From a560dffa321af2370454ffe54581ec2f1aedbe5d Mon Sep 17 00:00:00 2001
From: Qijia Liu <liumeo@pku.edu.cn>
Date: Mon, 7 Aug 2023 00:46:04 -0400
Subject: [PATCH] perf: replace rbtree with vector of pair (#684)

---
 src/rime/dict/entry_collector.cc | 10 ++++++++--
 src/rime/dict/entry_collector.h  |  6 ++++--
 2 files changed, 12 insertions(+), 4 deletions(-)

diff --git a/src/rime/dict/entry_collector.cc b/src/rime/dict/entry_collector.cc
index fe864fb31e..cdb7625c5e 100644
--- a/src/rime/dict/entry_collector.cc
+++ b/src/rime/dict/entry_collector.cc
@@ -4,7 +4,9 @@
 //
 // 2011-11-27 GONG Chen <chen.sst@gmail.com>
 //
+#include <algorithm>
 #include <fstream>
+#include <utility>
 #include <boost/algorithm/string.hpp>
 #include <boost/lexical_cast.hpp>
 #include <rime/algo/strings.h>
@@ -192,12 +194,14 @@ void EntryCollector::CreateEntry(const string& word,
   bool is_word = (e->raw_code.size() == 1);
   if (is_word) {
     auto& weights = words[e->text];
-    if (weights.find(code_str) != weights.end()) {
+    if (std::find_if(weights.begin(), weights.end(), [&](const auto& p) {
+          return p.first == code_str;
+        }) != weights.end()) {
       LOG(WARNING) << "duplicate word definition '" << e->text << "': ["
                    << code_str << "].";
       return;
     }
-    weights[code_str] += e->weight;
+    weights.push_back(std::make_pair(code_str, e->weight));
     total_weight[e->text] += e->weight;
   }
   entries.emplace_back(std::move(e));
@@ -214,6 +218,8 @@ bool EntryCollector::TranslateWord(const string& word, vector<string>* result) {
   }
   const auto& w = words.find(word);
   if (w != words.end()) {
+    std::sort(w->second.begin(), w->second.end(),
+              [](const auto& a, const auto& b) { return a.first < b.first; });
     for (const auto& v : w->second) {
       const double kMinimalWeight = 0.05;  // 5%
       double min_weight = total_weight[word] * kMinimalWeight;
diff --git a/src/rime/dict/entry_collector.h b/src/rime/dict/entry_collector.h
index b7b6105709..17f7267986 100644
--- a/src/rime/dict/entry_collector.h
+++ b/src/rime/dict/entry_collector.h
@@ -23,8 +23,10 @@ struct RawDictEntry {
 
 // code -> weight
 using WeightMap = map<string, double>;
-// word -> { code -> weight }
-using WordMap = hash_map<string, WeightMap>;
+// word -> [ { code, weight } ]
+// For the sake of memory usage, don't use word -> { code -> weight } as there
+// may be many words, but may not be many representations for a word
+using WordMap = hash_map<string, vector<pair<string, double>>>;
 // [ (word, weight), ... ]
 using EncodeQueue = std::queue<pair<string, string>>;