File Coverage

blib/lib/Mojo/Message.pm

Criterion	Covered	Total	%
statement	171	171	100.0
branch	81	88	92.0
condition	45	56	80.3
subroutine	44	44	100.0
pod	30	30	100.0
total	371	389	95.3

package Mojo::Message;
use Mojo::Base 'Mojo::EventEmitter';

use Carp qw(croak);
use Mojo::Asset::Memory;
use Mojo::Content::Single;
use Mojo::DOM;
use Mojo::JSON qw(j);
use Mojo::JSON::Pointer;
use Mojo::Parameters;
use Mojo::Upload;
use Mojo::Util qw(decode);

has content          => sub { Mojo::Content::Single->new };
has default_charset  => 'UTF-8';
has max_line_size    => sub { $ENV{MOJO_MAX_LINE_SIZE} || 8192 };
has max_message_size => sub { $ENV{MOJO_MAX_MESSAGE_SIZE} // 16777216 };
has version          => '1.1';

sub body {
  my $self = shift;

  # Get
  my $content = $self->content;
  return $content->is_multipart ? '' : $content->asset->slurp unless @_;

  # Set (multipart content needs to be downgraded)
  $content = $self->content(Mojo::Content::Single->new)->content if $content->is_multipart;
  $content->asset(Mojo::Asset::Memory->new->add_chunk(@_));

  return $self;
}

sub body_params {
  my $self = shift;

  return $self->{body_params} if $self->{body_params};
  my $params = $self->{body_params} = Mojo::Parameters->new;
  $params->charset($self->content->charset || $self->default_charset);

  # "application/x-www-form-urlencoded"
  my $type = $self->headers->content_type // '';
  if ($type =~ m!application/x-www-form-urlencoded!i) {
    $params->parse($self->content->asset->slurp);
  }

  # "multipart/form-data"
  elsif ($type =~ m!multipart/form-data!i) {
    $params->append(@$_[0, 1]) for @{$self->_parse_formdata};
  }

  return $params;
}

sub body_size { shift->content->body_size }

sub build_body       { shift->_build('get_body_chunk') }
sub build_headers    { shift->_build('get_header_chunk') }
sub build_start_line { shift->_build('get_start_line_chunk') }

sub cookie { shift->_cache('cookies', 0, @_) }

sub cookies { croak 'Method "cookies" not implemented by subclass' }

sub dom {
  my $self = shift;
  return undef if $self->content->is_multipart;
  my $dom = $self->{dom} ||= Mojo::DOM->new($self->text);
  return @_ ? $dom->find(@_) : $dom;
}

sub error {
  my $self = shift;
  return $self->{error} unless @_;
  $self->{error} = shift;
  return $self->finish;
}

sub every_cookie { shift->_cache('cookies', 1, @_) }
sub every_upload { shift->_cache('uploads', 1, @_) }

sub extract_start_line { croak 'Method "extract_start_line" not implemented by subclass' }

sub finish {
  my $self = shift;
  $self->{state} = 'finished';
  return $self->{finished}++ ? $self : $self->emit('finish');
}

sub fix_headers {
  my $self = shift;
  return $self if $self->{fix}++;

  # Content-Length or Connection (unless chunked transfer encoding is used)
  my $content = $self->content;
  my $headers = $content->headers;
  if    ($content->is_multipart)                           { $headers->remove('Content-Length') }
  elsif ($content->is_chunked || $headers->content_length) { return $self }
  if    ($content->is_dynamic)                             { $headers->connection('close') }
  else                                                     { $headers->content_length($self->body_size) }

  return $self;
}

sub get_body_chunk {
  my ($self, $offset) = @_;

  $self->emit('progress', 'body', $offset);
  my $chunk = $self->content->get_body_chunk($offset);
  return $chunk if !defined $chunk || length $chunk;
  $self->finish;

  return $chunk;
}

sub get_header_chunk {
  my ($self, $offset) = @_;
  $self->emit('progress', 'headers', $offset);
  return $self->fix_headers->content->get_header_chunk($offset);
}

sub get_start_line_chunk { croak 'Method "get_start_line_chunk" not implemented by subclass' }

sub header_size { shift->fix_headers->content->header_size }

sub headers { shift->content->headers }

sub is_finished { (shift->{state} // '') eq 'finished' }

sub is_limit_exceeded { !!shift->{limit} }

sub json {
  my ($self, $pointer) = @_;
  return undef if $self->content->is_multipart;
  my $data = $self->{json} //= j($self->body);
  return $pointer ? Mojo::JSON::Pointer->new($data)->get($pointer) : $data;
}

sub parse {
  my ($self, $chunk) = @_;

  return $self if $self->{error};
  $self->{raw_size} += length $chunk;
  $self->{buffer} .= $chunk;

  # Start-line
  unless ($self->{state}) {

    # Check start-line size
    my $len = index $self->{buffer}, "\x0a";
    $len = length $self->{buffer} if $len < 0;
    return $self->_limit('Maximum start-line size exceeded') if $len > $self->max_line_size;

    $self->{state} = 'content' if $self->extract_start_line(\$self->{buffer});
  }

  # Content
  my $state = $self->{state} // '';
  $self->content($self->content->parse(delete $self->{buffer})) if $state eq 'content' || $state eq 'finished';

  # Check message size
  my $max = $self->max_message_size;
  return $self->_limit('Maximum message size exceeded') if $max && $max < $self->{raw_size};

  # Check header size
  return $self->_limit('Maximum header size exceeded') if $self->headers->is_limit_exceeded;

  # Check buffer size
  return $self->_limit('Maximum buffer size exceeded') if $self->content->is_limit_exceeded;

  return $self->emit('progress')->content->is_finished ? $self->finish : $self;
}

sub save_to {
  my ($self, $path) = @_;
  my $content = $self->content;
  croak 'Multipart content cannot be saved to files' if $content->is_multipart;
  $content->asset->move_to($path);
  return $self;
}

sub start_line_size { croak 'Method "start_line_size" not implemented by subclass' }

sub text {
  my $self    = shift;
  my $body    = $self->body;
  my $charset = $self->content->charset || $self->default_charset;
  return $charset ? decode($charset, $body) // $body : $body;
}

sub to_string {
  my $self = shift;
  return $self->build_start_line . $self->build_headers . $self->build_body;
}

sub upload { shift->_cache('uploads', 0, @_) }

sub uploads {
  my $self = shift;

  my @uploads;
  for my $data (@{$self->_parse_formdata(1)}) {
    my $upload = Mojo::Upload->new(
      name     => $data->[0],
      filename => $data->[2],
      asset    => $data->[1]->asset,
      headers  => $data->[1]->headers
    );
    push @uploads, $upload;
  }

  return \@uploads;
}

sub _build {
  my ($self, $method) = @_;

  my ($buffer, $offset) = ('', 0);
  while (1) {

    # No chunk yet, try again
    next unless defined(my $chunk = $self->$method($offset));

    # End of part
    last unless my $len = length $chunk;

    $offset += $len;
    $buffer .= $chunk;
  }

  return $buffer;
}

sub _cache {
  my ($self, $method, $all, $name) = @_;

  # Cache objects by name
  unless ($self->{$method}) {
    $self->{$method} = {};
    push @{$self->{$method}{$_->name}}, $_ for @{$self->$method};
  }

  my $objects = $self->{$method}{$name} // [];
  return $all ? $objects : $objects->[-1];
}

sub _limit { ++$_[0]{limit} and return $_[0]->error({message => $_[1]}) }

sub _parse_formdata {
  my ($self, $upload) = @_;

  my @formdata;
  my $content = $self->content;
  return \@formdata unless $content->is_multipart;
  my $charset = $content->charset || $self->default_charset;

  # Check all parts recursively
  my @parts = ($content);
  while (my $part = shift @parts) {

    if ($part->is_multipart) {
      unshift @parts, @{$part->parts};
      next;
    }

    next unless my $disposition = $part->headers->content_disposition;
    my ($filename) = $disposition =~ /[; ]filename="((?:\\"|[^"])*)"/;
    next if $upload && !defined $filename || !$upload && defined $filename;
    my ($name) = $disposition =~ /[; ]name="((?:\\"|[^;"])*)"/;
    next if !defined $name;
    $part = $part->asset->slurp unless $upload;

    if ($charset) {
      $name     = decode($charset, $name)     // $name     if $name;
      $filename = decode($charset, $filename) // $filename if $filename;
      $part     = decode($charset, $part)     // $part unless $upload;
    }

    push @formdata, [$name, $part, $filename];
  }

  return \@formdata;
}

1;

=encoding utf8

=head1 NAME

Mojo::Message - HTTP message base class

=head1 SYNOPSIS

  package Mojo::Message::MyMessage;
  use Mojo::Base 'Mojo::Message';

  sub cookies              {...}
  sub extract_start_line   {...}
  sub get_start_line_chunk {...}
  sub start_line_size      {...}

=head1 DESCRIPTION

L is an abstract base class for HTTP message containers, based on L
7230|https://tools.ietf.org/html/rfc7230>, L and L
2388|https://tools.ietf.org/html/rfc2388>, like L and L.

=head1 EVENTS

L inherits all events from L and can emit the following new ones.

=head2 finish

  $msg->on(finish => sub ($msg) {...});

Emitted after message building or parsing is finished.

  my $before = time;
  $msg->on(finish => sub ($msg) { $msg->headers->header('X-Parser-Time' => time - $before) });

=head2 progress

  $msg->on(progress => sub ($msg) {...});

Emitted when message building or parsing makes progress.

  # Building
  $msg->on(progress => sub ($msg, $state, $offset) { say qq{Building "$state" at offset $offset} });

  # Parsing
  $msg->on(progress => sub ($msg) {
    return unless my $len = $msg->headers->content_length;
    my $size = $msg->content->progress;
    say 'Progress: ', $size == $len ? 100 : int($size / ($len / 100)), '%';
  });

=head1 ATTRIBUTES

L implements the following attributes.

=head2 content

  my $msg = $msg->content;
  $msg    = $msg->content(Mojo::Content::Single->new);

Message content, defaults to a L object.

=head2 default_charset

  my $charset = $msg->default_charset;
  $msg        = $msg->default_charset('UTF-8');

Default charset used by L and to extract data from C or
C message body, defaults to C.

=head2 max_line_size

  my $size = $msg->max_line_size;
  $msg     = $msg->max_line_size(1024);

Maximum start-line size in bytes, defaults to the value of the C environment variable or C<8192>
(8KiB).

=head2 max_message_size

  my $size = $msg->max_message_size;
  $msg     = $msg->max_message_size(1024);

Maximum message size in bytes, defaults to the value of the C environment variable or
C<16777216> (16MiB). Setting the value to C<0> will allow messages of indefinite size.

=head2 version

  my $version = $msg->version;
  $msg        = $msg->version('1.1');

HTTP version of message, defaults to C<1.1>.

=head1 METHODS

L inherits all methods from L and implements the following new ones.

=head2 body

  my $bytes = $msg->body;
  $msg      = $msg->body('Hello!');

Slurp or replace L.

=head2 body_params

  my $params = $msg->body_params;

C parameters extracted from C or C message body, usually
a L object. Note that this method caches all data, so it should not be called before the entire
message body has been received. Parts of the message body need to be loaded into memory to parse C parameters, so
you have to make sure it is not excessively large. There's a 16MiB limit for requests and a 2GiB limit for responses by
default.

  # Get POST parameter names and values
  my $hash = $msg->body_params->to_hash;

=head2 body_size

  my $size = $msg->body_size;

Content size in bytes.

=head2 build_body

  my $bytes = $msg->build_body;

Render whole body with L.

=head2 build_headers

  my $bytes = $msg->build_headers;

Render all headers with L.

=head2 build_start_line

  my $bytes = $msg->build_start_line;

Render start-line with L.

=head2 cookie

  my $cookie = $msg->cookie('foo');

Access message cookies, usually L or L objects. If there are multiple
cookies sharing the same name, and you want to access more than just the last one, you can use L. Note
that this method caches all data, so it should not be called before all headers have been received.

  # Get cookie value
  say $msg->cookie('foo')->value;

=head2 cookies

  my $cookies = $msg->cookies;

Access message cookies. Meant to be overloaded in a subclass.

=head2 dom

  my $dom        = $msg->dom;
  my $collection = $msg->dom('a[href]');

Retrieve message body from L and turn it into a L object, an optional selector can be used to call
the method L on it right away, which then returns a L object. Note that this method
caches all data, so it should not be called before the entire message body has been received. The whole message body
needs to be loaded into memory to parse it, so you have to make sure it is not excessively large. There's a 16MiB limit
for requests and a 2GiB limit for responses by default.

  # Perform "find" right away
  say $msg->dom('h1, h2, h3')->map('text')->join("\n");

  # Use everything else Mojo::DOM has to offer
  say $msg->dom->at('title')->text;
  say $msg->dom->at('body')->children->map('tag')->uniq->join("\n");

=head2 error

  my $err = $msg->error;
  $msg    = $msg->error({message => 'Parser error'});

Get or set message error, an C return value indicates that there is no error.

  # Connection or parser error
  $msg->error({message => 'Connection refused'});

  # 4xx/5xx response
  $msg->error({message => 'Internal Server Error', code => 500});

=head2 every_cookie

  my $cookies = $msg->every_cookie('foo');

Similar to L, but returns all message cookies sharing the same name as an array reference.

  # Get first cookie value
  say $msg->every_cookie('foo')->[0]->value;

=head2 every_upload

  my $uploads = $msg->every_upload('foo');

Similar to L, but returns all file uploads sharing the same name as an array reference.

  # Get content of first uploaded file
  say $msg->every_upload('foo')->[0]->asset->slurp;

=head2 extract_start_line

  my $bool = $msg->extract_start_line(\$str);

Extract start-line from string. Meant to be overloaded in a subclass.

=head2 finish

  $msg = $msg->finish;

Finish message parser/generator.

=head2 fix_headers

  $msg = $msg->fix_headers;

Make sure message has all required headers.

=head2 get_body_chunk

  my $bytes = $msg->get_body_chunk($offset);

Get a chunk of body data starting from a specific position. Note that it might not be possible to get the same chunk
twice if content was generated dynamically.

=head2 get_header_chunk

  my $bytes = $msg->get_header_chunk($offset);

Get a chunk of header data, starting from a specific position. Note that this method finalizes the message.

=head2 get_start_line_chunk

  my $bytes = $msg->get_start_line_chunk($offset);

Get a chunk of start-line data starting from a specific position. Meant to be overloaded in a subclass.

=head2 header_size

  my $size = $msg->header_size;

Size of headers in bytes. Note that this method finalizes the message.

=head2 headers

  my $headers = $msg->headers;

Message headers, usually a L object.

  # Longer version
  my $headers = $msg->content->headers;

=head2 is_finished

  my $bool = $msg->is_finished;

Check if message parser/generator is finished.

=head2 is_limit_exceeded

  my $bool = $msg->is_limit_exceeded;

Check if message has exceeded L, L, L or
L.

=head2 json

  my $value = $msg->json;
  my $value = $msg->json('/foo/bar');

Decode JSON message body directly using L if possible, an C return value indicates a bare C or
that decoding failed. An optional JSON Pointer can be used to extract a specific value with L.
Note that this method caches all data, so it should not be called before the entire message body has been received. The
whole message body needs to be loaded into memory to parse it, so you have to make sure it is not excessively large.
There's a 16MiB limit for requests and a 2GiB limit for responses by default.

  # Extract JSON values
  say $msg->json->{foo}{bar}[23];
  say $msg->json('/foo/bar/23');

=head2 parse

  $msg = $msg->parse('HTTP/1.1 200 OK...');

Parse message chunk.

=head2 save_to

  $msg = $msg->save_to('/some/path/index.html');

Save message body to a file.

=head2 start_line_size

  my $size = $msg->start_line_size;

Size of the start-line in bytes. Meant to be overloaded in a subclass.

=head2 text

  my $str = $msg->text;

Retrieve L and try to decode it with L or L.

=head2 to_string

  my $str = $msg->to_string;

Render whole message. Note that this method finalizes the message, and that it might not be possible to render the same
message twice if content was generated dynamically.

=head2 upload

  my $upload = $msg->upload('foo');

Access C file uploads, usually L objects. If there are multiple uploads sharing the
same name, and you want to access more than just the last one, you can use L. Note that this method
caches all data, so it should not be called before the entire message body has been received.

  # Get content of uploaded file
  say $msg->upload('foo')->asset->slurp;

=head2 uploads

  my $uploads = $msg->uploads;

All C file uploads, usually L objects.

  # Names of all uploads
  say $_->name for @{$msg->uploads};

=head1 SEE ALSO

L, L, L.

=cut