more deinflections + tests

1 files changed, 48 insertions, 8 deletions

diff --git a/language/tags.ts b/language/tags.ts
index 312a594..de1a3ea 100644
--- a/language/tags.ts
+++ b/language/tags.ts
@@ -131,7 +131,28 @@ export const Tag = {
 			Ra: "infl:cond:ra",
 		},
 		/** @constant makes a verb obligatory (e.g. 入ってはいけない) */
-		Obliged: "infl:must",
+		Obligatory: "infl:must",
+		/** @constant verbs that someone wants to do / be done */
+		Desirable: {
+			/** @constant 〜たい endings (e.g. 買いたい) */
+			Itai: "infl:desire:itai",
+			/** @constant 〜おう endings (e.g. 寝よう) */
+			Volitional: "infl:desire:volitional",
+		},
+		/** @constant makes a verb an attempt */
+		Attempt: {
+			/** @constant 〜みる to try something out (e.g. 飲んでみた) */
+			Miru: "infl:attempt:miru",
+			/** @constant 〜とする attempts (e.g. 入ろうとしている) */
+			ToSuru: "infl:attempt:tosuru",
+		},
+		/** @constant temporary tags (removed by parseTags) */
+		Temporary: {
+			/** @constant particle of obligatory conjugation (e.g. 行かない*と*だめ), or colloquial abbreviation */
+			ObligatoryParticle: "infl:tmp:must:prt",
+			/** @constant resulting action part of obligatory conjugation (e.g. 行かないと*だめ*) */
+			ObligatoryResult: "infl:tmp:must:res",
+		},
 	},
 	/** @constant uncategorized tags */
 	Auxiliary: {
@@ -165,12 +186,17 @@ export function parseTags(input: string) {
 		// 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;
+
+		// skip -te suffix tag if used for
+		if (tag == Tag.Inflection.Suffix.Te && filteredTags.anyOf([
+			Tag.Inflection.Tense.Continuous, // base for continuous tense
+			Tag.Inflection.Obligatory, // base for obligatory inflection
+			Tag.Inflection.Attempt.Miru, // base for 〜みる attempt
+		])) continue;
+	
+		// skip volitional tag if used for 〜とする attempt
+		if (tag == Tag.Inflection.Desirable.Volitional &&
+				filteredTags.anyOf([Tag.Inflection.Attempt.ToSuru])) continue;
 
 		// normalize multiple Inflection.Negative to single Inflection.Affirmative or Inflection.Negative
 		if (tag == Tag.Inflection.Negative) {
@@ -180,7 +206,21 @@ export function parseTags(input: string) {
 
 		filteredTags.push(tag);
 	}
+
+	// negative + と without resulting action = implicit affirmative obligatory
+	if (filteredTags.includes(Tag.Inflection.Temporary.ObligatoryParticle) &&
+			!filteredTags.includes(Tag.Inflection.Temporary.ObligatoryResult)) {
+		negationCount = 0; // -> make resulting tags affirmative
+	}
+
+	// normalize affirmative/negative
 	filteredTags.push(negationCount % 2 == 0 ? Tag.Inflection.Affirmative : Tag.Inflection.Negative);
-	return filteredTags.set().arr() as TokenTags; // make sure array doesn't contain duplicates
+
+	// filter any remaining temporary tags
+	type tempTag = typeof Tag.Inflection.Temporary[keyof typeof Tag.Inflection.Temporary];
+	filteredTags = filteredTags.filter(t => !Object.values(Tag.Inflection.Temporary).includes(t as tempTag));
+
+	// filter any duplicates
+	return filteredTags.set().arr() as TokenTags;
 }