root/lib/rbot/core/utils/httputil.rb @ 79d95448a9dd48c8273255be90945acfe53043d4

Revision 79d95448a9dd48c8273255be90945acfe53043d4, 22.4 KB (checked in by Matthias -apoc- Hecker <apoc@…>, 6 months ago)

Small httputil.rb fix. Wrong Argument port instead of pass.
  • Property mode set to 100644
Line 
1#-- vim:sw=2:et
2#++
3#
4# :title: rbot HTTP provider
5#
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>
9
10require 'resolv'
11require 'net/http'
12require 'cgi'
13require 'iconv'
14begin
15  require 'net/https'
16rescue LoadError => e
17  error "Couldn't load 'net/https':  #{e.pretty_inspect}"
18  error "Secured HTTP connections will fail"
19end
20
21# To handle Gzipped pages
22require 'stringio'
23require 'zlib'
24
25module ::Net
26  class HTTPResponse
27    attr_accessor :no_cache
28    unless method_defined? :raw_body
29      alias :raw_body :body
30    end
31
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
35
36      charsets = ['latin1'] # should be in config
37
38      if ctype.match(/charset=["']?([^\s"']+)["']?/i)
39        charsets << $1
40        debug "charset #{charsets.last} added from header"
41      end
42
43      case str
44      when /<\?xml\s[^>]*encoding=['"]([^\s"'>]+)["'][^>]*\?>/i
45        charsets << $1
46        debug "xml charset #{charsets.last} added from xml pi"
47      when /<(meta\s[^>]*http-equiv=["']?Content-Type["']?[^>]*)>/i
48        meta = $1
49        if meta =~ /charset=['"]?([^\s'";]+)['"]?/
50          charsets << $1
51          debug "html charset #{charsets.last} added from meta"
52        end
53      end
54      return charsets.uniq
55    end
56
57    def body_to_utf(str)
58      charsets = self.body_charset(str) or return str
59
60      charsets.reverse_each do |charset|
61        # XXX: this one is really ugly, but i don't know how to make it better
62        #  -jsn
63
64        0.upto(5) do |off|
65          begin
66            debug "trying #{charset} / offset #{off}"
67            return Iconv.iconv('utf-8//ignore',
68                               charset,
69                               str.slice(0 .. (-1 - off))).first
70          rescue
71            debug "conversion failed for #{charset} / offset #{off}"
72          end
73        end
74      end
75      return str
76    end
77
78    def decompress_body(str)
79      method = self['content-encoding']
80      case method
81      when nil
82        return str
83      when /gzip/ # Matches gzip, x-gzip, and the non-rfc-compliant gzip;q=\d sent by some servers
84        debug "gunzipping body"
85        begin
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
89          # partial read
90          debug "full gunzipping failed (#{e}), trying to recover as much as possible"
91          ret = ""
92          begin
93            Zlib::GzipReader.new(StringIO.new(str)).each_byte { |byte|
94              ret << byte
95            }
96          rescue
97          end
98          return ret
99        end
100      when 'deflate'
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)
105        begin
106          return inflater.inflate(str)
107        rescue Zlib::Error => e
108          raise e
109          # TODO
110          # debug "full inflation failed (#{e}), trying to recover as much as possible"
111        end
112      when /^(?:iso-8859-\d+|windows-\d+|utf-8|utf8)$/i
113        # B0rked servers (Freshmeat being one of them) sometimes return the charset
114        # in the content-encoding; in this case we assume that the document has
115        # a standarc content-encoding
116        old_hsh = self.to_hash
117        self['content-type']= self['content-type']+"; charset="+method.downcase
118        warning "Charset vs content-encoding confusion, trying to recover: from\n#{old_hsh.pretty_inspect}to\n#{self.to_hash.pretty_inspect}"
119        return str
120      else
121        debug self.to_hash
122        raise "Unhandled content encoding #{method}"
123      end
124    end
125
126    def cooked_body
127      return self.body_to_utf(self.decompress_body(self.raw_body))
128    end
129
130    # Read chunks from the body until we have at least _size_ bytes, yielding
131    # the partial text at each chunk. Return the partial body.
132    def partial_body(size=0, &block)
133
134      partial = String.new
135
136      if @read
137        debug "using body() as partial"
138        partial = self.body
139        yield self.body_to_utf(self.decompress_body(partial)) if block_given?
140      else
141        debug "disabling cache"
142        self.no_cache = true
143        self.read_body { |chunk|
144          partial << chunk
145          yield self.body_to_utf(self.decompress_body(partial)) if block_given?
146          break if size and size > 0 and partial.length >= size
147        }
148      end
149
150      return self.body_to_utf(self.decompress_body(partial))
151    end
152  end
153end
154
155Net::HTTP.version_1_2
156
157module ::Irc
158module Utils
159
160# class for making http requests easier (mainly for plugins to use)
161# this class can check the bot proxy configuration to determine if a proxy
162# needs to be used, which includes support for per-url proxy configuration.
163class HttpUtil
164    Bot::Config.register Bot::Config::IntegerValue.new('http.read_timeout',
165      :default => 10, :desc => "Default read timeout for HTTP connections")
166    Bot::Config.register Bot::Config::IntegerValue.new('http.open_timeout',
167      :default => 20, :desc => "Default open timeout for HTTP connections")
168    Bot::Config.register Bot::Config::BooleanValue.new('http.use_proxy',
169      :default => false, :desc => "should a proxy be used for HTTP requests?")
170    Bot::Config.register Bot::Config::StringValue.new('http.proxy_uri', :default => false,
171      :desc => "Proxy server to use for HTTP requests (URI, e.g http://proxy.host:port)")
172    Bot::Config.register Bot::Config::StringValue.new('http.proxy_user',
173      :default => nil,
174      :desc => "User for authenticating with the http proxy (if required)")
175    Bot::Config.register Bot::Config::StringValue.new('http.proxy_pass',
176      :default => nil,
177      :desc => "Password for authenticating with the http proxy (if required)")
178    Bot::Config.register Bot::Config::ArrayValue.new('http.proxy_include',
179      :default => [],
180      :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")
181    Bot::Config.register Bot::Config::ArrayValue.new('http.proxy_exclude',
182      :default => [],
183      :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")
184    Bot::Config.register Bot::Config::IntegerValue.new('http.max_redir',
185      :default => 5,
186      :desc => "Maximum number of redirections to be used when getting a document")
187    Bot::Config.register Bot::Config::IntegerValue.new('http.expire_time',
188      :default => 60,
189      :desc => "After how many minutes since last use a cached document is considered to be expired")
190    Bot::Config.register Bot::Config::IntegerValue.new('http.max_cache_time',
191      :default => 60*24,
192      :desc => "After how many minutes since first use a cached document is considered to be expired")
193    Bot::Config.register Bot::Config::BooleanValue.new('http.no_expire_cache',
194      :default => false,
195      :desc => "Set this to true if you want the bot to never expire the cached pages")
196    Bot::Config.register Bot::Config::IntegerValue.new('http.info_bytes',
197      :default => 8192,
198      :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.")
199
200  class CachedObject
201    attr_accessor :response, :last_used, :first_used, :count, :expires, :date
202
203    def self.maybe_new(resp)
204      debug "maybe new #{resp}"
205      return nil if resp.no_cache
206      return nil unless Net::HTTPOK === resp ||
207      Net::HTTPMovedPermanently === resp ||
208      Net::HTTPFound === resp ||
209      Net::HTTPPartialContent === resp
210
211      cc = resp['cache-control']
212      return nil if cc && (cc =~ /no-cache/i)
213
214      date = Time.now
215      if d = resp['date']
216        date = Time.httpdate(d)
217      end
218
219      return nil if resp['expires'] && (Time.httpdate(resp['expires']) < date)
220
221      debug "creating cache obj"
222
223      self.new(resp)
224    end
225
226    def use
227      now = Time.now
228      @first_used = now if @count == 0
229      @last_used = now
230      @count += 1
231    end
232
233    def expired?
234      debug "checking expired?"
235      if cc = self.response['cache-control'] && cc =~ /must-revalidate/
236        return true
237      end
238      return self.expires < Time.now
239    end
240
241    def setup_headers(hdr)
242      hdr['if-modified-since'] = self.date.rfc2822
243
244      debug "ims == #{hdr['if-modified-since']}"
245
246      if etag = self.response['etag']
247        hdr['if-none-match'] = etag
248        debug "etag: #{etag}"
249      end
250    end
251
252    def revalidate(resp = self.response)
253      @count = 0
254      self.use
255      self.date = resp.key?('date') ? Time.httpdate(resp['date']) : Time.now
256
257      cc = resp['cache-control']
258      if cc && (cc =~ /max-age=(\d+)/)
259        self.expires = self.date + $1.to_i
260      elsif resp.key?('expires')
261        self.expires = Time.httpdate(resp['expires'])
262      elsif lm = resp['last-modified']
263        delta = self.date - Time.httpdate(lm)
264        delta = 10 if delta <= 0
265        delta /= 5
266        self.expires = self.date + delta
267      else
268        self.expires = self.date + 300
269      end
270      # self.expires = Time.now + 10 # DEBUG
271      debug "expires on #{self.expires}"
272
273      return true
274    end
275
276    private
277    def initialize(resp)
278      @response = resp
279      begin
280        self.revalidate
281        self.response.raw_body
282      rescue Exception => e
283        error e
284        raise e
285      end
286    end
287  end
288
289  # Create the HttpUtil instance, associating it with Bot _bot_
290  #
291  def initialize(bot)
292    @bot = bot
293    @cache = Hash.new
294    @headers = {
295      'Accept-Charset' => 'utf-8;q=1.0, *;q=0.8',
296      'Accept-Encoding' => 'gzip;q=1, deflate;q=1, identity;q=0.8, *;q=0.2',
297      'User-Agent' =>
298        "rbot http util #{$version} (#{Irc::Bot::SOURCE_URL})"
299    }
300    debug "starting http cache cleanup timer"
301    @timer = @bot.timer.add(300) {
302      self.remove_stale_cache unless @bot.config['http.no_expire_cache']
303    }
304  end
305
306  # Clean up on HttpUtil unloading, by stopping the cache cleanup timer.
307  def cleanup
308    debug 'stopping http cache cleanup timer'
309    @bot.timer.remove(@timer)
310  end
311
312  # This method checks if a proxy is required to access _uri_, by looking at
313  # the values of config values +http.proxy_include+ and +http.proxy_exclude+.
314  #
315  # Each of these config values, if set, should be a Regexp the server name and
316  # IP address should be checked against.
317  #
318  def proxy_required(uri)
319    use_proxy = true
320    if @bot.config["http.proxy_exclude"].empty? && @bot.config["http.proxy_include"].empty?
321      return use_proxy
322    end
323
324    list = [uri.host]
325    begin
326      list.concat Resolv.getaddresses(uri.host)
327    rescue StandardError => err
328      warning "couldn't resolve host uri.host"
329    end
330
331    unless @bot.config["http.proxy_exclude"].empty?
332      re = @bot.config["http.proxy_exclude"].collect{|r| Regexp.new(r)}
333      re.each do |r|
334        list.each do |item|
335          if r.match(item)
336            use_proxy = false
337            break
338          end
339        end
340      end
341    end
342    unless @bot.config["http.proxy_include"].empty?
343      re = @bot.config["http.proxy_include"].collect{|r| Regexp.new(r)}
344      re.each do |r|
345        list.each do |item|
346          if r.match(item)
347            use_proxy = true
348            break
349          end
350        end
351      end
352    end
353    debug "using proxy for uri #{uri}?: #{use_proxy}"
354    return use_proxy
355  end
356
357  # _uri_:: URI to create a proxy for
358  #
359  # Return a net/http Proxy object, configured for proxying based on the
360  # bot's proxy configuration. See proxy_required for more details on this.
361  #
362  def get_proxy(uri, options = {})
363    opts = {
364      :read_timeout => @bot.config["http.read_timeout"],
365      :open_timeout => @bot.config["http.open_timeout"]
366    }.merge(options)
367
368    proxy = nil
369    proxy_host = nil
370    proxy_port = nil
371    proxy_user = nil
372    proxy_pass = nil
373
374    if @bot.config["http.use_proxy"]
375      if (ENV['http_proxy'])
376        proxy = URI.parse ENV['http_proxy'] rescue nil
377      end
378      if (@bot.config["http.proxy_uri"])
379        proxy = URI.parse @bot.config["http.proxy_uri"] rescue nil
380      end
381      if proxy
382        debug "proxy is set to #{proxy.host} port #{proxy.port}"
383        if proxy_required(uri)
384          proxy_host = proxy.host
385          proxy_port = proxy.port
386          proxy_user = @bot.config["http.proxy_user"]
387          proxy_pass = @bot.config["http.proxy_pass"]
388        end
389      end
390    end
391
392    h = Net::HTTP.new(uri.host, uri.port, proxy_host, proxy_port, proxy_user, proxy_pass)
393    h.use_ssl = true if uri.scheme == "https"
394
395    h.read_timeout = opts[:read_timeout]
396    h.open_timeout = opts[:open_timeout]
397    return h
398  end
399
400  # Internal method used to hanlde response _resp_ received when making a
401  # request for URI _uri_.
402  #
403  # It follows redirects, optionally yielding them if option :yield is :all.
404  #
405  # Also yields and returns the final _resp_.
406  #
407  def handle_response(uri, resp, opts, &block) # :yields: resp
408    if Net::HTTPRedirection === resp && opts[:max_redir] >= 0
409      if resp.key?('location')
410        raise 'Too many redirections' if opts[:max_redir] <= 0
411        yield resp if opts[:yield] == :all && block_given?
412        loc = resp['location']
413        new_loc = URI.join(uri.to_s, loc) rescue URI.parse(loc)
414        new_opts = opts.dup
415        new_opts[:max_redir] -= 1
416        case opts[:method].to_s.downcase.intern
417        when :post, :"net::http::post"
418          new_opts[:method] = :get
419        end
420        if resp['set-cookie']
421          debug "setting cookie #{resp['set-cookie']}"
422          new_opts[:headers] ||= Hash.new
423          new_opts[:headers]['Cookie'] = resp['set-cookie']
424        end
425        debug "following the redirect to #{new_loc}"
426        return get_response(new_loc, new_opts, &block)
427      else
428        warning ":| redirect w/o location?"
429      end
430    end
431    class << resp
432      undef_method :body
433      alias :body :cooked_body
434    end
435    unless resp['content-type']
436      debug "No content type, guessing"
437      resp['content-type'] =
438        case resp['x-rbot-location']
439        when /.html?$/i
440          'text/html'
441        when /.xml$/i
442          'application/xml'
443        when /.xhtml$/i
444          'application/xml+xhtml'
445        when /.(gif|png|jpe?g|jp2|tiff?)$/i
446          "image/#{$1.sub(/^jpg$/,'jpeg').sub(/^tif$/,'tiff')}"
447        else
448          'application/octetstream'
449        end
450    end
451    if block_given?
452      yield(resp)
453    else
454      # Net::HTTP wants us to read the whole body here
455      resp.raw_body
456    end
457    return resp
458  end
459
460  # _uri_::     uri to query (URI object or String)
461  #
462  # Generic http transaction method. It will return a Net::HTTPResponse
463  # object or raise an exception
464  #
465  # If a block is given, it will yield the response (see :yield option)
466  #
467  # Currently supported _options_:
468  #
469  # method::     request method [:get (default), :post or :head]
470  # open_timeout::     open timeout for the proxy
471  # read_timeout::     read timeout for the proxy
472  # cache::            should we cache results?
473  # yield::      if :final [default], calls the block for the response object;
474  #              if :all, call the block for all intermediate redirects, too
475  # max_redir::  how many redirects to follow before raising the exception
476  #              if -1, don't follow redirects, just return them
477  # range::      make a ranged request (usually GET). accepts a string
478  #              for HTTP/1.1 "Range:" header (i.e. "bytes=0-1000")
479  # body::       request body (usually for POST requests)
480  # headers::    additional headers to be set for the request. Its value must
481  #              be a Hash in the form { 'Header' => 'value' }
482  #
483  def get_response(uri_or_s, options = {}, &block) # :yields: resp
484    uri = uri_or_s.kind_of?(URI) ? uri_or_s : URI.parse(uri_or_s.to_s)
485    unless URI::HTTP === uri
486      if uri.scheme
487        raise "#{uri.scheme.inspect} URI scheme is not supported"
488      else
489        raise "don't know what to do with #{uri.to_s.inspect}"
490      end
491    end
492
493    opts = {
494      :max_redir => @bot.config['http.max_redir'],
495      :yield => :final,
496      :cache => true,
497      :method => :GET
498    }.merge(options)
499
500    resp = nil
501
502    req_class = case opts[:method].to_s.downcase.intern
503                when :head, :"net::http::head"
504                  opts[:max_redir] = -1
505                  Net::HTTP::Head
506                when :get, :"net::http::get"
507                  Net::HTTP::Get
508                when :post, :"net::http::post"
509                  opts[:cache] = false
510                  opts[:body] or raise 'post request w/o a body?'
511                  warning "refusing to cache POST request" if options[:cache]
512                  Net::HTTP::Post
513                else
514                  warning "unsupported method #{opts[:method]}, doing GET"
515                  Net::HTTP::Get
516                end
517
518    if req_class != Net::HTTP::Get && opts[:range]
519      warning "can't request ranges for #{req_class}"
520      opts.delete(:range)
521    end
522
523    cache_key = "#{opts[:range]}|#{req_class}|#{uri.to_s}"
524
525    if req_class != Net::HTTP::Get && req_class != Net::HTTP::Head
526      if opts[:cache]
527        warning "can't cache #{req_class.inspect} requests, working w/o cache"
528        opts[:cache] = false
529      end
530    end
531
532    debug "get_response(#{uri}, #{opts.inspect})"
533
534    cached = @cache[cache_key]
535
536    if opts[:cache] && cached
537      debug "got cached"
538      if !cached.expired?
539        debug "using cached"
540        cached.use
541        return handle_response(uri, cached.response, opts, &block)
542      end
543    end
544
545    headers = @headers.dup.merge(opts[:headers] || {})
546    headers['Range'] = opts[:range] if opts[:range]
547    headers['Authorization'] = opts[:auth_head] if opts[:auth_head]
548
549    if opts[:cache] && cached && (req_class == Net::HTTP::Get)
550      cached.setup_headers headers
551    end
552
553    req = req_class.new(uri.request_uri, headers)
554    if uri.user && uri.password
555      req.basic_auth(uri.user, uri.password)
556      opts[:auth_head] = req['Authorization']
557    end
558    req.body = opts[:body] if req_class == Net::HTTP::Post
559    debug "prepared request: #{req.to_hash.inspect}"
560
561    begin
562      get_proxy(uri, opts).start do |http|
563        http.request(req) do |resp|
564          resp['x-rbot-location'] = uri.to_s
565          if Net::HTTPNotModified === resp
566            debug "not modified"
567            begin
568              cached.revalidate(resp)
569            rescue Exception => e
570              error e
571            end
572            debug "reusing cached"
573            resp = cached.response
574          elsif Net::HTTPServerError === resp || Net::HTTPClientError === resp
575            debug "http error, deleting cached obj" if cached
576            @cache.delete(cache_key)
577          end
578
579          begin
580            return handle_response(uri, resp, opts, &block)
581          ensure
582            if cached = CachedObject.maybe_new(resp) rescue nil
583              debug "storing to cache"
584              @cache[cache_key] = cached
585            end
586          end
587        end
588      end
589    rescue Exception => e
590      error e
591      raise e.message
592    end
593  end
594
595  # _uri_::     uri to query (URI object or String)
596  #
597  # Simple GET request, returns (if possible) response body following redirs
598  # and caching if requested, yielding the actual response(s) to the optional
599  # block. See get_response for details on the supported _options_
600  #
601  def get(uri, options = {}, &block) # :yields: resp
602    begin
603      resp = get_response(uri, options, &block)
604      raise "http error: #{resp}" unless Net::HTTPOK === resp ||
605        Net::HTTPPartialContent === resp
606      return resp.body
607    rescue Exception => e
608      error e
609    end
610    return nil
611  end
612
613  # _uri_::     uri to query (URI object or String)
614  #
615  # Simple HEAD request, returns (if possible) response head following redirs
616  # and caching if requested, yielding the actual response(s) to the optional
617  # block. See get_response for details on the supported _options_
618  #
619  def head(uri, options = {}, &block) # :yields: resp
620    opts = {:method => :head}.merge(options)
621    begin
622      resp = get_response(uri, opts, &block)
623      # raise "http error #{resp}" if Net::HTTPClientError === resp ||
624      #   Net::HTTPServerError == resp
625      return resp
626    rescue Exception => e
627      error e
628    end
629    return nil
630  end
631
632  # _uri_::     uri to query (URI object or String)
633  # _data_::    body of the POST
634  #
635  # Simple POST request, returns (if possible) response following redirs and
636  # caching if requested, yielding the response(s) to the optional block. See
637  # get_response for details on the supported _options_
638  #
639  def post(uri, data, options = {}, &block) # :yields: resp
640    opts = {:method => :post, :body => data, :cache => false}.merge(options)
641    begin
642      resp = get_response(uri, opts, &block)
643      raise 'http error' unless Net::HTTPOK === resp or Net::HTTPCreated === resp
644      return resp
645    rescue Exception => e
646      error e
647    end
648    return nil
649  end
650
651  # _uri_::     uri to query (URI object or String)
652  # _nbytes_::  number of bytes to get
653  #
654  # Partial GET request, returns (if possible) the first _nbytes_ bytes of the
655  # response body, following redirs and caching if requested, yielding the
656  # actual response(s) to the optional block. See get_response for details on
657  # the supported _options_
658  #
659  def get_partial(uri, nbytes = @bot.config['http.info_bytes'], options = {}, &block) # :yields: resp
660    opts = {:range => "bytes=0-#{nbytes}"}.merge(options)
661    return get(uri, opts, &block)
662  end
663
664  def remove_stale_cache
665    debug "Removing stale cache"
666    now = Time.new
667    max_last = @bot.config['http.expire_time'] * 60
668    max_first = @bot.config['http.max_cache_time'] * 60
669    debug "#{@cache.size} pages before"
670    begin
671      @cache.reject! { |k, val|
672        (now - val.last_used > max_last) || (now - val.first_used > max_first)
673      }
674    rescue => e
675      error "Failed to remove stale cache: #{e.pretty_inspect}"
676    end
677    debug "#{@cache.size} pages after"
678  end
679
680end
681end
682end
683
684class HttpUtilPlugin < CoreBotModule
685  def initialize(*a)
686    super(*a)
687    debug 'initializing httputil'
688    @bot.httputil = Irc::Utils::HttpUtil.new(@bot)
689  end
690
691  def cleanup
692    debug 'shutting down httputil'
693    @bot.httputil.cleanup
694    @bot.httputil = nil
695    super
696  end
697end
698
699HttpUtilPlugin.new
Note: See TracBrowser for help on using the browser.