UNIX for Dummies Questions & Answers

Not certain the title is phrased correctly but here's the issue:

I need to scrape data out of a pdf file. I ftp the file from windows to unix(solaris 10) as a binary file. Open the file in vi and search for a string, Matrix, but no joy. Then I cat the same file, pipe it to grep 'Matrix' and the string is found.

Can anyone explain this behavior and, perhaps, how to get vi to act the same as grep. BTW I did not try awk, perl or sed; yet anyway.

Thanks,
rj

There are several possible reasons for this:

First, how did you issue the grep command? If it was "grep -i", then you switched on case INsensitivity, which is not so easily done in vi (certainly not with a command switch). Make sure you really search for the same string in both cases.

Second, The file might contain some (terminal-) control codes, which are "interpreted away" when you pipe it through cat.

Have a look at the unaltered file with "od -ax" to get even the non-displayable characters.

I hopoe this helps.

bakunin

Didn't use the "-i" switch with grep, nor the ":set ic" in vi but there are a bunch of non-printable characters in the file and maybe vi and cat/grep interpret them differently.

Thanks for the reply.

As i thought more about this i had some more ideas:

You might find the places you find with cat | grep in vi using the following formula (given, its cumbersome, but good enough for learning more about the problem):

To search for "abc" interspersed with an arbitrary number of code-characters you can search for:

Code:

/a[^b]*b[^c]c/

You search for the first character, then an arbitrary number of the NOTted following character, then the following character itself and so on. This way you should catch occurrences of "abc" even if they are drowned in control codes. The only prerequisite would be that the string is still on one line no newline character in between).

I hope this helps.

bakunin