html_entity_decode Not Working: The Usual Causes

Missing flags

Single quotes are the most common survivor. Without ENT_QUOTES, ' and ' are left untouched on older PHP defaults:

html_entity_decode($s);                                    // may skip quotes
html_entity_decode($s, ENT_QUOTES | ENT_HTML5, 'UTF-8');   // decodes everything

' specifically needs ENT_HTML5 — it is not part of the HTML 4 entity set that older defaults use.

Wrong charset

If the third argument does not match the actual encoding of the string, PHP may return an empty string rather than an error — which looks like a total failure rather than a mismatch.

Pass 'UTF-8' explicitly and make sure the input really is UTF-8. A Latin-1 string mislabelled as UTF-8 is a frequent cause of empty output.

Double encoding

If the source contains  , one decode gives you the literal text   — which looks exactly like nothing happened. Decoding again would produce the space, and that is precisely the pattern you must not automate.

Find where the double encoding happens instead. It is usually a value escaped once before storage and once on output.

It was never an entity

Entities require a trailing semicolon in most contexts. &nbsp without it may render in a browser through error recovery but will not decode.

Invented entities do not decode either — &arrow; is not in the HTML5 set no matter how reasonable it looks. Check the name against the specification, or use the numeric form.

And if the string is already decoded, a non-breaking space (U+00A0) looks identical to a regular space in most editors. Check the byte values before concluding the decode failed.