server/ext/tcp_http.ext

#! perl # optional depends=tcp

# http server - this tried to speak enough of http 1.1 (and 1.0)
# to work with browsers. it does not even attempt to be a complete
# implementation, although it should be mostly correct for that it does.

sub send {
   my $self = $_[0];

   if (length $self->{wbuf}) {
      $self->{wbuf} .= $_[1];
   } else {
      $self->{wbuf} .= $_[1];

      my $len = syswrite $self->{fh}, $self->{wbuf};
      substr $self->{wbuf}, 0, $len, "";

      $self->{ww} = AE::io $self->{fh}, 1, sub {
         my $len = syswrite $self->{fh}, $self->{wbuf};
         substr $self->{wbuf}, 0, $len, "";

         $self->{ww}->stop unless length $self->{wbuf};
      } if length $self->{wbuf};
   }
}

sub fatal {
   my ($self) = @_;

   $self->send ("HTTP/1.1 500 internal error\015\012");
}

sub respond {
   $_[0]->send ("HTTP/1.1 $_[1]\015\012"
                . "content-length: " . (0 + length $_[2]) . "\015\012"
                . "access-control-allow-origin: *\015\012"
                . $_[0]{ohdr}
                . "$_[3]\015\012" . ($_[0]{give_head} ? "" : $_[2]));
}

my $cache_headers = "cache-control: max-age=8640000\015\012"
                    . "etag: \"0\"\015\012";

sub content_type {
   return "content-type: image/png\015\012"  if $_[0] =~ /^\x89PNG/;
   return "content-type: image/jpeg\015\012" if $_[0] =~ /^......JFIF/s;
   return "content-type: audio/wav\015\012"  if $_[0] =~ /^RIFF/;
   return "content-type: audio/ogg\015\012"  if $_[0] =~ /^OggS/;
   return "content-type: text/html\015\012"  if $_[0] =~ /^</;

   "content-type: text/plain\015\012"
}

sub handle_con {
   my ($self) = @_;

   my ($method, $uri, $http, $req, $close, $len);

   while () {
      while ($self->{rbuf} =~ s/^( (?: [^\015]+ | . )+? \015\012)\015\012//xs) {
         $req = $1;

         # we ignore headers atm.

         $req =~ m%^(\S+) (\S+) HTTP/([0-9.]+)\015\012%ig
            or return $self->respond ("400 bad request");

         $method = uc $1;
         $uri    = $2;
         $http   = $3;

         $uri =~ s%^http://[^/]*%%i; # just in case
         $uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge; # %-decode

#         my %hdr;
#
#         $hdr{lc $1} .= ",$2"
#            while $req =~ /\G
#                  ([^:\000-\037]*):
#                  [\011\040]*
#                  ((?: [^\012]+ | \012[\011\040] )*)
#                  \012
#               /gxc;
#
#         $req =~ /\G$/
#           or return $self->respond ("400 bad request");
#
#         # remove the "," prefix we added to all headers above
#         substr $_, 0, 1, ""
#            for values %hdr;

         if ($http == 1.0) {
            if ($req =~ /^connection\s*:\s*keep-alive/mi) {
               $self->{ohdr} = "connection: keep-alive\015\012";
            } else {
               $self->{ohdr} = "connection: close\015\012";
               $close = 1;
            }
         }

         $self->{give_head} = $method eq "HEAD"
            and $method = "GET";

         cf::debug "HTTP $method: $self->{id} $uri";

         if ($method ne "GET") {
            $self->respond ("405 no $method");
            next;
         }

         # not needed for GET, but who knows
         $self->send ("HTTP/1.1 100 go on\015\012")
            if $req =~ /^expect:.*\b100-continue\b/i;

         if ($uri =~ m%^/([0-9a-f]+)(M?)$%) { # faces
            my $want_meta = $2;
            my $idx = $cf::FACEHASH{pack "H*", $1};

            $idx
               or do { $self->respond ("404 illegal face name"), next };

            if ($req =~ /if-none-match/i) { # dirtiest hack evar
               $self->respond ("304 not modified", "", $cache_headers);
               next;
            }

            my $type = cf::face::get_type $idx, 1;
            my $data = cf::face::get_data $idx, 1;

            (my $meta, $data) = unpack "(w/a*)*", $data
               if $type & 1;

            if ($want_meta) {
               if ($type & 1) {
                  $self->respond ("200 OK", $meta, "content-type: text/plain\015\012" . $cache_headers);
               } else {
                  $self->respond ("404 type $type has no metadata");
               }
            } else {
               $self->respond ("200 OK", $data, (content_type $data) . $cache_headers);
            }

         } elsif (my $idx = (cf::face::find "res/http$uri") || (cf::face::find "res/http${uri}index.html")) {
            # TODO: use etag (shudder)
            my $data = cf::face::get_data $idx, 1;
            $self->respond ("200 OK", $data, (content_type $data) . $cache_headers);

         } elsif (cf::face::find "res/http$uri/index.html") {
            $self->respond ("302 dirslash", "", "location: $uri/\015\012");

         } elsif ($uri eq "/debug") { # for debugging
            my @body = "<html><body>";

            for my $type (6, 5, 4, 3, 2, 1, 0) {
               push @body, "<h1>$type</h1>";

               for (1 .. cf::face::faces_size - 1) {
                  cf::cede_to_tick;

                  next if $type != cf::face::get_type $_;
                  my $name = cf::face::get_name $_;
                  my $id   = unpack "H*", cf::face::get_chksum $_, 1;
                  push @body, "$_ <a href='$id'>$name ($id)</a>";
                  push @body, " <a href='${id}M'>(meta)</a>" if $type & 1;
                  push @body, "<br>";
               }
            }

            push @body, "</body></html>";

            my $body = join "", @body;
            utf8::encode $body;
            $self->respond ("200 OK", $body, "content-type: text/html\015\012");

         } else {
            $self->respond ("404 not found");
         }

         cf::cede_to_tick;
      }

      return if $close; # http 1.0 only currently
      return if length $self->{rbuf} > 8192; # headers too long

      Coro::AnyEvent::readable $self->{fh}, 6;

      $len = sysread $self->{fh}, $self->{rbuf}, 4096, length $self->{rbuf};

      return if $len <= 0;
   };
}

our $DETECTOR = ext::tcp::register http => 64, sub {
   # regex avoids conflict with websockets, which use /ws
   m{^(?i:GET|HEAD|OPTIONS) \  (?! (?i:http://[^/]+)? /ws \  ) }x
}, sub {
   my $self = bless {
      id   => $_[0],
      fh   => $_[1],
      rbuf => $_[2],
      wbuf => "",
   };

   $self->{async} = Coro::async_pool {
      $Coro::current->nice (4);
      $Coro::current->{desc} = "http $self->{id}";

      $self->handle_con;

      # this is ultra-ugly, but overall, fast
#      Coro::AnyEvent::writable $self->{fh}
#         while length $self->{wbuf};

#      %$self = ();
   };
};

cf::register_exticmd http_faceurl => sub {
   my ($ns) = @_;

   "/"
};

Revision:	1.9
Committed:	Sat Nov 10 02:32:13 2012 UTC (11 years, 6 months ago) by root
Branch:	MAIN
Changes since 1.8:	+138 -85 lines
Log Message:	* empty log message *
#	User	Rev	Content
1	root	1.1	#! perl # optional depends=tcp
2
3	root	1.9	# http server - this tried to speak enough of http 1.1 (and 1.0)
4			# to work with browsers. it does not even attempt to be a complete
5			# implementation, although it should be mostly correct for that it does.
6	root	1.1
7			sub send {
8			my $self = $_[0];
9
10	root	1.9	if (length $self->{wbuf}) {
11			$self->{wbuf} .= $_[1];
12			} else {
13			$self->{wbuf} .= $_[1];
14	root	1.1
15			my $len = syswrite $self->{fh}, $self->{wbuf};
16			substr $self->{wbuf}, 0, $len, "";
17
18	root	1.9	$self->{ww} = AE::io $self->{fh}, 1, sub {
19			my $len = syswrite $self->{fh}, $self->{wbuf};
20			substr $self->{wbuf}, 0, $len, "";
21
22			$self->{ww}->stop unless length $self->{wbuf};
23			} if length $self->{wbuf};
24			}
25	root	1.1	}
26
27			sub fatal {
28			my ($self) = @_;
29
30			$self->send ("HTTP/1.1 500 internal error\015\012");
31			}
32
33			sub respond {
34			$_[0]->send ("HTTP/1.1 $_[1]\015\012"
35			. "content-length: " . (0 + length $_[2]) . "\015\012"
36			. "access-control-allow-origin: *\015\012"
37	root	1.6	. $_[0]{ohdr}
38	root	1.9	. "$_[3]\015\012" . ($_[0]{give_head} ? "" : $_[2]));
39	root	1.1	}
40
41			my $cache_headers = "cache-control: max-age=8640000\015\012"
42			. "etag: \"0\"\015\012";
43
44			sub content_type {
45			return "content-type: image/png\015\012" if $_[0] =~ /^\x89PNG/;
46			return "content-type: image/jpeg\015\012" if $_[0] =~ /^......JFIF/s;
47			return "content-type: audio/wav\015\012" if $_[0] =~ /^RIFF/;
48			return "content-type: audio/ogg\015\012" if $_[0] =~ /^OggS/;
49	root	1.7	return "content-type: text/html\015\012" if $_[0] =~ /^</;
50	root	1.1
51			"content-type: text/plain\015\012"
52			}
53
54	root	1.9	sub handle_con {
55	root	1.1	my ($self) = @_;
56
57	root	1.9	my ($method, $uri, $http, $req, $close, $len);
58	root	1.1
59	root	1.9	while () {
60			while ($self->{rbuf} =~ s/^( (?: [^\015]+ \| . )+? \015\012)\015\012//xs) {
61			$req = $1;
62
63			# we ignore headers atm.
64
65			$req =~ m%^(\S+) (\S+) HTTP/([0-9.]+)\015\012%ig
66			or return $self->respond ("400 bad request");
67
68			$method = uc $1;
69			$uri = $2;
70			$http = $3;
71
72			$uri =~ s%^http://[^/]*%%i; # just in case
73			$uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge; # %-decode
74
75			# my %hdr;
76			#
77			# $hdr{lc $1} .= ",$2"
78			# while $req =~ /\G
79			# ([^:\000-\037]*):
80			# [\011\040]*
81			# ((?: [^\012]+ \| \012[\011\040] )*)
82			# \012
83			# /gxc;
84			#
85			# $req =~ /\G$/
86			# or return $self->respond ("400 bad request");
87			#
88			# # remove the "," prefix we added to all headers above
89			# substr $_, 0, 1, ""
90			# for values %hdr;
91
92			if ($http == 1.0) {
93			if ($req =~ /^connection\s:\skeep-alive/mi) {
94			$self->{ohdr} = "connection: keep-alive\015\012";
95			} else {
96			$self->{ohdr} = "connection: close\015\012";
97			$close = 1;
98			}
99			}
100	root	1.1
101	root	1.9	$self->{give_head} = $method eq "HEAD"
102			and $method = "GET";
103	root	1.1
104	root	1.9	cf::debug "HTTP $method: $self->{id} $uri";
105	root	1.6
106	root	1.9	if ($method ne "GET") {
107			$self->respond ("405 no $method");
108			next;
109	root	1.6	}
110	root	1.1
111	root	1.9	# not needed for GET, but who knows
112			$self->send ("HTTP/1.1 100 go on\015\012")
113			if $req =~ /^expect:.*\b100-continue\b/i;
114
115			if ($uri =~ m%^/([0-9a-f]+)(M?)$%) { # faces
116			my $want_meta = $2;
117			my $idx = $cf::FACEHASH{pack "H*", $1};
118
119			$idx
120			or do { $self->respond ("404 illegal face name"), next };
121
122			if ($req =~ /if-none-match/i) { # dirtiest hack evar
123			$self->respond ("304 not modified", "", $cache_headers);
124			next;
125			}
126	root	1.1
127	root	1.9	my $type = cf::face::get_type $idx, 1;
128			my $data = cf::face::get_data $idx, 1;
129	root	1.1
130	root	1.9	(my $meta, $data) = unpack "(w/a)", $data
131			if $type & 1;
132	root	1.1
133	root	1.9	if ($want_meta) {
134			if ($type & 1) {
135			$self->respond ("200 OK", $meta, "content-type: text/plain\015\012" . $cache_headers);
136			} else {
137			$self->respond ("404 type $type has no metadata");
138			}
139			} else {
140			$self->respond ("200 OK", $data, (content_type $data) . $cache_headers);
141			}
142	root	1.1
143	root	1.9	} elsif (my $idx = (cf::face::find "res/http$uri") \|\| (cf::face::find "res/http${uri}index.html")) {
144			# TODO: use etag (shudder)
145			my $data = cf::face::get_data $idx, 1;
146			$self->respond ("200 OK", $data, (content_type $data) . $cache_headers);
147
148			} elsif (cf::face::find "res/http$uri/index.html") {
149			$self->respond ("302 dirslash", "", "location: $uri/\015\012");
150	root	1.1
151	root	1.9	} elsif ($uri eq "/debug") { # for debugging
152			my @body = "<html><body>";
153	root	1.1
154	root	1.9	for my $type (6, 5, 4, 3, 2, 1, 0) {
155			push @body, "<h1>$type</h1>";
156	root	1.1
157	root	1.9	for (1 .. cf::face::faces_size - 1) {
158			cf::cede_to_tick;
159
160			next if $type != cf::face::get_type $_;
161			my $name = cf::face::get_name $_;
162			my $id = unpack "H*", cf::face::get_chksum $_, 1;
163			push @body, "$_ <a href='$id'>$name ($id)</a>";
164			push @body, " <a href='${id}M'>(meta)</a>" if $type & 1;
165			push @body, "<br>";
166			}
167	root	1.1	}
168	root	1.9
169			push @body, "</body></html>";
170
171			my $body = join "", @body;
172			utf8::encode $body;
173			$self->respond ("200 OK", $body, "content-type: text/html\015\012");
174
175	root	1.1	} else {
176	root	1.9	$self->respond ("404 not found");
177	root	1.1	}
178
179	root	1.9	cf::cede_to_tick;
180			}
181	root	1.1
182	root	1.9	return if $close; # http 1.0 only currently
183			return if length $self->{rbuf} > 8192; # headers too long
184	root	1.1
185	root	1.9	Coro::AnyEvent::readable $self->{fh}, 6;
186	root	1.1
187	root	1.9	$len = sysread $self->{fh}, $self->{rbuf}, 4096, length $self->{rbuf};
188	root	1.1
189	root	1.9	return if $len <= 0;
190			};
191	root	1.1	}
192
193			our $DETECTOR = ext::tcp::register http => 64, sub {
194			# regex avoids conflict with websockets, which use /ws
195	root	1.2	m{^(?i:GET\|HEAD\|OPTIONS) \ (?! (?i:http://[^/]+)? /ws \ ) }x
196	root	1.1	}, sub {
197			my $self = bless {
198			id => $_[0],
199			fh => $_[1],
200			rbuf => $_[2],
201			wbuf => "",
202			};
203
204	root	1.9	$self->{async} = Coro::async_pool {
205			$Coro::current->nice (4);
206			$Coro::current->{desc} = "http $self->{id}";
207	root	1.1
208	root	1.9	$self->handle_con;
209	root	1.1
210	root	1.9	# this is ultra-ugly, but overall, fast
211			# Coro::AnyEvent::writable $self->{fh}
212			# while length $self->{wbuf};
213
214			# %$self = ();
215	root	1.1	};
216			};
217
218			cf::register_exticmd http_faceurl => sub {
219			my ($ns) = @_;
220
221			"/"
222			};
223