HNT-862-add-new-capitalization-rules

src/_shared/utils/applyApTitleCase.test.ts

@@ -137,6 +137,42 @@ describe('applyApTitleCase', () => {
       expect(applyApTitleCase(swc.result)).toEqual(swc.expected);
     });
   });
+
+  it('should handle new AP style stop words correctly', () => {
+    const testCases = [
+      {
+        result: 'The Dog Jumped Up And Over The Fence',
+        expected: 'The Dog Jumped Up and Over the Fence',
+      },
+      {
+        result: 'Work As A Team',
+        expected: 'Work as a Team',
+      },
+      {
+        result: 'If You Can Dream It',
+        expected: 'If You Can Dream It',
+      },
+      {
+        result: 'Turn Off The Lights',
+        expected: 'Turn off the Lights',
+      },
+      {
+        result: 'Going Out Tonight',
+        expected: 'Going out Tonight',
+      },
+      {
+        result: 'So What Do You Think',
+        expected: 'So What Do You Think',
+      },
+      {
+        result: 'Come If You Can So We Can Talk',
+        expected: 'Come if You Can so We Can Talk',
+      },
+    ];
+    testCases.forEach(({ result, expected }) => {
+      expect(applyApTitleCase(result)).toEqual(expected);
+    });
+  });
   it('should correctly format titles with curly apostrophes', () => {
     const testCases = [
       {
@@ -148,6 +184,42 @@ describe('applyApTitleCase', () => {
       expect(applyApTitleCase(result)).toEqual(expected);
     });
   });
+
+  it('should keep iPhone and similar Apple products with lowercase i', () => {
+    const testCases = [
+      {
+        result: 'The New IPhone Is Here',
+        expected: 'The New iPhone Is Here',
+      },
+      {
+        result: 'IPad Pro Vs IPad Air',
+        expected: 'iPad Pro vs iPad Air',
+      },
+      {
+        result: 'Using ICloud With Your IPod',
+        expected: 'Using iCloud With Your iPod',
+      },
+      {
+        result: 'IMac and MacBook Pro Comparison',
+        expected: 'iMac and MacBook Pro Comparison',
+      },
+      {
+        result: 'ITunes Is Now Apple Music',
+        expected: 'iTunes Is Now Apple Music',
+      },
+      {
+        result: 'Send IMessage From Your IPhone',
+        expected: 'Send iMessage From Your iPhone',
+      },
+      {
+        result: 'IBooks: The Complete Guide',
+        expected: 'iBooks: The Complete Guide',
+      },
+    ];
+    testCases.forEach(({ result, expected }) => {
+      expect(applyApTitleCase(result)).toEqual(expected);
+    });
+  });
 });
 
 describe('lowercaseAfterApostrophe', () => {

src/_shared/utils/applyApTitleCase.ts

@@ -1,5 +1,5 @@
 export const STOP_WORDS =
-  'a an and at but by for in nor of on or the to up yet';
+  'a an and as at but by for if in nor of off on or out so the to vs yet';
 
 // Matches a colon (:) and 0+ white spaces following after
 // Matches 1+ white spaces
@@ -11,12 +11,14 @@ export const stop = STOP_WORDS.split(' ');
 /**
  * Format a string: Match the letter after an apostrophe & capture the apostrophe and matched char.
  * Lowercase the captured letter & return the formatted string.
+ * Exception: O' prefix (like O'Hearn) should have the letter after apostrophe capitalized.
  * @param input
  * @returns {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;
+  // Negative lookbehind to exclude O' prefix
+  const regex = /(?<!^O)(?<![\s]O)(?<=\w)(['\u2018\u2019])(\w)/g;
   return input.replace(
     regex,
     (_, apostrophe, letter) => `${apostrophe}${letter.toLowerCase()}`,
@@ -63,8 +65,8 @@ export const applyApTitleCase = (value: string): string => {
         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 "
 
       if (
         index === 0 || // first word
@@ -79,5 +81,16 @@ export const applyApTitleCase = (value: string): string => {
       return word.toLowerCase();
     })
     .join(''); // join without additional spaces
-  return lowercaseAfterApostrophe(result);
+
+  // Apply special formatting rules
+  let formattedResult = lowercaseAfterApostrophe(result);
+
+  // Handle special cases like iPhone, iPad, iPod, etc.
+  // This regex looks for word boundaries followed by capital I and then Phone/Pad/Pod/etc.
+  formattedResult = formattedResult.replace(
+    /\bI(Phone|Pad|Pod|Mac|Cloud|Tunes|Books|Message)/g,
+    'i$1',
+  );
+
+  return formattedResult;
 };

src/curated-corpus/components/actions/RemoveSectionItemAction/RemoveSectionItemAction.test.tsx

@@ -4,9 +4,7 @@ import { render, screen } from '@testing-library/react';
 import { SnackbarProvider } from 'notistack';
 import { RemoveSectionItemAction } from './RemoveSectionItemAction';
 import { getTestApprovedItem } from '../../../helpers/approvedItem';
-import {
-  successMock,
-} from '../../../integration-test-mocks/removeSectionItem';
+import { successMock } from '../../../integration-test-mocks/removeSectionItem';
 import userEvent from '@testing-library/user-event';
 import { apolloCache } from '../../../../api/client';
 import {

temp/applyApTitleCase.js

@@ -0,0 +1,79 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.applyApTitleCase = exports.lowercaseAfterApostrophe = exports.stop = exports.SEPARATORS = exports.STOP_WORDS = void 0;
+exports.STOP_WORDS = 'a an and as at but by for if in nor of off on or out so the to yet';
+// Matches a colon (:) and 0+ white spaces following after
+// Matches 1+ white spaces
+// Matches special chars (i.e. hyphens, quotes, etc)
+exports.SEPARATORS = /(:\s*|\s+|[-‑–—,:;!?()“”'‘"])/; // Include curly quotes as separators
+exports.stop = exports.STOP_WORDS.split(' ');
+/**
+ * Format a string: Match the letter after an apostrophe & capture the apostrophe and matched char.
+ * Lowercase the captured letter & return the formatted string.
+ * Exception: O' prefix (like O'Hearn) should have the letter after apostrophe capitalized.
+ * @param input
+ * @returns {string}
+ */
+const lowercaseAfterApostrophe = (input) => {
+    // Match either an ASCII or curly apostrophe followed by a letter, after a word character.
+    // Negative lookbehind to exclude O' prefix
+    const regex = /(?<!^O)(?<![\s]O)(?<=\w)(['\u2018\u2019])(\w)/g;
+    return input.replace(regex, (_, apostrophe, letter) => `${apostrophe}${letter.toLowerCase()}`);
+};
+exports.lowercaseAfterApostrophe = lowercaseAfterApostrophe;
+/**
+ * Capitalize first character for string
+ *
+ * @param {string} value
+ * @returns {string}
+ */
+const capitalize = (value) => {
+    if (!value) {
+        return '';
+    }
+    return value.charAt(0).toUpperCase() + value.slice(1);
+};
+/**
+ * Helper to convert text to AP title case
+ * adapted from https://github.com/words/ap-style-title-case
+ * text should match https://headlinecapitalization.com/
+ *
+ * @param {string} [value]
+ * @returns {string}
+ */
+const applyApTitleCase = (value) => {
+    if (!value) {
+        return '';
+    }
+    // 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(exports.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 || // first word
+            index === all.length - 1 || // last word
+            isAfterColon || // capitalize the first word after a colon
+            isAfterQuote || // capitalize the first word after a quote
+            !exports.stop.includes(word.toLowerCase()) // not a stop word
+        ) {
+            return capitalize(word);
+        }
+        return word.toLowerCase();
+    })
+        .join(''); // join without additional spaces
+    // Apply special formatting rules
+    let formattedResult = (0, exports.lowercaseAfterApostrophe)(result);
+    // Handle special cases like iPhone, iPad, iPod, etc.
+    // This regex looks for word boundaries followed by capital I and then Phone/Pad/Pod/etc.
+    formattedResult = formattedResult.replace(/\bI(Phone|Pad|Pod|Mac|Cloud|Tunes|Books|Message)/g, 'i$1');
+    return formattedResult;
+};
+exports.applyApTitleCase = applyApTitleCase;