Description
The basic idea is that we want to sort obfuscated data without deobfuscating and reobfuscating it. That is, our input is an obfuscated file, and we could compute the output by deobfuscating the input, sorting the deobfuscated data, and then reobfuscating the resulting output—except that we do not want to obfuscate or deobfuscate anything.
Write a C function frobcmp that takes two arguments a and b as input and returns an int result that is negative, zero, or positive depending on whether a is less than, equal to, or greater than b. Each argument is of type char const \*, and each points to an array of non-space bytes that is followed by space byte. Use standard byte-by-byte lexicographic comparison on the non-space bytes, in the style of the function, except that you should assume that both arrays are frobnicated, (i.e., trivally obfuscated via [memfrob](https://www.gnu.org/software/libc/manual/html_node/Obfuscating-Data.html)) and should return the equivalent of running memcmp on the corresponding unfrobnicated arrays. If one unfrobnicated array is a prefix of the other, then consider the shorter to be less than the longer. The space bytes are not considered to be part of either array, so they do not participate in the comparison.
For example, frobcmp (“\*{\_CIA\\030\\031 “, “\*\`\_GZY\\v “) should return a positive int because “\*{\_CIA\\030\\031” is “\\0Quick23” frobnicated and “\*\`\_GZY\\v” is “\\0Jumps!” frobnicated, and “\\0Quick23” is greater than “\\0Jumps!” in the ASCII collating sequence. As the example demonstrates, null bytes ‘\\0’ are allowed in the byte arrays and do contribute to the comparison.
Your implementation should not invoke memfrob, as that would mean that memory would temporarily contain a copy of the unfrobnicated data. Instead, it should look only at frobnicated bytes one at a time, and unfrobnicate them “by hand”, so to speak.
Use your C function to write a program sfrob that reads frobnicated text lines from standard input, and writes a sorted version to standard output in frobnicated form. Frobnicated text lines consist of a series of non-space bytes followed by a single space; the spaces represent newlines in the original text. Your program should do all the sorting work itself, by calling frobcmp. If standard input ends in a partial record that does not have a trailing space, your program should behave as if a space were appended to the input.
Use [\<stdio.h\>](http://www.opengroup.org/onlinepubs/9699919799/basedefs/stdio.h.html) functions to do I/O. Use [malloc](http://www.opengroup.org/onlinepubs/9699919799/functions/malloc.html), [realloc](http://www.opengroup.org/onlinepubs/9699919799/functions/realloc.html) and [free](http://www.opengroup.org/onlinepubs/9699919799/functions/free.html) to allocate enough storage to hold all the input, and use [qsort](http://www.opengroup.org/onlinepubs/9699919799/functions/qsort.html) to sort the data. Do not assume that the input file is not growing: some other process may be appending to it while you’re reading, and your program should continue to read until it reaches end of file. For example, your program should work on the file /proc/self/status, a “file” that is constantly mutating: it always appears to be of size 0 when you ls it, but it always contains nonempty contents if you read it. You should make sure your program works on empty files, as well as on files that are relatively large, such as /usr/local/cs/jdk\*/jre/lib/rt.jar on SEASnet.
If the program encounters an error of any kind (including input, output or memory allocation failures, it should report the error to stderr and exit with status 1; otherwise, the program should succeed and exit with status 0. The program need not report stderr output errors.
For example, the shell command:
printf ‘*~BO *{_CIA *hXE]D *LER #@_GZY #E\\OX #^BO #FKPS #NEM\4’ |
./sfrob |
od -ta
should output:
0000000 * h X E ] D sp * { _ C I A sp * ~
0000020 B O sp * L E R sp # N E M eot sp # @
0000040 _ G Z Y sp # F K P S sp # E \ O X
0000060 sp # ^ B O sp
0000066
because frobnicating sfrob’s input and then appending a trailing newline (because the last frobnicated byte is not a newline) yields:
^@The
^@Quick
^@Brown
^@fox
^Ijumps
^Iover
^Ithe
^Ilazy
^Idog.
where \^@ denotes a null byte ‘\\0’, and \^I denotes a tab byte ‘\\t’. Sorting this yields:
^@Brown
^@Quick
^@The
^@fox
^Idog.
^Ijumps
^Ilazy
^Iover
^Ithe
and frobnicating this yields the output shown above.
