UNIX for Dummies Questions & Answers

Many thanks for the example. Unfortunately, it doesn't work. It returns file2 unchanged. Sorry...

As for the files - the values of columns 2 and 3 in file1 may overlap, this is why I need to match the files according to column1 first. Also, the files are not sorted, but I can do that before running a script.

Many thanks once again!

Why don't you give a more realistic example of your input file and the desired output?

Quick hack at it:

Code:

 
#!/usr/bin/ksh
 
cat file1 | while read a1 a2 a3 a4
do
 cat file2 | while read b1 b2 b3
 do
   if (( $b1 == $a1 && $b2 >= $a2 && $b2 <= $a3 ))
   then
     . . .
   fi
 done
done

Generates new file on stdout.

I am really sorry for the confusion. Here I will try to make a new, more clear example:

file1

file2

I need to match column1 in file2 to column1 if file2 first, then see if the number in column2 file2 is between the values of column2 and 3 of file1; and if so, ass column4 of file1 as an additional column to file2.
For example, in this file2, line1 - the number in column2 of file2, line1 is 1235, which is in the range of 1234 and 1240 of line1, columns 2 and 3 of file1; but it is also in the range of 1234 and 4560 of line3, columns 2 and 3 of file1. Because the first column of line 1, file2 matches the first column of line1, file1, the added value should be abc (and not ogh of line3, because the value of column1, line3, file1 does not match the value of column1, line1, file2). The output file should be like this:

What I meant by overlaping values in columns 2 and 3 of file1, is that the region covered by the first line (from 1234 to 1240) is also the part of the region in line3 (from 1223 to 4560), but they've got different values in column1 (1 and 6).

The values in columns 1 and 2 of file1 are sorted ascending. The values in columns1 and 2 of file2 are also sorted ascending.

Hope this is more clear and apologies for the confusion.

Many thanks for spending time on this!!!!

---------- Post updated at 03:57 PM ---------- Previous update was at 03:50 PM ----------

Sorry, I've noticed the typo in this paragraph. Here is the correct one:

For example, in this file2, line1 - the number in column2 of file2, line1 is 1235, which is in the range of 1234 and 1240 of line1, columns 2 and 3 of file1; but it is also in the range of 1223 and 4560 of line3, columns 2 and 3 of file1. Because the first column of line 1, file2 matches the first column of line1, file1, the added value should be abc (and not ogh of line3, because the value of column1, line3, file1 does not match the value of column1, line1, file2). The output file should be like this

Well, my loops do not exploit the sorted order, but they are sufficient, simple and sufficient.

To exploit the sort, you need to know any sides that are many-to-, so after a match you read the many file first looking for more matches. If many to many, then you still cannot solve with a simple merge. That is why the join command needs flat files -- it does seeks back to first line of key value.

why the need for cat? And this solution is O(n^2). Worse when the files are big.

Well, sometimes N^2 is low enough, and it is simple. The cat's are just me, too lazy to move the input down to a '<file1'; I think left to right when scripting! If the cats add enough overhead to matter, there is no hope for the n^2 !

Now, who has a solution with associative array scripting? Will that work with many to many?

A SQL solution would be great, too, but text tools like better record boundaries, like one row per line.