lib/mongo/util/bson.rb in mongo-0.0.3 vs lib/mongo/util/bson.rb in mongo-0.0.4

@@ -15,12 +15,15 @@
 # ++
 
 require 'base64'
 require 'mongo/util/byte_buffer'
 require 'mongo/util/ordered_hash'
-require 'mongo/objectid'
-require 'mongo/dbref'
+require 'mongo/types/binary'
+require 'mongo/types/dbref'
+require 'mongo/types/objectid'
+require 'mongo/types/regexp_of_holding'
+require 'mongo/types/undefined'
 
 # A BSON seralizer/deserializer.
 class BSON
 
   MINKEY = -1
@@ -41,12 +44,22 @@
   SYMBOL = 14
   CODE_W_SCOPE = 15
   NUMBER_INT = 16
   MAXKEY = 127
 
+  if RUBY_VERSION >= '1.9'
+    def self.to_utf8(str)
+      str.encode("utf-8")
+    end
+  else
+    def self.to_utf8(str)
+      str                       # TODO punt for now
+    end
+  end
+
   def self.serialize_cstr(buf, val)
-    buf.put_array(val.to_s.unpack("C*") + [0])
+    buf.put_array(to_utf8(val.to_s).unpack("C*") + [0])
   end
 
   def initialize(db=nil)
     # db is only needed during deserialization when the data contains a DBRef
     @db = db
@@ -65,11 +78,11 @@
     @buf.put_int(0)
 
     obj.each {|k, v|
       type = bson_type(v, k)
       case type
-      when STRING, CODE
+      when STRING, CODE, SYMBOL
         serialize_string_element(@buf, k, v, type)
       when NUMBER, NUMBER_INT
         serialize_number_element(@buf, k, v, type)
       when OBJECT
         serialize_object_element(@buf, k, v)
@@ -85,12 +98,10 @@
         serialize_date_element(@buf, k, v)
       when NULL
         serialize_null_element(@buf, k)
       when REF
         serialize_dbref_element(@buf, k, v)
-      when SYMBOL
-        serialize_symbol_element(@buf, k, v)
       when BINARY
         serialize_binary_element(@buf, k, v)
       when UNDEFINED
         serialize_undefined_element(@buf, k)
       when CODE_W_SCOPE
@@ -116,10 +127,13 @@
       type = @buf.get
       case type
       when STRING, CODE
         key = deserialize_cstr(@buf)
         doc[key] = deserialize_string_data(@buf)
+      when SYMBOL
+        key = deserialize_cstr(@buf)
+        doc[key] = deserialize_string_data(@buf).intern
       when NUMBER
         key = deserialize_cstr(@buf)
         doc[key] = deserialize_number_data(@buf)
       when NUMBER_INT
         key = deserialize_cstr(@buf)
@@ -149,13 +163,10 @@
         key = deserialize_cstr(@buf)
         doc[key] = XGen::Mongo::Driver::Undefined.new
       when REF
         key = deserialize_cstr(@buf)
         doc[key] = deserialize_dbref_data(@buf, key, parent)
-      when SYMBOL
-        key = deserialize_cstr(@buf)
-        doc[key] = deserialize_symbol_data(@buf)
       when BINARY
         key = deserialize_cstr(@buf)
         doc[key] = deserialize_binary_data(@buf)
       when CODE_W_SCOPE
         # TODO
@@ -168,10 +179,11 @@
     end
     @buf.rewind
     doc
   end
 
+  # For debugging.
   def hex_dump
     str = ''
     @buf.to_a.each_with_index { |b,i|
       if (i % 8) == 0
         str << "\n" if i > 0
@@ -219,17 +231,22 @@
     options_str = deserialize_cstr(buf)
     options = 0
     options |= Regexp::IGNORECASE if options_str.include?('i')
     options |= Regexp::MULTILINE if options_str.include?('m')
     options |= Regexp::EXTENDED if options_str.include?('x')
-    Regexp.new(str, options)
+    options_str.gsub!(/[imx]/, '') # Now remove the three we understand
+    XGen::Mongo::Driver::RegexpOfHolding.new(str, options, options_str)
   end
 
   def deserialize_string_data(buf)
     len = buf.get_int
     bytes = buf.get(len)
-    bytes[0..-2].pack("C*")
+    str = bytes[0..-2].pack("C*")
+    if RUBY_VERSION >= '1.9'
+      str.force_encoding("utf-8")
+    end
+    str
   end
 
   def deserialize_oid_data(buf)
     XGen::Mongo::Driver::ObjectID.new(buf.get(12))
   end
@@ -238,14 +255,10 @@
     ns = deserialize_cstr(buf)
     oid = deserialize_oid_data(buf)
     XGen::Mongo::Driver::DBRef.new(parent, key, @db, ns, oid)
   end
 
-  def deserialize_symbol_data(buf)
-    deserialize_cstr(buf).intern
-  end
-
   def deserialize_binary_data(buf)
     len = buf.get_int
     bytes = buf.get(len)
     str = ''
     bytes.each { |c| str << c.chr }
@@ -266,16 +279,10 @@
     self.class.serialize_cstr(buf, key)
     self.class.serialize_cstr(buf, val.namespace)
     buf.put_array(val.object_id.to_a)
   end
 
-  def serialize_symbol_element(buf, key, val)
-    buf.put(SYMBOL)
-    self.class.serialize_cstr(buf, key)
-    self.class.serialize_cstr(buf, val)
-  end
-
   def serialize_binary_element(buf, key, val)
     buf.put(BINARY)
     self.class.serialize_cstr(buf, key)
     buf.put_int(val.length)
     bytes = if RUBY_VERSION >= '1.9'
@@ -340,11 +347,13 @@
     options = val.options
     options_str = ''
     options_str << 'i' if ((options & Regexp::IGNORECASE) != 0)
     options_str << 'm' if ((options & Regexp::MULTILINE) != 0)
     options_str << 'x' if ((options & Regexp::EXTENDED) != 0)
-    self.class.serialize_cstr(buf, options_str)
+    options_str << val.extra_options_str if val.respond_to?(:extra_options_str)
+    # Must store option chars in alphabetical order
+    self.class.serialize_cstr(buf, options_str.split(//).sort.uniq.join)
   end
 
   def serialize_oid_element(buf, key, val)
     buf.put(OID)
     self.class.serialize_cstr(buf, key)
@@ -376,9 +385,12 @@
     chars = ""
     while 1
       b = buf.get
       break if b == 0
       chars << b.chr
+    end
+    if RUBY_VERSION >= '1.9'
+      chars.force_encoding("utf-8") # Mongo stores UTF-8
     end
     chars
   end
 
   def bson_type(o, key)