commit 56d772dd01d824fb28f81e53dde71439af0093f4
Merge: a92163a 11c6d38
Author: Daniel Swanson <popcorn.tomato.dude@gmail.com>
Date:   Fri Jun 11 11:38:04 2021 -0500

    Merge branch 'icu' of github.com:apertium/apertium into icu

diff --combined apertium/perceptron_spec.h
index 5d67748,cf021d1..d51a507
--- a/apertium/perceptron_spec.h
+++ b/apertium/perceptron_spec.h
@@@ -168,9 -168,9 +168,9 @@@ public
    #undef X
    static bool static_constructed;
    static unsigned char num_opcodes;
 -  static const UString opcode_names[];
 -  static const UString type_names[];
 -  static std::map<const UString, Opcode> opcode_values;
 +  static const std::string opcode_names[];
 +  static const std::string type_names[];
 +  static std::map<const std::string, Opcode> opcode_values;
    static std::vector<Morpheme> untagged_sentinel;
    static LexicalUnit token_wordoids_underflow;
    static LexicalUnit token_wordoids_overflow;
@@@ -188,13 -188,14 +188,12 @@@
            out << val.boolVal();
            break;
          case STRVAL:
-           out << val.str().c_str();
+           out << val.str();
            break;
          case STRARRVAL: {
            out << "[";
-           std::vector<std::string> &str_arr = val.strArr();
-           for (auto& it : str_arr) {
 -          std::vector<UString> &str_arr = val.strArr();
 -          std::vector<UString>::const_iterator it = str_arr.begin();
 -          for (; it != str_arr.end(); it++) {
 -            out << *it;
++          for (auto& it : val.strArr()) {
 +            out << it;
            }
            out << "]";
          } break;
@@@ -204,8 -205,9 +203,8 @@@
          case WRDARRVAL: {
            out << "[";
            std::vector<Morpheme> &wrd_arr = val.wrdArr();
 -          std::vector<Morpheme>::const_iterator it = wrd_arr.begin();
 -          for (; it != wrd_arr.end(); it++) {
 -            out << *it;
 +          for (auto& it : wrd_arr) {
 +            out << it;
            }
            out << "]";
          } break;
@@@ -228,11 -230,11 +227,11 @@@
        type = other.type;
        switch (type) {
          case STRVAL:
 -          payload.strval = new UString(*other.payload.strval);
 +          payload.strval = new std::string(*other.payload.strval);
            break;
          case STRARRVAL:
            payload.strarrval =
 -              new std::vector<UString>(*other.payload.strarrval);
 +              new std::vector<std::string>(*other.payload.strarrval);
            break;
          case WRDVAL:
            payload.wrdval = new Morpheme(*other.payload.wrdval);
@@@ -258,12 -260,12 +257,12 @@@
        payload.bval = bval;
        type = BVAL;
      }
 -    StackValue(const UString &strval) {
 -      payload.strval = new UString(strval);
 +    StackValue(const std::string& strval) {
 +      payload.strval = new std::string(strval);
        type = STRVAL;
      }
 -    StackValue(const std::vector<UString> &strarrval) {
 -      payload.strarrval = new std::vector<UString>(strarrval);
 +    StackValue(const std::vector<std::string> &strarrval) {
 +      payload.strarrval = new std::vector<std::string>(strarrval);
        type = STRARRVAL;
      }
      StackValue(const Morpheme &wordoid) {
@@@ -288,11 -290,11 +287,11 @@@
        payload.wrdarrval = new std::vector<Morpheme>(wordoids);
        type = WRDARRVAL;
      }
 -    StackValue(UString *strval) {
 +    StackValue(std::string* strval) {
        payload.strval = strval;
        type = STRVAL;
      }
 -    StackValue(std::vector<UString> *strarrval) {
 +    StackValue(std::vector<std::string>* strarrval) {
        payload.strarrval = strarrval;
        type = STRARRVAL;
      }
@@@ -329,11 -331,11 +328,11 @@@
        assert(type == BVAL);
        return payload.bval;
      }
 -    UString& str() const {
 +    std::string& str() const {
        assert(type == STRVAL);
        return *payload.strval;
      }
 -    std::vector<UString>& strArr() const {
 +    std::vector<std::string>& strArr() const {
        assert(type == STRARRVAL);
        return *payload.strarrval;
      }
@@@ -364,8 -366,8 +363,8 @@@
      union StackValueUnion {
        int intval;
        bool bval;
 -      UString* strval;
 -      std::vector<UString>* strarrval;
 +      std::string* strval;
 +      std::vector<std::string>* strarrval;
        Morpheme* wrdval;
        std::vector<Morpheme>* wrdarrval;
      } payload;
@@@ -377,8 -379,8 +376,8 @@@
      signed char intbyte : 8;
    };
    Optional<TaggerDataPercepCoarseTags> coarse_tags;
 -  static UString dot;
 -  std::vector<UString> str_consts;
 +  static std::string dot;
 +  std::vector<std::string> str_consts;
    std::vector<VMSet> set_consts;
    mutable std::vector<StackValue> global_results;
    std::vector<FeatureDefn> global_defns;
@@@ -388,10 -390,10 +387,10 @@@
      const TaggedSentence &tagged, const Sentence &untagged,
      int token_idx, int wordoid_idx,
      UnaryFeatureVec &feat_vec_out) const;
 -  UString coarsen(const Morpheme &wrd) const;
 +  std::string coarsen(const Morpheme &wrd) const;
    void clearCache() const;
    int beam_width;
 -  mutable std::map<const Morpheme, UString> coarsen_cache;
 +  mutable std::map<const Morpheme, std::string> coarsen_cache;
  private:
    class MachineStack {
      std::deque<StackValue> data;
@@@ -452,15 -454,15 +451,15 @@@
      };
      std::deque<LoopState> loop_stack;
      std::vector<StackValue> slots;
 -    void unimplemented_opcode(UString opstr);
 +    void unimplemented_opcode(std::string opstr);
      const LexicalUnit& get_token(const Sentence &untagged);
      const std::vector<Morpheme>& tagged_to_wordoids(const TaggedToken &tt);
      const Morpheme& get_wordoid(const TaggedSentence &tagged);
      const VMSet& get_set_operand();
      int get_int_operand();
      unsigned int get_uint_operand();
 -    const UString& get_str_operand();
 -    static UString get_tag(const Tag &in);
 +    const std::string& get_str_operand();
 +    static std::string get_tag(const Tag &in);
      bool execCommonOp(Opcode op);
    public:
      void traceMachineState();
@@@ -476,16 -478,16 +475,16 @@@
        int token_idx,
        int wordoid_idx);
    };
 -  struct In : public std::unary_function<const UString&, bool> {
 +  struct In : public std::unary_function<const std::string&, bool> {
      const VMSet& haystack;
      In(const VMSet &haystack);
 -    bool operator() (const UString &needle) const;
 +    bool operator() (const std::string &needle) const;
    };
    static void appendStr(UnaryFeatureVec &feat_vec,
 -                        const UString &tail_str);
 +                        const std::string &tail_str);
    static void appendStr(UnaryFeatureVec::iterator begin,
                          UnaryFeatureVec::iterator end,
 -                        const UString &tail_str);
 +                        const std::string &tail_str);
    void serialiseFeatDefn(
      std::ostream &serialised, const FeatureDefn &defn) const;
    void deserialiseFeatDefn(