NAME

APR::Bucket - Perl API for manipulating APR Buckets

Synopsis

use APR::Bucket ();
my $ba = $c->bucket_alloc;

$ba2 = APR::Bucket::alloc_create($pool);
APR::Bucket::alloc_destroy($ba2);

$b1 = APR::Bucket->new($ba, "aaa");
$b2 = APR::Bucket::eos_create($ba);
$b3 = APR::Bucket::flush_create($ba);

$b2->is_eos;
$b3->is_flush;

$len = $b1->length;
$len = $b1->read($data);
$type = $b1->type;

$b1->insert_after($b2);
$b1->insert_before($b3);
$b1->remove;
$b1->destroy;

$b2->delete; # remove+destroy

$b4 = APR::Bucket->new($ba, "to be setaside");
$b4->setaside($pool);

Description

APR::Bucket allows you to create, manipulate and delete APR buckets.

You will probably find the various insert methods confusing, the tip is to read the function right to left. The following code sample helps to visualize the operations:

my $bb = APR::Brigade->new($r->pool, $ba);
my $d1 = APR::Bucket->new($ba, "d1");
my $d2 = APR::Bucket->new($ba, "d2");
my $f1 = APR::Bucket::flush_create($ba);
my $f2 = APR::Bucket::flush_create($ba);
my $e1 = APR::Bucket::eos_create($ba);
                         # head->tail
$bb->insert_head(  $d1); # head->d1->tail
$d1->insert_after( $d2); # head->d1->d2->tail
$d2->insert_before($f1); # head->d1->f1->d2->tail
$d2->insert_after( $f2); # head->d1->f1->d2->f2->tail
$bb->insert_tail(  $e1); # head->d1->f1->d2->f2->e1->tail

API

APR::Bucket provides the following functions and/or methods:

delete

Tell the bucket to remove itself from the bucket brigade it belongs to, and destroy itself.

$bucket->delete();
obj: $bucket ( APR::Bucket object )
ret: no return value
since: 1.99_16

If the bucket is not attached to any bucket brigade then this operation just destroys the bucket.

delete is a convenience wrapper, internally doing:

$b->remove;
$b->destroy;

Examples:

Assuming that $bb already exists and filled with buckets, replace the existing data buckets with new buckets with upcased data;

for (my $b = $bb->first; $b; $b = $bb->next($b)) {
   if ($b->read(my $data)) {
        my $nb = APR::Bucket->new($bb->bucket_alloc, uc $data);
        $b->insert_before($nb);
        $b->delete;
        $b = $nb;
    }
}

destroy

Free the resources used by a bucket. If multiple buckets refer to the same resource it is freed when the last one goes away.

$bucket->destroy();
obj: $bucket ( APR::Bucket object )
ret: no return value
since: 1.99_16

A bucket needs to be destroyed if it was removed from a bucket brigade, to avoid memory leak.

If a bucket is linked to a bucket brigade, it needs to be removed from it, before it can be destroyed.

Usually instead of calling:

$b->remove;
$b->destroy;

it's better to call delete which does exactly that.

alloc_create

Create an APR::BucketAlloc object:

$ba = APR::Bucket::alloc_create($pool);
arg1: $pool ( APR::Pool object )

The pool used to create this object.

ret: $ba ( APR::BucketAlloc object )

The new object.

since: 1.99_17

This bucket allocation list (freelist) is used to create new buckets and bucket brigades. Normally it is not necesssary to create them, since the existing bucket brigades and/or connection objects in modperl 2.0 provide them automatically.

Example:

use APR::Bucket ();
use Apache::Connection ();
my $ba = APR::Bucket::alloc_create($c->pool);
my $eos_b = APR::Bucket::eos_create($ba);

alloc_destroy

Destroy an APR::BucketAlloc object:

APR::Bucket::alloc_destroy($ba);
arg1: $ba ( APR::BucketAlloc object )

The freelist to destroy.

ret: no return value
since: 1.99_17

Once destroyed this object may not be used again. Normally it is not necessary to destroy allocators, since the pool which created them will destroy them during pool cleanup.

eos_create

Create an EndOfStream bucket.

$b = APR::Bucket::eos_create($ba);
arg1: $ba ( APR::BucketAlloc object )

The freelist from which this bucket should be allocated

ret: $b ( APR::Bucket object )

The new bucket

since: 1.99_10

This bucket type indicates that there is no more data coming from down the filter stack. All filters should flush any buffered data at this point.

Example:

use APR::Bucket ();
use Apache::Connection ();
my $ba = $c->bucket_alloc;
my $eos_b = APR::Bucket::eos_create($ba);

flush_create

Create a flush bucket.

$b = APR::Bucket::flush_create($ba);
arg1: $ba ( APR::BucketAlloc object )

The freelist from which this bucket should be allocated

ret: $b ( APR::Bucket object )

The new bucket

since: 1.99_10

This bucket type indicates that filters should flush their data. There is no guarantee that they will flush it, but this is the best we can do.

insert_after

Insert a list of buckets after a specified bucket

$after_bucket->insert_after($add_bucket);
obj: $after_bucket ( APR::Bucket object )

The bucket to insert after

arg1: $add_bucket ( APR::Bucket object )

The buckets to insert. It says buckets, since $add_bucket may have more buckets attached after itself.

ret: no return value
since: 1.99_10

insert_before

Insert a list of buckets before a specified bucket

$before_bucket->insert_before($add_bucket);
obj: $before_bucket ( APR::Bucket object )

The bucket to insert before

arg1: $add_bucket ( APR::Bucket object )

The buckets to insert. It says buckets, since $add_bucket may have more buckets attached after itself.

ret: no return value
since: 1.99_10

is_eos

Determine if a bucket is an EOS bucket

$ret = $bucket->is_eos();
obj: $bucket ( APR::Bucket object )
ret: $ret ( boolean )
since: 1.99_10

is_flush

Determine if a bucket is a FLUSH bucket

$ret = $bucket->is_flush();
obj: $bucket ( APR::Bucket object )
ret: $ret ( boolean )
since: 1.99_10

length

Get the length of the data in the bucket.

$len = $b->length;
obj: $b ( APR::Bucket object )
ret: $len ( integer )

If the length is unknown, $len value will be -1.

since: 1.99_10

new

Create a new bucket and initialize it with data:

$nb = APR::Bucket->new($ba, $data);
$nb =          $b->new($ba, $data);
$nb = APR::Bucket->new($ba, $data, $offset);
$nb = APR::Bucket->new($ba, $data, $offset, $len);
obj: $b ( APR::Bucket object or class )
arg1: $ba ( APR::BucketAlloc object )
arg2: $data ( string )

The data to initialize with.

Important: in order to avoid unnecessary data copying the variable is stored in the bucket object. That means that if you modify $data after passing it to new() you will modify the data in the bucket as well. To avoid that pass to new() a copy which you won't modify.

opt arg3: $offset ( number )

Optional offset inside $data. Default: 0.

opt arg4: $len ( number )

Optional partial length to read.

If $offset is specified, then:

length $buffer - $offset;

will be used. Otherwise the default is to use:

length $buffer;
ret: $nb ( APR::Bucket object )

a newly created bucket object

since: 1.99_17

Examples:

  • Create a new bucket using a whole string:

    use APR::Bucket ();
    my $data = "my data";
    my $b = APR::Bucket->new($ba, $data);

    now the bucket contains the string 'my data'.

  • Create a new bucket using a sub-string:

    use APR::Bucket ();
    my $data   = "my data";
    my $offset = 3;
    my $b = APR::Bucket->new($ba, $data, $offset);

    now the bucket contains the string 'data'.

  • Create a new bucket not using the whole length and starting from an offset:

    use APR::Bucket ();
    my $data   = "my data";
    my $offset = 3;
    my $len    = 3;
    my $b = APR::Bucket->new($ba, $data, $offset, $len);

    now the bucket contains the string 'dat'.

read

Read the data from the bucket.

$len = $b->read($buffer);
$len = $b->read($buffer, $block);
obj: $b ( APR::Bucket object )

The bucket to read from

arg1: $buffer ( SCALAR )

The buffer to fill. All previous data will be lost.

opt arg2: $block ( APR::Const :read_type constant )

optional reading mode constant.

By default the read is blocking, via APR::BLOCK_READ constant.

ret: $len ( number )

How many bytes were actually read

$buffer gets populated with the string that is read. It will contain an empty string if there was nothing to read.

since: 1.99_15
excpt: APR::Error

It's important to know that certain bucket types (e.g. file bucket), may perform a split and insert extra buckets following the current one. Therefore never call $b->remove, before calling $b->read, or you may lose data.

Examples:

Blocking read:

my $len = $b->read(my $buffer);

Non-blocking read:

use APR::Const -compile 'NONBLOCK_READ';
my $len = $b->read(my $buffer, APR::NONBLOCK_READ);

remove

Tell the bucket to remove itself from the bucket brigade it belongs to.

$bucket->remove();
obj: $bucket ( APR::Bucket object )
ret: no return value
since: 1.99_10

If the bucket is not attached to any bucket brigade then this operation doesn't do anything.

When the bucket is removed, it's not not destroyed. Usually this is done in order to move the bucket to another bucket brigade. Or to copy the data way before destroying the bucket. If the bucket wasn't moved to another bucket brigade it must be destroyed.

Examples:

Assuming that $bb1 already exists and filled with buckets, move every odd bucket number to $bb2 and every even to $bb3:

my $bb2 = APR::Brigade->new($c->pool, $c->bucket_alloc);
my $bb3 = APR::Brigade->new($c->pool, $c->bucket_alloc);
my $count = 0;
while (my $bucket = $bb->first) {
    $count++;
    $bucket->remove;
    $count % 2
        ? $bb2->insert_tail($bucket)
        : $bb3->insert_tail($bucket);
}

setaside

Ensure the bucket's data lasts at least as long as the given pool:

my $status = $b->setaside($pool);
obj: $b ( APR::Bucket object )
arg1: $pool ( APR::Pool object )
ret: ( APR::Const status constant )

On success, APR::SUCCESS is returned. Otherwise a failure code is returned.

excpt: APR::Error

when your code deals only with mod_perl buckets, you don't have to ask for the return value. If this method is called in the VOID context, i.e.:

$b->setaside($pool);

mod_perl will do the error checking on your behalf, and if the return code is not APR::SUCCESS, an APR::Error exception will be thrown.

However if your code doesn't know which bucket types it may need to setaside, you may want to check the return code and deal with any errors. For example one of the possible error codes is APR::ENOTIMPL. As of this writing the pipe and socket buckets can't setaside(), in which case you may want to look at the ap_save_brigade() implementation.

since: 1.99_17

Usually setaside is called by certain output filters, in order to buffer socket writes of smaller buckets into a single write. This method works on all bucket types (not only the mod_perl bucket type), but as explained in the exceptions section, not all bucket types implement this method.

When a mod_perl bucket is setaside, its data is detached from the original perl scalar and copied into a pool bucket. That allows downstream filters to deal with the data originally owned by a Perl interpreter, making it possible for that interpreter to go away and do other things, or be destroyed.

type

Get the type of the data in the bucket.

$type = $b->type;
obj: $b ( APR::Bucket object )
ret: $type ( APR::BucketType object )
since: 1.99_10

You need to invoke APR::BucketType methods to access the data.

Example:

Create a flush bucket and read its type's name:

use APR::Bucket ();
use APR::BucketType ();
my $b = APR::Bucket::flush_create($ba);
my $type = $b->type;
my $type_name =  $type->name; # FLUSH

The type name will be 'FLUSH' in this example.

Unsupported API

APR::Socket also provides auto-generated Perl interface for a few other methods which aren't tested at the moment and therefore their API is a subject to change. These methods will be finalized later as a need arises. If you want to rely on any of the following methods please contact the the mod_perl development mailing list so we can help each other take the steps necessary to shift the method to an officially supported API.

data

$data = $b->data;

Gives a C pointer to the address of the data in the bucket. I can't see what use can be done of it in Perl.

obj: $b ( APR::Bucket object )
ret: $data ( C pointer )
since: subject to change

start

$start = $b->start;

It gives the offset to when a new bucket is created with a non-zero offset value:

my $b = APR::Bucket->new($ba, $data, $offset, $len);

So if the offset was 3. $start will be 3 too.

I fail to see what it can be useful for to the end user (it's mainly used internally).

obj: $b ( APR::Bucket object )
ret: $start ( offset number )
since: subject to change

See Also

mod_perl 2.0 documentation.

Copyright

mod_perl 2.0 and its core modules are copyrighted under The Apache Software License, Version 2.0.

Authors

The mod_perl development team and numerous contributors.