rack 2.0.9 → 2.1.0

data/lib/rack/multipart.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'rack/multipart/parser'
 
 module Rack
@@ -14,10 +16,10 @@ module Rack
     TOKEN = /[^\s()<>,;:\\"\/\[\]?=]+/
     CONDISP = /Content-Disposition:\s*#{TOKEN}\s*/i
     VALUE = /"(?:\\"|[^"])*"|#{TOKEN}/
-    BROKEN_QUOTED = /^#{CONDISP}.*;\sfilename="(.*?)"(?:\s*$|\s*;\s*#{TOKEN}=)/i
-    BROKEN_UNQUOTED = /^#{CONDISP}.*;\sfilename=(#{TOKEN})/i
+    BROKEN_QUOTED = /^#{CONDISP}.*;\s*filename="(.*?)"(?:\s*$|\s*;\s*#{TOKEN}=)/i
+    BROKEN_UNQUOTED = /^#{CONDISP}.*;\s*filename=(#{TOKEN})/i
     MULTIPART_CONTENT_TYPE = /Content-Type: (.*)#{EOL}/ni
-    MULTIPART_CONTENT_DISPOSITION = /Content-Disposition:.*\s+name=(#{VALUE})/ni
+    MULTIPART_CONTENT_DISPOSITION = /Content-Disposition:.*;\s*name=(#{VALUE})/ni
     MULTIPART_CONTENT_ID = /Content-ID:\s*([^#{EOL}]*)/ni
     # Updated definitions from RFC 2231
     ATTRIBUTE_CHAR = %r{[^ \t\v\n\r)(><@,;:\\"/\[\]?='*%]}

data/lib/rack/multipart/generator.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Rack
   module Multipart
     class Generator
@@ -27,21 +29,18 @@ module Rack
 
       private
       def multipart?
-        multipart = false
-
         query = lambda { |value|
           case value
           when Array
-            value.each(&query)
+            value.any?(&query)
           when Hash
-            value.values.each(&query)
+            value.values.any?(&query)
           when Rack::Multipart::UploadedFile
-            multipart = true
+            true
           end
         }
-        @params.values.each(&query)
 
-        multipart
+        @params.values.any?(&query)
       end
 
       def flattened_params

data/lib/rack/multipart/parser.rb

@@ -1,16 +1,24 @@
+# frozen_string_literal: true
+
 require 'rack/utils'
+require 'strscan'
+require 'rack/core_ext/regexp'
 
 module Rack
   module Multipart
     class MultipartPartLimitError < Errno::EMFILE; end
 
     class Parser
-      BUFSIZE = 16384
+      using ::Rack::RegexpExtensions
+
+      BUFSIZE = 1_048_576
       TEXT_PLAIN = "text/plain"
       TEMPFILE_FACTORY = lambda { |filename, content_type|
-        Tempfile.new(["RackMultipart", ::File.extname(filename.gsub("\0".freeze, '%00'.freeze))])
+        Tempfile.new(["RackMultipart", ::File.extname(filename.gsub("\0", '%00'))])
       }
 
+      BOUNDARY_REGEX = /\A([^\n]*(?:\n|\Z))/
+
       class BoundedIO # :nodoc:
         def initialize(io, content_length)
           @io             = io
@@ -18,15 +26,15 @@ module Rack
           @cursor = 0
         end
 
-        def read(size)
+        def read(size, outbuf = nil)
           return if @cursor >= @content_length
 
           left = @content_length - @cursor
 
           str = if left < size
-                  @io.read left
+                  @io.read left, outbuf
                 else
-                  @io.read size
+                  @io.read size, outbuf
                 end
 
           if str
@@ -61,13 +69,14 @@ module Rack
         return EMPTY unless boundary
 
         io = BoundedIO.new(io, content_length) if content_length
+        outbuf = String.new
 
         parser = new(boundary, tmpfile, bufsize, qp)
-        parser.on_read io.read(bufsize)
+        parser.on_read io.read(bufsize, outbuf)
 
         loop do
           break if parser.state == :DONE
-          parser.on_read io.read(bufsize)
+          parser.on_read io.read(bufsize, outbuf)
         end
 
         io.rewind
@@ -90,14 +99,14 @@ module Rack
               # those which give the lone filename.
               fn = filename.split(/[\/\\]/).last
 
-              data = {:filename => fn, :type => content_type,
-                      :name => name, :tempfile => body, :head => head}
+              data = { filename: fn, type: content_type,
+                      name: name, tempfile: body, head: head }
             elsif !filename && content_type && body.is_a?(IO)
               body.rewind
 
               # Generic multipart cases, not coming from a form
-              data = {:type => content_type,
-                      :name => name, :tempfile => body, :head => head}
+              data = { type: content_type,
+                      name: name, tempfile: body, head: head }
             end
 
             yield data
@@ -138,6 +147,7 @@ module Rack
           end
 
           @mime_parts[mime_index] = klass.new(body, head, filename, content_type, name)
+
           check_open_files
         end
 
@@ -163,25 +173,26 @@ module Rack
       attr_reader :state
 
       def initialize(boundary, tempfile, bufsize, query_parser)
-        @buf            = String.new
-
         @query_parser   = query_parser
         @params         = query_parser.make_params
         @boundary       = "--#{boundary}"
         @bufsize        = bufsize
 
-        @rx = /(?:#{EOL})?#{Regexp.quote(@boundary)}(#{EOL}|--)/n
-        @rx_max_size = EOL.size + @boundary.bytesize + [EOL.size, '--'.size].max
         @full_boundary = @boundary
         @end_boundary = @boundary + '--'
         @state = :FAST_FORWARD
         @mime_index = 0
         @collector = Collector.new tempfile
+
+        @sbuf = StringScanner.new("".dup)
+        @body_regex = /(.*?)(#{EOL})?#{Regexp.quote(@boundary)}(#{EOL}|--)/m
+        @rx_max_size = EOL.size + @boundary.bytesize + [EOL.size, '--'.size].max
+        @head_regex = /(.*?#{EOL})#{EOL}/m
       end
 
       def on_read content
         handle_empty_content!(content)
-        @buf << content
+        @sbuf.concat content
         run_parser
       end
 
@@ -192,7 +203,6 @@ module Rack
             @query_parser.normalize_params(@params, part.name, data, @query_parser.param_depth_limit)
           end
         end
-
         MultipartInfo.new @params.to_params_hash, @collector.find_all(&:file?).map(&:body)
       end
 
@@ -219,7 +229,7 @@ module Rack
         if consume_boundary
           @state = :MIME_HEAD
         else
-          raise EOFError, "bad content body" if @buf.bytesize >= @bufsize
+          raise EOFError, "bad content body" if @sbuf.rest_size >= @bufsize
           :want_read
         end
       end
@@ -227,19 +237,16 @@ module Rack
       def handle_consume_token
         tok = consume_boundary
         # break if we're at the end of a buffer, but not if it is the end of a field
-        if tok == :END_BOUNDARY || (@buf.empty? && tok != :BOUNDARY)
-          @state = :DONE
+        @state = if tok == :END_BOUNDARY || (@sbuf.eos? && tok != :BOUNDARY)
+          :DONE
         else
-          @state = :MIME_HEAD
+          :MIME_HEAD
         end
       end
 
       def handle_mime_head
-        if @buf.index(EOL + EOL)
-          i = @buf.index(EOL+EOL)
-          head = @buf.slice!(0, i+2) # First \r\n
-          @buf.slice!(0, 2)          # Second \r\n
-
+        if @sbuf.scan_until(@head_regex)
+          head = @sbuf[1]
           content_type = head[MULTIPART_CONTENT_TYPE, 1]
           if name = head[MULTIPART_CONTENT_DISPOSITION, 1]
             name = Rack::Auth::Digest::Params::dequote(name)
@@ -250,7 +257,7 @@ module Rack
           filename = get_filename(head)
 
           if name.nil? || name.empty?
-            name = filename || "#{content_type || TEXT_PLAIN}[]"
+            name = filename || "#{content_type || TEXT_PLAIN}[]".dup
           end
 
           @collector.on_mime_head @mime_index, head, filename, content_type, name
@@ -261,16 +268,19 @@ module Rack
       end
 
       def handle_mime_body
-        if i = @buf.index(rx)
-          # Save the rest.
-          @collector.on_mime_body @mime_index, @buf.slice!(0, i)
-          @buf.slice!(0, 2) # Remove \r\n after the content
+        if @sbuf.check_until(@body_regex) # check but do not advance the pointer yet
+          body = @sbuf[1]
+          @collector.on_mime_body @mime_index, body
+          @sbuf.pos += body.length + 2 # skip \r\n after the content
           @state = :CONSUME_TOKEN
           @mime_index += 1
         else
-          # Save the read body part.
-          if @rx_max_size < @buf.size
-            @collector.on_mime_body @mime_index, @buf.slice!(0, @buf.size - @rx_max_size)
+          # Save what we have so far
+          if @rx_max_size < @sbuf.rest_size
+            delta = @sbuf.rest_size - @rx_max_size
+            @collector.on_mime_body @mime_index, @sbuf.peek(delta)
+            @sbuf.pos += delta
+            @sbuf.string = @sbuf.rest
           end
           :want_read
         end
@@ -278,16 +288,13 @@ module Rack
 
       def full_boundary; @full_boundary; end
 
-      def rx; @rx; end
-
       def consume_boundary
-        while @buf.gsub!(/\A([^\n]*(?:\n|\Z))/, '')
-          read_buffer = $1
+        while read_buffer = @sbuf.scan_until(BOUNDARY_REGEX)
           case read_buffer.strip
           when full_boundary then return :BOUNDARY
           when @end_boundary then return :END_BOUNDARY
           end
-          return if @buf.empty?
+          return if @sbuf.eos?
         end
       end
 
@@ -308,8 +315,8 @@ module Rack
 
         return unless filename
 
-        if filename.scan(/%.?.?/).all? { |s| s =~ /%[0-9a-fA-F]{2}/ }
-          filename = Utils.unescape(filename)
+        if filename.scan(/%.?.?/).all? { |s| /%[0-9a-fA-F]{2}/.match?(s) }
+          filename = Utils.unescape_path(filename)
         end
 
         filename.scrub!
@@ -325,7 +332,7 @@ module Rack
         filename
       end
 
-      CHARSET   = "charset"
+      CHARSET = "charset"
 
       def tag_multipart_encoding(filename, content_type, name, body)
         name = name.to_s
@@ -342,10 +349,10 @@ module Rack
           if TEXT_PLAIN == type_subtype
             rest         = list.drop 1
             rest.each do |param|
-              k,v = param.split('=', 2)
+              k, v = param.split('=', 2)
               k.strip!
               v.strip!
-              v = v[1..-2] if v[0] == '"' && v[-1] == '"'
+              v = v[1..-2] if v.start_with?('"') && v.end_with?('"')
               encoding = Encoding.find v if k == CHARSET
             end
           end
@@ -355,7 +362,6 @@ module Rack
         body.force_encoding(encoding)
       end
 
-
       def handle_empty_content!(content)
         if content.nil? || content.empty?
           raise EOFError

data/lib/rack/multipart/uploaded_file.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Rack
   module Multipart
     class UploadedFile

data/lib/rack/null_logger.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Rack
   class NullLogger
     def initialize(app)

data/lib/rack/query_parser.rb

@@ -1,5 +1,11 @@
+# frozen_string_literal: true
+
+require_relative 'core_ext/regexp'
+
 module Rack
   class QueryParser
+    using ::Rack::RegexpExtensions
+
     DEFAULT_SEP = /[&;] */n
     COMMON_SEP = { ";" => /[;] */n, ";," => /[;,] */n, "&" => /[&] */n }
 
@@ -36,7 +42,7 @@ module Rack
 
       (qs || '').split(d ? (COMMON_SEP[d] || /[#{d}] */n) : DEFAULT_SEP).each do |p|
         next if p.empty?
-        k, v = p.split('='.freeze, 2).map!(&unescaper)
+        k, v = p.split('=', 2).map!(&unescaper)
 
         if cur = params[k]
           if cur.class == Array
@@ -49,7 +55,7 @@ module Rack
         end
       end
 
-      return params.to_params_hash
+      return params.to_h
     end
 
     # parse_nested_query expands a query string into structural types. Supported
@@ -61,13 +67,13 @@ module Rack
       return {} if qs.nil? || qs.empty?
       params = make_params
 
-      (qs || '').split(d ? (COMMON_SEP[d] || /[#{d}] */n) : DEFAULT_SEP).each do |p|
-        k, v = p.split('='.freeze, 2).map! { |s| unescape(s) }
+      qs.split(d ? (COMMON_SEP[d] || /[#{d}] */n) : DEFAULT_SEP).each do |p|
+        k, v = p.split('=', 2).map! { |s| unescape(s) }
 
         normalize_params(params, k, v, param_depth_limit)
       end
 
-      return params.to_params_hash
+      return params.to_h
     rescue ArgumentError => e
       raise InvalidParameterError, e.message
     end
@@ -79,22 +85,22 @@ module Rack
       raise RangeError if depth <= 0
 
       name =~ %r(\A[\[\]]*([^\[\]]+)\]*)
-      k = $1 || ''.freeze
-      after = $' || ''.freeze
+      k = $1 || ''
+      after = $' || ''
 
       if k.empty?
-        if !v.nil? && name == "[]".freeze
+        if !v.nil? && name == "[]"
           return Array(v)
         else
           return
         end
       end
 
-      if after == ''.freeze
+      if after == ''
         params[k] = v
-      elsif after == "[".freeze
+      elsif after == "["
         params[name] = v
-      elsif after == "[]".freeze
+      elsif after == "[]"
         params[k] ||= []
         raise ParameterTypeError, "expected Array (got #{params[k].class.name}) for param `#{k}'" unless params[k].is_a?(Array)
         params[k] << v
@@ -135,7 +141,7 @@ module Rack
     end
 
     def params_hash_has_key?(hash, key)
-      return false if key =~ /\[\]/
+      return false if /\[\]/.match?(key)
 
       key.split(/[\[\]]+/).inject(hash) do |h, part|
         next h if part == ''
@@ -171,22 +177,42 @@ module Rack
         @params.key?(key)
       end
 
-      def to_params_hash
-        hash = @params
-        hash.keys.each do |key|
-          value = hash[key]
-          if value.kind_of?(self.class)
-            if value.object_id == self.object_id
-              hash[key] = hash
-            else
-              hash[key] = value.to_params_hash
-            end
-          elsif value.kind_of?(Array)
-            value.map! {|x| x.kind_of?(self.class) ? x.to_params_hash : x}
+      # Recursively unwraps nested `Params` objects and constructs an object
+      # of the same shape, but using the objects' internal representations
+      # (Ruby hashes) in place of the objects. The result is a hash consisting
+      # purely of Ruby primitives.
+      #
+      #   Mutation warning!
+      #
+      #   1. This method mutates the internal representation of the `Params`
+      #      objects in order to save object allocations.
+      #
+      #   2. The value you get back is a reference to the internal hash
+      #      representation, not a copy.
+      #
+      #   3. Because the `Params` object's internal representation is mutable
+      #      through the `#[]=` method, it is not thread safe. The result of
+      #      getting the hash representation while another thread is adding a
+      #      key to it is non-deterministic.
+      #
+      def to_h
+        @params.each do |key, value|
+          case value
+          when self
+            # Handle circular references gracefully.
+            @params[key] = @params
+          when Params
+            @params[key] = value.to_h
+          when Array
+            value.map! { |v| v.kind_of?(Params) ? v.to_h : v }
+          else
+            # Ignore anything that is not a `Params` object or
+            # a collection that can contain one.
           end
         end
-        hash
+        @params
       end
+      alias_method :to_params_hash, :to_h
     end
   end
 end