Hi guys,

I have 2 text files with very similar contents, but in jumbled order. Does anyone know a good way to compare the lines of one to (all) the lines in the other?

eg.
File1 contains:
cat
dog
bird

File2 contains:
dog
bird
cat

Because diff compares these 2 files line by line it sees them as different - this is true, but I only want to see words that are missing from the document altogether.

Thanks in advance!

can you sort them then compare them ?

edit: else grep -f mite work ? ... or something like

Thanks schneidz!

That finally got my brain off the diff command.
Looks pretty simple and sweet, I'll give it a shot.

Note that it's not generally a good idea to read lines of input from a file or command with a for loop. You should generally use a while+read loop instead.

http://mywiki.wooledge.org/DontReadLinesWithFor
http://mywiki.wooledge.org/BashFAQ/001

Although in this particular case the expansion splitting the file into individual words results in the desired behavior. If the file were very large, however, it could possibly overwhelm the capacity of the terminal, as the whole list gets expanded before the for loop is run.


As mentioned, grep can also be used to test one file against another (on a per-line basis). This command prints every line in file2 that does not exist in file1:

Just run the command again with the files reversed to get all the unique lines in file1.

Thanks David,

I ended up using that exact grep command yesterday, but the files I was using seemed to be too large for it to handle. An strace on the process showed no activity at all. When I split the files to make them smaller it seemed to work well, and give me exactly the type of output I'm after. It seems to be the best option for now, as the very large files were just a one off - all the rest are smaller.

Thanks again.