separate fsm version from other

fsm/fsm.h

@@ -0,0 +1,295 @@
+#pragma once
+
+#include <type_traits>
+#include <utility>
+#include <variant>
+
+/** @brief Missing transition policy. */
+enum class missing_transition_policy {
+  /** @brief Missing transition is considered required by validation. */
+  strict,
+
+  /** @brief Missing transition is allowed (event ignored). */
+  ignore
+};
+
+/** @brief Marker for undefined transition. */
+struct no_transition {};
+
+/** @brief Default guard: always allow. */
+struct always_allow {
+  template <typename TContext, typename TFrom, typename TEvent>
+  bool operator() (TContext &, const TFrom &, const TEvent &) const noexcept {
+    return true;
+  }
+};
+
+/** @brief Default action: do nothing. */
+struct do_nothing {
+  template <typename TContext, typename TFrom, typename TTo, typename TEvent>
+  void operator() (TContext &, const TFrom &, const TTo &, const TEvent &) const noexcept {
+  }
+};
+
+/**
+ * @brief Transition descriptor.
+ *
+ * @tparam TTo Destination state type.
+ * @tparam TGuard Guard functor type.
+ * @tparam TAction Action functor type.
+ */
+template <typename TTo, typename TGuard = always_allow, typename TAction = do_nothing>
+struct transition_to {
+  using to_state = TTo;
+  using guard    = TGuard;
+  using action   = TAction;
+};
+
+/**
+ * @brief Transition table mapping: specialize to define transitions.
+ * Default: no_transition.
+ */
+template <typename TFrom, typename TEvent>
+struct transition {
+  using type = no_transition;
+};
+
+/** @brief Optional mapping to names for logs. */
+template <typename T>
+struct name_of {
+  static constexpr const char *value = "?";
+};
+
+/**
+ * @brief Logger interface (default no-op).
+ *
+ * You can implement:
+ *  - ignored_transition(from, event)
+ *  - unhandled_transition(from, event)   // policy=strict but missing transition at runtime
+ *  - guard_blocked(from, event)
+ */
+struct null_logger {
+  void ignored_transition (const char *, const char *) noexcept {}
+  void unhandled_transition (const char *, const char *) noexcept {}
+  void guard_blocked (const char *, const char *) noexcept {}
+};
+
+/** @brief Hooks: state-only. */
+template <typename TState>
+struct on_exit_state {
+  template <typename TContext>
+  static void call (TContext &, const TState &) noexcept {}
+};
+
+template <typename TState>
+struct on_entry_state {
+  template <typename TContext>
+  static void call (TContext &, const TState &) noexcept {}
+};
+
+/** @brief Hooks: event-specific. */
+template <typename TState, typename TEvent>
+struct on_exit_event {
+  template <typename TContext>
+  static void call (TContext &, const TState &, const TEvent &) noexcept {}
+};
+
+template <typename TState, typename TEvent>
+struct on_entry_event {
+  template <typename TContext>
+  static void call (TContext &, const TState &, const TEvent &) noexcept {}
+};
+
+/**
+ * @brief Factory: construct destination state.
+ *
+ * IMPORTANT: Now includes TContext in template parameters to avoid templated make<TContext>().
+ *
+ * Default: To{}.
+ * Specialize state_factory<TContext, From, Event, To> as needed.
+ */
+template <typename TContext, typename TFrom, typename TEvent, typename TTo>
+struct state_factory {
+  static TTo make (TContext &, const TFrom &, const TEvent &) {
+    return TTo{};
+  }
+};
+
+/**
+ * @brief Per (State,Event) missing transition policy.
+ *
+ * Default is ignore. Mark only required pairs strict.
+ */
+template <typename TFrom, typename TEvent>
+struct missing_transition {
+  static constexpr missing_transition_policy policy = missing_transition_policy::ignore;
+};
+
+/** @brief Helper: is T one of Ts... */
+template <typename T, typename... Ts>
+struct is_one_of : std::disjunction<std::is_same<T, Ts>...> {};
+
+/** @brief Helper: transition exists? */
+template <typename TFrom, typename TEvent>
+struct has_transition
+  : std::bool_constant < !std::is_same_v<typename transition<TFrom, TEvent>::type, no_transition >> {};
+
+/** @brief Factory validation (call-form). */
+template <typename TContext, typename TFrom, typename TEvent, typename TTo, typename = void>
+struct is_factory_invocable : std::false_type {};
+
+template <typename TContext, typename TFrom, typename TEvent, typename TTo>
+struct is_factory_invocable <
+  TContext, TFrom, TEvent, TTo,
+std::void_t<decltype (state_factory<TContext, TFrom, TEvent, TTo>::make (
+                        std::declval<TContext &>(),
+                        std::declval<const TFrom &>(),
+                        std::declval<const TEvent &>())) >>
+                            : std::bool_constant<std::is_same_v<
+                            decltype (state_factory<TContext, TFrom, TEvent, TTo>::make (
+                                        std::declval<TContext &>(),
+                                        std::declval<const TFrom &>(),
+                                        std::declval<const TEvent &>())),
+                                            TTo>> {};
+
+/**
+ * @brief FSM with runtime state storage and explicit compile-time validation.
+ *
+ * dispatch():
+ *  - never static_asserts on missing transitions (avoids State×Event explosion with std::visit)
+ *  - handles missing transitions via missing_transition<State,Event>::policy (ignore/strict)
+ *
+ * validate_events<Ev...>():
+ *  - compile-time checks ONLY for pairs marked strict.
+ */
+template <typename TContext, typename TLogger, typename... TStates>
+class fsm {
+  public:
+    using state_variant = std::variant<TStates...>;
+
+    template <typename TInitial,
+              typename = std::enable_if_t<is_one_of<TInitial, TStates...>::value>>
+    explicit fsm (TContext &ctx, TInitial initial, TLogger logger = TLogger{})
+      : m_ctx (ctx)
+      , m_logger (std::move (logger))
+      , m_state (std::move (initial)) {
+      std::visit (init_entry_visitor{*this}, m_state);
+    }
+
+    template <typename TEvent>
+    bool dispatch (const TEvent &ev) {
+      dispatch_visitor<TEvent> v{*this, ev};
+      return std::visit (v, m_state);
+    }
+
+    template <typename... TEvents>
+    static constexpr void validate_events() {
+      (validate_one_event<TEvents>(), ...);
+    }
+
+    const state_variant &state() const noexcept {
+      return m_state;
+    }
+    state_variant &state() noexcept {
+      return m_state;
+    }
+
+  private:
+    template <typename TEvent>
+    struct dispatch_visitor {
+      fsm &self;
+      const TEvent &ev;
+
+      template <typename TFrom>
+      bool operator() (TFrom &from) const {
+        return self.template dispatch_from<TFrom> (from, ev);
+      }
+    };
+
+    struct init_entry_visitor {
+      fsm &self;
+      template <typename TState>
+      void operator() (TState &st) const {
+        on_entry_state<TState>::call (self.m_ctx, st);
+      }
+    };
+
+    template <typename TEvent>
+    struct entry_after_commit_visitor {
+      fsm &self;
+      const TEvent &ev;
+
+      template <typename TState>
+      void operator() (TState &st) const {
+        on_entry_state<TState>::call (self.m_ctx, st);
+        on_entry_event<TState, TEvent>::call (self.m_ctx, st, ev);
+      }
+    };
+
+    template <typename TFrom, typename TEvent>
+    bool dispatch_from (TFrom &from, const TEvent &ev) {
+      using tr = typename transition<TFrom, TEvent>::type;
+
+      if constexpr (std::is_same_v<tr, no_transition>) {
+        if constexpr (missing_transition<TFrom, TEvent>::policy == missing_transition_policy::strict)
+          m_logger.unhandled_transition (name_of<TFrom>::value, name_of<TEvent>::value);
+        else
+          m_logger.ignored_transition (name_of<TFrom>::value, name_of<TEvent>::value);
+        return false;
+      } else {
+        using to_state = typename tr::to_state;
+        using guard_t  = typename tr::guard;
+        using action_t = typename tr::action;
+
+        static_assert (is_one_of<to_state, TStates...>::value,
+                       "FSM: transition target state not in FSM state list.");
+
+        static_assert (std::is_invocable_r_v<bool, guard_t, TContext &, const TFrom &, const TEvent &>,
+                       "FSM: guard must be callable as bool(Context&, const From&, const Event&).");
+
+        static_assert (std::is_invocable_r_v<void, action_t, TContext &, const TFrom &, const to_state &, const TEvent &>,
+                       "FSM: action must be callable as void(Context&, const From&, const To&, const Event&).");
+
+        static_assert (is_factory_invocable<TContext, TFrom, TEvent, to_state>::value,
+                       "FSM: state_factory<Context,From,Event,To>::make must be callable as "
+                       "To make(Context&, const From&, const Event&).");
+
+        guard_t guard{};
+        if (!guard (m_ctx, from, ev)) {
+          m_logger.guard_blocked (name_of<TFrom>::value, name_of<TEvent>::value);
+          return false;
+        }
+
+        on_exit_event<TFrom, TEvent>::call (m_ctx, from, ev);
+        on_exit_state<TFrom>::call (m_ctx, from);
+
+        to_state to = state_factory<TContext, TFrom, TEvent, to_state>::make (m_ctx, from, ev);
+
+        action_t action{};
+        action (m_ctx, from, to, ev);
+
+        m_state = std::move (to);
+
+        std::visit (entry_after_commit_visitor<TEvent> {*this, ev}, m_state);
+        return true;
+      }
+    }
+
+    template <typename TEvent>
+    static constexpr void validate_one_event() {
+      (validate_pair<TStates, TEvent>(), ...);
+    }
+
+    template <typename TState, typename TEvent>
+    static constexpr void validate_pair() {
+      if constexpr (missing_transition<TState, TEvent>::policy == missing_transition_policy::strict) {
+        static_assert (has_transition<TState, TEvent>::value,
+                       "FSM validation: required (strict) transition is missing for (State, Event).");
+      }
+    }
+
+  private:
+    TContext     &m_ctx;
+    TLogger       m_logger;
+    state_variant m_state;
+};

fsm/kindle2latex.pro

@@ -0,0 +1,10 @@
+TEMPLATE = app
+CONFIG += console c++17
+CONFIG -= app_bundle
+CONFIG -= qt
+
+SOURCES += \
+        main.cpp
+
+HEADERS += \
+  fsm.h

fsm/main.cpp

@@ -0,0 +1,372 @@
+#include "fsm.h"
+
+#include <cerrno>
+#include <cstdlib>
+#include <cstring>
+#include <fstream>
+#include <getopt.h>
+#include <iostream>
+#include <string>
+#include <unordered_map>
+#include <utility>
+#include <vector>
+
+// ------------------------------------------------------------
+// Data model
+// ------------------------------------------------------------
+
+struct Quote {
+  std::string meta;
+  std::vector<std::string> text_lines;
+};
+
+static std::string rstrip_cr (std::string s) {
+  if (!s.empty() && s.back() == '\r')
+    s.pop_back();
+  return s;
+}
+
+static std::string latex_escape (const std::string &s) {
+  // Conservative escaping for LaTeX special chars.
+  std::string out;
+  out.reserve (s.size() + s.size() / 8);
+  for (const unsigned char ch : s) {
+    switch (ch) {
+      case '\\':
+        out += "\\textbackslash{}";
+        break;
+      case '{':
+        out += "\\{";
+        break;
+      case '}':
+        out += "\\}";
+        break;
+      case '%':
+        out += "\\%";
+        break;
+      case '$':
+        out += "\\$";
+        break;
+      case '#':
+        out += "\\#";
+        break;
+      case '&':
+        out += "\\&";
+        break;
+      case '_':
+        out += "\\_";
+        break;
+      case '^':
+        out += "\\textasciicircum{}";
+        break;
+      case '~':
+        out += "\\textasciitilde{}";
+        break;
+      default:
+        out.push_back (static_cast<char> (ch));
+        break;
+    }
+  }
+  return out;
+}
+
+// ------------------------------------------------------------
+// FSM: parsing Kindle "My Clippings" style format
+// ------------------------------------------------------------
+
+struct StExpectTitle {};
+struct StExpectMeta {};
+struct StExpectBlank {};
+struct StCollectText {};
+
+struct EvTitle {
+  std::string s;
+};
+struct EvMeta  {
+  std::string s;
+};
+struct EvBlank {};
+struct EvText  {
+  std::string s;
+};
+struct EvSep   {};
+struct EvEof   {};
+
+struct ParserContext {
+  // group storage
+  std::vector<std::string> order;
+  std::unordered_map<std::string, std::vector<Quote>> by_title;
+  std::unordered_map<std::string, std::size_t> seen;
+
+  // currently parsed quote
+  std::string cur_title;
+  std::string cur_meta;
+  std::vector<std::string> cur_text;
+
+  void start_title (std::string t) {
+    cur_title = std::move (t);
+    cur_meta.clear();
+    cur_text.clear();
+  }
+  void set_meta (std::string m) {
+    cur_meta = std::move (m);
+  }
+  void add_text (std::string line) {
+    cur_text.push_back (std::move (line));
+  }
+
+  void finalize_quote_if_ready() {
+    if (cur_title.empty() || cur_meta.empty()) {
+      cur_meta.clear();
+      cur_text.clear();
+      return;
+    }
+
+    if (seen.find (cur_title) == seen.end()) {
+      seen.emplace (cur_title, order.size());
+      order.push_back (cur_title);
+    }
+
+    by_title[cur_title].push_back (Quote{cur_meta, cur_text});
+
+    cur_meta.clear();
+    cur_text.clear();
+  }
+};
+
+struct SilentLogger : null_logger {};
+
+// Actions
+struct ActSetTitle {
+  void operator() (ParserContext &ctx, const StExpectTitle &, const StExpectMeta &, const EvTitle &ev) const {
+    ctx.start_title (ev.s);
+  }
+};
+
+struct ActSetMeta {
+  void operator() (ParserContext &ctx, const StExpectMeta &, const StExpectBlank &, const EvMeta &ev) const {
+    ctx.set_meta (ev.s);
+  }
+};
+
+struct ActAddText {
+  template <typename TFrom>
+  void operator() (ParserContext &ctx, const TFrom &, const StCollectText &, const EvText &ev) const {
+    ctx.add_text (ev.s);
+  }
+};
+
+struct ActFinalizeOnSep {
+  template <typename TFrom>
+  void operator() (ParserContext &ctx, const TFrom &, const StExpectTitle &, const EvSep &) const {
+    ctx.finalize_quote_if_ready();
+  }
+};
+
+struct ActFinalizeOnEof {
+  template <typename TFrom>
+  void operator() (ParserContext &ctx, const TFrom &, const StExpectTitle &, const EvEof &) const {
+    ctx.finalize_quote_if_ready();
+  }
+};
+
+// Transitions
+template <> struct transition<StExpectTitle, EvTitle> {
+  using type = transition_to<StExpectMeta, always_allow, ActSetTitle>;
+};
+template <> struct transition<StExpectTitle, EvBlank> {
+  using type = transition_to<StExpectTitle>;
+};
+template <> struct transition<StExpectTitle, EvEof>   {
+  using type = transition_to<StExpectTitle, always_allow, ActFinalizeOnEof>;
+};
+
+template <> struct transition<StExpectMeta, EvMeta>   {
+  using type = transition_to<StExpectBlank, always_allow, ActSetMeta>;
+};
+
+template <> struct transition<StExpectBlank, EvBlank> {
+  using type = transition_to<StCollectText>;
+};
+template <> struct transition<StExpectBlank, EvText>  {
+  using type = transition_to<StCollectText, always_allow, ActAddText>;
+};
+
+template <> struct transition<StCollectText, EvText>  {
+  using type = transition_to<StCollectText, always_allow, ActAddText>;
+};
+template <> struct transition<StCollectText, EvSep>   {
+  using type = transition_to<StExpectTitle, always_allow, ActFinalizeOnSep>;
+};
+template <> struct transition<StCollectText, EvEof>   {
+  using type = transition_to<StExpectTitle, always_allow, ActFinalizeOnEof>;
+};
+
+// Strict contract for the pairs we dispatch.
+template <> struct missing_transition<StExpectTitle, EvTitle> {
+  static constexpr missing_transition_policy policy = missing_transition_policy::strict;
+};
+template <> struct missing_transition<StExpectTitle, EvBlank> {
+  static constexpr missing_transition_policy policy = missing_transition_policy::strict;
+};
+template <> struct missing_transition<StExpectTitle, EvEof>   {
+  static constexpr missing_transition_policy policy = missing_transition_policy::strict;
+};
+
+template <> struct missing_transition<StExpectMeta, EvMeta>   {
+  static constexpr missing_transition_policy policy = missing_transition_policy::strict;
+};
+
+template <> struct missing_transition<StExpectBlank, EvBlank> {
+  static constexpr missing_transition_policy policy = missing_transition_policy::strict;
+};
+template <> struct missing_transition<StExpectBlank, EvText>  {
+  static constexpr missing_transition_policy policy = missing_transition_policy::strict;
+};
+
+template <> struct missing_transition<StCollectText, EvText>  {
+  static constexpr missing_transition_policy policy = missing_transition_policy::strict;
+};
+template <> struct missing_transition<StCollectText, EvSep>   {
+  static constexpr missing_transition_policy policy = missing_transition_policy::strict;
+};
+template <> struct missing_transition<StCollectText, EvEof>   {
+  static constexpr missing_transition_policy policy = missing_transition_policy::strict;
+};
+
+using ParserFsm = fsm<ParserContext, SilentLogger, StExpectTitle, StExpectMeta, StExpectBlank, StCollectText>;
+
+static constexpr void validate_fsm_contract() {
+  ParserFsm::validate_events<EvTitle, EvMeta, EvBlank, EvText, EvSep, EvEof>();
+}
+
+// ------------------------------------------------------------
+// CLI
+// ------------------------------------------------------------
+
+struct CliArgs {
+  std::string input;
+  std::string output;
+};
+
+static void print_usage (const char *argv0) {
+  std::cerr << "Usage: " << argv0 << " --input <file> --output <file>\n";
+}
+
+static bool parse_args (int argc, char **argv, CliArgs &out) {
+  static option long_opts[] = {
+    {"input",  required_argument, nullptr, 'i'},
+    {"output", required_argument, nullptr, 'o'},
+    {"help",   no_argument,       nullptr, 'h'},
+    {nullptr,   0,                 nullptr,  0 }
+  };
+
+  int c = 0;
+  while ((c = ::getopt_long (argc, argv, "i:o:h", long_opts, nullptr)) != -1) {
+    switch (c) {
+      case 'i':
+        out.input = optarg;
+        break;
+      case 'o':
+        out.output = optarg;
+        break;
+      case 'h':
+        print_usage (argv[0]);
+        return false;
+      default:
+        print_usage (argv[0]);
+        return false;
+    }
+  }
+
+  if (out.input.empty() || out.output.empty()) {
+    print_usage (argv[0]);
+    return false;
+  }
+  return true;
+}
+
+// ------------------------------------------------------------
+// Conversion
+// ------------------------------------------------------------
+
+static int convert (const std::string &in_path, const std::string &out_path) {
+  std::ifstream in (in_path);
+  if (!in.is_open()) {
+    std::cerr << "Failed to open input file: " << in_path << " (" << std::strerror (errno) << ")\n";
+    return 2;
+  }
+
+  ParserContext ctx{};
+  ParserFsm fsm (ctx, StExpectTitle{}, SilentLogger{});
+
+  std::string line;
+  while (std::getline (in, line)) {
+    line = rstrip_cr (std::move (line));
+
+    if (line == "==========") {
+      fsm.dispatch (EvSep{});
+      continue;
+    }
+
+    const bool is_blank = line.empty();
+    const auto &st = fsm.state();
+
+    if (std::holds_alternative<StExpectTitle> (st)) {
+      if (is_blank)
+        fsm.dispatch (EvBlank{});
+      else
+        fsm.dispatch (EvTitle{line});
+    } else if (std::holds_alternative<StExpectMeta> (st)) {
+      // Kindle format expects meta here. If missing, keep going.
+      fsm.dispatch (EvMeta{line});
+    } else if (std::holds_alternative<StExpectBlank> (st)) {
+      if (is_blank)
+        fsm.dispatch (EvBlank{});
+      else
+        fsm.dispatch (EvText{line}); // some clippings have no blank line
+    } else {
+      // StCollectText
+      fsm.dispatch (EvText{line});
+    }
+  }
+
+  fsm.dispatch (EvEof{});
+
+  std::ofstream out (out_path, std::ios::trunc);
+  if (!out.is_open()) {
+    std::cerr << "Failed to open output file: " << out_path << " (" << std::strerror (errno) << ")\n";
+    return 3;
+  }
+
+  for (std::size_t i = 0; i < ctx.order.size(); ++i) {
+    const auto &title = ctx.order[i];
+    out << "\\section {" << latex_escape (title) << "}\n";
+
+    const auto it = ctx.by_title.find (title);
+    if (it == ctx.by_title.end())
+      continue;
+
+    for (const auto &q : it->second) {
+      out << "  \\subsection {" << latex_escape (q.meta) << "}\n";
+      for (const auto &tl : q.text_lines)
+        out << "    " << latex_escape (tl) << "\n";
+      out << "      \\subsubsection{notes}\n\n";
+    }
+
+    if (i + 1 < ctx.order.size())
+      out << "\n";
+  }
+
+  return 0;
+}
+
+int main (int argc, char **argv) {
+  validate_fsm_contract();
+
+  CliArgs args;
+  if (!parse_args (argc, argv, args))
+    return 1;
+
+  return convert (args.input, args.output);
+}