Skip to main content

Match Rules

Match Rules are a text-based filter based on one or more strings. They are used for rate limits and service matching.

Format

A Match Rule is optionally prefixed with one of the following:

  • regex:: parsed as regex, automatically anchored to beginning and end of input
  • raw:: exact match required (default)
  • raw_insensitive:: exact match required (case insensitive)
  • except:: can be used to negate previous rules, requires exact match like raw
  • except_regex:: same as regex, but also negates previous rules like except
  • internal:: Use a natively implemented matching function.
  • and: a list of rules. overrides the behavior of the group so that a match is only emitted if a rule in the group matches AND every rule in the and rule matches
  • correlate: Allows you to specify a sub-group of rules. Matches from the parent group will only be emitted when a match from the sub-group is found nearby.

Internal matchers

Internal matchers can be specified via the "internal" match rule. They allow you to use highly optimized dedicated matchers for common match cases or cases too complex for regex to easily handle. The currently implemented internal matchers are:

  • routing_number: Matches on bank routing numbers
  • credit_card: Matches on credit/debit card numbers. Supports almost ever major bank, and requires the number to have a valid LUHN checksum
  • int_phone: Matches on international phone numbers via googles libphonenumber library. Requires the country code to be specified beforehand (ie +1 or +33). Does not support IDD codes, does not support full RFC3966 syntax (like extensions).
  • national_phone: Matches on phone numbers specific to a country. Requires the country ID to be specified (ie US, NL). Since you specify the country in the policy, the number doesn't require the country code (ie +1) to be specified. NOTE: This is a "secondary matcher", meaning it can only be used inside and MatchRules. This means you need to specify a custom matcher to match for the phone number format of your desired country, and THEN specify national_phone in the and matcher to perform a full check.
  • hash: A bundle of regexes to match against easily identifiable hash types. You can optionally specify a list of which hash types to use, or leave it empty to use the default list. A full list can be found here
  • error_text: A bundle of regexes to match against easily identifiable error messages for various programming languages and frameworks. You can optionally specify a list of which error types to target, or leave it empty to use the default list. A full list can be found here
  • sql: Matches on SQL statements. To avoid false positives, extremely simple or benign statements are not considered a match.

Examples

(one rule per line, double line break is a separate set of rules)

regex: test[0-7]{3}
except: test000

regex: [a-z]+@[a-z]{2,15}\.[a-z]{2,5}
except: test@example.com
except: example@test.com
except_regex: (?:no-reply|noreply)@.*


regex: "\\b\\d{3}[ .-]\\d{2}[ .-]\\d{4}\\b"
correlate:
interest: primary
max_distance: 16
matches:
- raw_insensitive: ssn
correlate:
interest: secondary
max_distance: 16
matches:
- raw_insensitive: social
- raw_insensitive: security

national_phone matcher for the US:

phone:
- regex: "(?u)\\b(1[ .-]?)?[2-9]\\d{2}[ .-]?\\d{3}[ .-]?\\d{4}\\b|(\\b1[ .-]?)?\\([2-9]\\d{2}\\)[ .-]?\\d{3}[ .-]?\\d{4}\\b"
- and:
- internal: !national_phone US

supported hashes types

The following is a list of supported hash types. If no hashes are specified, then all types are used.

  • bcrypt
  • sha256_crypt
  • sha512_crypt
  • md5_crypt
  • sha1_crypt
  • sun_md5_crypt
  • argon2
  • bcrypt_sha256
  • phpass
  • pbkdf2_sha1
  • pbkdf2_sha256
  • pbkdf2_sha512
  • scram
  • scrypt
  • apr_md5_crypt
  • cta_pbkdf2_sha1
  • dlitz_pbkdf2_sha1
  • mssql2000
  • mssql2005
  • mysql141
  • postgres_md5
  • oracle11
  • bsd_nthash
  • django_pbkdf2_sha1
  • django_pbkdf2_sha256
  • django_salted_sha1
  • django_salted_md5
  • django_des_crypt
  • grub_pbkdf2_sha512

supported error_text types

The following is a list of the supported error text matchers. If no matchers are specified, then the ones marked with * are used.

    • js_error
    • py_error
    • java_error
    • rust_error
    • ruby_error
    • go_error
    • php_error
    • mysql_error
    • postgres_error
    • redis_error
    • mongodb_error
  • type_error
  • uncaught
  • socket_error
  • operation_not_supported
  • callback
  • segfault
  • runtime_error
  • out_of_memory
  • permission_denied
  • command_not_found
  • js_unknown_argument
  • js_invalid_invocation
  • js_bug_message
  • envoy_segfault