unsigned int n = 0; n < rest.size(); ++n) {
             char c = rest[rest.size() - n - 1];
-            unsigned char digit = reverseNix32Map[static_cast<unsigned char>(c)];
+            auto digit_opt = BaseNix32::lookupReverse(c);
 
-            if (digit == 0xFF)
+            if (!digit_opt)
                 throw BadHash("invalid base-32 hash: '%s'", rest);
 
+            uint8_t digit = std::move(*digit_opt);
+
             unsigned int b = n * 5;
             unsigned int i = b / 8;
             unsigned int j = b % 8;

src/libutil/include/nix/util/base-nix-32.hh

@@ -0,0 +1,45 @@
+#pragma once
+///@file
+
+#include <array>
+#include <cstdint>
+#include <optional>
+#include <string>
+#include <span>
+
+namespace nix {
+
+struct BaseNix32
+{
+    /// omitted: E O U T
+    constexpr static std::array<char, 32> characters = {'0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'a',
+                                                        'b', 'c', 'd', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm',
+                                                        'n', 'p', 'q', 'r', 's', 'v', 'w', 'x', 'y', 'z'};
+
+private:
+    static const std::array<uint8_t, 256> reverseMap;
+
+    const static constexpr uint8_t invalid = 0xFF;
+
+public:
+    static inline std::optional<uint8_t> lookupReverse(char base32)
+    {
+        uint8_t digit = reverseMap[static_cast<unsigned char>(base32)];
+        if (digit == invalid)
+            return std::nullopt;
+        else
+            return digit;
+    }
+
+    /**
+     * Returns the length of a base-32 representation of this hash.
+     */
+    static size_t encodedLength(size_t originalLength)
+    {
+        return (originalLength * 8 - 1) / 5 + 1;
+    }
+
+    static std::string encode(std::span<const uint8_t> originalData);
+};
+
+} // namespace nix

src/libutil/include/nix/util/hash.hh

@@ -35,8 +35,6 @@ constexpr inline size_t regularHashSize(HashAlgorithm type)
 
 extern const StringSet hashAlgorithms;
 
-extern const std::array<unsigned char, 256> reverseNix32Map;
-
 /**
  * @brief Enumeration representing the hash formats.
  */
@@ -44,7 +42,7 @@ enum struct HashFormat : int {
     /// @brief Base 64 encoding.
     /// @see [IETF RFC 4648, section 4](https://datatracker.ietf.org/doc/html/rfc4648#section-4).
     Base64,
-    /// @brief Nix-specific base-32 encoding. @see nix32Chars
+    /// @brief Nix-specific base-32 encoding. @see BaseNix32
     Nix32,
     /// @brief Lowercase hexadecimal encoding. @see base16Chars
     Base16,

src/libutil/include/nix/util/meson.build

@@ -8,6 +8,7 @@ headers = files(
   'archive.hh',
   'args.hh',
   'args/root.hh',
+  'base-nix-32.hh',
   'callback.hh',
   'canon-path.hh',
   'checked-arithmetic.hh',

src/libutil/meson.build

@@ -112,6 +112,7 @@ subdir('nix-meson-build-support/common')
 sources = [config_priv_h] + files(
   'archive.cc',
   'args.cc',
+  'base-nix-32.cc',
   'canon-path.cc',
   'compression.cc',
   'compute-levels.cc',

src/libutil/references.cc

@@ -1,6 +1,7 @@
 #include "nix/util/references.hh"
 #include "nix/util/hash.hh"
 #include "nix/util/archive.hh"
+#include "nix/util/base-nix-32.hh"
 
 #include <map>
 #include <cstdlib>
@@ -17,7 +18,7 @@ static void search(std::string_view s, StringSet & hashes, StringSet & seen)
         int j;
         bool match = true;
         for (j = refLength - 1; j >= 0; --j)
-            if (reverseNix32Map[(unsigned char) s[i + j]] == 0xFF) {
+            if (!BaseNix32::lookupReverse(s[i + j])) {
                 i += j + 1;
                 match = false;
                 break;