html-minifier 3.5.21 → 5.0.0

package/src/lib/whitespace.js

@@ -0,0 +1,431 @@
+/**
+ * Whitespace trimming and collapsing
+ */
+
+import {
+  RE_ALL_WS_NBSP,
+  RE_NBSP_LEADING_GROUP,
+  RE_NBSP_LEAD_GROUP,
+  RE_NBSP_TRAILING_GROUP,
+  RE_NBSP_TRAILING_STRIP,
+  inlineElementsToKeepWhitespace,
+  inlineElementsToKeepWhitespaceAround,
+  inlineElementsToKeepWhitespaceWithin,
+  inlineElementsToKeepWhitespaceBetween,
+  inlineElementsToKeepWhitespaceBetweenEnd,
+  inlineElementsTransparent,
+  inlineElementsTransparentAfter,
+  formControlElementsEither,
+  toEndTags
+} from './constants.js';
+
+const RE_ANY_WS_NBSP = /[ \n\r\t\f\xA0]/;
+const RE_TAB_NBSP = /[\t\xA0]/;
+const RE_ASCII_WS_RUN = /[ \n\r\f]+/g;
+// Every reason `collapseWhitespaceAllKnown` has to do any work at all, in one scan
+const RE_COLLAPSIBLE = /[\t\xA0\n\r\f]| {2}/;
+const RE_NON_WS = /\S/;
+// Any no-break space character (the `collapseNoBreakSpaces` early-exit scan)
+const RE_NO_BREAK_SPACE_CHARS = /[\xA0\u202F\u2007]/;
+
+/** @param {string} str */
+const trimWhitespace = str => {
+  if (!str) return str;
+  let start = 0;
+  let end = str.length;
+  while (start < end && isAsciiWs(str.charCodeAt(start))) start++;
+  while (end > start && isAsciiWs(str.charCodeAt(end - 1))) end--;
+  return start === 0 && end === str.length ? str : str.slice(start, end);
+};
+
+/** @param {number} code */
+function isAsciiWs(code) {
+  return code === 32 || code === 10 || code === 13 || code === 9 || code === 12;
+}
+
+/** @param {number} code */
+function isWsOrNbsp(code) {
+  return code === 32 || code === 10 || code === 13 || code === 9 || code === 12 || code === 160;
+}
+
+// Whether the string is empty or holds only what `\s` matches, scanned by code,
+// as text mostly ends the scan on its first character (where a regex costs more)
+/** @param {string} str */
+function isAllWhitespace(str) {
+  for (let i = 0; i < str.length; i++) {
+    const code = str.charCodeAt(i);
+    if (!((code >= 9 && code <= 13) || code === 32 || code === 0xA0 || code === 0x1680 ||
+        (code >= 0x2000 && code <= 0x200A) || code === 0x2028 || code === 0x2029 ||
+        code === 0x202F || code === 0x205F || code === 0x3000 || code === 0xFEFF)) {
+      return false;
+    }
+  }
+  return true;
+}
+
+// Whether the last character is whitespace, no-break space included
+/** @param {string} str */
+function endsWithWhitespace(str) {
+  return str.length > 0 && isWsOrNbsp(str.charCodeAt(str.length - 1));
+}
+
+// Module scope so the hot path doesn’t allocate a closure per call
+/** @param {string} spaces */
+function collapseRun(spaces) {
+  if (spaces === '\t') return '\t';
+  if (spaces.indexOf('\xA0') === -1) return ' ';
+  return spaces.replace(RE_NBSP_LEADING_GROUP, '$1 ');
+}
+
+/** @param {string} str */
+function collapseWhitespaceAll(str) {
+  if (!str) return str;
+  // Fast path: If there are no common whitespace characters, return early
+  if (!RE_ANY_WS_NBSP.test(str)) {
+    return str;
+  }
+  return collapseWhitespaceAllKnown(str);
+}
+
+// As `collapseWhitespaceAll`, for callers that already know `str` holds whitespace
+/** @param {string} str */
+function collapseWhitespaceAllKnown(str) {
+  // Most text reaching here is already collapsed, so settle that in one scan rather
+  // than two—single spaces with no tab and no no-break space leave nothing to do
+  if (!RE_COLLAPSIBLE.test(str)) {
+    return str;
+  }
+  // Only a tab or a no-break space makes the replacement depend on the run itself;
+  // without either, every run becomes one space and no per-match callback is needed
+  if (!RE_TAB_NBSP.test(str)) {
+    return str.replace(RE_ASCII_WS_RUN, ' ');
+  }
+  return str.replace(RE_ALL_WS_NBSP, collapseRun);
+}
+
+// Collapse whitespace into adjacent no-break spaces
+
+// A no-break space (`U+00A0`), narrow no-break space (`U+202F`), or figure space (`U+2007`),
+// literal or as a character reference; numeric references and `&nbsp` also work without
+// the semicolon in text
+const RE_NO_BREAK_SPACE_WS = /([ \n\r\t\f]*)(\xA0|\u202F|\u2007|&(?:nbsp;?|NonBreakingSpace;|numsp;|#0*(?:160|8239|8199)(?![0-9]);?|#[xX]0*(?:[aA]0|202[fF]|2007)(?![0-9a-fA-F]);?))([ \n\r\t\f]*)/g;
+// Decoded text holds references only as text, so there the characters alone count
+const RE_NO_BREAK_SPACE_CHAR_WS = /([ \n\r\t\f]*)([\xA0\u202F\u2007])([ \n\r\t\f]*)/g;
+const RE_LINE_BREAK = /[\n\r]/;
+
+/**
+ * @param {string} _match
+ * @param {string} before
+ * @param {string} noBreakSpace
+ * @param {string} after
+ */
+function keepLineBreaks(_match, before, noBreakSpace, after) {
+  return (RE_LINE_BREAK.test(before) ? before : '') + noBreakSpace + (RE_LINE_BREAK.test(after) ? after : '');
+}
+
+// Callers pass text whose whitespace runs are already collapsed, as each match
+// rescans the run in front of it
+/**
+ * @param {string} str
+ * @param {{decodeEntities?: boolean | undefined, preserveLineBreaks?: boolean | undefined}} options
+ */
+function collapseNoBreakSpaces(str, options) {
+  const decoded = Boolean(options.decodeEntities);
+  // One scan for all three no-break space characters, rather than an `indexOf` per character
+  if (!str || (!RE_NO_BREAK_SPACE_CHARS.test(str) && (decoded || str.indexOf('&') === -1))) {
+    return str;
+  }
+  const pattern = decoded ? RE_NO_BREAK_SPACE_CHAR_WS : RE_NO_BREAK_SPACE_WS;
+  if (options.preserveLineBreaks) {
+    return str.replace(pattern, keepLineBreaks);
+  }
+  return str.replace(pattern, '$2');
+}
+
+// Collapse whitespace with options
+
+/**
+ * @param {string} str
+ * @param {{preserveLineBreaks?: boolean | undefined, conservativeCollapse?: boolean | undefined}} options
+ * @param {boolean} trimLeft
+ * @param {boolean} trimRight
+ * @param {boolean} [collapseAll]
+ * @param {boolean} [keepUntrimmed] - Whether a side not trimmed keeps its line breaks as written
+ */
+function collapseWhitespace(str, options, trimLeft, trimRight, collapseAll = false, keepUntrimmed = false) {
+  if (!str) return str;
+
+  // Fast path: Nothing to do
+  if (!trimLeft && !trimRight && !collapseAll && (!options.preserveLineBreaks || keepUntrimmed)) {
+    return str;
+  }
+
+  // Fast path: No whitespace at all
+  if (!RE_ANY_WS_NBSP.test(str)) {
+    return str;
+  }
+
+  return collapseWhitespaceKnown(str, options, trimLeft, trimRight, collapseAll, keepUntrimmed);
+}
+
+// As `collapseWhitespace`, for callers that already know `str` is non-empty, holds
+// whitespace, and has something to do
+/**
+ * @param {string} str
+ * @param {{preserveLineBreaks?: boolean | undefined, conservativeCollapse?: boolean | undefined}} options
+ * @param {boolean} trimLeft
+ * @param {boolean} trimRight
+ * @param {boolean} collapseAll
+ * @param {boolean} [keepUntrimmed] - Whether a side not trimmed keeps its line breaks as written
+ */
+function collapseWhitespaceKnown(str, options, trimLeft, trimRight, collapseAll, keepUntrimmed = false) {
+  let lineBreakBefore = ''; let lineBreakAfter = '';
+
+  if (options.preserveLineBreaks) {
+    // Find leading/trailing whitespace containing line breaks manually
+    // (avoids polynomial backtracking with end-anchored lazy quantifiers);
+    // the line-break check scans the run’s char codes directly rather than
+    // testing a sliced copy, so the common case allocates nothing
+    if (trimLeft || !keepUntrimmed) {
+      let leadEnd = 0;
+      let hasLineBreak = false;
+      for (; leadEnd < str.length; leadEnd++) {
+        const code = str.charCodeAt(leadEnd);
+        if (!isAsciiWs(code)) break;
+        // Settling the line break in the same pass keeps the run from being walked twice
+        if (code === 10 || code === 13) hasLineBreak = true;
+      }
+      if (hasLineBreak) {
+        lineBreakBefore = '\n';
+        str = str.slice(leadEnd);
+      }
+    }
+    if (trimRight || !keepUntrimmed) {
+      let trailStart = str.length;
+      let hasLineBreak = false;
+      for (; trailStart > 0; trailStart--) {
+        const code = str.charCodeAt(trailStart - 1);
+        if (!isAsciiWs(code)) break;
+        if (code === 10 || code === 13) hasLineBreak = true;
+      }
+      if (hasLineBreak) {
+        lineBreakAfter = '\n';
+        str = str.slice(0, trailStart);
+      }
+    }
+  }
+
+  if (trimLeft) {
+    // Find the leading whitespace boundary with a loop, so the common case neither
+    // allocates a replacer closure nor runs the regex machinery
+    let start = 0;
+    while (start < str.length && isWsOrNbsp(str.charCodeAt(start))) {
+      start++;
+    }
+    if (start > 0) {
+      const spaces = str.slice(0, start);
+      const conservative = !lineBreakBefore && options.conservativeCollapse;
+      let replacement;
+      if (conservative && spaces === '\t') {
+        replacement = '\t';
+      } else if (spaces.indexOf('\xA0') === -1) {
+        // No no-break space: The whole run goes
+        replacement = conservative ? ' ' : '';
+      } else {
+        // No-break space is specifically handled via the nested regexes
+        replacement = spaces.replace(/^[^\xA0]+/, '').replace(RE_NBSP_LEAD_GROUP, '$1 ') || (conservative ? ' ' : '');
+      }
+      str = replacement + str.slice(start);
+    }
+  }
+
+  if (trimRight) {
+    // Find trailing whitespace boundary manually (avoids polynomial backtracking
+    // with `/[ \n\r\t\f\xA0]+$/` on strings with long internal whitespace runs)
+    let end = str.length;
+    while (end > 0 && isWsOrNbsp(str.charCodeAt(end - 1))) {
+      end--;
+    }
+    if (end < str.length) {
+      const spaces = str.slice(end);
+      const conservative = !lineBreakAfter && options.conservativeCollapse;
+      let replacement;
+      if (conservative && spaces === '\t') {
+        replacement = '\t';
+      } else if (spaces.indexOf('\xA0') === -1) {
+        // No no-break space: The whole run goes
+        replacement = conservative ? ' ' : '';
+      } else {
+        // No-break space is specifically handled via the nested regexes
+        replacement = spaces.replace(RE_NBSP_TRAILING_GROUP, ' $1').replace(RE_NBSP_TRAILING_STRIP, '') || (conservative ? ' ' : '');
+      }
+      str = str.slice(0, end) + replacement;
+    }
+  }
+
+  if (collapseAll && str) {
+    // Strip non-space whitespace then compress spaces to one
+    str = collapseWhitespaceAllKnown(str);
+  }
+
+  // Avoid string concatenation when no line breaks (common case)
+  if (!lineBreakBefore && !lineBreakAfter) return str;
+  if (!lineBreakBefore) return str + lineBreakAfter;
+  if (!lineBreakAfter) return lineBreakBefore + str;
+  return lineBreakBefore + str + lineBreakAfter;
+}
+
+// Collapse whitespace smartly based on surrounding tags
+
+/**
+ * @typedef {object} InlineSets
+ * @prop {Set<string>} around - Elements whose surrounding whitespace is kept
+ * @prop {Set<string>} aroundEnd - The same, spelled as end tags
+ * @prop {Set<string>} within - Elements whose inner whitespace is kept
+ * @prop {Set<string>} withinEnd - The same, spelled as end tags
+ * @prop {Set<string>} withinEither - `within` under both spellings
+ */
+
+/**
+ * Inline-element membership for the whitespace pass, each set paired with its end-tag
+ * spelling so a lookup takes the tag as it comes rather than slicing off the slash
+ * @param {Set<string>} around
+ * @param {Set<string>} within
+ * @returns {InlineSets}
+ */
+function buildInlineSets(around, within) {
+  const withinEnd = toEndTags(within);
+  return {
+    around,
+    aroundEnd: toEndTags(around),
+    within,
+    withinEnd,
+    withinEither: new Set([...within, ...withinEnd])
+  };
+}
+
+// The sets for a document without custom inline elements, built once
+const defaultInlineSets = buildInlineSets(inlineElementsToKeepWhitespaceAround, inlineElementsToKeepWhitespaceWithin);
+
+// Check if an input element has `type="hidden"`; module-scope so the hot path
+// doesn’t allocate a closure per text node
+/**
+ * @param {Array<{name: string, value?: string}>} attrs
+ */
+function isHiddenInput(attrs) {
+  if (!attrs || !attrs.length) return false;
+  for (const attr of attrs) {
+    // Attribute names and the `type` keyword are both ASCII case-insensitive
+    if (attr.name.toLowerCase() === 'type') {
+      return attr.value?.toLowerCase() === 'hidden';
+    }
+  }
+  return false;
+}
+
+/**
+ * @param {string} str
+ * @param {string} prevTag
+ * @param {string} nextTag
+ * @param {Array<{name: string, value?: string}>} prevAttrs
+ * @param {Array<{name: string, value?: string}>} nextAttrs
+ * @param {{preserveLineBreaks?: boolean, conservativeCollapse?: boolean, collapseInlineTagWhitespace?: boolean}} options
+ * @param {InlineSets} inlineSets
+ * @param {boolean} [collapseInside] - False where `canCollapseWhitespace` keeps runs inside the text as they are
+ */
+function collapseWhitespaceSmart(str, prevTag, nextTag, prevAttrs, nextAttrs, options, inlineSets, collapseInside = true) {
+  if (!str) return str;
+
+  // Fast path: No whitespace at all—every decision below would leave `str`
+  // untouched (`collapseWhitespace` returns such strings unchanged)
+  if (!RE_ANY_WS_NBSP.test(str)) {
+    return str;
+  }
+
+  let trimLeft = Boolean(prevTag) && !inlineElementsToKeepWhitespace.has(prevTag);
+  let trimRight = Boolean(nextTag) && !inlineElementsToKeepWhitespace.has(nextTag);
+
+  const prevTransparent = inlineElementsTransparentAfter.has(prevTag);
+  const nextTransparent = inlineElementsTransparent.has(nextTag);
+
+  // Every branch below that consults the text’s content needs a side left untrimmed, an element
+  // that doesn’t render, or `collapseInlineTagWhitespace`, so the scan for a non-whitespace
+  // character is only worth making then
+  const isPureWhitespace = (!trimLeft || !trimRight || prevTransparent || nextTransparent || Boolean(options.collapseInlineTagWhitespace)) && !RE_NON_WS.test(str);
+
+  // Next to text, spacing is part of the text, so aggressive collapsing only applies between tags
+  const inlineOption = Boolean(options.collapseInlineTagWhitespace) && isPureWhitespace;
+
+  if (isPureWhitespace) {
+    // Smart default behavior: Collapse space around non-rendering elements
+    // (`type="hidden"`); this happens even in basic `collapseWhitespace` mode
+    if (!trimLeft && prevTag === 'input' && isHiddenInput(prevAttrs)) {
+      trimLeft = true;
+    }
+    if (!trimRight && nextTag === 'input' && isHiddenInput(nextAttrs)) {
+      trimRight = true;
+    }
+    // Aggressive mode: Collapse between all form controls
+    if (inlineOption && (!trimLeft || !trimRight) &&
+        formControlElementsEither.has(prevTag) && formControlElementsEither.has(nextTag)) {
+      trimLeft = true;
+      trimRight = true;
+    }
+  }
+
+  // Aggressive mode keeps whitespace around standard text-level elements only, not
+  // around all inline elements
+  if (trimLeft) {
+    trimLeft = prevTag.charCodeAt(0) === 47 /* / */ ? !(inlineOption ? inlineElementsToKeepWhitespaceBetweenEnd : inlineSets.aroundEnd).has(prevTag) : !inlineSets.within.has(prevTag);
+  }
+
+  if (trimRight) {
+    trimRight = nextTag.charCodeAt(0) === 47 /* / */ ? !inlineSets.withinEnd.has(nextTag) : !(inlineOption ? inlineElementsToKeepWhitespaceBetween : inlineSets.around).has(nextTag);
+  }
+
+  // Next to an element that doesn’t render, whitespace may separate what lies beyond it, so the
+  // element leaves the decision to the text or to a tag on the other side that renders
+  if (prevTransparent && (!isPureWhitespace || (Boolean(nextTag) && !nextTransparent))) {
+    trimLeft = false;
+  }
+  if (nextTransparent && (!isPureWhitespace || (Boolean(prevTag) && !prevTransparent))) {
+    trimRight = false;
+  }
+
+  const collapseAll = collapseInside && Boolean(prevTag && nextTag);
+  if (!trimLeft && !trimRight && !collapseAll && (!options.preserveLineBreaks || !collapseInside)) {
+    return str;
+  }
+  return collapseWhitespaceKnown(str, options, trimLeft, trimRight, collapseAll, !collapseInside);
+}
+
+// Collapse/trim whitespace for given tag
+
+// `xmp`, `listing`, and `plaintext` render preformatted, as `pre` does
+// https://html.spec.whatwg.org/multipage/rendering.html#the-page
+const noCollapseWhitespaceTags = new Set(['script', 'style', 'pre', 'textarea', 'xmp', 'listing', 'plaintext']);
+const noTrimWhitespaceTags = new Set(['pre', 'textarea', 'xmp', 'listing', 'plaintext']);
+
+/** @param {string} tag */
+function canCollapseWhitespace(tag) {
+  return !noCollapseWhitespaceTags.has(tag);
+}
+
+/** @param {string} tag */
+function canTrimWhitespace(tag) {
+  return !noTrimWhitespaceTags.has(tag);
+}
+
+export {
+  buildInlineSets,
+  defaultInlineSets,
+  endsWithWhitespace,
+  trimWhitespace,
+  collapseWhitespaceAll,
+  collapseNoBreakSpaces,
+  collapseWhitespace,
+  collapseWhitespaceSmart,
+  canCollapseWhitespace,
+  canTrimWhitespace,
+  isAllWhitespace
+};

package/src/presets.js

@@ -0,0 +1,58 @@
+/**
+ * Preset configurations
+ *
+ * Presets provide curated option sets for common use cases:
+ * - `conservative`: Safe minification suitable for most projects
+ * - `comprehensive`: Aggressive minification for maximum file size reduction
+ */
+
+export const presets = {
+  conservative: {
+    caseSensitive: true,
+    collapseBooleanAttributes: true,
+    collapseWhitespace: true,
+    conservativeCollapse: true,
+    preserveLineBreaks: true,
+    removeComments: true,
+    removeDefaultTypeAttributes: true,
+    useShortDoctype: true
+  },
+  comprehensive: {
+    collapseAttributeWhitespace: true,
+    collapseBooleanAttributes: true,
+    collapseEmptyAttributes: true,
+    collapseWhitespace: true,
+    decodeEntities: true,
+    mergeScripts: true,
+    minifyCSS: true,
+    minifyJS: true,
+    minifyJSON: true,
+    minifySVG: true,
+    minifyURLs: true,
+    removeAttributeQuotes: true,
+    removeComments: true,
+    removeDefaultTypeAttributes: true,
+    removeEmptyAttributes: true,
+    removeOptionalTags: true,
+    removeRedundantAttributes: true,
+    selfCloseForeignElements: true,
+    useShortDoctype: true
+  }
+};
+
+/**
+ * @param {string} name - Preset name (“conservative” or “comprehensive”)
+ * @returns {object|null} Preset options object or null if not found
+ */
+export function getPreset(name) {
+  if (!name) return null;
+  const normalizedName = name.toLowerCase();
+  return /** @type {Record<string, object>} */ (presets)[normalizedName] || null;
+}
+
+/**
+ * @returns {string[]} Array of preset names
+ */
+export function getPresetNames() {
+  return Object.keys(presets);
+}