Safe Haskell | Safe |
---|
Du
Description
Third assignment for IB016, semester spring 2019
Implementing du
In this assignment you will implement a simplified version of the Unix
utility du
. This utility can be used to detect filesystem usage by
files and directories. You should create a standalone executable module.
This time, only the outer specification in given, you have to do the
functional decomposition yourself.
Note: Doing the functional decomposition is actually a part of your task. There is a separate submission folder (odevzdávárna) to which you are required to upload a file with function names and type signatures. See the interactive outline for details.
du usage / commandline options
Your program usage should be:
./du [options] [files-or-directories]
You should implement following options:
-a, --all write counts for all files, not just directories -c, --total produce a grand total -d, --max-depth=N print the total for a directory (or file, with --all) only if it is N or fewer levels below the command line argument; --max-depth=0 is the same as --summarize -h, --human-readable print sizes in human readable format (e.g., 1.0KiB 234MiB 2GiB) --si like -h, but use powers of 1000 not 1024 (e.g., 1.0kB, 245MB, 2.1GB) -s, --summarize display only a total for each argument --help display help and exit
You don't need to handle invalid options, you can ignore them.
You may assume that the short -d
option is always immediately followed by its
argument and the long --depth
option is separated from its argument by a
single equal sign ('='
). E.g. -d0 -d10 --max-depth=0 --max-depth=10
.
Combination of -s
and -a
is not valid, and need not be handled, the same
holds for --max-depth=0 and --all.
du behaviour
Then du
is run without any options, it prints sizes of all its commandline
arguments: for files the size is printed directly, for directories their
size is summarized recursively. By default, files inside directories are
not printed. If any options are given, they precede all files and
directories (you can take that for granted and don't have to check).
If no files or directories are given, du
should work with
current working directory (./).
By default, sizes are printed in kibibytes without the unit (1 KiB = 1024 B). With --human-readable (or -h) sizes are printed with an appropriate unit (using binary prefixes https://en.wikipedia.org/wiki/Binary_prefix) such that the value is between 1 and 1023. If --si is given, sizes are handled similarly but using 1000-based SI prefixes (and value should be between 1 and 999). If the value is less then 10, one decimal place is displayed, otherwise there are no decimals (though rounding can be arbitrary).
All the other command line options should be handled according to usage given above. If --help is given, a short help summarizing options and usage should be displayed and all other options should be ignored.
Bonus (+3 points)
Optionally, as a bonus, you can also implement:
--exclude=PATTERN exclude files and directories that match PATTERN
PATTERN
is a shell pattern (not a regular expression): ?
matches any one
character and *
matches any string (including an empty one). For example,
./du --exclude="*.swp
" does not count any files that end in (or are in a
subdirectory ending in) .swp
. (You may use module
System.FilePath.Glob.
It supports somewhat richer patterns, but so does the actual du
.)
The option may be given more than once. Files are excluded if they match any of the patterns. Command line arguments are subject to this option.
Further notes
- In basic execution (without
-s
), subdirectories are printed. - In case of error (such as permission error or directory vanishing before
it can be explored) program should not stop but print an error message
(on
stderr
), you should handle onlyIOException
and you can use that it is an instance ofShow
. - You can ignore anything that is neither file, nor directory (such as devices, symlinks, pipes,…).
- You should NOT ignore hidden files (on Unix beginning with
.
). - The original Linux
du
is calculating file sizes based on disk allocation, sizes reported byhFileSize
can differ (that is OK). - You can assume no files or directories are named as valid options (e.g. there is no file named '--all').
Module and package constraints
You can use any modules from any packages, but all used packaged (except base) must be noted in the header of this file next to your name and UČO.
Tips and tricks
- For the recursive traversal, functions from
System.Directory
may be handy. - File size can be obtained by calling
hFileSize
fromSystem.IO
or alternatively byfileSize
fromSystem.Posix.Files
. However, the latter case is not multiplatform). - You can use
Text.Printf
for formating. - Think twice before you start writing the code. Doing a proper functional decomposition will save you a lot of work/refactoring. Think of the functions you'll need, write their type signatures, submit them, and only then start programming.
- You may use monoids for command line arguments processing, but you don't have to.
- There are many useful general-purpose packages on Hackage, see for example the MissingH package. If there is something reasonably common you want, try to search Hackage first (but do not install everything just for a small function).
Examples
Order of files and directories on same level in hierarchy is not relevant and can differ on your system. Also the output in case of error need not match literally.
$ ./du --help usage: du [options] [files] -a, --all write counts for all files, not just directories --si like -h, but use powers of 1000 not 1024 -h, --human-readable print sizes in human readable format (e.g., 1K 234M 2G) -s, --summarize display only a total for each argument -d, --max-depth print the total for a directory (or file, with --all) only if it is N or fewer levels below the command line argument; --max-depth=0 is the same as --summarize -c, --total produce a grand total --help display this help and exit $ mkdir test; cd test $ mkdir -p first/second third $ dd if=/dev/zero of=a bs=1024 count=100 &> /dev/null $ dd if=/dev/zero of=first/b bs=1024 count=200 &> /dev/null $ dd if=/dev/zero of=first/c bs=1024 count=300 &> /dev/null $ dd if=/dev/zero of=first/second/d bs=1024 count=1024 &> /dev/null $ ../du 0 ./third 1024 ./first/second 1524 ./first 1624 . $ ../du first third 1024 first/second 1524 first 0 third $ ../du -c first third 1024 first/second 1524 first 0 third 1524 total $ ../du first 1024 first/second 1524 first $ ../du -s first 1524 first $ ../du --summarize first 1524 first $ ../du -h first 1.0 MiB first/second 1.4 MiB first $ ../du --si first 1.1 MB first/second 1.5 MB first $ ../du -h -s -c first a 1.4 MiB first 100 KiB a 1.5 MiB total $ ../du -a -h first 200 KiB first/b 300 KiB first/c 1.0 MiB first/second/d 1.0 MiB first/second 1.4 MiB first $ mkdir fourth && chmod -r fourth $ ../du fourth first error: fourth: getDirectoryContents: permission denied (Permission denied) 1024 first/second 1524 first/ $ ../du fifth first error: fifth: openFile: does not exist (No such file or directory) 1024 first/second 1524 first $ ../du -d1 error: ./fourth: getDirectoryContents: permission denied (Permission denied) 0 ./third 1524 ./first 1624 . $ ../du --max-depth=1 --human-readable error: ./fourth: getDirectoryContents: permission denied (Permission denied) 0.0 B ./third 1.4 MiB ./first 1.5 MiB $ ../du -a --exclude="*th*" --exclude="?eco??" 100 ./a 300 ./first/c 200 ./first/b 500 ./first 600 .