fix(user_dictionary, contextual_translation): fix user phrase quality…

…; order contextual suggestions by type
rime · Jun 17, 2019 · 69d5c32 · 69d5c32
1 parent d9fd0cc
commit 69d5c32
Show file tree

Hide file tree

Showing 6 changed files with 21 additions and 15 deletions.
diff --git a/src/rime/dict/user_dictionary.cc b/src/rime/dict/user_dictionary.cc
@@ -485,8 +485,7 @@ an<DictEntry> UserDictionary::CreateDictEntry(const string& key,
                                   (double)v.commits / present_tick,
                                   (double)present_tick,
                                   v.dee);
-  constexpr double kUser = 13;  // log(1e8) - log(200)
-  e->weight = kUser + log(weight > 0 ? weight : DBL_EPSILON) + credibility;
+  e->weight = log(weight > 0 ? weight : DBL_EPSILON) + credibility;
   if (full_code) {
     *full_code = key.substr(0, separator_pos);
   }

diff --git a/src/rime/gear/contextual_translation.cc b/src/rime/gear/contextual_translation.cc
@@ -10,14 +10,17 @@ const int kContextualSearchLimit = 32;
 bool ContextualTranslation::Replenish() {
   vector<of<Phrase>> queue;
   size_t end_pos = 0;
+  std::string last_type;
   while (!translation_->exhausted() &&
          cache_.size() + queue.size() < kContextualSearchLimit) {
     auto cand = translation_->Peek();
     DLOG(INFO) << cand->text() << " cache/queue: "
                << cache_.size() << "/" << queue.size();
-    if (cand->type() == "phrase" || cand->type() == "table") {
-      if (end_pos != cand->end()) {
+    if (cand->type() == "phrase" || cand->type() == "user_phrase" ||
+	cand->type() == "table" || cand->type() == "user_table") {
+      if (end_pos != cand->end() || last_type != cand->type()) {
         end_pos = cand->end();
+	last_type = cand->type();
         AppendToCache(queue);
       }
       queue.push_back(Evaluate(As<Phrase>(cand)));

diff --git a/src/rime/gear/script_translator.cc b/src/rime/gear/script_translator.cc
@@ -491,7 +491,7 @@ void ScriptTranslation::PrepareCandidate() {
     DLOG(INFO) << "user phrase '" << entry->text
                << "', code length: " << user_phrase_code_length;
     cand = New<Phrase>(translator_->language(),
-                       "phrase",
+                       "user_phrase",
                        start_,
                        start_ + user_phrase_code_length,
                        entry);

diff --git a/src/rime/gear/single_char_filter.cc b/src/rime/gear/single_char_filter.cc
@@ -40,7 +40,7 @@ bool SingleCharFirstTranslation::Rearrange() {
   while (!translation_->exhausted()) {
     auto cand = translation_->Peek();
     auto phrase = As<Phrase>(Candidate::GetGenuineCandidate(cand));
-    if (!phrase || phrase->type() != "table") {
+    if (!phrase || phrase->type() != "table" || phrase->type() != "user_table") {
       break;
     }
     if (unistrlen(cand->text()) == 1) {

diff --git a/src/rime/gear/speller.cc b/src/rime/gear/speller.cc
@@ -33,14 +33,18 @@ static bool reached_max_code_length(const an<Candidate>& cand,
   return code_length >= max_code_length;
 }
 
+static inline bool is_table_entry(const an<Candidate>& cand) {
+  const auto& type = Candidate::GetGenuineCandidate(cand)->type();
+  return type == "table" || type == "user_table";
+}
+
 static bool is_auto_selectable(const an<Candidate>& cand,
                                const string& input,
                                const string& delimiters) {
   return
       // reaches end of input
       cand->end() == input.length() &&
-      // is table entry
-      Candidate::GetGenuineCandidate(cand)->type() == "table" &&
+      is_table_entry(cand) &&
       // no delimiters
       input.find_first_of(delimiters, cand->start()) == string::npos;
 }

diff --git a/src/rime/gear/table_translator.cc b/src/rime/gear/table_translator.cc
@@ -74,14 +74,12 @@ an<Candidate> TableTranslation::Peek() {
   if (options_) {
     options_->comment_formatter().Apply(&comment);
   }
-  auto phrase = New<Phrase>(
-      language_,
-      e->remaining_code_length == 0 ? "table" : "completion",
-      start_, end_, e);
+  bool incomplete = e->remaining_code_length != 0;
+  auto type = incomplete ? "completion" : is_user_phrase ? "user_table" : "table";
+  auto phrase = New<Phrase>(language_, type, start_, end_, e);
   if (phrase) {
     phrase->set_comment(comment);
     phrase->set_preedit(preedit_);
-    bool incomplete = e->remaining_code_length != 0;
     phrase->set_quality(exp(e->weight) +
                         options_->initial_quality() +
                         (incomplete ? -1 : 0) +
@@ -345,6 +343,7 @@ bool TableTranslator::Memorize(const CommitEntry& commit_entry) {
         string phrase;
         for (; it != history.rend(); ++it) {
           if (it->type != "table" &&
+              it->type != "user_table" &&
               it->type != "sentence" &&
               it->type != "uniquified")
             break;
@@ -464,7 +463,8 @@ an<Candidate> SentenceTranslation::Peek() {
   }
   size_t code_length = 0;
   an<DictEntry> entry;
-  if (PreferUserPhrase()) {
+  bool is_user_phrase = PreferUserPhrase();
+  if (is_user_phrase) {
     auto r = user_phrase_collector_.rbegin();
     code_length = r->first;
     entry = r->second[user_phrase_index_];
@@ -476,7 +476,7 @@ an<Candidate> SentenceTranslation::Peek() {
   }
   auto result = New<Phrase>(
       translator_ ? translator_->language() : NULL,
-      "table",
+      is_user_phrase ? "user_table" : "table",
       start_,
       start_ + code_length,
       entry);