Jaro Winkler string similarity for Progress Openedge

/* *************************** Definitions ************************** */ define input parameter v-one as character no-undo. define input parameter v-two as character no-undo. define variable v-shorter as character no-undo. define variable v-longer as character no-undo. define variable v-searchrange as integer no-undo. define variable v-aux1 as character no-undo. define variable v-aux2 as character no-undo. define variable v-distance as decimal no-undo. define variable v-t as decimal no-undo. define variable v-calc as decimal no-undo. define variable v-cnt as integer no-undo. define variable v-l as integer no-undo. define temp-table tt-word field letter as character field pos as integer. define temp-table tt-word2 field letter as character field pos as integer. /* *************************** Main Block *************************** */ procedure jaro.ip: define input parameter v-first as character. define input parameter v-second as character. define input parameter v-limit as integer. define input-output parameter v-new as character. define input-output parameter v-trans as decimal. define variable v-calculated as character no-undo. define variable v-thischar as character no-undo. define variable v-found as logical no-undo initial ?. define variable v-cnt as integer no-undo. define variable v-min as integer no-undo. define variable v-max as integer no-undo. define variable v-diff as integer no-undo. empty temp-table tt-word. empty temp-table tt-word2. assign v-new = "" v-trans = 0. repeat v-cnt = 1 to length(v-first) : create tt-word. assign tt-word.pos = v-cnt tt-word.letter = substring(v-first,v-cnt,1). end. repeat v-cnt = 1 to length(v-second) : create tt-word2. assign tt-word2.pos = v-cnt tt-word2.letter = substring(v-second,v-cnt,1). end. v-new = "". for each tt-word exclusive-lock: find first tt-word2 no-lock where tt-word2.pos = tt-word.pos no-error. if available tt-word2 then do: if tt-word2.letter = tt-word.letter then do: v-new = v-new + tt-word.letter. end. end. find first tt-word2 no-lock where tt-word2.letter = tt-word.letter and tt-word2.pos <> tt-word.pos no-error. if available tt-word2 then do: /*see if it is in search range*/ v-diff = absolute(tt-word2.pos - tt-word.pos). if v-diff <= v-limit then do: assign v-trans = v-trans + 1 v-new = v-new + tt-word.letter. end. end. end. end procedure. /*********************************************************/ /*MAIN*/ /*********************************************************/ if length(v-one) > length(v-two) then do: assign v-longer = lc(v-one) v-shorter = lc(v-two). end. else do: assign v-longer = lc(v-two) v-shorter = lc(v-one). end. assign v-searchrange = ( length(v-longer) / 2) - 1. run jaro.ip(v-shorter,v-longer, v-searchrange,input-output v-aux1,input-output v-t). run jaro.ip(v-longer,v-shorter, v-searchrange,input-output v-aux2,input-output v-t). v-t = v-t / 2. if length(v-aux1) = 0 or length(v-aux2) = 0 then do: v-distance = 0. end. if length(v-aux1) <> length(v-aux2) then do: v-distance = 0. end. v-distance = (( length(v-aux1) / length(v-shorter) ) + ( length(v-aux2) / length(v-longer) ) + ((length(v-aux1) - v-t) / length(v-aux1))) / 3. if v-distance > 0.7 then do: constantblock: repeat v-cnt = 1 to length(v-shorter): if substring(v-shorter,v-cnt,1) = substring(v-longer,v-cnt,1) then do: v-l = v-l + 1. end. else do: leave constantblock. end. end. end. v-calc = v-distance + v-l * 0.1 * (1 - v-distance).