Module: TingYun::Agent::Database::Obfuscator::ObfuscationHelpers

Included in:
TingYun::Agent::Database::Obfuscator
Defined in:
lib/ting_yun/agent/database.rb

Constant Summary collapse

REVERSE_SINGLE_QUOTES_REGEX =

Note that the following two regexes are applied to a reversed version of the query. This is why the backslash escape sequences (' and ") appear reversed within them.

Note that some database adapters (notably, PostgreSQL with standard_conforming_strings on and MySQL with NO_BACKSLASH_ESCAPES on) do not apply special treatment to backslashes within quoted string literals. We don’t have an easy way of determining whether the database connection from which a query was captured was operating in one of these modes, but the obfuscation is done in such a way that it should not matter.

Reversing the query string before obfuscation allows us to get around the fact that a ' appearing within a string may or may not terminate the string, because we know that a string cannot start with a '.

/'(?:''|'\\|[^'])*'/
REVERSE_ANY_QUOTES_REGEX = 
/'(?:''|'\\|[^'])*'|"(?:""|"\\|[^"])*"/
NUMERICS_REGEX = 
/\b\d+\b/
SQL_COMMENT_REGEX =

We take a conservative, overly-aggressive approach to obfuscating comments, and drop everything from the query after encountering any character sequence that could be a comment initiator. We do this after removal of string literals to avoid accidentally over-obfuscating when a string literal contains a comment initiator.

Regexp.new('(?:/\*|--|#).*', Regexp::MULTILINE).freeze
LITERAL_SINGLE_QUOTE =

We use these to check whether the query contains any quote characters after obfuscation. If so, that’s a good indication that the original query was malformed, and so our obfuscation can’t reliabily find literals. In such a case, we’ll replace the entire query with a placeholder.

"'".freeze
LITERAL_DOUBLE_QUOTE = 
'"'.freeze
PLACEHOLDER = 
'?'.freeze

Instance Method Summary collapse

Instance Method Details

#contains_quotes?(str) ⇒ Boolean 



401
402
403
 
# File 'lib/ting_yun/agent/database.rb', line 401

def contains_quotes?(str)
  str.include?(LITERAL_SINGLE_QUOTE) || str.include?(LITERAL_DOUBLE_QUOTE)
end

#contains_single_quotes?(str) ⇒ Boolean 



397
398
399
 
# File 'lib/ting_yun/agent/database.rb', line 397

def contains_single_quotes?(str)
  str.include?(LITERAL_SINGLE_QUOTE)
end

#obfuscate_numeric_literals(sql) ⇒ Object 



389
390
391
 
# File 'lib/ting_yun/agent/database.rb', line 389

def obfuscate_numeric_literals(sql)
  sql.gsub(NUMERICS_REGEX, PLACEHOLDER)
end

#obfuscate_quoted_literals(sql) ⇒ Object 



382
383
384
385
386
387
 
# File 'lib/ting_yun/agent/database.rb', line 382

def obfuscate_quoted_literals(sql)
  obfuscated = sql.reverse
  obfuscated.gsub!(REVERSE_ANY_QUOTES_REGEX, PLACEHOLDER)
  obfuscated.reverse!
  obfuscated
end

#obfuscate_single_quote_literals(sql) ⇒ Object 



375
376
377
378
379
380
 
# File 'lib/ting_yun/agent/database.rb', line 375

def obfuscate_single_quote_literals(sql)
  obfuscated = sql.reverse
  obfuscated.gsub!(REVERSE_SINGLE_QUOTES_REGEX, PLACEHOLDER)
  obfuscated.reverse!
  obfuscated
end

#remove_comments(sql) ⇒ Object 



393
394
395
 
# File 'lib/ting_yun/agent/database.rb', line 393

def remove_comments(sql)
  sql.gsub(SQL_COMMENT_REGEX, PLACEHOLDER)
end