Amazon-S3-Lite

 view release on metacpan or  search on metacpan

lib/Amazon/S3/Lite.pm  view on Meta::CPAN

package Amazon::S3::Lite;

use strict;
use warnings;

use Amazon::Signature4::Lite;
use Amazon::S3::Lite::Credentials;
use Amazon::S3::Lite::Logger;
use Carp qw(croak);
use Data::Dumper;
use Digest::MD5 qw(md5_base64 md5);
use English qw(-no_match_vars);
use HTTP::Tiny;
use List::Util qw(pairs);
use MIME::Base64 qw(encode_base64);
use Scalar::Util qw(blessed openhandle);
use URI::Escape qw(uri_escape_utf8);
use XML::Twig;

use Readonly;
Readonly our $TRUE  => 1;
Readonly our $FALSE => 0;

our $VERSION = '1.2.2';

########################################################################
sub new {
########################################################################
  my ( $class, @args ) = @_;

  my $options = ref $args[0] ? $args[0] : {@args};

  my $self = bless $options, $class;

  $self->{host}    //= 's3.amazonaws.com';
  $self->{secure}  //= $TRUE;
  $self->{timeout} //= 30;
  $self->{region}  //= 'us-east-1';

  $self->_init_logger;
  $self->_init_credentials;
  $self->_init_ua;

  return $self;
}

########################################################################
# Logger setup
# Priority: caller-supplied object -> Log::Log4perl (if available) ->
#           minimal STDERR logger
########################################################################
sub _init_logger {
########################################################################
  my ($self) = @_;

  my $logger = $self->{logger};

  if ( $logger && blessed $logger ) {
    # Validate it quacks like a logger
    for my $method (qw(trace debug info warn error)) {
      croak "logger object must implement '$method'"
        if !$logger->can($method);
    }

    return;
  }

  my $log4perl = eval {
    require Log::Log4perl;
    1;
  };

  my $log_level = $self->{log_level} // 'warn';
  $self->{log_level} = $log_level;

  if ($log4perl) {
    if ( !Log::Log4perl->initialized ) {

lib/Amazon/S3/Lite.pm  view on Meta::CPAN

# Build a fresh signer from current credentials.
# Called per-request so that rotating credentials (Lambda IAM roles)
# are always current.
########################################################################
sub _signer {
########################################################################
  my ( $self, $region ) = @_;

  my $creds = $self->credentials;

  my $access_key
    = $creds->can('get_aws_access_key_id')
    ? $creds->get_aws_access_key_id
    : $creds->aws_access_key_id;

  my $secret_key
    = $creds->can('get_aws_secret_access_key')
    ? $creds->get_aws_secret_access_key
    : $creds->aws_secret_access_key;

  my $token_sub = $creds->can('get_token') // $creds->can('token');
  my $token     = $token_sub ? $token_sub->($creds) : undef;

  return Amazon::Signature4::Lite->new(
    access_key    => $access_key,
    secret_key    => $secret_key,
    session_token => $token,
    region        => $region // $self->region,
    service       => 's3',
  );
}

########################################################################
# Build the endpoint URL for a bucket/key
########################################################################
sub _endpoint {
########################################################################
  my ( $self, $bucket, $key ) = @_;

  my $scheme = $self->{secure} ? 'https' : 'http';
  my $host   = $self->host;

  # Path-style URL: https://s3.amazonaws.com/bucket/key
  # (virtual-hosted style omitted for simplicity; path-style works
  # everywhere and avoids SSL cert issues with dotted bucket names)
  my $url = "$scheme://$host";

  $url .= "/$bucket"              if defined $bucket && length $bucket;
  $url .= '/' . _encode_key($key) if defined $key    && length $key;

  return $url;
}

########################################################################
# URI-encode an S3 key, preserving '/' separators
########################################################################
sub _encode_key {
########################################################################
  my ($key) = @_;

  return join '/', map { uri_escape_utf8( $_, '^A-Za-z0-9\-._~' ) }
    split m{/}, $key, -1;
}

########################################################################
sub _request {
########################################################################
  my ( $self, $method, $url, $headers, $content, $extra, $region ) = @_;

  $region  //= $self->region;
  $headers //= {};
  $content //= q{};
  $extra   //= {};

  my $content_is_coderef = ref $content eq 'CODE';

  # sign — returns merged headers ready for HTTP::Tiny
  my $signed = $self->_signer($region)->sign(
    method  => $method,
    url     => $url,
    headers => $headers,
    payload => $content_is_coderef ? q{} : $content,
  );

  # HTTP::Tiny sets Host itself — remove to avoid duplicate header error
  delete $signed->{host};

  $self->logger->debug("$method $url");

  my $options = { headers => $signed };

  if ( length $content || $content_is_coderef ) {
    $options->{content} = $content;
  }

  if ( $extra->{data_callback} ) {
    $options->{data_callback} = $extra->{data_callback};
  }

  my $response = $self->ua->request( $method, $url, $options );

  $self->logger->debug( sprintf 'Response: %s %s', $response->{status}, $response->{reason} );

  return $response;
}

########################################################################
# head_object( $bucket, $key )
#
# Fetches metadata for an object without retrieving the body.
# Returns undef if the key does not exist (404).
# Returns a hashref with content_type, content_length, etag,
# last_modified, and metadata (x-amz-meta-* headers).
########################################################################
sub head_object {
########################################################################
  my ( $self, $bucket, $key ) = @_;

  croak 'bucket is required' if !defined $bucket || !length $bucket;
  croak 'key is required'    if !defined $key    || !length $key;

lib/Amazon/S3/Lite.pm  view on Meta::CPAN

  if ( defined $filename ) {
    # Open the destination file before making the request so we catch
    # permission errors early, before network round-trip
    open my $fh, '>', $filename
      or croak "cannot open '$filename' for writing: $!";

    $extra->{data_callback} = sub {
      my ($data) = @_;
      print {$fh} $data
        or croak "write to '$filename' failed: $!";
    };

    my $response = $self->_request( 'GET', $url, \%headers, q{}, $extra );

    close $fh
      or croak "close of '$filename' failed: $!";

    return undef ## no critic (Subroutines::ProhibitExplicitReturnUndef)
      if _is_not_found($response);

    $self->_croak_on_error( $response, 'get_object' );

    # Return metadata only — content is on disk
    return $self->_extract_object_metadata( $response->{headers} );
  }

  # In-memory path
  my $response = $self->_request( 'GET', $url, \%headers );

  return undef ## no critic (Subroutines::ProhibitExplicitReturnUndef)
    if _is_not_found($response);

  $self->_croak_on_error( $response, 'get_object' );

  my $result = $self->_extract_object_metadata( $response->{headers} );
  $result->{content} = $response->{content};

  return $result;
}

########################################################################
# delete_object( $bucket, $key, %options )
#
# Deletes an object from S3. Options:
#   version_id => $vid    delete a specific version
#
# Returns true on success. Note S3 returns 204 for both successful
# deletes and deletes of non-existent keys — no distinction is made.
# Croaks on network or server errors.
########################################################################
sub delete_object {
########################################################################
  my ( $self, $bucket, $key, %options ) = @_;

  croak 'bucket is required' if !defined $bucket || !length $bucket;
  croak 'key is required'    if !defined $key    || !length $key;

  my $url = $self->_endpoint( $bucket, $key );

  if ( defined $options{version_id} ) {
    $url .= '?versionId=' . uri_escape_utf8( $options{version_id} );
  }

  my $response = $self->_request( 'DELETE', $url );

  $self->_croak_on_error( $response, 'delete_object' );

  return 1;
}

########################################################################
# create_bucket( $bucket, %options )
#
# Creates a new S3 bucket.
#
# us-east-1 is the S3 default region — the CreateBucketConfiguration
# body must NOT be sent for us-east-1 (S3 will error). All other regions
# require it with LocationConstraint set to the target region.
#
# Options: acl, region
#
# Returns true on success. Croaks on failure.
########################################################################
sub create_bucket {
########################################################################
  my ( $self, $bucket, %options ) = @_;

  croak 'bucket is required'
    if !defined $bucket || !length $bucket;

  my $region = $options{region} // $self->region;
  my $url    = $self->_endpoint($bucket);
  my %headers;

  $headers{'x-amz-acl'} = $options{acl} if $options{acl};

  my $content = q{};

  # us-east-1 is the implicit default — sending LocationConstraint for it
  # causes an error. All other regions require it.
  if ( $region ne 'us-east-1' ) {
    $content
      = sprintf '<CreateBucketConfiguration '
      . 'xmlns="http://s3.amazonaws.com/doc/2006-03-01/">'
      . '<LocationConstraint>%s</LocationConstraint>'
      . '</CreateBucketConfiguration>',
      $region;
    $headers{'Content-Type'}   = 'application/xml';
    $headers{'Content-Length'} = length $content;
  }

  my $response = $self->_request( 'PUT', $url, \%headers, $content, {}, $region );

  $self->_croak_on_error( $response, 'create_bucket' );

  return 1;
}

########################################################################
# list_buckets()
#

lib/Amazon/S3/Lite.pm  view on Meta::CPAN

      my $n = read( $data, $buf, $chunk_size );
      return $buf if $n;
      return q{};
    };
  }
  elsif ( ref $data eq 'SCALAR' ) {
    # --- Scalar ref path ---
    $body                      = ${$data};
    $headers{'Content-Length'} = length $body;
    $headers{'Content-MD5'}    = encode_base64( md5($body), q{} );
  }
  else {
    # --- Plain scalar path ---
    $body                      = $data;
    $headers{'Content-Length'} = length $body;
    $headers{'Content-MD5'}    = encode_base64( md5($body), q{} );
  }

  my $response = $self->_request( 'PUT', $url, \%headers, $body );

  $self->_croak_on_error( $response, 'put_object' );

  my $etag = $response->{headers}{etag};
  $etag =~ s/\A"|"\z//gxsm if defined $etag;

  return $etag;
}

########################################################################
# list_objects_v2( $bucket, %options )
#
# Lists objects in a bucket using the S3 ListObjectsV2 API.
# Returns a hashref with keys: bucket, prefix, key_count, max_keys,
# is_truncated, next_continuation_token, objects, common_prefixes.
########################################################################
sub list_objects_v2 {
########################################################################
  my ( $self, $bucket, %options ) = @_;

  croak 'bucket is required'
    if !defined $bucket || !length $bucket;

  # Map our option names to S3 query parameter names
  my %param_map = (
    prefix             => 'prefix',
    delimiter          => 'delimiter',
    max_keys           => 'max-keys',
    continuation_token => 'continuation-token',
    start_after        => 'start-after',
  );

  my %params = ( 'list-type' => '2' );

  for my $opt ( keys %param_map ) {
    if ( defined $options{$opt} ) {
      $params{ $param_map{$opt} } = $options{$opt};
    }
  }

  # Build query string
  my $query = join q{&}, map { uri_escape_utf8($_) . q{=} . uri_escape_utf8( $params{$_} ) }
    sort keys %params;

  my $url = $self->_endpoint($bucket) . q{?} . $query;

  my $response = $self->_request( 'GET', $url );

  return undef ## no critic (Subroutines::ProhibitExplicitReturnUndef)
    if _is_not_found($response);

  $self->_croak_on_error( $response, 'list_objects_v2' );

  return $self->_parse_list_objects_v2( $response->{content} );
}

########################################################################
# Parse the XML body of a ListObjectsV2 response
########################################################################
########################################################################
sub _parse_list_objects_v2 {
########################################################################
  my ( $self, $xml ) = @_;

  my ( @objects, @common_prefixes );
  my ( $bucket, $prefix, $key_count, $max_keys, $is_truncated, $next_token );

  XML::Twig->new(
    twig_handlers => {
      'Name'                    => sub { $bucket       = $_[1]->text },
      'ListBucketResult/Prefix' => sub { $prefix       = $_[1]->text },
      'KeyCount'                => sub { $key_count    = $_[1]->text + 0 },
      'MaxKeys'                 => sub { $max_keys     = $_[1]->text + 0 },
      'IsTruncated'             => sub { $is_truncated = $_[1]->text eq 'true' ? 1 : 0 },
      'NextContinuationToken'   => sub { $next_token   = $_[1]->text },
      'Contents'                => sub {
        my ( $t, $node ) = @_;
        my $etag = $node->first_child_text('ETag') // q{};
        $etag =~ s/\A"|"\z//gxsm;
        push @objects,
          {
          key           => $node->first_child_text('Key'),
          size          => $node->first_child_text('Size') + 0,
          last_modified => $node->first_child_text('LastModified'),
          etag          => $etag,
          storage_class => $node->first_child_text('StorageClass'),
          };
        $t->purge;  # free memory as we go - important for large listings
      },
      'CommonPrefixes' => sub {
        my ( $t, $node ) = @_;
        push @common_prefixes, $node->first_child_text('Prefix');
      },
    }
  )->parse($xml);

  return {
    bucket                  => $bucket,
    prefix                  => $prefix,
    key_count               => $key_count,
    max_keys                => $max_keys,
    is_truncated            => $is_truncated,