server/ext/tcp_http.ext

#! perl # optional depends=tcp

# http server - this tries 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, "";

         delete $self->{ww} unless $len; # in case of errors, stop
         delete $self->{ww} unless length $self->{wbuf};
      } if length $self->{wbuf};
   }
}

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

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

sub respond_304 {
   $_[0]->send ("HTTP/1.1 304 cache\015\012"
                . $_[0]{ohdr}
                . "\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: public, max-age=8640000\015\012"
                    . "etag: \"0\"\015\012";

my $revalidate_headers = "cache-control: public, max-age=60, must-revalidate\015\012";

my $nocache_headers = "cache-control: no-cache, max-age=0\015\012";

sub content_type($$) {
   return "content-type: image/png\015\012"                             if $_[1] =~ /^\x89PNG/;
   return "content-type: image/jpeg\015\012"                            if $_[1] =~ /^......JFIF/s;
   return "content-type: audio/wav\015\012"                             if $_[1] =~ /^RIFF/;
   return "content-type: audio/ogg\015\012"                             if $_[1] =~ /^OggS/;
   return "content-type: text/css; charset=utf-8\015\012"               if $_[0] =~ /\.css$/;
   return "content-type: application/javascript; charset=utf-8\015\012" if $_[0] =~ /\.js$/;
   return "content-type: text/html; charset=utf-8\015\012"              if $_[1] =~ /^</;

   "content-type: text/plain; charset=utf-8\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

         if (0) {
         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::face::HASH{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;
               next;
            }

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

            (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; charset=utf-8\015\012" . $cache_headers);
               } else {
                  $self->respond ("404 type $type has no metadata");
               }
            } else {
               $self->respond ("200 OK", $data, (content_type $uri, $data) . $cache_headers);
            }

         } elsif (my $idx = (cf::face::find "res/http$uri") || (cf::face::find "res/http${uri}index.html")) {
            my $data =              cf::face::get_data $idx;
            my $csum = unpack "H*", cf::face::get_csum $idx;
            my $hdr  = "etag: \"$csum\"\015\012" . $revalidate_headers;

            if ($req =~ /if-none-match: "$csum"/i) { # dirtiest hack evar
               $self->respond_304;
               next;
            }

            $self->respond ("200 OK", $data, (content_type $uri, $data) . $hdr);

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

         } elsif ($uri eq "/debug") { # for debugging
            my @body = <<EOF;
<html><head><style>
th:nth-child(1) { text-align: right; }
th:nth-child(2) { text-align: left;  }
th:nth-child(3) { text-align: right; }
th:nth-child(4) { text-align: left;  }
th:nth-child(5) { text-align: left;  }
td:nth-child(1) { text-align: right;  font-family: monospace;}
td:nth-child(2) { text-align: left;   font-family: monospace;}
td:nth-child(3) { text-align: right;  font-family: monospace;}
td:nth-child(4) { text-align: left;  }
td:nth-child(5) { text-align: left;  }
</style><body>
EOF
            for my $type (6, 5, 4, 3, 2, 1, 0) {
               push @body, "<h1>$type</h1><table><tr><th>#</th><th>csum</th><th>size</th><th>name</th><th>meta</th></tr>";

               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_csum $_, 0;
                  push @body, "<tr><td>$_</td><td>$id</td><td>$cf::face::SIZE[0][$_]</td><td><a href='$id'>$name</a></td>";
                  push @body, "<td><a href='${id}M'>meta</a>" if $type & 1;
                  push @body, "</tr>";
               }

               push @body, "</table>";
            }

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

            my $body = join "", @body;
            utf8::encode $body;
            $self->respond ("200 OK", $body, "content-type: text/html; charset=utf-8\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;
   };
};

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

   "/"
};

Revision:	1.18
Committed:	Sat Jan 28 22:38:45 2017 UTC (7 years, 3 months ago) by root
Branch:	MAIN
CVS Tags:	HEAD
Changes since 1.17:	+8 -2 lines
Log Message:	* empty log message *
#	User	Rev	Content
1	root	1.1	#! perl # optional depends=tcp
2
3	root	1.14	# http server - this tries to speak enough of http 1.1 (and 1.0)
4	root	1.9	# 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	root	1.13	delete $self->{ww} unless $len; # in case of errors, stop
23	root	1.11	delete $self->{ww} unless length $self->{wbuf};
24	root	1.9	} if length $self->{wbuf};
25			}
26	root	1.1	}
27
28			sub fatal {
29			my ($self) = @_;
30
31			$self->send ("HTTP/1.1 500 internal error\015\012");
32			}
33
34	root	1.18	sub respond_304 {
35			$_[0]->send ("HTTP/1.1 304 cache\015\012"
36			. $_[0]{ohdr}
37			. "\015\012");
38			}
39
40	root	1.1	sub respond {
41			$_[0]->send ("HTTP/1.1 $_[1]\015\012"
42			. "content-length: " . (0 + length $_[2]) . "\015\012"
43			. "access-control-allow-origin: *\015\012"
44	root	1.6	. $_[0]{ohdr}
45	root	1.9	. "$_[3]\015\012" . ($_[0]{give_head} ? "" : $_[2]));
46	root	1.1	}
47
48	root	1.17	my $cache_headers = "cache-control: public, max-age=8640000\015\012"
49	root	1.1	. "etag: \"0\"\015\012";
50
51	root	1.17	my $revalidate_headers = "cache-control: public, max-age=60, must-revalidate\015\012";
52
53			my $nocache_headers = "cache-control: no-cache, max-age=0\015\012";
54
55	root	1.15	sub content_type($$) {
56	root	1.16	return "content-type: image/png\015\012" if $_[1] =~ /^\x89PNG/;
57			return "content-type: image/jpeg\015\012" if $_[1] =~ /^......JFIF/s;
58			return "content-type: audio/wav\015\012" if $_[1] =~ /^RIFF/;
59			return "content-type: audio/ogg\015\012" if $_[1] =~ /^OggS/;
60			return "content-type: text/css; charset=utf-8\015\012" if $_[0] =~ /\.css$/;
61			return "content-type: application/javascript; charset=utf-8\015\012" if $_[0] =~ /\.js$/;
62			return "content-type: text/html; charset=utf-8\015\012" if $_[1] =~ /^</;
63	root	1.1
64	root	1.16	"content-type: text/plain; charset=utf-8\015\012"
65	root	1.1	}
66
67	root	1.9	sub handle_con {
68	root	1.1	my ($self) = @_;
69
70	root	1.9	my ($method, $uri, $http, $req, $close, $len);
71	root	1.1
72	root	1.9	while () {
73			while ($self->{rbuf} =~ s/^( (?: [^\015]+ \| . )+? \015\012)\015\012//xs) {
74			$req = $1;
75
76			# we ignore headers atm.
77
78			$req =~ m%^(\S+) (\S+) HTTP/([0-9.]+)\015\012%ig
79			or return $self->respond ("400 bad request");
80
81			$method = uc $1;
82			$uri = $2;
83			$http = $3;
84
85			$uri =~ s%^http://[^/]*%%i; # just in case
86			$uri =~ s/%([0-9a-fA-F][0-9a-fA-F])/chr hex $1/ge; # %-decode
87
88	root	1.14	if (0) {
89			my %hdr;
90
91			$hdr{lc $1} .= ",$2"
92			while $req =~ /\G
93			([^:\000-\037]*):
94			[\011\040]*
95			((?: [^\012]+ \| \012[\011\040] )*)
96			\012
97			/gxc;
98
99			$req =~ /\G$/
100			or return $self->respond ("400 bad request");
101
102			# remove the "," prefix we added to all headers above
103			substr $_, 0, 1, ""
104			for values %hdr;
105			}
106	root	1.9
107			if ($http == 1.0) {
108			if ($req =~ /^connection\s:\skeep-alive/mi) {
109			$self->{ohdr} = "connection: keep-alive\015\012";
110			} else {
111			$self->{ohdr} = "connection: close\015\012";
112			$close = 1;
113			}
114			}
115	root	1.1
116	root	1.9	$self->{give_head} = $method eq "HEAD"
117			and $method = "GET";
118	root	1.1
119	root	1.9	cf::debug "HTTP $method: $self->{id} $uri";
120	root	1.6
121	root	1.9	if ($method ne "GET") {
122			$self->respond ("405 no $method");
123			next;
124	root	1.6	}
125	root	1.1
126	root	1.9	# not needed for GET, but who knows
127			$self->send ("HTTP/1.1 100 go on\015\012")
128			if $req =~ /^expect:.*\b100-continue\b/i;
129
130			if ($uri =~ m%^/([0-9a-f]+)(M?)$%) { # faces
131			my $want_meta = $2;
132	root	1.12	my $idx = $cf::face::HASH{pack "H*", $1};
133	root	1.9
134			$idx
135			or do { $self->respond ("404 illegal face name"), next };
136
137			if ($req =~ /if-none-match/i) { # dirtiest hack evar
138	root	1.18	$self->respond_304;
139	root	1.9	next;
140			}
141	root	1.1
142	root	1.12	my $type = cf::face::get_type $idx;
143			my $data = cf::face::get_data $idx;
144	root	1.1
145	root	1.9	(my $meta, $data) = unpack "(w/a)", $data
146			if $type & 1;
147	root	1.1
148	root	1.9	if ($want_meta) {
149			if ($type & 1) {
150	root	1.16	$self->respond ("200 OK", $meta, "content-type: text/plain; charset=utf-8\015\012" . $cache_headers);
151	root	1.9	} else {
152			$self->respond ("404 type $type has no metadata");
153			}
154			} else {
155	root	1.15	$self->respond ("200 OK", $data, (content_type $uri, $data) . $cache_headers);
156	root	1.9	}
157	root	1.1
158	root	1.9	} elsif (my $idx = (cf::face::find "res/http$uri") \|\| (cf::face::find "res/http${uri}index.html")) {
159	root	1.17	my $data = cf::face::get_data $idx;
160			my $csum = unpack "H*", cf::face::get_csum $idx;
161			my $hdr = "etag: \"$csum\"\015\012" . $revalidate_headers;
162
163			if ($req =~ /if-none-match: "$csum"/i) { # dirtiest hack evar
164	root	1.18	$self->respond_304;
165	root	1.17	next;
166			}
167
168			$self->respond ("200 OK", $data, (content_type $uri, $data) . $hdr);
169	root	1.9
170			} elsif (cf::face::find "res/http$uri/index.html") {
171	root	1.17	$self->respond ("302 dirslash", "", "location: $uri/\015\012" . $nocache_headers);
172	root	1.1
173	root	1.9	} elsif ($uri eq "/debug") { # for debugging
174	root	1.14	my @body = <<EOF;
175			<html><head><style>
176			th:nth-child(1) { text-align: right; }
177			th:nth-child(2) { text-align: left; }
178			th:nth-child(3) { text-align: right; }
179			th:nth-child(4) { text-align: left; }
180			th:nth-child(5) { text-align: left; }
181			td:nth-child(1) { text-align: right; font-family: monospace;}
182			td:nth-child(2) { text-align: left; font-family: monospace;}
183			td:nth-child(3) { text-align: right; font-family: monospace;}
184			td:nth-child(4) { text-align: left; }
185			td:nth-child(5) { text-align: left; }
186			</style><body>
187			EOF
188	root	1.9	for my $type (6, 5, 4, 3, 2, 1, 0) {
189	root	1.12	push @body, "<h1>$type</h1><table><tr><th>#</th><th>csum</th><th>size</th><th>name</th><th>meta</th></tr>";
190	root	1.1
191	root	1.9	for (1 .. cf::face::faces_size - 1) {
192			cf::cede_to_tick;
193
194			next if $type != cf::face::get_type $_;
195			my $name = cf::face::get_name $_;
196	root	1.12	my $id = unpack "H*", cf::face::get_csum $_, 0;
197			push @body, "<tr><td>$_</td><td>$id</td><td>$cf::face::SIZE[0][$_]</td><td><a href='$id'>$name</a></td>";
198			push @body, "<td><a href='${id}M'>meta</a>" if $type & 1;
199			push @body, "</tr>";
200	root	1.9	}
201	root	1.12
202			push @body, "</table>";
203	root	1.1	}
204	root	1.9
205			push @body, "</body></html>";
206
207			my $body = join "", @body;
208			utf8::encode $body;
209	root	1.16	$self->respond ("200 OK", $body, "content-type: text/html; charset=utf-8\015\012");
210	root	1.9
211	root	1.1	} else {
212	root	1.9	$self->respond ("404 not found");
213	root	1.1	}
214
215	root	1.9	cf::cede_to_tick;
216			}
217	root	1.1
218	root	1.9	return if $close; # http 1.0 only currently
219			return if length $self->{rbuf} > 8192; # headers too long
220	root	1.1
221	root	1.9	Coro::AnyEvent::readable $self->{fh}, 6;
222	root	1.1
223	root	1.9	$len = sysread $self->{fh}, $self->{rbuf}, 4096, length $self->{rbuf};
224	root	1.1
225	root	1.9	return if $len <= 0;
226			};
227	root	1.1	}
228
229			our $DETECTOR = ext::tcp::register http => 64, sub {
230			# regex avoids conflict with websockets, which use /ws
231	root	1.2	m{^(?i:GET\|HEAD\|OPTIONS) \ (?! (?i:http://[^/]+)? /ws \ ) }x
232	root	1.1	}, sub {
233			my $self = bless {
234			id => $_[0],
235			fh => $_[1],
236			rbuf => $_[2],
237			wbuf => "",
238			};
239
240	root	1.9	$self->{async} = Coro::async_pool {
241			$Coro::current->nice (4);
242			$Coro::current->{desc} = "http $self->{id}";
243	root	1.1
244	root	1.9	$self->handle_con;
245	root	1.1	};
246			};
247
248			cf::register_exticmd http_faceurl => sub {
249			my ($ns) = @_;
250
251			"/"
252			};
253