akita-har_logger 0.2.3 → 0.2.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 5a4d073bcb2347b0527e3dab531c77770a29c29326bf76cb2ca09724206af8e5
-  data.tar.gz: 2fc8b6e0b4721699f87a08ba098b3d6137835b14c01eeb73b945a453a98267f2
+  metadata.gz: b2030633ff794a611ceb1fddba8b274e5617765c0644dcdebbe6743515dd0576
+  data.tar.gz: c3b1eb76987d0db9dcf63830a67882cbdfbaceeba937c1a9a20a874ddc4efd42
 SHA512:
-  metadata.gz: f07983494e000f9ed64ffdfe346869bdd0b574db21210a39de0199d1df81f877262c2623c72ca918c792cff77a29872224a5ba3d5f3ac12691381d8caf89bc17
-  data.tar.gz: 513be52cfaf7f114cb95ed33df45528336db9f9945ac5252db12a9abfe30103308cd1d7d92585ec6ac8897c231129e0d83cee596a3df3034946c88b5c59dcde6
+  metadata.gz: 7b7e2cbf0e7580c044ca178d58af63fc41059318475f033314bf9fbba9ce3546da02808b2c237c6cef6ada40d6b6eebaf905ed2201f943367deea8fc49342d5b
+  data.tar.gz: c8637bc4282a1571b0366e1dbff4948771a7d10913b26c7df595996b4096dcd18e8119448cf3f578c9d33f3acf12034c05d92b62ce1900eff34b4c5adb71fe9f

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    akita-har_logger (0.2.3)
+    akita-har_logger (0.2.7)
 
 GEM
   remote: https://rubygems.org/

data/lib/akita/har_logger/har_utils.rb

@@ -4,13 +4,32 @@ module Akita
   module HarLogger
     class HarUtils
       # Rack apparently uses 8-bit ASCII for everything, even when the string
-      # is not 8-bit ASCII. This reinterprets the given string as UTF-8.
+      # is not 8-bit ASCII. This reinterprets 8-bit ASCII strings as UTF-8.
+      #
+      # If we are unable to do this reinterpretation, return the string
+      # unchanged, but log a warning that points to the caller.
       def self.fixEncoding(v)
-        if v == nil || v.encoding == Encoding::UTF_8 then
-          v
-        else
-          String.new(v).force_encoding(Encoding::UTF_8)
+        if v == nil then
+          return v
         end
+
+        if !(v.is_a? String) then
+          Rails.logger.warn "[#{caller_locations(1, 1)}] fixEncoding was not given a string. This might cause JSON serialization to fail."
+          return v
+        end
+
+        # Only re-interpret 8-bit ASCII.
+        if v.encoding != Encoding::ASCII_8BIT then
+          return v
+        end
+
+        forced = String.new(v).force_encoding(Encoding::UTF_8)
+        if forced.valid_encoding? then
+          return forced
+        end
+
+        Rails.logger.warn "[#{caller_locations(1, 1)}] Unable to fix encoding: not a valid UTF-8 string. This will likely cause JSON serialization to fail."
+        v
       end
 
       # Converts a Hash into a list of Hash objects. Each entry in the given

data/lib/akita/har_logger/http_request.rb

@@ -85,7 +85,7 @@ module Akita
           return Encoding::ISO_8859_1
         end
 
-        Encoding.default_external
+        Encoding::UTF_8
       end
 
       # Obtains the posted data from an HTTP environment.
@@ -117,8 +117,27 @@ module Akita
             # body when the request specifies UTF-8. Reinterpret the content
             # body according to what the request says it is, and re-encode into
             # UTF-8.
-            result[:text] = req.body.string.encode(Encoding::UTF_8,
-                getPostDataCharSet(env))
+            #
+            # Gracefully handle any characters that are invalid in the source
+            # encoding and characters that have no UTF-8 representation by
+            # replacing with '?'. Log a warning when this happens.
+            source = req.body.string.force_encoding(getPostDataCharSet(env))
+            utf8EncodingSuccessful = false
+            if source.valid_encoding? then
+              begin
+                result[:text] = source.encode(Encoding::UTF_8)
+                utf8EncodingSuccessful = true
+              rescue Encoding::UndefinedConversionError
+                Rails.logger.warn "[#{caller_locations(0, 1)}] Unable to losslessly convert request body from #{source.encoding} to UTF-8. Characters undefined in UTF-8 will be replaced with '?'."
+              end
+            else
+              Rails.logger.warn "[#{caller_locations(0, 1)}] Request body is not valid #{source.encoding}. Invalid characters and characters undefined in UTF-8 will be replaced with '?'."
+            end
+
+            if !utf8EncodingSuccessful then
+              result[:text] = source.encode(Encoding::UTF_8,
+                  invalid: :replace, undef: :replace, replace: '?')
+            end
           end
 
           result
@@ -144,7 +163,7 @@ module Akita
         # for the CRLF on the blank line.
         getHeaders(env).reduce(line_length + 2) { |accum, entry|
           # Header-Name: header value<CR><LF>
-          accum + entry[:name].length + 2 + entry[:value].length + 2
+          accum + entry[:name].length + 2 + entry[:value].to_s.length + 2
         }
       end
 

data/lib/akita/har_logger/http_response.rb

@@ -25,7 +25,7 @@ module Akita
 
       # Obtains the status text corresponding to a status code.
       def getStatusText(status)
-        Rack::Utils::HTTP_STATUS_CODES[status]
+        HarUtils.fixEncoding(Rack::Utils::HTTP_STATUS_CODES[status])
       end
 
       # Obtains the HTTP version in the response.
@@ -65,8 +65,8 @@ module Akita
           if match then cookie_value = match[1] end
 
           result << {
-            name: cookie_name,
-            value: cookie_value,
+            name: HarUtils.fixEncoding(cookie_name),
+            value: HarUtils.fixEncoding(cookie_value),
           }
         }
 
@@ -74,19 +74,56 @@ module Akita
       end
 
       def getContent(headers, body)
-        # XXX Handle compression & encoding.
+        # XXX Handle compression
+        # XXX Figure out how to properly join together multi-part bodies.
 
+        # Try to convert the body into UTF-8. If this fails, assume the body is
+        # binary data.
+        # XXX TODO Take charset part of Content-Type header into account.
         text = +""
+        haveBinaryData = false
         body.each { |part|
-          # XXX Figure out how to join together multi-part bodies.
-          text << part;
+          partStr = part.to_s
+
+          if partStr.encoding == Encoding::ASCII_8BIT then
+            # Have 8-bit ASCII data. Try to interpret as UTF-8. If this fails,
+            # treat as binary data.
+            forced = String.new(partStr).force_encoding(Encoding::UTF_8)
+            if forced.valid_encoding? then
+              text << forced
+              next
+            end
+
+            haveBinaryData = true
+            break
+          end
+
+          if !partStr.valid_encoding? then
+            # Source encoding is not valid. Treat as binary data.
+            haveBinaryData = true
+            break
+          end
+
+          # Try to re-encode as UTF-8. If this fails, treat as binary data.
+          begin
+            text << partStr.encode(Encoding::UTF_8)
+          rescue Encoding::UndefinedConversionError
+            haveBinaryData = true
+            break
+          end
         }
 
+        if haveBinaryData then
+          # TODO Encode binary body data with base64.
+          # XXX Omit for now.
+          text = ""
+        end
+
         {
           size: getBodySize(body),
 
           # XXX What to use when no Content-Type is given?
-          mimeType: headers['Content-Type'],
+          mimeType: HarUtils.fixEncoding(headers['Content-Type']),
 
           text: text,
         }
@@ -95,7 +132,9 @@ module Akita
       def getRedirectUrl(headers)
         # Use the "Location" header if it exists. Otherwise, based on some HAR
         # examples found online, it looks like an empty string is used.
-        headers.key?('Location') ? headers['Location'] : ''
+        headers.key?('Location') ?
+          HarUtils.fixEncoding(headers['Location']) :
+          ''
       end
 
       def getHeadersSize(env, status, headers)
@@ -114,7 +153,7 @@ module Akita
         # for the CRLF on the blank line.
         headers.reduce(status_length + 2) { |accum, (k, v)|
           # Header-Name: header value<CR><LF>
-          accum + k.length + 2 + v.length + 2
+          accum + k.length + 2 + v.to_s.length + 2
         }
       end
 

data/lib/akita/har_logger/version.rb

@@ -2,6 +2,6 @@
 
 module Akita
   module HarLogger
-    VERSION = "0.2.3"
+    VERSION = "0.2.7"
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: akita-har_logger
 version: !ruby/object:Gem::Version
-  version: 0.2.3
+  version: 0.2.7
 platform: ruby
 authors:
 - Jed Liu
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2021-07-11 00:00:00.000000000 Z
+date: 2021-07-15 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rspec