twl-generator 1.2.8 → 1.2.9

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "twl-generator",
-  "version": "1.2.8",
+  "version": "1.2.9",
   "description": "Generate term-to-article lists from unfoldingWord en_tw archive for Bible books. Works in both Node.js (CLI) and React.js (browser) environments.",
   "main": "src/index.js",
   "bin": {
@@ -56,4 +56,4 @@
       "optional": true
     }
   }
-}
+}

package/src/utils/twl-matcher.js

@@ -5,59 +5,61 @@ function generateVariants(term) {
   const variants = new Set([term]);
 
   const nouns = ['doe', 'deer', 'father', 'Father'];
+  const do_not_pluralize = ['doe'];
+  const do_not_depluralize = [];
 
   // Handle pluralization - simple 's' removal (but not for words ending in 'ss')
-  if (term.endsWith('s') && term.length > 2 && !term.endsWith('ss') && !term.endsWith('es')) {
+  if (term.endsWith('s') && term.length > 2 && !term.endsWith('ss') && !term.endsWith('es') && !do_not_depluralize.includes(term)) {
     variants.add(term.slice(0, -1)); // dogs -> dog (but not does -> doe)
-  } else {
+  } else if (!do_not_pluralize.includes(term)) {
     variants.add(term + 's'); // dog -> dogs
   }
 
   // Handle 'es' endings - but only for legitimate plural patterns
-  if (term.endsWith('es') && term.length > 4) {
+  if (term.endsWith('es') && term.length > 4 && !do_not_depluralize.includes(term)) {
     const base = term.slice(0, -2);
     // Only if the base word would naturally take 'es' plural
     if (/[sxz]$|[cs]h$/.test(base)) {
       variants.add(base); // horses -> horse, churches -> church
     }
-  } else if (term.endsWith('e')) {
+  } else if (term.endsWith('e') && !do_not_pluralize.includes(term)) {
     variants.add(term + 's'); // horse -> horses
-  } else if (/[sxz]$|[cs]h$/.test(term)) {
+  } else if (/[sxz]$|[cs]h$/.test(term) && !do_not_pluralize.includes(term)) {
     variants.add(term + 'es'); // church -> churches
   }
 
   // Handle 'ies' endings for words ending in 'y'
-  if (term.endsWith('ies') && term.length > 4) {
+  if (term.endsWith('ies') && term.length > 4 && !do_not_depluralize.includes(term)) {
     variants.add(term.slice(0, -3) + 'y'); // cities -> city
-  } else if (term.endsWith('y') && term.length > 2 && !/[aeiou]y$/.test(term)) {
+  } else if (term.endsWith('y') && term.length > 2 && !/[aeiou]y$/.test(term) && !do_not_pluralize.includes(term)) {
     variants.add(term.slice(0, -1) + 'ies'); // city -> cities
   }
 
-  // // Handle possessive forms
+  // // Handle possessive forms -- // Commented out since we use curly quotes
   // variants.add(term + "'s");
   // variants.add(term + "'");
   // if (term.endsWith('s')) {
   //   variants.add(term + "'");
   // }
 
-  // Handle -ed forms - but only for legitimate verb patterns
-  if (term.endsWith('ed') && term.length > 4) {
-    const base = term.slice(0, -2);
-    // Only create base form if it looks like a legitimate verb stem
-    if (base.length > 2) {
-      variants.add(base); // walked -> walk
+  if (!nouns.includes(term)) {
+    // Handle -ed forms - but only for legitimate verb patterns
+    if (term.endsWith('ed') && term.length > 4) {
+      const base = term.slice(0, -2);
+      // Only create base form if it looks like a legitimate verb stem
+      if (base.length > 2) {
+        variants.add(base); // walked -> walk
+      }
     }
-  }
 
-  // Handle -ing forms
-  if (term.endsWith('ing') && term.length > 5) {
-    const base = term.slice(0, -3);
-    if (base.length > 2) {
-      variants.add(base); // walking -> walk
+    // Handle -ing forms
+    if (term.endsWith('ing') && term.length > 5) {
+      const base = term.slice(0, -3);
+      if (base.length > 2) {
+        variants.add(base); // walking -> walk
+      }
     }
-  }
 
-  if (!nouns.includes(term)) {
     // Double consonant handling for -ed/-ing
     if (/[bcdfghjklmnpqrstvwxyz][aeiou][bcdfghjklmnpqrstvwxyz]$/.test(term)) {
       variants.add(term + term.slice(-1) + 'ed'); // stop -> stopped