fix: tighten emoji lookup contracts

dict.c

@@ -33,7 +33,7 @@ dictlookup(Dictreq *req, Dictres *res)
 		while(p < e && *p != ' ')
 			p++;
 		sinit(&tmp, sp, p - sp);
-		if(scmp(&tmp, &req->key) != 0)
+		if(req->lang == LangEMOJI || scmp(&tmp, &req->key) != 0)
 			res->kouho[res->nkouho++] = tmp;
 		if(p < e)
 			p++;

map/emoji.dict

@@ -75,9 +75,14 @@ bl	😊
 blu	😊
 blus	😊
 blush	😊
-c	χ
+c	χ ☕️
 ch	χ
 chi	χ
+co	☕️
+cof	☕️
+coff	☕️
+coffe	☕️
+coffee	☕️
 d	° δ Δ ↓
 de	Δ ° δ
 deg	°
@@ -198,6 +203,65 @@ zet	ζ
 zeta	ζ
 ~	≈
 ~=	≈
+あ	❤️
+あい	❤️
+い	👍
+いい	👍
+いいね	👍
+え	😀 😄 😊
+えが	😀 😄 😊
+えがお	😀 😄 😊
+こ	☕️
+こー	☕️
+こーひ	☕️
+こーひー	☕️
+さ	👍
+さん	👍
+さんせ	👍
+さんせい	👍
+ば	😂
+ばく	😂
+ばくし	😂
+ばくしょ	😂
+ばくしょう	😂
+ほ	😄 😊
+ほほ	😄 😊
+ほほえ	😄 😊
+ほほえみ	😄 😊
+わ	😀 😂
+わら	😀 😂
+わらい	😀 😂
+ア	❤️
+アイ	❤️
+イ	👍
+イイ	👍
+イイネ	👍
+エ	😀 😄 😊
+エガ	😀 😄 😊
+エガオ	😀 😄 😊
+コ	☕️
+コー	☕️
+コーヒ	☕️
+コーヒー	☕️
+サ	👍
+サン	👍
+サンセ	👍
+サンセイ	👍
+ハ	❤️
+ハー	❤️
+ハート	❤️
+バ	😂
+バク	😂
+バクシ	😂
+バクショ	😂
+バクショウ	😂
+ホ	😄 😊
+ホホ	😄 😊
+ホホエ	😄 😊
+ホホエミ	😄 😊
+ワ	😀 😂
+ワラ	😀 😂
+ワライ	😀 😂
 따	👍
 따봉	👍
 미	😄 😊
@@ -220,6 +284,8 @@ zeta	ζ
 좋아요	👍
 찬	👍
 찬성	👍
+커	☕️
+커피	☕️
 폭	😂
 폭소	😂
 하	❤️

map/emoji.src

@@ -103,9 +103,10 @@
 ζ	zeta
 
 # Small multilingual emoji seed. Shared aliases intentionally produce choices.
-😀	웃음	웃다	스마일	smile	grin
-😄	웃음	미소	스마일	smile	happy
-😂	웃음	폭소	웃겨	joy	lol
-😊	미소	방긋	smile	blush	happy
-❤️	사랑	하트	heart	love
-👍	좋아요	찬성	따봉	엄지	thumbsup	like
+😀	웃음	웃다	스마일	えがお	エガオ	わらい	ワライ	smile	grin
+😄	웃음	미소	스마일	えがお	エガオ	ほほえみ	ホホエミ	smile	happy
+😂	웃음	폭소	웃겨	ばくしょう	バクショウ	わらい	ワライ	joy	lol
+😊	미소	방긋	ほほえみ	ホホエミ	えがお	エガオ	smile	blush	happy
+❤️	사랑	하트	あい	アイ	ハート	heart	love
+👍	좋아요	찬성	따봉	엄지	いいね	イイネ	さんせい	サンセイ	thumbsup	like
+☕️	커피	こーひー	コーヒー	coffee

map/mkemoji

@@ -38,7 +38,7 @@ def read(path):
             for field in fields[1:]:
                 alias = fold(field)
                 if (not alias or len(alias) > MAXRUNES or hascontrol(alias)
-                        or alias != alias.strip()):
+                        or alias != alias.strip() or alias.startswith(";")):
                     raise ValueError(f"{path}:{lineno}: bad alias")
                 entries.append((result, alias))
     return entries

tests/dict_test.c

@@ -89,3 +89,28 @@ dictionary_misses_clear_result(struct ct *t)
 	hmapfree(lang->dict);
 	lang->dict = saved;
 }
+
+void
+dictionary_emoji_identity(struct ct *t)
+{
+	Dictreq req;
+	Dictres res;
+	Hmap *saved;
+	Lang *lang;
+	Str key;
+
+	lang = getlang(LangEMOJI);
+	saved = lang->dict;
+	lang->dict = hmapalloc(2);
+	key = mkstr("é");
+	hmapset(&lang->dict, &key, "é", strlen("é"));
+	memset(&req, 0, sizeof req);
+	req.key = key;
+	req.pre = key;
+	req.lang = LangEMOJI;
+	dictlookup(&req, &res);
+	if(CT_EQ_INT(t, 1, res.nkouho))
+		checkstr(t, "emoji identity candidate", "é", &res.kouho[0]);
+	hmapfree(lang->dict);
+	lang->dict = saved;
+}

tests/engine_test.c

@@ -404,6 +404,10 @@ emojibegin(Emojifix *f, int showpre)
 	setemoji(&l->dict, "smil", "😀 😄");
 	setemoji(&l->dict, "smile", "😀 😄");
 	setemoji(&l->dict, "웃음", "😀 😄 😂");
+	setemoji(&l->dict, "えがお", "😀 😄 😊");
+	setemoji(&l->dict, "エガオ", "😀 😄 😊");
+	setemoji(&l->dict, "heart", "❤️");
+	setemoji(&l->dict, "coffee", "☕️");
 	setemoji(&l->dict, "^", "¹ ² ³ ⁴ ⁵ ⁶ ⁷ ⁸ ⁹ ⁽");
 	setemoji(&l->dict, "^0", "⁰");
 	setemoji(&l->dict, "_", "₁ ₂ ₃ ₄ ₅ ₆ ₇ ₈ ₉ ₍");
@@ -698,6 +702,53 @@ engine_emoji_queries(struct ct *t)
 	emojiend(&f);
 }
 
+void
+engine_emoji_japanese_and_multirune(struct ct *t)
+{
+	Emojifix f;
+	Str com;
+
+	emojibegin(&f, 0);
+	init();
+	im.l = getlang(LangJP);
+	sclear(&com);
+	CT_CHECK(t, keystroke('e', Mctrl, &com));
+	if(typekeys(t, "egao", &com)){
+		checkstr(t, "localized Hiragana query", "えがお", &search.text);
+		CT_EQ_INT(t, 3, im.nkouho);
+		CT_CHECK(t, keystroke(Kesc, 0, &com));
+	}
+
+	init();
+	im.l = getlang(LangJPK);
+	sclear(&com);
+	CT_CHECK(t, keystroke('e', Mctrl, &com));
+	if(typekeys(t, "egao", &com)){
+		checkstr(t, "localized Katakana query", "エガオ", &search.text);
+		CT_EQ_INT(t, 3, im.nkouho);
+		CT_CHECK(t, keystroke(Kesc, 0, &com));
+	}
+
+	init();
+	im.l = getlang(LangEN);
+	sclear(&com);
+	CT_CHECK(t, keystroke('e', Mctrl, &com));
+	if(typekeys(t, "heart", &com)){
+		CT_EQ_INT(t, 1, im.nkouho);
+		CT_CHECK(t, keystroke(Kret, 0, &com));
+		checkstr(t, "variation-selector emoji", "❤️", &com);
+	}
+
+	sclear(&com);
+	CT_CHECK(t, keystroke('e', Mctrl, &com));
+	if(typekeys(t, "coffee", &com)){
+		CT_EQ_INT(t, 1, im.nkouho);
+		CT_CHECK(t, keystroke(Kret, 0, &com));
+		checkstr(t, "multi-codepoint emoji", "☕️", &com);
+	}
+	emojiend(&f);
+}
+
 void
 engine_emoji_digit_aliases(struct ct *t)
 {

tests/mkemoji_test.py

@@ -66,6 +66,7 @@ class MkemojiTest(unittest.TestCase):
             "x\t a\n",
             "\0\ta\n",
             "x\ta\0b\n",
+            "x\t;hidden\n",
             "x" * 65 + "\ta\n",
             "x\t" + "a" * 65 + "\n",
         ]

tests/test.h

@@ -43,6 +43,7 @@ void	engine_japanese_backspace_and_boundaries(struct ct*);
 void	engine_katakana_sequences(struct ct*);
 void	engine_emoji_single_candidate(struct ct*);
 void	engine_emoji_queries(struct ct*);
+void	engine_emoji_japanese_and_multirune(struct ct*);
 void	engine_emoji_digit_aliases(struct ct*);
 void	engine_emoji_navigation(struct ct*);
 void	engine_emoji_preedit_languages(struct ct*);
@@ -50,6 +51,7 @@ void	engine_emoji_start_and_unknown(struct ct*);
 void	engine_emoji_dictionary_identity(struct ct*);
 void	dictionary_candidates(struct ct*);
 void	dictionary_misses_clear_result(struct ct*);
+void	dictionary_emoji_identity(struct ct*);
 void	ipc_masks_modifiers(struct ct*);
 
 #endif

tests/unit_test.c

@@ -92,6 +92,7 @@ static const struct ct_test tests[] = {
 	{ "engine/katakana-sequences", engine_katakana_sequences },
 	{ "engine/emoji-single-candidate", engine_emoji_single_candidate },
 	{ "engine/emoji-queries", engine_emoji_queries },
+	{ "engine/emoji-japanese-multirune", engine_emoji_japanese_and_multirune },
 	{ "engine/emoji-digit-aliases", engine_emoji_digit_aliases },
 	{ "engine/emoji-navigation", engine_emoji_navigation },
 	{ "engine/emoji-preedit-languages", engine_emoji_preedit_languages },
@@ -99,6 +100,7 @@ static const struct ct_test tests[] = {
 	{ "engine/emoji-dictionary-identity", engine_emoji_dictionary_identity },
 	{ "dict/candidates", dictionary_candidates },
 	{ "dict/misses-clear-result", dictionary_misses_clear_result },
+	{ "dict/emoji-identity", dictionary_emoji_identity },
 	{ "ipc/masks-modifiers", ipc_masks_modifiers },
 };