emoji: repair one-shot candidate selection

Makefile

@@ -42,5 +42,7 @@ test:
 
 verify-map:
 	python3 map/mktelex.py | cmp - map/telex.map
+	python3 -B map/mkemoji | cmp - map/emoji.dict
+	python3 -B tests/mkemoji_test.py
 
 .PHONY: all check test verify-map clean xim bench

README.md

@@ -29,9 +29,9 @@ mirrors must provide the same sibling; initialize it after cloning with
 
 The suite covers the fixed-size UTF-8 string, hash table, trie fixtures,
 Japanese/Hangul/Telex state transitions, dictionary candidates, engine
-selection state, and production map loading. Tests use explicit case tables,
+selection state, IPC request fields, and production map loading. Tests use
-boundary values, and small regressions for repaired core contracts. IPC and GUI
+explicit case tables, boundary values, and small regressions for repaired core
-protocol stacks are outside this suite and require separate integration checks.
+contracts. Full IPC and GUI protocol stacks require separate integration checks.
 
 ## Run
 
@@ -66,11 +66,35 @@ Switch input modes with Ctrl + key:
 	S  Hangul
 	T  English
 	V  Vietnamese (Telex)
-	E  Emoji
+	P  Toggle preedit echo
-	P  Toggle preedit/candidate popup window
 
 Type romanized input. Select candidates with 1-9 or arrow keys.
-Tab or Enter to commit.
+Outside one-shot search, Tab or Enter commits.
+
+### Emoji and symbols
+
+Ctrl+E starts a one-shot symbol search; it is not a mode. Any current preedit
+is committed, then the previous language is kept for the next input. Type an
+English alias or the phonetic keys of the current layout (Korean aliases work,
+for example). The candidate list is shown even when Ctrl+P has hidden normal
+preedit echo.
+
+	안녕 → Ctrl+E → smile → Enter → 😀
+
+There is no auto-insert. Enter commits the first candidate (or the current
+one); arrows and Tab move, and Esc cancels. Bare 1 through 9 select an existing
+numbered row; a bare digit with no such row and 0 remain query text. Digit
+aliases are kept in matching slots, so `^1`, `_2`, and `<3` produce `¹`, `₂`,
+and `♥`.
+
+`map/emoji.src` is the source of truth: each UTF-8 TSV row is a result followed
+by one or more tab-separated aliases. Aliases may be multilingual and ASCII
+letters are case-insensitive. Results and aliases are limited to 64 Unicode
+code points; results cannot contain whitespace. Run `map/mkemoji > map/emoji.dict`
+after editing it; `make verify-map` checks the generated dictionary and the
+Telex map.
+
+	😀	웃음	웃다	스마일	smile	grin
 
 ## Architecture
 

dat.h

@@ -156,6 +156,7 @@ struct Dictres
 	Str	key;
 	Str	kouho[Maxkouho];
 	int	nkouho;
+	int	lang;
 };
 
 extern Lang langs[];

dict.c

@@ -12,6 +12,7 @@ dictlookup(Dictreq *req, Dictres *res)
 
 	res->key = req->pre;
 	res->nkouho = 0;
+	res->lang = req->lang;
 	if(req->key.n == 0)
 		return;
 	l = getlang(req->lang);

ipc.c

@@ -38,7 +38,7 @@ ipcunpackreq(const unsigned char req[Ipcreqsz], int *want, unsigned int *mod,
 	unsigned int *key)
 {
 	*want = req[0] != 0;
-	*mod = req[1];
+	*mod = req[1] & Mmask;
 	*key = (unsigned int)req[2] |
 		((unsigned int)req[3] << 8) |
 		((unsigned int)req[4] << 16) |

ipc.h

@@ -7,6 +7,7 @@
 
 /*
  * Request: [want-preedit, modifiers, key byte 0, ..., key byte 3].
+ * Only Mmask modifier bits are retained.
  * Response: [eaten, commit-length-low, commit-length-high, commit...],
  * followed, when requested, by
  * [preedit-length-low, preedit-length-high, preedit...].
@@ -22,13 +23,14 @@ enum
 	Kesc = Kspec|0x1b,
 	Kup = Kspec|0x52,
 	Kdown = Kspec|0x54,
-	Kshift = Kspec|0xe1,
+	Kmodfirst = Kspec|0xe1,
-	Ksuper = Kspec|0xeb,
+	Kmodlast = Kspec|0xee,
 
 	Mshift = 1<<0,
 	Mctrl = 1<<2,
 	Malt = 1<<3,
 	Msuper = 1<<6,
+	Mmask = Mshift|Mctrl|Malt|Msuper,
 
 	Ipcreqsz = 6,
 	Ipclensz = 2,

map/emoji.dict

@@ -0,0 +1,226 @@
+!	⚠ ≠
+!!	⚠
+!=	≠
+*	★
+**	★
++	±
++-	±
+-	→
+->	→
+.	· … ÷
+..	· …
+...	…
+./	÷
+:	☹ ☺
+:(	☹
+:)	☺
+<	← ≤ ♥ ≠
+<-	←
+<3	♥
+<=	≤
+<>	≠
+=	≡ ⇒
+==	≡
+=>	⇒
+>	≥
+>=	≥
+^	¹ ² ³ ⁴ ⁵ ⁶ ⁷ ⁸ ⁹ ⁽ ⁾ ⁺ ⁻ ⁼ ⁰ ⁱ ⁿ
+^(	⁽
+^)	⁾
+^+	⁺
+^-	⁻
+^0	⁰
+^1	¹
+^2	²
+^3	³
+^4	⁴
+^5	⁵
+^6	⁶
+^7	⁷
+^8	⁸
+^9	⁹
+^=	⁼
+^i	ⁱ
+^n	ⁿ
+_	₁ ₂ ₃ ₄ ₅ ₆ ₇ ₈ ₉ ₍ ₎ ₊ ₋ ₌ ₀ ₐ ₑ ₒ ₓ
+_(	₍
+_)	₎
+_+	₊
+_-	₋
+_0	₀
+_1	₁
+_2	₂
+_3	₃
+_4	₄
+_5	₅
+_6	₆
+_7	₇
+_8	₈
+_9	₉
+_=	₌
+_a	ₐ
+_e	ₑ
+_o	ₒ
+_x	ₓ
+a	α
+al	α
+alp	α
+alph	α
+alpha	α
+b	β 😊
+be	β
+bet	β
+beta	β
+bl	😊
+blu	😊
+blus	😊
+blush	😊
+c	χ
+ch	χ
+chi	χ
+d	° δ Δ ↓
+de	Δ ° δ
+deg	°
+del	δ
+delt	δ
+delta	δ
+dn	↓
+e	ε η
+ep	ε
+eps	ε
+et	η
+eta	η
+f	凸
+fu	凸
+fuc	凸
+fuck	凸
+g	γ Γ 😀
+ga	Γ γ
+gam	γ
+gamm	γ
+gamma	γ
+gr	😀
+gri	😀
+grin	😀
+h	😄 😊 ❤️
+ha	😄 😊
+hap	😄 😊
+happ	😄 😊
+happy	😄 😊
+he	❤️
+hea	❤️
+hear	❤️
+heart	❤️
+i	∫ ∞ ι
+ii	∫
+in	∞
+inf	∞
+io	ι
+iot	ι
+iota	ι
+j	😂
+jo	😂
+joy	😂
+k	κ
+ka	κ
+kap	κ
+kapp	κ
+kappa	κ
+l	λ Λ 😂 ❤️ 👍
+la	Λ λ
+lam	λ
+lamb	λ
+lambd	λ
+lambda	λ
+li	👍
+lik	👍
+like	👍
+lo	😂 ❤️
+lol	😂
+lov	❤️
+love	❤️
+m	× μ
+mu	μ ×
+mul	×
+n	ν
+nu	ν
+o	ω Ω ●
+om	Ω ω
+ome	ω
+omeg	ω
+omega	ω
+oo	●
+p	φ Φ π Π ∏ ψ Ψ
+ph	Φ φ
+phi	φ
+pi	π Π
+pp	∏
+ps	Ψ ψ
+psi	ψ
+r	ρ
+rh	ρ
+rho	ρ
+s	σ Σ √ ∑ 😀 😄 😊
+si	Σ σ
+sig	σ
+sigm	σ
+sigma	σ
+sm	😀 😄 😊
+smi	😀 😄 😊
+smil	😀 😄 😊
+smile	😀 😄 😊
+sq	√
+ss	∑
+t	τ θ Θ 👍
+ta	τ
+tau	τ
+th	Θ θ 👍
+the	θ
+thet	θ
+theta	θ
+thu	👍
+thum	👍
+thumb	👍
+thumbs	👍
+thumbsu	👍
+thumbsup	👍
+u	↑ υ
+up	↑ υ
+ups	υ
+v	✓
+vv	✓
+x	ξ Ξ ✗
+xi	ξ Ξ
+xx	✗
+z	ζ
+ze	ζ
+zet	ζ
+zeta	ζ
+~	≈
+~=	≈
+따	👍
+따봉	👍
+미	😄 😊
+미소	😄 😊
+방	😊
+방긋	😊
+사	❤️
+사랑	❤️
+스	😀 😄
+스마	😀 😄
+스마일	😀 😄
+엄	👍
+엄지	👍
+웃	😀 😄 😂
+웃겨	😂
+웃다	😀
+웃음	😀 😄 😂
+좋	👍
+좋아	👍
+좋아요	👍
+찬	👍
+찬성	👍
+폭	😂
+폭소	😂
+하	❤️
+하트	❤️

map/emoji.map

@@ -1,183 +0,0 @@
-!	!
-!!	⚠
-!=	≠
-*	*
-**	★
-+	+
-+-	±
--	-
-->	→
-.	.
-..	·
-...	…
-./	÷
-:	:
-:(	☹
-:)	☺
-<	<
-<-	←
-<3	♥
-<=	≤
-<>	≠
-=	=
-==	≡
-=>	⇒
->	>
->=	≥
-D	D
-De	Δ
-G	G
-Ga	Γ
-I	I
-II	∫
-L	L
-La	Λ
-O	O
-Om	Ω
-P	P
-PP	∏
-Ph	Φ
-Pi	Π
-Ps	Ψ
-S	S
-SS	∑
-Si	Σ
-T	T
-Th	Θ
-X	X
-Xi	Ξ
-^	^
-^(	⁽
-^)	⁾
-^+	⁺
-^-	⁻
-^0	⁰
-^1	¹
-^2	²
-^3	³
-^4	⁴
-^5	⁵
-^6	⁶
-^7	⁷
-^8	⁸
-^9	⁹
-^=	⁼
-^i	ⁱ
-^n	ⁿ
-_	_
-_(	₍
-_)	₎
-_+	₊
-_-	₋
-_0	₀
-_1	₁
-_2	₂
-_3	₃
-_4	₄
-_5	₅
-_6	₆
-_7	₇
-_8	₈
-_9	₉
-_=	₌
-_a	ₐ
-_e	ₑ
-_o	ₒ
-_x	ₓ
-a	a
-al	al
-alp	alp
-alph	alph
-alpha	α
-b	b
-be	be
-bet	bet
-beta	β
-c	c
-ch	ch
-chi	χ
-d	d
-de	de
-deg	°
-del	del
-delt	delt
-delta	δ
-dn	↓
-e	e
-ep	ep
-eps	ε
-et	et
-eta	η
-f	f
-fu	fu
-fuc	fuc
-fuck	凸
-g	g
-ga	ga
-gam	gam
-gamm	gamm
-gamma	γ
-i	i
-in	in
-inf	∞
-io	io
-iot	iot
-iota	ι
-k	k
-ka	ka
-kap	kap
-kapp	kapp
-kappa	κ
-l	l
-la	la
-lam	lam
-lamb	lamb
-lambd	lambd
-lambda	λ
-m	m
-mu	μ
-mul	×
-n	n
-nu	ν
-o	o
-om	om
-ome	ome
-omeg	omeg
-omega	ω
-oo	●
-p	p
-ph	ph
-phi	φ
-pi	π
-ps	ps
-psi	ψ
-r	r
-rh	rh
-rho	ρ
-s	s
-si	si
-sig	sig
-sigm	sigm
-sigma	σ
-sq	√
-t	t
-ta	ta
-tau	τ
-th	th
-the	the
-thet	thet
-theta	θ
-u	u
-up	↑
-ups	υ
-v	v
-vv	✓
-x	x
-xi	ξ
-xx	✗
-z	z
-ze	ze
-zet	zet
-zeta	ζ
-~	~
-~=	≈

map/emoji.src

@@ -1,100 +1,111 @@
-!!	⚠
+# Result first, followed by one or more TAB-separated aliases.
-!=	≠
+⚠	!!
-**	★
+★	**
-+-	±
+±	+-
-->	→
+→	->
-..	·
+·	..
-...	…
+…	...
-./	÷
+÷	./
-:(	☹
+☹	:(
-:)	☺
+☺	:)
-<-	←
+
-<=	≤
+# Bare 1-9 choose prefix candidates; keep digit aliases in matching slots.
-<>	≠
+←	<-
-==	≡
+≤	<=
-=>	⇒
+♥	<3
->=	≥
+≠	<>	!=
-^(	⁽
+≡	==
-^)	⁾
+⇒	=>
-^+	⁺
+≥	>=
-^-	⁻
+¹	^1
-^=	⁼
+²	^2
-_(	₍
+³	^3
-_)	₎
+⁴	^4
-_+	₊
+⁵	^5
-_-	₋
+⁶	^6
-_=	₌
+⁷	^7
-~=	≈
+⁸	^8
-^0	⁰
+⁹	^9
-_0	₀
+₁	_1
-^1	¹
+₂	_2
-_1	₁
+₃	_3
-^2	²
+₄	_4
-_2	₂
+₅	_5
-<3	♥
+₆	_6
-^3	³
+₇	_7
-_3	₃
+₈	_8
-^4	⁴
+₉	_9
-_4	₄
+
-^5	⁵
+⁽	^(
-_5	₅
+⁾	^)
-^6	⁶
+⁺	^+
-_6	₆
+⁻	^-
-^7	⁷
+⁼	^=
-_7	₇
+₍	_(
-^8	⁸
+₎	_)
-_8	₈
+₊	_+
-^9	⁹
+₋	_-
-_9	₉
+₌	_=
-_a	ₐ
+≈	~=
-alpha	α
+⁰	^0
-beta	β
+₀	_0
-chi	χ
+ₐ	_a
-deg	°
+α	alpha
-delta	δ
+β	beta
-De	Δ
+χ	chi
-dn	↓
+°	deg
-_e	ₑ
+δ	delta
-eps	ε
+Δ	De
-eta	η
+↓	dn
-fuck	凸
+ₑ	_e
-gamma	γ
+ε	eps
-Ga	Γ
+η	eta
-^i	ⁱ
+凸	fuck
-II	∫
+γ	gamma
-inf	∞
+Γ	Ga
-iota	ι
+ⁱ	^i
-kappa	κ
+∫	II
-lambda	λ
+∞	inf
-La	Λ
+ι	iota
-mul	×
+κ	kappa
-mu	μ
+λ	lambda
-^n	ⁿ
+Λ	La
-nu	ν
+×	mul
-omega	ω
+μ	mu
-Om	Ω
+ⁿ	^n
-oo	●
+ν	nu
-_o	ₒ
+ω	omega
-phi	φ
+Ω	Om
-Ph	Φ
+●	oo
-pi	π
+ₒ	_o
-Pi	Π
+φ	phi
-PP	∏
+Φ	Ph
-psi	ψ
+π	pi
-Ps	Ψ
+Π	Pi
-rho	ρ
+∏	PP
-sigma	σ
+ψ	psi
-Si	Σ
+Ψ	Ps
-sq	√
+ρ	rho
-SS	∑
+σ	sigma
-tau	τ
+Σ	Si
-theta	θ
+√	sq
-Th	Θ
+∑	SS
-up	↑
+τ	tau
-ups	υ
+θ	theta
-vv	✓
+Θ	Th
-xi	ξ
+↑	up
-Xi	Ξ
+υ	ups
-xx	✗
+✓	vv
-_x	ₓ
+ξ	xi
-zeta	ζ
+Ξ	Xi
+✗	xx
+ₓ	_x
+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

map/mkemoji

@@ -1,39 +1,80 @@
-#!/bin/sh
+#!/usr/bin/env python3
+"""Write emoji.dict to stdout from a result-first UTF-8 TSV file."""
 
-awk '
+import sys
-BEGIN { FS = "\t" }
+import unicodedata
-/^$/ { next }
+from pathlib import Path
-{
-	key = $1
-	val = $2
-	map[key] = val
-	for(i = 1; i < length(key); i++){
-		p = substr(key, 1, i)
-		if(!(p in map))
-			map[p] = p
-	}
-}
-END {
-	for(k in map)
-		print k "\t" map[k]
-}
-' $1 | sort -k1,1 > emoji.map
 
-awk '
+
-BEGIN { FS = "\t" }
+SOURCE = Path(__file__).with_name("emoji.src")
-/^$/ { next }
+MAXRUNES = 64
-{
+
-	key = $1
+
-	val = $2
+def fold(s):
-	for(i = 1; i <= length(key); i++){
+    s = "".join(chr(ord(c) + 32) if "A" <= c <= "Z" else c for c in s)
-		p = substr(key, 1, i)
+    return unicodedata.normalize("NFC", s)
-		dict[p] = dict[p] " " val
+
-	}
+
-}
+def hascontrol(s):
-END {
+    return any(unicodedata.category(c) == "Cc" for c in s)
-	for(p in dict){
+
-		sub(/^ /, "", dict[p])
+
-		print p "\t" dict[p]
+def read(path):
-	}
+    entries = []
-}
+    with path.open(encoding="utf-8") as src:
-' $1 | sort -k1,1 > emoji.dict
+        for lineno, raw in enumerate(src, 1):
+            line = raw.rstrip("\r\n")
+            if not line.strip():
+                continue
+            if "\t" not in line and line.lstrip().startswith("#"):
+                continue
+            fields = line.split("\t")
+            if len(fields) < 2:
+                raise ValueError(f"{path}:{lineno}: need a result and an alias")
+            result = unicodedata.normalize("NFC", fields[0])
+            if (not result or len(result) > MAXRUNES or hascontrol(result)
+                    or any(c.isspace() for c in result)):
+                raise ValueError(f"{path}:{lineno}: bad result")
+            for field in fields[1:]:
+                alias = fold(field)
+                if (not alias or len(alias) > MAXRUNES or hascontrol(alias)
+                        or alias != alias.strip()):
+                    raise ValueError(f"{path}:{lineno}: bad alias")
+                entries.append((result, alias))
+    return entries
+
+
+def add(table, key, result):
+    values = table.setdefault(key, [])
+    if result not in values:
+        values.append(result)
+
+
+def build(entries):
+    exact = {}
+    prefix = {}
+    for result, alias in entries:
+        for n in range(1, len(alias) + 1):
+            add(exact if n == len(alias) else prefix, alias[:n], result)
+    for key in sorted(exact.keys() | prefix.keys()):
+        values = exact.get(key, []) + prefix.get(key, [])
+        values = list(dict.fromkeys(values))
+        yield f"{key}\t{' '.join(values)}"
+
+
+def main():
+    if len(sys.argv) > 2:
+        print(f"usage: {sys.argv[0]} [emoji.src]", file=sys.stderr)
+        return 2
+    path = Path(sys.argv[1]) if len(sys.argv) == 2 else SOURCE
+    try:
+        for line in build(read(path)):
+            print(line)
+    except (OSError, UnicodeError, ValueError) as error:
+        print(error, file=sys.stderr)
+        return 1
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())

strans.c

@@ -3,8 +3,18 @@
 
 static Im im;
 static int popup = 0;
+static int visible = 0;
 static void dictqmap(Im*);
 
+typedef struct Search Search;
+struct Search
+{
+	int	on;
+	Str	raw;
+	Str	text;
+};
+static Search search;
+
 static void
 backmap(Im *im)
 {
@@ -16,7 +26,7 @@ Lang langs[] = {
 	{LangJP,    "hira",    "kanji",  transmap, backmap, dictqmap, nil, nil},
 	{LangJPK,   "kata",    "kanji",  transmap, backmap, dictqmap, nil, nil},
 	{LangKO,    "hangul",  nil,      transko,  backko,  dictqmap, nil, nil},
-	{LangEMOJI, "emoji",   "emoji",  transmap, backmap, dictqmap, nil, nil},
+	{LangEMOJI, nil,       "emoji",  nil,      nil,     nil,      nil, nil},
 	{LangVI,    "telex",   nil,      transvi,  backvi,  dictqmap, nil, nil},
 };
 int nlang = nelem(langs);
@@ -43,14 +53,17 @@ setkouho(Dictres *res)
 static void
 show(void)
 {
+	static Drawcmd old;
 	Drawcmd dc;
+	Str *pre;
 	int i, first, n;
 
-	if(!popup)
-		return;
 	sclear(&dc.pre);
-	if(!mapget(im.l->map, &im.pre, &dc.pre))
+	pre = search.on ? &search.text : &im.pre;
-		dc.pre = im.pre;
+	if(popup){
+		if(search.on || im.l->map == nil || !mapget(im.l->map, pre, &dc.pre))
+			dc.pre = *pre;
+	}
 	first = im.sel >= Maxdisp ? im.sel - Maxdisp + 1 : 0;
 	n = im.nkouho - first;
 	if(n > Maxdisp) n = Maxdisp;
@@ -58,7 +71,13 @@ show(void)
 	dc.sel = im.sel >= 0 ? im.sel - first : -1;
 	for(i = 0; i < n; i++)
 		dc.kouho[i] = im.kouho[first + i];
-	chansend(drawc, &dc);
+	if(dc.pre.n == 0 && n == 0 && !visible)
+		return;
+	visible = dc.pre.n != 0 || n != 0;
+	/* imthread is the sole producer; keep only the newest pending draw. */
+	while(channbrecv(drawc, &old) > 0)
+		;
+	channbsend(drawc, &dc);
 }
 
 static void
@@ -110,7 +129,7 @@ commit(Str *com)
 {
 	Str val;
 
-	if(mapget(im.l->map, &im.pre, &val))
+	if(im.l->map != nil && mapget(im.l->map, &im.pre, &val))
 		sappend(com, &val);
 	else
 		sappend(com, &im.pre);
@@ -209,13 +228,248 @@ transmap(Im *im, Rune c)
 	return e;
 }
 
+static Rune
+ctrlkey(Rune c)
+{
+	if(c >= 'A' && c <= 'Z')
+		c += 'a' - 'A';
+	if(c >= 'a' && c <= 'z')
+		return c - 'a' + 1;
+	return c;
+}
+
+static int
+ismodkey(u32int ks)
+{
+	return ks >= Kmodfirst && ks <= Kmodlast;
+}
+
+static void
+foldascii(Str *s)
+{
+	int i;
+
+	for(i = 0; i < s->n; i++)
+		if(s->r[i] >= 'A' && s->r[i] <= 'Z')
+			s->r[i] += 'a' - 'A';
+}
+
+static void
+transstr(Lang *l, Str *raw, Str *out)
+{
+	static Im q;
+	Emit e;
+	Str mapped;
+	int i;
+
+	sclear(out);
+	if(l == nil || l->trans == nil || (l->trans == transmap && l->map == nil)){
+		*out = *raw;
+		return;
+	}
+	memset(&q, 0, sizeof q);
+	q.l = l;
+	for(i = 0; i < raw->n; i++){
+		e = l->trans(&q, raw->r[i]);
+		sappend(out, &e.s);
+		q.pre = e.next;
+		if(!e.eat && out->n < Maxrunes)
+			sputr(out, raw->r[i]);
+	}
+	if(q.pre.n == 0)
+		return;
+	if(l->map != nil && mapget(l->map, &q.pre, &mapped))
+		sappend(out, &mapped);
+	else
+		sappend(out, &q.pre);
+}
+
+static void
+addkouho(Str *s)
+{
+	int i;
+
+	for(i = 0; i < im.nkouho; i++)
+		if(scmp(&im.kouho[i], s) == 0)
+			return;
+	if(im.nkouho < Maxkouho)
+		im.kouho[im.nkouho++] = *s;
+}
+
+static void
+emojilookup(Str *key, Dictres *res)
+{
+	Dictreq req;
+
+	req.key = *key;
+	req.pre = req.key;
+	req.lang = LangEMOJI;
+	dictlookup(&req, res);
+}
+
+static void
+emojiquery(void)
+{
+	static Dictres res;
+	Str raw, local;
+	int i, rawhit, localhit;
+
+	raw = search.raw;
+	foldascii(&raw);
+	transstr(im.l, &search.raw, &local);
+	foldascii(&local);
+	clearkouho();
+	emojilookup(&raw, &res);
+	rawhit = res.nkouho != 0;
+	for(i = 0; i < res.nkouho; i++)
+		addkouho(&res.kouho[i]);
+	localhit = 0;
+	if(scmp(&raw, &local) != 0){
+		emojilookup(&local, &res);
+		localhit = res.nkouho != 0;
+		for(i = 0; i < res.nkouho; i++)
+			addkouho(&res.kouho[i]);
+	}
+	search.text = rawhit || !localhit ? raw : local;
+	show();
+}
+
+static void
+endsearch(void)
+{
+	search.on = 0;
+	sclear(&search.raw);
+	sclear(&search.text);
+	clearkouho();
+	show();
+}
+
+static void
+startsearch(Str *com)
+{
+	if(im.pre.n > 0)
+		commit(com);
+	sclear(&im.pre);
+	sclear(&im.raw);
+	clearkouho();
+	search.on = 1;
+	sclear(&search.raw);
+	sclear(&search.text);
+	show();
+}
+
+static void
+commitsearch(Str *com)
+{
+	sappend(com, &search.text);
+	endsearch();
+}
+
+static int
+pickemoji(int n, Str *com)
+{
+	if(n < 0 || n >= im.nkouho)
+		return 0;
+	sappend(com, &im.kouho[n]);
+	endsearch();
+	return 1;
+}
+
+static int
+searchkey(u32int ks, u32int mod, Str *com)
+{
+	int n, off;
+	Rune c;
+
+	if(ismodkey(ks))
+		return 0;
+	if(ks == Kdown || ks == Kup){
+		if(im.nkouho == 0)
+			return 1;
+		if(ks == Kdown && im.sel < im.nkouho - 1)
+			im.sel++;
+		if(ks == Kup && im.sel > 0)
+			im.sel--;
+		show();
+		return 1;
+	}
+	n = ks - '1';
+	off = im.sel >= Maxdisp ? im.sel - Maxdisp + 1 : 0;
+	if(n >= 0 && n < Maxdisp && off + n < im.nkouho && mod == 0){
+		pickemoji(off + n, com);
+		return 1;
+	}
+	if(ks == Ktab){
+		if(im.nkouho == 0)
+			return 1;
+		if(mod & Mshift){
+			if(im.sel <= 0)
+				im.sel = im.nkouho - 1;
+			else
+				im.sel--;
+		}else if(im.sel >= im.nkouho - 1)
+			im.sel = 0;
+		else
+			im.sel++;
+		show();
+		return 1;
+	}
+	if(ks == Kret){
+		if(im.nkouho > 0)
+			return pickemoji(im.sel >= 0 ? im.sel : 0, com);
+		commitsearch(com);
+		return 1;
+	}
+	if(ks == Kback){
+		if(search.raw.n == 0)
+			return 1;
+		spopr(&search.raw);
+		emojiquery();
+		return 1;
+	}
+	if(ks == Kesc){
+		endsearch();
+		return 1;
+	}
+	if(mod & Mctrl){
+		c = ctrlkey(ks);
+		if(c == LangEMOJI){
+			endsearch();
+			return 1;
+		}
+		if(c == 0x10){
+			popup = !popup;
+			show();
+			return 1;
+		}
+		endsearch();
+		if(setlang(c))
+			return 1;
+		return 0;
+	}
+	if(ks >= Kspec || ks < ' ' || (mod & (Malt|Msuper))){
+		commitsearch(com);
+		return 0;
+	}
+	if(search.raw.n >= Maxrunes){
+		commitsearch(com);
+		return 0;
+	}
+	sputr(&search.raw, ks);
+	emojiquery();
+	return 1;
+}
+
 static int
 keystroke(u32int ks, u32int mod, Str *com)
 {
 	int n, off;
+	Rune c;
 
-	if(ks == Ksuper || ks == Kshift)
+	if(ismodkey(ks))
-		return 1;
+		return 0;
+	if(search.on)
+		return searchkey(ks, mod, com);
 	if(ks == Kdown || ks == Kup){
 		if(im.nkouho == 0)
 			return 0;
@@ -271,14 +525,18 @@ keystroke(u32int ks, u32int mod, Str *com)
 		return 0;
 	}
 	if(mod & Mctrl){
-		reset();
+		c = ctrlkey(ks);
-		if(ks >= 'a' && ks <= 'z')
+		if(c == LangEMOJI){
-			ks -= 'a' - 1;
+			startsearch(com);
-		if(ks == 0x10){
-			popup = !popup;
 			return 1;
 		}
-		if(setlang(ks))
+		if(c == 0x10){
+			popup = !popup;
+			show();
+			return 1;
+		}
+		reset();
+		if(setlang(c))
 			return 1;
 		return 0;
 	}
@@ -305,6 +563,8 @@ init(void)
 {
 	memset(&im, 0, sizeof(im));
 	im.l = getlang(LangEN);
+	memset(&search, 0, sizeof search);
+	visible = 0;
 }
 
 static void
@@ -316,15 +576,25 @@ imhandlekey(Keyreq *kr)
 	sclear(&res.preedit);
 	res.eaten = keystroke(kr->ks, kr->mod, &res.commit);
 	if(kr->want)
-		res.preedit = im.pre;
+		res.preedit = search.on ? search.text : im.pre;
 	chansend(kr->reply, &res);
 }
 
+static void
+dictresult(Dictres *res)
+{
+	if(search.on || res->lang != im.l->lang ||
+	   scmp(&res->key, &im.pre) != 0)
+		return;
+	setkouho(res);
+	show();
+}
+
 void
 imthread(void*)
 {
+	static Dictres res;
 	Keyreq kr;
-	Dictres res;
 	Alt alts[] = {
 		{keyc, &kr, CHANRCV, nil},
 		{dictresc, &res, CHANRCV, nil},
@@ -339,10 +609,7 @@ imthread(void*)
 			imhandlekey(&kr);
 			break;
 		case 1:
-			if(scmp(&res.key, &im.pre) == 0){
+			dictresult(&res);
-				setkouho(&res);
-				show();
-			}
 			break;
 		}
 	}

tests/Makefile

@@ -5,9 +5,9 @@ LIBS = -lthread -lbio
 
 PROG = unit_test
 TESTSRC = unit_test.c test_util.c str_test.c hash_test.c trie_test.c \
-	ko_test.c vi_test.c engine_test.c dict_test.c
+	ko_test.c vi_test.c engine_test.c dict_test.c ipc_test.c
 TESTOBJ = $(TESTSRC:.c=.o)
-PARENTSRC = str.c hash.c trie.c dict.c ko.c vi.c
+PARENTSRC = str.c hash.c trie.c dict.c ko.c vi.c ipc.c
 PARENTOBJ = $(PARENTSRC:%.c=unit_%.o)
 OBJS = $(TESTOBJ) $(PARENTOBJ)
 

tests/dict_test.c

@@ -25,6 +25,7 @@ dictionary_candidates(struct ct *t)
 	dictlookup(&req, &res);
 	if(!CT_EQ_INT(t, 2, res.nkouho))
 		goto cleanup;
+	CT_EQ_INT(t, LangJP, res.lang);
 	CT_EQ_INT(t, 0, scmp(&req.pre, &res.key));
 	checkstr(t, "candidate 1", "候補1", &res.kouho[0]);
 	checkstr(t, "candidate 2", "候補2", &res.kouho[1]);
@@ -44,6 +45,7 @@ dictionary_candidates(struct ct *t)
 	dictlookup(&req, &res);
 	if(!CT_EQ_INT(t, Maxkouho, res.nkouho))
 		goto cleanup;
+	CT_EQ_INT(t, LangJP, res.lang);
 	CT_EQ_INT(t, 0, scmp(&req.pre, &res.key));
 	checkstr(t, "first capped candidate", "c00", &res.kouho[0]);
 	checkstr(t, "last capped candidate", "c31", &res.kouho[31]);
@@ -81,6 +83,7 @@ dictionary_misses_clear_result(struct ct *t)
 		if(res.nkouho != 0)
 			CT_ERRORF(t, "%s: want 0 candidates, got %d",
 				cases[i].name, res.nkouho);
+		CT_EQ_INT(t, LangJP, res.lang);
 		checkstr(t, cases[i].name, cases[i].pre, &res.key);
 	}
 	hmapfree(lang->dict);

tests/engine_test.c

@@ -255,3 +255,449 @@ engine_language_switch_state(struct ct *t)
 	}
 	vi->map = saved;
 }
+
+typedef struct Emojifix Emojifix;
+struct Emojifix
+{
+	Im	im;
+	Search	search;
+	Hmap	*dict;
+	int	popup;
+	int	visible;
+};
+
+static int
+draindraw(Drawcmd *last)
+{
+	Drawcmd dc = {0};
+	int n;
+
+	n = 0;
+	while(channbrecv(drawc, &dc) > 0){
+		n++;
+		if(last != nil)
+			*last = dc;
+	}
+	return n;
+}
+
+static void
+setemoji(Hmap **dict, char *key, char *val)
+{
+	Str s;
+
+	s = mkstr(key);
+	hmapset(dict, &s, val, strlen(val));
+}
+
+static void
+emojibegin(Emojifix *f, int showpre)
+{
+	Lang *l;
+
+	l = getlang(LangEMOJI);
+	f->im = im;
+	f->search = search;
+	f->dict = l->dict;
+	f->popup = popup;
+	f->visible = visible;
+	draindraw(nil);
+	l->dict = hmapalloc(32);
+	setemoji(&l->dict, "a", "A B");
+	setemoji(&l->dict, "ㅁ", "B C");
+	setemoji(&l->dict, "alpha", "α");
+	setemoji(&l->dict, "smil", "😀 😄");
+	setemoji(&l->dict, "smile", "😀 😄");
+	setemoji(&l->dict, "웃음", "😀 😄 😂");
+	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 ₍");
+	setemoji(&l->dict, "<", "← ≤ ♥ ≠");
+	setemoji(&l->dict, "many",
+		"c1 c2 c3 c4 c5 c6 c7 c8 c9 c10 c11 c12");
+	init();
+	im.l = getlang(LangKO);
+	popup = showpre;
+}
+
+static void
+emojiend(Emojifix *f)
+{
+	Lang *l;
+	Hmap *dict;
+
+	l = getlang(LangEMOJI);
+	dict = l->dict;
+	l->dict = f->dict;
+	hmapfree(dict);
+	draindraw(nil);
+	im = f->im;
+	search = f->search;
+	popup = f->popup;
+	visible = f->visible;
+}
+
+static int
+typekeys(struct ct *t, char *keys, Str *com)
+{
+	for(; *keys != '\0'; keys++)
+		if(!CT_CHECK(t, keystroke((uchar)*keys, 0, com)))
+			return 0;
+	return 1;
+}
+
+void
+engine_emoji_single_candidate(struct ct *t)
+{
+	Drawcmd dc = {0};
+	Emojifix f;
+	Str com;
+
+	emojibegin(&f, 1);
+	sclear(&com);
+	CT_CHECK(t, keystroke('p', Mctrl, &com));
+	CT_EQ_INT(t, 0, popup);
+	CT_CHECK(t, keystroke('e', Mctrl, &com));
+	CT_CHECK(t, search.on);
+	CT_EQ_INT(t, LangKO, im.l->lang);
+	if(typekeys(t, "alpha", &com)){
+		checkstr(t, "no automatic commit", "", &com);
+		checkstr(t, "raw English query", "alpha", &search.text);
+		CT_CHECK(t, search.on);
+		CT_EQ_INT(t, 1, im.nkouho);
+		checkstr(t, "single candidate", "α", &im.kouho[0]);
+		CT_CHECK(t, draindraw(&dc) > 0);
+		CT_EQ_INT(t, 0, dc.pre.n);
+		CT_EQ_INT(t, 1, dc.nkouho);
+		CT_CHECK(t, keystroke(Kret, 0, &com));
+		checkstr(t, "explicit candidate commit", "α", &com);
+		CT_CHECK(t, !search.on);
+		CT_EQ_INT(t, LangKO, im.l->lang);
+		show();
+		CT_CHECK(t, draindraw(&dc) > 0);
+		CT_EQ_INT(t, 0, dc.pre.n);
+		CT_EQ_INT(t, 0, dc.nkouho);
+	}
+	emojiend(&f);
+}
+
+void
+engine_emoji_queries(struct ct *t)
+{
+	Drawcmd dc = {0};
+	Emojifix f;
+	Str com;
+
+	emojibegin(&f, 1);
+	sclear(&com);
+	CT_CHECK(t, keystroke('e', Mctrl, &com));
+	CT_CHECK(t, keystroke('a', 0, &com));
+	checkstr(t, "raw query wins", "a", &search.text);
+	CT_EQ_INT(t, 3, im.nkouho);
+	checkstr(t, "raw candidate", "A", &im.kouho[0]);
+	checkstr(t, "deduplicated candidate", "B", &im.kouho[1]);
+	checkstr(t, "local candidate", "C", &im.kouho[2]);
+	CT_CHECK(t, draindraw(&dc) > 0);
+	checkstr(t, "prefix popup query", "a", &dc.pre);
+	CT_EQ_INT(t, 3, dc.nkouho);
+	CT_CHECK(t, keystroke(Kesc, 0, &com));
+
+	CT_CHECK(t, keystroke('e', Mctrl, &com));
+	if(typekeys(t, "smile", &com)){
+		checkstr(t, "English alias", "smile", &search.text);
+		CT_EQ_INT(t, 2, im.nkouho);
+		CT_CHECK(t, keystroke(Kback, 0, &com));
+		checkstr(t, "backspaced query", "smil", &search.text);
+		CT_EQ_INT(t, 2, im.nkouho);
+		CT_CHECK(t, keystroke('e', 0, &com));
+		CT_CHECK(t, keystroke('2', 0, &com));
+		checkstr(t, "selected English alias", "😄", &com);
+		CT_EQ_INT(t, LangKO, im.l->lang);
+	}
+
+	sclear(&com);
+	CT_CHECK(t, keystroke('e', Mctrl, &com));
+	if(typekeys(t, "SMILE", &com)){
+		checkstr(t, "folded English alias", "smile", &search.text);
+		CT_EQ_INT(t, 2, im.nkouho);
+		CT_CHECK(t, keystroke(Kesc, 0, &com));
+	}
+
+	sclear(&com);
+	CT_CHECK(t, keystroke('e', Mctrl, &com));
+	if(typekeys(t, "dntdma", &com)){
+		checkstr(t, "localized Korean query", "웃음", &search.text);
+		CT_EQ_INT(t, 3, im.nkouho);
+		CT_CHECK(t, draindraw(&dc) > 0);
+		checkstr(t, "popup query", "웃음", &dc.pre);
+		CT_EQ_INT(t, 3, dc.nkouho);
+		CT_CHECK(t, keystroke(Kesc, 0, &com));
+		checkstr(t, "cancel commit", "", &com);
+		CT_CHECK(t, !search.on);
+		CT_EQ_INT(t, LangKO, im.l->lang);
+	}
+	emojiend(&f);
+}
+
+void
+engine_emoji_digit_aliases(struct ct *t)
+{
+	static const struct {
+		char *name;
+		Rune prefix;
+		Rune digit;
+		char *want;
+	} cases[] = {
+		{ "^1", '^', '1', "¹" },
+		{ "_2", '_', '2', "₂" },
+		{ "<3", '<', '3', "♥" },
+	};
+	Emojifix f;
+	Str com;
+	int i;
+
+	emojibegin(&f, 0);
+	for(i = 0; i < nelem(cases); i++){
+		sclear(&com);
+		CT_CHECK(t, keystroke('e', Mctrl, &com));
+		CT_CHECK(t, keystroke(cases[i].prefix, Mshift, &com));
+		CT_CHECK(t, im.nkouho >= cases[i].digit - '0');
+		CT_CHECK(t, keystroke(cases[i].digit, 0, &com));
+		checkstr(t, cases[i].name, cases[i].want, &com);
+		CT_CHECK(t, !search.on);
+		CT_EQ_INT(t, LangKO, im.l->lang);
+	}
+
+	sclear(&com);
+	CT_CHECK(t, keystroke('e', Mctrl, &com));
+	CT_CHECK(t, keystroke('^', Mshift, &com));
+	CT_CHECK(t, keystroke('0', 0, &com));
+	checkstr(t, "zero query", "^0", &search.text);
+	checkstr(t, "zero does not commit", "", &com);
+	CT_EQ_INT(t, 1, im.nkouho);
+	CT_CHECK(t, keystroke(Kret, 0, &com));
+	checkstr(t, "zero alias", "⁰", &com);
+
+	sclear(&com);
+	CT_CHECK(t, keystroke('e', Mctrl, &com));
+	if(typekeys(t, "smile", &com)){
+		CT_CHECK(t, keystroke('9', 0, &com));
+		checkstr(t, "missing number query", "smile9", &search.text);
+		checkstr(t, "missing number commit", "", &com);
+		CT_CHECK(t, search.on);
+		CT_EQ_INT(t, 0, im.nkouho);
+		CT_CHECK(t, keystroke(Kret, 0, &com));
+		checkstr(t, "missing number text", "smile9", &com);
+	}
+	emojiend(&f);
+}
+
+void
+engine_emoji_navigation(struct ct *t)
+{
+	Drawcmd dc = {0};
+	Emojifix f;
+	Str com;
+	int i;
+
+	emojibegin(&f, 0);
+	sclear(&com);
+	CT_CHECK(t, keystroke('e', Mctrl, &com));
+	if(typekeys(t, "many", &com)){
+		CT_EQ_INT(t, 12, im.nkouho);
+		for(i = 0; i < Maxdisp+1; i++)
+			CT_CHECK(t, keystroke(Kdown, 0, &com));
+		CT_EQ_INT(t, Maxdisp, im.sel);
+		CT_CHECK(t, draindraw(&dc) > 0);
+		checkstr(t, "scrolled first row", "c2", &dc.kouho[0]);
+		checkstr(t, "scrolled ninth row", "c10", &dc.kouho[8]);
+		CT_EQ_INT(t, 8, dc.sel);
+		CT_CHECK(t, keystroke('1', 0, &com));
+		checkstr(t, "scrolled row one", "c2", &com);
+	}
+
+	sclear(&com);
+	CT_CHECK(t, keystroke('e', Mctrl, &com));
+	if(typekeys(t, "many", &com)){
+		for(i = 0; i < 12; i++)
+			CT_CHECK(t, keystroke(Kdown, 0, &com));
+		CT_EQ_INT(t, 11, im.sel);
+		CT_CHECK(t, keystroke('9', 0, &com));
+		checkstr(t, "scrolled row nine", "c12", &com);
+	}
+
+	sclear(&com);
+	CT_CHECK(t, keystroke('e', Mctrl, &com));
+	if(typekeys(t, "smile", &com)){
+		CT_CHECK(t, keystroke(Ktab, 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(Kup, 0, &com));
+		CT_EQ_INT(t, 0, im.sel);
+		CT_CHECK(t, keystroke(Ktab, Mshift, &com));
+		CT_EQ_INT(t, 1, im.sel);
+		CT_CHECK(t, keystroke(Kret, 0, &com));
+		checkstr(t, "navigated candidate", "😄", &com);
+	}
+	emojiend(&f);
+}
+
+void
+engine_emoji_preedit_languages(struct ct *t)
+{
+	Emojifix f;
+	Str com, shown;
+	int p;
+
+	emojibegin(&f, 0);
+	init();
+	im.l = getlang(LangJP);
+	sclear(&com);
+	if(typekeys(t, "ka", &com)){
+		CT_CHECK(t, !keystroke(Kmodfirst+2, 0, &com));
+		checkstr(t, "modifier preserves Japanese", "ka", &im.pre);
+		CT_CHECK(t, keystroke('E', Mctrl|Mshift, &com));
+		checkstr(t, "committed Japanese preedit", "か", &com);
+		CT_CHECK(t, search.on);
+		CT_EQ_INT(t, LangJP, im.l->lang);
+		CT_CHECK(t, keystroke('a', 0, &com));
+		p = popup;
+		CT_CHECK(t, !keystroke(Kmodfirst+2, 0, &com));
+		CT_CHECK(t, search.on);
+		checkstr(t, "modifier preserves search", "a", &search.text);
+		CT_CHECK(t, keystroke('P', Mctrl|Mshift, &com));
+		CT_EQ_INT(t, !p, popup);
+		CT_CHECK(t, search.on);
+		CT_CHECK(t, !keystroke(Kmodfirst+2, 0, &com));
+		CT_CHECK(t, keystroke('E', Mctrl|Mshift, &com));
+		CT_CHECK(t, !search.on);
+		sclear(&com);
+		if(typekeys(t, "na", &com)){
+			shown = shownpre(&im);
+			checkstr(t, "Japanese resumed", "な", &shown);
+		}
+	}
+
+	init();
+	im.l = getlang(LangKO);
+	sclear(&com);
+	if(typekeys(t, "rk", &com)){
+		CT_CHECK(t, keystroke('e', Mctrl, &com));
+		checkstr(t, "committed Korean preedit", "가", &com);
+		if(typekeys(t, "smile", &com)){
+			CT_CHECK(t, keystroke('1', 0, &com));
+			checkstr(t, "Korean selection", "가😀", &com);
+			sclear(&com);
+			if(typekeys(t, "rk", &com))
+				checkstr(t, "Korean resumed", "가", &im.pre);
+		}
+	}
+
+	init();
+	im.l = &testvi;
+	sclear(&com);
+	if(typekeys(t, "as", &com)){
+		CT_CHECK(t, keystroke('e', Mctrl, &com));
+		checkstr(t, "committed Telex preedit", "á", &com);
+		CT_CHECK(t, keystroke(Kesc, 0, &com));
+		CT_CHECK(t, !search.on);
+		CT_EQ_PTR(t, &testvi, im.l);
+		sclear(&com);
+		if(typekeys(t, "as", &com)){
+			shown = shownpre(&im);
+			checkstr(t, "Telex resumed", "á", &shown);
+			checkstr(t, "Telex raw resumed", "as", &im.raw);
+		}
+	}
+	emojiend(&f);
+}
+
+void
+engine_emoji_start_and_unknown(struct ct *t)
+{
+	Emojifix f;
+	Str com;
+	char want[Maxrunes+1];
+	int i;
+
+	emojibegin(&f, 0);
+	sclear(&com);
+	if(typekeys(t, "dkssud", &com)){
+		CT_CHECK(t, keystroke('e', Mctrl, &com));
+		checkstr(t, "committed Korean preedit", "안녕", &com);
+		CT_CHECK(t, search.on);
+		CT_EQ_INT(t, LangKO, im.l->lang);
+		if(typekeys(t, "xyz", &com)){
+			checkstr(t, "unknown displayed query", "xyz", &search.text);
+			CT_EQ_INT(t, 0, im.nkouho);
+			CT_CHECK(t, keystroke(Kret, 0, &com));
+			checkstr(t, "unknown query commit", "안녕xyz", &com);
+			CT_CHECK(t, !search.on);
+			CT_EQ_INT(t, LangKO, im.l->lang);
+		}
+	}
+
+	sclear(&com);
+	CT_CHECK(t, keystroke('e', Mctrl, &com));
+	if(typekeys(t, "xyz", &com)){
+		CT_CHECK(t, !keystroke(Kspec|0x51, 0, &com));
+		checkstr(t, "special commits shown query", "xyz", &com);
+		CT_CHECK(t, !search.on);
+		CT_EQ_INT(t, LangKO, im.l->lang);
+	}
+
+	sclear(&com);
+	CT_CHECK(t, keystroke('e', Mctrl, &com));
+	for(i = 0; i < Maxrunes; i++)
+		CT_CHECK(t, keystroke('q', 0, &com));
+	CT_EQ_INT(t, Maxrunes, search.raw.n);
+	CT_CHECK(t, search.on);
+	CT_CHECK(t, !keystroke('z', 0, &com));
+	memset(want, 'q', Maxrunes);
+	want[Maxrunes] = '\0';
+	checkstr(t, "bounded query commit", want, &com);
+	CT_CHECK(t, !search.on);
+	emojiend(&f);
+}
+
+void
+engine_emoji_dictionary_identity(struct ct *t)
+{
+	Dictres res;
+	Emojifix f;
+
+	emojibegin(&f, 0);
+	im.pre = mkstr("same");
+	im.kouho[0] = mkstr("old");
+	im.nkouho = 1;
+	im.sel = 0;
+	memset(&res, 0, sizeof res);
+	res.key = im.pre;
+	res.kouho[0] = mkstr("wrong");
+	res.nkouho = 1;
+	res.lang = LangJP;
+	dictresult(&res);
+	CT_EQ_INT(t, 1, im.nkouho);
+	checkstr(t, "wrong language ignored", "old", &im.kouho[0]);
+
+	res.lang = LangKO;
+	res.key = mkstr("other");
+	dictresult(&res);
+	CT_EQ_INT(t, 1, im.nkouho);
+	checkstr(t, "wrong preedit ignored", "old", &im.kouho[0]);
+
+	res.key = im.pre;
+	search.on = 1;
+	dictresult(&res);
+	CT_EQ_INT(t, 1, im.nkouho);
+	checkstr(t, "search response ignored", "old", &im.kouho[0]);
+
+	search.on = 0;
+	res.kouho[0] = mkstr("right");
+	dictresult(&res);
+	CT_EQ_INT(t, 1, im.nkouho);
+	CT_EQ_INT(t, -1, im.sel);
+	checkstr(t, "matching response accepted", "right", &im.kouho[0]);
+	emojiend(&f);
+}

tests/ipc_test.c

@@ -0,0 +1,15 @@
+#include "test.h"
+
+void
+ipc_masks_modifiers(struct ct *t)
+{
+	uchar buf[Ipcreqsz];
+	u32int key, mod;
+	int want;
+
+	ipcpackreq(buf, 1, Mctrl|Malt|(1<<1)|(1<<4)|(1<<7), 0x1f642);
+	ipcunpackreq(buf, &want, &mod, &key);
+	CT_EQ_INT(t, 1, want);
+	CT_EQ_UINT(t, Mctrl|Malt, mod);
+	CT_EQ_UINT(t, 0x1f642, key);
+}

tests/mkemoji_test.py

@@ -0,0 +1,85 @@
+#!/usr/bin/env python3
+
+import subprocess
+import sys
+import tempfile
+import unittest
+from pathlib import Path
+
+
+ROOT = Path(__file__).resolve().parents[1]
+MKEMOJI = ROOT / "map" / "mkemoji"
+
+
+def generate(source=None):
+    args = [sys.executable, "-B", str(MKEMOJI)]
+    if source is not None:
+        args.append(str(source))
+    return subprocess.run(args, capture_output=True, text=True, check=False)
+
+
+def table(output):
+    return dict(line.split("\t", 1) for line in output.splitlines())
+
+
+class MkemojiTest(unittest.TestCase):
+    def source(self, text):
+        tmp = tempfile.TemporaryDirectory()
+        path = Path(tmp.name) / "emoji.src"
+        path.write_text(text, encoding="utf-8")
+        self.addCleanup(tmp.cleanup)
+        return path
+
+    def test_production_digit_slots(self):
+        result = generate()
+        self.assertEqual(result.returncode, 0, result.stderr)
+        data = table(result.stdout)
+        self.assertEqual(data["^"].split()[:9], list("¹²³⁴⁵⁶⁷⁸⁹"))
+        self.assertEqual(data["_"].split()[:9], list("₁₂₃₄₅₆₇₈₉"))
+        self.assertEqual(data["<"].split()[2], "♥")
+        self.assertEqual(data["^1"], "¹")
+        self.assertEqual(data["_2"], "₂")
+        self.assertEqual(data["<3"], "♥")
+
+    def test_fold_normalize_and_exact_first(self):
+        source = self.source(
+            "β\tALPHABET\n"
+            "α\talpha\n"
+            "e\u0301\tE\u0301\n"
+            "#\thash\n"
+        )
+        first = generate(source)
+        second = generate(source)
+        self.assertEqual(first.returncode, 0, first.stderr)
+        self.assertEqual(first.stdout, second.stdout)
+        data = table(first.stdout)
+        self.assertEqual(data["alpha"].split(), ["α", "β"])
+        self.assertEqual(data["al"].split(), ["β", "α"])
+        self.assertEqual(data["é"], "é")
+        self.assertEqual(data["hash"], "#")
+
+    def test_rejects_malformed_and_oversized_rows(self):
+        bad = [
+            "x\n",
+            "bad result\ta\n",
+            "x\t\n",
+            "x\t a\n",
+            "\0\ta\n",
+            "x\ta\0b\n",
+            "x" * 65 + "\ta\n",
+            "x\t" + "a" * 65 + "\n",
+        ]
+        for text in bad:
+            with self.subTest(text=repr(text)):
+                result = generate(self.source(text))
+                self.assertEqual(result.returncode, 1)
+
+    def test_accepts_runtime_boundary(self):
+        source = self.source("x" * 64 + "\t" + "a" * 64 + "\n")
+        result = generate(source)
+        self.assertEqual(result.returncode, 0, result.stderr)
+        self.assertIn("a" * 64 + "\t" + "x" * 64, result.stdout)
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test.h

@@ -33,7 +33,15 @@ 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_emoji_single_candidate(struct ct*);
+void	engine_emoji_queries(struct ct*);
+void	engine_emoji_digit_aliases(struct ct*);
+void	engine_emoji_navigation(struct ct*);
+void	engine_emoji_preedit_languages(struct ct*);
+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	ipc_masks_modifiers(struct ct*);
 
 #endif

tests/unit_test.c

@@ -80,8 +80,16 @@ 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/emoji-single-candidate", engine_emoji_single_candidate },
+	{ "engine/emoji-queries", engine_emoji_queries },
+	{ "engine/emoji-digit-aliases", engine_emoji_digit_aliases },
+	{ "engine/emoji-navigation", engine_emoji_navigation },
+	{ "engine/emoji-preedit-languages", engine_emoji_preedit_languages },
+	{ "engine/emoji-start-and-unknown", engine_emoji_start_and_unknown },
+	{ "engine/emoji-dictionary-identity", engine_emoji_dictionary_identity },
 	{ "dict/candidates", dictionary_candidates },
 	{ "dict/misses-clear-result", dictionary_misses_clear_result },
+	{ "ipc/masks-modifiers", ipc_masks_modifiers },
 };
 
 void

win.c

@@ -90,16 +90,18 @@ fill(u32int *buf, int n, u32int color)
 static void
 drawkouho(Drawcmd *dc, int first, int n, int w, int h)
 {
-	int sely, y, i;
+	int npre, sely, y, i;
 	Str *s;
 
 	fill(img, w * h, Colbg);
-	drawstr(img, 0, 0, dc->pre.r, dc->pre.n, w, h);
+	npre = dc->pre.n != 0;
-	if(dc->sel >= 0){
+	if(npre)
-		sely = Fontsz + dc->sel * Fontsz;
+		drawstr(img, 0, 0, dc->pre.r, dc->pre.n, w, h);
+	if(dc->sel >= 0 && dc->sel < n){
+		sely = (npre + dc->sel) * Fontsz;
 		fill(img + sely * w, Fontsz * w, Colsel);
 	}
-	for(i = 0, y = Fontsz; i < n; i++, y += Fontsz){
+	for(i = 0, y = npre * Fontsz; i < n; i++, y += Fontsz){
 		s = &dc->kouho[first+i];
 		putfont(img, w, h, 0, y, '1' + i + Asciitofull);
 		drawstr(img, 2*Fontsz, y, s->r, s->n, w, h);
@@ -124,13 +126,18 @@ winhide(void)
 static void
 winshow(Drawcmd *dc)
 {
-	int px, py, w, h, i, n, maxw;
+	int npre, px, py, w, h, i, n, maxw;
 	u32int vals[4];
 	xcb_query_pointer_reply_t *ptr;
 	xcb_query_pointer_cookie_t cookie;
 
+	n = min(max(dc->nkouho, 0), Maxdisp);
+	npre = dc->pre.n != 0;
+	if(n == 0 && npre == 0){
+		winhide();
+		return;
+	}
 	cookie = xcb_query_pointer(conn, scr->root);
-	n = dc->nkouho;
 	maxw = dc->pre.n;
 	for(i = 0; i < n; i++)
 		maxw = max(maxw, dc->kouho[i].n);
@@ -140,7 +147,7 @@ winshow(Drawcmd *dc)
 	px = ptr->root_x + 10;
 	py = ptr->root_y + 10;
 	free(ptr);
-	vals[3] = h = (n + 1) * Fontsz;
+	vals[3] = h = (n + npre) * Fontsz;
 	vals[2] = w = (maxw + 3) * Fontsz;
 	vals[1] = py = max(0, min(py, scr->height_in_pixels - h));
 	vals[0] = px = max(0, min(px, scr->width_in_pixels - w));