Skip to content

Functions

The Cloudflare Rules language provides functions for manipulating and validating values in an expression:

Transformation functions

The Rules language supports several functions that transform values extracted from HTTP requests. A common use case for transformation functions is the conversion of a string of characters to uppercase or lowercase, since by default, string evaluation is case-sensitive.

For example, the lower() function converts all uppercase characters in a string to lowercase.

In the expression below, the lower() function transforms http.host values to lowercase so that they match the target value "www.cloudflare.com":

lower(http.host) == "www.cloudflare.com"

Transformation functions that do not take arrays as an argument type require the [*] index notation. Refer to Arrays for more information.

The Rules language supports these transformation functions:

any

any(Array<Boolean>)Boolean

Returns true when the comparison operator in the argument returns true for any of the values in the argument array. Returns false otherwise.

Example:

any(url_decode(http.request.body.form.values[*])[*] contains "an xss attack")

all

all(Array<Boolean>)Boolean

Returns true when the comparison operator in the argument returns true for all values in the argument array. Returns false otherwise.

Example:

all(http.request.headers["content-type"][*] == "application/json")

cidr

cidr(address (IP address), ipv4_network_bits (Integer), ipv6_network_bits (Integer))IP address

Returns the network address corresponding to an IP address (IPv4 or IPv6), given the provided IPv4 and IPv6 network bits (which determine the corresponding netmasks).

The address parameter must be a field, that is, it cannot be a literal String.

The ipv4_network_bits value must be between 1 and 32, and the ipv6_network_bits value must be between 1 and 128.

Examples:

  • If ip.src is 113.10.0.2, cidr(ip.src, 24, 24) will return 113.10.0.0.
  • If ip.src is 2001:0000:130F:0000:0000:09C0:876A:130B, cidr(ip.src, 24, 24) will return 2001:0000:0000:0000:0000:0000:0000:0000.

cidr6

cidr6(address (IP address), ipv6_network_bits (Integer))IP address

Returns the IPv6 network address corresponding to an IPv6 address, given the provided network bits (which determine the netmask). If you provide an IPv4 address in the first parameter, it will be returned unchanged.

The address parameter must be a field, that is, it cannot be a literal String.

The ipv6_network_bits value must be between 1 and 128.

This function is equivalent to: cidr(<address>, 32, <ipv6_network_bits>).

Examples:

  • If ip.src is 2001:0000:130F:0000:0000:09C0:876A:130B, cidr6(ip.src, 24) will return 2001:0000:0000:0000:0000:0000:0000:0000.
  • If ip.src is 113.10.0.2, cidr6(ip.src, 24) will return 113.10.0.2 (unchanged).

concat

concat(String | Integer | Bytes | Array elements)String

Takes a comma-separated list of values. Concatenates the argument values into a single String.

For example, concat("String1", " ", "String", 2) will return "String1 String2".

decode_base64

decode_base64(source (String))String

Decodes a Base64-encoded String specified in source.

source must be a field, that is, it cannot be a literal String.

For example, with the following HTTP request header: client_id: MTIzYWJj, (any(decode_base64(http.request.headers["client_id"][*])[*] eq "123abc")) would return true.

ends_with

ends_with(source (String), substring (String))Boolean

Returns true when the source ends with a given substring. Returns false otherwise. The source cannot be a literal value (like "foo").

For example, if http.request.uri.path is "/welcome.html", then ends_with(http.request.uri.path, ".html") will return true.

len

len(String | Bytes)Integer

Returns the byte length of a String or Bytes field.

For example, if http.host is "example.com", then len(http.host) will return 11.

lookup_json_integer

lookup_json_integer(field (String), key (String | Integer) [, key (String | Integer), ...])Integer

Returns the integer value associated with the supplied key in field.

The field must be a string representation of a valid JSON document.

The key can be an attribute name, a zero-based position number in a JSON array, or a combination of these two options (as extra function parameters), while following the hierarchy of the JSON document to obtain a specific integer value.

Note: This function only works for plain integers. For example, it will not work for floating numbers with a zero decimal part such as 42.0.

Examples:

  • Given the following JSON object contained in the http.request.body.raw field:
    { "record_id": "aed53a", "version": 2 }
    Then lookup_json_integer(http.request.body.raw, "version") will return 2.

  • Given the following nested object:
    { "product": { "id": 356 } }
    Then lookup_json_integer(http.request.body.raw, "product", "id") will return 356.

  • Given the following JSON array at the root level:
    ["first_item", -234]
    Then lookup_json_integer(http.request.body.raw, 1) will return -234.

  • Given the following array in a JSON object attribute:
    { "network_ids": [123, 456] }
    Then lookup_json_integer(http.request.body.raw, "network_ids", 0) will return 123.

  • Given the following root-level array of JSON objects:
    [{ "product_id": 123 }, { "product_id": 456 }]
    Then lookup_json_integer(http.request.body.raw, 1, "product_id") will return 456.

lookup_json_string

lookup_json_string(field (String) key (String | Integer) [, key (String | Integer), ...])String

Returns the string value associated with the supplied key in field.

The field must be a string representation of a valid JSON document.

The key can be an attribute name, a zero-based position number in a JSON array, or a combination of these two options (as extra function parameters), while following the hierarchy of the JSON document to obtain a specific value.

Examples:

  • Given the following JSON object contained in the http.request.body.raw field:
    { "company": "cloudflare", "product": "rulesets" }
    Then lookup_json_string(http.request.body.raw, "company") == "cloudflare" will return true.

  • Given the following nested object:
    { "network": { "name": "cloudflare" } }
    Then lookup_json_string(http.request.body.raw, "network", "name") == "cloudflare" will return true.

  • Given the following JSON array at the root level:
    ["other_company", "cloudflare"]
    Then lookup_json_string(http.request.body.raw, 1) == "cloudflare" will return true.

  • Given the following array in a JSON object attribute:
    { "networks": ["other_company", "cloudflare"] }
    Then lookup_json_string(http.request.body.raw, "networks", 1) == "cloudflare" will return true.

  • Given the following root-level array of JSON objects:
    [{ "network": "other_company" }, { "network": "cloudflare" }]
    Then lookup_json_string(http.request.body.raw, 1, "network") == "cloudflare" will return true.

lower

lower(String)String

Converts a string field to lowercase. Only uppercase ASCII bytes are converted. All other bytes are unaffected.

For example, if http.host is "WWW.cloudflare.com", then lower(http.host) == "www.cloudflare.com" will return true.

regex_replace

regex_replace(source (String), regular_expression (String), replacement (String))String

Replaces a part of a source string matched by a regular expression with a replacement string, returning the result. The replacement string can contain references to regular expression capture groups (for example, ${1} and ${2}), up to eight replacement references.

Examples:

  • Literal match replace:
    regex_replace("/foo/bar", "/bar$", "/baz") == "/foo/baz"

  • If there is no match, the input string does not change:
    regex_replace("/x", "^/y$", "/mumble") == "/x"

  • Match is case-sensitive by default:
    regex_replace("/foo", "^/FOO$", "/x") == "/foo"

  • When there are multiple matches, only one replacement occurs (the first one):
    regex_replace("/a/a", "/a", "/b") == "/b/a"

  • Escape a $ in the replacement string by prefixing it with another $:
    regex_replace("/b", "^/b$", "/b$$") == "/b$"

  • Replace with capture groups:
    regex_replace("/foo/a/path", "^/foo/([^/]*)/(.*)$", "/bar/${2}/${1}") == "/bar/path/a/"

Create capture groups by putting part of the regular expression in parentheses. Then, reference a capture group using ${<NUMBER>} in the replacement string, where <NUMBER> is the number of the capture group.

remove_bytes

remove_bytes(Bytes)Bytes

Returns a new byte array with all the occurrences of the given bytes removed.

For example, if http.host is "www.cloudflare.com", then remove_bytes(http.host, "\x2e\x77") will return "cloudflarecom".

starts_with

starts_with(source (String), substring (String))Boolean

Returns true when the source starts with a given substring. Returns false otherwise. The source cannot be a literal value (like "foo").

For example, if http.request.uri.path is "/blog/first-post", then starts_with(http.request.uri.path, "/blog") will return true.

substring

substring(field (String | Bytes), start (Integer) [, end (Integer)])String

Returns part of the field value (the value of a String or Bytes field) from the start byte index up to (but excluding) the end byte index. The first byte in field has index 0. If you do not provide the optional end index, the function returns the part of the string from start index to the end of the string.

The start and end indexes can be negative integer values, which allows you to access characters from the end of the string instead of the beginning.

Examples:

// If http.request.body.raw is "asdfghjk":
substring(http.request.body.raw, 2, 5) will return "dfg"
substring(http.request.body.raw, 2) will return "dfghjk"
substring(http.request.body.raw, -2) will return "jk"
substring(http.request.body.raw, 0, -2) will return "asdfgh"

to_string

to_string(Integer | Boolean | IP address)String

Returns the string representation of an Integer, Boolean, or IP address value.

Examples:

// If cf.bot_management.score is 5:
to_string(cf.bot_management.score) will return "5"
// If ssl is true:
to_string(ssl) will return "true"

upper

upper(String)String

Converts a string field to uppercase. Only lowercase ASCII bytes are converted. All other bytes are unaffected.

For example, if http.host is"www.cloudflare.com", then upper(http.host) will return "WWW.CLOUDFLARE.COM".

url_decode

url_decode(source (String)[, options (String)])String

Decodes a URL-formatted string defined in source, as in the following:

  • %20 and + decode to a space character ( ).

  • %E4%BD decodes to ä½.

The source must be a field, that is, it cannot be a literal string.

The options parameter is optional. You must provide any options as a single string wrapped in quotes, such as "r" or "ur". The available options are the following:

  • r: Applies recursive decoding. For example, %2520 will be decoded twice (recursively) to a space character ( ).
  • u: Enables Unicode percent decoding. The result will be encoded in UTF-8. For example, "%u2601" would be decoded to a cloud emoji (☁️) encoded in UTF-8 ("\xe2\x98\x81", with a size of 3 bytes).

Examples:

url_decode("John%20Doe") will return "John Doe"
url_decode("John+Doe") will return "John Doe"
url_decode("%2520") will return "%20"
url_decode("%2520", "r") will return " "
// Using url_decode() with the any() function:
any(url_decode(http.request.body.form.values[*])[*] contains "an xss attack")

uuidv4

uuidv4(Bytes)String

Generates a random UUIDv4 (Universally Unique Identifier, version 4) based on the given argument (a source of randomness). To obtain an array of random bytes, use the cf.random_seed field.

For example, uuidv4(cf.random_seed) will return a UUIDv4 similar to 49887398-6bcf-485f-8899-f15dbef4d1d5.

wildcard_replace

wildcard_replace(source (Bytes), wildcard_pattern (Bytes), replacement (Bytes) [, flags (Bytes)])String

Replaces a source string, matched by a literal with zero or more * wildcard metacharacters, with a replacement string, returning the result. The replacement string can contain references to wildcard capture groups (for example, ${1} and ${2}), up to eight replacement references.

If there is no match, the function will return source unchanged.

The source parameter must be a field (it cannot be a literal string). Additionally, the entire source value must match the wildcard_pattern parameter (it cannot match only part of the field value).

To enter a literal * character in the wildcard_pattern parameter, you must escape it using \*. Additionally, you must also escape \ using \\. Two unescaped * characters in a row (**) in this parameter are considered invalid and cannot be used. If you need to perform character escaping, it is recommended that you use the raw string syntax for the wildcard_pattern parameter.

To enter a literal $ character in the replacement parameter, you must escape it using $$.

To perform case-sensitive wildcard matching, set the flags parameter to "s".

This function uses lazy matching, that is, it tries to match each * metacharacter with the shortest possible string.

Examples:

  • If the full URI is https://apps.example.com/calendar/admin?expand=true,
    wildcard_replace(http.request.full_uri, "https://*.example.com/*/*", "https://example.com/${1}/${2}/${3}") will return https://example.com/apps/calendar/admin?expand=true

  • If the full URI is https://example.com/applications/app1,
    wildcard_replace(http.request.full_uri, "/applications/*", "/apps/${1}") will return https://example.com/applications/app1 (unchanged value, since there is no match for the full URI value; you should use the http.request.uri.path field for URI path matching).

  • If the URI path is /calendar,
    wildcard_replace(http.request.uri.path, "/*", "/apps/${1}") will return /apps/calendar.

  • If the URI path is /Apps/calendar,
    wildcard_replace(http.request.uri.path, "/apps/*", "/${1}") will return /calendar (case-insensitive match by default).

  • If the URI path is /Apps/calendar,
    wildcard_replace(http.request.uri.path, "/apps/*", "/${1}", "s") will return /Apps/calendar (unchanged value) because there is no case-sensitive match.

  • If the URI path is /apps/calendar/login,
    wildcard_replace(http.request.uri.path, "/apps/*/login", "/${1}/login") will return /calendar/login.

For more examples of wildcard matching, refer to Wildcard matching.

Magic Firewall Functions

bit_slice

bit_slice(protocol (String), offset_start (Number), offset_end (Number))Number

This function looks for matches on a given slice of bits.

The offset starts on the given protocol header. For example, to match on the first bit of payload for a UDP packet, you must set offset_start to 64.

This is primarily intended for use with ip, udp, and tcp.

The slice (offset_endoffset_start) cannot be longer than 32 bits, but multiple calls can be joined together by using logical expressions.

The bit_slice offset cannot exceed 2,040 bits.

HMAC validation

Overview

You can validate hash-based message authentication code (HMAC) tokens in a rule expression by using the is_timed_hmac_valid_v0() function, which has this signature:

is_timed_hmac_valid_v0(
<String literal as Key>,
<String field as MessageMAC>,
<Integer literal as ttl>,
<Integer as currentTimeStamp>,
<Optional Integer literal as lengthOfSeparator, default: 0>,
<Optional String literal as flags>
) -> <Bool as result>

The is_timed_hmac_valid_v0() function has these parameter definitions:

  • Key (String literal)

    • Specifies the secret cryptographic key for validating the HMAC.
  • MessageMAC (String)

    • Contains a concatenation of these HMAC elements: message, separator, timestamp, mac. For a definition and an example, refer to MessageMAC.
  • ttl (Integer literal)

    • Defines the time-to-live for the HMAC token, expressed in seconds. Determines how long the token is valid, relative to the time it was issued.
  • currentTimeStamp (Integer)

    • Represents the UNIX timestamp when Cloudflare received the request, expressed in seconds. Pass the http.request.timestamp.sec field as an approximate value to this argument.
  • lengthOfSeparator (Integer literal, optional)

    • Specifies the length of the separator between the timestamp and the message in the MessageMAC. Expressed in bytes, with a default value of 0.
  • flags (String literal, optional)

    • When you set this optional argument to 's', the function expects the value of the Base64-encoded mac in the MessageMAC argument to use the URL-safe character set with no padding.

    • When you do not set the value of flags to 's', you must URL encode the Base64 value for mac in the MessageMAC argument.

Usage

The is_timed_hmac_valid_v0() function uses the supplied Key to generate a message authentication code (MAC) from the message and the timestamp regions of the MessageMAC. When the generated MAC matches the mac region of the MessageMAC and the token has not expired, the HMAC is valid and the function returns true.

For example, the following expression matches requests to downloads.example.com that do not include valid HMAC tokens:

http.host == "downloads.example.com"
and not is_timed_hmac_valid_v0("mysecretkey", http.request.uri, 100000, http.request.timestamp.sec, 8)

For examples of rules that use HMAC validation, refer to Configure token authentication in the WAF documentation.

MessageMAC

A valid MessageMAC satisfies the following regular expression:

(.+)(.*)(\d{10})-(.{43,})

and is composed of these parentheses-delimited expressions:

ExpressionDescriptionExample
(.+)The message to validate./download/cat.jpg
(.*)The separator between message and timestamp, commonly a parameter name.&verify=
(\d{10})The 10-digit UNIX timestamp when the MAC was issued, expressed in seconds.1484063137
(.{43,})A Base64-encoded version of the mac. When you do not set the value of the urlSafe argument in the HMAC validation function to 's', you must URL-encode the Base64 value for mac. When the Base64 MAC encoding is URL-safe, the mac value contains 43 bytes. Otherwise, the value will be 44 bytes or more, because of URL encoding.IaLGSmELTvlhfd0ItdN6PhhHTFhzx73EX8uy%2FcSDiIU%3D

For details on generating a MessageMAC, refer to HMAC token generation.

HMAC validation examples

MessageMAC in a single field

Consider the case where the MessageMAC is contained entirely within a single field, as in this example URI path:

/download/cat.jpg?verify=1484063787-IaLGSmELTvlhfd0ItdN6PhhHTFhzx73EX8uy%2FcSDiIU%3D

Note how the URI maps to the elements of the MessageMAC:

ElementValue
message/download/cat.jpg
separator?verify= (with length 8)
timestamp1484063787
macIaLGSmELTvlhfd0ItdN6PhhHTFhzx73EX8uy%2FcSDiIU%3D

When the MessageMAC is contained entirely within a single field such as http.request.uri, pass the field name to the MessageMAC argument of the HMAC validation function:

is_timed_hmac_valid_v0(
"mysecretkey",
http.request.uri,
100000,
http.request.timestamp.sec,
8
)

Concatenated MessageMAC argument

To compose a MessageMAC from more than one field, use the concat() function.

This example constructs the value of the MessageMAC argument by concatenating the request URI and two header fields:

is_timed_hmac_valid_v0(
"mysecretkey",
concat(
http.request.uri,
http.request.headers["timestamp"][0],
"-",
http.request.headers["mac"][0]),
100000,
http.request.timestamp.sec,
0
)