fix(script_translator): excluded words cause crashes

Author: ksqsf

src/rime/dict/dictionary.cc

@@ -197,7 +197,7 @@ bool DictEntryIterator::Next() {
       return false;
     }
   } while (filter_ && !filter_(Peek()));
-  return true;
+  return entry_.get();
 }
 
 // Note: does not apply filters

src/rime/gear/script_translator.cc

@@ -569,82 +569,93 @@ inline static bool prefer_user_phrase(
 }
 
 bool ScriptTranslation::PrepareCandidate() {
-  if (exhausted()) {
-    candidate_source_ = kUninitialized;
-    candidate_ = nullptr;
-    return false;
-  }
-  if (sentence_) {
-    candidate_source_ = kSentence;
-    candidate_ = sentence_;
-    return true;
-  }
-  const size_t full_code_length = end_of_input_ - start_;
-  size_t user_phrase_code_length = 0;
-  if (user_phrase_ && user_phrase_iter_ != user_phrase_->rend()) {
-    user_phrase_code_length = user_phrase_iter_->first;
-  }
-  size_t phrase_code_length = 0;
-  if (phrase_ && phrase_iter_ != phrase_->rend()) {
-    phrase_code_length = phrase_iter_->first;
-  }
-  if (user_phrase_code_length > 0 &&
-      prefer_user_phrase(
-          user_phrase_code_length, phrase_code_length,
-          // 編碼長度相同時, 用戶詞優先
-          [this, full_code_length, phrase_code_length]() {
-            UserDictEntryIterator& uter = user_phrase_iter_->second;
-            DictEntryIterator& iter = phrase_iter_->second;
-            // 但若用戶詞爲糾錯結果, 則免去優先權, 回歸權重比較
-            bool user_is_correction = syllabifier_->IsCorrection(
-                uter.Peek()->code, phrase_code_length);
-            if (user_is_correction) {
-              // 因對糾錯候選施加了較大懲罰, 通常結果爲 系統原文 > 用戶糾錯
-              // 兩者皆糾錯時, 用戶詞的動態權重有助達成 用戶糾錯 > 系統糾錯
-              return uter.Peek()->weight >= iter.Peek()->weight;
-            }
-            // 長詞聯想之前須至少出一個嚴格匹配的候選
-            // 故確定首選之際, 系統嚴格匹配 > 用戶長詞聯想
-            const int kNumExactMatchOnTop = 1;
-            return candidate_index_ >= kNumExactMatchOnTop ||
-                   prefer_user_phrase(
-                       has_exact_match_phrase(user_phrase_, user_phrase_iter_,
-                                              full_code_length),
-                       has_exact_match_phrase(phrase_, phrase_iter_,
-                                              full_code_length));
-          })) {
-    UserDictEntryIterator& uter = user_phrase_iter_->second;
-    const auto& entry = uter.Peek();
-    DLOG(INFO) << "user phrase '" << entry->text
-               << "', code length: " << user_phrase_code_length;
-    candidate_source_ = kUserPhrase;
-    candidate_ =
-        New<Phrase>(translator_->language(),
-                    entry->IsPredictiveMatch() ? "completion" : "user_phrase",
-                    start_, start_ + user_phrase_code_length, entry);
-    candidate_->set_quality(std::exp(entry->weight) +
-                            translator_->initial_quality() +
-                            (entry->quality_len / full_code_length));
-    return true;
-  } else if (phrase_code_length > 0) {
-    DictEntryIterator& iter = phrase_iter_->second;
-    const auto& entry = iter.Peek();
-    DLOG(INFO) << "phrase '" << entry->text
-               << "', code length: " << phrase_code_length;
-    candidate_source_ = kSysPhrase;
-    candidate_ =
-        New<Phrase>(translator_->language(),
-                    entry->IsPredictiveMatch() ? "completion" : "phrase",
-                    start_, start_ + phrase_code_length, entry);
-    candidate_->set_quality(std::exp(entry->weight) +
-                            translator_->initial_quality() +
-                            (entry->quality_len / full_code_length));
-    return true;
-  } else {
-    candidate_source_ = kUninitialized;
-    candidate_ = nullptr;
-    return false;
-  }
+  do {
+    if (exhausted()) {
+      candidate_source_ = kUninitialized;
+      candidate_ = nullptr;
+      return false;
+    }
+    if (sentence_) {
+      candidate_source_ = kSentence;
+      candidate_ = sentence_;
+      return true;
+    }
+    const size_t full_code_length = end_of_input_ - start_;
+    size_t user_phrase_code_length = 0;
+    if (user_phrase_ && user_phrase_iter_ != user_phrase_->rend()) {
+      user_phrase_code_length = user_phrase_iter_->first;
+    }
+    size_t phrase_code_length = 0;
+    if (phrase_ && phrase_iter_ != phrase_->rend()) {
+      phrase_code_length = phrase_iter_->first;
+    }
+    if (user_phrase_code_length > 0 &&
+        prefer_user_phrase(
+            user_phrase_code_length, phrase_code_length,
+            // 編碼長度相同時, 用戶詞優先
+            [this, full_code_length, phrase_code_length]() {
+              UserDictEntryIterator& uter = user_phrase_iter_->second;
+              DictEntryIterator& iter = phrase_iter_->second;
+              // 但若用戶詞爲糾錯結果, 則免去優先權, 回歸權重比較
+              bool user_is_correction = syllabifier_->IsCorrection(
+                  uter.Peek()->code, phrase_code_length);
+              if (user_is_correction) {
+                // 因對糾錯候選施加了較大懲罰, 通常結果爲 系統原文 > 用戶糾錯
+                // 兩者皆糾錯時, 用戶詞的動態權重有助達成 用戶糾錯 > 系統糾錯
+                return uter.Peek()->weight >= iter.Peek()->weight;
+              }
+              // 長詞聯想之前須至少出一個嚴格匹配的候選
+              // 故確定首選之際, 系統嚴格匹配 > 用戶長詞聯想
+              const int kNumExactMatchOnTop = 1;
+              return candidate_index_ >= kNumExactMatchOnTop ||
+                     prefer_user_phrase(
+                         has_exact_match_phrase(user_phrase_, user_phrase_iter_,
+                                                full_code_length),
+                         has_exact_match_phrase(phrase_, phrase_iter_,
+                                                full_code_length));
+            })) {
+      UserDictEntryIterator& uter = user_phrase_iter_->second;
+      const auto& entry = uter.Peek();
+      if (!entry) {
+        ++user_phrase_iter_;
+        continue;
+      }
+      DLOG(INFO) << "user phrase '" << entry->text
+                 << "', code length: " << user_phrase_code_length;
+      candidate_source_ = kUserPhrase;
+      candidate_ =
+          New<Phrase>(translator_->language(),
+                      entry->IsPredictiveMatch() ? "completion" : "user_phrase",
+                      start_, start_ + user_phrase_code_length, entry);
+      candidate_->set_quality(std::exp(entry->weight) +
+                              translator_->initial_quality() +
+                              (entry->quality_len / full_code_length));
+      return true;
+    } else if (phrase_code_length > 0) {
+      DictEntryIterator& iter = phrase_iter_->second;
+      const auto& entry = iter.Peek();
+      if (!entry) {
+        ++phrase_iter_;
+        continue;
+      }
+      DLOG(INFO) << "phrase '" << entry->text
+                 << "', code length: " << phrase_code_length;
+      candidate_source_ = kSysPhrase;
+      candidate_ =
+          New<Phrase>(translator_->language(),
+                      entry->IsPredictiveMatch() ? "completion" : "phrase",
+                      start_, start_ + phrase_code_length, entry);
+      candidate_->set_quality(std::exp(entry->weight) +
+                              translator_->initial_quality() +
+                              (entry->quality_len / full_code_length));
+      return true;
+    } else {
+      candidate_source_ = kUninitialized;
+      candidate_ = nullptr;
+      return false;
+    }
+  } while (!exhausted() && !candidate_);
+  return false;
 }
 
 bool ScriptTranslation::CheckEmpty() {