4 # :title: rbot HTTP provider
6 # Author:: Tom Gilbert <tom@linuxbrit.co.uk>
7 # Author:: Giuseppe "Oblomov" Bilotta <giuseppe.bilotta@gmail.com>
8 # Author:: Dmitry "jsn" Kim <dmitry point kim at gmail point com>
17 error "Couldn't load 'net/https': #{e.pretty_inspect}"
18 error "Secured HTTP connections will fail"
21 # To handle Gzipped pages
27 attr_accessor :no_cache
28 if !instance_methods.include?('raw_body')
32 def body_charset(str=self.raw_body)
33 ctype = self['content-type'] || 'text/html'
34 return nil unless ctype =~ /^text/i || ctype =~ /x(ht)?ml/i
36 charsets = ['latin1'] # should be in config
38 if ctype.match(/charset=["']?([^\s"']+)["']?/i)
40 debug "charset #{charsets.last} added from header"
44 when /<\?xml\s[^>]*encoding=['"]([^\s"'>]+)["'][^>]*\?>/i
46 debug "xml charset #{charsets.last} added from xml pi"
47 when /<(meta\s[^>]*http-equiv=["']?Content-Type["']?[^>]*)>/i
49 if meta =~ /charset=['"]?([^\s'";]+)['"]?/
51 debug "html charset #{charsets.last} added from meta"
58 charsets = self.body_charset(str) or return str
60 charsets.reverse_each do |charset|
61 # XXX: this one is really ugly, but i don't know how to make it better
66 debug "trying #{charset} / offset #{off}"
67 return Iconv.iconv('utf-8//ignore',
69 str.slice(0 .. (-1 - off))).first
71 debug "conversion failed for #{charset} / offset #{off}"
78 def decompress_body(str)
79 method = self['content-encoding']
83 when /gzip/ # Matches gzip, x-gzip, and the non-rfc-compliant gzip;q=\d sent by some servers
84 debug "gunzipping body"
86 return Zlib::GzipReader.new(StringIO.new(str)).read
87 rescue Zlib::Error => e
88 # If we can't unpack the whole stream (e.g. because we're doing a
90 debug "full gunzipping failed (#{e}), trying to recover as much as possible"
93 Zlib::GzipReader.new(StringIO.new(str)).each_byte { |byte|
101 debug "inflating body"
102 # From http://www.koders.com/ruby/fid927B4382397E5115AC0ABE21181AB5C1CBDD5C17.aspx?s=thread:
103 # -MAX_WBITS stops zlib from looking for a zlib header
104 inflater = Zlib::Inflate.new(-Zlib::MAX_WBITS)
106 return inflater.inflate(str)
107 rescue Zlib::Error => e
110 # debug "full inflation failed (#{e}), trying to recover as much as possible"
113 raise "Unhandled content encoding #{method}"
118 return self.body_to_utf(self.decompress_body(self.raw_body))
121 # Read chunks from the body until we have at least _size_ bytes, yielding
122 # the partial text at each chunk. Return the partial body.
123 def partial_body(size=0, &block)
128 debug "using body() as partial"
130 yield self.body_to_utf(self.decompress_body(partial)) if block_given?
132 debug "disabling cache"
134 self.read_body { |chunk|
136 yield self.body_to_utf(self.decompress_body(partial)) if block_given?
137 break if size and size > 0 and partial.length >= size
141 return self.body_to_utf(self.decompress_body(partial))
146 Net::HTTP.version_1_2
151 # class for making http requests easier (mainly for plugins to use)
152 # this class can check the bot proxy configuration to determine if a proxy
153 # needs to be used, which includes support for per-url proxy configuration.
155 Bot::Config.register Bot::Config::IntegerValue.new('http.read_timeout',
156 :default => 10, :desc => "Default read timeout for HTTP connections")
157 Bot::Config.register Bot::Config::IntegerValue.new('http.open_timeout',
158 :default => 20, :desc => "Default open timeout for HTTP connections")
159 Bot::Config.register Bot::Config::BooleanValue.new('http.use_proxy',
160 :default => false, :desc => "should a proxy be used for HTTP requests?")
161 Bot::Config.register Bot::Config::StringValue.new('http.proxy_uri', :default => false,
162 :desc => "Proxy server to use for HTTP requests (URI, e.g http://proxy.host:port)")
163 Bot::Config.register Bot::Config::StringValue.new('http.proxy_user',
165 :desc => "User for authenticating with the http proxy (if required)")
166 Bot::Config.register Bot::Config::StringValue.new('http.proxy_pass',
168 :desc => "Password for authenticating with the http proxy (if required)")
169 Bot::Config.register Bot::Config::ArrayValue.new('http.proxy_include',
171 :desc => "List of regexps to check against a URI's hostname/ip to see if we should use the proxy to access this URI. All URIs are proxied by default if the proxy is set, so this is only required to re-include URIs that might have been excluded by the exclude list. e.g. exclude /.*\.foo\.com/, include bar\.foo\.com")
172 Bot::Config.register Bot::Config::ArrayValue.new('http.proxy_exclude',
174 :desc => "List of regexps to check against a URI's hostname/ip to see if we should use avoid the proxy to access this URI and access it directly")
175 Bot::Config.register Bot::Config::IntegerValue.new('http.max_redir',
177 :desc => "Maximum number of redirections to be used when getting a document")
178 Bot::Config.register Bot::Config::IntegerValue.new('http.expire_time',
180 :desc => "After how many minutes since last use a cached document is considered to be expired")
181 Bot::Config.register Bot::Config::IntegerValue.new('http.max_cache_time',
183 :desc => "After how many minutes since first use a cached document is considered to be expired")
184 Bot::Config.register Bot::Config::IntegerValue.new('http.no_expire_cache',
186 :desc => "Set this to true if you want the bot to never expire the cached pages")
187 Bot::Config.register Bot::Config::IntegerValue.new('http.info_bytes',
189 :desc => "How many bytes to download from a web page to find some information. Set to 0 to let the bot download the whole page.")
192 attr_accessor :response, :last_used, :first_used, :count, :expires, :date
194 def self.maybe_new(resp)
195 debug "maybe new #{resp}"
196 return nil if resp.no_cache
197 return nil unless Net::HTTPOK === resp ||
198 Net::HTTPMovedPermanently === resp ||
199 Net::HTTPFound === resp ||
200 Net::HTTPPartialContent === resp
202 cc = resp['cache-control']
203 return nil if cc && (cc =~ /no-cache/i)
207 date = Time.httpdate(d)
210 return nil if resp['expires'] && (Time.httpdate(resp['expires']) < date)
212 debug "creating cache obj"
219 @first_used = now if @count == 0
225 debug "checking expired?"
226 if cc = self.response['cache-control'] && cc =~ /must-revalidate/
229 return self.expires < Time.now
232 def setup_headers(hdr)
233 hdr['if-modified-since'] = self.date.rfc2822
235 debug "ims == #{hdr['if-modified-since']}"
237 if etag = self.response['etag']
238 hdr['if-none-match'] = etag
239 debug "etag: #{etag}"
243 def revalidate(resp = self.response)
246 self.date = resp.key?('date') ? Time.httpdate(resp['date']) : Time.now
248 cc = resp['cache-control']
249 if cc && (cc =~ /max-age=(\d+)/)
250 self.expires = self.date + $1.to_i
251 elsif resp.key?('expires')
252 self.expires = Time.httpdate(resp['expires'])
253 elsif lm = resp['last-modified']
254 delta = self.date - Time.httpdate(lm)
255 delta = 10 if delta <= 0
257 self.expires = self.date + delta
259 self.expires = self.date + 300
261 # self.expires = Time.now + 10 # DEBUG
262 debug "expires on #{self.expires}"
272 self.response.raw_body
273 rescue Exception => e
280 # Create the HttpUtil instance, associating it with Bot _bot_
286 'Accept-Charset' => 'utf-8;q=1.0, *;q=0.8',
287 'Accept-Encoding' => 'gzip;q=1, deflate;q=1, identity;q=0.8, *;q=0.2',
289 "rbot http util #{$version} (#{Irc::Bot::SOURCE_URL})"
291 debug "starting http cache cleanup timer"
292 @timer = @bot.timer.add(300) {
293 self.remove_stale_cache unless @bot.config['http.no_expire_cache']
297 # Clean up on HttpUtil unloading, by stopping the cache cleanup timer.
299 debug 'stopping http cache cleanup timer'
300 @bot.timer.remove(@timer)
303 # This method checks if a proxy is required to access _uri_, by looking at
304 # the values of config values +http.proxy_include+ and +http.proxy_exclude+.
306 # Each of these config values, if set, should be a Regexp the server name and
307 # IP address should be checked against.
309 def proxy_required(uri)
311 if @bot.config["http.proxy_exclude"].empty? && @bot.config["http.proxy_include"].empty?
317 list.concat Resolv.getaddresses(uri.host)
318 rescue StandardError => err
319 warning "couldn't resolve host uri.host"
322 unless @bot.config["http.proxy_exclude"].empty?
323 re = @bot.config["http.proxy_exclude"].collect{|r| Regexp.new(r)}
333 unless @bot.config["http.proxy_include"].empty?
334 re = @bot.config["http.proxy_include"].collect{|r| Regexp.new(r)}
344 debug "using proxy for uri #{uri}?: #{use_proxy}"
348 # _uri_:: URI to create a proxy for
350 # Return a net/http Proxy object, configured for proxying based on the
351 # bot's proxy configuration. See proxy_required for more details on this.
353 def get_proxy(uri, options = {})
355 :read_timeout => @bot.config["http.read_timeout"],
356 :open_timeout => @bot.config["http.open_timeout"]
365 if @bot.config["http.use_proxy"]
366 if (ENV['http_proxy'])
367 proxy = URI.parse ENV['http_proxy'] rescue nil
369 if (@bot.config["http.proxy_uri"])
370 proxy = URI.parse @bot.config["http.proxy_uri"] rescue nil
373 debug "proxy is set to #{proxy.host} port #{proxy.port}"
374 if proxy_required(uri)
375 proxy_host = proxy.host
376 proxy_port = proxy.port
377 proxy_user = @bot.config["http.proxy_user"]
378 proxy_pass = @bot.config["http.proxy_pass"]
383 h = Net::HTTP.new(uri.host, uri.port, proxy_host, proxy_port, proxy_user, proxy_port)
384 h.use_ssl = true if uri.scheme == "https"
386 h.read_timeout = opts[:read_timeout]
387 h.open_timeout = opts[:open_timeout]
391 # Internal method used to hanlde response _resp_ received when making a
392 # request for URI _uri_.
394 # It follows redirects, optionally yielding them if option :yield is :all.
396 # Also yields and returns the final _resp_.
398 def handle_response(uri, resp, opts, &block) # :yields: resp
399 if Net::HTTPRedirection === resp && opts[:max_redir] >= 0
400 if resp.key?('location')
401 raise 'Too many redirections' if opts[:max_redir] <= 0
402 yield resp if opts[:yield] == :all && block_given?
403 loc = resp['location']
404 new_loc = URI.join(uri.to_s, loc) rescue URI.parse(loc)
406 new_opts[:max_redir] -= 1
407 case opts[:method].to_s.downcase.intern
408 when :post, :"net::http::post"
409 new_opts[:method] = :get
411 if resp['set-cookie']
412 debug "setting cookie #{resp['set-cookie']}"
413 new_opts[:headers] ||= Hash.new
414 new_opts[:headers]['Cookie'] = resp['set-cookie']
416 debug "following the redirect to #{new_loc}"
417 return get_response(new_loc, new_opts, &block)
419 warning ":| redirect w/o location?"
424 alias :body :cooked_body
426 unless resp['content-type']
427 debug "No content type, guessing"
428 resp['content-type'] =
429 case resp['x-rbot-location']
435 'application/xml+xhtml'
436 when /.(gif|png|jpe?g|jp2|tiff?)$/i
437 "image/#{$1.sub(/^jpg$/,'jpeg').sub(/^tif$/,'tiff')}"
439 'application/octetstream'
445 # Net::HTTP wants us to read the whole body here
451 # _uri_:: uri to query (URI object or String)
453 # Generic http transaction method. It will return a Net::HTTPResponse
454 # object or raise an exception
456 # If a block is given, it will yield the response (see :yield option)
458 # Currently supported _options_:
460 # method:: request method [:get (default), :post or :head]
461 # open_timeout:: open timeout for the proxy
462 # read_timeout:: read timeout for the proxy
463 # cache:: should we cache results?
464 # yield:: if :final [default], calls the block for the response object;
465 # if :all, call the block for all intermediate redirects, too
466 # max_redir:: how many redirects to follow before raising the exception
467 # if -1, don't follow redirects, just return them
468 # range:: make a ranged request (usually GET). accepts a string
469 # for HTTP/1.1 "Range:" header (i.e. "bytes=0-1000")
470 # body:: request body (usually for POST requests)
471 # headers:: additional headers to be set for the request. Its value must
472 # be a Hash in the form { 'Header' => 'value' }
474 def get_response(uri_or_s, options = {}, &block) # :yields: resp
475 uri = uri_or_s.kind_of?(URI) ? uri_or_s : URI.parse(uri_or_s.to_s)
477 :max_redir => @bot.config['http.max_redir'],
486 req_class = case opts[:method].to_s.downcase.intern
487 when :head, :"net::http::head"
488 opts[:max_redir] = -1
490 when :get, :"net::http::get"
492 when :post, :"net::http::post"
494 opts[:body] or raise 'post request w/o a body?'
495 warning "refusing to cache POST request" if options[:cache]
498 warning "unsupported method #{opts[:method]}, doing GET"
502 if req_class != Net::HTTP::Get && opts[:range]
503 warning "can't request ranges for #{req_class}"
507 cache_key = "#{opts[:range]}|#{req_class}|#{uri.to_s}"
509 if req_class != Net::HTTP::Get && req_class != Net::HTTP::Head
511 warning "can't cache #{req_class.inspect} requests, working w/o cache"
516 debug "get_response(#{uri}, #{opts.inspect})"
518 if opts[:cache] && cached = @cache[cache_key]
523 return handle_response(uri, cached.response, opts, &block)
527 headers = @headers.dup.merge(opts[:headers] || {})
528 headers['Range'] = opts[:range] if opts[:range]
529 headers['Authorization'] = opts[:auth_head] if opts[:auth_head]
531 cached.setup_headers(headers) if cached && (req_class == Net::HTTP::Get)
532 req = req_class.new(uri.request_uri, headers)
533 if uri.user && uri.password
534 req.basic_auth(uri.user, uri.password)
535 opts[:auth_head] = req['Authorization']
537 req.body = opts[:body] if req_class == Net::HTTP::Post
538 debug "prepared request: #{req.to_hash.inspect}"
541 get_proxy(uri, opts).start do |http|
542 http.request(req) do |resp|
543 resp['x-rbot-location'] = uri.to_s
544 if Net::HTTPNotModified === resp
547 cached.revalidate(resp)
548 rescue Exception => e
551 debug "reusing cached"
552 resp = cached.response
553 elsif Net::HTTPServerError === resp || Net::HTTPClientError === resp
554 debug "http error, deleting cached obj" if cached
555 @cache.delete(cache_key)
558 return handle_response(uri, resp, opts, &block)
560 if cached = CachedObject.maybe_new(resp) rescue nil
561 debug "storing to cache"
562 @cache[cache_key] = cached
567 return handle_response(uri, resp, opts, &block)
570 rescue Exception => e
576 # _uri_:: uri to query (URI object or String)
578 # Simple GET request, returns (if possible) response body following redirs
579 # and caching if requested, yielding the actual response(s) to the optional
580 # block. See get_response for details on the supported _options_
582 def get(uri, options = {}, &block) # :yields: resp
584 resp = get_response(uri, options, &block)
585 raise "http error: #{resp}" unless Net::HTTPOK === resp ||
586 Net::HTTPPartialContent === resp
588 rescue Exception => e
594 # _uri_:: uri to query (URI object or String)
596 # Simple HEAD request, returns (if possible) response head following redirs
597 # and caching if requested, yielding the actual response(s) to the optional
598 # block. See get_response for details on the supported _options_
600 def head(uri, options = {}, &block) # :yields: resp
601 opts = {:method => :head}.merge(options)
603 resp = get_response(uri, opts, &block)
604 raise "http error #{resp}" if Net::HTTPClientError === resp ||
605 Net::HTTPServerError == resp
607 rescue Exception => e
613 # _uri_:: uri to query (URI object or String)
614 # _data_:: body of the POST
616 # Simple POST request, returns (if possible) response following redirs and
617 # caching if requested, yielding the response(s) to the optional block. See
618 # get_response for details on the supported _options_
620 def post(uri, data, options = {}, &block) # :yields: resp
621 opts = {:method => :post, :body => data, :cache => false}.merge(options)
623 resp = get_response(uri, opts, &block)
624 raise 'http error' unless Net::HTTPOK === resp
626 rescue Exception => e
632 # _uri_:: uri to query (URI object or String)
633 # _nbytes_:: number of bytes to get
635 # Partial GET request, returns (if possible) the first _nbytes_ bytes of the
636 # response body, following redirs and caching if requested, yielding the
637 # actual response(s) to the optional block. See get_response for details on
638 # the supported _options_
640 def get_partial(uri, nbytes = @bot.config['http.info_bytes'], options = {}, &block) # :yields: resp
641 opts = {:range => "bytes=0-#{nbytes}"}.merge(options)
642 return get(uri, opts, &block)
645 def remove_stale_cache
646 debug "Removing stale cache"
648 max_last = @bot.config['http.expire_time'] * 60
649 max_first = @bot.config['http.max_cache_time'] * 60
650 debug "#{@cache.size} pages before"
652 @cache.reject! { |k, val|
653 (now - val.last_used > max_last) || (now - val.first_used > max_first)
656 error "Failed to remove stale cache: #{e.pretty_inspect}"
658 debug "#{@cache.size} pages after"
665 class HttpUtilPlugin < CoreBotModule
668 debug 'initializing httputil'
669 @bot.httputil = Irc::Utils::HttpUtil.new(@bot)
673 debug 'shutting down httputil'
674 @bot.httputil.cleanup