refactor(natives): removed sanitizeText from pi-natives, use JS impl

- Moved @oh-my-pi/pi-natives from devDependencies to dependencies in pi-utils.
- Deleted Rust sanitizeText implementation and its JS bindings/types.
- Updated sanitize benchmark to use pi-utils implementation as baseline.

bun.lock

@@ -180,13 +180,13 @@
       "name": "@oh-my-pi/pi-utils",
       "version": "15.1.2",
       "dependencies": {
+        "@oh-my-pi/pi-natives": "catalog:",
         "beautiful-mermaid": "catalog:",
         "handlebars": "catalog:",
         "winston": "catalog:",
         "winston-daily-rotate-file": "catalog:",
       },
       "devDependencies": {
-        "@oh-my-pi/pi-natives": "catalog:",
         "@types/bun": "catalog:",
       },
     },

crates/pi-natives/src/text.rs

@@ -1263,89 +1263,6 @@ pub fn extract_segments(
 	})
 }
 
-// ============================================================================
-// sanitizeText
-// ============================================================================
-
-/// Strip ANSI escape sequences, remove control characters / lone surrogates,
-/// and normalize line endings.
-#[napi]
-pub fn sanitize_text(text: JsString<'_>) -> Result<Either<JsString<'_>, Utf16String>> {
-	let original = text;
-	let text_u16 = text.into_utf16()?;
-	let data = text_u16.as_slice();
-
-	let mut did_change = false;
-	let mut out: Vec<u16> = Vec::new();
-	let mut last = 0usize;
-	let mut i = 0usize;
-	let len = data.len();
-
-	while i < len {
-		let u = data[i];
-
-		// Allow tab + newline; normalize CR by removing it.
-		if u == 0x09 || u == 0x0a {
-			i += 1;
-			continue;
-		}
-
-		let mut remove_len = if u == ESC
-			&& let Some(seq_len) = ansi_seq_len_u16(data, i)
-		{
-			seq_len
-		} else {
-			0usize
-		};
-
-		if remove_len == 0 {
-			// Drop CR to normalize line endings.
-			if u == 0x0d {
-				remove_len = 1;
-			} else if u <= 0x1f || u == 0x7f || (0x80..=0x9f).contains(&u) {
-				// C0 + DEL + C1 controls.
-				remove_len = 1;
-			} else if (0xd800..=0xdbff).contains(&u) {
-				// High surrogate: keep only if followed by a valid low surrogate.
-				if i + 1 < len {
-					let lo = data[i + 1];
-					if (0xdc00..=0xdfff).contains(&lo) {
-						i += 2;
-						continue;
-					}
-				}
-				remove_len = 1;
-			} else if (0xdc00..=0xdfff).contains(&u) {
-				// Lone low surrogate.
-				remove_len = 1;
-			}
-		}
-
-		if remove_len == 0 {
-			i += 1;
-			continue;
-		}
-
-		if !did_change {
-			did_change = true;
-			out = Vec::with_capacity(len);
-		}
-		if last != i {
-			out.extend_from_slice(&data[last..i]);
-		}
-		i += remove_len;
-		last = i;
-	}
-
-	if !did_change {
-		return Ok(Either::A(original));
-	}
-	if last < len {
-		out.extend_from_slice(&data[last..]);
-	}
-	Ok(Either::B(build_utf16_string(out)))
-}
-
 // ============================================================================
 // visibleWidth
 // ============================================================================

packages/natives/native/index.d.ts

@@ -1154,12 +1154,6 @@ export interface PtyStartOptions {
  */
 export declare function readImageFromClipboard(): Promise<ClipboardImage | undefined | null>
 
-/**
- * Strip ANSI escape sequences, remove control characters / lone surrogates,
- * and normalize line endings.
- */
-export declare function sanitizeText(text: string): string
-
 /**
  * Search content for a pattern (one-shot, compiles pattern each time).
  * For repeated searches with the same pattern, use [`grep`] with file filters.

packages/natives/native/index.js

@@ -56,7 +56,6 @@ export const matchesLegacySequence = nativeBindings.matchesLegacySequence;
 export const parseKey = nativeBindings.parseKey;
 export const parseKittySequence = nativeBindings.parseKittySequence;
 export const readImageFromClipboard = nativeBindings.readImageFromClipboard;
-export const sanitizeText = nativeBindings.sanitizeText;
 export const search = nativeBindings.search;
 export const sliceWithWidth = nativeBindings.sliceWithWidth;
 export const summarizeCode = nativeBindings.summarizeCode;

packages/natives/test/native.test.ts

@@ -15,7 +15,6 @@ import {
 	listWorkspace,
 	MacOSPowerAssertion,
 	PtySession,
-	sanitizeText,
 	summarizeCode,
 	truncateToWidth,
 	visibleWidth,
@@ -584,28 +583,11 @@ describe("pi-natives", () => {
 		});
 	});
 
-	describe("sanitizeText", () => {
-		it("should strip ANSI, remove control chars and normalize CR", () => {
-			const input = "\x1b[31mred\x1b[0m\ra\u0000b\tline\ncarriage\r\u0001\u0085";
-			expect(sanitizeText(input)).toBe("redab\tline\ncarriage");
-		});
-
-		it("should remove lone surrogates but keep valid pairs", () => {
-			expect(sanitizeText(`a\ud800b\udc00c`)).toBe("abc");
-			const validPair = "a\u{1f600}b";
-			expect(sanitizeText(validPair)).toBe(validPair);
-		});
-
-		it("should strip OSC sequences", () => {
-			const input = "\x1b]0;title\x07hello";
-			expect(sanitizeText(input)).toBe("hello");
-		});
-		describe("MacOSPowerAssertion", () => {
-			it("should create a stoppable power assertion handle", () => {
-				const assertion = MacOSPowerAssertion.start({ reason: "pi-natives test" });
-				assertion.stop();
-				assertion.stop();
-			});
+	describe("MacOSPowerAssertion", () => {
+		it("should create a stoppable power assertion handle", () => {
+			const assertion = MacOSPowerAssertion.start({ reason: "pi-natives test" });
+			assertion.stop();
+			assertion.stop();
 		});
 	});
 });

packages/tui/bench/sanitize.ts

@@ -1,4 +1,3 @@
-import { sanitizeText as nativeSanitizeText } from "@oh-my-pi/pi-natives";
 import { sanitizeText as currentSanitizeText } from "@oh-my-pi/pi-utils/sanitize-text";
 
 const STRIP_RE = new RegExp(
@@ -313,39 +312,35 @@ console.log(`Text layout benchmark (${ITERATIONS} iterations)\n`);
 
 for (const name in samples) {
 	const text = samples[name as keyof typeof samples];
+	const baseline = currentSanitizeText(text);
 	const jsResult = jsSanitizeText(text);
-	const nativeResult = nativeSanitizeText(text);
-	const currentResult = currentSanitizeText(text);
 	const regexResult = regexSanitizeText(text);
-	if (jsResult !== nativeResult) {
-		console.log(`MISMATCH js/native ${name}`);
+	if (jsResult !== baseline) {
+		console.log(`MISMATCH js/current ${name}`);
 	}
-	if (currentResult !== nativeResult) {
-		console.log(`MISMATCH current/native ${name}: current=${JSON.stringify(currentResult)} native=${JSON.stringify(nativeResult)}`);
-	}
-	if (regexResult !== nativeResult) {
-		console.log(`MISMATCH regex/native ${name}: regex=${JSON.stringify(regexResult)} native=${JSON.stringify(nativeResult)}`);
+	if (regexResult !== baseline) {
+		console.log(`MISMATCH regex/current ${name}: regex=${JSON.stringify(regexResult)} baseline=${JSON.stringify(baseline)}`);
 	}
 	const gatedResult = gatedSanitizeText(text);
 	const skipResult = skipRunSanitizeText(text);
- 	const removalStartResult = removalStartSanitizeText(text);
- 	const wellFormedControlResult = wellFormedControlSanitizeText(text);
- 	const lazyWellFormedResult = lazyWellFormedSanitizeText(text);
-	if (gatedResult !== nativeResult) {
-		console.log(`MISMATCH gated/native ${name}`);
+	const removalStartResult = removalStartSanitizeText(text);
+	const wellFormedControlResult = wellFormedControlSanitizeText(text);
+	const lazyWellFormedResult = lazyWellFormedSanitizeText(text);
+	if (gatedResult !== baseline) {
+		console.log(`MISMATCH gated/current ${name}`);
 	}
-	if (skipResult !== nativeResult) {
-		console.log(`MISMATCH skip/native ${name}: skip=${JSON.stringify(skipResult)} native=${JSON.stringify(nativeResult)}`);
+	if (skipResult !== baseline) {
+		console.log(`MISMATCH skip/current ${name}: skip=${JSON.stringify(skipResult)} baseline=${JSON.stringify(baseline)}`);
+	}
+	if (removalStartResult !== baseline) {
+		console.log(`MISMATCH removalStart/current ${name}: removalStart=${JSON.stringify(removalStartResult)} baseline=${JSON.stringify(baseline)}`);
+	}
+	if (wellFormedControlResult !== baseline) {
+		console.log(`MISMATCH wellFormedControl/current ${name}: wellFormedControl=${JSON.stringify(wellFormedControlResult)} baseline=${JSON.stringify(baseline)}`);
+	}
+	if (lazyWellFormedResult !== baseline) {
+		console.log(`MISMATCH lazyWellFormed/current ${name}: lazyWellFormed=${JSON.stringify(lazyWellFormedResult)} baseline=${JSON.stringify(baseline)}`);
 	}
- 	if (removalStartResult !== nativeResult) {
- 		console.log(`MISMATCH removalStart/native ${name}: removalStart=${JSON.stringify(removalStartResult)} native=${JSON.stringify(nativeResult)}`);
- 	}
- 	if (wellFormedControlResult !== nativeResult) {
- 		console.log(`MISMATCH wellFormedControl/native ${name}: wellFormedControl=${JSON.stringify(wellFormedControlResult)} native=${JSON.stringify(nativeResult)}`);
- 	}
- 	if (lazyWellFormedResult !== nativeResult) {
- 		console.log(`MISMATCH lazyWellFormed/native ${name}: lazyWellFormed=${JSON.stringify(lazyWellFormedResult)} native=${JSON.stringify(nativeResult)}`);
- 	}
 
 	bench(`jsSanitizeText/${name}`, () => {
 		jsSanitizeText(text);
@@ -362,17 +357,14 @@ for (const name in samples) {
 	bench(`skipRunSanitizeText/${name}`, () => {
 		skipRunSanitizeText(text);
 	});
- 	bench(`removalStartSanitizeText/${name}`, () => {
- 		removalStartSanitizeText(text);
- 	});
- 	bench(`wellFormedControlSanitizeText/${name}`, () => {
- 		wellFormedControlSanitizeText(text);
- 	});
- 	bench(`lazyWellFormedSanitizeText/${name}`, () => {
- 		lazyWellFormedSanitizeText(text);
- 	});
-	bench(`nativeSanitizeText/${name}`, () => {
-		nativeSanitizeText(text);
+	bench(`removalStartSanitizeText/${name}`, () => {
+		removalStartSanitizeText(text);
+	});
+	bench(`wellFormedControlSanitizeText/${name}`, () => {
+		wellFormedControlSanitizeText(text);
+	});
+	bench(`lazyWellFormedSanitizeText/${name}`, () => {
+		lazyWellFormedSanitizeText(text);
 	});
 	console.log();
 }