Purpose
The purpose of this assignment is to understand the use of sockets and simple client-server communication interactions.
Overview
In Assignment 4, you were asked to browse through a particular directory tree, locate and create a catalog of a certain subset of the files. However, this catalog was not accessible over a network. In this assignment, you are tasked with writing two programs, a client and server, which communicate with each other via TCP sockets and facilitate sharing of files over a network. Since files downloaded over a network have a high probability of being corrupted, you will also implement error checking using a widely used hashing algorithm.
Assignment
In this assignment, you are supposed to implement two C programs, a server and a client, that have the following functionalities:
- Files (either image or text) can be transmitted between the server and the client through TCP socket.
- A catalog file will be generated at the server and record all the image files that can be downloaded by the client.
- MD5 checksum is used at the client to verify the correctness of transmission.
- An HTML file, named “download.html” is used to show all the downloaded files.
- Deploy the server and client program on two different computers in the computer lab.
Program Invocation
The server program will be invoked as follows:
./server server.config
where server.config is like
Port = <port>
Dir = <server directory>
- The server program when started, should open a TCP socket on the specified port and wait for incoming client connections.
- The server should browse the directory specified in [server directory] to find image directory and prepare a catalog.
The client program will be invoked as follows:
./client client.config
where client.config is like
Server = <IP_address>
Port = <port>
Chunk_Size = <chunk size>
ImageType = <type> (only for passive mode)
- ip_address is an IPv4 address in dotted decimal format, which specifies the IP address of the machine on which your server program is running.
- port indicates the port on which your server is listening for connections.
- The chunk size is the size of chunks that we separate image files into for transmission.
- The type is only for passive mode. You need to transmit all the files satisfying the given image type. You give only one image type at a time.
Server design
The server first reads a configuration file (i.e., server.config), and get necessary configuration data. Then it browses through a specified directory (i.e., [server directory]) and searches for image files of type jpg, png, gif, and tiff (in the assignment we only give jpg files as an example, but the program needs to support all the image types). The server then creates a catalog file, named catalog.csv, which contains a record for each image file found (Details on creating this file are specified later in this handout). The server then starts listening for client connections on a specified port (i.e., [port]).
Client design
The client first reads a configuration file (i.e., client.config) to get necessary configuration information and then runs in 2 modes: Interactive and passive. In both modes, the client must first attempt to connect to a server through IP address and port (i.e., [IP address] and [port]). After the socket connection has been successfully set up, the client must download catalog.csv from the server. The contents of this file must then be displayed to the user on the screen/console.
Passive mode: In passive mode, the file extension will be provided in the configuration file. This will be one of jpg, png, gif, or tiff. The client program must then download all files corresponding to the specified type from the server.
Interactive mode: In the interactive mode, the client must allow the user to choose files for download. You must allow the user to download multiple image files in this mode of execution.
When you are downloading the files, separate image files into small chunks, because the size of image file may be large. Each chunk has [chunk size] bytes. After a file is downloaded, you need to output the number of chunks you have downloaded.
Record available images through “catalog.csv”
This is a CSV formatted file. You need to use a comma(,) as the delimiter. The header line for this file should be
filename, size, checksum
Following the header line, each image file found in the directory must have its own record in the file on a separate line. The filename must be stripped down to the last / (e.g., image01.png instead of /home/grad03/student/hw5/image01.png) . You need to compute the checksum for each image file. You may use the provided API for this. You may refer to catalog.csv in the attached tar ball.
Verify the correctness of downloading through checksum
You have been provided with an API which computes the MD5 hash for a given file. You need to compile md5sum.c and link it with your server and client programs. You may use the provided md5_test.c and Makefile as a reference. (This program is identical to the md5sum command in Linux). Do NOT modify the md5sum.[c/h] files. You will lose significant points if you do so.
Show the downloaded files through “download.html”
This is an HTML document which should list the names of all downloaded files with hyperlinks to the full size images. Additionally, “Checksum match” or “Checksum mismatch” must be displayed next to each filename depending on the result of the checksum comparison. (Note: If, for a particular file there is a checksum mismatch, you do not need to provide a link to the full size image, since the image is probably corrupted). You can find a sample download.html in the attached tar ball.
Deployment of server/client
Although it is possible to run both the server and the client on the same machine, for this assignment you need to ensure that the server and the client both run on different machines. For instance, you can run the server on one of the machines in Lind40 and the client on one of the machines in Keller 4250. (This is how your assignment will be graded, by running your server and client programs on physically different machines, so make sure that it works).
Assumptions
General
- You are not allowed to use commands in shell scripts, such as “find” & “grep”, use functions in C libraries, such as “opendir()”and “readdir()”.
- You should not change md5sum.[c/h], but you can put them in your own directory.
- Maximum length of file names [ 255 bytes; maximum length of dir path [ 1024 bytes
- There’s only one client and one server.
- We assume the catalog.csv file will transmit successfully.
Server-side
- The directory path in server’s configuration file to the server can be absolute or relative.
- The specified directory can contain files other than the image files.
- The specified directory can contain any number of subdirectories. You need to go through all the subdirectories and check for image files while preparing the catalogue.
- The filenames encountered in the directory are guaranteed to be unique.
- If catalog.csv exists, you must truncate/delete it and create a new file.
Client-side
- The images subdirectory may or may not exist initially. If it exists, it may contain some files. However, these files will not be found on the server (i.e. There will not be any filename conflicts during the download phase)
- You can place all downloaded images in the top-level images subdirectory. There is no need to recreate the directory structure as per what is present on the server.
- If download.html exists, you must truncate/delete the file and create a new file.
- You may safely assume that the server program is always started before the client program starts.
Trouble shooting
- Error in compiling “fatal error: openssl/md5.h: No such file or directory”
- First check whether you have installed libssl-dev, and openssl
- Then check your make file, remember to (1) compile server.c with linkers: -lcrypto - lssr, and (2) use linker -pthread for threads
- Sometimes you need to convert signed char/unsigned char to char*, you can try “sprintf”.
- When you are reading/writing from/to sockets, try to use unbuffered I/O, such as read(), write(), send(), recv(), instead of buffered I/O, such as fprintf() and fread().
What you need to turn in
- The source files of your programs ( *.[c/h] files)
- The configuration files ([server/client].config)
- A README.txt which contains the name and student ID of the team members. This file must also contain instructions about the way your programs must be run.
- A Makefile for compiling the two programs. The two executables must be named server and client respectively.
- Please do not include download.html, catalog.csv, images/and other image files in your submissions.