Support strings of all lowercase

Author: sirlancelot

lib/index.test.ts

@@ -16,6 +16,12 @@ for (const [language, locale, corpus] of [
 			"Dr. John Smith, Jr. gave a lecture.",
 			"As of 2.14.2023, the average price of apples is $1.31 per pound!",
 			"There's pool, pinball, Ms. Pac Man, tvs for Football (or playoff baseball!!!!) and they are kid friendly on one half of the bar.",
+			// all lowercase
+			"i just spoke with michelle at cinergy.",
+			"she said she's hearing a rumor that lonestar may be doing more work at the end of the month.",
+			"kevin, could you please call jeff or gary to see if the rumor is true.",
+			"i'd like to have a \"heads up\" if possible.",
+
 		],
 	],
 	[

lib/index.ts

@@ -17,17 +17,19 @@ const makeSegmenter = cached((locale: string) => ({
  * Split a string into sentences, respecting common abbreviations.
  */
 export function* splitBySentence(
-	input: string,
+	rawInput: string,
 	locale: Intl.LocalesArgument = "en"
 ): Generator<Intl.SegmentData> {
-	if (!input || typeof input !== "string")
+	if (!rawInput || typeof rawInput !== "string")
 		throw new TypeError("input must be a string")
 
 	const { abbreviations, segmenter } = makeSegmenter(locale.toString())
 	const rLastWord = /(?<=\s|^)\S+(?=\s+$)/
+	const input = rawInput.replaceAll(/(?<=\.\s+)\S/g, (char) =>
+		char.toLocaleUpperCase()
+	)
 
-	let continuationIndex: number | undefined
-	let continuation = ""
+	let left = 0
 	for (const { segment, index } of segmenter.segment(input)) {
 		const match = segment.match(rLastWord)
 
@@ -37,20 +39,17 @@ export function* splitBySentence(
 			(abbreviations.has(match[0].toLocaleLowerCase(locale)) ||
 				// 2. A closing parenthesis without a period.
 				match[0].endsWith(")"))
-		) {
-			continuationIndex = continuationIndex ?? index
-			continuation += segment
+		)
 			continue
-		}
 
+		const right = index + segment.length
 		yield {
-			segment: continuation + segment,
-			index: continuationIndex ?? index,
-			input,
+			segment: rawInput.slice(left, right),
+			index: left,
+			input: rawInput,
 		}
 
-		continuation = ""
-		continuationIndex = undefined
+		left = right
 	}
 }