104.7 Lesson 1

Temporary files are files used by programs to store data that are only needed for a short time. This can be the data of running processes, crash logs, scratch files from an autosave, intermediary files used during a file conversion, cache files and such.

You can use the parameters below to search for files with specific attributes, like ones that are writable by your user, have a specific set of permissions or are of a certain size:

You can add a - before NNNN to check for files that have at least the permission specified. For example, -perm -644 would match files that have at least 644 (rw-r—​r--) permissions. This includes a file with 664 (rw-rw-r--) or even 775 (rwxrwx-r-x).

Again, you can add the + or - prefixes (here meaning bigger than and smaller than) to search for relative sizes. For example, -size -10M will match any file less than 10 MiB in size.

For example, to search for files under your home directory which contain the case insensitive pattern report in any part of the name, have 0644 permissions, have been accessed 10 days ago and which size is at least 1 Mib, you could use

Besides searching for attributes you can also perform searches by time, finding files that were accessed, had their attributes changed or were modified during a specific period of time. The parameters are:

For -cmin N and -ctime N, any attribute change will cause a match, including a change in permissions, reading or writing to the file. This makes these parameters especially powerful, since practically any operation involving the file will trigger a match.

The following example would match any file in the current directory which has been modified less than 24 hours ago and is bigger than 100 MiB:

locate and updatedb are commands that can be used to quickly find a file matching a given pattern on a Linux system. But unlike find, locate will not search the filesystem for the pattern: instead, it looks it up on a database built by running the updatedb command. This gives you very quick results, but they may be imprecise depending on when the database was last updated.

The simplest way to use locate is to just give it a pattern to search for. For example, to find every JPEG image on your system, you would use locate jpg. The list of results can be quite lengthy, but should look like this:

When asked for the jpg pattern, locate will show anything that contains this pattern, no matter what comes before or after it. You can see an example of this in the file jpg_specs.doc in the listing above: it contains the pattern, but the extension is not jpg.

By default the pattern is case sensitive. This means that files containing .JPG would not be shown since the pattern is in lowercase. To avoid this, pass the -i parameter to locate. Repeating our previous example:

Notice that the file Mate1_old.JPG, in bold above, was not present in the previous listing.

You can pass multiple patterns to locate, just separate them with spaces. The example below would do a case-insensitive search for any files matching the zip and jpg patterns:

When using multiple patterns, you can request locate to show only files that match all of them. This is done with the -A option. The following example would show any file matching the .jpg and the .zip patterns:

If you wish to count the number of files that match a given pattern instead of showing their full path you can use the -c option. For example, to count the number of .jpg files on a system:

One problem with locate is that it only shows entries present in the database generated by updatedb (located in /var/lib/mlocate.db). If the database is outdated, the output could show files that have been deleted since the last time it was updated. One way to avoid this is to add the -e parameter, which will make it check to see if the file still exists before showing it on the output.

Of course, this will not solve the problem of files created after the last database update not showing up. For this you will have to update the database with the command updatedb. How long this will take will depend on the amount of files of your disk.

which is a very useful command that shows the full path to an executable. For example, if you want to locate the executable for bash, you could use:

If the -a option is added the command will show all pathnames that match the executable. Observe the difference:

type is a similar command which will show information about a binary, including where it is located and its type. Just use type followed by the command name:

The -a parameter works in the same way as in which, showing all pathnames that match the executable. Like so:

And the -t parameter will show the file type of the command which can be alias, keyword, function, builtin or file. For example:

The command whereis is more versatile and besides binaries can also be used to show the location of man pages or even source code for a program (if available in your system). Just type whereis followed by the binary name:

The results above include binaries (/usr/bin/locate) and compressed manual pages (/usr/share/man/man1/locate.1.gz).

You can quickly filter the results using commandline switches like -b, which will limit them to only the binaries, -m, which will limit them to only man pages, or -s, which will limit them to only the source code. Repeating the example above, you would get: