Merge remote-tracking branch 'fork/master' into disconnect_userdict

Author: shewer

.github/workflows/linux-build.yml

@@ -9,7 +9,7 @@ on:
 
 jobs:
   build:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-24.04
     strategy:
       fail-fast: false
       matrix:

README.md

@@ -41,7 +41,7 @@ Build dependencies
   - compiler with C++17 support
   - cmake>=3.12
   - libboost>=1.74
-  - libglog (optional)
+  - libglog>=0.7 (optional)
   - libleveldb
   - libmarisa
   - libopencc>=1.0.2
@@ -75,6 +75,7 @@ Official:
 Community:
   - [emacs-rime](https://github.com/DogLooksGood/emacs-rime): frontend for Emacs
   - [coc-rime](https://github.com/tonyfettes/coc-rime): frontend for Vim
+  - [zsh-rime](https://github.com/Freed-Wu/zsh-rime): frontend for Zsh
   - [fcitx-rime](https://github.com/fcitx/fcitx-rime): Fcitx frontend for Linux
   - [fcitx5-rime](https://github.com/fcitx/fcitx5-rime): Fcitx5 frontend for Linux
   - [Hamster](https://github.com/imfuxiao/Hamster): frontend for iOS

src/rime/common.h

@@ -14,9 +14,9 @@
 #include <list>
 #include <map>
 #include <memory>
+#include <optional>
 #include <set>
 #include <string>
-#include <utility>
 #include <unordered_map>
 #include <unordered_set>
 #include <utility>

src/rime/config/config_data.cc

@@ -66,15 +66,16 @@ bool ConfigData::LoadFromFile(const path& file_path, ConfigCompiler* compiler) {
   modified_ = false;
   root.reset();
   if (!std::filesystem::exists(file_path)) {
-    LOG(WARNING) << "nonexistent config file '" << file_path << "'.";
+    if (!boost::ends_with(file_path.u8string(), ".custom.yaml"))
+      LOG(WARNING) << "nonexistent config file '" << file_path << "'.";
     return false;
   }
   LOG(INFO) << "loading config file '" << file_path << "'.";
   try {
     YAML::Node doc = YAML::LoadFile(file_path.string());
     root = ConvertFromYaml(doc, compiler);
   } catch (YAML::Exception& e) {
-    LOG(ERROR) << "Error parsing YAML: " << e.what();
+    LOG(ERROR) << "Error parsing YAML \"" << file_path << "\" : " << e.what();
     return false;
   }
   return true;

src/rime/context.cc

@@ -143,27 +143,27 @@ bool Context::Highlight(size_t index) {
   return true;
 }
 
-bool Context::DeleteCandidate(
-    function<an<Candidate>(Segment& seg)> get_candidate) {
+bool Context::DeleteCandidate(std::optional<size_t> index) {
   if (composition_.empty())
     return false;
   Segment& seg(composition_.back());
-  if (auto cand = get_candidate(seg)) {
-    DLOG(INFO) << "Deleting candidate: '" << cand->text();
-    delete_notifier_(this);
-    return true;  // CAVEAT: this doesn't mean anything is deleted for sure
+  auto cand = index ? seg.GetCandidateAt(*index) : seg.GetSelectedCandidate();
+  if (!cand)
+    return false;
+  DLOG(INFO) << "Deleting candidate: " << cand->text();
+  if (index) {
+    seg.selected_index = *index;
   }
-  return false;
+  delete_notifier_(this);
+  return true;  // CAVEAT: this doesn't mean anything is deleted for sure
 }
 
 bool Context::DeleteCandidate(size_t index) {
-  return DeleteCandidate(
-      [index](Segment& seg) { return seg.GetCandidateAt(index); });
+  return DeleteCandidate(std::optional{index});
 }
 
 bool Context::DeleteCurrentSelection() {
-  return DeleteCandidate(
-      [](Segment& seg) { return seg.GetSelectedCandidate(); });
+  return DeleteCandidate({});
 }
 
 bool Context::ConfirmCurrentSelection() {

src/rime/context.h

@@ -74,6 +74,8 @@ class RIME_API Context {
   bool get_option(const string& name) const;
   void set_property(const string& name, const string& value);
   string get_property(const string& name) const;
+  const map<string, bool>& options() const { return options_; }
+  const map<string, string>& properties() const { return properties_; }
   // options and properties starting with '_' are local to schema;
   // others are session scoped.
   void ClearTransientOptions();
@@ -92,7 +94,7 @@ class RIME_API Context {
 
  private:
   string GetSoftCursor() const;
-  bool DeleteCandidate(function<an<Candidate>(Segment& seg)> get_candidate);
+  bool DeleteCandidate(std::optional<size_t> index);
 
   string input_;
   size_t caret_pos_ = 0;

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]);
@@ -168,15 +174,19 @@ 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;
   } else if (!weight_str.empty()) {  // absolute weight
     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;
     }
   }

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

src/rime/gear/matcher.cc

@@ -16,8 +16,11 @@ Matcher::Matcher(const Ticket& ticket) : Segmentor(ticket) {
   // read schema settings
   if (!ticket.schema)
     return;
+  if (name_space_ == "segmentor") {
+    name_space_ = "recognizer";
+  }
   Config* config = ticket.schema->config();
-  patterns_.LoadConfig(config);
+  patterns_.LoadConfig(config, name_space_);
 }
 
 bool Matcher::Proceed(Segmentation* segmentation) {

src/rime/gear/recognizer.cc

@@ -32,8 +32,8 @@ static void load_patterns(RecognizerPatterns* patterns, an<ConfigMap> map) {
   }
 }
 
-void RecognizerPatterns::LoadConfig(Config* config) {
-  load_patterns(this, config->GetMap("recognizer/patterns"));
+void RecognizerPatterns::LoadConfig(Config* config, const string& name_space) {
+  load_patterns(this, config->GetMap(name_space + "/patterns"));
 }
 
 RecognizerMatch RecognizerPatterns::GetMatch(
@@ -72,9 +72,12 @@ RecognizerMatch RecognizerPatterns::GetMatch(
 Recognizer::Recognizer(const Ticket& ticket) : Processor(ticket) {
   if (!ticket.schema)
     return;
+  if (name_space_ == "processor") {
+    name_space_ = "recognizer";
+  }
   if (Config* config = ticket.schema->config()) {
-    patterns_.LoadConfig(config);
-    config->GetBool("recognizer/use_space", &use_space_);
+    patterns_.LoadConfig(config, name_space_);
+    config->GetBool(name_space_ + "/use_space", &use_space_);
   }
 }
 

src/rime/gear/recognizer.h

@@ -29,7 +29,7 @@ struct RecognizerMatch {
 
 class RecognizerPatterns : public map<string, boost::regex> {
  public:
-  void LoadConfig(Config* config);
+  void LoadConfig(Config* config, const string& name_space);
   RecognizerMatch GetMatch(const string& input,
                            const Segmentation& segmentation) const;
 };

src/rime/gear/speller.cc

@@ -38,12 +38,17 @@ static inline bool is_table_entry(const an<Candidate>& cand) {
   return type == "table" || type == "user_table";
 }
 
+static inline bool is_simple_candidate(const an<Candidate>& cand) {
+  return bool(As<SimpleCandidate>(cand));
+}
+
 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(cand) &&
+      cand->end() == input.length() &&
+      (is_table_entry(cand) || is_simple_candidate(cand)) &&
       // no delimiters
       input.find_first_of(delimiters, cand->start()) == string::npos;
 }

src/rime/setup.cc

@@ -84,7 +84,11 @@ RIME_API void SetupLogging(const char* app_name,
   // Do not allow other users to read/write log files created by current
   // process.
   FLAGS_logfile_mode = 0600;
-  google::InitGoogleLogging(app_name);
+  if (google::IsGoogleLoggingInitialized()) {
+    LOG(WARNING) << "Glog is already initialized.";
+  } else {
+    google::InitGoogleLogging(app_name);
+  }
 #endif  // RIME_ENABLE_LOGGING
 }
 

tools/rime_api_console.cc

@@ -165,6 +165,26 @@ bool execute_special_command(const char* line, RimeSessionId session_id) {
   if (!strcmp(line, "synchronize")) {
     return rime->sync_user_data();
   }
+  const char* kDeleteCandidateOnCurrentPage = "delete on current page ";
+  command_length = strlen(kDeleteCandidateOnCurrentPage);
+  if (!strncmp(line, kDeleteCandidateOnCurrentPage, command_length)) {
+    const char* index_str = line + command_length;
+    int index = atoi(index_str);
+    if (!rime->delete_candidate_on_current_page(session_id, index)) {
+      fprintf(stderr, "failed to delete\n");
+    }
+    return true;
+  }
+  const char* kDeleteCandidate = "delete ";
+  command_length = strlen(kDeleteCandidate);
+  if (!strncmp(line, kDeleteCandidate, command_length)) {
+    const char* index_str = line + command_length;
+    int index = atoi(index_str);
+    if (!rime->delete_candidate(session_id, index)) {
+      fprintf(stderr, "failed to delete\n");
+    }
+    return true;
+  }
   return false;
 }