rename conditionals and deconjugate obligatory inflections

4 files changed, 53 insertions, 13 deletions

diff --git a/db/dict/deinflections.sql b/db/dict/deinflections.sql
index a6070a8..7d67b02 100644
--- a/db/dict/deinflections.sql
+++ b/db/dict/deinflections.sql
@@ -142,17 +142,23 @@ insert into deinflection_temp values
 	('infl:potential', 'ありえる', 'ある', 'a', ''), -- exception
 
 	-- conditionals <https://guidetojapanese.org/learn/grammar/conditionals>
-	('infl:suffix:ba', 'えば', 'う', 'a', 'u'),
-	('infl:suffix:ba', 'けば', 'く', 'a', 'u'),
-	('infl:suffix:ba', 'げば', 'ぐ', 'a', 'u'),
-	('infl:suffix:ba', 'せば', 'す', 'a', 'u'),
-	('infl:suffix:ba', 'てば', 'つ', 'a', 'u'),
-	('infl:suffix:ba', 'ねば', 'ぬ', 'a', 'u'),
-	('infl:suffix:ba', 'べば', 'ぶ', 'a', 'u'),
-	('infl:suffix:ba', 'めば', 'む', 'a', 'u'),
-	('infl:suffix:ba', 'れば', 'る', 'a', 'u ru'),
-	('infl:suffix:ba', 'ければ', 'い', 'a', 'a'),
+	('infl:cond:ba',   'えば', 'う', 'a', 'u'),
+	('infl:cond:ba',   'けば', 'く', 'a', 'u'),
+	('infl:cond:ba',   'げば', 'ぐ', 'a', 'u'),
+	('infl:cond:ba',   'せば', 'す', 'a', 'u'),
+	('infl:cond:ba',   'てば', 'つ', 'a', 'u'),
+	('infl:cond:ba',   'ねば', 'ぬ', 'a', 'u'),
+	('infl:cond:ba',   'べば', 'ぶ', 'a', 'u'),
+	('infl:cond:ba',   'めば', 'む', 'a', 'u'),
+	('infl:cond:ba',   'れば', 'る', 'a', 'u ru'),
+	('infl:cond:ba', 'ければ', 'い', 'a', 'a'),
 	-- TODO: 〜であれば (deconjugates to です i think?)
+	('infl:cond:ra', 'ら', '', 'a', 'a'), -- TODO: constrain valid matches to after past conjugation?
+
+	-- obligation <https://guidetojapanese.org/learn/grammar/must>
+	('infl:must infl:negative', 'はだめ', '', 'a', 'a'), -- built-in negative because だめ can't be deconjugated
+	('infl:must', 'はいける', '', 'a', 'a'), -- はいけない -> positive
+	('infl:must', 'はなる', '', 'a', 'a'), -- はならない -> positive
 
 	-- auxiliary rules
 	('class:verb:suru-included', 'する', '', 's', ''); -- deconjugate suru verbs into stem
diff --git a/import/jmdict/jmdict.ts b/import/jmdict/jmdict.ts
index 6109c9b..1d14ade 100644
--- a/import/jmdict/jmdict.ts
+++ b/import/jmdict/jmdict.ts
@@ -21,6 +21,7 @@ const tagLookup = {
 	["class/v1"]: Tag.Class.Verb.Ru,
 	["class/v5"]: Tag.Class.Verb.U,
 	["class/v5k"]: Tag.Class.Verb.U,
+	["class/v5k-s"]: Tag.Class.Verb.U,
 	["class/v5uru"]: Tag.Class.Verb.U,
 	["class/v5r-i"]: Tag.Class.Verb.U,
 	["class/v5u-s"]: Tag.Class.Verb.U,
diff --git a/language/tags.ts b/language/tags.ts
index a9fc5ca..312a594 100644
--- a/language/tags.ts
+++ b/language/tags.ts
@@ -67,6 +67,15 @@ export const Tag = {
 	 * e.g. 来ない -> 来る [infl:negative]
 	 */
 	Inflection: {
+		/**
+		 * @constant affirmative conjugations
+		 *
+		 * This conjugation should not be added by any deconjugation rules, but is
+		 * calculated based on the amount of negations. Even counts of negative
+		 * inflections (including 0) add this tag, while odd counts don't add this
+		 * tag.
+		 */
+		Affirmative: "infl:affirmative",
 		/** @constant negative conjugations */
 		Negative: "infl:negative",
 		/** @constant time-related conjugations */
@@ -89,8 +98,6 @@ export const Tag = {
 			Te: "infl:suffix:te",
 			/** @constant -tari ending (e.g. 遊んだり) */
 			Tari: "infl:suffix:tari",
-			/** @constant -ba ending for conditionals (e.g. 泳げれば)*/
-			Ba: "infl:suffix:ba",
 		},
 		/** @constant internal deinflection rules */
 		Reason: {
@@ -116,6 +123,15 @@ export const Tag = {
 		Causative: "infl:causative",
 		/** @constant imperative form (e.g. 聞け) */
 		Command: "infl:command",
+		/** @constant conditional forms */
+		Conditional: {
+			/** @constant -ba ending (e.g. 泳げれば) */
+			Ba: "infl:cond:ba",
+			/** @constant -ra ending (e.g. 取ったら) */
+			Ra: "infl:cond:ra",
+		},
+		/** @constant makes a verb obligatory (e.g. 入ってはいけない) */
+		Obliged: "infl:must",
 	},
 	/** @constant uncategorized tags */
 	Auxiliary: {
@@ -140,16 +156,31 @@ export type TokenTags = Array<TokenTag>;
 export function parseTags(input: string) {
 	var tags = input.replaceAll(/ +/g, " ").trim().split(" ") as TokenTag[];
 	var filteredTags: TokenTag[] = [];
+	var negationCount = 0;
 	for (var tag of tags) {
+		// conjugations that are used as "stepping stones" for others should be
+		// filtered in this loop. checking if a combination of tags is valid should
+		// be done in ./parser.ts
+
 		// skip past tense tag if used as step for -te and -tari inflection
 		if (tag == Tag.Inflection.Tense.Past &&
 				filteredTags.anyOf([Tag.Inflection.Suffix.Te, Tag.Inflection.Suffix.Tari])) continue;
 		// skip -te suffix tag if it's a base for continuous tense
 		if (tag == Tag.Inflection.Suffix.Te &&
 				filteredTags.anyOf([Tag.Inflection.Tense.Continuous])) continue;
+		// skip -te suffix tag if it's a base for obligatory inflection
+		if (tag == Tag.Inflection.Suffix.Te &&
+				filteredTags.anyOf([Tag.Inflection.Obliged])) continue;
+
+		// normalize multiple Inflection.Negative to single Inflection.Affirmative or Inflection.Negative
+		if (tag == Tag.Inflection.Negative) {
+			negationCount++;
+			continue;
+		}
 
 		filteredTags.push(tag);
 	}
+	filteredTags.push(negationCount % 2 == 0 ? Tag.Inflection.Affirmative : Tag.Inflection.Negative);
 	return filteredTags.set().arr() as TokenTags; // make sure array doesn't contain duplicates
 }
 
diff --git a/test/deinflection/cases.ts b/test/deinflection/cases.ts
index e0b2137..143e2a4 100644
--- a/test/deinflection/cases.ts
+++ b/test/deinflection/cases.ts
@@ -38,8 +38,10 @@ export default [
 	{ input: "落ちられる", mustHave: [ Inflection.Potential ], mustNotHave: [], },
 	{ input: "言われる", mustHave: [ Inflection.Passive ], mustNotHave: [], },
 	{ input: "喋っている", mustHave: [ Inflection.Tense.Continuous ], mustNotHave: [ Inflection.Suffix.Te ], },
-	{ input: "泳げれば", mustHave: [ Inflection.Suffix.Ba ], mustNotHave: [], },
+	{ input: "泳げれば", mustHave: [ Inflection.Conditional.Ba ], mustNotHave: [], },
 	{ input: "取らなければ", mustHave: [ Inflection.Potential, Inflection.Negative ], mustNotHave: [], },
+	{ input: "入ってはいけない", mustHave: [ Inflection.Obliged, Inflection.Negative ], mustNotHave: [ Inflection.Affirmative ], }, // obliged + negative = must not ~
+	{ input: "行かなくてはなりません", mustHave: [ Inflection.Obliged, Inflection.Affirmative ], mustNotHave: [ Inflection.Negative ], }, // obliged + affirmative = must ~
 	// TODO: りゃ for いることは
 	// TODO: じゃ for では
 	// TODO: なきゃ + なくちゃ