Princeton University
COS 217: Introduction to Programming Systems

Purpose

The purpose of this assignment is to help you learn C dynamic memory management, and how to create abstract data types (ADTs) in C. It also will give you the opportunity to gain more experience with the GNU/Linux programming tools, especially bash, emacs, gdb, and make.

Ground Rules

Implementing hash table expansion (as described below) is the "on your own" part of this assignment. That part is worth 10% of this assignment.

Background

A symbol table is an unordered collection of bindings. A binding consists of a key and a value. A key is a string that uniquely identifies its binding; a value is data that is somehow pertinent to its key. A symbol table allows its client to insert (put) new bindings, to retrieve (get) the values of bindings with specified keys, and to remove bindings with specified keys. Symbol tables are used often in programming systems; compilers, assemblers, and execution profilers use them extensively.

There are several reasonable ways to implement a symbol table. A simple implementation might store the bindings in a linked list. Linked lists are described in Section 2.7 of The Practice of Programming (Kernighan and Pike) and Section 17.5 of C Programming: A Modern Approach (King). A more efficient implementation might use a hash table. Hash tables are described in Section 2.9 of The Practice of Programming (Kernighan & Pike) and Chapter 14 of Algorithms in C, Parts 1-4 (Sedgewick).

Your Task

Your task in this assignment is to create an ADT named SymTable. Each SymTable object should be a symbol table. A SymTable object should be generic. That is, a SymTable object should contain values which are void pointers, and thus can point to data of any type.

You should create two implementations of your SymTable ADT: one that uses a linked list and another that uses a hash table.

The SymTable Interface

Store your SymTable interface in a file named symtable.h. It should contain these function declarations:

SymTable_T SymTable_new(void);

void SymTable_free(SymTable_T oSymTable);

int SymTable_getLength(SymTable_T oSymTable);

int SymTable_put(SymTable_T oSymTable, const char *pcKey, const void *pvValue);

void *SymTable_replace(SymTable_T oSymTable, const char *pcKey,
   const void *pvValue);

int SymTable_contains(SymTable_T oSymTable, const char *pcKey);

void *SymTable_get(SymTable_T oSymTable, const char *pcKey);

void *SymTable_remove(SymTable_T oSymTable, const char *pcKey);

void SymTable_map(SymTable_T oSymTable,
   void (*pfApply)(const char *pcKey, void *pvValue, void *pvExtra),
   const void *pvExtra);

SymTable_new should return a new SymTable object that contains no bindings, or NULL if insufficient memory is available.

SymTable_free should free all memory occupied by oSymTable.

SymTable_getLength should return the number of bindings in oSymTable.

If oSymTable does not contain a binding with key pcKey, then SymTable_put should add a new binding to oSymTable consisting of key pcKey and value pvValue and return 1 (TRUE). Otherwise the function should leave oSymTable unchanged and return 0 (FALSE). If insufficient memory is available, then the function should leave oSymTable unchanged and return 0 (FALSE).

If oSymTable contains a binding with key pcKey, then SymTable_replace should replace the binding's value with pvValue and return the old value. Otherwise it should leave oSymTable unchanged and return NULL.

SymTable_contains should return 1 (TRUE) if oSymTable contains a binding whose key is pcKey, and 0 (FALSE) otherwise.

SymTable_get should return the value of the binding within oSymTable whose key is pcKey, or NULL if no such binding exists.

If oSymTable contains a binding with key pcKey, then SymTable_remove should remove that binding from oSymTable and return the binding's value. Otherwise the function should not change oSymTable and return NULL.

SymTable_map should apply function *pfApply to each binding in oSymTable, passing pvExtra as an extra parameter. That is,the function should call (*pfApply)(pcKey, pvValue, pvExtra) for each pcKey/pvValue binding in oSymTable.

A SymTable object should "own" its keys. That is, a SymTable object is responsible for allocating and freeing the memory in which its keys reside. Specifically, SymTable_put should not simply store the value of pcKey within the binding that it creates. Rather, SymTable_put should make a copy of the string pointed to by pcKey, and store the address of that copy within the new binding. You will find the standard C functions strlen, malloc, and strcpy useful for making the copy. A SymTable object also should free the memory in which its keys reside when that memory is no longer required.

Conversely, a SymTable object should not own its values. Indeed it cannot own its values; since it cannot determine the types of its values, it cannot create copies of them.

The SymTable Linked List Implementation

Your SymTable linked list implementation should:

• Reside in a file named symtablelist.c.
• Validate function parameters by calling the standard assert macro.
• Contain no memory leaks. Your SymTable ADT will use dynamically allocated memory. It should explicitly free all dynamically allocated memory when that memory is no longer required; for each call of malloc or calloc in your ADT, eventually there should be exactly one call of free.
• Avoid gross inefficiencies. In particular, it would be grossly inefficient to traverse the list multiple times when a single time would suffice.

The SymTable Hash Table Implementation

Your SymTable hash table implementation should:

• Reside in a file named symtablehash.c.
• Contain 509 buckets initially.
• Use a reasonable hash function. You are welcome to use this one. See page 578 of the Sedgewick textbook for an alternative.

enum {HASH_MULTIPLIER = 65599};
...
static int SymTable_hash(const char *pcKey, int iBucketCount)

/* Return a hash code for pcKey that is between 0 and iBucketCount-1,
   inclusive.  Adapted from the COS 217 lecture notes. */

{
   int i;
   unsigned int uiHash = 0U;
   for (i = 0; pcKey[i] != '\0'; i++)
      uiHash = uiHash * (unsigned int)HASH_MULTIPLIER
               + (unsigned int)pcKey[i];
   return (int)(uiHash % (unsigned int)iBucketCount);
}

• Validate function parameters by calling the standard assert macro.
• Contain no memory leaks. Your SymTable ADT will use dynamically allocated memory. It should explicitly free all dynamically allocated memory when that memory is no longer required; for each call of malloc or calloc in your ADT, eventually there should be exactly one call of free.
• Avoid gross inefficiencies. In particular, it would be grossly inefficient to hash the given key multiple times when a single time would suffice.
• Expand. That is, for efficiency your implementation should dynamically increase the number of buckets (and, necessarily, reposition all bindings) whenever the number of bindings becomes too large.

Concerning hash table expansion:

• Page 577 of the Sedgewick textbook provides a sequence of integers that are appropriate bucket counts: 509, 1021, 2039, 4093, 8191, 16381, 32749, and 65521. When SymTable_put detects that the new binding count exceeds 509, it should increase the SymTable object's bucket count to 1021. When the function detects that the new binding count exceeds 1021, it should increase the bucket count to 2039. Etc. When SymTable_put detects that the new binding count exceeds 65521, it should not increase the bucket count. Thus 65521 is the maximum number of buckets that the SymTable object should contain.
• You will receive a better grade if you submit a working non-expanding hash table implementation instead of a non-working expanding hash table implementation. If your attempts to develop the expansion code fail, then leave the expansion code in your symtablehash.c file as comments, and describe your attempts in your readme file.

Logistics

Develop on hats using emacs to create source code, make to build, and gdb to debug.

A client program named testsymtable.c is available in the /u/cos217/Assignment3 directory on hats. That program requires you to provide a single command-line argument, which should be an integer that specifies a binding count. The program tests your SymTable ADT by manipulating several SymTable objects. One of those SymTable objects contains the specified number of bindings. The program prints to the standard output stream an indication of how much CPU time it consumed while manipulating that SymTable object. You should use testsymtable.c to test your SymTable ADT. You should create additional test programs, as you deem necessary, to test your SymTable ADT.

Create a makefile. The first dependency rule of the makefile should build two executable files: testsymtablelist (using testsymtable.c and symtablelist.c) and testsymtablehash (using testsymtable.c and symtablehash.c). That is, the first dependency rule of your makefile should be:

all: testsymtablelist testsymtablehash

The makefile that you submit should:

• Maintain object (.o) files to allow for partial builds of both executable binary files.
• Encode the dependencies among the files that comprise your program.
• Build using the gcc217 command, but define a macro that allows the user to switch to the gcc217m command easily.

We recommend that you create your makefile early in your development process. Doing so will allow you to use and test your makefile during development.

Create a readme text file that contains:

• Your name.
• A description of whatever help (if any) you received from others while doing the assignment, and the names of any individuals with whom you collaborated, as prescribed by the course "Policies" web page.
• This statement: "Under penalty of the University's Honor Code, I affirm that I received help from no one on the 'on your own' part of this assignment."
• The CPU times reported by testsymtable.c with binding counts 100, 1000, 10000, 100000, and 1000000 using (1) your linked list implementation, (2) your non-expanding hash table implementation, and (3) your expanding hash table implementation. You can create a non-expanding hash table implementation by temporarily commenting-out your expansion code; don't forget to "comment in" that code before submitting your work. If the time consumed is more than 5 minutes, you may abort execution and report "Greater than 5 minutes."
• (Optionally) An indication of how much time you spent doing the assignment.
• (Optionally) Your assessment of the assignment.
• (Optionally) Any information that will help us to grade your work in the most favorable light. In particular you should describe all known bugs.

Submit your work electronically on hats via the command:

submit 3 symtable.h symtablelist.c symtablehash.c makefile readme

Grading

We will grade your work on quality from the user's point of view and from the programmer's point of view. To encourage good coding practices, we will deduct points if gcc217 generates warning messages.

From the user's point of view, your module has quality if it behaves as it should. The correct behavior of the SymTable ADT is defined by the previous sections of this assignment specification.

In part, style is defined by the rules given in The Practice of Programming (Kernighan and Pike), as summarized by the Rules of Programming Style document. These additional rules apply:

Names: You should use a clear and consistent style for variable and function names. One example of such a style is to prefix each variable name with characters that indicate its type. For example, the prefix c might indicate that the variable is of type char, i might indicate int, pc might mean char*, ui might mean unsigned int, etc. But it is fine to use another style -- a style which does not include the type of a variable in its name -- as long as the result is a readable program.

Line lengths: Limit line lengths in your source code to 72 characters. Doing so allows us to print your work in two columns, thus saving paper.

Comments: Each source code file should begin with a comment that includes your name, the number of the assignment, and the name of the file.

Comments: Each function should begin with a comment that describes what the function does from the caller's point of view. The function comment should:

• Explicitly refer to the function's parameters (by name) and the function's return value.
• State what, if anything, the function reads from standard input or any other stream, and what, if anything, the function writes to standard output, standard error, or any other stream.
• State which global variables the function uses or affects.
• Appear in both the interface (.h) file for the sake of the clients of the function and the implementation (.c) file for the sake of the maintainers of the function.

Comments: Each structure type definition and each structure field definition should have a comment that describes it.

Comments: The interface of an ADT should contain a comment that describes what an object of that type is. It would be reasonable to place that comment adjacent to the definition of the opaque pointer.