nx_html_escape.nx

html_escape.nx -- HTML entity encoder + decoder. Escapes the five characters that are NEVER safe in HTML text content or attribute values: < &lt; > &gt; & &amp; " &quot; ' &#39; (apostrophe -- &apos; is XML only, not HTML4) This is the XSS-prevention primitive. Any user-supplied text rendered into an HTML template MUST go through this first. Nishi-pages, every sovereign web handler, and any template engine built on top all rely on correct escaping. Decoder handles the five named entities + numeric character references (&#NN; and &#xHH;) bounded to the ASCII range. Full Unicode entity decoding is out of scope (would need the HTML5 named-character-reference table of ~2000 entries); callers with that need should pre-process server-side into one of the five safe forms. Invariants: HE1 Encoding output is never larger than 6x input (worst case: every byte is '>' which becomes "&gt;" = 4 bytes). Caller sizes the output buffer at src_len * 6 + 1. HE2 Encode is idempotent on already-safe bytes (ASCII alnum, space, punctuation except the five). HE3 Decode rejects malformed entities by leaving them literal (per WHATWG tolerant parsing); unknown named entities also pass through literal. HE4 Round trip: decode(encode(x)) == x for any ASCII x.