feat: Levenshtein distance (#513)

Author: benthevining

libchess/src/uci/EngineBase.cpp

@@ -13,6 +13,7 @@
  */
 
 #include <algorithm>
+#include <array>
 #include <atomic>
 #include <cassert>
 #include <expected>
@@ -25,9 +26,11 @@
 #include <libchess/uci/Printing.hpp>
 #include <libutil/Strings.hpp>
 #include <print>
+#include <ranges>
 #include <string>
 #include <string_view>
 #include <utility>
+#include <vector>
 
 namespace chess::uci {
 
@@ -48,6 +51,29 @@ using util::strings::trim;
 // defined out-of-line to address -Wweak-vtables
 EngineBase::~EngineBase() = default;
 
+namespace {
+    // returns name of closest known command
+    [[nodiscard]] auto find_nearest_command(
+        const string_view input, const EngineBase::CommandList standardCommands, const EngineBase::CommandList customCommands)
+        -> string_view
+    {
+        // map commands to pair of: command name, Levenshtein distance from input
+        const auto mapped
+            = std::views::join(std::array { standardCommands, customCommands })
+            | std::views::transform([input](const EngineCommand& command) {
+                  return std::make_pair(
+                      command.name,
+                      util::strings::levenshtein_distance(input, command.name));
+              })
+            | std::ranges::to<std::vector>();
+
+        const auto closest = std::ranges::min(
+            mapped, std::ranges::less { }, [](const auto& item) { return item.second; });
+
+        return closest.first;
+    }
+} // namespace
+
 void EngineBase::handle_command(const string_view command)
 {
     auto [firstWord, rest] = split_at_first_space(command);
@@ -72,7 +98,12 @@ void EngineBase::handle_command(const string_view command)
         return;
     }
 
-    info_string(std::format("Unknown UCI command: '{}'", firstWord));
+    info_string(std::format(
+        "Unknown UCI command: '{}'", firstWord));
+
+    info_string(std::format(
+        "The closest known command is: {}",
+        find_nearest_command(firstWord, standardUCICommands, customCommands)));
 }
 
 void EngineBase::respond_to_uci()
@@ -154,6 +185,29 @@ void EngineBase::handle_setpos(const string_view arguments)
               });
 }
 
+namespace {
+    // returns name of closest known option
+    [[nodiscard]] auto find_nearest_option(
+        const string_view input, const EngineBase::OptionList standardOptions, const EngineBase::OptionList customOptions)
+        -> string_view
+    {
+        // map options to pair of: option name, Levenshtein distance from input
+        const auto mapped
+            = std::views::join(std::array { standardOptions, customOptions })
+            | std::views::transform([input](const Option* option) {
+                  return std::make_pair(
+                      option->get_name(),
+                      util::strings::levenshtein_distance(input, option->get_name()));
+              })
+            | std::ranges::to<std::vector>();
+
+        const auto closest = std::ranges::min(
+            mapped, std::ranges::less { }, [](const auto& item) { return item.second; });
+
+        return closest.first;
+    }
+} // namespace
+
 void EngineBase::handle_setoption(const string_view arguments)
 {
     auto [firstWord, rest] = split_at_first_space(arguments);
@@ -205,7 +259,12 @@ void EngineBase::handle_setoption(const string_view arguments)
     if (update_option(get_custom_uci_options()))
         return;
 
-    info_string(std::format("Attempted to set unknown option '{}'", name));
+    info_string(std::format(
+        "Attempted to set unknown option '{}'", name));
+
+    info_string(std::format(
+        "The closest known option is: {}",
+        find_nearest_option(name, standardUCIOptions, get_custom_uci_options())));
 }
 
 void EngineBase::loop()

libutil/include/libutil/Strings.hpp

@@ -130,6 +130,10 @@ void write_integer(
  */
 [[nodiscard]] auto words_view(string_view text);
 
+/** Computes the Levenshtein distance between the two strings. */
+[[nodiscard, gnu::const]] auto levenshtein_distance(
+    string_view first, string_view second) -> size_t;
+
 /// @}
 
 /*

libutil/src/Strings.cpp

@@ -23,6 +23,7 @@
 #include <ranges>
 #include <string>
 #include <string_view>
+#include <vector>
 
 namespace {
 
@@ -150,4 +151,31 @@ auto split_at_first_space_or_newline(const string_view input) -> StringViewPair
     };
 }
 
+auto levenshtein_distance(
+    const string_view first, const string_view second) -> size_t
+{
+    const auto size_a = first.size();
+    const auto size_b = second.size();
+
+    auto distances = std::views::iota(0uz, size_b + 1uz)
+                   | std::ranges::to<std::vector>();
+
+    for (auto i = 0uz; i < size_a; ++i) {
+        auto prevDist = 0uz;
+
+        for (auto j = 0uz; j < size_b; ++j) {
+            const auto next = distances.at(j + 1uz);
+
+            const auto dist = std::exchange(prevDist, next)
+                            + (first.at(i) == second.at(j) ? 0uz : 1uz);
+
+            distances.at(j + 1uz) = std::min({ dist,
+                distances.at(j) + 1uz,
+                next + 1uz });
+        }
+    }
+
+    return distances.at(size_b);
+}
+
 } // namespace util::strings

tests/unit/libutil/Strings.cpp

@@ -219,3 +219,20 @@ TEST_CASE("Strings - words_view()", TAGS)
         REQUIRE(words.back() == "456");
     }
 }
+
+TEST_CASE("Strings - Levenshtein distance", TAGS)
+{
+    using util::strings::levenshtein_distance;
+
+    REQUIRE(
+        levenshtein_distance("kitten", "sitting") == 3uz);
+
+    REQUIRE(
+        levenshtein_distance("corporate", "cooperation") == 5uz);
+
+    REQUIRE(
+        levenshtein_distance("123", { }) == 0uz);
+
+    REQUIRE(
+        levenshtein_distance({ }, { }) == 0uz);
+}