From 1e14721a50e3dfc852ec89eee1efbd314b223001 Mon Sep 17 00:00:00 2001
From: fxliang <fx.liang@outlook.com>
Date: Tue, 18 Jun 2024 19:56:17 +0800
Subject: [PATCH] chore: log dict file info and line number info when deploying
 warning (#874)

---
 src/rime/dict/entry_collector.cc | 18 ++++++++++++++----
 src/rime/dict/entry_collector.h  |  4 ++++
 2 files changed, 18 insertions(+), 4 deletions(-)

diff --git a/src/rime/dict/entry_collector.cc b/src/rime/dict/entry_collector.cc
index e9fa7404a9..48d9402753 100644
--- a/src/rime/dict/entry_collector.cc
+++ b/src/rime/dict/entry_collector.cc
@@ -56,6 +56,8 @@ void EntryCollector::LoadPresetVocabulary(DictSettings* settings) {
 
 void EntryCollector::Collect(const path& dict_file) {
   LOG(INFO) << "collecting entries from " << dict_file;
+  current_dict_file = dict_file.u8string();
+  line_number = 0;
   // read table
   std::ifstream fin(dict_file.c_str());
   DictSettings settings;
@@ -69,13 +71,15 @@ void EntryCollector::Collect(const path& dict_file) {
   int weight_column = settings.GetColumnIndex("weight");
   int stem_column = settings.GetColumnIndex("stem");
   if (text_column == -1) {
-    LOG(ERROR) << "missing text column definition.";
+    LOG(ERROR) << "missing text column definition in file: " << dict_file
+               << ".";
     return;
   }
   bool enable_comment = true;
   string line;
   while (getline(fin, line)) {
     boost::algorithm::trim_right(line);
+    line_number++;
     // skip empty lines and comments
     if (line.empty())
       continue;
@@ -90,7 +94,9 @@ void EntryCollector::Collect(const path& dict_file) {
     auto row = strings::split(line, "\t");
     int num_columns = static_cast<int>(row.size());
     if (num_columns <= text_column || row[text_column].empty()) {
-      LOG(WARNING) << "Missing entry text at #" << num_entries << ".";
+      LOG(WARNING) << "Missing entry text at #" << num_entries
+                   << ", line: " << line_number
+                   << " of file: " << current_dict_file << ".";
       continue;
     }
     const auto& word(row[text_column]);
@@ -168,7 +174,9 @@ void EntryCollector::CreateEntry(const string& word,
     try {
       percentage = std::stod(weight_str.substr(0, weight_str.length() - 1));
     } catch (...) {
-      LOG(WARNING) << "invalid entry definition at #" << num_entries << ".";
+      LOG(WARNING) << "invalid entry definition at #" << num_entries
+                   << ", line: " << line_number
+                   << " of file: " << current_dict_file << ".";
       percentage = 100.0;
     }
     e->weight *= percentage / 100.0;
@@ -176,7 +184,9 @@ void EntryCollector::CreateEntry(const string& word,
     try {
       e->weight = std::stod(weight_str);
     } catch (...) {
-      LOG(WARNING) << "invalid entry definition at #" << num_entries << ".";
+      LOG(WARNING) << "invalid entry definition at #" << num_entries
+                   << ", line: " << line_number
+                   << " of file: " << current_dict_file << ".";
       e->weight = 0.0;
     }
   }
diff --git a/src/rime/dict/entry_collector.h b/src/rime/dict/entry_collector.h
index d79cc51d75..dc39cb799c 100644
--- a/src/rime/dict/entry_collector.h
+++ b/src/rime/dict/entry_collector.h
@@ -71,6 +71,10 @@ class EntryCollector : public PhraseCollector {
   set<string /* word */> collection;
   WordMap words;
   WeightMap total_weight;
+
+ private:
+  string current_dict_file;
+  size_t line_number;
 };
 
 }  // namespace rime