Msql(3)
NAME
Msql / Mysql - Perl interfaces to the mSQL and mysql databases
SYNOPSIS
use Msql;
$dbh = Msql->connect($host);
$dbh = Msql->connect($host, $database);
or
use Mysql;
$dbh = Mysql->connect(undef, $database, $user, $password);
$dbh = Mysql->connect($host, $database, $user, $password);
or
$dbh = Msql1->connect($host);
$dbh = Msql1->connect($host, $database);
$dbh->selectdb($database);
@arr = $dbh->listdbs;
@arr = $dbh->listtables;
$quoted_string = $dbh->quote($unquoted_string);
$error_message = $dbh->errmsg;
$error_number = $dbh->errno; # MySQL only
$sth = $dbh->listfields($table);
$sth = $dbh->query($sql_statement);
@arr = $sth->fetchrow; # Array context
$firstcol = $sth->fetchrow; # Scalar context
@arr = $sth->fetchcol($col_number);
%hash = $sth->fetchhash;
$sth->dataseek($row_number);
$sth->as_string;
@indices = $sth->listindices # only in mSQL 2.0
@arr = $dbh->listindex($table,$index) # only in mSQL 2.0
($step,$value) = $dbh->getsequenceinfo($table) # only in mSQL 2.0
$rc = $dbh->shutdown();
$rc = $dbh->createdb($database);
$rc = $dbh->dropdb($database);
OBSOLETE SOFTWARE
As of Msql-Mysql-modules 1.19_10 M(y)sqlPerl is no longer a separate
module. Instead it is emulated using the DBI drivers. You are strongly
encouraged to implement new code with DBI directly. See "COMPATIBILITY
NOTES" below.
DESCRIPTION
This package is designed as close as possible to its C API counterpart.
The manual that comes with mSQL or MySQL describes most things you
need. Due to popular demand it was decided though, that this interface
does not use StudlyCaps (see below).
As of March 1998, the Msql and Mysql modules are obsoleted by the DBI
drivers DBD::mSQL and DBD::mysql, respectively. You are strongly
encouraged to implement new code with the DBI drivers. In fact, Msql
and Mysql are currently implemented as emulations on top of the DBI
drivers.
Internally you are dealing with the two classes "Msql" and
"Msql::Statement" or "Mysql" and "Mysql::Statement", respectively. You
will never see the latter, because you reach it through a statement
handle returned by a query or a listfields statement. The only class
you name explicitly is Msql or Mysql. They offer you the connect com-
mand:
$dbh = Msql->connect($host);
$dbh = Msql->connect($host, $database);
or
$dbh = Mysql->connect($host, undef, $user, $password);
$dbh = Mysql->connect($host, $database, $user, $password);
or
$dbh = Msql1->connect($host);
$dbh = Msql1->connect($host, $database);
This connects you with the desired host/database. With no argument or
with an empty string as the first argument it connects to the UNIX
socket, which has a much better performance than the TCP counterpart. A
database name as the second argument selects the chosen database within
the connection. The return value is a database handle if the connect
succeeds, otherwise the return value is undef.
You will need this handle to gain further access to the database.
$dbh->selectdb($database);
If you have not chosen a database with the "connect" command, or if you
want to change the connection to a different database using a database
handle you have got from a previous "connect", then use selectdb.
$sth = $dbh->listfields($table);
$sth = $dbh->query($sql_statement);
These two work rather similar as descibed in the mSQL or MySQL manual.
They return a statement handle which lets you further explore what the
server has to tell you. On error the return value is undef. The object
returned by listfields will not know about the size of the table, so a
numrows() on it will return the string "N/A";
@arr = $dbh->listdbs();
@arr = $dbh->listtables;
An array is returned that contains the requested names without any fur-
ther information.
@arr = $sth->fetchrow;
returns an array of the values of the next row fetched from the server.
Be carefull with context here! In scalar context the method behaves
different than expected and returns the first column:
$firstcol = $sth->fetchrow; # Scalar context!
Similar does
%hash = $sth->fetchhash;
return a complete hash. The keys in this hash are the column names of
the table, the values are the table values. Be aware, that when you
have a table with two identical column names, you will not be able to
use this method without trashing one column. In such a case, you should
use the fetchrow method.
@arr = $sth->fetchcol($colnum);
returns an array of the values of each row for column $colnum. Note
that this reads the entire table and leaves the row offset at the end
of the table; be sure to use $sth->dataseek() to reset it if you want
to re-examine the table.
$sth->dataseek($row_number);
lets you specify a certain offset of the data associated with the
statement handle. The next fetchrow will then return the appropriate
row (first row being 0).
No close statement
Whenever the scalar that holds a database or statement handle loses its
value, Msql chooses the appropriate action (frees the result or closes
the database connection). So if you want to free the result or close
the connection, choose to do one of the following:
undef the handle
use the handle for another purpose
let the handle run out of scope
exit the program.
Error messages
Both drivers, Msql and Mysql implement a method ->errmsg(), which
returns a textual error message. Mysql additionally supports a method
->errno returning the corresponding error number.
Usually you do fetch error messages with
$errmsg = $dbh->errmsg();
In situations where a $dbh is not available (for example when connect()
failed) you may instead do a
$errmsg = Msql->errmsg();
or
$errmsg = Mysql->errmsg();
or
$errmsg = Msql1->errmsg();
The "-w" switch
With Msql and Mysql the "-w" switch is your friend! If you call your
perl program with the "-w" switch you get the warnings from ->errmsg on
STDERR. This is a handy method to get the error messages from the msql
server without coding it into your program.
If you want to know in greater detail what's going on, set the environ-
ment variables that are described in David's manual. David's debugging
aid is excellent, there's nothing to be added.
By default errors are printed as warnings. You can suppress this behav-
iour by using the PrintError attribute of the respective handles:
$dbh->{'dbh'}->{'PrintError'} = 0;
->quote($str [, $length])
returns the argument enclosed in single ticks ('') with any special
character escaped according to the needs of the API.
For mSQL this means, any single tick within the string is escaped with
a backslash and backslashes are doubled. Currently (as of msql-1.0.16)
the API does not allow to insert NUL's (ASCII 0) into tables. The quote
method does not fix this deficiency.
MySQL allows NUL's or any other kind of binary data in strings. Thus
the quote method will additionally escape NUL's as \0.
If you pass undefined values to the quote method, it returns the string
"NULL".
If a second parameter is passed to "quote", the result is truncated to
that many characters.
NULL fields
NULL fields in tables are returned to perl as undefined values.
Metadata
Now lets reconsider the above methods with regard to metadata.
Database Handle
As said above you get a database handle with the connect() method. The
database handle knows about the socket, the host, and the database it
is connected to.
You get at the three values with the methods
$scalar = $dbh->sock;
$scalar = $dbh->host;
$scalar = $dbh->database;
Mysql additionally supports
$scalar = $dbh->user;
$scalar = $dbh->sockfd;
where the latter is the file descriptor of the socket used by the data-
base connection. This is the same as $dbh->sock for mSQL.
Statement Handle
Two constructor methods return a statement handle:
$sth = $dbh->listfields($table);
$sth = $dbh->query($sql_statement);
$sth knows about all metadata that are provided by the API:
$scalar = $sth->numrows;
$scalar = $sth->numfields;
@arr = $sth->table; the names of the tables of each column
@arr = $sth->name; the names of the columns
@arr = $sth->type; the type of each column, defined in msql.h
and accessible via Msql::CHAR_TYPE,
&Msql::INT_TYPE, &Msql::REAL_TYPE or
&Mysql::FIELD_TYPE_STRING,
&Mysql::FIELD_TYPE_LONG, ...
@arr = $sth->isnotnull; array of boolean
@arr = $sth->isprikey; array of boolean
@arr = $sth->isnum; array of boolean
@arr = $sth->length; array of the possibble maximum length of each
field in bytes
@arr = $sth->maxlength; array of the actual maximum length of each field
in bytes. Be careful when using this attribute
under MsqlPerl: The server doesn't offer this
attribute, thus it is calculated by fetching
all rows. This might take a long time and you
might need to call $sth->dataseek.
Mysql additionally supports
$scalar = $sth->affectedrows number of rows in database affected by query
$scalar = $sth->insertid the unique id given to a auto_increment field.
$string = $sth->info() more info from some queries (ALTER TABLE...)
$arrref = $sth->isblob; array of boolean
The array methods (table, name, type, is_not_null, is_pri_key, length,
affected_rows, is_num and blob) return an array in array context and an
array reference (see perlref and perlldsc for details) when called in a
scalar context. The scalar context is useful, if you need only the name
of one column, e.g.
$name_of_third_column = $sth->name->[2]
which is equivalent to
@all_column_names = $sth->name;
$name_of_third_column = $all_column_names[2];
New in mSQL 2.0
The query() function in the API returns the number of rows affected by
a query. To cite the mSQL API manual, this means...
If the return code is greater than 0, not only does it imply
success, it also indicates the number of rows "touched" by the query
(i.e. the number of rows returned by a SELECT, the number of rows
modified by an update, or the number of rows removed by a delete).
As we are returning a statement handle on selects, we can easily check
the number of rows returned. For non-selects we behave just the same as
mSQL-2.
To find all indices associated with a table you can call the
"listindices()" method on a statement handle. To find out the columns
included in an index, you can call the "listindex($table,$index)"
method on a database handle.
There are a few new column types in mSQL 2. You can access their
numeric value with these functions defined in the Msql package:
IDENT_TYPE, NULL_TYPE, TEXT_TYPE, DATE_TYPE, UINT_TYPE, MONEY_TYPE,
TIME_TYPE, IDX_TYPE, SYSVAR_TYPE.
You cannot talk to a 1.0 server with a 2.0 client.
You cannot link to a 1.0 library and to a 2.0 library at the same time.
So you may want to build two different Msql modules at a time, one for
1.0, another for 2.0, and load whichever you need. Check out what the
"-I" switch in perl is for.
Everything else seems to remain backwards compatible.
@EXPORT
For historical reasons the constants CHAR_TYPE, INT_TYPE, and REAL_TYPE
are in @EXPORT instead of @EXPORT_OK. This means, that you always have
them imported into your namespace. I consider it a bug, but not such a
serious one, that I intend to break old programs by moving them into
EXPORT_OK.
Displaying whole tables in one go
A handy method to show the complete contents of a statement handle is
the as_string method. This works similar to the msql monitor with a few
exceptions:
the width of a column
is calculated by examining the width of all entries in that column
control characters
are mapped into their backslashed octal representation
backslashes
are doubled ("\\ instead of \")
numeric values
are adjusted right (both integer and floating point values)
The differences are illustrated by the following table:
Input to msql (a real carriage return here replaced with ^M):
CREATE TABLE demo (
first_field CHAR(10),
second_field INT
) \g
INSERT INTO demo VALUES ('new
line',2)\g
INSERT INTO demo VALUES ('back\\slash',1)\g
INSERT INTO demo VALUES ('cr^Mcrnl
nl',3)\g
Output of msql:
+-------------+--------------+
| first_field | second_field |
+-------------+--------------+
| new
line | 2 |
| back\slash | 1 |
crnlr
nl | 3 |
+-------------+--------------+
Output of pmsql:
+----------------+------------+
|first_field |second_field|
+----------------+------------+
|new\012line | 2|
|back\\slash | 1|
|cr\015crnl\012nl| 3|
+----------------+------------+
Version information
The version of Msql and Mysql is always stored in $Msql::VERSION or
$Mysql::VERSION as it is perl standard.
The mSQL API implements methods to access some internal configuration
parameters: gethostinfo, getserverinfo, and getprotoinfo. All three
are available both as class methods or via a database handle. But under
no circumstances they are associated with a database handle. All three
return global variables that reflect the last connect() command within
the current program. This means, that all three return empty strings or
zero before the first call to connect().
This situation is better with MySQL: The methods are valid only in con-
nection with a database handle.
Administration
shutdown, createdb, dropdb, reloadacls are all accessible via a data-
base handle and implement the corresponding methods to what msqladmin
does.
The mSQL and MySQL engines do not permit that these commands are
invoked by users without sufficient privileges. So please make sure to
check the return and error code when you issue one of them.
$rc = $dbh->shutdown();
$rc = $dbh->createdb($database);
$rc = $dbh->dropdb($database);
It should be noted that database deletion is not prompted for in any
way. Nor is it undo-able from within Perl.
B<Once you issue the dropdb() method, the database will be gone!>
These methods should be used at your own risk.
StudlyCaps
Real Perl Programmers (C) usually don't like to type ListTables but
prefer list_tables or listtables. The mSQL API uses StudlyCaps every-
where and so did early versions of MsqlPerl. Beginning with $VERSION
1.06 all methods are internally in lowercase, but may be written how-
ever you please. Case is ignored and you may use the underline to
improve readability.
The price for using different method names is neglectible. Any method
name you use that can be transformed into a known one, will only be
defined once within a program and will remain an alias until the pro-
gram terminates. So feel free to run fetch_row or connecT or ListDBs as
in your old programs. These, of course, will continue to work.
PREREQUISITES
mSQL is a database server and an API library written by David Hughes.
To use the adaptor you definitely have to install these first.
MySQL is a libmysqlclient.a library written by Michael Widenius This
was originally inspired by MySQL.
COMPATIBILITY NOTES
M(y)sql used to be a separate module written in C. This is no longer
the case, instead the old modules are emulated by their corresponding
DBI drivers. I did my best to remove any incompatibilities, but the
following problems are known to remain:
Static methods
For whatever reason, mSQL implements some functions independent
from the respective database connection that really depend on it.
This made it possible to implement
Msql->errmsg
or
Msql->getserverinfo
as static methods. This is no longer the case, it never was for
MysqlPerl. Instead you have to use
$dbh->errmsg
or
$dbh->getserverinfo
$M(Y)SQL::QUIET
This variable used to turn off the printing of error messages.
Unfortunately DBI uses a completely different mechanism for that:
The "PrintError" attribute of the database and/or statement han-
dles. We try to emulate the old behaviour by setting the "PrintEr-
ror" attribute to the current value of $M(Y)SQL::QUIET when a han-
dle is created, that is when M(y)sql->connect or $dbh->query() are
called.
You can overwrite this by using something like
$dbh->{'dbh'}->{'PrintError'} = 1;
or
$sth->{'PrintError'} = 0;
AUTHORS
Andreas Koenig "koenig@franz.ww.TU-Berlin.DE" wrote the original
MsqlPerl. Jochen Wiedmann wrote the M(y)sqlPerl emulation using DBI.
SEE ALSO
Alligator Descartes wrote a database driver for Tim Bunce's DBI. I rec-
ommend anybody to carefully watch the development of this module
("DBD::mSQL"). Msql is a simple, stable, and fast module, and it will
be supported for a long time. But it's a dead end. I expect in the
medium term, that the DBI efforts result in a richer module family with
better support and more functionality. Alligator maintains an interest-
ing page on the DBI development:
http://www.symbolstone.org/technology/perl/DBI
perl v5.8.6 2004-07-13 Mysql(3)
Man(1) output converted with
man2html