chocoholic (this spelling is only slightly more common than 'chocaholic' according to google. both are domain names, but chocoholic.com is much fancier.)
really, whatever-you-like-aholic. so we seem to append the morpheme -aholic (according to certain phonological/phonotactic rules) to mean 'addicted to that noun.' any reasonable person might guess that '-aholic' means something like 'addicted to.'
but consider the form 'alcoholic'. here (and I think most people would argue that this is the origination of the -aholic morpheme), we're actually just appending 'ic'. so how did -ahol become attached from 'alcohol'?
the suffix -aholic has its own listing in the OED online. it reads:
The final element of WORKAHOLIC (after ALCOHOLIC n. 2) used as a suffix forming ns., as computerholic, newsaholic, spendaholic, etc., (chiefly humorous nonce-words) denoting one who appears to be addicted to the object, activity, etc., specified; a person subject to an inordinate craving for or obsession with (something).
according to wikipedia:
Etymologically, "chocoholic" is a blend of "chocolate" and "alcoholic", though some linguists complain that the word, by construction, implies addiction to "chocohol" rather than "chocolate", suggesting that chocolatic is a more appropriate neologism than chocoholic.
under this etymologic history, the word is a portmanteau, but I would argue that -aholic is productive enough as a suffix to merit giving it that label.
the fallacy of the linguists' argument as presented in the wikipedia quotation lies in the idea that construction should imply definite relationships. sure, construction often does and can imply things about meaning (maybe because speakers are aware that novel utterances are more likely to be understood if they follow established patterns?), but the idea that some neologisms are 'more appropriate' than others stems from the mistaken assumption that causality matters in language change. again, speakers tend to make novel utterances predictable, not necessarily logical. it's correlation, not causation, that produces meaning.