From 74d74210f83ec7fb54c818c7769830d2d98a604f Mon Sep 17 00:00:00 2001
From: groverlynn <groverlynn@gmail.com>
Date: Fri, 9 Feb 2024 14:05:06 +0100
Subject: [PATCH] chain conversion (#715)

---
 src/rime/gear/simplifier.cc | 66 ++++++++++++++++++++++++++-----------
 1 file changed, 47 insertions(+), 19 deletions(-)

diff --git a/src/rime/gear/simplifier.cc b/src/rime/gear/simplifier.cc
index b257721355..073e3f0e75 100644
--- a/src/rime/gear/simplifier.cc
+++ b/src/rime/gear/simplifier.cc
@@ -65,11 +65,30 @@ class Opencc {
         opencc::Optional<const opencc::DictEntry*> item =
             dict->Match(original_word);
         if (item.IsNull()) {
-          // Current dictionary doesn't convert the word. We need to keep it for
-          // other dicts in the chain. e.g. s2t.json expands 里 to 里 and 裏,
-          // then t2tw.json passes 里 as-is and converts 裏 to 裡.
-          if (word_set.insert(original_word).second) {
-            converted_words.push_back(original_word);
+          // There is no exact match, but still need to convert partially
+          // matched in a chain conversion. Here apply default (max. seg.)
+          // match to get the most probable conversion result
+          std::ostringstream buffer;
+          for (const char* wstr = original_word.c_str(); *wstr != '\0';) {
+            opencc::Optional<const opencc::DictEntry*> matched =
+                dict->MatchPrefix(wstr);
+            size_t matched_length;
+            if (matched.IsNull()) {
+              matched_length = opencc::UTF8Util::NextCharLength(wstr);
+              buffer << opencc::UTF8Util::FromSubstr(wstr, matched_length);
+            } else {
+              matched_length = matched.Get()->KeyLength();
+              buffer << matched.Get()->GetDefault();
+            }
+            wstr += matched_length;
+          }
+          const string& converted_word = buffer.str();
+          // Even if current dictionary doesn't convert the word
+          // (converted_word == original_word), we still need to keep it for
+          // subsequent dicts in the chain. e.g. s2t.json expands 里 to 里 and
+          // 裏, then t2tw.json passes 里 as-is and converts 裏 to 裡.
+          if (word_set.insert(converted_word).second) {
+            converted_words.push_back(converted_word);
           }
           continue;
         }
@@ -94,23 +113,32 @@ class Opencc {
   bool RandomConvertText(const string& text, string* simplified) {
     if (dict_ == nullptr)
       return false;
+    const list<opencc::ConversionPtr> conversions =
+        converter_->GetConversionChain()->GetConversions();
     const char* phrase = text.c_str();
-    std::ostringstream buffer;
-    for (const char* pstr = phrase; *pstr != '\0';) {
-      opencc::Optional<const opencc::DictEntry*> matched =
-          dict_->MatchPrefix(pstr);
-      size_t matchedLength;
-      if (matched.IsNull()) {
-        matchedLength = opencc::UTF8Util::NextCharLength(pstr);
-        buffer << opencc::UTF8Util::FromSubstr(pstr, matchedLength);
-      } else {
-        matchedLength = matched.Get()->KeyLength();
-        size_t i = rand() % (matched.Get()->NumValues());
-        buffer << matched.Get()->Values().at(i);
+    for (auto conversion : conversions) {
+      opencc::DictPtr dict = conversion->GetDict();
+      if (dict == nullptr) {
+        return false;
+      }
+      std::ostringstream buffer;
+      for (const char* pstr = phrase; *pstr != '\0';) {
+        opencc::Optional<const opencc::DictEntry*> matched =
+            dict->MatchPrefix(pstr);
+        size_t matched_length;
+        if (matched.IsNull()) {
+          matched_length = opencc::UTF8Util::NextCharLength(pstr);
+          buffer << opencc::UTF8Util::FromSubstr(pstr, matched_length);
+        } else {
+          matched_length = matched.Get()->KeyLength();
+          size_t i = rand() % (matched.Get()->NumValues());
+          buffer << matched.Get()->Values().at(i);
+        }
+        pstr += matched_length;
       }
-      pstr += matchedLength;
+      *simplified = buffer.str();
+      phrase = simplified->c_str();
     }
-    *simplified = buffer.str();
     return *simplified != text;
   }