freebsd-dev/contrib/perl5/ext/SDBM_File/sdbm/sdbm.3

.\" $Id: sdbm.3,v 1.2 90/12/13 13:00:57 oz Exp $
.TH SDBM 3 "1 March 1990"
.SH NAME
sdbm, sdbm_open, sdbm_prep, sdbm_close, sdbm_fetch, sdbm_store, sdbm_delete, sdbm_exists, sdbm_firstkey, sdbm_nextkey, sdbm_hash, sdbm_rdonly, sdbm_error, sdbm_clearerr, sdbm_dirfno, sdbm_pagfno \- data base subroutines
.SH SYNOPSIS
.nf
.ft B
#include <sdbm.h>
.sp
typedef struct {
	char *dptr;
	int dsize;
} datum;
.sp
datum nullitem = { NULL, 0 };
.sp
\s-1DBM\s0 *sdbm_open(char *file, int flags, int mode)
.sp
\s-1DBM\s0 *sdbm_prep(char *dirname, char *pagname, int flags, int mode)
.sp
void sdbm_close(\s-1DBM\s0 *db)
.sp
datum sdbm_fetch(\s-1DBM\s0 *db, key)
.sp
int sdbm_store(\s-1DBM\s0 *db, datum key, datum val, int flags)
.sp
int sdbm_delete(\s-1DBM\s0 *db, datum key)
.sp
int sdbm_exists(\s-1DBM\s0 *db, datum key)
.sp
datum sdbm_firstkey(\s-1DBM\s0 *db)
.sp
datum sdbm_nextkey(\s-1DBM\s0 *db)
.sp
long sdbm_hash(char *string, int len)
.sp
int sdbm_rdonly(\s-1DBM\s0 *db)
int sdbm_error(\s-1DBM\s0 *db)
sdbm_clearerr(\s-1DBM\s0 *db)
int sdbm_dirfno(\s-1DBM\s0 *db)
int sdbm_pagfno(\s-1DBM\s0 *db)
.ft R
.fi
.SH DESCRIPTION
.IX "database library" sdbm "" "\fLsdbm\fR"
.IX sdbm_open "" "\fLsdbm_open\fR \(em open \fLsdbm\fR database"
.IX sdbm_prep "" "\fLsdbm_prep\fR \(em prepare \fLsdbm\fR database"
.IX sdbm_close "" "\fLsdbm_close\fR \(em close \fLsdbm\fR routine"
.IX sdbm_fetch "" "\fLsdbm_fetch\fR \(em fetch \fLsdbm\fR database data"
.IX sdbm_store "" "\fLsdbm_store\fR \(em add data to \fLsdbm\fR database"
.IX sdbm_delete "" "\fLsdbm_delete\fR \(em remove data from \fLsdbm\fR database"
.IX sdbm_exists "" "\fLsdbm_exists\fR \(em test \fLsdbm\fR key existence"
.IX sdbm_firstkey "" "\fLsdbm_firstkey\fR \(em access \fLsdbm\fR database"
.IX sdbm_nextkey "" "\fLsdbm_nextkey\fR \(em access \fLsdbm\fR database"
.IX sdbm_hash "" "\fLsdbm_hash\fR \(em string hash for \fLsdbm\fR database"
.IX sdbm_rdonly "" "\fLsdbm_rdonly\fR \(em return \fLsdbm\fR database read-only mode"
.IX sdbm_error "" "\fLsdbm_error\fR \(em return \fLsdbm\fR database error condition"
.IX sdbm_clearerr "" "\fLsdbm_clearerr\fR \(em clear \fLsdbm\fR database error condition"
.IX sdbm_dirfno "" "\fLsdbm_dirfno\fR \(em return \fLsdbm\fR database bitmap file descriptor"
.IX sdbm_pagfno "" "\fLsdbm_pagfno\fR \(em return \fLsdbm\fR database data file descriptor"
.IX "database functions \(em \fLsdbm\fR"  sdbm_open  ""  \fLsdbm_open\fP
.IX "database functions \(em \fLsdbm\fR"  sdbm_prep  ""  \fLsdbm_prep\fP
.IX "database functions \(em \fLsdbm\fR"  sdbm_close  ""  \fLsdbm_close\fP
.IX "database functions \(em \fLsdbm\fR"  sdbm_fetch  ""  \fLsdbm_fetch\fP
.IX "database functions \(em \fLsdbm\fR"  sdbm_store  ""  \fLsdbm_store\fP
.IX "database functions \(em \fLsdbm\fR"  sdbm_delete  ""  \fLsdbm_delete\fP
.IX "database functions \(em \fLsdbm\fR"  sdbm_firstkey  ""  \fLsdbm_firstkey\fP
.IX "database functions \(em \fLsdbm\fR"  sdbm_nextkey  ""  \fLsdbm_nextkey\fP
.IX "database functions \(em \fLsdbm\fR"  sdbm_rdonly  ""  \fLsdbm_rdonly\fP
.IX "database functions \(em \fLsdbm\fR"  sdbm_error  ""  \fLsdbm_error\fP
.IX "database functions \(em \fLsdbm\fR"  sdbm_clearerr  ""  \fLsdbm_clearerr\fP
.IX "database functions \(em \fLsdbm\fR"  sdbm_dirfno  ""  \fLsdbm_dirfno\fP
.IX "database functions \(em \fLsdbm\fR"  sdbm_pagfno  ""  \fLsdbm_pagfno\fP
.LP
This package allows an application to maintain a mapping of <key,value> pairs
in disk files.  This is not to be considered a real database system, but is
still useful in many simple applications built around fast retrieval of a data
value from a key.  This implementation uses an external hashing scheme,
called Dynamic Hashing, as described by Per-Aake Larson in BIT 18 (1978) pp.
184-201.  Retrieval of any item usually requires a single disk access.
The application interface is compatible with the
.IR ndbm (3)
library.
.LP
An
.B sdbm
database is kept in two files usually given the extensions
.B \.dir
and
.BR \.pag .
The
.B \.dir
file contains a bitmap representing a forest of binary hash trees, the leaves
of which indicate data pages in the
.B \.pag
file.
.LP
The application interface uses the
.B datum
structure to describe both
.I keys
and
.IR value s.
A
.B datum
specifies a byte sequence of
.I dsize
size pointed to by
.IR dptr .
If you use
.SM ASCII
strings as
.IR key s
or
.IR value s,
then you must decide whether or not to include the terminating
.SM NUL
byte which sometimes defines strings.  Including it will require larger
database files, but it will be possible to get sensible output from a
.IR strings (1)
command applied to the data file.
.LP
In order to allow a process using this package to manipulate multiple
databases, the applications interface always requires a
.IR handle ,
a
.BR "DBM *" ,
to identify the database to be manipulated.  Such a handle can be obtained
from the only routines that do not require it, namely
.BR sdbm_open (\|)
or
.BR sdbm_prep (\|).
Either of these will open or create the two necessary files.  The
difference is that the latter allows explicitly naming the bitmap and data
files whereas
.BR sdbm_open (\|)
will take a base file name and call
.BR sdbm_prep (\|)
with the default extensions.
The
.I flags
and
.I mode
parameters are the same as for
.BR open (2).
.LP
To free the resources occupied while a database handle is active, call
.BR sdbm_close (\|).
.LP
Given a handle, one can retrieve data associated with a key by using the
.BR sdbm_fetch (\|)
routine, and associate data with a key by using the
.BR sdbm_store (\|)
routine.
.BR sdbm_exists (\|)
will say whether a given key exists in the database.
.LP
The values of the
.I flags
parameter for
.BR sdbm_store (\|)
can be either
.BR \s-1DBM_INSERT\s0 ,
which will not change an existing entry with the same key, or
.BR \s-1DBM_REPLACE\s0 ,
which will replace an existing entry with the same key.
Keys are unique within the database.
.LP
To delete a key and its associated value use the
.BR sdbm_delete (\|)
routine.
.LP
To retrieve every key in the database, use a loop like:
.sp
.nf
.ft B
for (key = sdbm_firstkey(db); key.dptr != NULL; key = sdbm_nextkey(db))
        ;
.ft R
.fi
.LP
The order of retrieval is unspecified.
.LP
If you determine that the performance of the database is inadequate or
you notice clustering or other effects that may be due to the hashing
algorithm used by this package, you can override it by supplying your
own
.BR sdbm_hash (\|)
routine.  Doing so will make the database unintelligable to any other
applications that do not use your specialized hash function.
.sp
.LP
The following macros are defined in the header file:
.IP
.BR sdbm_rdonly (\|)
returns true if the database has been opened read\-only.
.IP
.BR sdbm_error (\|)
returns true if an I/O error has occurred.
.IP
.BR sdbm_clearerr (\|)
allows you to clear the error flag if you think you know what the error
was and insist on ignoring it.
.IP
.BR sdbm_dirfno (\|)
returns the file descriptor associated with the bitmap file.
.IP
.BR sdbm_pagfno (\|)
returns the file descriptor associated with the data file.
.SH SEE ALSO
.IR open (2).
.SH DIAGNOSTICS
Functions that return a
.B "DBM *"
handle will use
.SM NULL
to indicate an error.
Functions that return an
.B int
will use \-1 to indicate an error.  The normal return value in that case is 0.
Functions that return a
.B datum
will return
.B nullitem
to indicate an error.
.LP
As a special case of
.BR sdbm_store (\|),
if it is called with the
.B \s-1DBM_INSERT\s0
flag and the key already exists in the database, the return value will be 1.
.LP
In general, if a function parameter is invalid,
.B errno
will be set to
.BR \s-1EINVAL\s0 .
If a write operation is requested on a read-only database,
.B errno
will be set to
.BR \s-1ENOPERM\s0 .
If a memory allocation (using
.IR malloc (3))
failed,
.B errno
will be set to
.BR \s-1ENOMEM\s0 .
For I/O operation failures
.B errno
will contain the value set by the relevant failed system call, either
.IR read (2),
.IR write (2),
or
.IR lseek (2).
.SH AUTHOR
.IP "Ozan S. Yigit" (oz@nexus.yorku.ca)
.SH BUGS
The sum of key and value data sizes must not exceed
.B \s-1PAIRMAX\s0
(1008 bytes).
.LP
The sum of the key and value data sizes where several keys hash to the
same value must fit within one bitmap page.
.LP
The
.B \.pag
file will contain holes, so its apparent size is larger than its contents.
When copied through the filesystem the holes will be filled.
.LP
The contents of
.B datum
values returned are in volatile storage.  If you want to retain the values
pointed to, you must copy them immediately before another call to this package.
.LP
The only safe way for multiple processes to (read and) update a database at
the same time, is to implement a private locking scheme outside this package
and open and close the database between lock acquisitions.  It is safe for
multiple processes to concurrently access a database read-only.
.SH APPLICATIONS PORTABILITY
For complete source code compatibility with the Berkeley Unix
.IR ndbm (3)
library, the 
.B sdbm.h
header file should be installed in
.BR /usr/include/ndbm.h .
.LP
The
.B nullitem
data item, and the
.BR sdbm_prep (\|),
.BR sdbm_hash (\|),
.BR sdbm_rdonly (\|),
.BR sdbm_dirfno (\|),
and
.BR sdbm_pagfno (\|)
functions are unique to this package.