C library for creating and accessing a dynamic array of words, using a word as an index.
cc [flags] sourcefiles -lJudy #include <Judy.h> int Rc_int; // return code - integer Word_t Rc_word; // return code - unsigned word Word_t Index, Index1, Index2, Nth; PWord_t PValue; // pointer to return value Pvoid_t PJLArray = (Pvoid_t) NULL; // initialize JudyL array JLI( PValue, PJLArray, Index); // JudyLIns() JLD( Rc_int, PJLArray, Index); // JudyLDel() JLG( PValue, PJLArray, Index); // JudyLGet() JLC( Rc_word, PJLArray, Index1, Index2); // JudyLCount() JLBC(PValue, PJLArray, Nth, Index); // JudyLByCount() JLFA(Rc_word, PJLArray); // JudyLFreeArray() JLMU(Rc_word, PJLArray); // JudyLMemUsed() JLF( PValue, PJLArray, Index); // JudyLFirst() JLN( PValue, PJLArray, Index); // JudyLNext() JLL( PValue, PJLArray, Index); // JudyLLast() JLP( PValue, PJLArray, Index); // JudyLPrev() JLFE(Rc_int, PJLArray, Index); // JudyLFirstEmpty() JLNE(Rc_int, PJLArray, Index); // JudyLNextEmpty() JLLE(Rc_int, PJLArray, Index); // JudyLLastEmpty() JLPE(Rc_int, PJLArray, Index); // JudyLPrevEmpty()
A JudyL array is the equivalent of an array of word-sized values. A Value is addressed by an Index (key). The array may be sparse, and the Index may be any word-sized number. Memory to support the array is allocated as index/value pairs are inserted, and released as index/value pairs are deleted. A JudyL array can also be thought of as a mapper, that is "map" a word to another word/pointer.
As with an ordinary array, there are no duplicate indexes in a JudyL array.
The value may be used as a scalar, or a pointer to a structure or block of data (or even another Judy array).
A JudyL array is allocated with a NULL pointer
Pvoid_t PJLArray = (Pvoid_t) NULL;
Using the macros described here, rather than the JudyL function calls, the default error handling sends a message to the standard error and terminates the program with exit(1);. For other error handling methods, see the ERRORS section. JLI( PValue, PJLArray, Index); // JudyLIns()
Because the macro forms are sometimes faster and have a simpler error handling interface than the equivalent JudyL functions, they are the preferred way of calling the JudyL functions.
JLI(PValue, PJLArray, Index) // JudyLIns()
Insert an Index and Value into the JudyL array PJLArray. If the Index is successfully inserted, the Value is initialized to 0. If the Index was already present, the Value is not modified.
Return PValue pointing to Value. Your program can use this pointer to read or modify Value until the next JLI() (insert), JLD() (delete) or JLFA() (freearray) is executed on PJLArray. Examples:
*PValue = 1234; Value = *PValue;
Return PValue set to PJERR if a malloc() fail occured. Note: JLI() and JLD() reorganize the JudyL array. Therefore, PValue returned from previous JudyL calls become invalid and must be re-acquired.
JLD(Rc_int, PJLArray, Index) // JudyLDel()
Delete the Index/Value pair from the JudyL array.
Return Rc_int set to 1 if successful. Return Rc_int set to 0 if Index was not present. Return Rc_int set to JERR if a malloc() fail occured.
JLG(PValue, PJLArray, Index) // JudyLGet()
Get the pointer PValue associated with Index in the PJLArray Judy array.
Return PValue pointing to Value. Return PValue set to NULL if the Index was not present. Return PValue set to PJERR if a malloc() fail occured.
JLC(Rc_word, PJLArray, Index1, Index2) // JudyLCount()
Count the number of indexes present in the JudyL array PJLArray between Index1 and Index2 (inclusive).
Return Rc_word set to the count. A return value of 0 can be valid as a count.
To count all indexes present in a JudyL array, use:
JLC(Rc_word, PJLArray, 0, -1);
JLBC(PValue, PJLArray, Nth, Index) // JudyLByCount()
Locate the Nth index that is present in the JudyL array PJLArray (Nth = 1 returns the first index present).
Return PValue pointing to its Value and Index set to the Nth index if found, otherwise return PValue set to NULL (the value of Index is undefined).
JLFA(Rc_word, PJLArray) // JudyLFreeArray()
Given a pointer to a JudyL array, free the entire array (much faster than using a JLN(), JLD() loop).
Return Rc_word set to the number of bytes freed and PJLArray set to NULL.
JLMU(Rc_word, PJLArray) // JudyLMemUsed()
Return Rc_word set to the number of bytes of memory malloc()'ed by PJLArray. This is a very fast routine, and may be used before and after a JLI() or JLD() call with little performance impact.
JudyL Search Functions
JLF(), JLN(), JLL(), JLP() allow you to search for indexes in the array. You may search inclusively or exclusively, in either forward or reverse directions. If successful, Index is returned set to the found index, and PValue is returned set to a pointer to Index's Value. If unsuccessful, PValue is returned set to NULL, and Index contains no useful information. PValue must be tested for non-NULL prior to using Index, since a search failure is possible.
JLFE(), JLNE(), JLLE(), JLPE() allow you to search for indexes that are not present ("empty") in the array. You may search inclusively or exclusively, in either forward or reverse directions. If successful, Index is returned set to a not present ("empty") index, and Rc_int is returned set to 1. If unsuccessful, Rc_int is returned set to 0, and and Index contains no useful information. Rc_int must be checked prior to using Index, since a search failure is possible.
JLF(PValue, PJLArray, Index) // JudyLFirst()
Search (inclusive) for the first index present that is equal to or greater than the passed Index. (Start with Index = 0 to find the first index in the array.) JLF() is typically used to begin a sorted-order scan of the indexes present in a JudyL array.
JLN(PValue, PJLArray, Index) // JudyLNext()
Search (exclusive) for the next index present that is greater than the passed Index. JLN() is typically used to continue a sorted-order scan of the indexes present in a JudyL array, or to locate a "neighbor" of a given index.
JLL(PValue, PJLArray, Index) // JudyLLast()
Search (inclusive) for the last index present that is equal to or less than the passed Index. (Start with Index = -1, that is, all ones, to find the last index in the array.) JLL() is typically used to begin a reverse-sorted-order scan of the indexes present in a JudyL array.
JLP(PValue, PJLArray, Index) // JudyLPrev()
Search (exclusive) for the previous index present that is less than the passed Index. JLP() is typically used to continue a reverse-sorted-order scan of the indexes present in a JudyL array, or to locate a "neighbor" of a given index.
JLFE(Rc_int, PJLArray, Index) // JudyLFirstEmpty()
Search (inclusive) for the first index absent that is equal to or greater than the passed Index. (Start with Index = 0 to find the first index absent in the array.)
JLNE(Rc_int, PJLArray, Index) // JudyLNextEmpty()
Search (exclusive) for the next index absent that is greater than the passed Index.
JLLE(Rc_int, PJLArray, Index) // JudyLLastEmpty()
Search (inclusive) for the last index absent that is equal to or less than the passed Index. (Start with Index = -1, that is, all ones, to find the last index absent in the array.)
JLPE(Rc_int, PJLArray, Index) // JudyLPrevEmpty()
Search (exclusive) for the previous index absent that is less than the passed Index.
Storing a pointer to another JudyL array in a JudyL array's Value is a simple way to support dynamic multi-dimensional arrays. These arrays (or trees) built using JudyL arrays are very fast and memory efficient. (In fact, that is how JudySL and JudyHS are implemented). An arbitrary number of dimensions can be realized this way. To terminate the number of dimensions (or tree), the Value pointer is marked to NOT point to another Judy array. A JLAP_INVALID flag is used in the least significant bit(s) of the pointer. After the flag JLAP_INVALID is removed, it is used as a pointer to the users data. The Judy.h header file defines JLAP_INVALID. See code fragment below.
Note: The current version of Judy.h changed this flag from 0x4 to 0x1 to allow for a malloc() that does not deliver memory on an 8 byte aligned boundry (such as old versions of valgrind).
The following example code segment can be used to determine whether or not a pointer points to another JudyL:
PValue = (PWord_t)PMultiDimArray;
for (Dim = 0; ;Dim++)
{
if (PValue == (PWord_t)NULL) goto IndexNotFound;
/* Advance to next dimension in array */
JLG(PValue, (Pvoid_t)*PValue, Index[Dim]);
/* Check if pointer to user buffer: */
if (*PValue & JLAP_INVALID)) break;
}
UPointer = (UPointer_t) (*PValue & ~JLAP_INVALID); // mask and cast.
printf("User object pointer is 0x%lx\n", (Word_t) UPointer);
&.&.&.
Note: This works because malloc() guarantees to return a pointer with the least bit(s) == 0x0. You must remove JLAP_INVALID before using the pointer.
Read a series of index/value pairs from the standard input, store in a JudyL array, and then print out in sorted order.
#include <stdio.h>
#include <Judy.h>
Word_t Index; // array index
Word_t Value; // array element value
Word_t * PValue; // pointer to array element value
int Rc_int; // return code
Pvoid_t PJLArray = (Pvoid_t) NULL; // initialize JudyL array
while (scanf("%lu %lu", &Index, &Value))
{
JLI(PValue, PJLArray, Index);
If (PValue == PJERR) goto process_malloc_failure;
*PValue = Value; // store new value
}
// Next, visit all the stored indexes in sorted order, first ascending,
// then descending, and delete each index during the descending pass.
Index = 0;
JLF(PValue, PJLArray, Index);
while (PValue != NULL)
{
printf("%lu %lu\n", Index, *PValue));
JLN(PValue, PJLArray, Index);
}
Index = -1;
JLL(PValue, PJLArray, Index);
while (PValue != NULL)
{
printf("%lu %lu\n", Index, *PValue));
JLD(Rc_int, PJLArray, Index);
if (Rc_int == JERR) goto process_malloc_failure;
JLP(PValue, PJLArray, Index);
}
Judy was invented by Doug Baskins and implemented -Packard.