feat(editor): implement SimpleEngine which doesn't use intelligent ph…

…rasing

capi/src/io.rs

@@ -9,16 +9,16 @@ use std::{
 };
 
 use chewing::{
-    conversion::{ChewingEngine, Interval, Symbol},
+    conversion::{ChewingEngine, Interval, SimpleEngine, Symbol},
     dictionary::{Layered, SystemDictionaryLoader, UserDictionaryLoader},
     editor::{
         keyboard::{AnyKeyboardLayout, KeyCode, KeyboardLayout, Modifiers, Qwerty},
         zhuyin_layout::{
             DaiChien26, Et, Et26, GinYieh, Hsu, Ibm, KeyboardLayoutCompat, Pinyin, Standard,
             SyllableEditor,
         },
-        BasicEditor, CharacterForm, Editor, EditorKeyBehavior, LanguageMode, LaxUserFreqEstimate,
-        UserPhraseAddDirection,
+        BasicEditor, CharacterForm, ConversionEngineKind, Editor, EditorKeyBehavior, LanguageMode,
+        LaxUserFreqEstimate, UserPhraseAddDirection,
     },
     zhuyin::Syllable,
 };
@@ -171,7 +171,7 @@ pub unsafe extern "C" fn chewing_new2(
     let estimate = LaxUserFreqEstimate::max_from(user_dictionary.as_ref());
 
     let dict = Layered::new(dictionaries, user_dictionary);
-    let conversion_engine = ChewingEngine::new();
+    let conversion_engine = Box::new(ChewingEngine::new());
     let kb_compat = KeyboardLayoutCompat::Default;
     let keyboard = AnyKeyboardLayout::Qwerty(Qwerty);
     let editor = Editor::new(conversion_engine, dict, estimate, abbrev, sym_sel);
@@ -303,7 +303,8 @@ pub unsafe extern "C" fn chewing_config_has_option(
         | "chewing.selection_keys"
         | "chewing.character_form"
         | "chewing.space_is_select_key"
-        | "chewing.fuzzy_search_mode" => true,
+        | "chewing.fuzzy_search_mode"
+        | "cheiwng.conversion_engine" => true,
         _ => false,
     };
 
@@ -346,6 +347,10 @@ pub unsafe extern "C" fn chewing_config_get_int(
         },
         "chewing.space_is_select_key" => option.space_is_select_key as c_int,
         "chewing.fuzzy_search_mode" => option.fuzzy_search as c_int,
+        "chewing.conversion_engine" => match option.conversion_engine {
+            ConversionEngineKind::ChewingEngine => 0,
+            ConversionEngineKind::SimpleEngine => 1,
+        },
         _ => ERROR,
     }
 }
@@ -438,6 +443,21 @@ pub unsafe extern "C" fn chewing_config_set_int(
             ensure_bool!(value);
             options.fuzzy_search = value > 0;
         }
+        "chewing.conversion_engine" => {
+            options.conversion_engine = match value {
+                0 => {
+                    ctx.editor
+                        .set_conversion_engine(Box::new(ChewingEngine::new()));
+                    ConversionEngineKind::ChewingEngine
+                }
+                1 => {
+                    ctx.editor
+                        .set_conversion_engine(Box::new(SimpleEngine::new()));
+                    ConversionEngineKind::SimpleEngine
+                }
+                _ => return ERROR,
+            }
+        }
         _ => return ERROR,
     };
 

src/conversion/chewing.rs

@@ -9,7 +9,7 @@ use log::trace;
 
 use crate::dictionary::{Dictionary, Phrase};
 
-use super::{Composition, Gap, Interval, Symbol};
+use super::{Composition, ConversionEngine, Gap, Interval, Symbol};
 
 /// TODO: doc
 #[derive(Debug, Default)]
@@ -51,6 +51,76 @@ impl ChewingEngine {
     }
 }
 
+impl ConversionEngine for ChewingEngine {
+    fn convert<'a>(
+        &'a self,
+        dict: &'a dyn Dictionary,
+        comp: &'a Composition,
+    ) -> Box<dyn Iterator<Item = Vec<Interval>> + 'a> {
+        Box::new(ChewingEngine::convert(self, dict, comp))
+    }
+}
+
+/// Simple engine does not perform any intelligent conversion.
+#[derive(Debug, Default)]
+pub struct SimpleEngine;
+
+impl SimpleEngine {
+    pub fn new() -> SimpleEngine {
+        SimpleEngine
+    }
+    pub fn convert<'a>(
+        &'a self,
+        dict: &'a dyn Dictionary,
+        comp: &'a Composition,
+    ) -> impl Iterator<Item = Vec<Interval>> + Clone + 'a {
+        let mut intervals = vec![];
+
+        for (i, sym) in comp.symbols().iter().enumerate() {
+            if comp
+                .selections
+                .iter()
+                .any(|selection| selection.intersect_range(i, i + 1))
+            {
+                continue;
+            }
+            if sym.is_char() {
+                intervals.push(Interval {
+                    start: i,
+                    end: i + 1,
+                    is_phrase: false,
+                    str: sym.to_char().unwrap().to_string().into_boxed_str(),
+                });
+            } else {
+                let phrase = dict.lookup_first_phrase(&[sym.to_syllable().unwrap()]);
+                let phrase_str = phrase.map_or_else(
+                    || sym.to_syllable().unwrap().to_string(),
+                    |phrase| phrase.to_string(),
+                );
+                intervals.push(Interval {
+                    start: i,
+                    end: i + 1,
+                    is_phrase: true,
+                    str: phrase_str.into_boxed_str(),
+                })
+            }
+        }
+        intervals.extend_from_slice(comp.selections());
+        intervals.sort_by_key(|int| int.start);
+        iter::once(intervals)
+    }
+}
+
+impl ConversionEngine for SimpleEngine {
+    fn convert<'a>(
+        &'a self,
+        dict: &'a dyn Dictionary,
+        comp: &'a Composition,
+    ) -> Box<dyn Iterator<Item = Vec<Interval>> + 'a> {
+        Box::new(SimpleEngine::convert(self, dict, comp))
+    }
+}
+
 fn glue_fn(com: &Composition, mut acc: Vec<Interval>, interval: Interval) -> Vec<Interval> {
     if acc.is_empty() {
         acc.push(interval);

src/conversion/mod.rs

@@ -9,11 +9,22 @@ use std::{
     fmt::Debug,
 };
 
-use crate::zhuyin::{Syllable, SyllableSlice};
+use crate::{
+    dictionary::Dictionary,
+    zhuyin::{Syllable, SyllableSlice},
+};
 
-pub use self::chewing::ChewingEngine;
+pub use self::chewing::{ChewingEngine, SimpleEngine};
 pub(crate) use self::symbol::{full_width_symbol_input, special_symbol_input};
 
+pub trait ConversionEngine: Debug {
+    fn convert<'a>(
+        &'a self,
+        dict: &'a dyn Dictionary,
+        comp: &'a Composition,
+    ) -> Box<dyn Iterator<Item = Vec<Interval>> + 'a>;
+}
+
 /// TODO: doc
 #[derive(Default, PartialEq, Eq, PartialOrd, Ord, Hash, Clone)]
 pub struct Interval {

src/editor/mod.rs

@@ -19,7 +19,10 @@ pub use estimate::{LaxUserFreqEstimate, UserFreqEstimate};
 use log::{debug, info, trace, warn};
 
 use crate::{
-    conversion::{full_width_symbol_input, special_symbol_input, ChewingEngine, Interval, Symbol},
+    conversion::{
+        full_width_symbol_input, special_symbol_input, ChewingEngine, ConversionEngine, Interval,
+        Symbol,
+    },
     dictionary::{
         Dictionary, DictionaryMut, Layered, LookupStrategy, SystemDictionaryLoader,
         UpdateDictionaryError, UserDictionaryLoader,
@@ -53,6 +56,12 @@ pub enum UserPhraseAddDirection {
     Backward,
 }
 
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum ConversionEngineKind {
+    ChewingEngine,
+    SimpleEngine,
+}
+
 #[derive(Debug, Clone, Copy)]
 pub struct EditorOptions {
     pub easy_symbol_input: bool,
@@ -67,6 +76,7 @@ pub struct EditorOptions {
     pub character_form: CharacterForm,
     pub user_phrase_add_dir: UserPhraseAddDirection,
     pub fuzzy_search: bool,
+    pub conversion_engine: ConversionEngineKind,
 }
 
 impl Default for EditorOptions {
@@ -84,6 +94,8 @@ impl Default for EditorOptions {
             character_form: CharacterForm::Halfwidth,
             user_phrase_add_dir: UserPhraseAddDirection::Forward,
             fuzzy_search: false,
+            // FIXME may be out of sync with the engine used
+            conversion_engine: ConversionEngineKind::ChewingEngine,
         }
     }
 }
@@ -160,7 +172,7 @@ impl Error for EditorError {}
 pub(crate) struct SharedState {
     com: CompositionEditor,
     syl: Box<dyn SyllableEditor>,
-    conv: ChewingEngine,
+    conv: Box<dyn ConversionEngine>,
     dict: Layered,
     abbr: AbbrevTable,
     sym_sel: SymbolSelector,
@@ -180,15 +192,15 @@ impl Editor {
         let user_dict = UserDictionaryLoader::new().load()?;
         let estimate = LaxUserFreqEstimate::max_from(user_dict.as_ref());
         let dict = Layered::new(system_dict, user_dict);
-        let conversion_engine = ChewingEngine::new();
+        let conversion_engine = Box::new(ChewingEngine::new());
         let abbrev = SystemDictionaryLoader::new().load_abbrev()?;
         let sym_sel = SystemDictionaryLoader::new().load_symbol_selector()?;
         let editor = Editor::new(conversion_engine, dict, estimate, abbrev, sym_sel);
         Ok(editor)
     }
 
     pub fn new(
-        conv: ChewingEngine,
+        conv: Box<dyn ConversionEngine>,
         dict: Layered,
         estimate: LaxUserFreqEstimate,
         abbr: AbbrevTable,
@@ -218,6 +230,10 @@ impl Editor {
         self.shared.syl = syl;
         info!("Set syllable editor: {:?}", self.shared.syl);
     }
+    pub fn set_conversion_engine(&mut self, engine: Box<dyn ConversionEngine>) {
+        self.shared.conv = engine;
+        info!("Set conversion engine: {:?}", self.shared.conv);
+    }
     pub fn clear(&mut self) {
         self.state = Box::new(Entering);
         self.shared.clear();
@@ -1486,7 +1502,7 @@ mod tests {
             vec![Box::new(TrieBuf::new_in_memory())],
             Box::new(TrieBuf::new_in_memory()),
         );
-        let conversion_engine = ChewingEngine::new();
+        let conversion_engine = Box::new(ChewingEngine::new());
         let estimate = LaxUserFreqEstimate::new(0);
         let abbrev = AbbrevTable::new();
         let sym_sel = SymbolSelector::default();
@@ -1513,7 +1529,7 @@ mod tests {
             vec![("冊", 100)],
         )]);
         let dict = Layered::new(vec![Box::new(dict)], Box::new(TrieBuf::new_in_memory()));
-        let conversion_engine = ChewingEngine::new();
+        let conversion_engine = Box::new(ChewingEngine::new());
         let estimate = LaxUserFreqEstimate::new(0);
         let abbrev = AbbrevTable::new();
         let sym_sel = SymbolSelector::default();
@@ -1546,7 +1562,7 @@ mod tests {
             vec![("冊", 100)],
         )]);
         let dict = Layered::new(vec![Box::new(dict)], Box::new(TrieBuf::new_in_memory()));
-        let conversion_engine = ChewingEngine::new();
+        let conversion_engine = Box::new(ChewingEngine::new());
         let estimate = LaxUserFreqEstimate::new(0);
         let abbrev = AbbrevTable::new();
         let sym_sel = SymbolSelector::default();
@@ -1589,7 +1605,7 @@ mod tests {
             vec![("冊", 100)],
         )]);
         let dict = Layered::new(vec![Box::new(dict)], Box::new(TrieBuf::new_in_memory()));
-        let conversion_engine = ChewingEngine::new();
+        let conversion_engine = Box::new(ChewingEngine::new());
         let estimate = LaxUserFreqEstimate::new(0);
         let abbrev = AbbrevTable::new();
         let sym_sel = SymbolSelector::default();
@@ -1647,7 +1663,7 @@ mod tests {
             vec![("冊", 100)],
         )]);
         let dict = Layered::new(vec![Box::new(dict)], Box::new(TrieBuf::new_in_memory()));
-        let conversion_engine = ChewingEngine::new();
+        let conversion_engine = Box::new(ChewingEngine::new());
         let estimate = LaxUserFreqEstimate::new(0);
         let abbrev = AbbrevTable::new();
         let sym_sel = SymbolSelector::default();
@@ -1685,7 +1701,7 @@ mod tests {
         let keyboard = Qwerty;
         let dict = TrieBuf::new_in_memory();
         let dict = Layered::new(vec![Box::new(dict)], Box::new(TrieBuf::new_in_memory()));
-        let conversion_engine = ChewingEngine::new();
+        let conversion_engine = Box::new(ChewingEngine::new());
         let estimate = LaxUserFreqEstimate::new(0);
         let abbrev = AbbrevTable::new();
         let sym_sel = SymbolSelector::default();

tests/genkeystroke.c

@@ -352,7 +352,9 @@ int main(int argc, char *argv[])
         mvaddstr(9, 20, "Ctrl + b : toggle Eng/Chi mode");
         mvaddstr(10, 0, "F1, F2, F3, ..., F9 : Add user defined phrase");
         mvaddstr(11, 0, "Ctrl + h : toggle Full/Half shape mode");
-        mvaddstr(12, 0, "Ctrl + n/p : Next / Previous keyboard layout");
+        mvaddstr(12, 0, "Ctrl + f : toggle Fuzzy Search mode");
+        mvaddstr(13, 0, "Ctrl + s : toggle Simple mode");
+        mvaddstr(14, 0, "Ctrl + n/p : Next / Previous keyboard layout");
         show_commit_string(14, 0, ctx);
         show_userphrase(7, 14, ctx);
         show_edit_buffer(1, 0, ctx);
@@ -459,6 +461,12 @@ int main(int argc, char *argv[])
             else
                 chewing_config_set_int(ctx, "chewing.fuzzy_search_mode", TRUE);
             break;
+        case KEY_CTRL_('S'):
+            if (chewing_config_get_int(ctx, "chewing.conversion_engine") == 1)
+                chewing_config_set_int(ctx, "chewing.conversion_engine", 0);
+            else
+                chewing_config_set_int(ctx, "chewing.conversion_engine", 1);
+            break;
         case KEY_CTRL_('H'):   /* emulate Shift */
             if (chewing_get_ShapeMode(ctx) == FULLSHAPE_MODE)
                 chewing_set_ShapeMode(ctx, HALFSHAPE_MODE);

tests/test-bopomofo.c

@@ -1350,6 +1350,28 @@ void test_FuzzySearchMode_Hanyu()
     chewing_delete(ctx);
 }
 
+void test_SimpleEngine()
+{
+    const TestData SIMPLE_INPUT[] = {
+        {"ru0320 5j4up ai6g4<E>", "簡單住因模市" },
+        {"ru0320 5j4<D>4up <D>2ai6g4<D><D>2<E>", "簡單注音模式" },
+    };
+    size_t i;
+    ChewingContext *ctx;
+
+    ctx = chewing_new();
+    start_testcase(ctx, fd);
+    chewing_set_maxChiSymbolLen(ctx, 16);
+    chewing_config_set_int(ctx, "chewing.conversion_engine", 1);
+
+    for (i = 0; i < ARRAY_SIZE(SIMPLE_INPUT); ++i) {
+        type_keystroke_by_string(ctx, SIMPLE_INPUT[i].token);
+        ok_commit_buffer(ctx, SIMPLE_INPUT[i].expected);
+    }
+
+    chewing_delete(ctx);
+}
+
 void test_get_phoneSeq()
 {
     static const struct {
@@ -2366,6 +2388,7 @@ int main(int argc, char *argv[])
     test_Space();
     test_FuzzySearchMode();
     test_FuzzySearchMode_Hanyu();
+    test_SimpleEngine();
 
     test_get_phoneSeq();
     test_bopomofo_buffer();

tests/test-config.c

@@ -55,6 +55,7 @@ void test_has_option()
         ,"chewing.character_form"
         ,"chewing.space_is_select_key"
         ,"chewing.fuzzy_search_mode"
+        ,"chewing.conversion_engine"
     };
 
     ctx = chewing_new();