More Logos RMAC expansion fixes

Also, the unused expansion pattern detection was faulty - fix it and adjust patterns accordingly
schierlm · Feb 15, 2024 · bfde314 · bfde314
1 parent 786bdc9
commit bfde314
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 11 deletions.
diff --git a/biblemulticonverter-logos/src/main/java/biblemulticonverter/logos/format/LogosHTML.java b/biblemulticonverter-logos/src/main/java/biblemulticonverter/logos/format/LogosHTML.java
@@ -539,7 +539,7 @@ private static String tagOne(String str, Pattern pattern, String languageCode) {
 
 	protected static String convertMorphology(String rmac) {
 		rmac = rmac.replaceFirst("^(S-[123])[SP]([NVGDA][SP][MFN](-(S|C|ABB|I|N|K|ATT|ARAM|HEB))?)$", "$1$2");
-		Matcher m = Utils.compilePattern("([NARCDTKIXQFSP])(-([123]?)([NVGDA][SP][MFN]?))?(?:-([PLT]|[PL]G|LI|NUI))?(-(S|C|ABB|I|N|K|ATT|ARAM|HEB))?").matcher(rmac);
+		Matcher m = Utils.compilePattern("([NARCDTKIXQFSP])(-([123]?)([NVGDA][SP][MFN]?))?(?:-([PLT]|[PL]G|LI|NUI))?(?:-(S|C|ABB|I|N|K|ATT|ARAM|HEB))?").matcher(rmac);
 		if (m.matches()) {
 			if (rmac.startsWith("N-LI"))
 				return "XL";
@@ -590,7 +590,7 @@ else if (flags.length() == 2 && !cops.isEmpty())
 			case 'T': // @D[ADGNV][DPS][FMN]
 				return "D" + flags;
 			case 'Q': // correlative or interrogative -> RK or RI
-				return "RX" + person + flags;
+				return person.isEmpty() ? "R" : "R?" + person + flags;
 			case 'R': // @R[CDFIKNPRSX][123][ADGNV][DPS][FMN][AP]
 			case 'X':
 			case 'F':

diff --git a/...lticonverter-logos/src/test/java/biblemulticonverter/logos/format/RMACConversionTest.java b/...lticonverter-logos/src/test/java/biblemulticonverter/logos/format/RMACConversionTest.java
@@ -38,17 +38,17 @@ public class RMACConversionTest {
 	};
 
 	private static final String[] LOGOS_PATTERNS = {
-			"J[ADGNV?][DPS?][FMN?][COPS?]", // Adjective
-			"B[CEIKNPSX?]", // Adverb
-			"D[ADGNV?][DPS?][FMN?]", // Article
+			"J[ADGNV?][PS?][FMN?][COS?]", // Adjective "J[ADGNV?][DPS?][FMN?][COPS?]"
+			"B[CIKNS?]", // Adverb "B[CEIKNPSX?]"
+			"D[ADGNV?][PS?][FMN?]", // Article "D[ADGNV?][DPS?][FMN?]"
 			"I", // Interjection
-			"C[[ACADALAMANAPARATAZLALCLDLILKLMLNLTLXSCSE??]]", // Conjunction
-			"T[CEIKNPSX?]", // Particle
+			"C", // Conjunction C[[ACADALAMANAPARATAZLALCLDLILKLMLNLTLXSCSE??]]
+			"T[CIN?]", // Particle "T[CEIKNPSX?]"
 			"P", // Preposition
-			"R[CDFIKNPRSX][123?][ADGNV?][DPS?][FMN?][AP?]", // Pronoun
-			"N[ADGNV?][DPS?][FMN?][COPS?]", // Noun
+			"R[CDFIKPRSX?][123?][ADGNV?][PS?][FMN?]", // Pronoun "R[CDFIKNPRSX?][123?][ADGNV?][DPS?][FMN?][AP?]"
+			"N[ADGNV?][PS?][FMN?][COS?]", // Noun "N[ADGNV?][DPS?][FMN?][COPS?]"
 			"X[FLNOP]", // Indeclinable
-			"V[AFILPRT?][AMPU?][IMNOPS][123?][DPS?][ADGNV?][FMN?]", // Verb
+			"V[AFILPR?][AMPU?][IMNOPS][123][PS]", "V[AFILPRT?][AMPU?][IMNOPS]?[DPS?][ADGNV?][FMN?]", // Verb "V[AFILPRT?][AMPU?][IMNOPS][123?][DPS?][ADGNV?][FMN?]"
 	};
 
 	protected static List<String> computePatterns() {
@@ -116,7 +116,7 @@ public void testLogosConversion() {
 		Set<String> logosExpansions = new HashSet<>(), unusedExpansions = new HashSet<>();
 		for (String logosPattern : LOGOS_PATTERNS) {
 			for (String expanded : expandPattern(logosPattern)) {
-				if (!logosPattern.contains("?")) {
+				if (!expanded.contains("?")) {
 					unusedExpansions.add(expanded);
 				} else {
 					expanded = expanded.replaceAll("\\?+$", "");
@@ -146,6 +146,8 @@ private void testLogosWord(String rmac, String logos) {
 	public void testLogosRegressions() {
 		testLogosWord("N-GSM-P", "NGSM:XP");
 		testLogosWord("V-XXM-2P", "V??M2P");
+		testLogosWord("Q", "R");
+		testLogosWord("N-NPN-C", "NNPNC");
 	}
 
 }