fix: preserve complete Japanese readings

tests/engine_test.c

@@ -2,6 +2,8 @@
 #include "../strans.c"
 #include "test.h"
 
+static int typekeys(struct ct*, char*, Str*);
+
 void
 vietnamese_state_lifetime(struct ct *t)
 {
@@ -73,7 +75,7 @@ engine_backspace_clears_candidates(struct ct *t)
 	im.sel = 0;
 	CT_CHECK(t, keystroke(Kback, 0, &com));
 	checkstr(t, "backspace commit", "", &com);
-	checkstr(t, "raw preedit after backspace", "n", &im.pre);
+	checkstr(t, "pending romaji after backspace", "n", &im.raw);
 	shown = shownpre(&im);
 	checkstr(t, "shown preedit after backspace", "ん", &shown);
 	CT_EQ_INT(t, 0, im.nkouho);
@@ -113,10 +115,11 @@ engine_commit_contract(struct ct *t)
 
 	init();
 	im.l = getlang(LangJP);
-	im.pre = mkstr("ka");
 	im.nkouho = 1;
 	im.sel = -1;
 	sclear(&com);
+	if(!typekeys(t, "ka", &com))
+		return;
 	CT_CHECK(t, keystroke(Kret, 0, &com));
 	checkstr(t, "return commit", "か", &com);
 	CT_EQ_INT(t, 0, im.pre.n);
@@ -126,12 +129,13 @@ engine_commit_contract(struct ct *t)
 
 	init();
 	im.l = getlang(LangJP);
-	im.pre = mkstr("ka");
+	sclear(&com);
+	if(!typekeys(t, "ka", &com))
+		return;
 	im.kouho[0] = mkstr("c1");
 	im.kouho[1] = mkstr("c2");
 	im.nkouho = 2;
 	im.sel = 1;
-	sclear(&com);
 	CT_CHECK(t, keystroke(Kret, 0, &com));
 	checkstr(t, "selected candidate", "c2", &com);
 	CT_EQ_INT(t, 0, im.pre.n);
@@ -140,10 +144,11 @@ engine_commit_contract(struct ct *t)
 
 	init();
 	im.l = getlang(LangJP);
-	im.pre = mkstr("ka");
+	sclear(&com);
+	if(!typekeys(t, "ka", &com))
+		return;
 	im.nkouho = 1;
 	im.sel = -1;
-	sclear(&com);
 	CT_CHECK(t, !keystroke('q', 0, &com));
 	checkstr(t, "uneaten-key commit", "か", &com);
 	CT_EQ_INT(t, 0, im.pre.n);
@@ -153,8 +158,9 @@ engine_commit_contract(struct ct *t)
 
 	init();
 	im.l = getlang(LangJP);
-	im.pre = mkstr("ka");
 	sclear(&com);
+	if(!typekeys(t, "ka", &com))
+		return;
 	CT_CHECK(t, keystroke(0xf008, 0, &com));
 	CT_EQ_INT(t, 2, com.n);
 	CT_EQ_UINT(t, 0x304b, com.r[0]);
@@ -346,6 +352,135 @@ typekeys(struct ct *t, char *keys, Str *com)
 	return 1;
 }
 
+void
+engine_japanese_readings(struct ct *t)
+{
+	static const struct {
+		char *keys;
+		char *want;
+	} cases[] = {
+		{ "nya", "にゃ" },
+		{ "n'ya", "んや" },
+		{ "nnya", "んにゃ" },
+		{ "kin'youbi", "きんようび" },
+	};
+	Str com, shown;
+	int i;
+
+	for(i = 0; i < nelem(cases); i++){
+		init();
+		im.l = getlang(LangJP);
+		sclear(&com);
+		if(!typekeys(t, cases[i].keys, &com))
+			continue;
+		checkstr(t, cases[i].keys, "", &com);
+		shown = shownpre(&im);
+		checkstr(t, cases[i].keys, cases[i].want, &shown);
+		CT_CHECK(t, keystroke(Kret, 0, &com));
+		checkstr(t, "committed reading", cases[i].want, &com);
+	}
+}
+
+static int
+newestrequest(Dictreq *req)
+{
+	int n;
+
+	n = 0;
+	while(channbrecv(dictreqc, req) > 0)
+		n++;
+	return n;
+}
+
+void
+engine_japanese_candidates(struct ct *t)
+{
+	Dictreq req;
+	Dictres res;
+	Hmap *saved;
+	Lang *jp;
+	Str com, key, shown;
+
+	jp = getlang(LangJP);
+	saved = jp->dict;
+	jp->dict = hmapalloc(8);
+	key = mkstr("かんじ");
+	hmapset(&jp->dict, &key, "漢字 幹事", strlen("漢字 幹事"));
+	init();
+	im.l = jp;
+	sclear(&com);
+	if(!typekeys(t, "kanji", &com))
+		goto cleanup;
+	shown = shownpre(&im);
+	checkstr(t, "complete Japanese reading", "かんじ", &shown);
+	if(!CT_CHECK(t, newestrequest(&req) > 0))
+		goto cleanup;
+	checkstr(t, "dictionary lookup key", "かんじ", &req.key);
+	checkstr(t, "dictionary request identity", "かんじ", &req.pre);
+	dictlookup(&req, &res);
+	dictresult(&res);
+	if(!CT_EQ_INT(t, 2, im.nkouho))
+		goto cleanup;
+	checkstr(t, "candidate one", "漢字", &im.kouho[0]);
+	checkstr(t, "candidate two", "幹事", &im.kouho[1]);
+	CT_CHECK(t, keystroke(Kdown, 0, &com));
+	CT_EQ_INT(t, 0, im.sel);
+	CT_CHECK(t, keystroke(Kdown, 0, &com));
+	CT_EQ_INT(t, 1, im.sel);
+	CT_CHECK(t, keystroke(Kret, 0, &com));
+	checkstr(t, "selected Kanji", "幹事", &com);
+cleanup:
+	newestrequest(&req);
+	hmapfree(jp->dict);
+	jp->dict = saved;
+}
+
+void
+engine_japanese_backspace_and_boundaries(struct ct *t)
+{
+	Str com, shown;
+
+	init();
+	im.l = getlang(LangJP);
+	sclear(&com);
+	if(typekeys(t, "kanji", &com)){
+		CT_CHECK(t, keystroke(Kback, 0, &com));
+		shown = shownpre(&im);
+		checkstr(t, "backspace pending i", "かんj", &shown);
+		CT_CHECK(t, keystroke(Kback, 0, &com));
+		shown = shownpre(&im);
+		checkstr(t, "backspace pending j", "かん", &shown);
+		CT_CHECK(t, keystroke(Kback, 0, &com));
+		shown = shownpre(&im);
+		checkstr(t, "backspace accumulated kana", "か", &shown);
+	}
+
+	init();
+	im.l = getlang(LangJP);
+	sclear(&com);
+	if(typekeys(t, "kanji", &com)){
+		CT_CHECK(t, keystroke('k', Mctrl, &com));
+		checkstr(t, "Hiragana mode boundary", "かんじ", &com);
+		CT_EQ_INT(t, LangJPK, im.l->lang);
+		CT_EQ_INT(t, 0, im.pre.n);
+		CT_EQ_INT(t, 0, im.raw.n);
+		sclear(&com);
+		if(typekeys(t, "ka", &com)){
+			CT_CHECK(t, keystroke('n', Mctrl, &com));
+			checkstr(t, "Katakana mode boundary", "カ", &com);
+			CT_EQ_INT(t, LangJP, im.l->lang);
+		}
+	}
+
+	init();
+	im.l = getlang(LangJP);
+	sclear(&com);
+	if(typekeys(t, "kanji", &com)){
+		CT_CHECK(t, keystroke(' ', 0, &com));
+		checkstr(t, "real input boundary", "かんじ ", &com);
+	}
+}
+
 void
 engine_emoji_single_candidate(struct ct *t)
 {
@@ -556,7 +691,8 @@ engine_emoji_preedit_languages(struct ct *t)
 	sclear(&com);
 	if(typekeys(t, "ka", &com)){
 		CT_CHECK(t, !keystroke(Kmodfirst+2, 0, &com));
-		checkstr(t, "modifier preserves Japanese", "ka", &im.pre);
+		shown = shownpre(&im);
+		checkstr(t, "modifier preserves Japanese", "か", &shown);
 		CT_CHECK(t, keystroke('E', Mctrl|Mshift, &com));
 		checkstr(t, "committed Japanese preedit", "か", &com);
 		CT_CHECK(t, search.on);

tests/test.h

@@ -33,6 +33,9 @@ void	engine_selects_visible_candidate(struct ct*);
 void	engine_commit_contract(struct ct*);
 void	engine_language_switch_state(struct ct*);
 void	engine_telex_history_bound(struct ct*);
+void	engine_japanese_readings(struct ct*);
+void	engine_japanese_candidates(struct ct*);
+void	engine_japanese_backspace_and_boundaries(struct ct*);
 void	engine_emoji_single_candidate(struct ct*);
 void	engine_emoji_queries(struct ct*);
 void	engine_emoji_digit_aliases(struct ct*);

tests/test_util.c

@@ -24,8 +24,18 @@ checkstr(struct ct *t, char *where, char *want, Str *got)
 Str
 shownpre(Im *state)
 {
-	Str shown;
+	Str mapped, shown;
 
+	if(state->l->lang == LangJP || state->l->lang == LangJPK){
+		shown = state->pre;
+		if(state->raw.n == 0)
+			return shown;
+		if(mapget(state->l->map, &state->raw, &mapped))
+			sappend(&shown, &mapped);
+		else
+			sappend(&shown, &state->raw);
+		return shown;
+	}
 	if(state->l->map != nil && mapget(state->l->map, &state->pre, &shown))
 		return shown;
 	return state->pre;

tests/trie_test.c

@@ -76,10 +76,14 @@ transmap_states(struct ct *t)
 	};
 	Emit e;
 	Im state;
+	Trie *fixture, *saved;
 	int i;
 
+	fixture = trieopen("data/hira.map");
 	memset(&state, 0, sizeof state);
 	state.l = getlang(LangJP);
+	saved = state.l->map;
+	state.l->map = fixture;
 	for(i = 0; i < nelem(cases); i++){
 		state.pre = mkstr(cases[i].pre);
 		e = transmap(&state, cases[i].key);
@@ -90,4 +94,6 @@ transmap_states(struct ct *t)
 		checkstr(t, "next", cases[i].next, &e.next);
 		checkstr(t, "mapped", cases[i].mapped, &e.dict);
 	}
+	trieclose(fixture);
+	state.l->map = saved;
 }

tests/unit_test.c

@@ -46,12 +46,14 @@ erealloc(void *p, ulong n)
 static void
 testmapinit(void)
 {
-	Lang *jp;
+	Lang *jp, *kata;
 
 	jp = getlang(LangJP);
-	if(jp == nil)
+	kata = getlang(LangJPK);
+	if(jp == nil || kata == nil)
 		die("test language is not registered");
-	jp->map = trieopen("data/hira.map");
+	jp->map = trieopen("../map/hira.map");
+	kata->map = trieopen("../map/kata.map");
 	memset(&testvi, 0, sizeof testvi);
 	testvi.lang = LangVI;
 	testvi.mapname = "telex";
@@ -80,6 +82,9 @@ static const struct ct_test tests[] = {
 	{ "engine/commit-contract", engine_commit_contract },
 	{ "engine/language-switch-state", engine_language_switch_state },
 	{ "engine/telex-history-bound", engine_telex_history_bound },
+	{ "engine/japanese-readings", engine_japanese_readings },
+	{ "engine/japanese-candidates", engine_japanese_candidates },
+	{ "engine/japanese-backspace-boundaries", engine_japanese_backspace_and_boundaries },
 	{ "engine/emoji-single-candidate", engine_emoji_single_candidate },
 	{ "engine/emoji-queries", engine_emoji_queries },
 	{ "engine/emoji-digit-aliases", engine_emoji_digit_aliases },