.\"	$OpenBSD: ohash_init.3,v 1.6 2003/05/07 08:52:43 jmc Exp $
.\"
.\" Copyright (c) 1999 Marc Espie.
.\"
.\" Code written for the OpenBSD project.
.\"
.\" Redistribution and use in source and binary forms, with or without
.\" modification, are permitted provided that the following conditions
.\" are met:
.\" 1. Redistributions of source code must retain the above copyright
.\"    notice, this list of conditions and the following disclaimer.
.\" 2. Redistributions in binary form must reproduce the above copyright
.\"    notice, this list of conditions and the following disclaimer in the
.\"    documentation and/or other materials provided with the distribution.
.\"
.\" THIS SOFTWARE IS PROVIDED BY THE OPENBSD PROJECT AND CONTRIBUTORS
.\" ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
.\" LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
.\" A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE OPENBSD
.\" PROJECT OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
.\" SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
.\" DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
.\" THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
.\" (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
.\" OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
.\"
.Dd November 3, 1999
.Dt OPEN_HASH 3
.Os
.Sh NAME
.Nm ohash_init ,
.Nm ohash_delete ,
.Nm ohash_lookup_interval ,
.Nm ohash_lookup_memory ,
.Nm ohash_find ,
.Nm ohash_remove ,
.Nm ohash_insert ,
.Nm ohash_first ,
.Nm ohash_next ,
.Nm ohash_entries
.Nd light-weight open hashing
.Sh SYNOPSIS
.Fd #include <sys/types.h>
.Fd #include <stddef.h>
.Fd #include <ohash.h>
.Ft void
.Fn ohash_init "struct ohash *h" "unsigned int size" "struct ohash_info *info"
.Ft void
.Fn ohash_delete "struct ohash *h"
.Ft "unsigned int"
.Fn ohash_lookup_interval "struct ohash *h" "const char *start" "const char *end" "u_int32_t hv"
.Ft "unsigned int"
.Fn ohash_lookup_memory "struct ohash *h" "const char *k" "size_t s" "u_int32_t hv"
.Ft void *
.Fn ohash_find "struct ohash *h" "unsigned int i"
.Ft void *
.Fn ohash_remove "struct ohash *h" "unsigned int i"
.Ft void *
.Fn ohash_insert "struct ohash *h" "unsigned int i" "void *p"
.Ft void *
.Fn ohash_first "struct ohash *h" "unsigned int *i"
.Ft void *
.Fn ohash_next "struct ohash *h" "unsigned int *i"
.Ft "unsigned int"
.Fn ohash_entries "struct ohash *h"
.Sh DESCRIPTION
Those functions have been designed as a fast, extensible alternative to
the usual hash table functions.
They provide storage and retrieval of records indexed by keys,
where a key is a contiguous sequence of bytes at a fixed position in
each record.
Keys can either be null-terminated strings or fixed-size memory areas.
All functions take a pointer to an ohash structure as the
.Fa h
function argument.
Storage for this structure should be provided by user code.
.Pp
.Fn ohash_init
initializes the table to store roughly 2 to the power
.Fa size
elements.
.Fa info
holds the position of the key in each record, and two pointers to
.Xr calloc 3
and
.Xr free 3 Ns No -like
functions, to use for managing the table internal storage.
.Pp
.Fn ohash_delete
frees storage internal to
.Fa h .
Elements themselves should be freed by the user first, using for instance
.Fn ohash_first
and
.Fn ohash_next .
.Pp
.Fn ohash_lookup_interval
and
.Fn ohash_lookup_memory
are the basic look-up element functions.
The hashing function result is provided by the user as
.Fa hv .
These return a
.Qq slot
in the ohash table
.Fa h ,
to be used with
.Fn ohash_find ,
.Fn ohash_insert ,
or
.Fn ohash_remove .
This slot is only valid up to the next call to
.Fn ohash_insert
or
.Fn ohash_remove .
.Pp
.Fn ohash_lookup_interval
handles string-like keys.
.Fn ohash_lookup_interval
assumes the key is the interval between
.Fa start
and
.Fa end ,
exclusive,
though the actual elements stored in the table should only contain
null-terminated keys.
.Pp
.Fn ohash_lookup_memory
assumes the key is the memory area starting at
.Fa k
of size
.Fa s .
All bytes are significant in key comparison.
.Pp
.Fn ohash_find
retrieves an element from a slot
.Fa i
returned by the
.Fn ohash_lookup*
functions.
It returns
.Dv NULL
if the slot is empty.
.Pp
.Fn ohash_insert
inserts a new element
.Fa p
at slot
.Fa i .
Slot
.Fa i
must be empty and element
.Fa p
must have a key corresponding to the
.Fn ohash_lookup*
call.
.Pp
.Fn ohash_remove
removes the element at slot
.Fa i .
It returns the removed element, for user code to dispose of, or
.Dv NULL
if the slot was empty.
.Pp
.Fn ohash_first
and
.Fn ohash_next
can be used to access all elements in an ohash table, like this:
.Pp
.Bd -literal
	for (n = ohash_first(h, &i); n != NULL; n = ohash_next(h, &i))
		do_something_with(n);
.Ed
.Pp
.Fa i
points to an auxiliary unsigned integer used to record the current position
in the ohash table.
Those functions are safe to use even while entries are added to/removed
from the table, but in such a case they don't guarantee that new entries
will be returned.
As a special case, they can safely be used to free elements in the table.
.Pp
.Fn ohash_entries
returns the number of elements in the hash table.
.Sh STORAGE HANDLING
Only
.Fn ohash_init ,
.Fn ohash_insert ,
.Fn ohash_remove
and
.Fn ohash_delete
may call the user-supplied memory functions.
It is the responsibility of the user memory allocation code to verify
that those calls did not fail.
.Pp
If memory allocation fails,
.Fn ohash_init
returns a useless hash table.
.Fn ohash_insert
and
.Fn ohash_remove
still perform the requested operation, but the returned table should be
considered read-only.
It can still be accessed by
.Fn ohash_lookup* ,
.Fn ohash_find ,
.Fn ohash_first
and
.Fn ohash_next
to dump relevant information to disk before aborting.
.Sh THREAD SAFETY
The open hashing functions are not thread-safe by design.
In particular, in a threaded environment, there is no guarantee that a
.Qq slot
will not move between a
.Fn ohash_lookup*
and a
.Fn ohash_find ,
.Fn ohash_insert
or
.Fn ohash_remove
call.
.Pp
Multi-threaded applications should explicitly protect ohash table access.
.Sh SEE ALSO
.Rs
.%A Donald E. Knuth
.%B The Art of Computer Programming
.%V Vol. 3
.%P pp 506-550
.%D 1973
.Re
.Xr ohash_interval 3
.Sh STANDARDS
Those functions are completely non-standard and should be avoided in
portable programs.
.Sh HISTORY
Those functions were designed and written for
.Ox
.Xr make 1
by Marc Espie in 1999.