Incorrect escaping of & without a trailing space

[TimG1964]

Maybe this is somehow connected with #17. Or maybe I'm simply misunderstanding something.

I'm using XML.jl to work directly with Excel files. A string like this is a valid cell entry in Excel:

"https://myhouse.sharepoint.com/:i:/r/sites/CorporateServices-COR08PublicAffairsandRelations/Shared%20Documents/COR08%20Public%20Affairs%20and%20Relations/Case%20Studies/Photos/My%20favourite%20society/MAT_5946%20(1).jpg?csf=1&web=1&e=tIywhD"

I can paste it directly in to an Excel workbook by hand.

But I cannot use XML.jl to add it to a cell programmatically because XML.escape() won't escape the two ampersands. The escape function looks for a space after the ampersand, and so these two aren't found.

julia> const escape_chars = ('&' => "&amp;", '<' => "&lt;", '>' => "&gt;", "'" => "&apos;", '"' => "&quot;")
('&' => "&amp;", '<' => "&lt;", '>' => "&gt;", "'" => "&apos;", '"' => "&quot;")

julia> function escape(x::String)
           result = replace(x, r"&(?=\s;)" => "&amp;")
           for (pat, r) in escape_chars[2:end]
                   result = replace(result, pat => r)
           end
           return result
       end
escape (generic function with 1 method)

julia> x = "https://myhouse.sharepoint.com/:i:/r/sites/CorporateServices-COR08PublicAffairsandRelations/Shared%20Documents/COR08%20Public%20Affairs%20and%20Relations/Case%20Studies/Photos/My%20favourite%20society/MAT_5946%20(1).jpg?csf=1&web=1&e=tIywhD"
"https://myhouse.sharepoint.com/:i:/r/sites/CorporateServices-COR08PublicAffairsandRelations/Shared%20Documents/COR08%20Public%20Affairs%20and%20Relations/Case%20Studies/Photos/My%20favourite%20society/MAT_5946%20(1).jpg?csf=1&web=1&e=tIywhD"

julia> escape(x)
"https://myhouse.sharepoint.com/:i:/r/sites/CorporateServices-COR08PublicAffairsandRelations/Shared%20Documents/COR08%20Public%20Affairs%20and%20Relations/Case%20Studies/Photos/My%20favourite%20society/MAT_5946%20(1).jpg?csf=1&web=1&e=tIywhD"

julia> 

A simple change of the regex to r"&(?!amp;|quot;|apos;|gt;|lt;)" seems to work for me, but I'm not sure if this is a general solution.