Magic Banana woke me up with: "Input excerpt and expected output..."
OK. The beginning-5-col-file shows all five tab-separated fields.
The output-2-column-file actually has an added third column listing the
proper IPv4 addresses that resolve to the hostnames in the first column.
You will see that a couple of the server owners have active imaginations:
3rd row: 53.198.27.67 to 198.27.67.35
14th row: 10.181.14.194 to 181.14.194.10
Took a while to solve those two.
I initially converted the five-column file to two columns in Leafpad by
converting the
tabs to dots and then converting the .com. to .com[space], then .net. ...
.br. ... .cn. ...
and so on, with a couple of unintended consequences ... not for a 5,000 row
file. I also
had to correct the hostnames of two entries that had been truncated at their
beginnings
by reference to the original four-month list. These hostnames all appear in
online
Webalizer- and spam-statistics files.
I noticed that when the nslookup fails to return the correct hostname, the
reverse order
can be grep'ed from the four octets at the beginning of the arpa-format
reply. That would
have gotten the right answers in all but two of this list of hostnames.