refactor: Unify all types of NFA transitions into NfaSpontaneousTransition. (#76)

Co-authored-by: Lin Zhihao <59785146+LinZhihao-723@users.noreply.github.com>

Author: SharafMohamed

CMakeLists.txt

@@ -73,13 +73,14 @@ set(SOURCE_FILES
     src/log_surgeon/finite_automata/DfaState.hpp
     src/log_surgeon/finite_automata/DfaStatePair.hpp
     src/log_surgeon/finite_automata/Nfa.hpp
+    src/log_surgeon/finite_automata/NfaSpontaneousTransition.hpp
     src/log_surgeon/finite_automata/NfaState.hpp
     src/log_surgeon/finite_automata/PrefixTree.cpp
     src/log_surgeon/finite_automata/PrefixTree.hpp
     src/log_surgeon/finite_automata/RegexAST.hpp
     src/log_surgeon/finite_automata/RegisterHandler.hpp
     src/log_surgeon/finite_automata/StateType.hpp
-    src/log_surgeon/finite_automata/TaggedTransition.hpp
+    src/log_surgeon/finite_automata/TagOperation.hpp
     src/log_surgeon/finite_automata/UnicodeIntervalTree.hpp
     src/log_surgeon/finite_automata/UnicodeIntervalTree.tpp
     src/log_surgeon/Lalr1Parser.cpp

src/log_surgeon/Lexer.hpp

@@ -13,6 +13,7 @@
 #include <log_surgeon/Constants.hpp>
 #include <log_surgeon/finite_automata/Dfa.hpp>
 #include <log_surgeon/finite_automata/DfaState.hpp>
+#include <log_surgeon/finite_automata/NfaState.hpp>
 #include <log_surgeon/finite_automata/RegexAST.hpp>
 #include <log_surgeon/LexicalRule.hpp>
 #include <log_surgeon/ParserInputBuffer.hpp>

src/log_surgeon/finite_automata/Dfa.hpp

@@ -5,6 +5,7 @@
 #include <map>
 #include <memory>
 #include <set>
+#include <stack>
 #include <vector>
 
 #include <log_surgeon/finite_automata/DfaStatePair.hpp>

src/log_surgeon/finite_automata/Nfa.hpp

@@ -3,6 +3,7 @@
 
 #include <cstdint>
 #include <memory>
+#include <optional>
 #include <queue>
 #include <string>
 #include <unordered_map>
@@ -11,9 +12,12 @@
 #include <vector>
 
 #include <fmt/core.h>
+#include <fmt/format.h>
 
 #include <log_surgeon/Constants.hpp>
-#include <log_surgeon/finite_automata/NfaState.hpp>
+#include <log_surgeon/finite_automata/Capture.hpp>
+#include <log_surgeon/finite_automata/TagOperation.hpp>
+#include <log_surgeon/finite_automata/UnicodeIntervalTree.hpp>
 #include <log_surgeon/LexicalRule.hpp>
 #include <log_surgeon/types.hpp>
 #include <log_surgeon/UniqueIdGenerator.hpp>
@@ -35,32 +39,31 @@ class Nfa {
     explicit Nfa(std::vector<LexicalRule<TypedNfaState>> const& rules);
 
     /**
-     * Creates a unique_ptr for an NFA state with no tagged transitions and adds it to `m_states`.
-     * @return TypedNfaState*
+     * @return A pointer to the newly created NFA state with no spontaneous transitions.
      */
     [[nodiscard]] auto new_state() -> TypedNfaState*;
 
     /**
-     * Creates a unique_ptr for an NFA state with a negative tagged transition and adds it to
-     * `m_states`.
-     * @param captures
-     * @param dest_state
-     * @return TypedNfaState*
+     * @param captures A vector containing the captures of all alternate paths.
+     * @param dest_state The destination state to arrive at after negating the captures.
+     * @return A pointer to the newly created NFA state with a spontaneous transition to
+     * `dest_state`negating all the tags associated with `captures`.
      */
-    [[nodiscard]] auto new_state_with_negative_tagged_transition(
+    [[nodiscard]] auto new_state_from_negative_captures(
             std::vector<Capture const*> const& captures,
             TypedNfaState const* dest_state
     ) -> TypedNfaState*;
 
     /**
-     * Creates the start and end states for a capture group.
-     * @param capture The capture associated with the capture group.
-     * @param dest_state
-     * @return A pair of states:
-     * - A new state with a positive tagged start transition from `m_root`.
-     * - A new state with a positive tagged end transition to `dest_state`.
+     * @param capture The positive capture to be tracked.
+     * @param dest_state The destination state to arrive at after tracking the capture.
+     * @return A pair of pointers to the two newly created NFA states:
+     * - A state arrived at from a spontaneous transition out of `m_root` that sets a tag to track
+     * the capture's start position.
+     * - A state with a spontaneous transition to `dest_state` that sets a tag to track the
+     * capture's end position
      */
-    [[nodiscard]] auto new_start_and_end_states_with_positive_tagged_transitions(
+    [[nodiscard]] auto new_start_and_end_states_from_positive_capture(
             Capture const* capture,
             TypedNfaState const* dest_state
     ) -> std::pair<TypedNfaState*, TypedNfaState*>;
@@ -72,9 +75,10 @@ class Nfa {
     [[nodiscard]] auto get_bfs_traversal_order() const -> std::vector<TypedNfaState const*>;
 
     /**
-     * @return A string representation of the NFA.
+     * @return A string representation of the NFA on success.
+     * @return Forwards `NfaState::serialize`'s return value (`std::nullopt`) on failure.
      */
-    [[nodiscard]] auto serialize() const -> std::string;
+    [[nodiscard]] auto serialize() const -> std::optional<std::string>;
 
     auto add_root_interval(Interval interval, TypedNfaState* dest_state) -> void {
         m_root->add_interval(interval, dest_state);
@@ -100,18 +104,6 @@ class Nfa {
     [[nodiscard]] auto get_or_create_capture_tag_pair(Capture const* capture
     ) -> std::pair<tag_id_t, tag_id_t>;
 
-    /**
-     * Creates a `unique_ptr` for an NFA state with a positive tagged end transition and adds it to
-     * `m_states`.
-     * @param tag_id
-     * @param dest_state
-     * @return A new state with a positive tagged end transition to `dest_state`.
-     */
-    [[nodiscard]] auto new_state_with_positive_tagged_end_transition(
-            tag_id_t tag_id,
-            TypedNfaState const* dest_state
-    ) -> TypedNfaState*;
-
     std::vector<std::unique_ptr<TypedNfaState>> m_states;
     // TODO: Lexer currently enforces unique naming across capture groups. However, this limits use
     // cases. Possibly initialize this in the lexer and pass it in during construction.
@@ -146,39 +138,35 @@ auto Nfa<TypedNfaState>::new_state() -> TypedNfaState* {
 }
 
 template <typename TypedNfaState>
-auto Nfa<TypedNfaState>::new_state_with_positive_tagged_end_transition(
-        tag_id_t const tag_id,
-        TypedNfaState const* dest_state
-) -> TypedNfaState* {
-    m_states.emplace_back(std::make_unique<TypedNfaState>(tag_id, dest_state));
-    return m_states.back().get();
-}
-
-template <typename TypedNfaState>
-auto Nfa<TypedNfaState>::new_state_with_negative_tagged_transition(
+auto Nfa<TypedNfaState>::new_state_from_negative_captures(
         std::vector<Capture const*> const& captures,
         TypedNfaState const* dest_state
 ) -> TypedNfaState* {
     std::vector<tag_id_t> tags;
-    for (auto const capture : captures) {
+    for (auto const* capture : captures) {
         auto const [start_tag, end_tag]{get_or_create_capture_tag_pair(capture)};
         tags.push_back(start_tag);
         tags.push_back(end_tag);
     }
 
-    m_states.emplace_back(std::make_unique<TypedNfaState>(std::move(tags), dest_state));
+    m_states.emplace_back(
+            std::make_unique<TypedNfaState>(TagOperationType::Negate, std::move(tags), dest_state)
+    );
     return m_states.back().get();
 }
 
 template <typename TypedNfaState>
-auto Nfa<TypedNfaState>::new_start_and_end_states_with_positive_tagged_transitions(
+auto Nfa<TypedNfaState>::new_start_and_end_states_from_positive_capture(
         Capture const* capture,
         TypedNfaState const* dest_state
 ) -> std::pair<TypedNfaState*, TypedNfaState*> {
     auto const [start_tag, end_tag]{get_or_create_capture_tag_pair(capture)};
-    auto* start_state = new_state();
-    m_root->add_positive_tagged_start_transition(start_tag, start_state);
-    auto* end_state{new_state_with_positive_tagged_end_transition(end_tag, dest_state)};
+    auto* start_state{new_state()};
+    m_root->add_spontaneous_transition(TagOperationType::Set, {start_tag}, start_state);
+    m_states.emplace_back(
+            std::make_unique<TypedNfaState>(TagOperationType::Set, std::vector{end_tag}, dest_state)
+    );
+    auto* end_state{m_states.back().get()};
     return {start_state, end_state};
 }
 
@@ -208,33 +196,15 @@ auto Nfa<TypedNfaState>::get_bfs_traversal_order() const -> std::vector<TypedNfa
                 add_to_queue_and_visited(dest_state);
             }
         }
-        for (auto const* dest_state : current_state->get_epsilon_transitions()) {
-            add_to_queue_and_visited(dest_state);
-        }
-        for (auto const& positive_tagged_start_transition :
-             current_state->get_positive_tagged_start_transitions())
-        {
-            add_to_queue_and_visited(positive_tagged_start_transition.get_dest_state());
-        }
-
-        auto const& optional_positive_tagged_end_transition
-                = current_state->get_positive_tagged_end_transition();
-        if (optional_positive_tagged_end_transition.has_value()) {
-            add_to_queue_and_visited(optional_positive_tagged_end_transition.value().get_dest_state(
-            ));
-        }
-
-        auto const& optional_negative_tagged_transition
-                = current_state->get_negative_tagged_transition();
-        if (optional_negative_tagged_transition.has_value()) {
-            add_to_queue_and_visited(optional_negative_tagged_transition.value().get_dest_state());
+        for (auto const& spontaneous_transition : current_state->get_spontaneous_transitions()) {
+            add_to_queue_and_visited(spontaneous_transition.get_dest_state());
         }
     }
     return visited_order;
 }
 
 template <typename TypedNfaState>
-auto Nfa<TypedNfaState>::serialize() const -> std::string {
+auto Nfa<TypedNfaState>::serialize() const -> std::optional<std::string> {
     auto const traversal_order = get_bfs_traversal_order();
 
     std::unordered_map<TypedNfaState const*, uint32_t> state_ids;
@@ -244,10 +214,11 @@ auto Nfa<TypedNfaState>::serialize() const -> std::string {
 
     std::vector<std::string> serialized_states;
     for (auto const* state : traversal_order) {
-        // `state_ids` is well-formed as its generated from `get_bfs_traversal_order` so we can
-        // safely assume `state->serialize(state_ids)` will return a valid value.
-        // NOLINTNEXTLINE(bugprone-unchecked-optional-access)
-        serialized_states.emplace_back(state->serialize(state_ids).value());
+        auto const optional_serialized_state{state->serialize(state_ids)};
+        if (false == optional_serialized_state.has_value()) {
+            return std::nullopt;
+        }
+        serialized_states.emplace_back(optional_serialized_state.value());
     }
     return fmt::format("{}\n", fmt::join(serialized_states, "\n"));
 }

src/log_surgeon/finite_automata/NfaSpontaneousTransition.hpp

@@ -0,0 +1,65 @@
+#ifndef LOG_SURGEON_FINITE_AUTOMATA_NFASPONTANEOUSTRANSITION_HPP
+#define LOG_SURGEON_FINITE_AUTOMATA_NFASPONTANEOUSTRANSITION_HPP
+
+#include <cstdint>
+#include <optional>
+#include <ranges>
+#include <string>
+#include <unordered_map>
+#include <utility>
+#include <vector>
+
+#include <fmt/format.h>
+
+#include <log_surgeon/finite_automata/TagOperation.hpp>
+
+namespace log_surgeon::finite_automata {
+/**
+ * Represents an NFA transition with a collection of tag operations to be performed during the
+ * transition.
+ *
+ * @tparam TypedNfaState Specifies the type of transition (bytes or UTF-8 characters).
+ */
+template <typename TypedNfaState>
+class NfaSpontaneousTransition {
+public:
+    NfaSpontaneousTransition(std::vector<TagOperation> tag_ops, TypedNfaState const* dest_state)
+            : m_tag_ops{std::move(tag_ops)},
+              m_dest_state{dest_state} {}
+
+    [[nodiscard]] auto get_tag_ops() const -> std::vector<TagOperation> const& { return m_tag_ops; }
+
+    [[nodiscard]] auto get_dest_state() const -> TypedNfaState const* { return m_dest_state; }
+
+    /**
+     * @param state_ids A map of states to their unique identifiers.
+     * @return A string representation of the spontaneous transition on success.
+     * @return std::nullopt if `m_dest_state` is not in `state_ids`.
+     */
+    [[nodiscard]] auto serialize(std::unordered_map<TypedNfaState const*, uint32_t> const& state_ids
+    ) const -> std::optional<std::string>;
+
+private:
+    std::vector<TagOperation> m_tag_ops;
+    TypedNfaState const* m_dest_state;
+};
+
+template <typename TypedNfaState>
+auto NfaSpontaneousTransition<TypedNfaState>::serialize(
+        std::unordered_map<TypedNfaState const*, uint32_t> const& state_ids
+) const -> std::optional<std::string> {
+    if (false == state_ids.contains(m_dest_state)) {
+        return std::nullopt;
+    }
+    auto transformed_operations
+            = m_tag_ops | std::ranges::views::transform(&TagOperation::serialize);
+
+    return fmt::format(
+            "{}[{}]",
+            state_ids.at(m_dest_state),
+            fmt::join(transformed_operations, ",")
+    );
+}
+}  // namespace log_surgeon::finite_automata
+
+#endif  // LOG_SURGEON_FINITE_AUTOMATA_NFASPONTANEOUSTRANSITION_HPP