todo

executables like /bin/true for every file name

https://github.com/ziglang/zig/issues/4812

infinite bell nestings

folder with symlink to every single object created by angryfiles

every single --option and -o as every type of fs object

max length /../something/../something/../

zfs reservers .zfs/ for snapshots etc. so... need to make all 4 char file names:)

https://github.com/openzfs/zfs/issues/13980

all paris of of complementary chars () {} [] <> /\ +- -+ "" '' `` (and all of these quoted)

all 1 pixel images

byte strings that are not valid file names

add unique key to each file content in tree to make app prove it read each path

getconf ARG_MAX

angry file name oracle

angryos
    every file name is obfusciated
        makes file names and the os orthoginal

accept a root_dir to iterate through in any order, using those files as template= args for every time angryfiles needs a template file

'un-tab completable files"
'In honor of @Acosta getting embarrassed on live television, not to mention how disrespectful he is, I want everybody to see this unearthed video of him doing #Arsenio back in the day! ‬Totally u...-9cfsjdg2zyw11.mp4'
fuck, even that is a (copy/paste) bug, the real file name is:
b'/home/cfg/_myapps/angryfiles/\xe2\x80\xaaIn honor of @Acosta getting embarrassed on live television, not to mention how disrespectful he is, I want everybody to see this unearthed video of him doing #Arsenio back in the day! \xe2\x80\xac \xe2\x80\xaaTotally u...-9cfsjdg2zyw11.mp4'


"bloom" mode, each byte folder, nested, nested, nested ...
    for zfstester
    same idea, but with sequential numbering of folders


every unicode .dotfile

executable version of every single file, execution prints the contents or the hash
    quines

files named like args -- -0 -h --help

write inode # to each file

DT_UNKNOWN https://github.com/ziglang/zig/issues/5123

all variation of perms bits, including +i and xattr

valid binaries (/bin/true) with invalid unicode names

all valid file names counting up in binary

dirs with every # of files up to max argc (minimal_cee testing for args overflows after 255)

all files in their msgpacked name
    file names that are indistuningishable from properly messagepacked data
        possible to construct a valid dir what's also valid msgpack when concatinated?
            ditto for \0 joined names, and \n too

all files with the contents being the sha3 of the N byte file name

angryfiles for every type of dent. refactor to just deal with dent types, should be able to test some arb program with files names of every dent type, and every byte name(s)

a deliberate FS BLACK HOLE.. no iterating out of it, it always goes deeper

/dev/byte/0-255 that can be symlinked to

symlink_to_bell -> /dev/bell (cat /dev/bell -> BELL)



file names that draw ascii color pictures when written to the terminal

stuff that breaks recursion depth maxes on different programming language implementations

add max_depth, min_depth, min/max everything, file size, inode#, path element length, least perms, most perms

all other encodings of each of the bytes and unicode's

all unicode points with unicodehaz

+i and other attributes, bad permissions no +x on dirs etc

malicious file names from mitre CVE's

folders for deduplication tests, where a/ and b/ have >1 elements each and b has one or more duplicates of files in a, where not all files in a are in b

FUSE files that do funky things... like have the max file size for different old FS's, or have max/min dates for all dates, or minimal or maximal perms, or change in size when read

paths that are unicode, and then the same path, but in int codepoints

files that look like escaped bytes \x5d

files that are all beeps

files with dangerious names "rm -rf ."

file one
file\n two
(# [[ $'\0' == '' ]] && echo meow) # bash cant parse NULL

files that look like results from "file": PAR.txt: Par archive data

files that start and end with 'A' '''A''' "A" """A""" ""A""

files named --help/-help/-h (make a option to gather these from a comand)

names with unclosed quotes ' " ` and paren ( [ { and !'s

all whitespace paths

huge selection of url's all unrolled into the filesystem path

lots of home/~/home/user~~/~/../~./.~/'"`s

byte strings that match surrogate encoded paths

max nested /~/~/~/~/ for every single byte file name

every length file starting with ~

various lath looking things with files literally named "~"
bup fs to demonstrate why getdents64 kills find

big example hash tree filesystems.

files named "None" and "NOTHING" (etc) Zero NULL Nul NUL NaN nan bla bla

files names the uescaped names of the bytes: f = Path(os.fsdecode(bytes([254])))

nice help with cmdline examples ready to use to break stuff

write those unicode test files

unicode emoji's

term color escape codes

bytes root's to start with to attempt to feed kcl dirops iterate a bytes root from command line

http://sourceware-org.1504.n7.nabble.com/BZ-6895-Describe-DT-LNK-td5495.html


 $ ls /mnt/t420s_256GB_samsung_ssd_S2R5NX0J707260P/.iridb/database.local/iri_index/e/1/3/e132bae070e951b89305b6a943e94b0d82be4a8c/tags
total 8.0K
drwxr-xr-x 2 user user 4.0K May 19 12:05 .
drwxr-xr-x 4 user user 4.0K May 19 12:05 ..
lrwxrwxrwx 1 root root   40 May 19 12:05 9a3e67212ba7d06bafdc8c75df84684a2ab26a68 -> 9a3e67212ba7d06bafdc8c75df84684a2ab26a68


all hash filenames for every combination ob bytes output, so sha1, sha256, sha3 for all combination sof 1 byte, and 2byte etc

attrs
xattrs (treat like files, run through same tests)


symlinks to char device
symlinks to block device
block and char devices with bytes names

filenames that look like the output of ls (symlinks -> etc) and other command line apps

data classes, to test things like "file" with valid data

broken symlinks where the target folder exists but the file does not (so readlink -f happily resolves it anyway)

relative symlinks with and without path compoinents

all length newline filenames and dirnames (for both kinds of newline)

all shaX prefixes (iterate /dev/urnadom to get N files per folder where N is the nesting level)

maximal length paths

maximal depth trees

base128/192

all whitesapce filenames of every length

unicode snowman

iri's and uri's with testcases as filenames

ln -P

symlink->hardlink->file

symlinks to different mountpoint (like the iridb abspath vs realpath problem) (and the non-shortest link dest example)

files with all n byte combinations


files of different sizes, all 1-1k, then 1k-1M in 1k inc etc..

add command line option to skip all the asserts for speed

all 1 byte symlinks to all 1 byte dirs
all 1 byte symlinks to -> all 1 byte symlinks to all 1 byte dirs

maximally nested symlink tree byte->byte->byte->byte

maximally nested dir tree (to max out path length)
randomly generated tree

1 byte symlink to symlink to file

[b-a] to see if glob.glob() is being used without glob.escape()


2015-11-20_03:21:58 root@xeonbox0 ~/cfg/_myapps $ ls _badyt_copy
total 222M
drwxr-xr-x  2 user user 4.0K Nov  5 07:29 .
drwxrwxr-x 40 user user 128K Nov 17 02:44 ..
-rw-r--r--  1 user user 1.7K Nov  5 06:26 StormCloudsGathering__StormCloudsGathering__20130911__9_11 - Echoes of Darkness [Mini-documentary]__youtube__5PY_qM28rnA.description
-rw-------  1 user user  28K Nov  5 06:26 StormCloudsGathering__StormCloudsGathering__20130911__9_11 - Echoes of Darkness [Mini-documentary]__youtube__5PY_qM28rnA.info.json
-rw-r--r--  1 user user  12K Nov  5 06:26 StormCloudsGathering__StormCloudsGathering__NA__9_11 - Echoes of Darkness [Mini-documentary]__youtube__5PY_qM28rnA.info.json
-rw-r--r--  1 user user 222M Nov  5 06:26 StormCloudsGathering__StormCloudsGathering__NA__9_11 - Echoes of Darkness [Mini-documentary]__youtube__5PY_qM28rnA.mp4
-rw-r--r--  1 user user 1.7K Nov  5 06:26 StormCloudsGathering__StormCloudsGathering__NA__9_11 - Echoes of Darkness [Mini-documentary]__youtube__5PY_qM28rnA.mp4.description






2015-12-11_20:54:53 root@xeonbox0 /home/user/cfg/input $ ll
total 24
drwxr-xr-x   5 user user 4096 Dec  9 20:30 .
drwxrwxr-x 105 user user 4096 Dec  9 19:38 ..
drwxr-xr-x   2 user user 4096 Dec 11 04:30 alphagrip
drwxrwxr-x   4 user user 4096 Dec  9 22:15 keyboard
drwxr-xr-x   2 user user 4096 Dec 10 23:47 mouse
-rwxr-xr-x   1 user user   25 Dec  9 20:08 xinput_list
2015-12-11_20:54:55 root@xeonbox0 /home/user/cfg/input $ ln -s ../linux/eudev/ udev
2015-12-11_20:56:54 root@xeonbox0 /home/user/cfg/input $ ll
total 24
drwxr-xr-x   5 user user 4096 Dec 11 20:56 .
drwxrwxr-x 105 user user 4096 Dec  9 19:38 ..
drwxr-xr-x   2 user user 4096 Dec 11 04:30 alphagrip
drwxrwxr-x   4 user user 4096 Dec  9 22:15 keyboard
drwxr-xr-x   2 user user 4096 Dec 10 23:47 mouse
lrwxrwxrwx   1 root root   15 Dec 11 20:56 udev -> ../linux/eudev/
-rwxr-xr-x   1 user user   25 Dec  9 20:08 xinput_list
2015-12-11_20:56:56 root@xeonbox0 /home/user/cfg/input $ rm udev/
rm: cannot remove ‘udev/’: Is a directory
2015-12-11_20:57:20 root@xeonbox0 /home/user/cfg/input $ rm udev
rm: remove symbolic link ‘udev’? y
2015-12-11_20:57:29 root@xeonbox0 /home/user/cfg/input $ ln -s ../linux/eudev udev
2015-12-11_20:57:35 root@xeonbox0 /home/user/cfg/input $ ll
total 24
drwxr-xr-x   5 user user 4096 Dec 11 20:57 .
drwxrwxr-x 105 user user 4096 Dec  9 19:38 ..
drwxr-xr-x   2 user user 4096 Dec 11 04:30 alphagrip
drwxrwxr-x   4 user user 4096 Dec  9 22:15 keyboard
drwxr-xr-x   2 user user 4096 Dec 10 23:47 mouse
lrwxrwxrwx   1 root root   14 Dec 11 20:57 udev -> ../linux/eudev
-rwxr-xr-x   1 user user   25 Dec  9 20:08 xinput_list
2015-12-11_20:57:36 root@xeonbox0 /home/user/cfg/input $

 # TODO
    # folders with #'s of files in them 2**0, 2**1, 2**2, files/dirs etc

    Windows-1252
        http://www.i18nqa.com/debug/utf8-debug.html

    make fifos that look like files that when opened feed ugly data (attach to other fuzzers)
    pure unicode filenames
    control char file names (terminal color codes etc)

    fifos plugged into /dev/zero
    fifos plugged into /dev/null
    symlinks to fifos
    symlinks to block devices
    symlinks to char devices
    device nodes
    every major,minor device
    files with weird attributes
    1 byte files
    2 byte files
    symlinks to /
    files with odd mtimes and atimes
    files that start with - and --
    files that are all unprintable chars
    files named exit/quit/exit()/quit()
    crazy multiline unicode, odd unicode space mix
    self ref symlinks
    self ref links
    all lengths of null
    hard links
    files that disapperar when you access them (inotify triggers)
    files containing unicode, ascii, binary
    network sockets
    deeply nested trees, select a level for example 3 == 0/0/0 if it's indexed by hash filenames
    expanmd to benchmarking, be able to test what's fastest to copy/count in a dir tree nesting level n=3, n=4, n=5 for hash index
    hardlink for dir's (ln -F)
    hardlink to symlink (ln -P)
    device files
    make a generator that does not write files, it just outputs the filenames for use with other stuff that should accept arb byte2
    make files for every magic "file" byte signature to "fool" the "file" command
    max depth dir tree
    ext2/ext3/ext4 filesystem limits
    symlink -> symlink -> symlink ... -> [file,dir.broken_symlink,fifo,device,socket] for "get_symlink_level(n)" testing
    broken symlinks to bytes destinations
    bytes broken symlinks to unicode destinations
    bytes broken symlinks to invalid unicode destinations
    bytes broken symlinks to bytes destinations

    simple tests:
        single folder
            empty single char folder
            single char folder
                empty file
            single char folder
                2x empty files

all byte folders nested in byte order

files with weird flags/dates/metadata/perms
symlinks back to /

import keyword
keyword.kwlist

all file sizes up to x
files guranteed to take up more than one block
all attributes

all numeric file names
all caps file names
all lowercase file names
all symbol file names