MC-1583:"Fix Title” for English titles not capitalizing first word after single quote or colon (#1223)

* MC-1583:"Fix Title” for English headlines not capitalizing first word after single quote or colon

Author: katerinachinnappan

src/_shared/utils/applyApTitleCase.test.ts

@@ -1,4 +1,4 @@
-import { applyApTitleCase } from './applyApTitleCase';
+import { applyApTitleCase, lowercaseAfterApostrophe } from './applyApTitleCase';
 
 // examples taken from https://www.grammarly.com/blog/capitalization-in-the-titles/
 // tested at https://headlinecapitalization.com/ (AP style)
@@ -113,4 +113,39 @@ describe('applyApTitleCase', () => {
       expect(applyApTitleCase(swc.result)).toEqual(swc.expected);
     });
   });
+
+  it('should differentiate between strings in quotes and apostrophe', () => {
+    const sentencesWithContractions = [
+      {
+        result: "Here's what you haven't noticed 'foo bar' foo'S",
+        expected: "Here's What You Haven't Noticed 'Foo Bar' Foo's",
+      },
+    ];
+    sentencesWithContractions.forEach((swc) => {
+      expect(applyApTitleCase(swc.result)).toEqual(swc.expected);
+    });
+  });
+
+  it('should capitalize after a colon (:)', () => {
+    const sentencesWithContractions = [
+      {
+        result: "Here's what you haven't noticed 'foo bar' foo'S: foo Bar",
+        expected: "Here's What You Haven't Noticed 'Foo Bar' Foo'S: Foo Bar",
+      },
+    ];
+    sentencesWithContractions.forEach((swc) => {
+      expect(applyApTitleCase(swc.result)).toEqual(swc.expected);
+    });
+  });
+});
+
+describe('lowercaseAfterApostrophe', () => {
+  it('lowercase letter after apostrophe & return new string', () => {
+    const result = lowercaseAfterApostrophe("foo'S");
+    expect(result).toEqual("foo's");
+  });
+  it('lowercase letter after apostrophe, ignore string in quotes, & return new string', () => {
+    const result = lowercaseAfterApostrophe("'Foo' foo'S DaY's");
+    expect(result).toEqual("'Foo' foo's DaY's");
+  });
 });

src/_shared/utils/applyApTitleCase.ts

@@ -1,10 +1,31 @@
 export const STOP_WORDS =
-  'a an and at but by for in nor of on or so the to up yet';
+  'a an and at but by for in nor of on or the to up yet';
 
-export const SEPARATORS = /(\s+|[-‑–—,:;!?()“”"])/;
+// Matches a colon (:) and 0+ white spaces following after
+// Matches 1+ white spaces
+// Matches special chars (i.e. hyphens, quotes, etc)
+export const SEPARATORS = /(:\s*|\s+|[-‑–—,:;!?()“”'‘"])/; // Include curly quotes as separators
 
 export const stop = STOP_WORDS.split(' ');
 
+/**
+ * Format a string: Capture the letter after an apostrophe at the end of a
+ * sentence (without requiring a space) or with a white space following the letter.
+ * Lowercase the captured letter & return the formatted string.
+ * @param input
+ * @returns {string}
+ */
+export const lowercaseAfterApostrophe = (input: string): string => {
+  // matches a char (num or letter) right after an apostrophe,
+  // only if the apostrophe is preceded by a char & is followed
+  // by a space or end of the str.
+  const regex = /(?<=\w)'(\w)(?=\s|$)/g;
+
+  return input.replace(regex, (match, p1) => {
+    return `'${p1.toLowerCase()}`; // Replace with the apostrophe and the lowercase letter
+  });
+};
+
 /**
  * Capitalize first character for string
  *
@@ -30,19 +51,36 @@ export const applyApTitleCase = (value: string): string => {
   if (!value) {
     return '';
   }
-  // split by separators, check if word is first or last
-  // or not blacklisted, then capitalize
-  return value
-    .split(SEPARATORS)
+
+  // Split and filter empty strings
+  // Boolean here acts as a callback, evaluates each word:
+  // If it's a non-empty string, keep the word in the array;
+  // If it's an empty string (or falsy), remove from array.
+  const allWords = value.split(SEPARATORS).filter(Boolean); // Split and filter empty strings
+
+  const result = allWords
     .map((word, index, all) => {
+      const isAfterColon = index > 0 && all[index - 1].trim() === ':';
+
+      const isAfterQuote =
+        index > 0 &&
+        (allWords[index - 1] === "'" ||
+          allWords[index - 1] === '"' ||
+          allWords[index - 1] === '\u2018' || // Opening single quote ’
+          allWords[index - 1] === '\u201C'); // Opening double quote “
+
       if (
-        index === 0 ||
-        index === all.length - 1 ||
-        !stop.includes(word.toLowerCase())
+        index === 0 || // first word
+        index === all.length - 1 || // last word
+        isAfterColon || // capitalize the first word after a colon
+        isAfterQuote || // capitalize the first word after a quote
+        !stop.includes(word.toLowerCase()) // not a stop word
       ) {
         return capitalize(word);
       }
+
       return word.toLowerCase();
     })
-    .join('');
+    .join(''); // join without additional spaces
+  return lowercaseAfterApostrophe(result);
 };