Paperclip file spoofing corrected/modified.

Dockerfile

@@ -2,7 +2,13 @@ FROM rails:4.2.5
 MAINTAINER BioSistemika <info@biosistemika.com>
 
 # additional dependecies
-RUN apt-get update -qq && apt-get install -y default-jre-headless unison sudo graphviz --no-install-recommends && rm -rf /var/lib/apt/lists/*
+RUN apt-get update -qq && \
+  apt-get install -y \
+  default-jre-headless \
+  unison \
+  sudo graphviz --no-install-recommends \
+  sudo libfile-mimeinfo-perl && \
+  rm -rf /var/lib/apt/lists/*
 
 # heroku tools
 RUN wget -O- https://toolbelt.heroku.com/install-ubuntu.sh | sh

app/assets/javascripts/direct-upload.js

@@ -53,9 +53,9 @@
   }
 
   /*
-   * The server checks if files are OK (correct file type, presence,
-   * size and spoofing) and only then generates posts for S3 server file
-   * uploading (each post for different size/style of the same file).
+   * The server checks if files are OK (presence, size and spoofing)
+   * and only then generates posts for S3 server file uploading
+   * (each post for different size/style of the same file).
    */
   function fetchUploadSignature(ev, fileInput, file, signUrl) {
     var formData = new FormData();
@@ -69,17 +69,14 @@
       contentType: false,
       error: function (xhr) {
         try {
-          var data = JSON.parse(xhr.responseText);
-          if (data.status === "error") {
-            // File error
-            var errMsg = jsonToValuesArray(data.errors);
-            renderFormError(ev, fileInput, errMsg);
-          }
+          // File error
+          var jsonData = $.parseJSON(xhr.responseText);
+          var errMsg = jsonToValuesArray(jsonData.errors);
         } catch(err) {
           // Connection error
           var errMsg = I18n.t("general.file.upload_failure");
-          renderFormError(ev, fileInput, errMsg);
         }
+        renderFormError(ev, fileInput, errMsg);
       }
     });
   }
@@ -102,9 +99,15 @@
       data: formData,
       processData: false,
       contentType: false,
-      error: function () {
-        // Connection error
-        var errMsg = I18n.t("general.file.upload_failure");
+      error: function (xhr) {
+        try {
+          // File error
+          var $xmlData = $(xhr.responseText);
+          var errMsg = $xmlData.find("Message").text().strToErrorFormat();
+        } catch(err) {
+          // Connection error
+          var errMsg = I18n.t("general.file.upload_failure");
+        }
         renderFormError(ev, fileInput, errMsg);
       }
     });

app/controllers/assets_controller.rb

@@ -7,12 +7,8 @@ class AssetsController < ApplicationController
   def signature
     respond_to do |format|
       format.json {
-
         asset = Asset.new(asset_params)
         if asset.errors.any?
-          # We need to validate, although 'new' already does it, so that
-          # asset's after_validation gets triggered, which modifies errors
-          asset.valid?
           render json: {
             status: 'error',
             errors: asset.errors

app/models/asset.rb

@@ -42,9 +42,11 @@ class Asset < ActiveRecord::Base
   has_many :report_elements, inverse_of: :asset, dependent: :destroy
   has_one :asset_text_datum, inverse_of: :asset, dependent: :destroy
 
-  # Specific file errors propagate to "fire" error hash key,
+  # Specific file errors propagate to "file" error hash key,
   # so use just these errors
   after_validation :filter_paperclip_errors
+  # Needed because Paperclip validatates on creation
+  after_create :filter_paperclip_errors
 
   attr_accessor :file_content, :file_info, :preview_cached
 
@@ -343,5 +345,4 @@ class Asset < ActiveRecord::Base
 
     self.file = data
   end
-
 end

config/initializers/paperclip.rb

@@ -57,3 +57,115 @@ Paperclip::Attachment.class_eval do
     Paperclip.io_adapters.for self
   end
 end
+
+module Paperclip
+  # Checks file for spoofing
+  class MediaTypeSpoofDetector
+    def spoofed?
+      if has_name? && has_extension? && (media_type_mismatch? ||
+        mapping_override_mismatch?)
+        Paperclip.log("Content Type Spoof: Filename #{File.basename(@name)} "\
+          "(#{supplied_content_type} from Headers, #{content_types_from_name} "\
+          'from Extension), content type discovered: '\
+          "#{calculated_content_type}. See documentation to allow this "\
+          'combination.')
+        true
+      else
+        false
+      end
+    end
+
+    private
+
+    # Determine file content type from its name
+    def content_types_from_name
+      @content_types_from_name ||=
+        Paperclip.run('mimetype', '-b :file_name', file_name: @name).chomp
+    end
+
+    # Determine file media type from its name
+    def media_types_from_name
+      @media_types_from_name ||= extract_media_type content_types_from_name
+    end
+
+    # Determine file content type from mimetype command
+    def type_from_mimetype_command
+      @type_from_mimetype_command ||=
+        Paperclip.run('mimetype', '-b :file', file: @file.path).chomp
+    end
+
+    # Determine file media type from mimetype command
+    def media_type_from_mimetype_command
+      @media_type_from_mimetype_command ||=
+        extract_media_type type_from_mimetype_command
+    end
+
+    # Determine file content type from it's content (file and mimetype command)
+    def type_from_file_command
+      unless defined? @type_from_file_command
+        @type_from_file_command =
+          Paperclip.run('file', '-b --mime :file', file: @file.path)
+                   .split(/[:;]\s+/).first
+
+        if allowed_spoof_exception?(@type_from_file_command,
+                                    media_type_from_file_command) ||
+           (@type_from_file_command.in?(%w(text/plain text/html)) &&
+            media_type_from_mimetype_command.in?(%w(text application)))
+          # File content type is generalized, so rely on file extension for
+          # correct/more specific content type
+          @type_from_file_command = type_from_mimetype_command
+        end
+      end
+      @type_from_file_command
+    rescue Cocaine::CommandLineError
+      ''
+    end
+
+    # Determine file media type from it's content (file and mimetype command)
+    def media_type_from_file_command
+      @media_type_from_file_command ||=
+        extract_media_type type_from_file_command
+    end
+
+    def extract_media_type(content_type)
+      if content_type.empty?
+        ''
+      else
+        content_type.split('/').first
+      end
+    end
+
+    # Checks file media type mismatch between file's name and header
+    def supplied_type_mismatch?
+      !allowed_spoof_exception?(supplied_content_type, supplied_media_type) &&
+        media_types_from_name != supplied_media_type
+    end
+
+    # Checks file media type mismatch between file's name and content
+    def calculated_type_mismatch?
+      !allowed_spoof_exception?(calculated_content_type,
+                                calculated_media_type) &&
+        media_types_from_name != calculated_media_type
+    end
+
+    # Checks file content type mismatch between file's name and content
+    def mapping_override_mismatch?
+      !allowed_spoof_exception?(calculated_content_type,
+                                calculated_media_type) &&
+        content_types_from_name != calculated_content_type
+    end
+
+    # Check if we have a file spoof exception which is allowed/safe
+    def allowed_spoof_exception?(content_type, media_type)
+      content_type == 'application/octet-stream' ||
+        (content_type == 'inode/x-empty' && @file.size.zero?) ||
+        (content_type == 'text/x-c' &&
+         content_types_from_name == 'text/x-java') ||
+        (media_type.in?(%w(image audio video)) &&
+         media_type == media_types_from_name) ||
+        (content_types_from_name.in? %W(#{}
+                                        text/plain
+                                        application/octet-stream))
+    end
+  end
+end