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>
10 # Copyright:: (C) 2002-2005 Tom Gilbert
11 # Copyright:: (C) 2006 Tom Gilbert, Giuseppe Bilotta
12 # Copyright:: (C) 2007 Giuseppe Bilotta, Dmitry Kim
21 error "Couldn't load 'net/https': #{e.pretty_inspect}"
22 error "Secured HTTP connections will fail"
25 # To handle Gzipped pages
31 attr_accessor :no_cache
32 if !instance_methods.include?('raw_body')
36 def body_charset(str=self.raw_body)
37 ctype = self['content-type'] || 'text/html'
38 return nil unless ctype =~ /^text/i || ctype =~ /x(ht)?ml/i
40 charsets = ['latin1'] # should be in config
42 if self['content-type'].match(/charset=["']?([^\s"']+)["']?/i)
44 debug "charset #{charsets.last} added from header"
48 when /<\?xml\s[^>]*encoding=['"]([^\s"'>]+)["'][^>]*\?>/i
50 debug "xml charset #{charsets.last} added from xml pi"
51 when /<(meta\s[^>]*http-equiv=["']?Content-Type["']?[^>]*)>/i
53 if meta =~ /charset=['"]?([^\s'";]+)['"]?/
55 debug "html charset #{charsets.last} added from meta"
62 charsets = self.body_charset(str) or return str
64 charsets.reverse_each do |charset|
65 # XXX: this one is really ugly, but i don't know how to make it better
70 debug "trying #{charset} / offset #{off}"
71 return Iconv.iconv('utf-8//ignore',
73 str.slice(0 .. (-1 - off))).first
75 debug "conversion failed for #{charset} / offset #{off}"
82 def decompress_body(str)
83 method = self['content-encoding']
87 when /gzip/ # Matches gzip, x-gzip, and the non-rfc-compliant gzip;q=\d sent by some servers
88 debug "gunzipping body"
90 return Zlib::GzipReader.new(StringIO.new(str)).read
91 rescue Zlib::Error => e
92 # If we can't unpack the whole stream (e.g. because we're doing a
94 debug "full gunzipping failed (#{e}), trying to recover as much as possible"
97 Zlib::GzipReader.new(StringIO.new(str)).each_byte { |byte|
105 raise "Unhandled content encoding #{method}"
110 return self.body_to_utf(self.decompress_body(self.raw_body))
113 # Read chunks from the body until we have at least _size_ bytes, yielding
114 # the partial text at each chunk. Return the partial body.
115 def partial_body(size=0, &block)
120 self.read_body { |chunk|
122 yield self.body_to_utf(self.decompress_body(partial)) if block_given?
123 break if size and size > 0 and partial.length >= size
126 return self.body_to_utf(self.decompress_body(partial))
131 Net::HTTP.version_1_2
136 # class for making http requests easier (mainly for plugins to use)
137 # this class can check the bot proxy configuration to determine if a proxy
138 # needs to be used, which includes support for per-url proxy configuration.
140 Bot::Config.register Bot::Config::BooleanValue.new('http.use_proxy',
141 :default => false, :desc => "should a proxy be used for HTTP requests?")
142 Bot::Config.register Bot::Config::StringValue.new('http.proxy_uri', :default => false,
143 :desc => "Proxy server to use for HTTP requests (URI, e.g http://proxy.host:port)")
144 Bot::Config.register Bot::Config::StringValue.new('http.proxy_user',
146 :desc => "User for authenticating with the http proxy (if required)")
147 Bot::Config.register Bot::Config::StringValue.new('http.proxy_pass',
149 :desc => "Password for authenticating with the http proxy (if required)")
150 Bot::Config.register Bot::Config::ArrayValue.new('http.proxy_include',
152 :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")
153 Bot::Config.register Bot::Config::ArrayValue.new('http.proxy_exclude',
155 :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")
156 Bot::Config.register Bot::Config::IntegerValue.new('http.max_redir',
158 :desc => "Maximum number of redirections to be used when getting a document")
159 Bot::Config.register Bot::Config::IntegerValue.new('http.expire_time',
161 :desc => "After how many minutes since last use a cached document is considered to be expired")
162 Bot::Config.register Bot::Config::IntegerValue.new('http.max_cache_time',
164 :desc => "After how many minutes since first use a cached document is considered to be expired")
165 Bot::Config.register Bot::Config::IntegerValue.new('http.no_expire_cache',
167 :desc => "Set this to true if you want the bot to never expire the cached pages")
168 Bot::Config.register Bot::Config::IntegerValue.new('http.info_bytes',
170 :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.")
173 attr_accessor :response, :last_used, :first_used, :count, :expires, :date
175 def self.maybe_new(resp)
176 debug "maybe new #{resp}"
177 return nil if resp.no_cache
178 return nil unless Net::HTTPOK === resp ||
179 Net::HTTPMovedPermanently === resp ||
180 Net::HTTPFound === resp ||
181 Net::HTTPPartialContent === resp
183 cc = resp['cache-control']
184 return nil if cc && (cc =~ /no-cache/i)
188 date = Time.httpdate(d)
191 return nil if resp['expires'] && (Time.httpdate(resp['expires']) < date)
193 debug "creating cache obj"
200 @first_used = now if @count == 0
206 debug "checking expired?"
207 if cc = self.response['cache-control'] && cc =~ /must-revalidate/
210 return self.expires < Time.now
213 def setup_headers(hdr)
214 hdr['if-modified-since'] = self.date.rfc2822
216 debug "ims == #{hdr['if-modified-since']}"
218 if etag = self.response['etag']
219 hdr['if-none-match'] = etag
220 debug "etag: #{etag}"
224 def revalidate(resp = self.response)
227 self.date = resp.key?('date') ? Time.httpdate(resp['date']) : Time.now
229 cc = resp['cache-control']
230 if cc && (cc =~ /max-age=(\d+)/)
231 self.expires = self.date + $1.to_i
232 elsif resp.key?('expires')
233 self.expires = Time.httpdate(resp['expires'])
234 elsif lm = resp['last-modified']
235 delta = self.date - Time.httpdate(lm)
236 delta = 10 if delta <= 0
238 self.expires = self.date + delta
240 self.expires = self.date + 300
242 # self.expires = Time.now + 10 # DEBUG
243 debug "expires on #{self.expires}"
253 self.response.raw_body
254 rescue Exception => e
261 # Create the HttpUtil instance, associating it with Bot _bot_
267 'Accept-Charset' => 'utf-8;q=1.0, *;q=0.8',
268 'Accept-Encoding' => 'gzip;q=1, identity;q=0.8, *;q=0.2',
270 "rbot http util #{$version} (http://linuxbrit.co.uk/rbot/)"
272 debug "starting http cache cleanup timer"
273 @timer = @bot.timer.add(300) {
274 self.remove_stale_cache unless @bot.config['http.no_expire_cache']
278 # Clean up on HttpUtil unloading, by stopping the cache cleanup timer.
280 debug 'stopping http cache cleanup timer'
281 @bot.timer.remove(@timer)
284 # This method checks if a proxy is required to access _uri_, by looking at
285 # the values of config values +http.proxy_include+ and +http.proxy_exclude+.
287 # Each of these config values, if set, should be a Regexp the server name and
288 # IP address should be checked against.
290 def proxy_required(uri)
292 if @bot.config["http.proxy_exclude"].empty? && @bot.config["http.proxy_include"].empty?
298 list.concat Resolv.getaddresses(uri.host)
299 rescue StandardError => err
300 warning "couldn't resolve host uri.host"
303 unless @bot.config["http.proxy_exclude"].empty?
304 re = @bot.config["http.proxy_exclude"].collect{|r| Regexp.new(r)}
314 unless @bot.config["http.proxy_include"].empty?
315 re = @bot.config["http.proxy_include"].collect{|r| Regexp.new(r)}
325 debug "using proxy for uri #{uri}?: #{use_proxy}"
329 # _uri_:: URI to create a proxy for
331 # Return a net/http Proxy object, configured for proxying based on the
332 # bot's proxy configuration. See proxy_required for more details on this.
334 def get_proxy(uri, options = {})
346 if @bot.config["http.use_proxy"]
347 if (ENV['http_proxy'])
348 proxy = URI.parse ENV['http_proxy'] rescue nil
350 if (@bot.config["http.proxy_uri"])
351 proxy = URI.parse @bot.config["http.proxy_uri"] rescue nil
354 debug "proxy is set to #{proxy.host} port #{proxy.port}"
355 if proxy_required(uri)
356 proxy_host = proxy.host
357 proxy_port = proxy.port
358 proxy_user = @bot.config["http.proxy_user"]
359 proxy_pass = @bot.config["http.proxy_pass"]
364 h = Net::HTTP.new(uri.host, uri.port, proxy_host, proxy_port, proxy_user, proxy_port)
365 h.use_ssl = true if uri.scheme == "https"
367 h.read_timeout = opts[:read_timeout]
368 h.open_timeout = opts[:open_timeout]
372 # Internal method used to hanlde response _resp_ received when making a
373 # request for URI _uri_.
375 # It follows redirects, optionally yielding them if option :yield is :all.
377 # Also yields and returns the final _resp_.
379 def handle_response(uri, resp, opts, &block) # :yields: resp
380 if Net::HTTPRedirection === resp && opts[:max_redir] >= 0
381 if resp.key?('location')
382 raise 'Too many redirections' if opts[:max_redir] <= 0
383 yield resp if opts[:yield] == :all && block_given?
384 loc = resp['location']
385 new_loc = URI.join(uri.to_s, loc) rescue URI.parse(loc)
387 new_opts[:max_redir] -= 1
388 case opts[:method].to_s.downcase.intern
389 when :post, :"net::http::post"
390 new_opts[:method] = :get
392 if resp['set-cookie']
393 debug "setting cookie #{resp['set-cookie']}"
394 new_opts[:headers] ||= Hash.new
395 new_opts[:headers]['Cookie'] = resp['set-cookie']
397 debug "following the redirect to #{new_loc}"
398 return get_response(new_loc, new_opts, &block)
400 warning ":| redirect w/o location?"
405 alias :body :cooked_body
410 # Net::HTTP wants us to read the whole body here
416 # _uri_:: uri to query (URI object or String)
418 # Generic http transaction method. It will return a Net::HTTPResponse
419 # object or raise an exception
421 # If a block is given, it will yield the response (see :yield option)
423 # Currently supported _options_:
425 # method:: request method [:get (default), :post or :head]
426 # open_timeout:: open timeout for the proxy
427 # read_timeout:: read timeout for the proxy
428 # cache:: should we cache results?
429 # yield:: if :final [default], calls the block for the response object;
430 # if :all, call the block for all intermediate redirects, too
431 # max_redir:: how many redirects to follow before raising the exception
432 # if -1, don't follow redirects, just return them
433 # range:: make a ranged request (usually GET). accepts a string
434 # for HTTP/1.1 "Range:" header (i.e. "bytes=0-1000")
435 # body:: request body (usually for POST requests)
436 # headers:: additional headers to be set for the request. Its value must
437 # be a Hash in the form { 'Header' => 'value' }
439 def get_response(uri_or_s, options = {}, &block) # :yields: resp
440 uri = uri_or_s.kind_of?(URI) ? uri_or_s : URI.parse(uri_or_s.to_s)
442 :max_redir => @bot.config['http.max_redir'],
451 req_class = case opts[:method].to_s.downcase.intern
452 when :head, :"net::http::head"
453 opts[:max_redir] = -1
455 when :get, :"net::http::get"
457 when :post, :"net::http::post"
459 opts[:body] or raise 'post request w/o a body?'
460 warning "refusing to cache POST request" if options[:cache]
463 warning "unsupported method #{opts[:method]}, doing GET"
467 if req_class != Net::HTTP::Get && opts[:range]
468 warning "can't request ranges for #{req_class}"
472 cache_key = "#{opts[:range]}|#{req_class}|#{uri.to_s}"
474 if req_class != Net::HTTP::Get && req_class != Net::HTTP::Head
476 warning "can't cache #{req_class.inspect} requests, working w/o cache"
481 debug "get_response(#{uri}, #{opts.inspect})"
483 if opts[:cache] && cached = @cache[cache_key]
488 return handle_response(uri, cached.response, opts, &block)
492 headers = @headers.dup.merge(opts[:headers] || {})
493 headers['Range'] = opts[:range] if opts[:range]
494 headers['Authorization'] = opts[:auth_head] if opts[:auth_head]
496 cached.setup_headers(headers) if cached && (req_class == Net::HTTP::Get)
497 req = req_class.new(uri.request_uri, headers)
498 if uri.user && uri.password
499 req.basic_auth(uri.user, uri.password)
500 opts[:auth_head] = req['Authorization']
502 req.body = opts[:body] if req_class == Net::HTTP::Post
503 debug "prepared request: #{req.to_hash.inspect}"
505 get_proxy(uri, opts).start do |http|
506 http.request(req) do |resp|
507 resp['x-rbot-location'] = uri.to_s
508 if Net::HTTPNotModified === resp
511 cached.revalidate(resp)
512 rescue Exception => e
515 debug "reusing cached"
516 resp = cached.response
517 elsif Net::HTTPServerError === resp || Net::HTTPClientError === resp
518 debug "http error, deleting cached obj" if cached
519 @cache.delete(cache_key)
522 return handle_response(uri, resp, opts, &block)
524 if cached = CachedObject.maybe_new(resp) rescue nil
525 debug "storing to cache"
526 @cache[cache_key] = cached
531 return handle_response(uri, resp, opts, &block)
536 # _uri_:: uri to query (URI object or String)
538 # Simple GET request, returns (if possible) response body following redirs
539 # and caching if requested, yielding the actual response(s) to the optional
540 # block. See get_response for details on the supported _options_
542 def get(uri, options = {}, &block) # :yields: resp
544 resp = get_response(uri, options, &block)
545 raise "http error: #{resp}" unless Net::HTTPOK === resp ||
546 Net::HTTPPartialContent === resp
548 rescue Exception => e
554 # _uri_:: uri to query (URI object or String)
556 # Simple HEAD request, returns (if possible) response head following redirs
557 # and caching if requested, yielding the actual response(s) to the optional
558 # block. See get_response for details on the supported _options_
560 def head(uri, options = {}, &block) # :yields: resp
561 opts = {:method => :head}.merge(options)
563 resp = get_response(uri, opts, &block)
564 raise "http error #{resp}" if Net::HTTPClientError === resp ||
565 Net::HTTPServerError == resp
567 rescue Exception => e
573 # _uri_:: uri to query (URI object or String)
574 # _data_:: body of the POST
576 # Simple POST request, returns (if possible) response following redirs and
577 # caching if requested, yielding the response(s) to the optional block. See
578 # get_response for details on the supported _options_
580 def post(uri, data, options = {}, &block) # :yields: resp
581 opts = {:method => :post, :body => data, :cache => false}.merge(options)
583 resp = get_response(uri, opts, &block)
584 raise 'http error' unless Net::HTTPOK === resp
586 rescue Exception => e
592 # _uri_:: uri to query (URI object or String)
593 # _nbytes_:: number of bytes to get
595 # Partia GET request, returns (if possible) the first _nbytes_ bytes of the
596 # response body, following redirs and caching if requested, yielding the
597 # actual response(s) to the optional block. See get_response for details on
598 # the supported _options_
600 def get_partial(uri, nbytes = @bot.config['http.info_bytes'], options = {}, &block) # :yields: resp
601 opts = {:range => "bytes=0-#{nbytes}"}.merge(options)
602 return get(uri, opts, &block)
605 def remove_stale_cache
606 debug "Removing stale cache"
608 max_last = @bot.config['http.expire_time'] * 60
609 max_first = @bot.config['http.max_cache_time'] * 60
610 debug "#{@cache.size} pages before"
612 @cache.reject! { |k, val|
613 (now - val.last_used > max_last) || (now - val.first_used > max_first)
616 error "Failed to remove stale cache: #{e.pretty_inspect}"
618 debug "#{@cache.size} pages after"
625 class HttpUtilPlugin < CoreBotModule
628 debug 'initializing httputil'
629 @bot.httputil = Irc::Utils::HttpUtil.new(@bot)
633 debug 'shutting down httputil'
634 @bot.httputil.cleanup