I think Colin hit the nail on the head with this: I think what Art is saying is that all the column names are there but many of the fields have no data for many rows. The problem is how to automate so my usual fix of manually adding enough dummy data in the first several empty fields isn't going to help.
There is no automated solution for those problems if fields don't have all the data and are also missing separators that could give enough implied order. Suppose you have a file that purports to represent a table of schools, with the name of the school, the first and last name of the principal and the name of the town, all separated by tab characters. The first row has the names of the fields: "School"<tab>"Fname"<tab>"Lname"<tab>"Town" That makes sense if all of the fields are in place: "Wardley"<tab>"Henry"<tab>"Brown"<tab>"Middletown" It even makes sense if separators are in place to imply NULL values: "Wardley"<tab><tab>"Brown"<tab>"Middletown" But what if a line consists of: "Harrison" ... is that the name of the school, a first name, a last name or a town? or a line that consists of... "Winston"<tab>"Rush"<tab>"Reston" Is that the name of the school followed by a last name followed by a town? Or is it some other combination like the principal Winston Rush of an unnamed school in the town of Reston?
|