The goal: create several versions of a process that updates and saves a binary file as a new file.
The Setup
This project will be done in 4 parts. To keep them separate, I implemented a factory pattern so
that you (and the autograder) can test each copying method separately. It will look like this:
FileModifierFactory: creates
1. Part1SimpleFileModifier: fill in during part 1
2. Part2MultiProcessModifier: fill in during part 2
3. Part3ThreadedModifier: fill in during part 3
4. Part4SocketModifier: fill in during part 4
You will be given (and not need to modify):
1. main.cpp. Launches the appropriate test based on the arguments
2. Util.cpp/h. Includes some useful attributes.
3. FileModifierFactory.cpp & .h. These build the proper PartXModifier based on
the argument.
4. PipeCopier.cpp & .h. Helps you with the pipe for part 2.
While each part will be tested separately, you are encouraged to reuse code as much of it will
be useful for multiple parts.
The File
The file you are to read, modify, and save is a binary file that contains a sales list. A binary file is
a non-text file, meaning some things (like numbers) aren’t stored as digits but as the ints/floats
you use as variables. The name of the files to read and write will be in Util.h.
The file will be structured like this:
Field
Size
HEADER
NumEntries
4 bytes
Type Purpose
Integer Tells you how many
entries you need to
read
Timestamp (#
seconds since
1/1/1970)
Item’s code
Name of the item
sold
ENTRY (repeated
NumEntries times)
Date/Time sizeof(time_t) Time
Item ID
Item Name Sizeof(int)
50 bytes Integer
char*Item Quantity
Item Price
Sizeof(int)
Sizeof(float)
Integer
Float
Number sold
Price of the
products
What You’ll Do
In each part, the goal is to copy the file, adding two additional sales:
1. The Sobell book:
a. Time 1612195200 (2/1/2020 4 PM GMT)
b. ID 4636152
c. Name “A Programming Guide to Linux Commands, Editors, and Shell
Programming by Sobell” [warning: this is more than 49 characters, so you have
to truncate it—I say 49 because you need a null terminator]
d. Quantity: 70
e. Price: 70.99
2. The Advanced Programming book
a. Time 1613412000 (2/15/2020 6 PM GMT)
b. ID 6530927
c. Name “Advanced Programming in the UNIX Environment by Stevens and Rago”
[warning: more than 49 characters again]d. Quantity: 68
e. Price: 89.99
Be sure to update the total number of entries to account for these new ones.
Part 1 (Due 3/29)
You will read in the file (Util::inputFileName), add the two entries, and save the file
(Util::outputFileName) using open(), close(), read(), and write().
You must use the low-level functions we will talk about in APUE chapter 3 (open, close, read,
write). Failure to do so will result in 0 points for this part of the project.
It is highly recommended that you do the file reading and writing in class(es) outside
Part1SimpleFileModifier, as that code will be useful later.
Part 2 (Due 4/12)
In this case you will spawn a new process using fork() and exec(), and split the responsibilities
like this:
1. The original process will read the file (2 nd argument=2) and then write the data over a
pipe to the child process.
2. The child process will read the file from the pipe (which will be set to standard input)
and write the data to the output file.
3. PipeMaker will take care of the pipe setup for you:a. Create PipeMaker before the fork.
b. In the parent process, call setUpToWrite() to ready this process for writing. You’ll
get back the file descriptor to write to. Write the file data to that file descriptor
(hint: it’s just like writing to a file).
c. In the child process, before execing, call setUpToRead() to dup the pipe output
to standard input. You can then exec the process (21S_CS3377_Project) with the
write option (2 nd argument=3), read the data from standard input (just a file
descriptor, remember!), and write to the output file.
d. Either the parent or the child process can do the update (but not both,
obviously).
When calling exec, use the command 21S_CS3377_Project 2 3. This will trigger main to
give you the proper setup for the child process. You will be responsible for spotting the
Util::IOType of WRITE (3), and read from standard input rather than the input file.
Part 3 (Due 4/26)
In this part you will create a thread and pass the file data from one thread to the other. The
threads will be like this:
1. Main thread: read the data, create the thread, and pass the data along
2. Created thread: receive the data and output it to the file
I did all of this inside of Part3ThreadedModifier. Create a mutex and condition for both threads
to share, and pass a pointer to the Part3ThreadedModifier object to pthread_create (and read
it in the other thread). Then you can use the shared mutex and condition to coordinate passing
the data.
The easiest way to pass the data is to use a variable inside Part3ThreadedModifier (type
EntryInfo). The main thread should lock the mutex before creating the receiving thread (and the
receiving thread should attempt to lock the mutex right after it starts up) to ensure the proper
ordering. Then do a loop in each thread like this:
Main (sending) thread
Wait on shared condition (for writing
thread to be ready)
Receiving thread
Signal condition to say we’re ready
Wait on condition (for an entry to be
ready)
Update the variable with the next entry
Signal the condition
Loop around and wait on the condition
again
Retrieve the info, save it for later writingLoop around and signal the condition
again
Once you’ve passed all the entries (5 or 7 depending on where you want to add the new ones),
unlock the mutex on both sides.
Part 4 (Due 5/10)
Here you will use two processes again, this time with a socket connecting them.
A port number to use (12345) is at Util::portNumber.
• Note: if you get an error that the port is already in use, it’s likely because you just ran
the project and the operating system hasn’t released the port yet. You can either wait a
bit (a few minutes at most) or change the port number (12346, etc.).
When I did this step, I reversed the setup from part 2: the main process here reads from the
socket (writing to the output file) and the spawned process writes to the socket (reading from
the input file). Again, it is up to you where you want to add the two new entries.
When you spawn the 2 nd process, use 21S_CS3377_Project 4 3.
After the fork, the socket reading process (parent process for me) creates a socket and listens
on that socket using the port number above. The socket writing process (child process for me)
creates a socket and connects to the listen socket. Depending on the timing of things the listen
socket may not be ready the first time; here is code to repeatedly wait for the listen socket to
be available:
int amountToWait = 1;
while ( connect(fileDescriptor, (struct sockaddr*) &serverAddress,
sizeof(serverAddress))) {
if ( errno != ECONNREFUSED) {
// Something unexpected happened
throw FileModifyException(“Error connecting”);
}
std::cout << "Not ready to connect yet...n";
// Exponential backoff
sleep(amountToWait);
amountToWait = amountToWait * 2;
}
Once the connection is made (reader gets back a file descriptor from accept() and the writer
gets out of the loop above) you can transfer the data. Remember that a socket is just a file
descriptor, so your code to write/read from earlier parts will work here, too.