HBASE-28549 Make shell commands support column qualifiers with colons (…

…#5849)

Signed-off-by: Duo Zhang <zhangduo@apache.org>

hbase-shell/src/main/ruby/hbase/table.rb

@@ -135,7 +135,7 @@ def close
     # Put a cell 'value' at specified table/row/column
     def _put_internal(row, column, value, timestamp = nil, args = {})
       p = org.apache.hadoop.hbase.client.Put.new(row.to_s.to_java_bytes)
-      family, qualifier = parse_column_name(column)
+      family, qualifier = split_column_name(column)
       if args.any?
         attributes = args[ATTRIBUTES]
         set_attributes(p, attributes) if attributes
@@ -188,14 +188,14 @@ def _createdelete_internal(row, column = nil,
       end
       if column != ""
         if column && all_version
-          family, qualifier = parse_column_name(column)
+          family, qualifier = split_column_name(column)
           if qualifier
             d.addColumns(family, qualifier, timestamp)
           else
             d.addFamily(family, timestamp)
           end
         elsif column && !all_version
-          family, qualifier = parse_column_name(column)
+          family, qualifier = split_column_name(column)
           if qualifier
             d.addColumn(family, qualifier, timestamp)
           else
@@ -273,7 +273,7 @@ def _incr_internal(row, column, value = nil, args = {})
       value = 1 if value.is_a?(Hash)
       value ||= 1
       incr = org.apache.hadoop.hbase.client.Increment.new(row.to_s.to_java_bytes)
-      family, qualifier = parse_column_name(column)
+      family, qualifier = split_column_name(column)
       if args.any?
         attributes = args[ATTRIBUTES]
         visibility = args[VISIBILITY]
@@ -296,7 +296,7 @@ def _incr_internal(row, column, value = nil, args = {})
     # appends the value atomically
     def _append_internal(row, column, value, args = {})
       append = org.apache.hadoop.hbase.client.Append.new(row.to_s.to_java_bytes)
-      family, qualifier = parse_column_name(column)
+      family, qualifier = split_column_name(column)
       if args.any?
         attributes = args[ATTRIBUTES]
         visibility = args[VISIBILITY]
@@ -491,7 +491,7 @@ def _get_internal(row, *args)
     #----------------------------------------------------------------------------------------------
     # Fetches and decodes a counter value from hbase
     def _get_counter_internal(row, column)
-      family, qualifier = parse_column_name(column.to_s)
+      family, qualifier = split_column_name(column.to_s)
       # Format get request
       get = org.apache.hadoop.hbase.client.Get.new(row.to_s.to_java_bytes)
       get.addColumn(family, qualifier)
@@ -833,16 +833,16 @@ def convert_bytes(bytes, converter_class = nil, converter_method = nil)
       eval(converter_class).method(converter_method).call(bytes)
     end
 
-    def convert_bytes_with_position(bytes, offset, len, converter_class, converter_method)
-      # Avoid nil
-      converter_class ||= 'org.apache.hadoop.hbase.util.Bytes'
-      converter_method ||= 'toStringBinary'
-      eval(converter_class).method(converter_method).call(bytes, offset, len)
-    end
-
     # store the information designating what part of a column should be printed, and how
     ColumnFormatSpec = Struct.new(:family, :qualifier, :converter)
 
+    # Use this instead of parse_column_name if the name cannot contain a converter
+    private def split_column_name(column)
+      # NOTE: We use 'take(2)' instead of 'to_a' to avoid type coercion of the nested byte arrays.
+      # https://github.com/jruby/jruby/blob/9.3.13.0/core/src/main/java/org/jruby/java/proxies/ArrayJavaProxy.java#L484-L488
+      org.apache.hadoop.hbase.CellUtil.parseColumn(column.to_java_bytes).take(2)
+    end
+
     ##
     # Parse the column specification for formatting used by shell commands like :scan
     #
@@ -856,21 +856,29 @@ def convert_bytes_with_position(bytes, offset, len, converter_class, converter_m
     # @param [String] column
     # @return [ColumnFormatSpec] family, qualifier, and converter as Java bytes
     private def parse_column_format_spec(column)
-      split = org.apache.hadoop.hbase.CellUtil.parseColumn(column.to_java_bytes)
-      family = split[0]
-      qualifier = nil
       converter = nil
-      if split.length > 1
-        parts = org.apache.hadoop.hbase.CellUtil.parseColumn(split[1])
-        qualifier = parts[0]
-        if parts.length > 1
-          converter = parts[1]
+      family, qualifier = split_column_name(column)
+      if qualifier
+        delim = org.apache.hadoop.hbase.KeyValue.getDelimiterInReverse(
+          qualifier, 0, qualifier.length, org.apache.hadoop.hbase.KeyValue::COLUMN_FAMILY_DELIMITER
+        )
+        if delim >= 0
+          prefix, suffix = qualifier[0...delim], qualifier[delim+1..-1]
+          if converter?(suffix.to_s)
+            qualifier = prefix
+            converter = suffix
+          end
         end
       end
 
       ColumnFormatSpec.new(family, qualifier, converter)
     end
 
+    # Check if the expression can be a converter
+    private def converter?(expr)
+      expr =~ /^c\(.+\)\..+/ || Bytes.respond_to?(expr)
+    end
+
     private def set_column_converter(family, qualifier, converter)
       @converters["#{String.from_java_bytes(family)}:#{String.from_java_bytes(qualifier)}"] = String.from_java_bytes(converter)
     end

hbase-shell/src/test/ruby/hbase/table_test.rb

@@ -242,6 +242,42 @@ def teardown
     define_test "get_counter should return nil for non-existent counters" do
       assert_nil(@test_table._get_counter_internal(12345, 'x:qqqq'))
     end
+
+    define_test "should work with qualifiers with colons" do
+      rowkey = "123"
+
+      # Two columns with multiple colons in their qualifiers with the same prefix
+      col1 = "x:foo:bar:c1"
+      col2 = "x:foo:bar:c2"
+
+      # Make sure that no data is present
+      @test_table.deleteall(rowkey)
+
+      # Put two columns with colons in their qualifiers
+      @test_table.put(rowkey, col1, org.apache.hadoop.hbase.util.Bytes.toBytes(1))
+      @test_table.put(rowkey, col2, org.apache.hadoop.hbase.util.Bytes.toBytes(2))
+      assert_equal(2, @test_table._get_internal(rowkey).length)
+
+      # Increment the second column by 10 => 2 + 10 => 12
+      @test_table.incr(rowkey, col2, 10)
+      assert_equal(12, @test_table._get_counter_internal(rowkey, col2))
+
+      # Check the counter value using toLong converter
+      %w[:toLong :c(org.apache.hadoop.hbase.util.Bytes).toLong].each do |suffix|
+        res = @test_table._get_internal(rowkey, { COLUMNS => [col2 + suffix] })
+        assert_not_nil(res)
+        assert_kind_of(Hash, res)
+        assert_not_nil(/value=12/.match(res[col2]))
+      end
+
+      # Delete the first column
+      @test_table.delete(rowkey, col1)
+      assert_equal(1, @test_table._get_internal(rowkey).length)
+
+      # Append twice to the deleted column
+      @test_table.append(rowkey, col1, '123')
+      assert_equal("123123", @test_table._append_internal(rowkey, col1, '123'))
+    end
   end
 
   # Complex data management methods tests