text-sanctifier 1.3.4 → 1.3.6

package/src/sanctifyText.js

@@ -2,6 +2,8 @@
 
 import { CONTROL_CHARS_REGEX, INVISIBLE_TRASH_REGEX } from './categories.generated.js';
 export { CONTROL_CHARS_REGEX, INVISIBLE_TRASH_REGEX };
+import { ASCII_KEYBOARD_SAFE_REGEX, EMOJI_REGEX, normalizeTypographicJank } from './textPatterns.js';
+export { ASCII_KEYBOARD_SAFE_REGEX, EMOJI_REGEX, normalizeTypographicJank };
 
 
 /**
@@ -18,6 +20,8 @@ export { CONTROL_CHARS_REGEX, INVISIBLE_TRASH_REGEX };
  * @property {boolean} [finalTrim]
  * @property {number} [maxLength]
  * @property {'throw'|'truncate'|'noop'} [onMaxLength]
+ * @property {boolean} [normalizeSpaces] - Replace look-alike spaces with an ordinary space (runs first).
+ * @property {boolean} [purgeZeroWidthSpaces] - Remove U+200B, U+2060 and U+FEFF only, keeping joiners and selectors.
  */
 
 
@@ -39,9 +43,13 @@ export { CONTROL_CHARS_REGEX, INVISIBLE_TRASH_REGEX };
  * @param {boolean} [defaultOptions.finalTrim]
  * @param {number} [defaultOptions.maxLength=Infinity] - Hard input length cap (UTF-16 code units).
  * @param {'throw' | 'truncate' | 'noop'} [defaultOptions.onMaxLength='throw'] - Behavior when input exceeds maxLength.
+ * @param {boolean} [defaultOptions.normalizeSpaces]
+ * @param {boolean} [defaultOptions.purgeZeroWidthSpaces]
  * @returns {(text: string) => string}
  */
-export function summonSanctifier(defaultOptions = {}) {
+function createSanctifier(defaultOptions = {}) {
+    const normalizeSpaces = !!defaultOptions.normalizeSpaces;
+    const purgeZeroWidthSpaces = !!defaultOptions.purgeZeroWidthSpaces;
     const purgeInvisibleChars = !!defaultOptions.purgeInvisibleChars;
     const purgeEmojis = !!defaultOptions.purgeEmojis;
     const nukeControls = !!defaultOptions.nukeControls;
@@ -79,11 +87,16 @@ export function summonSanctifier(defaultOptions = {}) {
             collapseSpaces,
             finalTrim,
             maxLength,
-            onMaxLength
+            onMaxLength,
+            normalizeSpaces,
+            purgeZeroWidthSpaces
         );
 }
 
-// --- Added Presets ---
+// --- Presets ---
+//
+// Each preset is its own export, so a bundle that imports one preset leaves out the
+// others. summonSanctifier.strict and the rest are the same functions.
 
 /**
  * Strict sanitizer:
@@ -91,21 +104,24 @@ export function summonSanctifier(defaultOptions = {}) {
  * - Collapse all newlines
  * - Collapse spaces
  * - Nuke control characters
+ * @param {string} text
+ * @returns {string}
  */
-/** @param {string} text @returns {string} */
-summonSanctifier.strict = text => sanctifyText(
-    text,
-    true,   // purgeInvisibleChars
-    true,   // purgeEmojis
-    true,   // nukeControls
-    false,  // keyboardOnlyFilter
-    true,   // normalizeNewlines
-    true,   // trimSpacesAroundNewlines
-    true,   // collapseNewLines
-    false,  // preserveParagraphs
-    true,   // collapseSpaces
-    true    // finalTrim
-);
+export function sanctifyStrict(text) {
+    return sanctifyText(
+        text,
+        true,   // purgeInvisibleChars
+        true,   // purgeEmojis
+        true,   // nukeControls
+        false,  // keyboardOnlyFilter
+        true,   // normalizeNewlines
+        true,   // trimSpacesAroundNewlines
+        true,   // collapseNewLines
+        false,  // preserveParagraphs
+        true,   // collapseSpaces
+        true    // finalTrim
+    );
+}
 
 
 /**
@@ -113,21 +129,24 @@ summonSanctifier.strict = text => sanctifyText(
  * - Collapse spaces
  * - Preserve paragraphs
  * - Normalize newlines
+ * @param {string} text
+ * @returns {string}
  */
-/** @param {string} text @returns {string} */
-summonSanctifier.loose = text => sanctifyText(
-    text,
-    false,  // purgeInvisibleChars
-    false,  // purgeEmojis
-    false,  // nukeControls
-    false,  // keyboardOnlyFilter
-    true,   // normalizeNewlines
-    true,   // trimSpacesAroundNewlines
-    true,   // collapseNewLines
-    true,   // preserveParagraphs
-    true,   // collapseSpaces
-    true    // finalTrim
-);
+export function sanctifyLoose(text) {
+    return sanctifyText(
+        text,
+        false,  // purgeInvisibleChars
+        false,  // purgeEmojis
+        false,  // nukeControls
+        false,  // keyboardOnlyFilter
+        true,   // normalizeNewlines
+        true,   // trimSpacesAroundNewlines
+        true,   // collapseNewLines
+        true,   // preserveParagraphs
+        true,   // collapseSpaces
+        true    // finalTrim
+    );
+}
 
 
 /**
@@ -135,21 +154,24 @@ summonSanctifier.loose = text => sanctifyText(
  * - Keeps emojis and printable ASCII
  * - Strips non-standard characters
  * - Normalizes typographic trash
+ * @param {string} text
+ * @returns {string}
  */
-/** @param {string} text @returns {string} */
-summonSanctifier.keyboardOnlyEmoji = text => sanctifyText(
-    text,
-    false,  // purgeInvisibleChars
-    false,  // purgeEmojis
-    false,  // nukeControls
-    true,   // keyboardOnlyFilter
-    true,   // normalizeNewlines
-    true,   // trimSpacesAroundNewlines
-    false,  // collapseNewLines
-    false,  // preserveParagraphs
-    false,  // collapseSpaces
-    true    // finalTrim
-);
+export function sanctifyKeyboardOnlyEmoji(text) {
+    return sanctifyText(
+        text,
+        false,  // purgeInvisibleChars
+        false,  // purgeEmojis
+        false,  // nukeControls
+        true,   // keyboardOnlyFilter
+        true,   // normalizeNewlines
+        true,   // trimSpacesAroundNewlines
+        false,  // collapseNewLines
+        false,  // preserveParagraphs
+        false,  // collapseSpaces
+        true    // finalTrim
+    );
+}
 
 
 /**
@@ -157,21 +179,97 @@ summonSanctifier.keyboardOnlyEmoji = text => sanctifyText(
  * - Removes emojis
  * - Collapses all whitespace
  * - Restricts to printable ASCII only
+ * @param {string} text
+ * @returns {string}
  */
-/** @param {string} text @returns {string} */
-summonSanctifier.keyboardOnly = text => sanctifyText(
-    text,
-    true,   // purgeInvisibleChars
-    true,   // purgeEmojis
-    true,   // nukeControls
-    true,   // keyboardOnlyFilter
-    true,   // normalizeNewlines
-    true,   // trimSpacesAroundNewlines
-    true,   // collapseNewLines
-    false,  // preserveParagraphs
-    true,   // collapseSpaces
-    true    // finalTrim
-);
+export function sanctifyKeyboardOnly(text) {
+    return sanctifyText(
+        text,
+        true,   // purgeInvisibleChars
+        true,   // purgeEmojis
+        true,   // nukeControls
+        true,   // keyboardOnlyFilter
+        true,   // normalizeNewlines
+        true,   // trimSpacesAroundNewlines
+        true,   // collapseNewLines
+        false,  // preserveParagraphs
+        true,   // collapseSpaces
+        true    // finalTrim
+    );
+}
+
+
+/**
+ * Pasted text and model output:
+ * - Look-alike spaces (NBSP, narrow NBSP, typographic spaces) become ordinary spaces,
+ *   so words never run together
+ * - Zero-width spaces, word joiners and BOMs are removed
+ * - Newlines are normalized to `\n`
+ * Keeps emoji joiners, variation selectors, bidi marks, indentation and line breaks.
+ * @param {string} text
+ * @returns {string}
+ */
+export function sanctifyPasted(text) {
+    return sanctifyText(
+        text,
+        false,  // purgeInvisibleChars
+        false,  // purgeEmojis
+        false,  // nukeControls
+        false,  // keyboardOnlyFilter
+        true,   // normalizeNewlines
+        false,  // trimSpacesAroundNewlines
+        false,  // collapseNewLines
+        false,  // preserveParagraphs
+        false,  // collapseSpaces
+        false,  // finalTrim
+        Infinity, // maxLength
+        'throw',  // onMaxLength
+        true,   // normalizeSpaces
+        true    // purgeZeroWidthSpaces
+    );
+}
+
+
+/**
+ * Summons a customized sanctifier function with pre-bound booleans.
+ *
+ * Accepts full flag names and returns a text-cleaning function.
+ *
+ * @param {Object} [defaultOptions={}]
+ * @param {boolean} [defaultOptions.purgeInvisibleChars]
+ * @param {boolean} [defaultOptions.purgeEmojis]
+ * @param {boolean} [defaultOptions.nukeControls]
+ * @param {boolean} [defaultOptions.keyboardOnlyFilter]
+ * @param {boolean} [defaultOptions.normalizeNewlines]
+ * @param {boolean} [defaultOptions.trimSpacesAroundNewlines]
+ * @param {boolean} [defaultOptions.collapseNewLines]
+ * @param {boolean} [defaultOptions.preserveParagraphs]
+ * @param {boolean} [defaultOptions.collapseSpaces]
+ * @param {boolean} [defaultOptions.finalTrim]
+ * @param {number} [defaultOptions.maxLength=Infinity] - Hard input length cap (UTF-16 code units).
+ * @param {'throw' | 'truncate' | 'noop'} [defaultOptions.onMaxLength='throw'] - Behavior when input exceeds maxLength.
+ * @param {boolean} [defaultOptions.normalizeSpaces]
+ * @param {boolean} [defaultOptions.purgeZeroWidthSpaces]
+ * @returns {(text: string) => string}
+ *
+ * The presets are attached as properties (`summonSanctifier.strict`, `.loose`,
+ * `.keyboardOnlyEmoji`, `.keyboardOnly`, `.pasted`) and are also standalone exports.
+ * Built in a call marked pure, so a bundle that imports only a preset leaves it out.
+ * @type {typeof createSanctifier & {
+ *   strict: typeof sanctifyStrict,
+ *   loose: typeof sanctifyLoose,
+ *   keyboardOnlyEmoji: typeof sanctifyKeyboardOnlyEmoji,
+ *   keyboardOnly: typeof sanctifyKeyboardOnly,
+ *   pasted: typeof sanctifyPasted,
+ * }}
+ */
+export const summonSanctifier = /* @__PURE__ */ Object.assign(createSanctifier, {
+    strict: sanctifyStrict,
+    loose: sanctifyLoose,
+    keyboardOnlyEmoji: sanctifyKeyboardOnlyEmoji,
+    keyboardOnly: sanctifyKeyboardOnly,
+    pasted: sanctifyPasted,
+});
 
 
 /**
@@ -198,6 +296,9 @@ summonSanctifier.keyboardOnly = text => sanctifyText(
  *   - `'throw'`: throw a `RangeError`
  *   - `'truncate'`: slice to `maxLength` before processing
  *   - `'noop'`: return the original input unchanged (skips processing)
+ * @param {boolean} [normalizeSpaces=false] - Replace look-alike spaces with an ordinary space. Runs first,
+ *   so `purgeInvisibleChars` no longer joins the words around them.
+ * @param {boolean} [purgeZeroWidthSpaces=false] - Remove zero-width spaces, word joiners and BOMs only.
  * @returns {string}
  * @throws {TypeError} If `text` is not a string.
  * @throws {RangeError} If input exceeds `maxLength` and `onMaxLength` is `'throw'`.
@@ -215,7 +316,9 @@ export function sanctifyText(
     collapseSpaces = false,
     finalTrim = false,
     maxLength = Infinity,
-    onMaxLength = 'throw'
+    onMaxLength = 'throw',
+    normalizeSpaces = false,
+    purgeZeroWidthSpaces = false
 ) {
     if (typeof text !== 'string') {
         throw new TypeError('sanctifyText expects a string input.');
@@ -242,6 +345,12 @@ export function sanctifyText(
 
     let cleaned = text;
 
+    // Look-alike spaces become ordinary spaces, before anything can delete them
+    if (normalizeSpaces) cleaned = cleaned.replace(SPACE_LOOKALIKES_REGEX, ' ');
+
+    // Zero-width spaces, word joiners and BOMs (not joiners or selectors)
+    if (purgeZeroWidthSpaces) cleaned = cleaned.replace(ZERO_WIDTH_SPACES_REGEX, '');
+
     //  Purge invisible Unicode trash (zero-width, non-breaking, bidi junk, etc.)
     if (purgeInvisibleChars) cleaned = purgeInvisibleTrash(cleaned);
 
@@ -274,6 +383,16 @@ export function sanctifyText(
 // --- Micro helpers ---
 
 
+/**
+ * Look-alike spaces: the `nonbreaking_space` category except U+3000, the ordinary
+ * space of CJK text. A test checks this against the category table.
+ */
+export const SPACE_LOOKALIKES_REGEX = /[\u00A0\u2000-\u200A\u202F\u205F]/g;
+
+/** Zero-width space, word joiner and BOM: invisible and never needed in plain text. */
+export const ZERO_WIDTH_SPACES_REGEX = /[\u200B\u2060\uFEFF]/g;
+
+
 /**
  * Removes the invisible Unicode characters in INVISIBLE_TRASH_REGEX: non-breaking and
  * typographic spaces, zero-width characters, bidi controls, invisible format characters,
@@ -317,13 +436,6 @@ function extractEmojiSequences(text) {
     return out;
 }
 
-/**
- * Matches any character that is NOT:
- * - Printable ASCII (U+0020–U+007E)
- * - Newline characters: \n (LF), \r (CR)
- * This allows for \n, \r, and \r\n line endings.
- */
-export const ASCII_KEYBOARD_SAFE_REGEX = /[^\x20-\x7E\r\n]+/gu;
 
 /**
  * Removes all non-keyboard characters, preserving:
@@ -345,64 +457,6 @@ function purgeNonKeyboardChars(text, purgeEmojis = false) {
 }
 
 
-// Smart apostrophes (single quote variations)
-const SMART_SINGLE_QUOTES_REGEX = /[\u2018\u2019\u201A\u201B\u2032\u2035]/g;
-
-// Smart double quotes (including primes and guillemets)
-const SMART_DOUBLE_QUOTES_REGEX = /[\u201C\u201D\u201E\u201F\u2033\u2036\u00AB\u00BB]/g;
-
-// Dash variants: en dash, em dash, horizontal bar, figure dash, minus
-const UNICODE_DASHES_REGEX = /[\u2012\u2013\u2014\u2015\u2212]/g;
-
-// Ellipsis character
-const ELLIPSIS_REGEX = /\u2026/g;
-
-// Bullets and middle dot
-const BULLETS_REGEX = /[\u2022\u00B7]/g;
-
-// Full-width ASCII punctuation: U+FF01 - U+FF5E
-const FULLWIDTH_PUNCTUATION_REGEX = /[\uFF01-\uFF5E]/g;
-
-/**
- * Normalizes typographic Unicode punctuation into ASCII equivalents.
- * @param {string} text
- * @returns {string}
- */
-export function normalizeTypographicJank(text) {
-    return text
-        .replace(SMART_SINGLE_QUOTES_REGEX, "'")
-        .replace(SMART_DOUBLE_QUOTES_REGEX, '"')
-        .replace(UNICODE_DASHES_REGEX, '-')
-        .replace(ELLIPSIS_REGEX, '...')
-        .replace(BULLETS_REGEX, '*')
-        .replace(FULLWIDTH_PUNCTUATION_REGEX, m =>
-            String.fromCharCode(m.charCodeAt(0) - 0xFEE0)
-        );
-}
-
-
-/** @type {RegExp} */
-export let EMOJI_REGEX;
-
-/**
- * Try Unicode property escape regex (preferred).
- * Fallback to basic emoji ranges if unsupported.
- */
-try {
-    EMOJI_REGEX = new RegExp(
-        '(?:\\p{Extended_Pictographic}(?:\\uFE0F|\\uFE0E)?(?:\\u200D(?:\\p{Extended_Pictographic}|\\w)+)*)',
-        'gu'
-    );
-} catch {
-    // Fallback: wide-range emoji component match (flags, tones, symbols)
-    //   * Covers:
-    //   * - Emoji base chars (1F300–1FAFF)
-    //   * - Dingbats (2700–27BF) like ❌, ✅, ☑️
-    //   * - Skin tones (1F3FB–1F3FF)
-    //   * - ZWJ (200D), variation selectors (FE0F)
-    //   * - Regional indicators (1F1E6–1F1FF)
-    EMOJI_REGEX = /[\u{2700}-\u{27BF}\u{1F300}-\u{1FAFF}\u{1F3FB}-\u{1F3FF}\u200D\uFE0F\u{1F1E6}-\u{1F1FF}]/gu;
-}
 
 /**
  * Removes all emoji characters using Unicode property escapes.

package/src/textPatterns.js

@@ -0,0 +1,78 @@
+// src/textPatterns.js
+//
+// Patterns shared by the inspector and the sanitizer. They live apart from
+// sanctifyText.js so that importing an inspector does not pull in the sanitizer
+// engine and its presets (see the size probes in scripts/check-size.js).
+
+
+/**
+ * Matches any character that is NOT:
+ * - Printable ASCII (U+0020–U+007E)
+ * - Newline characters: \n (LF), \r (CR)
+ * This allows for \n, \r, and \r\n line endings.
+ */
+export const ASCII_KEYBOARD_SAFE_REGEX = /[^\x20-\x7E\r\n]+/gu;
+
+
+// Smart apostrophes (single quote variations)
+const SMART_SINGLE_QUOTES_REGEX = /[\u2018\u2019\u201A\u201B\u2032\u2035]/g;
+
+// Smart double quotes (including primes and guillemets)
+const SMART_DOUBLE_QUOTES_REGEX = /[\u201C\u201D\u201E\u201F\u2033\u2036\u00AB\u00BB]/g;
+
+// Dash variants: en dash, em dash, horizontal bar, figure dash, minus
+const UNICODE_DASHES_REGEX = /[\u2012\u2013\u2014\u2015\u2212]/g;
+
+// Ellipsis character
+const ELLIPSIS_REGEX = /\u2026/g;
+
+// Bullets and middle dot
+const BULLETS_REGEX = /[\u2022\u00B7]/g;
+
+// Full-width ASCII punctuation: U+FF01 - U+FF5E
+const FULLWIDTH_PUNCTUATION_REGEX = /[\uFF01-\uFF5E]/g;
+
+/**
+ * Normalizes typographic Unicode punctuation into ASCII equivalents.
+ * @param {string} text
+ * @returns {string}
+ */
+export function normalizeTypographicJank(text) {
+    return text
+        .replace(SMART_SINGLE_QUOTES_REGEX, "'")
+        .replace(SMART_DOUBLE_QUOTES_REGEX, '"')
+        .replace(UNICODE_DASHES_REGEX, '-')
+        .replace(ELLIPSIS_REGEX, '...')
+        .replace(BULLETS_REGEX, '*')
+        .replace(FULLWIDTH_PUNCTUATION_REGEX, m =>
+            String.fromCharCode(m.charCodeAt(0) - 0xFEE0)
+        );
+}
+
+
+/**
+ * Try Unicode property escape regex (preferred).
+ * Fallback to basic emoji ranges if unsupported.
+ * Built in a function marked pure, so bundlers drop it when nothing uses it.
+ * @returns {RegExp}
+ */
+function buildEmojiRegex() {
+    try {
+        return new RegExp(
+            '(?:\\p{Extended_Pictographic}(?:\\uFE0F|\\uFE0E)?(?:\\u200D(?:\\p{Extended_Pictographic}|\\w)+)*)',
+            'gu'
+        );
+    } catch {
+        // Fallback: wide-range emoji component match (flags, tones, symbols)
+        //   * Covers:
+        //   * - Emoji base chars (1F300–1FAFF)
+        //   * - Dingbats (2700–27BF) like ❌, ✅, ☑
+        //   * - Skin tones (1F3FB–1F3FF)
+        //   * - ZWJ (200D), variation selectors (FE0F)
+        //   * - Regional indicators (1F1E6–1F1FF)
+        return /[\u{2700}-\u{27BF}\u{1F300}-\u{1FAFF}\u{1F3FB}-\u{1F3FF}\u200D\uFE0F\u{1F1E6}-\u{1F1FF}]/gu;
+    }
+}
+
+/** @type {RegExp} */
+export const EMOJI_REGEX = /* @__PURE__ */ buildEmojiRegex();

package/src/version.generated.js

@@ -0,0 +1,5 @@
+// src/version.generated.js
+// Generated from package.json by scripts/gen-version.js (`npm run gen`); do not edit.
+
+/** The installed version of text-sanctifier, as in its package.json. */
+export const VERSION = '1.3.6';