.TH JudyHS 3 .SH NAME JudyHS \- C library for creating and accessing a dynamic array, using an array-of-bytes of \fBLength\fP as an \fBIndex\fP and a word as a \fBValue\fP. .PP .SH SYNOPSIS .PP .nf .ps +1 .ft B cc [flags] \fIsourcefiles\fP -lJudy .PP .ft B #include .PP .ft B Word_t * PValue; // JudyHS array element int Rc_int; // return flag Word_t Rc_word; // full word return value Pvoid_t PJHSArray = (Pvoid_t) NULL; // initialize JudyHS array uint8_t * Index; // array-of-bytes pointer Word_t Length; // number of bytes in Index .PP .ft B JHSI( PValue, PJHSArray, Index, Length); // JudyHSIns() JHSD( Rc_int, PJHSArray, Index, Length); // JudyHSDel() JHSG( PValue, PJHSArray, Index, Length); // JudyHSGet() JHSFA(Rc_word, PJHSArray); // JudyHSFreeArray() .ft P .ps .fi .SH DESCRIPTION A JudyHS array is the equivalent of an array of word-sized value/pointers. An \fBIndex\fP is a pointer to an array-of-bytes of specified length: \fBLength\fP. Rather than using a null terminated string, this difference from \fIJudySL\fP(3) allows strings to contain all bits (specifically the null character). This new addition (May 2004) to Judy arrays is a hybird using the best capabilities of hashing and Judy methods. \fBJudyHS\fP does not have a poor performance case where knowledge of the hash algorithm can be used to degrade the performance. .PP Since JudyHS is based on a hash method, \fBIndexes\fP are not stored in any particular order. Therefore the JudyHSFirst(), JudyHSNext(), JudyHSPrev() and JudyHSLast() neighbor search functions are not practical. The \fBLength\fP of each array-of-bytes can be from 0 to the limits of \fImalloc()\fP (about 2GB). .PP The hallmark of \fBJudyHS\fP is speed with scalability, but memory efficiency is excellent. The speed is very competitive with the best hashing methods. The memory efficiency is similar to a linked list of the same \fBIndexes\fP and \fBValues\fP. \fBJudyHS\fP is designed to scale from 0 to billions of \fBIndexes\fP. .PP A JudyHS array is allocated with a \fBNULL\fP pointer .PP .nf .ps +1 Pvoid_t PJHSArray = (Pvoid_t) NULL; .ps .fi .PP Because the macro forms of the API have a simpler error handling interface than the equivalent \fIfunctions\fP, they are the preferred way to use JudyHS. .PP .SH \fBJHSI(PValue, PJHSArray, Index, Length)\fP // \fBJudyHSIns()\fP Given a pointer to a JudyHS array (\fBPJHSArray\fP), insert an \fBIndex\fP string of length: \fBLength\fP and a \fBValue\fP into the JudyHS array: \fBPJHSArray\fP. If the \fBIndex\fP is successfully inserted, the \fBValue\fP is initialized to 0. If the \fBIndex\fP was already present, the \fBValue\fP is not modified. .PP Return \fBPValue\fP pointing to \fBValue\fP. Your program should use this pointer to read or modify the \fBValue\fP, for example: .PP .nf .ps +1 Value = *PValue; *PValue = 1234; .ps .fi .PP \fBNote\fP: \fBJHSI()\fP and \fBJHSD\fP can reorganize the JudyHS array. Therefore, pointers returned from previous \fBJudyHS\fP calls become invalid and must be re-acquired (using \fBJHSG()\fP). .PP .SH \fBJHSD(Rc_int, PJHSArray, Index, Length)\fP // \fBJudyHSDel()\fP Given a pointer to a JudyHS array (\fBPJHSArray\fP), delete the specified \fBIndex\fP along with the \fBValue\fP from the JudyHS array. .PP Return \fBRc_int\fP set to 1 if successfully removed from the array. Return \fBRc_int\fP set to 0 if \fBIndex\fP was not present. .PP .SH \fBJHSG(PValue, PJHSArray, Index, Length)\fP // \fBJudyHSGet()\fP Given a pointer to a JudyHS array (\fBPJHSArray\fP), find \fBValue\fP associated with \fBIndex\fP. .PP Return \fBPValue\fP pointing to \fBIndex\fP's \fBValue\fP. Return \fBPValue\fP set to \fBNULL\fP if the \fBIndex\fP was not present. .PP .SH \fBJHSFA(Rc_word, PJHSArray)\fP // \fBJudyHSFreeArray()\fP Given a pointer to a JudyHS array (\fBPJHSArray\fP), free the entire array. .PP Return \fBRc_word\fP set to the number of bytes freed and \fBPJHSArray\fP set to NULL. .PP .SH \fBERRORS:\fP See: \fIJudy_3.htm#ERRORS\fP .PP .SH EXAMPLES Show how to program with the JudyHS macros. This program will print duplicate lines and their line number from \fIstdin\fP. .PP .PP .nf .ps +1 #include #include #include #include .PP // Compiled: // cc -O PrintDupLines.c -lJudy -o PrintDupLines .PP #define MAXLINE 1000000 /* max fgets length of line */ uint8_t Index[MAXLINE]; // string to check .PP int // Usage: PrintDupLines < file main() { Pvoid_t PJArray = (PWord_t)NULL; // Judy array. PWord_t PValue; // Judy array element pointer. Word_t Bytes; // size of JudyHS array. Word_t LineNumb = 0; // current line number Word_t Dups = 0; // number of duplicate lines .PP while (fgets(Index, MAXLINE, stdin) != (char *)NULL) { LineNumb++; // line number .PP // store string into array JHSI(PValue, PJArray, Index, strlen(Index)); if (PValue == PJERR) // See ERRORS section { fprintf(stderr, "Out of memory -- exit\\n"); exit(1); } if (*PValue == 0) // check if duplicate { Dups++; printf("Duplicate lines %lu:%lu:%s", *PValue, LineNumb, Index); } else { *PValue = LineNumb; // store Line number } } printf("%lu Duplicates, free JudyHS array of %lu Lines\\n", Dups, LineNumb - Dups); JHSFA(Bytes, PJArray); // free JudyHS array printf("JudyHSFreeArray() free'ed %lu bytes of memory\\n", Bytes); return (0); } .ps .fi .PP .SH AUTHOR JudyHS was invented and implemented by Doug Baskins after retiring \-Packard. .PP .SH SEE ALSO \fIJudy\fP(3), \fIJudy1\fP(3), \fIJudyL\fP(3), \fIJudySL\fP(3), .br \fImalloc()\fP, .br the Judy website, \fIhttp://judy.sourceforge.net\fP, for further information and Application Notes.