Pointers

Assignment: Pointers

• Assign: Saturday 24 February
• Checkpoint: (Graded on completion) Complete the Checkpoint Questions by Wednesday 28 February
• Due: Thursday 7 March
• Policy: Pair graded synthesis assignment
• Partner search: Find a partner here.
• Code: cs240 start pointers (if this doesn't work, use cs240.s24 start pointers)
• Submit: git commit and git push your completed code.
• Reference:
  • Programming with Memory
  • Reasoning about Programs
  • C resources
  • debugging resources

Contents

• Assignment: Pointers
• Overview
  • Goals
  • Advice
  • Time Reports
• Setup
• Tasks
• Preparatory Exercises
  • Checkpoint Questions
  • After the Pointers lab
• Command Structure
  • Command Lines
  • Command Arrays
• Specification: your tasks
  • Function Specifications
  • Memory Rules
  • Coding Style Rules
• Implementation Workflow
• Compiling and Testing
  • Test Cases
  • Test Drivers
  • Assertions
• Debugging
  • valgrind
  • gdb
  • Backtracking
• Submission
• Grading
• More Tips
  • printf
  • C Function Headers and Declaration Order
• Extra Fun Extensions
  • Extra Fun Extend the Command Language with Quoting
  • Extra Fun Replace Multi-Pass Loop Code with Single There-and-Back-Again Recursive Code
• License

Overview

The main task of this assignment is to implement a shell command parser as a small library in C. The shell is the program that reads and interprets the commands you type at a command line terminal. This assignment has you implement the first step of such a program: splitting a command line into its meaningful parts.

The parser takes a command line, a single string, and converts it to a command array, an array of strings representing the command and its space-separated arguments. The string split function available in many programming languages’ standard libraries would accomplish most of the task, but in this assignment you will essentially implement split at a low level. We prohibit the use of string manipulation functions from the C standard library to force you to confront how strings are implemented and manipulated in memory.

Goals

• To become familiar with the byte-addressable memory model through pointers and arrays at the C programming language level of abstraction.
• To understand the link between pointer arithmetic and array indexing and representation.
• To practice principled reasoning about program execution, using assertions, testing, and structured debugging.
• To practice using tools for memory safety checking and debugging.
• To have at least one aha! moment in appreciating how storage abstractions in your favorite programming language relate to (and hide!) details of the byte-addressable memory model.

Advice

Programming with C pointers has a tendency go wrong in entertaining or painful ways unless you practice careful discipline in programming.

For each stage in the suggested workflow:

1. Plan carefully before typing any code.
2. Implement one step at a time (with useful assertions).
3. Test each step extensively, including with tools for detecting pointer errors, before implementing the next.
4. Commit each tested step before implementing more.

This careful process will save time by catching bugs early and making sure you can go back to previously working versions if things start going wrong later.

Time Reports

According to self-reported times on this assignment from a previous offering:

• 25% of students spent <= 8 hours.
• 50% of students spent <= 12 hours.
• 75% of students spent <= 15 hours.

Setup

Get your repository with cs240 start pointers, whether working alone or with a teammate.

Your repository initially contains the following files:

• For the command library implementation:
  • command.c: a file where you will implement a command parsing library for simple shell commands
  • command.h: a C header file describing the API of the command parsing library in command.c
• For testing, demonstration, and automation:
  • command_lines.c: a file where you will add test cases for your command parser to be used by all testing drivers
  • command_demo.c: a simple demonstration of how the command parsing library functions are used, also useful for incremental testing during development
  • command_show_test.c, command_free_test.c, command_parse_test.c: drivers for testing parts of the library
  • Makefile: recipes to compile the various parts

Compile all command-parsing code and test drivers with the command make. This produces several executables for different testing purposes during development. Run tests with individual test drivers. Some tests automatically report failures; others require manual inspection of results.

Tasks

You will modify command.c to implement a small library for parsing shell commands with the interface declared in command.h and the functionality described by the Specification section below.

char** command_parse(char* line, int* status);
void   command_show(char** command);
void   command_free(char** command);

The core of your library is the command_parse function, which takes a command line – a string capturing the command typed in the terminal – and parses it to produce a command array – an array of strings capturing the individual words of the command line. You will implement the command_parse function, plus three other simple functions for displaying and freeing command arrays.

A few provided test drivers include logic to help automate testing of your parsing library. These drivers all use command line test cases from command_lines.c. You will add several additional command line test cases for robust testing of your command parsing library implementation.

Grading considers design, documentation, style, correctness, and performance.

The remainder of this document describes:

1. Preparatory exercises to help you navigate and plan the assignment.
2. The definition of command lines and command arrays.
3. The specification of the functions you must implement.
4. The suggested implementation workflow.
5. Documentation of how to compile and test your implementation.
6. Tools and strategies for debugging.
7. The grading criteria.
8. Additional tips.

Preparatory Exercises

Checkpoint Questions

These questions are on what the assignment is asking you to do, and can be answered with a basic understanding of strings in general, even if you do not yet fully understand how strings are implemented in C.

Valid commands

1. Can spaces appear at the start or end of a valid command lines?
2. Is the & character ever considered part of a command word?
3. Can a command line with zero &s be valid?
4. Can a command line with two or more &s be valid?
5. Is a command line that consists of just zero or more spaces valid?
6. Is the command line consisting of just & valid?

Status field

1. For a valid command with zero ampersands, what should be stored in status?
2. For a valid command with one ampersand, what should be stored in status?
3. For any invalid command, should anything be stored in status?

command_parse and command_show

1. Is command_parse allowed to mutate the contents of a command line string it receives as an argument?

2. Consider the lines printed bycommand_show for the command array that results from calling command_parse on the command string "cp -r ../tests files/test-cases",

   1. How many total lines should be printed?
   2. How many total open/close curly braces ({ and }) should be printed?
   3. How many total commas should be printed?
   4. How many total double quote characters " should be printed?

If you are working on the assignment itself with a partner, be sure to consult with each other about the answers to these questions to before writing your implementation code.

After the Pointers lab

Review the lab exercises for Lab 6 on Pointers/Arrays in C.

Command Structure

Software driven by a command line interface (CLI), such as most software used in CS 240, interprets textual commands typed by a user at a command prompt. To interpret a command, a system must extract the command’s parts to understand its structure. In this assignment, your command parsing library will convert between two representations of simple multi-word commands:

• A command line is a string capturing the full text of a command, just as the user types it in a terminal.
• A command array is an array of strings, where each string captures one meaningful token in the command, in order.

The remainder of this section defines the structure of commands and these two representations as used in this assignment.

Command Lines

A command line is a C string containing zero or more space-separated words, with an optional single background status indicator character, '&', after all words.

• Like all C strings, the command line is terminated by the null character '\0'.
• Spaces may appear anywhere in a command line.
• Word characters include all printable characters excluding (1) ampersand ('&'), and (2) whitespace characters, which include space (' '), tab ('\t'), newline ('\n'), and carriage return ('\r').
• A command word is a contiguous sequence of word characters delineated on each end by a non-word character ('&' or whitespace) or the beginning or end of the string.
• Ampersand ('&') is a special status indicator character, not a word character.
  • In a valid command, ampersand ('&') may appear at most once, and only zero or more spaces after the final word, followed only by zero or more spaces to the end of the string. Any other placement of ampersand is invalid.¹
  • A valid command containing no ampersand is a foreground command.
  • A valid command containing one ampersand is a background command.

Command Line Examples:

Here are two typical command lines:

• The command line "ls -l cs240-pointers" indicates a foreground command containing the three command words "ls", "-l", and "cs240-pointers".
• The command line "emacs cs240-pointers/command.c &" indicates a background command containing the two command words "emacs" and "cs240-pointers/command.c".

The definition above allows any spacing. The following lines are all valid command lines indicating the foreground command containing the three words "ls", "-l", and "cs240-pointers":

"ls -l cs240-pointers"
"ls       -l cs240-pointers  "
"       ls   -l    cs240-pointers   "

The definition above requires that, if present, ampersand ('&') must be the last non-space character of the command line. It may appear adjacent to or separate from the last word. Regardless of placement, ampersand ('&') is not a word character and is never part of any command word. For example, these valid command lines all indicate the background command containing the two words "emacs" and "cs240-pointers/command.c".

"emacs cs240-pointers/command.c &"
"emacs cs240-pointers/command.c&"
"  emacs  cs240-pointers/command.c   & "
" emacs   cs240-pointers/command.c&       "

The following are some examples of invalid command lines, with ampersand misused:

"&uhoh"
"  & uh oh"
"uh & oh"
"uh& &oh"
"uh oh & &"
"   uh oh &&  "

Command Arrays

The parser converts a command line into a command array, an array of strings representing the words of the command line, in order, terminated by NULL. Recall that a string in C is not a special type; it is just an array of chars terminated by a null character ('\0'). A command array is thus an array of pointers to arrays of chars and has the type char**. All arrays in this structure must be null-terminated, using the right notion of nullness for each array.

Here is an example command array for the command line string "ls -l cs240pointers":

Command Array:             Null-Terminated Strings
Index   Contents           (stored elsewhere in memory)
      +----------+
    0 |  ptr  *----------> "ls"
      +----------+
    1 |  ptr  *----------> "-l"
      +----------+
    2 |  ptr  *----------> "cs240-pointers"
      +----------+
    3 |   NULL   |
      +----------+

Here is the same array drawn another way and showing how the array is arranged in memory, with each element’s offset from the base address of the array. Addresses grow left to right, and are assumed to be 64 bits (8 bytes), so indices are related to offsets by a factor of 8.

Index:        0       1       2       3 
Offset:   +0      +8      +16     +24     +32
          +-------+-------+-------+-------+
Contents: |   *   |   *   |   *   | NULL  |
          +---|---+---|---+---|---+-------+
              |       |       |
              V       V       V
              "ls"    "-l"    "cs240-pointers"

Although we draw “strings” in the above pictures, this is an abstraction. Each string is actually represented by a '\0'-terminated array of 1-byte characters in memory. Since each element is one byte, the addresses of adjacent characters differ by 1 and the offset is identical to the index. Below is the memory representation of the string "ls":

Index:      0     1     2
Offset:  +0    +1    +2    +3
         +-----+-----+-----+
         | 'l' | 's' |'\0' |
         +-----+-----+-----+

Note: Practice reading memory diagrams every which way. Pay attention to what direction addresses grow. We are intentionally using different conventions in different drawings to help you get used to thinking with different abstractions used in practice.

Specification: your tasks

Your task for this assignment is to write three functions (plus any necessary helper functions) supporting command parsing in command.c according to the headers in command.h.

If you write helper functions, be sure to read about how C function declarations work. We give one possible plan for implementation and testing shortly.

Function Specifications

• char** command_parse(char* line, int* status)

  Parse the command line string, line, and:

  • If the command line is valid:
    • Store 0 (false) in memory at the address given by the pointer status if the command is a background command or store 1 (true) at this location if the command is a foreground command.
    • Build a command array containing the words of the command line, in order, and return its base address.
  • If the command line is invalid, return NULL without setting foreground/background status.

  The command_parse function must not mutate the memory representing the original command line string.

• void command_show(char** commandArray)

  Print the structure of a commandArray according to the following format to aid in debugging and data inspection. You will need to learn how to use printf.

  The output should use the exactly the following format, which makes clear what strings the command array holds. For example, this C code:

  int status_here;
  command_show(command_parse("ls -l cs240-pointers ", &status_here));
  

  Should cause command_show to print this text:

  {
    "ls",
    "-l",
    "cs240-pointers",
    NULL
  }
  
  

  Note that the use of added quotes in the output format distinguishes a correct command array where all spaces are stripped away, such as above, from an incorrect command array that contains some spaces in words, as below:

  // Incorrect command_parse: whitespace left in 
  {
    "ls ",
    "-l ",
    "cs240-pointers   ",
    NULL
  }
    
  

  More generally, the output of command_show is the following sequence of lines, each ending with a newline character ('\n'):

  1. An initial line containing only a single left curly brace {.
  2. One line for each word in the command array, in order, where:
     • The line starts with two spaces   ; followed by
     • a double quote "; followed by
     • the characters of the command word in sequence; followed by
     • a double quote "; followed by
     • a comma ,.
  3. A line containing two spaces    followed by the four uppercase letters NULL, corresponding to the null-pointer terminator at the end of the command array.
  4. A final line containing only a single right curly brace }. This line should end with a newline character, just like all the others.

  This is valid C array initializer syntax. In other words, command_show prints a valid C source code description of the command array it is given, as long as the command array words do not contain characters in need of escape: " ('"'), \ ('\\'), tab ('\t'), carriage return ('\r'), newline ('\n'), etc. We will not test command lines that contain such characters, so supporting them is not required.

  Note that this output will be automatically graded, so you should match this format exactly.

• void command_free(char** commandArray)

  Free all parts of the given commandArray array structure that were allocated by command_parse, including the individual command word strings and the top-level array.

Memory Rules

Your code should be free of memory safety violations as detected by the valgrind memory error detector:

• No use of unallocated, uninitialized, or freed memory contents.
• No out-of-bounds accesses.
• No calls to free free on pointers not returned by malloc.
• No calling free multiple times for a given pointer.

In addition, your code must follow these stricter memory allocation rules:

• Clients of the command library own and manage the memory representing command line strings.
  • Command library functions must never free nor mutate command line strings.
• The command_parse function must allocate command array structures dynamically with malloc and return them to the client code that invoked the library functions. Once returned, these structures are owned by the client. You may assume that:
  • Clients will not mutate command array structures returned by command_parse.
  • Clients will call command_free to free an allocated command array (returned by command_parse) at most once and will not otherwise free any parts of the command array structures returned by command_parse.
• Only command_parse may allocate memory, and it must allocate no more memory than is necessary to represent the final command array structure at the time it is returned.
• Assuming that the client code eventually calls command_free on every command array returned by command_parse, all memory allocated by the command library functions must be freed.

Coding Style Rules

Follow this general C style guide as a baseline, along with the following specific rules:

• Do not use any C standard library string functions in command.c. You must define all string processing from scratch.
• For full credit, eliminate array indexing syntax in this assignment. Use idiomatic pointer style instead.
• Document in comments how your code implements the specifications provided, connecting your C implementation to the specification.
  • Our sample solution for command_parse is under 70 lines of code and 20–30 lines of comments.
• Use assertions in your code to check assumptions where reasonable.
• Use well-scoped helper functions where reasonable.
  • Clean solutions are possible with and without helper functions.
  • If using helper functions, match them to well-defined sub-problems with clear arguments and results.
  • Be sure to order function headers properly.
• Prefer simple efficient code over complicated code.
• Ensure your code has no compiler warnings or compiler errors.
• Ensure your code has no run-time errors, including memory errors such as segmentation faults, as well as more subtle memory safety violations and leaks as detected by valgrind.

Idiomatic Pointer Style

For full credit, your submitted code must use only pointers and pointer arithmetic, with no array indexing notation. We require this to get you to confront the real memory representation of arrays in this assignment. Outside this assignment, array indexing is clearer style in many cases.

A simple way to think with arrays but write with pointers is to use *(a + i) wherever you think a[i]. However, this simple transformation rarely matches an idiomatic pointer style, so you should not simply replace each occurrence of a[i] by *(a + i).

An idiomatic loop over an array with array indexing typically uses an integer index variable incremented on each iteration. Keep the scope of the index variable (or “loop variable”) as small as possible for the task.

// pre:  s is the address of a null-terminated string.
// post: replaces all characters in s by 'Z'
for (int i = 0; s[i] != '\0'; i++) {
    s[i] = 'Z';
}

An idiomatic loop over an array with pointer arithmetic typically uses a cursor pointer that is incremented to point to the next element on each iteration. Keep the scope of the cursor pointer variable as small as possible for the task.

// pre:  s is the address of a null-terminated string.
// post: replaces all characters in s by 'Z'
for (char* cursor = s; *cursor != '\0'; cursor++) {
    *cursor = 'Z';
}

In the above example, the cursor pointer is explicitly named cursor, which we think is a good name to highlight its purpose. However, in many real C programs the name cursor might be replaced by a shorter name like p or ptr.

Your final code should contain zero array[index] operations.

Implementation Workflow

This is one possible workflow to understand and implement command_parse and friends. Regardless of how you decide to approach this assignment, be sure to commit your changes after each step!

It’s a good idea to study the test driver section to understand what each test command is doing and other options you have for using the test drivers.

1. Finish all the prep exercises before starting to change any files.

2. Add several more valid command line test case strings in command_lines.c.

3. Use make test-expected to show the expected results for all test command lines in command_lines.c. This will use a solution implementation to illustrate how command_parse, command_show, and command_free should behave on each of the test cases in command_lines.c.

4. Implement command_show. Test with make test-show at this stage. This will use a solution implementation for command_parse to parse each of the test cases in command_lines.c. Then, for each of the resulting command arrays, it will display the printed output of your actual implementation of command_show on the command array followed by the expected printed output on the command array from the solution implementation of command_show.

   Compare the printed outputs carefully, and make sure that your actual output matches the actual output character for character! Pay special attention to braces, spaces, quotation marks, and commas!

5. Implement command_free. Test with make test-free at this stage. This will use a solution implementation for command_parse to parse each of the test cases in command_lines.c, but will use your implementation of command_free to free the result in command arrays and their contained strings.

   Study the output to see if it contains any valgrind error messages (these will begin with a notation like ==29223==, where the the particular number may vary) or memory leaks (listed under the LEAK SUMMARY: section at the end of the valgrind output.

   Not quite sure what to do here? It’s fine to implement command_parse first and then come back to this step.

6. Add many more valid and invalid command line test cases to command_lines.c to test all aspects of the specification extensively and adversarially.

7. Implement and test command_parse in stages. Test each stage and commit a working version before continuing. The basic algorithm should proceed in this order:

   1. Count the number of words in line and detect use of &, returning NULL for invalid commands and marking the foreground/background status for valid commands.
      • Note that the stencil code has a command_char_is_space helper function already provided.
      • Test with valgrind ./command_demo_actual.bin and some printing of intermediate results in command_parse or command_demo at this stage.
   2. Allocate the top-level command array based on the counted number of words.
   3. For each word in line, allocate properly sized space to hold the word as a null-terminated string, copy the word into this space, and save it into the command array.
      • Test with valgrind ./command_parse_test_actual.bin and valgrind ./command_adversarial_test_actual.bin at this stage.

8. Once everything is implemented:

   • Run make test-demo to test all of your functions together at once.
   • Comment out or remove any diagnostic printing you added during development. This is important because the diagonstic printing will confuse the autograder.

Compiling and Testing

To compile all test drivers, run make.

You are writing a command parsing library, i.e., a collection of related useful functions. The library itself cannot be run directly; it is not a complete program. We provide several test driver programs that act as clients of the library, calling its functions. You will write several command line test cases to be used by these drivers.

Test Cases

Develop your own large suite of test cases by adding command line strings to COMMAND_LINES in command_lines.c. By large, we mean perhaps dozens. Be adversarial. Create test cases for as many corner cases as possible in both valid and invalid commands. Your test cases will contribute to your grade, and preparing and using them will help you ensure your code is correct and efficient. All test drivers use your tests as their source of additional test commands.

Test Drivers

Each provided test driver is useful for different testing tasks. All driver executable files end with .bin (for binary executable). Each driver has two variants:

• The actual variant uses your command.c library. Run this variant to test your code.
• The expected variant uses a complete and correct (but hidden) solution implementation of the command library. Use it to see expected behavior or compare your code’s behavior to expected behavior.

For both variants, the executable can be run in two ways:

• Without any arguments, the executable will run its test on each command line from COMMAND_LINES in command_lines.c. Example:

  $ valgrind ./command_demo_actual.bin
  

• Given exactly one argument, the executable will run its on test using the single argument as a command line. To test on a command line with multiple words, be sure to quote it so it is captured as a single string, rather than one argument per word. (Talk about meta, we have to be careful with how your terminal parses your command line to make sure we are passing the string we expect to your command line parser!)

  $ valgrind ./command_demo_actual.bin "Hello world!"
  

In general, always run these executables under valgrind to help catch memory errors early. The drivers are listed below with their purpose, methods for checking correctness, source code file, command to compile just that driver, and the names of the relevant executables.

1. demo:
   • Purpose: demonstration of library functions, testing command_parse before completion.
     • This is a simple driver that is easy to modify and extend by changing the the function command_demo in the command_demo.c file. For example, you might invoke helper functions directly in addition to command_parse. If doing so, you must copy their function headers to command.h to make them visible from command_demo.c.
   • Correctness: manually inspect outputs, check the output of valgrind for memory errors or memory leaks, compare total memory allocations reported by valgrind for the actual and expected versions, optionally add custom additional checks to the driver code
   • Source: command_demo.c
   • Compile: make demo
   • Executables: ./command_demo_actual.bin and ./command_demo_expected.bin
   • Shorthands:
     • make test-expected is equivalent to make demo followed by valgrind ./command_demo_expected.bin.
     • make test-demo is equivalent to make demo followed by valgrind ./command_demo_actual.bin.
2. show:
   • Purpose: testing command_show
   • Correctness: manually inspect outputs, check valgrind for memory errors
   • Source: command_show_test.c
   • Compile: make show
   • Executables: ./command_show_test_actual.bin
   • Shorthand: make test-show is equivalent to make show followed by valgrind ./command_show_test_actual.bin.)
3. free:
   • Purpose: testing command_free
   • Correctness: check the output of valgrind for memory errors and memory leaks
   • Source: command_free_test.c
   • Compile: make free
   • Executables: ./command_free_test_actual.bin and ./command_free_test_expected.bin
   • Shorthand: make test-free is equivalent to make free followed by valgrind ./command_free_test_actual.bin.
4. parse:
   • Purpose: testing a complete command_parse implementation under valgrind
   • Correctness: check reports of mismatches between actual and expected behavior, check valgrind for memory errors and memory leaks
   • Source: command_parse_test.c
   • Compile: make parse
   • Executables: ./command_parse_test_actual.bin and ./command_parse_test_expected.bin
   • Shorthand: make test-parse is equivalent to make parse followed by valgrind ./command_parse_test_actual.bin.
5. adversarial:
   • Purpose: testing a complete command_parse implementation
     • This driver replaces malloc and free with correct–but more adversarial–versions that make buggy command_parse implementations more likely to crash or cause obvious data corruption. Basically, allocated blocks are pre-filled with '%' characters (byte value 0x25) and the immediately adjacent (out of bounds) chunks of memory are filled with '@' characters (byte value 0x40) before the allocated block and filled with '^' characters (byte value 0x5E) after the allocated block, while freed blocks are mostly filled with '$' characters (byte value 0x24). This makes use of values from uninitialized ('%' == 0x25), out of bounds ('@' == 0x40 and '^' == 0x5E), and recently freed ('$' == 0x24) locations less likely to get “lucky” by finding '\0' and more likely to cause obvious corruption or segmentation faults.
     • It is reasonable to run this driver under valgrind, but valgrind may report fewer errors than for the other test drivers due to the extra manipulation that the driver is doing.
   • Correctness: check reports of mismatches between actual and expected behavior, check for crashes and obvious data corruption, check valgrind for memory errors and memory leaks
   • Source: command_parse_test.c
   • Compile: make adversarial
   • Executables: ./command_adversarial_test_actual.bin and ./command_adversarial_test_expected.bin
   • Shorthand: make test-adversarial is equivalent to make adversarial followed valgrind ./command_adversarial_test_actual.bin.

In addition, the following command will check that there are no compiler warnings. This is an important step for this project, since some compiler warnings may indicate issues with incorrect type annotations that cause unexpected allocation size behavior:

1. check-warnings:
   • Purpose: checking that you have no compiler warnings.
   • Correctness: check for error reports. With no warnings, the output should be a single line starting with gcc.
   • Compile: make check-warnings

Assertions

Assertions are “executable documentation” or “executable specifications”: they document rules about how code should be used or how data should be structured, but they also make it easier to detect violations of these rules (a.k.a. bugs!). Use the assert(...) statement in C by including assert.h and asserting expected properties.

The provided starter code already includes code that asserts that the arguments to command_ functions are not NULL. command_parse begins with

assert(line != NULL);
assert(status != NULL);

and both command_show and command_free begin with

assert(commandArray != NULL);

If a NULL argument is ever passed to these functions, an error message will be printed and execution will halt immediately. Detecting errors early like this (vs. crashing or corrupting data later wherever the code depends on this assumption) saves a lot of time. Add assertions to make the “rules” of your code clear wherever you make assumptions.

Note that assertions are clearer when the declaration involves a boolean expression. Although assert(line != NULL); and assert(line) have the same behavior, the first version is preferred because it explicitly indicates that the assertion verifies that line is not NULL.

Debugging

valgrind and gdb are the tools of choice. (See the Pointer Practice exercises, which introduce their usage.) Avoid print-based debugging. Backtrack from symptoms to cause.

valgrind

Valgrind is an extended memory error checker. It helps catch pointer errors, misuse of malloc and free, and more. Run valgrind on your compiled program like this: valgrind ./command_demo_actual.bin. (Replace ./command_demo_actual.bin with any other executable to check others.) The valgrind tool will run your program and observe its execution to detect errors at run time. See the tools page for additional reference links.

Be sure to periodically run your code under valgrind during development and testing to help catch memory errors early.

gdb

Use GDB to help debug your programs when you need more information than valgrind offers. Review previous lab activities for tips. When debugging programs with pointers, pay special attention to the pointer values your program generates. Inspect them like other variables or use the address-of (&) and dereference (*) operators at the gdb prompt to help explore program state.

Documentation and quick reference for gdb:

• CS 240 GDB reference sheet (pdf, txt)
• GDB Quick Reference card
• GDB manual
• For documentation from the command line, type help at the gdb command prompt, or type man gdb, or info gdb in the shell. Some people also like to run gdb under gdb-mode in Emacs.

Does GDB tell you that it cannot display something due to compiler optimizations? If so, turn off compiler optimizations by adding -O0 (that’s “space dash Capital-Oh zero”) at the end of the CFLAGS = ... line in your Makefile. The CFLAGS variable here determines what flags are passed to the compiler. -O0 enables level zero of optimization (i.e., none). Run make clean and then make to compile your code again without optimizations.

Backtracking

Did your program hit an explicit error? An assertion failure? A segfault? Debug by collecting evidence and backtracking methodically to find the cause of this symptom.

• Localize the crash. Exactly what line of code, what operation in this line, what value(s) used by this operation manifested what error and crashed? This symptom does not necessarily indicate the cause, but it marks where to begin the search.
  • valgrind will provide most of this information for every memory error.
  • If more information is needed, use gdb to run the code until it crashes, then find the line that crashed (and the context in which it executed) with bt, backtrace, or where.
  • Inspect the most relevant line of code and the error report to determine what the error means. Was it a segfault? If it was an assertion failure, what was the assertion checking?
  • Inspect the arguments shown in the backtrace or print variables to determine what ill-formed pointer was dereferenced to cause a segmentation fault or what illegal values failed an assertion.
• Trace backward through the dependencies of this broken operation to the original logic error.
  • Invalid values: Did any of this operation’s arguments hold invalid values? Did this operation load any invalid values from memory?
    • What operations produced these invalid values?
      • This is easy to answer for invalid arguments.
      • For invalid values loaded from memory, consider what earlier operations (perhaps in completely separate function calls) may have saved (or failed to save) this value.
    • Continue tracing backwards, treating these producer operations as broken operations.
  • Invalid choices: Was it invalid to apply this operation here given correct arguments (or regardless of the arguments)?
    • What control flow decision led to execution of this operation?
    • Is the logic of the decision correct?
    • If so, continue tracing backwards treating the control flow decision as a broken operation.
  • Callers: If using helper functions, you may need to trace back beyond the function where the crash happened.
    • Get a backtrace to see where the current function was called. (Don’t just assume you know which function called this one. Get the truth.)
    • If you want to inspect local variables in functions that called this one, use up and down in gdb to step up and down the call stack, then print what you want to see.
• Minimize the input. Try to write a small command line that captures the essence of the command line on which your command parser encounters an error. A smaller input is easier to reason about in full.

Submission

Submit: The course staff will collect your work directly from your hosted repository. To submit your work:

1. Test your source code files one last time. Make sure that, at a minimum, submitted source code is free of syntax errors and any other static errors (such as static type errors or name/scope errors). In other words: the code does not need to complete the correct computation when invoked, but it must be a valid program. We will not grade files that do not pass this bar.

2. Make sure you have committed your latest changes. (Replace FILES with the files you changed and MESSAGE with your commit message.)

   $ git add FILES
   $ git commit -m "MESSAGE"
   

3. Run the command cs240 sign to sign your work and respond to any assignment survey questions.

   $ cs240 sign
   

   (If this encounters an error, instead execute cs240.s24 sign.)

4. Push your signature and your latest local commits to the hosted repository.

   $ git push
   

Confirm: All local changes have been submitted if the output of git status shows both:

• Your branch is up to date with 'origin/main', meaning all local commits have been pushed
• nothing to commit, meaning all local changes have been committed

Resubmit: If you realize you need to change something later, just repeat this process.

Grading

Your grade will be derived as follows:

• Functional Correctness (55 points):
  • Your code behaves according to the specification.
• Memory Safety and Efficiency (20 points):
  • Your code is free of memory safety violations.
  • Your code allocates no more memory than strictly necessary.
  • Your code does not suffer memory leaks (as detected by valgrind).
• Design, style, documentation, and test cases (25 points):
  • You completed the Checkpoint Questions by the required deadline. (3 points)

  • Your command_lines.c file contains sufficient test cases for normal and boundary cases for valid and invalid commands. (5 points)

    You should extend command_lines.c with your own large suite of test inputs and run test drivers under valgrind. Careful and extensive testing will help you check that your code meets the specification and runs free of memory safety violations and memory leaks. Take this seriously.

  • Your code in command.c is well-organized, easy to read, and uses clear and uncomplicated control flow. Helper functions are used to simplify code where appropriate. (7 points)
  • Your code uses idiomatic pointer style rather than using array indexing. (5 points)
  • Your code is well documented, using appropriate comments to highlight aspects that may not be obvious from the code itself. (5 points)

We compute the correctness, memory safety, and efficiency components of your grade by running your code on a private suite of test inputs under valgrind to detect memory safety violations, measure memory allocation, and detect memory leaks.

More Tips

printf

The standard way of printing output in C is the printf function from the stdio library. The printf function accepts a format string (a template) and additional arguments to format according to each of the format specifiers (holes in the template with directions about how to fill them given the right kind of value). Use this linked documentation/introduction to get started with printf. Here’s an example:

void display_line_info(char* command_line) {
  // Note: strlen computes the length of a string,
  // but you may not use it in this assignment.
  printf("Command line is \"%s\" (%d characters).\n",
         command_line, strlen(command_line));
}

Calling display_line_info("Hello world!") would print this output:

Command line is "Hello world!" (12 characters).

Some important things to know when using printf:

• printf prints exactly what you tell it to, not more: you must use a newline character ('\n')if you want a new line to be printed.
• Instead of figuring out how to build a larger string just to print it, either use the printf format string to do the building or use multiple printf calls with smaller individual strings.

• Even when printing a preexisting string value, an explicit format string should generally be used:

  char* my_string = computed_somehow();
  // Use this:
  printf("%s", my_string);
  // Not this:
  printf(my_string);
  

• Since displaying individual characters to the screen is expensive, standard output (stdout), the channel to which printf prints to display in the terminal, is buffered: individual pieces of output are gathered up into larger chunks and flushed to the display all at once. This improves efficiency. For most printing tasks, this implementation detail is entirely imperceptible and not necessary to understand. However, if you are debugging code that interleaves small printfs with other operations that could crash or cause valgrind errors (e.g. memory operations), there’s a chance you will be confused if you are not aware of it.
  • The newline character ('\n') typically flushes all buffer text. In other words, printing a newline causes any text that has already been printed–but has not yet appeared in the output–to appear in the output immediately, in the order it was printed.
  • Strings without a newline may not appear immediately – perhaps only when the next newline is printed.
  • To flush the buffer explicitly at any time, you can use fflush(stdout);. This is not needed for a working command.c implementation, but if you run into seemingly truncated output while debugging, well-placed fflush calls might help clarify things during debugging. In general, this is a good reason to use a debugger instead of print-based debugging.

C Function Headers and Declaration Order

In C, a function is allowed to be used only after (i.e., later in the file than) its declaration. This differs from Java, which allows you to refer to later methods. When declaring helper functions, you can do one of a few things to deal with this:

1. Just declare your helper function before the functions that use it.

2. Write a function header earlier in the file and the actual definition later in the file. The function header just describes the name and type of the function, much like an interface method in Java. For example:

    // A function header declares that such a function exists,
    // and will be implemented elsewhere.
    int helper(int x, int y);
   
    // Parameter names are optional in headers.
    int helper2(char*);
   
    void needsHelp() {
        // OK, because header precedes this point in file.
        helper(7, 8);
        helper2("hello");
    }
   
    int helper(int x, int y) {
        return x + y;
    }
   
    int helper2(char* str) {
        return 7;
    }
   

3. If the functions would likely get used elsewhere, then put the header in a header file, a file ending in .h that contains only function headers (for related functions) and data type declarations. For example, if you added another general function (not just a helper function) for manipulating commands (not required in this assignment), it would be best to place a function header for it with the other function headers in command.h so that users of your command library can call it.

   Header files are included (essentially programmatically copy-pasted) by the #include directive you often see at the tops of C source files. Then these functions can be used and their implementations will later be found elsewhere if compiled correctly.

Extra Fun Extensions

Before trying either of these extensions:

1. Finish and submit your work for the required parts with a descriptive commit message.
2. Make a copy of your command_parse function and rename the copy command_parse_standard.
3. Replace entire the body of the command_parse function with return command_parse_standard(line, status);. Now, command_parse is just a wrapper for whatever version you are working on.

After completing a working extension, make sure to test again before submitting any work on the extensions.

Extra Fun Extend the Command Language with Quoting

Start by making a copy of command_parse_standard and renaming it command_parse_quotes, then changing the body of command_parse to call this new function.

Extend your command-line parser to support quoting with the single-quote (') or double-quote (") characters. Note:

• After a starting quote character, all characters (including ampersand & and space) until the matching close quote are part of the current word.
• The delimiting quote characters are not part of the word.
• A quoted section immediately adjacent to a non-quoted word (with no intervening space) together form a single word.
• A literal quote character may be used as part of a quoted or unquoted word by using the escape character \ to prefix it.

Here are some examples using single-quote:

• Basic quoting example with the single-quote character:
  • Command line:
    • As a C string literal: "echo 'hello && world' si'ngle word' &"
    • When printed: echo 'hello && world' si'ngle word' &
  • Command array:
    • "echo"
    • "hello && world"
    • "single word"
    • NULL
  • Command status: background
• With escape characters to allow quote characters inside quoted elements:
  • Command line:
    • As a C string literal: "echo 'Don\\'t unquote until here,' I\\'m sure."
    • When printed: echo 'Don\'t unquote until here,' I\'m sure.
  • Command array:
    • "echo"
    • "Don't unquote until here,"
    • "I'm"
    • "sure."
    • NULL
  • Command status: foreground

Extra Fun Replace Multi-Pass Loop Code with Single There-and-Back-Again² Recursive Code

Define a small family of recursive helper functions that accomplish the processes of counting words, counting letters, allocating, copying, etc., without using any loops. Use only recursion. Replace the body of command_parse with a single call to the top-level recursive function you define.

There are many levels of progressively more sophisticated recursive formulations.

Your loop-based code probably made several repeated scans of the command-line string or its subsections in order to avoid additional allocations. In the most sophisticated recursive versions, it is possible to accomplish the entire parsing process with only a single pass over the command-line string, with exactly one recursive function call and one command-line character array access for each character of the command line (plus null terminator). Local variables at each layer of recursion can remember (as recursion progresses deeper) and later reuse relevant input characters and other information (as recursion returns from deeper levels). This is not entirely free: no extra heap space or command-line scans are required, but the call stack will store information about each recursive call in memory.

License

Pointers by Benjamin P. Wood at Wellesley College is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

Source code and evaluation infrastructure for Pointers are available upon request for reuse by instructors in other courses or institutions.