dict: lookups take a trie; a prefix search is its own function; no self-key rule

dictlookup took a Lang for two reasons that were not its business: to
know whether to walk below the key (the emoji dictionary) and to drop a
candidate equal to the key except there.  No dictionary lists its key
among its candidates — the rule was left from a first version that put
the reading first itself — so it goes, and the walk is dictprefix(),
called by the emoji search alone.

dict.c

@@ -1,13 +1,9 @@
 #include "dat.h"
 #include "fn.h"
 
-/*
- * Appends the space-separated words of a node's entry to out[], up to
- * max: its candidates, minus the key itself, except in the emoji
- * dictionary, where a query may name its own answer.
- */
+/* Appends the space-separated words of a node's entry to out[], up to max. */
 static int
-words(Lang *l, Str *key, Tnode *nd, Str *out, int n, int max)
+words(Tnode *nd, Str *out, int n, int max)
 {
 	char *p, *e, *sp;
 	Str tmp;
@@ -22,8 +18,7 @@ words(Lang *l, Str *key, Tnode *nd, Str *out, int n, int max)
 		sp = p;
 		while(p < e && *p != ' ')
 			p++;
-		if(sinit(&tmp, sp, p - sp) && tmp.n > 0 &&
-		   (l->lang == LangEMOJI || scmp(&tmp, key) != 0))
+		if(sinit(&tmp, sp, p - sp) && tmp.n > 0)
 			out[n++] = tmp;
 	}
 	return n;
@@ -31,33 +26,40 @@ words(Lang *l, Str *key, Tnode *nd, Str *out, int n, int max)
 
 /* The entries at and below a node, in the dictionary's order. */
 static int
-below(Lang *l, Str *key, int ni, Str *out, int n, int max)
+below(Trie *t, int ni, Str *out, int n, int max)
 {
 	Tnode *nd;
 	int ci;
 
-	nd = &l->dict->nodes[ni];
-	n = words(l, key, nd, out, n, max);
-	for(ci = nd->child; ci >= 0 && n < max; ci = l->dict->nodes[ci].sibling)
-		n = below(l, key, ci, out, n, max);
+	nd = &t->nodes[ni];
+	n = words(nd, out, n, max);
+	for(ci = nd->child; ci >= 0 && n < max; ci = t->nodes[ci].sibling)
+		n = below(t, ci, out, n, max);
 	return n;
 }
 
-/*
- * Fills out[] with up to max candidates for key.  The emoji dictionary is
- * searched by prefix, the key's own entry first; the others exactly.
- */
+/* Fills out[] with up to max candidates for key: the words of its entry. */
 int
-dictlookup(Lang *l, Str *key, Str *out, int max)
+dictlookup(Trie *t, Str *key, Str *out, int max)
 {
 	int ni;
 
-	ni = key->n == 0 ? -1 : trienode(l->dict, key);
+	ni = key->n == 0 ? -1 : trienode(t, key);
 	if(ni < 0)
 		return 0;
-	if(l->lang == LangEMOJI)
-		return below(l, key, ni, out, 0, max);
-	return words(l, key, &l->dict->nodes[ni], out, 0, max);
+	return words(&t->nodes[ni], out, 0, max);
+}
+
+/* As dictlookup, for every entry key is a prefix of, key's own first. */
+int
+dictprefix(Trie *t, Str *key, Str *out, int max)
+{
+	int ni;
+
+	ni = key->n == 0 ? -1 : trienode(t, key);
+	if(ni < 0)
+		return 0;
+	return below(t, ni, out, 0, max);
 }
 
 static Trie*

fn.h

@@ -21,7 +21,8 @@ int	trielookup(Trie*, Str*, char**, int*);
 Lang*	getlang(int);
 void	langinit(char*);
 
-int	dictlookup(Lang*, Str*, Str*, int);
+int	dictlookup(Trie*, Str*, Str*, int);
+int	dictprefix(Trie*, Str*, Str*, int);
 
 void	drawthread(void*);
 void	popuparea(Area*, int, Area*, int, int, Area*);

strans.c

@@ -289,7 +289,7 @@ dictqokuri(Str *reading)
 		key = *reading;
 		key.n = i;
 		sputr(&key, c);
-		n = dictlookup(im.l, &key, kouho, Maxkouho);
+		n = dictlookup(im.l->dict, &key, kouho, Maxkouho);
 		sclear(&okuri);
 		for(j = i - (reading->r[i-1] == L'っ'); j < reading->n; j++)
 			sputr(&okuri, reading->r[j]);
@@ -310,7 +310,7 @@ dictqjp(void)
 	if(!isjp(&im) || !jpreading(im.l->map, &im.pre, &im.raw, &reading) ||
 	   reading.n == 0)
 		return;
-	im.nkouho = dictlookup(im.l, &reading, im.kouho, Maxkouho);
+	im.nkouho = dictlookup(im.l->dict, &reading, im.kouho, Maxkouho);
 	dictqokuri(&reading);
 }
 
@@ -572,13 +572,13 @@ emojiquery(void)
 	transstr(im.l, nil, &search.raw, &local);
 	foldascii(&local);
 	clearkouho();
-	n = dictlookup(getlang(LangEMOJI), &raw, kouho, Maxkouho);
+	n = dictprefix(getlang(LangEMOJI)->dict, &raw, kouho, Maxkouho);
 	rawhit = n != 0;
 	for(i = 0; i < n; i++)
 		addkouho(&kouho[i]);
 	localhit = 0;
 	if(scmp(&raw, &local) != 0){
-		n = dictlookup(getlang(LangEMOJI), &local, kouho, Maxkouho);
+		n = dictprefix(getlang(LangEMOJI)->dict, &local, kouho, Maxkouho);
 		localhit = n != 0;
 		for(i = 0; i < n; i++)
 			addkouho(&kouho[i]);
@@ -593,8 +593,8 @@ hanjaquery(void)
 {
 	transstr(im.l, &search.seed, &search.raw, &search.text);
 	clearkouho();
-	im.nkouho = dictlookup(getlang(LangHANJA), &search.text, im.kouho,
-		Maxkouho);
+	im.nkouho = dictlookup(getlang(LangHANJA)->dict, &search.text,
+		im.kouho, Maxkouho);
 	selectfirst();
 }
 

tests/dict_test.c

@@ -15,10 +15,10 @@ dictionary_candidates(struct ct *t)
 	lang = getlang(LangJP);
 	saved = lang->dict;
 	lang->dict = trienew();
-	trieput(lang->dict, "かな", strlen("かな"), "  候補1  かな 候補2  ",
-		strlen("  候補1  かな 候補2  "));
+	trieput(lang->dict, "かな", strlen("かな"), "  候補1  候補2  ",
+		strlen("  候補1  候補2  "));
 	key = mkstr("かな");
-	n = dictlookup(lang, &key, kouho, Maxkouho);
+	n = dictlookup(lang->dict, &key, kouho, Maxkouho);
 	if(CT_EQ_INT(t, 2, n)){
 		checkstr(t, "candidate 1", "候補1", &kouho[0]);
 		checkstr(t, "candidate 2", "候補2", &kouho[1]);
@@ -34,13 +34,13 @@ dictionary_candidates(struct ct *t)
 	*p = '\0';
 	trieput(lang->dict, "key", 3, many, strlen(many));
 	key = mkstr("key");
-	n = dictlookup(lang, &key, kouho, Maxkouho);
+	n = dictlookup(lang->dict, &key, kouho, Maxkouho);
 	if(CT_EQ_INT(t, Maxkouho, n)){
 		checkstr(t, "first capped candidate", "c00", &kouho[0]);
 		snprint(item, sizeof item, "c%02d", Maxkouho-1);
 		checkstr(t, "last capped candidate", item, &kouho[Maxkouho-1]);
 	}
-	CT_EQ_INT(t, 3, dictlookup(lang, &key, kouho, 3));
+	CT_EQ_INT(t, 3, dictlookup(lang->dict, &key, kouho, 3));
 	trieclose(lang->dict);
 	lang->dict = saved;
 }
@@ -56,28 +56,33 @@ dictionary_misses(struct ct *t)
 	saved = lang->dict;
 	lang->dict = trienew();
 	key = mkstr("");
-	CT_EQ_INT(t, 0, dictlookup(lang, &key, kouho, Maxkouho));
+	CT_EQ_INT(t, 0, dictlookup(lang->dict, &key, kouho, Maxkouho));
 	key = mkstr("missing");
-	CT_EQ_INT(t, 0, dictlookup(lang, &key, kouho, Maxkouho));
-	CT_EQ_INT(t, 0, dictlookup(getlang(LangKO), &key, kouho, Maxkouho));
+	CT_EQ_INT(t, 0, dictlookup(lang->dict, &key, kouho, Maxkouho));
+	CT_EQ_INT(t, 0, dictlookup(getlang(LangKO)->dict, &key, kouho, Maxkouho));
 	trieclose(lang->dict);
 	lang->dict = saved;
 }
 
 void
-dictionary_emoji_identity(struct ct *t)
+dictionary_prefix(struct ct *t)
 {
 	Str kouho[Maxkouho], key;
-	Trie *saved;
-	Lang *lang;
+	Trie *dict;
 
-	lang = getlang(LangEMOJI);
-	saved = lang->dict;
-	lang->dict = trienew();
-	trieput(lang->dict, "é", strlen("é"), "é", strlen("é"));
-	key = mkstr("é");
-	if(CT_EQ_INT(t, 1, dictlookup(lang, &key, kouho, Maxkouho)))
-		checkstr(t, "emoji identity candidate", "é", &kouho[0]);
-	trieclose(lang->dict);
-	lang->dict = saved;
+	dict = trienew();
+	trieput(dict, "smile", 5, "A B", 3);
+	trieput(dict, "smiley", 6, "B C", 3);
+	trieput(dict, "sad", 3, "D", 1);
+	key = mkstr("smile");
+	if(CT_EQ_INT(t, 4, dictprefix(dict, &key, kouho, Maxkouho))){
+		checkstr(t, "own entry first", "A", &kouho[0]);
+		checkstr(t, "entry below", "C", &kouho[3]);
+	}
+	key = mkstr("s");
+	CT_EQ_INT(t, 5, dictprefix(dict, &key, kouho, Maxkouho));
+	CT_EQ_INT(t, 2, dictprefix(dict, &key, kouho, 2));
+	key = mkstr("x");
+	CT_EQ_INT(t, 0, dictprefix(dict, &key, kouho, Maxkouho));
+	trieclose(dict);
 }

tests/test.h

@@ -95,7 +95,7 @@ void	engine_randomized_stress(struct ct*);
 void	engine_full_boundary_passthrough(struct ct*);
 void	dictionary_candidates(struct ct*);
 void	dictionary_misses(struct ct*);
-void	dictionary_emoji_identity(struct ct*);
+void	dictionary_prefix(struct ct*);
 void	ipc_masks_modifiers(struct ct*);
 void	ipc_control_and_caret_frames(struct ct*);
 void	ipc_runtime_path(struct ct*);

tests/unit_test.c

@@ -118,7 +118,7 @@ static const struct ct_test tests[] = {
 	{ "engine/full-boundary-passthrough", engine_full_boundary_passthrough },
 	{ "dict/candidates", dictionary_candidates },
 	{ "dict/misses", dictionary_misses },
-	{ "dict/emoji-identity", dictionary_emoji_identity },
+	{ "dict/prefix", dictionary_prefix },
 	{ "ipc/masks-modifiers", ipc_masks_modifiers },
 	{ "ipc/control-caret-frames", ipc_control_and_caret_frames },
 	{ "ipc/runtime-path", ipc_runtime_path },