awaforge

htmlEntities

Apache-2.0Release 1L3

HTML5 entities decoder — strict and lenient, WHATWG-compliant.

Module htmlEntities | Source packages/front/fw/src/io/text/html-entities.js | Deps none | Worker-safe yes

Resolve

const ent = runtime.resolve('htmlEntities');
// Returns: { decodeHtmlStrict, decodeHtml, decodeCodePoint, HTML5_ENTITIES }

API

Method Signature Returns
decodeHtmlStrict (s: string) => string Decodes all HTML entity references (semicolon-terminated only). Unknown entities left as-is.
decodeHtml (s: string) => string Decodes HTML entities in lenient mode (semicolon optional).
decodeCodePoint (cp: number) => string Decodes a numeric code point to a string. Returns U+FFFD for forbidden values.
HTML5_ENTITIES Record<string, string> Full WHATWG table — 2,125 semicolon-terminated entries.

Examples

const ent = runtime.resolve('htmlEntities');

// Named entities
ent.decodeHtmlStrict('&amp;');          // → '&'
ent.decodeHtmlStrict('&aacute;');       // → 'á'
ent.decodeHtmlStrict('&Alpha;');        // → 'Α'
ent.decodeHtmlStrict('&unknown;');      // → '&unknown;' (unchanged)

// Numeric
ent.decodeHtmlStrict('&#42;');          // → '*'
ent.decodeHtmlStrict('&#x2A;');         // → '*'
ent.decodeHtmlStrict('&#0;');           // → '�' (U+FFFD)
ent.decodeHtmlStrict('&#128;');         // → '€' (Windows-1252 override)

// Lenient (without semicolon)
ent.decodeHtml('&amp');                 // → '&'

// Low level
ent.decodeCodePoint(42);               // → '*'
ent.decodeCodePoint(0xD800);           // → '�' (surrogate)

Worker Usage

const worker = fw.createWorker(
    function ({ libs, args }) {
        const ent = libs.htmlEntities;
        self.postMessage(ent.decodeHtmlStrict(args[0]));
    },
    { dependencies: ['htmlEntities'], args: ['&lt;p&gt;Hello&lt;/p&gt;'] }
);

Notes

  • Implements CommonMark §6.2: surrogates (0xD800–0xDFFF), &#0;, and code points > 0x10FFFF are all replaced by U+FFFD.
  • Applies the 27 Windows-1252 remappings defined by the HTML5 spec (e.g. &#128; → €).
  • The WHATWG table (2,125 entries) is loaded once at module level and shared across all factory instances — no memory duplication.
  • Strict mode only: semicolon-less forms like &amp are not decoded by decodeHtmlStrict (use decodeHtml for lenient mode).

See also

  • str — escapeHTML for the reverse encoding (encode <, >, &, " as HTML entities)
  • unicode — Unicode normalization and collation