arachni 0.4.5.2 → 0.4.6

data/lib/arachni/http/cookie_jar.rb

@@ -1,5 +1,5 @@
 =begin
-    Copyright 2010-2013 Tasos Laskos <tasos.laskos@gmail.com>
+    Copyright 2010-2014 Tasos Laskos <tasos.laskos@gmail.com>
 
     Licensed under the Apache License, Version 2.0 (the "License");
     you may not use this file except in compliance with the License.
@@ -100,9 +100,15 @@ class CookieJar
         [cookies].flatten.compact.each do |c|
             self << case c
                         when String
-                            Cookie.from_string( ::Arachni::Options.url.to_s, c )
+                            begin
+                                Cookie.from_string( ::Arachni::Options.url.to_s, c )
+                            rescue
+                                Cookie.from_set_cookie( ::Arachni::Options.url.to_s, c )
+                            end
+
                         when Hash
                             Cookie.new( ::Arachni::Options.url.to_s, c ) if c.any?
+
                         when Cookie
                             c
                     end

data/lib/arachni/issue.rb

@@ -1,5 +1,5 @@
 =begin
-    Copyright 2010-2013 Tasos Laskos <tasos.laskos@gmail.com>
+    Copyright 2010-2014 Tasos Laskos <tasos.laskos@gmail.com>
 
     Licensed under the Apache License, Version 2.0 (the "License");
     you may not use this file except in compliance with the License.
@@ -146,6 +146,9 @@ class Issue
     #   method merged with a `Hash` holding {Issue} attributes.
     #
     def initialize( opts = {} )
+        # Make sure we're dealing with UTF-8 data.
+        opts = opts.recode
+
         @verification = false
         @references   = {}
         @opts         = { regexp: '' }

data/lib/arachni/mixins/observable.rb

@@ -1,5 +1,5 @@
 =begin
-    Copyright 2010-2013 Tasos Laskos <tasos.laskos@gmail.com>
+    Copyright 2010-2014 Tasos Laskos <tasos.laskos@gmail.com>
 
     Licensed under the Apache License, Version 2.0 (the "License");
     you may not use this file except in compliance with the License.

data/lib/arachni/mixins/progress_bar.rb

@@ -1,5 +1,5 @@
 =begin
-    Copyright 2010-2013 Tasos Laskos <tasos.laskos@gmail.com>
+    Copyright 2010-2014 Tasos Laskos <tasos.laskos@gmail.com>
 
     Licensed under the Apache License, Version 2.0 (the "License");
     you may not use this file except in compliance with the License.

data/lib/arachni/mixins/terminal.rb

@@ -1,5 +1,5 @@
 =begin
-    Copyright 2010-2013 Tasos Laskos <tasos.laskos@gmail.com>
+    Copyright 2010-2014 Tasos Laskos <tasos.laskos@gmail.com>
 
     Licensed under the Apache License, Version 2.0 (the "License");
     you may not use this file except in compliance with the License.

data/lib/arachni/module.rb

@@ -1,5 +1,5 @@
 =begin
-    Copyright 2010-2013 Tasos Laskos <tasos.laskos@gmail.com>
+    Copyright 2010-2014 Tasos Laskos <tasos.laskos@gmail.com>
 
     Licensed under the Apache License, Version 2.0 (the "License");
     you may not use this file except in compliance with the License.

data/lib/arachni/module/auditor.rb

@@ -1,5 +1,5 @@
 =begin
-    Copyright 2010-2013 Tasos Laskos <tasos.laskos@gmail.com>
+    Copyright 2010-2014 Tasos Laskos <tasos.laskos@gmail.com>
 
     Licensed under the Apache License, Version 2.0 (the "License");
     you may not use this file except in compliance with the License.
@@ -142,7 +142,7 @@ module Auditor
         # Elements to audit.
         #
         # If no elements have been passed to audit methods, candidates will be
-        # determined by {#candidate_elements}.
+        # determined by {#each_candidate_element}.
         #
         elements: [Element::LINK, Element::FORM,
                    Element::COOKIE, Element::HEADER,
@@ -413,18 +413,18 @@ module Auditor
         element = opts[:element] || opts[:elem]
 
         msg = "In #{element}"
-        msg << " var '#{var}'" if var
+        msg << " input '#{var}'" if var
         print_ok "#{msg} ( #{url} )"
 
         print_verbose( "Injected string:\t" + opts[:injected] ) if opts[:injected]
         print_verbose( "Verified string:\t" + opts[:match].to_s ) if opts[:match]
-        print_verbose( "Matched regular expression: " + opts[:regexp].to_s ) if opts[:regexp]
+        print_verbose( 'Matched regular expression: ' + opts[:regexp].to_s ) if opts[:regexp]
         print_debug( 'Request ID: ' + res.request.id.to_s ) if res
         print_verbose( '---------' ) if only_positives?
 
         # Platform identification by vulnerability.
         platform_type = nil
-        if platform = opts[:platform]
+        if (platform = opts[:platform])
             Platform::Manager[url] << platform if Options.fingerprint?
             platform_type = Platform::Manager[url].find_type( platform )
         end
@@ -484,6 +484,7 @@ module Auditor
         false
     end
 
+    # Passes each element prepared for audit to the block.
     #
     # If no element types have been specified in `opts`, it will use the elements
     # from the module's {Base.info} hash.
@@ -495,9 +496,9 @@ module Auditor
     # @option opts  [Array]  :elements
     #   Element types to audit (see {OPTIONS}`[:elements]`).
     #
-    # @return   [Array<Arachni::Element>]   Prepared elements.
-    #
-    def candidate_elements( opts = {} )
+    # @yield       [element]  Each candidate element.
+    # @yieldparam [Arachni::Element]
+    def each_candidate_element( opts = {} )
         if !opts.include?( :elements) || !opts[:elements] || opts[:elements].empty?
             opts[:elements] = self.class.info[:elements]
         end
@@ -529,14 +530,19 @@ module Auditor
             end
         end
 
-        elements.map { |e| d = e.dup; d.auditor = self; d }
+        while (e = elements.pop)
+            next if e.auditable.empty?
+            d = e.dup
+            d.auditor = self
+            yield d
+        end
     end
 
     #
     # If a block has been provided it calls {Arachni::Element::Capabilities::Auditable#audit}
     # for every element, otherwise, it defaults to {#audit_taint}.
     #
-    # Uses {#candidate_elements} to decide which elements to audit.
+    # Uses {#each_candidate_element} to decide which elements to audit.
     #
     # @see OPTIONS
     # @see Arachni::Element::Capabilities::Auditable#audit
@@ -547,7 +553,7 @@ module Auditor
         if !block_given?
             audit_taint( payloads, opts )
         else
-            candidate_elements( opts ).each { |e| e.audit( payloads, opts, &block ) }
+            each_candidate_element( opts ) { |e| e.audit( payloads, opts, &block ) }
         end
     end
 
@@ -555,40 +561,40 @@ module Auditor
     # Provides easy access to element auditing using simple taint analysis
     # and automatically logs results.
     #
-    # Uses {#candidate_elements} to decide which elements to audit.
+    # Uses {#each_candidate_element} to decide which elements to audit.
     #
     # @see OPTIONS
     # @see Arachni::Element::Capabilities::Auditable::Taint
     #
     def audit_taint( payloads, opts = {} )
         opts = OPTIONS.merge( opts )
-        candidate_elements( opts ).each { |e| e.taint_analysis( payloads, opts ) }
+        each_candidate_element( opts ) { |e| e.taint_analysis( payloads, opts ) }
     end
 
     #
     # Audits elements using differential analysis and automatically logs results.
     #
-    # Uses {#candidate_elements} to decide which elements to audit.
+    # Uses {#each_candidate_element} to decide which elements to audit.
     #
     # @see OPTIONS
     # @see Arachni::Element::Capabilities::Auditable::RDiff
     #
     def audit_rdiff( opts = {}, &block )
         opts = OPTIONS.merge( opts )
-        candidate_elements( opts ).each { |e| e.rdiff_analysis( opts, &block ) }
+        each_candidate_element( opts ) { |e| e.rdiff_analysis( opts, &block ) }
     end
 
     #
     # Audits elements using timing attacks and automatically logs results.
     #
-    # Uses {#candidate_elements} to decide which elements to audit.
+    # Uses {#each_candidate_element} to decide which elements to audit.
     #
     # @see OPTIONS
     # @see Arachni::Element::Capabilities::Auditable::Timeout
     #
     def audit_timeout( payloads, opts = {} )
         opts = OPTIONS.merge( opts )
-        candidate_elements( opts ).each { |e| e.timeout_analysis( payloads, opts ) }
+        each_candidate_element( opts ) { |e| e.timeout_analysis( payloads, opts ) }
     end
 
     private

data/lib/arachni/module/base.rb

@@ -1,5 +1,5 @@
 =begin
-    Copyright 2010-2013 Tasos Laskos <tasos.laskos@gmail.com>
+    Copyright 2010-2014 Tasos Laskos <tasos.laskos@gmail.com>
 
     Licensed under the Apache License, Version 2.0 (the "License");
     you may not use this file except in compliance with the License.

data/lib/arachni/module/manager.rb

@@ -1,5 +1,5 @@
 =begin
-    Copyright 2010-2013 Tasos Laskos <tasos.laskos@gmail.com>
+    Copyright 2010-2014 Tasos Laskos <tasos.laskos@gmail.com>
 
     Licensed under the Apache License, Version 2.0 (the "License");
     you may not use this file except in compliance with the License.
@@ -173,17 +173,17 @@ class Manager < Arachni::Component::Manager
         on_register_results_blocks_raw.each { |block| block.call( results ) }
 
         unique = dedup( results )
-        return 0 if unique.empty?
+        return results if unique.empty?
 
         # Don't allow multiple variations of the same audit-type issue,
         # only allow variations for recon modules.
         unique.each { |issue| issue_set << issue.unique_id if issue.audit? }
 
         on_register_results_blocks.each { |block| block.call( unique ) }
-        return 0 if !store?
+        return results if !store?
 
         unique.each { |issue| self.results << issue }
-        unique.size
+        results
     end
     def register_results( results )
         self.class.register_results( results )

data/lib/arachni/module/output.rb

@@ -1,5 +1,5 @@
 =begin
-    Copyright 2010-2013 Tasos Laskos <tasos.laskos@gmail.com>
+    Copyright 2010-2014 Tasos Laskos <tasos.laskos@gmail.com>
 
     Licensed under the Apache License, Version 2.0 (the "License");
     you may not use this file except in compliance with the License.

data/lib/arachni/module/utilities.rb

@@ -1,5 +1,5 @@
 =begin
-    Copyright 2010-2013 Tasos Laskos <tasos.laskos@gmail.com>
+    Copyright 2010-2014 Tasos Laskos <tasos.laskos@gmail.com>
 
     Licensed under the Apache License, Version 2.0 (the "License");
     you may not use this file except in compliance with the License.

data/lib/arachni/options.rb

@@ -1,5 +1,5 @@
 =begin
-    Copyright 2010-2013 Tasos Laskos <tasos.laskos@gmail.com>
+    Copyright 2010-2014 Tasos Laskos <tasos.laskos@gmail.com>
 
     Licensed under the Apache License, Version 2.0 (the "License");
     you may not use this file except in compliance with the License.
@@ -210,6 +210,14 @@ class Options
     #
     attr_accessor :http_req_limit
 
+    # Maximum amount of requests to keep in the queue.
+    #
+    # Bigger size means better scheduling and bette performance, smaller means
+    # less RAM consumption.
+    #
+    # @return    [Integer]
+    attr_accessor :http_queue_size
+
     #
     # Should Arachni audit links?
     #
@@ -417,12 +425,15 @@ class Options
     # @return   [String]   Path to the UNIX socket to use.
     attr_accessor :rpc_socket
 
-    # @return   [Integer]   port for the RPC server to listen to
+    # @return   [Integer]   port for the RPC server to listen to.
     attr_accessor :rpc_port
 
-    # @return   [String]   (hostname or IP) address for the RPC server to bind to
+    # @return   [String]   Hostname or IP address for the RPC server to bind to.
     attr_accessor :rpc_address
 
+    # @return   [String]   External (hostname or IP) address for the RPC server to bind to.
+    attr_accessor :rpc_external_address
+
     # @return   [Array<Integer>]
     #   Range of ports to use when spawning instances,
     #   first element should be the lowest port number, last the max port number.
@@ -581,9 +592,10 @@ class Options
         @lsmod  = []
         @lsrep  = []
 
-        @http_req_limit = 20
-        @http_username = nil
-        @http_password = nil
+        @http_req_limit  = 20
+        @http_queue_size = 500
+        @http_username   = nil
+        @http_password   = nil
 
         @mods = []
 
@@ -974,6 +986,7 @@ class Options
             [ '--exclude-vector',          GetoptLong::REQUIRED_ARGUMENT ],
             [ '--include',           '-i', GetoptLong::REQUIRED_ARGUMENT ],
             [ '--http-req-limit',          GetoptLong::REQUIRED_ARGUMENT ],
+            [ '--http-queue-size',         GetoptLong::REQUIRED_ARGUMENT ],
             [ '--http-timeout',            GetoptLong::REQUIRED_ARGUMENT ],
             [ '--follow-subdomains', '-f', GetoptLong::NO_ARGUMENT ],
             [ '--debug',             '-w', GetoptLong::NO_ARGUMENT ],
@@ -988,6 +1001,7 @@ class Options
             [ '--node-ssl-cert',          GetoptLong::REQUIRED_ARGUMENT ],
             [ '--ssl-ca',                 GetoptLong::REQUIRED_ARGUMENT ],
             [ '--address',                GetoptLong::REQUIRED_ARGUMENT ],
+            [ '--external-address',       GetoptLong::REQUIRED_ARGUMENT ],
             [ '--reroute-to-logfile',     GetoptLong::NO_ARGUMENT ],
             [ '--pool-size',              GetoptLong::REQUIRED_ARGUMENT ],
             [ '--neighbour',              GetoptLong::REQUIRED_ARGUMENT ],
@@ -1114,6 +1128,9 @@ class Options
                     when '--http-req-limit'
                         @http_req_limit = arg.to_i
 
+                    when '--http-queue-size'
+                        @http_queue_size = arg.to_i
+
                     when '--http-timeout'
                         @http_timeout = arg.to_i
 
@@ -1234,6 +1251,9 @@ class Options
                     when '--address'
                         @rpc_address = arg.to_s
 
+                    when '--external-address'
+                        @rpc_external_address = arg.to_s
+
                     when '--pool-size'
                         @pool_size = arg.to_i
 
@@ -1401,6 +1421,7 @@ class Options
     #
     def load( filepath )
         opts = YAML::load( IO.read( filepath ) )
+        opts = self.deep_clone.merge!( opts ) if opts.is_a? Hash
 
         if opts.restrict_paths_filepath
             opts.restrict_paths = paths_from_file( opts.restrict_paths_filepath )
@@ -1450,7 +1471,7 @@ class Options
         options.to_hash.each_pair do |k, v|
             next if !v
             next if ( v.is_a?( Array ) || v.is_a?( Hash ) ) && v.empty?
-            send( "#{k.to_s}=", v )
+            send( "#{k.to_s}=", v ) rescue NoMethodError
         end
         self
     end

data/lib/arachni/page.rb

@@ -1,5 +1,5 @@
 =begin
-    Copyright 2010-2013 Tasos Laskos <tasos.laskos@gmail.com>
+    Copyright 2010-2014 Tasos Laskos <tasos.laskos@gmail.com>
 
     Licensed under the Apache License, Version 2.0 (the "License");
     you may not use this file except in compliance with the License.
@@ -146,8 +146,8 @@ class Page
     end
 
     def marshal_dump
-        @document = nil
         instance_variables.inject( {} ) do |h, iv|
+            next h if iv == :@document
             h[iv] = instance_variable_get( iv )
             h
         end
@@ -171,9 +171,8 @@ class Page
     # @return   [Hash]  Converts the page data to a hash.
     def to_h
         instance_variables.reduce({}) do |h, iv|
-            if iv != :@document
-                h[iv.to_s.gsub( '@', '').to_sym] = try_dup( instance_variable_get( iv ) )
-            end
+            next h if iv == :@document
+            h[iv.to_s.gsub( '@', '').to_sym] = try_dup( instance_variable_get( iv ) )
             h
         end
     end

data/lib/arachni/parser.rb

@@ -1,5 +1,5 @@
 =begin
-    Copyright 2010-2013 Tasos Laskos <tasos.laskos@gmail.com>
+    Copyright 2010-2014 Tasos Laskos <tasos.laskos@gmail.com>
 
     Licensed under the Apache License, Version 2.0 (the "License");
     you may not use this file except in compliance with the License.
@@ -134,7 +134,8 @@ class Parser
                 body:             @html,
                 request_headers:  @response.request ? @response.request.headers : {},
                 response_headers: @response_headers,
-                text:             false
+                text:             false,
+                links:            [self_link]
             )
             Platform::Manager.fingerprint( page ) if Options.fingerprint?
             return page

data/lib/arachni/platform.rb

@@ -1,5 +1,5 @@
 =begin
-    Copyright 2010-2013 Tasos Laskos <tasos.laskos@gmail.com>
+    Copyright 2010-2014 Tasos Laskos <tasos.laskos@gmail.com>
 
     Licensed under the Apache License, Version 2.0 (the "License");
     you may not use this file except in compliance with the License.

data/lib/arachni/platform/fingerprinter.rb

@@ -1,5 +1,5 @@
 =begin
-    Copyright 2010-2013 Tasos Laskos <tasos.laskos@gmail.com>
+    Copyright 2010-2014 Tasos Laskos <tasos.laskos@gmail.com>
 
     Licensed under the Apache License, Version 2.0 (the "License");
     you may not use this file except in compliance with the License.

data/lib/arachni/platform/list.rb

@@ -1,5 +1,5 @@
 =begin
-    Copyright 2010-2013 Tasos Laskos <tasos.laskos@gmail.com>
+    Copyright 2010-2014 Tasos Laskos <tasos.laskos@gmail.com>
 
     Licensed under the Apache License, Version 2.0 (the "License");
     you may not use this file except in compliance with the License.
@@ -28,12 +28,12 @@ class List
     #   Valid platforms for this list.
     def initialize( valid_platforms )
         @valid_platforms = normalize!( valid_platforms )
-        @platforms       = Set.new
+        @platforms       = []
     end
 
     # @return   [Array<Symbol>] Supported platforms.
     def valid
-        @valid ||= Set.new( hierarchical? ? flat_list : @valid_platforms )
+        hierarchical? ? @valid_platforms.find_symbol_keys_recursively : @valid_platforms
     end
 
     # Selects appropriate data depending on the applicable platforms
@@ -126,39 +126,31 @@ class List
     # @return   [Platform] `self`
     # @raise    [Error::Invalid]  On {#invalid?} platforms.
     def <<( platform )
-        @platforms << normalize( platform )
+        @platforms |= [normalize( platform )]
         self
     end
 
     # @param    [Platform, Enumerable] enum
     #   Enumerable object containing platforms.
-    # @return   [Platform] Updated copy of `self`.
+    #   New {Platform} built by merging `self` and the elements of the
+    #   given enumerable object.
     # @raise    [Error::Invalid]  On {#invalid?} platforms.
     def merge( enum )
         dup.merge!( enum )
     end
+    alias + merge
+    alias | merge
 
     # @param    [Enumerable] enum
     #   Enumerable object containing platforms.
     # @return   [Platform] Updated `self`.
     # @raise    [Error::Invalid]  On {#invalid?} platforms.
     def merge!( enum )
-        @platforms.merge normalize( enum )
+        @platforms |= normalize( enum )
         self
     end
     alias update merge!
 
-    # @param    [Platform, Enumerable] enum
-    #   {Platform} or enumerable object containing platforms.
-    # @return   [Platform]
-    #   New {Platform} built by merging `self` and the elements of the
-    #   given enumerable object.
-    # @raise    [Error::Invalid]  On {#invalid?} platforms.
-    def |( enum )
-        dup.merge( enum )
-    end
-    alias + |
-
     # @param    [Block] block   Block to be passed each platform.
     # @return   [Enumerator, Platform]
     #   `Enumerator` if no `block` is given, `self` otherwise.
@@ -217,28 +209,18 @@ class List
     protected
 
     def platforms=( enum )
-        @platforms = Set.new( enum )
+        @platforms = enum.to_a
     end
 
     private
 
-    # @return   [Array<Symbol>] Flat list of supported {OS operating systems}.
-    def flat_list( hash = @valid_platforms )
-        flat = []
-        hash.each do |k, v|
-            flat << k
-            flat |= flat_list( v ) if v.any?
-        end
-        flat.reject { |i| !i.is_a? Symbol }
-    end
-
     def find_children( platform, hash = @valid_platforms )
         return [] if hash.empty?
 
         children = []
         hash.each do |k, v|
             if k == platform
-                children |= flat_list( v )
+                children |= v.find_symbol_keys_recursively
             elsif v.is_a? Hash
                 children |= find_children( platform, v )
             end