hello

I have a directory containing mostly pdfs which I have gathered from all over the internet. I now want to sort them into directories in a more structured mannar based on their contents. To do this, I am running pdftotext to preview the contents then moving them to another dir based on what I read.

Now, I am particularly lazy and am trying to work out a fast way to repeat this process for different files. (Even though the length of time I have spent trying to work this out is longer that it would have taken me to do it anyway...)

The main aim is to take the second line and replace the first argument with a different file name.

Much reading as lead me to this route:
However, this has 2 flaws.
1) I don't want to have to type the old argument
2) I don't really want to have to type all of the new argument either. Bash's Tab completion doesn't seem to work at the end of a string like that.

The first problem can be partly overcome with:
Which will yank the first argument of the previous command. It doesn't solve the second one though.

This seems a very indirect route to perform this substitution. Is there a better solution? Perhaps the s/ operator can take a regexp that refers to the argument rather than a string?

I don't have pdf2txt installed, so I tried pdf2dsc which produces a postscript document containing structure information.
ex:
%!PS-Adobe-3.0
%%Title: (LinuxSysCalls.fm)

So a sed command '/Title/s/.*(\(.*\)\.fm)/\1/' could be used to extract the title.
You could put a command like this in a bash loop for every pdf file.

I just tried this line:
pdf2dsc LexAnalysis.pdf /dev/stdout | head | sed -n '/Title/s/.*(\(.*\).fm)/\1/p'
LexAnalysis

This isn't the best example since in this case, the Title and Filename match.
Opps! The .fm isn't in another pdf document I tested, so another sed command may be needed. And some have spaces before or after and in the title.

Here is a sed script that handles these three cases.
Often, you will run into cases you haven't considered, so using a sed script allows you to make additions later without changing your main bash script.

Sorry, accidentally double posted when the wireless started acting up.

Thanks for your reply, but I think you've missed the point slightly. My question is more general. The fact that I was going through PDF documents is almost beside the point but was important to provide context. (although you might have pdftotext installed, note that it is written in full, part of xpdf I think)

Say, for example, that I had just typed a long string at the bash prompt with lots of arguments and filenames, pipes and whatnot:

So, a fairly general command but something I'm not wanting to type more than once. If I wanted to do that to a lot of files in a completly automatic way it could be achieved easily with a for loop:

which would result in only typing it once.

In this case, I needed to intervene in the middle to decide what I was going to do with the files so a loop wouldn't be that helpful. I was tring to recall that line interactively from bash's history and alter one argument. I could do this by hitting the up arrow a couple of times and then moving the cursor with the arrow keys. Seeing as there are some interesting ways to yank arguments from previous lines in the history, I thought there might be a means to easily substitute the nth argument with a different value. My s/../../ substitution in my first post was part of a command to bash's history rather than part of a sed script.

In the absence of a quick and dirty method to substitute an individual argument, the method I'd probably use in future would be to write a sing line script to wrap the whole line into a single command and pass the argument to the script in place of the filename.

You don't need to type the rest of the line, just the part that changes.
!!-2:s/afilename/anewfilename/

I thought that you wanted to extract the title of the pdf from the file itself and use that as the name. That would allow you to perform this operation automatically.