We address the problem of automatically harmonizing a leadsheet in the style of any arranger. We model the arranging style as a Markov model estimated from a corpus of non-annotated MIDI files. We consider a vertical approach to harmonization, in which chords are all taken from the arranger corpus. We show that standard Markov models, using various vertical viewpoints are not adapted for such a task, because the problem is basically over constrained. We propose the concept of fioriture to better capture the subtleties of an arranging style. Fioritures are ornaments of the given melody during which the arranging style can be expressed more freely than for melody notes. Fioritures are defined as random walks with unary constraints and can be implemented with the technique of Markov constraints. We claim that fioritures lead to musically more interesting harmonizations than previous approaches and discuss why. We focus on the style of Take 6, arguably the most sophisticated arranging style in the jazz genre, and we demonstrate the validity of our approach by harmonizing a large corpus of standard leadsheets.