fix: Don't truncate speech containing , (#26)

Author: eps1lon

.changeset/quick-goats-sniff.md

@@ -0,0 +1,8 @@
+---
+"screen-reader-testing-library": patch
+---
+
+Speech containing a comma is no longer truncated
+
+For example, `Choose time, selected time is ...` previously resulted in `["Choose tim"]`.
+The previous speech parser only used a regular expression which cannot handle context-sensitive languages.

src/__tests__/speechLogParser.js

@@ -200,8 +200,7 @@ it("ignores commands", () => {
 		    "Search…",
 		  ],
 		  Array [
-		    "Too much output to announc",
-		    "",
+		    "Too much output to announce, navigate to rows manually to readteardown JestAssertionError: expect(received).toEqual(expected) // deep equality - Expected - 2 + Received + 2 Array [ Array [ + \\"angChangeCommand (\\\\",
 		  ],
 		  Array [
 		    "nvda.ini (Working Tree) - nvda-snapshot-testing - Visual Studio Code",
@@ -217,7 +216,7 @@ it("ignores commands", () => {
 		    "complementary landmark",
 		  ],
 		  Array [
-		    "Terminal ",
+		    "Terminal 2, PowerShell Integrated Console",
 		    "edit",
 		    "blank",
 		  ],
@@ -234,3 +233,20 @@ it("ignores commands", () => {
 		]
 	`);
 });
+
+it("is context sensitive", () => {
+	expect(
+		extractSpeechLines(
+			"Speaking [LangChangeCommand ('en'), 'clickable', 'main landmark', 'button', 'Choose time, selected time is 12:00 AM']"
+		)
+	).toMatchInlineSnapshot(`
+		Array [
+		  Array [
+		    "clickable",
+		    "main landmark",
+		    "button",
+		    "Choose time, selected time is 12:00 AM",
+		  ],
+		]
+	`);
+});

src/logParser.js

@@ -10,27 +10,44 @@ function extractSpeechLines(nvdaLog) {
 		})
 		.map((line) => {
 			// In: "Speaking ['speech1', 'speech2', 'speech 3']"
-			// Out: "['speech1', 'speech2', 'speech 3']"
+			// Out: "'speech1', 'speech2', 'speech 3'"
 			const listText = line.trim().replace(/^Speaking \[([^\]]+)\]/, "$1");
 
-			// In: "['speech1', 'speech2', 'speech 3']"
-			// Out: the corresponding array structure in JS with added quotes
-			return listText
-				.split(",")
-				.map((rawSpeech) => {
-					const speech = rawSpeech.trim();
-					if (speech.startsWith("'")) {
-						return speech.slice(1, -1);
+			// light-weight parser for
+			// In: "'speech1', 'speech2, other', 'speech 3'"
+			// Put: ['speech1', 'speech2, other', 'speech 3']
+			/**
+			 * @type {string[]}
+			 */
+			const spoken = [];
+			/**
+			 * @type {'type' | 'command' | 'speech'}
+			 */
+			let currentlyParsing = "type";
+			let speech = "";
+			for (const token of listText) {
+				if (currentlyParsing === "type") {
+					if (token === "'") {
+						currentlyParsing = "speech";
+					} else if (!/(\s|,)/.test(token)) {
+						currentlyParsing = "command";
 					}
-					// ignore commands for now
-					return null;
-				})
-				.filter(
-					/**
-					 * @returns {speech is string}
-					 */
-					(speech) => speech !== null
-				);
+				} else if (currentlyParsing === "command") {
+					if (token === ",") {
+						currentlyParsing = "type";
+					}
+				} else if (currentlyParsing === "speech") {
+					if (token === "'") {
+						spoken.push(speech);
+						speech = "";
+						currentlyParsing = "type";
+					} else {
+						speech += token;
+					}
+				}
+			}
+
+			return spoken;
 		});
 }