Fix singularizer issues and add test cases

ballerina-platform · Mar 1, 2024 · 65c5f05 · 65c5f05
1 parent 3e8e999
commit 65c5f05
Show file tree

Hide file tree

Showing 3 changed files with 73 additions and 34 deletions.
diff --git a/persist-cli/src/main/java/io/ballerina/persist/inflector/Rules.java b/persist-cli/src/main/java/io/ballerina/persist/inflector/Rules.java
@@ -85,9 +85,13 @@ private Rules() {}
             {"pickaxe", "pickaxes"},
             {"passerby", "passersby"},
             {"canvas", "canvases"},
-            {"status", "statuses"}
+            {"mango", "mangoes"}
     });
 
+    static final String[][] SINGULAR_IRREGULAR_RULES = {
+            {"have", "have"}
+    };
+
     // Pluralization rules.
     static final String[][] PLURALIZATION_RULES = (new String[][]{
             {"[^\u0000-\u007F]$", "$0"},
@@ -120,33 +124,32 @@ private Rules() {}
     });
 
     static final String[][] SINGULARIZATION_RULES = (new String[][]{
-            {"(s|si|ni)([aeiouy])s$", "$1$2"},
-            {"([^aeiou])ese$", "$1"},
-            {"(ax|test)es$", "$1is"},
-            {"(alias|[^aou]us|t[lm]as|gas|ris)es$", "$1"},
-            {"(e[mn]u)s$", "$1"},
-            {"([^l]ias|[aeiou]las|[ejzr]as|[iu]am)s$", "$1"},
-            {"(alumn|syllab|vir|radi|nucle|fung|cact|stimul|termin|bacill|foc|uter|loc|strat)i$", "$1us"},
-            {"(alumn|alg|vertebr)ae$", "$1a"},
-            {"(seraph|cherub)im$", "$1"},
-            {"(her|at|gr)oes$", "$1o"},
-            {"(agend|addend|millenni|dat|extrem|bacteri|desiderat|strat|candelabr|errat|ov|" +
-                    "symposi|curricul|automat|quor)a$", "$1um"},
-            {"(apheli|hyperbat|periheli|asyndet|noumen|phenomen|criteri|organ|prolegomen|" +
-                    "hedr|automat)a$", "$1on"},
-            {"ses$", "sis"},
-            {"(?:(kni|wi|li)fe|(ar|l|ea|eo|oa|hoo)f)ves$", "$1$2"},
-            {"([^aeiouy]|qu)ies$", "$1y"},
-            {"([^ch][ieo][ln])eys$", "$1ey"},
-            {"(x|ch|ss|sh|zz)es$", "$1"},
-            {"(matr|cod|mur|sil|vert|ind|append)ices$", "$1ix"},
+            {"(ss)$", "$1"},
+            {"(wi|kni|(?:after|half|high|low|mid|non|night|[^\\w]|^)li)ves$", "$1fe"},
+            {"(ar|(?:wo|[ae])l|[eo][ao])ves$", "$1f"},
+            {"ies$", "y"},
+            {"(dg|ss|ois|lk|ok|wn|mb|th|ch|ec|oal|is|ck|ix|sser|ts|wb)ies$", "$1ie"},
+            {"\b(l|(?:neck|cross|hog|aun)?t|coll|faer|food|gen|goon|group|hipp|junk|vegg|(?:pork)?p|charl|calor|cut" +
+                    ")ies$", "$1ie"},
+            {"\b(mon|smil)ies$", "$1ey"},
             {"\b((?:tit)?m|l)ice$", "$1ouse"},
-            {"(pe)ople$", "$1rson"},
+            {"(seraph|cherub)im$", "$1"},
+            {"(x|ch|ss|sh|zz|tto|go|cho|alias|[^aou]us|t[lm]as|gas|(?:her|at|gr)o|[aeiou]ris)(?:es)?$", "$1"},
+            {"(analy|diagno|parenthe|progno|synop|the|empha|cri|ne)(?:sis|ses)$", "$1sis"},
+            {"(movie|twelve|abuse|e[mn]u)s$", "$1"},
+            {"(test)(?:is|es)$", "$1is"},
+            {"(alumn|syllab|vir|radi|nucle|fung|cact|stimul|termin|bacill|foc|uter|loc|strat)(?:us|i)$", "$1us"},
+            {"(agend|addend|millenni|dat|extrem|bacteri|desiderat|strat|candelabr|errat|ov|symposi|curricul|" +
+                    "quor)a$", "$1um"},
+            {"(apheli|hyperbat|periheli|asyndet|noumen|phenomen|criteri|organ|prolegomen|hedr|automat)a$", "$1on"},
+            {"(alumn|alg|vertebr)ae$", "$1a"},
+            {"(cod|mur|sil|vert|ind)ices$", "$1ex"},
+            {"(matr|append)ices$", "$1ix"},
+            {"(pe)(rson|ople)$", "$1rson"},
             {"(child)ren$", "$1"},
-            {"eaux$", "$0"},
+            {"(eau)x?$", "$1"},
             {"men$", "man"},
-            {"you", "thou"},
-            {"s$", ""}
+            {"s$", ""},
     });
 
 
@@ -171,6 +174,7 @@ private Rules() {}
             "^[a-z]*measles$",
             "^[a-z]*o[iu]s$", // "carnivorous"
             "^[a-z]*pox$", // "chickpox", "smallpox"
-            "^[a-z]*sheep$"
+            "^[a-z]*sheep$",
+            "species"
     );
 }
diff --git a/persist-cli/src/main/java/io/ballerina/persist/inflector/Singularizer.java b/persist-cli/src/main/java/io/ballerina/persist/inflector/Singularizer.java
@@ -30,6 +30,14 @@ public static String singularize(String word) {
                 return word;
             }
         }
+        for (String[] irregularRule: Rules.SINGULAR_IRREGULAR_RULES) {
+            if (irregularRule[0].equals(word)) {
+                return word;
+            }
+            if (irregularRule[1].equals(word)) {
+                return irregularRule[0];
+            }
+        }
         for (String[] irregularRule: Rules.IRREGULAR_RULES) {
             if (irregularRule[0].equals(word)) {
                 return word;

diff --git a/persist-cli/src/test/java/io/ballerina/persist/UnitTest.java b/persist-cli/src/test/java/io/ballerina/persist/UnitTest.java
@@ -31,7 +31,7 @@
  * A unit test class for singular to plural functions.
  */
 public class UnitTest {
-    List<String> singularWords = Arrays.asList(
+    List<String> singularInput = Arrays.asList(
             "boy", "girl", "bird", "cod", "commerce", "quiz", "lemma", "dingo", "echo", "yes",
             "tornado", "have", "dingo", "bus", "child", "mom", "dad", "bottle", "sticker", "moss",
             "wolf", "wife", "life", "leaf", "woman", "mouse", "goose", "baby", "toy", "kidney",
@@ -43,7 +43,7 @@ public class UnitTest {
             "medium", "bacterium", "kangaroo", "cherry", "sky", "monkey", "berry", "video", "studio", "mango",
             "tornado", "tuxedo", "volcano", "house", "sister", "item", "thing", "computer", "flower", "roof", "has"
     );
-    List<String> pluralWords = Arrays.asList(
+    List<String> pluralOutput = Arrays.asList(
             "boys", "girls", "birds", "cod", "commerce", "quizzes", "lemmata", "dingoes", "echoes", "yeses",
             "tornadoes", "have", "dingoes", "buses", "children", "moms", "dads", "bottles", "stickers", "mosses",
             "wolves", "wives", "lives", "leaves", "women", "mouses", "geese", "babies", "toys", "kidneys",
@@ -53,20 +53,47 @@ public class UnitTest {
             "cactuses", "focuses", "series", "species", "mouses", "feet", "nannies", "studies", "foxes", "pouches",
             "brushes", "quizzes", "roofs", "trucks", "bugs", "pens", "books", "vegetables", "chairs", "bacteria",
             "mediums", "bacteria", "kangaroos", "cherries", "skies", "monkeys", "berries", "videos", "studios",
-            "mangos", "tornadoes", "tuxedos", "volcanoes", "houses", "sisters", "items", "things", "computers",
+            "mangoes", "tornadoes", "tuxedos", "volcanoes", "houses", "sisters", "items", "things", "computers",
             "flowers", "roofs", "have"
     );
+
+    List<String> pluralInput = Arrays.asList(
+            "boys", "girls", "birds", "cod", "commerce", "quizzes", "lemmata", "dingoes", "echoes", "yeses",
+            "tornadoes", "have", "dingoes", "buses", "children", "moms", "dads", "bottles", "stickers", "mosses",
+            "wolves", "wives", "lives", "leaves", "women", "mouses", "geese", "babies", "toys", "kidneys",
+            "potatoes", "memos", "stereos", "sheep", "deer", "series", "species", "windows", "stickers", "desks",
+            "pencils", "cups", "milks", "choices", "boxes", "thieves", "armies", "women", "friends", "daisies",
+            "bosses", "marshes", "classes", "lunches", "beliefs", "chefs", "cities", "rays", "photos", "pianos",
+            "cactuses", "focuses", "series", "species", "mouses", "feet", "nannies", "studies", "foxes", "pouches",
+            "brushes", "quizzes", "roofs", "trucks", "bugs", "pens", "books", "vegetables", "chairs", "bacteria",
+            "mediums", "bacteria", "kangaroos", "cherries", "skies", "monkeys", "berries", "videos", "studios",
+            "mangoes", "tornadoes", "tuxedos", "volcanoes", "houses", "sisters", "items", "things", "computers",
+            "flowers", "roofs", "have"
+    );
+
+    List<String> singularOutput = Arrays.asList(
+            "boy", "girl", "bird", "cod", "commerce", "quiz", "lemma", "dingo", "echo", "yes",
+            "tornado", "have", "dingo", "bus", "child", "mom", "dad", "bottle", "sticker", "moss",
+            "wolf", "wife", "life", "leaf", "woman", "mouse", "goose", "baby", "toy", "kidney",
+            "potato", "memo", "stereo", "sheep", "deer", "series", "species", "window", "sticker", "desk",
+            "pencil", "cup", "milk", "choice", "box", "thief", "army", "woman", "friend", "daisy",
+            "boss", "marsh", "class", "lunch", "belief", "chef", "city", "ray", "photo", "piano",
+            "cactus", "focus", "series", "species", "mouse", "foot", "nanny", "study", "fox", "pouch",
+            "brush", "quiz", "roof", "truck", "bug", "pen", "book", "vegetable", "chair", "bacterium",
+            "medium", "bacterium", "kangaroo", "cherry", "sky", "monkey", "berry", "video", "studio", "mango",
+            "tornado", "tuxedo", "volcano", "house", "sister", "item", "thing", "computer", "flower", "roof", "have"
+    );
     @Test
     public void testSingularToPlural() {
         long startTime = System.nanoTime();
-        List<String> outputs = singularWords.stream().map(Pluralizer::pluralize).toList();
-        assertResults(startTime, outputs, pluralWords);
+        List<String> outputs = singularInput.stream().map(Pluralizer::pluralize).toList();
+        assertResults(startTime, outputs, pluralOutput);
     }
-    @Test(enabled = false)
+    @Test(enabled = true)
     public void testPluralToSingular() {
         long startTime = System.nanoTime();
-        List<String> outputs = pluralWords.stream().map(Singularizer::singularize).toList();
-        assertResults(startTime, outputs, singularWords);
+        List<String> outputs = pluralInput.stream().map(Singularizer::singularize).toList();
+        assertResults(startTime, outputs, singularOutput);
     }
     private void assertResults(long startTime, List<String> outputs, List<String> expectedOutputs) {
         PrintStream print = System.out;