Sha256: d11a80d63f42aad1c7119d2f8c367d0a3c5be4c0d7caf3812c70c0e5f38afae6
Contents?: true
Size: 1.1 KB
Versions: 76
Compression:
Stored size: 1.1 KB
Contents
module FlydataCore
module Redshift
class String
NULL_CHAR_CODE_POINTS = "\\0".codepoints
# Redshift supports UTF-8 but it enforces stricter rule than other
# implementations such as MySQL or Ruby. This method returns a
# Redshift-safe string from the given string.
def self.encode(string, options = {})
encoded_str = string.encoding == Encoding::UTF_8 ?
string.encode(Encoding::UTF_16, options).encode(Encoding::UTF_8) :
string.encode(Encoding::UTF_8, options)
ret_cpts = []
rep_str_cpts = (options[:replace] || "\uFFFD").codepoints
replaced = false
encoded_str.each_codepoint do |cp|
# Per Redshift document
# http://docs.aws.amazon.com/redshift/latest/dg/multi-byte-character-load-errors.html
if cp >= 0xFDD0 && cp <= 0xFDEF || cp == 0xFFFE || cp == 0xFFFF
ret_cpts.concat(rep_str_cpts)
replaced = true
elsif cp == 0x0000
ret_cpts.concat(NULL_CHAR_CODE_POINTS)
replaced = true
else
ret_cpts << cp
end
end
replaced ? ret_cpts.pack("U*") : encoded_str
end
end
end
end
Version data entries
76 entries across 76 versions & 1 rubygems