-- unique.e v0.2 by CChris <oedoc@free.fr>
-- exports the unique_filename() function

sequence alphabet
alphabet="abcdefghijklmnopqrstuvwxyz0123456789"
integer lmax
lmax=0

function separate(sequence s1,sequence s2)
--converts s1 and s2 to intermediate form and checks whether they are consecutive
--Returns "" if so, else the pair of intermediate forms
    sequence S1,S2
    integer p

    S2=repeat(0,lmax)
    S1=repeat(0,lmax)
    for i=1 to length(s1) do
        S1[i]=find(s1[i],alphabet)
    end for
    for i=1 to length(s2) do
        S2[i]=find(s2[i],alphabet)
    end for
    p=find(0,S1=S2)
    if length(s1)=lmax and length(s2)=lmax and p=lmax and S2[p]=S1[p]+1 then 
        return {}
    else
        return {S1,S2,p,s1,s2}
    end if
end function

function intermediate(sequence s)
-- s is not empty and returned from separate
integer p
sequence S1,S2

    S1=s[1]
    S2=s[2]
    p=s[3]
    if p<=length(s[4]) and p<=length(s[5]) and S1[p]<S2[p]-1 then
        -- the first diffreent letters are not consecutive
        -- form a name using the common part & the first letter in between
        return s[4][1..p-1]&alphabet[S1[p]+1]
    elsif p>length(s[4]) then
        -- 2nd file name starts with 1st one
        if length(s[5])>p then
            -- second filename is longer by 2 or more chars, return a slice in between
            return s[5][1..p]
        elsif S2[p]>1 then   -- replace last letter with 'a'
            return s[4]&'a'
        elsif S2[p-1]<length(alphabet) then
            return s[4][1..p-2]&alphabet[S2[p-1]+1]
            -- else they are consecutive
        end if
    elsif length(s[4])>p then -- first name is longer by 2 chars or more
        return s[4][1..p]
    elsif S1[length(s[5])]<S2[length(s[5])]-1 then
        -- last common letter is not the last available one
        s[4][p-1]=alphabet[S1[p-1]+1]
        return s[4]
    elsif S1[p]<length(alphabet) then
        -- last letter of first name is not the last available one
        s[4][p]=alphabet[S1[p]+1]
        return s[4]
    elsif length(s[4])<lmax then
        there is some room at the end
        return s[4]&'a'
    end if
    -- shouldn't reach here
end function

include file.e

global function unique_filename(sequence currentDir,integer recurse)
--finds a name that doesn't match any in the directory or subdirs.
--The returned name will use only the 26 lower case letters and the 10 digits.
-- It won't be longer than the longest reported name
-- It will not contain any dot.
-- If the recurse flag is nonzero, files in subdirectories will be taken into account.
-- If the supplied directory is "", the current directory will be used.
    sequence filenames,s
    integer k

    if not length(currentDir) then
        currentDir=current_dir()
    end if
    filenames=dir(currentDir)
    if recurse then  -- add recursively all filenames in subdirs
        k=1
        while k<=length(filenames) do
            if find('d',filenames[k][D_ATTRIBUTES]) and not find(filenames[k],{".",".."}) then
                filenames&=dir(filenames[k][D_NAME])
            end if
            filenames[k]=filenames[k][D_NAME]
            k+=1
        end while
    end if
    k=1
    for i=1 to length(filenames) do
        -- consider all names, so that any extension can be added to the returned name
        s=filenames[i]
        filenames[k]=or_bits(s,and_bits(s,64)/2)  -- fast lower(); work a litle harder to accommodate accented chars
        k+=1
        if length(s)>lmax then lmax=length(s) end if
        for j=1 to length(s) do
            if not find(s[j],alphabet) then
                alphabet&=s[j]
            end if
        end for
    end for
    filenames=sort(filenames[1..k])
    for i=1 to length(filenames)-1 do
        -- find first pair of non consecutive names
        s=separate(filenames[i],filenames[i+1])
        if length(s) then
            return intermediate(s)
        end if
    end for
    -- all filenames are consecutive - a rare case !
    if length(filenames[1])>1 then
        return filenames[1][1..length(filenames[1])-1]
    elsif compare(filenames[1],"a") then
        return "a"
    elsif length(filenames[length(filenames)])<lmax then
        return filenames[length(filenames)]&'a'
    elsif filenames[length(filenames)][length(filenames[length(filenames)])]!=alphabet[length(alphabet)] then
        -- change last letter of greatest name
        filenames[length(filenames)][length(filenames[length(filenames)])]=alphabet[length(alphabet)]
        return filenames[length(filenames)]
    else
        -- this one can't exist. If it is not allowable, it means that all possible names were taken
        return repeat('a',lmax+1)
    end if
end function