refactor(poet): support making sentences in left associate fashion

src/rime/gear/poet.cc

@@ -6,6 +6,8 @@
 //
 // 2011-10-06 GONG Chen <chen.sst@gmail.com>
 //
+#include <algorithm>
+#include <functional>
 #include <rime/candidate.h>
 #include <rime/config.h>
 #include <rime/dict/vocabulary.h>
@@ -21,12 +23,24 @@ inline static Grammar* create_grammar(Config* config) {
   return nullptr;
 }
 
-Poet::Poet(const Language* language, Config* config)
+Poet::Poet(const Language* language, Config* config, Compare compare)
     : language_(language),
-      grammar_(create_grammar(config)) {}
+      grammar_(create_grammar(config)),
+      compare_(compare) {}
 
 Poet::~Poet() {}
 
+bool Poet::LeftAssociateCompare(const Sentence& one, const Sentence& other) {
+  return one.weight() < other.weight() || (  // left associate if even
+      one.weight() == other.weight() && (
+          one.size() > other.size() || (  // less components is more favorable
+              one.size() == other.size() &&
+              std::lexicographical_compare(one.syllable_lengths().begin(),
+                                           one.syllable_lengths().end(),
+                                           other.syllable_lengths().begin(),
+                                           other.syllable_lengths().end()))));
+}
+
 an<Sentence> Poet::MakeSentence(const WordGraph& graph,
                                 size_t total_length) {
   // TODO: save more intermediate sentence candidates
@@ -49,9 +63,10 @@ an<Sentence> Poet::MakeSentence(const WordGraph& graph,
         auto new_sentence = New<Sentence>(*sentences[start_pos]);
         new_sentence->Extend(*entry, end_pos, is_rear, grammar_.get());
         if (sentences.find(end_pos) == sentences.end() ||
-            sentences[end_pos]->weight() < new_sentence->weight()) {
-          DLOG(INFO) << "updated sentences " << end_pos << ") with '"
-                     << new_sentence->text() << "', " << new_sentence->weight();
+            compare_(*sentences[end_pos], *new_sentence)) {
+          DLOG(INFO) << "updated sentences " << end_pos << ") with "
+                     << new_sentence->text() << " weight: "
+                     << new_sentence->weight();
           sentences[end_pos] = std::move(new_sentence);
         }
       }

src/rime/gear/poet.h

@@ -23,14 +23,24 @@ class Language;
 
 class Poet {
  public:
-  Poet(const Language* language, Config* config);
+  // sentence "less", used to compare sentences of the same input range.
+  using Compare = function<bool (const Sentence&, const Sentence&)>;
+
+  static bool CompareWeight(const Sentence& one, const Sentence& other) {
+    return one.weight() < other.weight();
+  }
+  static bool LeftAssociateCompare(const Sentence& one, const Sentence& other);
+
+  Poet(const Language* language, Config* config,
+       Compare compare = CompareWeight);
   ~Poet();
 
   an<Sentence> MakeSentence(const WordGraph& graph, size_t total_length);
 
- protected:
+ private:
   const Language* language_;
   the<Grammar> grammar_;
+  Compare compare_;
 };
 
 }  // namespace rime

src/rime/gear/table_translator.cc

@@ -221,7 +221,7 @@ TableTranslator::TableTranslator(const Ticket& ticket)
     config->GetInt(name_space_ + "/max_homographs",
                    &max_homographs_);
     if (enable_sentence_ || sentence_over_completion_) {
-      poet_.reset(new Poet(language(), config));
+      poet_.reset(new Poet(language(), config, Poet::LeftAssociateCompare));
     }
   }
   if (enable_encoder_ && user_dict_) {

src/rime/gear/translator_commons.cc

@@ -92,7 +92,7 @@ void Sentence::Extend(const DictEntry& entry,
                       size_t end_pos,
                       bool is_rear,
                       Grammar* grammar) {
-  entry_->weight += Grammar::Evaluate(entry_->text, entry, is_rear, grammar);
+  entry_->weight += Grammar::Evaluate(text(), entry, is_rear, grammar);
   entry_->text.append(entry.text);
   entry_->code.insert(entry_->code.end(),
                       entry.code.begin(),
@@ -101,7 +101,7 @@ void Sentence::Extend(const DictEntry& entry,
   syllable_lengths_.push_back(end_pos - end());
   set_end(end_pos);
   DLOG(INFO) << "extend sentence " << end_pos << ") "
-             << entry_->text << " : " << entry_->weight;
+             << text() << " weight: " << weight();
 }
 
 void Sentence::Offset(size_t offset) {

src/rime/gear/translator_commons.h

@@ -71,7 +71,9 @@ class Language;
 class Phrase : public Candidate {
  public:
   Phrase(const Language* language,
-         const string& type, size_t start, size_t end,
+         const string& type,
+         size_t start,
+         size_t end,
          const an<DictEntry>& entry)
       : Candidate(type, start, end),
         language_(language),
@@ -112,9 +114,7 @@ class Grammar;
 class Sentence : public Phrase {
  public:
   Sentence(const Language* language)
-      : Phrase(language, "sentence", 0, 0, New<DictEntry>()) {
-    entry_->weight = 0.0;
-  }
+      : Phrase(language, "sentence", 0, 0, New<DictEntry>()) {}
   Sentence(const Sentence& other)
       : Phrase(other),
         components_(other.components_),
@@ -127,6 +127,14 @@ class Sentence : public Phrase {
               Grammar* grammar);
   void Offset(size_t offset);
 
+  bool empty() const {
+    return components_.empty();
+  }
+
+  size_t size() const {
+    return components_.size();
+  }
+
   const vector<DictEntry>& components() const {
     return components_;
   }