feat: update title capitalization rules

packages/content-common/src/index.spec.ts

@@ -130,6 +130,90 @@ describe('content-common', () => {
         expect(applyApTitleCase(result)).toEqual(expected);
       });
     });
+
+    it('should capitalize "a" and "the" after sentence-ending punctuation', () => {
+      const testCases = [
+        {
+          result: 'Nazi Persecution Scattered My Family. a Lost Archive Brought Us Together',
+          expected: 'Nazi Persecution Scattered My Family. A Lost Archive Brought Us Together',
+        },
+        {
+          result: 'This is the end! the beginning starts now',
+          expected: 'This Is the End! The Beginning Starts Now',
+        },
+        {
+          result: 'What happened? a miracle occurred',
+          expected: 'What Happened? A Miracle Occurred',
+        },
+        {
+          result: 'She said "Hello." the crowd cheered',
+          expected: 'She Said "Hello." The Crowd Cheered',
+        },
+      ];
+      testCases.forEach(({ result, expected }) => {
+        expect(applyApTitleCase(result)).toEqual(expected);
+      });
+    });
+
+    it('should always format iPhone correctly', () => {
+      const testCases = [
+        {
+          result: 'the new Iphone is amazing',
+          expected: 'The New iPhone Is Amazing',
+        },
+        {
+          result: 'IPHONE users love their devices',
+          expected: 'iPhone Users Love Their Devices',
+        },
+        {
+          result: 'my iphone broke yesterday',
+          expected: 'My iPhone Broke Yesterday',
+        },
+      ];
+      testCases.forEach(({ result, expected }) => {
+        expect(applyApTitleCase(result)).toEqual(expected);
+      });
+    });
+
+    it('should always lowercase "vs."', () => {
+      const testCases = [
+        {
+          result: 'Apple Vs. Samsung: the battle continues',
+          expected: 'Apple vs. Samsung: The Battle Continues',
+        },
+        {
+          result: 'Batman VS. Superman was a movie',
+          expected: 'Batman vs. Superman Was a Movie',
+        },
+        {
+          result: 'Good vs Evil: a timeless struggle',
+          expected: 'Good vs. Evil: A Timeless Struggle',
+        },
+      ];
+      testCases.forEach(({ result, expected }) => {
+        expect(applyApTitleCase(result)).toEqual(expected);
+      });
+    });
+
+    it('should not capitalize "as" in title case', () => {
+      const testCases = [
+        {
+          result: 'Working As a Team Is Important',
+          expected: 'Working as a Team Is Important',
+        },
+        {
+          result: 'As The Sun Sets',
+          expected: 'As the Sun Sets',
+        },
+        {
+          result: 'She Sees It As An Opportunity',
+          expected: 'She Sees It as an Opportunity',
+        },
+      ];
+      testCases.forEach(({ result, expected }) => {
+        expect(applyApTitleCase(result)).toEqual(expected);
+      });
+    });
   });
   describe('lowercaseAfterApostrophe', () => {
     it('lowercase letter after apostrophe & return new string', () => {

packages/content-common/src/index.ts

@@ -56,7 +56,10 @@ export const sanitizeText = (input: string, maxLength: number): string => {
 export const lowercaseAfterApostrophe = (input: string): string => {
   // Match either an ASCII or curly apostrophe followed by a letter, after a word character.
   const regex = /(?<=\w)(['\u2018\u2019])(\w)/g;
-  return input.replace(regex, (_, apostrophe, letter) => `${apostrophe}${letter.toLowerCase()}`);
+  return input.replace(
+    regex,
+    (_, apostrophe, letter) => `${apostrophe}${letter.toLowerCase()}`,
+  );
 };
 
 /**
@@ -89,22 +92,65 @@ export const applyApTitleCase = (value: string): string => {
 
   const result = allWords
     .map((word, index, all) => {
+      // Check if the previous non-empty element is a sentence-ending punctuation
+      const isAfterSentenceEnd =
+        index > 0 &&
+        (() => {
+          // Look for the previous non-empty element
+          for (let i = index - 1; i >= 0; i--) {
+            const prev = all[i].trim();
+            if (prev) {
+              // Check if it ends with sentence-ending punctuation
+              return (
+                /[.!?]$/.test(prev) ||
+                // Or if it's a closing quote after sentence-ending punctuation
+                (i > 0 &&
+                  (prev === '"' ||
+                    prev === '\u201D' ||
+                    prev === "'" ||
+                    prev === '\u2019') &&
+                  /[.!?]$/.test(all[i - 1].trim()))
+              );
+            }
+          }
+          return false;
+        })();
+
       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 “
+          allWords[index - 1] === '\u2018' || // Opening single quote '
+          allWords[index - 1] === '\u201C'); // Opening double quote "
+
+      // Special case handling for specific words
+      const lowerWord = word.toLowerCase();
+
+      // Handle iPhone
+      if (lowerWord === 'iphone') {
+        return 'iPhone';
+      }
+
+      // Handle vs.
+      if (lowerWord === 'vs.' || lowerWord === 'vs') {
+        return 'vs.';
+      }
 
+      // Check if we should capitalize this word
       if (
         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
+        isAfterSentenceEnd || // capitalize after sentence-ending punctuation
+        (!stop.includes(lowerWord) && lowerWord !== 'as') // not a stop word and not 'as'
       ) {
+        // Special handling for 'a' and 'the' after sentence-ending punctuation
+        if (isAfterSentenceEnd && (lowerWord === 'a' || lowerWord === 'the')) {
+          return capitalize(word);
+        }
         return capitalize(word);
       }