sshell$

Summary:

This program sshell$ was aimed to help us understand UNIX system call through the development of our own shell. A shell is a command-line interpreter that accpets user input in the form of command lines and executes them. sshell will perform a set of core processes; Execute, Selection, Redirection, and Composition.

Implementation:

The implmentation of this prorgam follows 9 distinct steps:

0. Setting up the make file and understand the code for system()
1. Program the fork()+exec()+wait() method
2. Execute command lines containing programs and their arguments
3. Implement builtin commands; pwd, cd
4. Parsing the output and redirecting it to a running specified program
5. Run commands concurrently to allow data to be transmitted from one program to the next
6. Implement sls retrieves current working directory and prints file along with its size.
7. Redirect standard error specified by two meta-characters to a pipeline or specified file
8. Handle errors if incorrect command line is supplied by user

Phase 1: fork()+exec()+wait()

The beginning of this process the shell fork() and creates a child process. The child process then runs the specified command with execvp() while the parent process waits using waitpid() until the child process has completed and the parent is able collect the exit status which is returned by retval.

Phase 2: Command arguments

At the beginning of the commands execution we are parsing the command and its arguments, sperating the string using strsep(). This process seperates the string when specified characters are found. A set counter argc is used to track the argument postion and insert it into an array of arguments argv[]. When command line has been parsed, the command argv[0] is initially checked for any builtins commands followed by its arguments argv[1]. Then it will flow into a fork()+exec()+wait() and execute parsed command into stdout.

Phase 3: Builtin commands

Once the configuration structure for parsing commands is built, it can now detect if command argv[0] is a builtin command.

Two helper functions have been created to optimize the code flow.

cd(): if argv[0] is equal to cd, then we will exectute the function. Unlike pwd(), this command takes arguments to change the current working directory of process to the directory specificed by $PATH$. if no argument !argv[1] then it will return an out of bound exit code of 255. If valid argument is passed in then return 0, else return 1.

pwd(); similar to the inital step for cd(). Once we have enterted the function we check argv for any following arguments that the command line may have. If there is an arg present in argv[] besides argv[0] this will return an out of bound exit code of 255. If no other arguments are inserted then we use a buffer to get current working directory with getcwd() and print out to stdout.

Phase 4: Output redirection

We have three pointers to keep track of the filename, delimiter, and array of arguments from the command. The process checks if there is an output direction > or >&, remove the meta-character, then filename, and finally store it.

We begin the output redirection by checking for a one of the specific meta characters > or >& using strstr(). Once an occurance of the first character is found we set two flags stdout_redirect and stderr_redirect to a 0 or 1. strstr() will return the first occurance in the string if any, is detected.

Below shows how the flags and delimiter are set for each case, to distingish between operations needed to be done:

>&: Indicates standard error and standard output should be redirected. If encountered, stdout_redirect and stderr_redirect are set to 1 and delimiter is assigned to >.

>: Indicates that the command right before meta character is to write output to the specified file instead of the shell's standard output. If encountered, stdout_redirect is set to 1 and delimiter is assigned to >.

Once the flags and delimiter are set, the delimiter is checked if empty (NULL) . If empty this indicates that command has no redirection. Otherwise, we split the argument into an array of arguments using the splitStrtoArr(). Once there , arguments are seperated and assigned to either cmd or filename. If stderr_redirect flag is set then we check for next character to gather filename in its entirety.

Once filename and command are avaliable, they are ran through different edge cases prior to opening and creating file to make sure that there is indeed a file or command to perform the process. If one or both are missing, then errormessage() will send error message out to stderr. Then the shell checks if a file is able to be opened or created. Finally if successful, we continue with execution.

Phase 5: Pipeline commands

If we have more than 1 command, then for each additional process, we need to fork in order to execute them concurrently. We also need to pipe the previous process's stdout and/or stderr into the next process's stdin.

First we create a duplicate of the existing stdout, stderr, and stdin so we can restore them to their initial state when we are done. Then we go into a for loop, and for each process, we create a pipe and fork(), then in the parent we wait for the child to finish, then we copy the current stdin into our newly created pipe's read end. Next, we check for the exit code returned by the child, perform any error handling, then proceed to the next iteration of the for loop. For the child, we need to change the stdin to the previous pipe's read end. Then depending on the total number of processes, which process is currently executing, and whether we have an output redirection. We redirect stdout and/or stderr to either write to the pipe, write onto the terminal, or write into the provided file. Then we execute the command and exit.

Finally, after all the processes are executed, we restore stdin, stdout and stderr, and print out the + completed ... message using completed_process().

Phase 6: Extra feature(s)

Standard error redirection:

1. Explained in Phase 4 under >&
2. |&: we first check whether we need to pipe error. We do this by checking whether there is a & after |. Then when we need to pipe error, we change stderr to write to the pipe together with stdout.
3. sls (simple ls): We create object for dp and sb which is our data path and size signal. We retrieve our currently working directory using opendir(.). Once the directory is present, we read each file in directory and as long as it is not a hidden file. We print the file and file size to stdout.

Phase 7: Helper functions

Function splitStrtoArr() performs a method that splits a string based on a delimter using strsep(). Checks for any lingering white-spaces by calling trimspaces() then stores split arguments into an array of arguments.

Function trimspaces() trims leading and trailing spaces of a string by doing some pointer arithmatics and assigns ending element as null terminator if none is present.

Function error_message() performs the parsing error handling. By running specific edges cases, a switch statement is used to parse the correct error message onto stderr specific to the condtion it failed to meet.

Function completed_process() retrieves the command and exit status retval of the completed proccess and prints the finished statment with exit status onto stderr.

Function char_switch() goes through command and finds occurances of | or &. If found, it performs a swap of characters and switches those for spaces.

Function occur() goes through the entirety of the string to find occurances of a specific character and keeps track of amount when detected.

Function execute_process() first check if the process being asked to execute is a built-in function. Otherwise, it forks itself and the parent waits for the child to return an exit status. The child execute the command.

References:

1. Split string into array: https://c-for-dummies.com/blog/?p=1769
2. Trimming whitespace: https://stackoverflow.com/questions/122616/ how-do-i-trim-leading-trailing-whitespace-in-a-standard-way
3. Pipe implementation: https://stackoverflow.com/questions/60804552/ pipe-two-or-more-shell-commands-in-c-using-a-loop
4. forking and waiting: https://stackoverflow.com/questions/13654315/two- forks-and-the-use-of-wait
5. Ignoring hidden files: https://stackoverflow.com/questions/845556/ how-to-ignore-hidden-files-with-opendir-and-readdir-in-c-library
6. CWD info: https://stackoverflow.com/questions/59983516/ what-does-signify-in-c-as-argument-to-opendir
7. GNUC Libary: https://www.gnu.org/software/libc/manual/html_mono/libc. html#Running-a-Command

Authors:

Created by Jeffery Zhang and Sergio Santoyo