In this MP, you will write code with less guidance than you likely had in the past. In particular, you will
The specific tasks of this MP involve working with PNG files, a common image file format. PNG files, like many other file formats, are not text-based: unlike JSON, XML, or source code you cannot simply open them in a text editor to explore their contents. You will need to consult the PNG documentation to understand the meaning of the bytes in a PNG file.
PNG is a chunk-based file format. It is an open format, completely specified in a public royalty-free document, https://www.w3.org/TR/png/. The goal of a specification like that is to leave no ambiguity in how to create a file or interpret one you receive, which can lead to them being rather dense and difficult to read. Various parties sometimes make less-complete but easier-to-read summaries, such as https://en.wikipedia.org/wiki/PNG, which can be an easier place for someone new to the format to start.
You will need to consult either the official spec of a simpler summary to complete this MP: we will not summarize or teach you that content directly.
| Official PNG specification | https://www.w3.org/TR/png/ |
| Simpler incomplete summary | https://en.wikipedia.org/wiki/PNG |
In 1985 Electronic Arts introduced a concept for storing media files as a sequence of packets or chunks of binary data, each preceded by a short label stating what type of data the chunk contained. Their Interchange File Format was adopted with slight tweaks by Microsoft as the Resource Interchange File Format (RIFF) on top of which Windows 3.1’s video, audio, and user interface element file formats were built. Since then that model has come to dominate the media file formats: many formats have the general form of a sequence of chunks, each with a header that specifies its type and size, coupled with specific rules for interpreting the bytes of certain chunk types.
One benefit of a chunk-based design is that basic applications can skip chunks they don’t understand and still get partial functionality. Additionally, updates to the format can add new features in new chunk types without breaking any existing files or applications.The initial files we provide have very little structure in them, not much more than picking out the command line arguments. We have a general guide on coding from blank files that might be helpful to some of you.
A learning goal of this MP is working without us giving you a design or guide to getting started on this MP specifically. Please don’t ask course staff to bypass that goal by telling you how to design your code.
what comes next?and do that.
mp3.zip contains header files, testers, and a Makefile. You will add to the following files:
pngchunklist.cextractuiucchunk.cinsertuiucchunk.cpnglib.c and pnglib.h (optional, provided to allow sharing code between the other three files)The first three of these will be command-line application; we provide the basic command-line argument count checking for you and you write the rest. The pnglib files will not be directly tested, but can be used to keep your code organized.
pngchunklistThis application lists the type and size of each chunk in the PNG file in the order that they appear in the file. The format of this should be one printed-out line per chunk, each with the type, a space, and the length.
You can implement it in whatever way you wish, provided that you write all the code yourself without copying from any AI system or third party. We recommend using the following libraries:
fopenfseekfreadntohlprintf or an alternative display function. Remember that if you are printing a string it needs to be null-terminated; for example a 4-byte string is stored in 5 bytes, with the last one being 0.Coding this application shouldn’t require malloc and should fseek past most of the bytes in the file.
For extra credit, add the following features to this application:
IHDR or IEND chunks or has them in the wrong position in the file, print an error message to stderr and exit with code 4width or height, a space, and then the corresponding value. Print these two lines in the same order they appear within the PNG file.Extra credit is graded all-or-nothing and is only awarded on an MP if all normal credit is, so focus on passing the regular tests first.
extractuiucchunkThis application looks for a special chunk with type uiuc; if found, it copies the data inside that chunk into a separate file. If there is more than one such chunk, it copies the first and then exits with code 0. If there is no such chunk, prints an error message to stderr and exists with code 4.
You can implement it in whatever way you wish, provided that you write all the code yourself without copying from any AI system or third party. We recommend sharing significant code from pngchunklist (ideally by putting the shared code in library functions in pnglib.c and pnglib.h), and additionally using the following libraries:
insertuiucchunkMakes a copy of a PNG file with a new uiuc chunk. If the input PNG file already has a uiuc chunk, replace it. If not, add it between two existing chunks.
You can implement it in whatever way you wish, provided that you write all the code yourself without copying from any AI system or third party. We recommend sharing significant code with the other two applications (ideally by putting the shared code in library functions in pnglib.c and pnglib.h).
You’ll also need to generate a crc32 checksum; this is a computation based in coding theory which most CS programs either don’t teach at all or only teach it in an elective course. Perhaps that is why the code you need is given in the PNG specification in Appendix D. Note there are three functions in that appendix: the first to be run just once, the second to be given bytes of a chunk (either all at once or a few bytes at a time, it makes no difference), and the third a reference showing how you’d use the second if you have the entire chunk in one array of unsigned characters. Also note that the appendix is somewhat old and assumes that long means a 32-bit integer; currently int (or int32_t from #include <stdint.h>) is the type with that size.
insertuiucchunkPut a secret payload inside a PNG of your choice and share it on the CampusWire post Post Your MP3 Images Here!
and its accompanying Drive folder. This is an opportunity to share a bit about yourself and your interests; please keep it (both the image and its hidden contents) inoffensive.
Submit on the submission site. Submit pngchunklist.c, extractuiucchunk.c, and insertuiucchunk.c; if you changed them, also submit pnglib.c and pnglib.h. Do not change or submit the Makefile, tests, or example images.
Automated tests are provided in the make test target. These use specific images in the provided img/ directory to run. On the testing server, a different set of images will be used; that should have no impact on your grading if you’ve parsed the PNG files, but will reduce your score if you’ve hard coded solutions to specific files or test cases.
The three programs are weighted as follows:
Valgrind will also be run using
valgrind --leak-check=full --show-leak-kinds=all ./yourprogram whatever other arguments you wantfor each of the three programs; if it gives errors other points will be reduced by 10–25%, depending on how many programs have memory errors.
If you fail a test case, you should look inside tests.py and search for the message associated with the string, then look for the command that was run to create that error message (which are strings that begin ./ and the name of one of your three programs) and try running that yourself. Note that some tests depend on the outcome of previous instructions; for example, one of the commands (line 185) is
./extractuiucchunk tmp2 tmp4where tmp2 was previously created (on line 143) by
./insertuiucchunk img/onered.png tmp tmp2and tmp was previously created (on line 140) by
echo "this is a test" >> tmpecho and ./insertuiucchunk tests first, then run the extract test that failed in the debugger.