Expressions

                                                                                                                                                                                                                                                                                                                                      Back to [Book]     [Topic]

Expressions

Expressions allow you to dynamically construct SAF parameters and to select which SAFs to execute on a request-by-request basis. Expressions are constructed from literals, variables, functions, and operators. Use expressions in If and ElseIf tags (see If, ElseIf, and Else), in log format strings (see Appendix C: Using the Custom Log File Format), and SAF parameters (see String Interpolation).

This section contains the following sections:

• Expression Syntax
• Expression Results as Booleans
• Expression Literals
• Expression Variables
• Expression Operators
• Expression Functions
• Regular Expressions

Expression Syntax

The expression syntax is similar to the syntax used in Perl. Expressions are constructed from literals, variables, functions, and operators.

The following example shows an expression used in an If tag:

<If not $internal
    and $uri =~ "^/private/(.*)$" 
    and $referer !~ "^https?://example.com/"> 
NameTrans fn="redirect" 
          url="http://example.com/denied.jsp?file=$1" 
</If> 

This example expression checks to see if a request meets certain criteria, for example if it is an internal request. If it does not meet the criteria, the server redirects the request to a request denied URL.

The expression contains the following components:

Expression Results as Booleans

In some circumstances, for example, after evaluating an If or ElseIf expression, the server must treat the result of an expression as a Boolean. The server uses the following rules when converting a numeric value to a Boolean:

• Numeric 0 (zero) evaluates to false
• All other numeric values evaluate to true

The server uses the following rules when converting a string to a Boolean:

• Zero-length strings evaluate to false
• The string 0 (zero) evaluates to false
• All other strings evaluate to true

Expression Literals

Expression literals are divided into string and numeric literals.

String Literals

A string literal is bracketed by either single quotes (') or double quotes ("). When single quotes bracket a string literal, the value of the literal is the value within the quotes. When double quotes are used, any references to variables or expressions within the quotes are interpolated. For more information, see String Interpolation.

The following expression examples show the use of single and double quotes.

# This expression evaluates to true
('foo' eq "foo")

# This expression evaluates to false
('foo' eq "bar")

# This expression evaluates to true
('foo' eq "f$(lc('O'))o")

# This expression may evaluate to true or false,
# depending on the value of the variable $foo
('$foo' eq "$foo")

To include an uninterpolated $ character in a double quote string, use the $$ or \$ escape sequences.

When a double quote character appears within a literal bracketed by double quotes, it must be prefixed with a backslash. When a single backslash ( \ ) character appears within a literal bracketed by double quotes, it must be prefixed with a backslash. When a single quote character appears within a literal bracketed by single quotes, it must be prefixed with a backslash.

The following examples show valid and invalid literals:

# The following are examples of valid literals
'this string literal is bracketed by single quotes'
"this string literal is bracketed by double quotes"
"the backslash, \\, escapes characters in double quote string literals"
'it\'s easy to use strings literals'

# The following are examples of invalid literals
'it's important to escape quote characters'
"any \ characters in double quote string literals must be escaped"

Numeric Literals

A numeric literal can consist of decimal digits and an optional decimal point, a leading zero followed by octal digits, or a leading 0x prefix followed by hexadecimal digits. Hexadecimal and octal numbers are automatically converted to decimal form.

The following examples show expressions that use numeric literals:

# The following expressions evaluate to true
(1 < 2)
(0x10 == "16")
(1 == 1.00)

# The following expressions evaluate to false
(1 > 2)
("0x10" == 16)
(1 != 1.00)

Expression Variables

Any $variable can be used as a variable in an expression. To mirror the Client tag syntax, the $ prefix is optional for predefined variable names in expressions. For example, the following three portions of obj.conf are all semantically equivalent:

<If $uri = "*.html">
...
</If>

<If uri = "*.html">
...
</If>

<Client uri = "*.html">
...
</Client>

Any variable names you define must use the $ prefix. For example, the following expression is invalid even if somecustomvariable is defined in a server.xml variable element:

<If somecustomvariable = "foo">
...
</If>

To make this expression valid, add the dollar sign prefix:

<If $somecustomvariable = "foo">
...
</If>

Expression Operators

The following table lists the operators that are used in expressions.

Table A-2 List of Expression Operators

||

The following table lists the precedence of operators within expressions from highest to lowest precedence.

Table A-3 Operator Precedence

=, =~, !~

||

The numeric operators (<, <=, >, >=, ==, and !=) are intended to operate on numbers and not strings. To facilitate comparing numbers, dates, and timestamps, the numeric operators ignore any white space, colons, slashes, and commas in their arguments. Dashes after the first digit are also ignored.

Note - It is generally incorrect to use the numeric operators on non-numeric values.

For example, the following expression evaluates to true:

# The following expression evaluates to true because both
# "foo" and "bar" are numerically equivalent to 0
("foo" == "bar")

Expression Functions

Expression functions manipulate data for use in expressions. Expression functions are different from SAFs. While SAFs perform the actual work associated with an HTTP request, expression functions are used to select which SAFs run and what parameters to pass to the SAFs.

Some expression functions require one or more arguments. An expression function's argument list is bracketed by parentheses (()) and the individual arguments are separated by commas (,).

The individual expression functions are listed in the following sections:

• atime
• choose
• ctime
• escape
• external
• httpdate
• lc
• length
• lookup
• mtime
• owner
• uc
• unescape
• uuid

atime

The atime function returns the time of the last access for the specified file or directory.

Syntax

atime(path)

Arguments

The following table describes the argument for the expression function.

Table A-4 atime Argument

See Also

• ctime
• mtime

choose

The choose function parses pipe-separated values from values and returns one at random.

Syntax

choose(values)

Arguments

The following table describes the argument for the expression function.

Table A-5 choose Argument

(|)

Example

The following obj.conf code demonstrates the use of choose to randomly select one of three images:

NameTrans fn="rewrite"
          from="/images/random"
          path="/images/$(choose('iwsvi.jpg|0061.jpg|webservervii.jpg'))"

ctime

The ctime function returns the time of the last status change for the specified file or directory.

Syntax

ctime(path)

Arguments

The following table describes the argument for the expression function.

Table A-6 ctime Argument

See Also

• atime
• mtime

escape

The escape function encodes the URI using util_uri_escape, converting special octets to their %-encoded equivalents, and returns the result.

Syntax

escape(uri)

Arguments

The following table describes the argument for the expression function.

Table A-7 escape Argument

See Also

unescape

external

The external function passes a value to an external rewriting program and returns the result.

Each invocation of external results in a single newline-terminated line being written to the external rewriting program's stdin. For each line of input, the program must produce a single line of output. When developing an external rewriting program, it is important to avoid buffering stdout. In Perl, for example, $| = 1; should be used to disable buffering. Because the server expects the external rewriting program to produce one line of output for each line of input, the server can hang if the external rewriting program buffers its output.

Syntax

external(program, value)

Arguments

The expression function has the following arguments.

Table A-8 external Arguments

Argument	Description
program	The program argument is the file name of an external rewriting program. Because program is executed using the operating system's default shell (/bin/sh on Unix/Linux) or the command interpreter (CMD.EXE on Windows), program should be an absolute path or the name of a program in the operating system's PATH. The server starts the external rewriting program on demand. A given server process never executes more than one instance of the program at a time. Note - The server may start multiple instances of a given external rewriting program when the server is running in multiprocess mode.
value	The value passed to the rewrite program.

Example

The following is an example of an external rewriting program rewrite.pl, used to change the prefix /home/ to /u/:

#!/usr/bin/perl
$| = 1;
while (<STDIN>) {
    s|^/home/|/u/|;
    print $_;
}

In this example, the external expression function used to invoke rewrite.pl is as follows:

NameTrans fn="rewrite" path="$(external('rewrite.pl', $path))"

httpdate

The httpdate function returns an RFC 1123 date/time stamp for use in HTTP header fields such as Expires.

Syntax

httpdate(time)

Arguments

The following table describes the argument for the expression function.

Table A-9 httpdate Argument

Example

The following obj.conf code could be used to set an Expires header that indicates a response is not cached for more than one day:

ObjectType fn="set-variable"
           insert-srvhdrs="$(httpdate($time + 86400))"

lc

The lc function converts all the US ASCII characters in the string to lowercase and returns the result.

Syntax

lc(string)

Arguments

The following table describes the argument for the expression function.

Table A-10 lc Argument

Example

The following obj.conf code can be used to redirect clients who erroneously used uppercase characters in the request URI to the equivalent lowercase URI:

<If code == 404 and not -e path and -e lc(path)>
Error fn="redirect" uri="$(lc($uri))"
</If>

See Also

uc

length

The length function returns the length of its argument, that is, a number representing the length of the string.

Syntax

length(string)

Arguments

The following table describes the argument for the expression function.

Table A-11 length Argument

Example

The following obj.conf code can be used to send a 404 Not found error to clients that request URIs longer than 255 bytes:

<If length($uri) > 255)>
PathCheck fn="deny-existence"
</If>

lookup

The lookup function inspects a text file for a name-value pair with name name and returns the corresponding value. The name-value pairs in the file are separated by white space.

If the file does not contain a name-value pair with the specified name, this function returns the value of defaultvalue, if specified, or returns an empty string.

Syntax

lookup(filename, name, defaultvalue)

Arguments

The expression function has the following arguments:

Table A-12 lookup Arguments

Example

The following example shows a text file called urimap.conf that could be used with the lookup function to map shortcut URIs to URIs:

# This file contains URI mappings for Web Server.
# Lines beginning with # are treated as comments.
# All other lines consist of a shortcut URI, whitespace, and canonical URI.
/webserver /software/products/web_srvr/home_web_srvr.html
/solaris   /software/solaris/
/java      /software/java/

Using the sample text file above, you could use the following lookup expression to implement shortcut URIs for commonly accessed resources:

<If lookup('urimap.conf', uri)>
NameTrans fn="redirect" url="$(lookup('urimap.conf', uri))"
</If>

mtime

The mtime function returns the time of the last data modification for the specified file or directory.

Syntax

mtime(path)

Arguments

The following table describes the argument for the expression function.

Table A-13 mtime Argument

See Also

• atime
• ctime

owner

The owner function returns the owner of a file.

Syntax

owner(path)

Arguments

The following table describes the argument for the expression function.

Table A-14 owner Argument

uc

The uc function converts all the US ASCII characters in string to uppercase and returns the result.

Syntax

uc(string)

Arguments

The following table describes the argument for the expression function.

Table A-15 uc Argument

See Also

lc

unescape

The unescape function decodes the URI using util_uri_unescape, converting %-encoded octets to their unencoded form, and returns the result.

Syntax

unescape(uri)

Arguments

The following table describes the argument for the expression function.

Table A-16 unescape Argument

See Also

escape

uuid

The uuid function returns a UUID as a string. No two calls to uuid return the same UUID. Because they are guaranteed to be unique, UUIDs are useful for constructing client-specific cookie values.

Syntax

uuid()

Regular Expressions

The If and ElseIf expressions may evaluate regular expressions using the =~ and !~ regular expression matching operators. These regular expressions use the Perl-compatible syntax implemented by Perl-compatible Regular Expressions (PCRE).

By default, regular expressions are case sensitive. The (?i) option flag can be added to the beginning of a regular expression to request case insensitivity. For example:

$uri =~ '^/[Ff][Ii][Ll][Ee][Nn][Aa][Mm][Ee]$'

$uri =~ '(?i)^/filename$'

When an If or ElseIf expression contains a regular expression, regular expression backreferences can appear within arguments in the container body. Regular expression backreferences are of the form $n where n is an integer between 1 and 9 corresponding to the capturing subpattern of the regular expression.

For example:

<If $path =~ '^(.*)(\.html|\.htm)$'>
NameTrans fn="rewrite" path="$1.shtml"
</If>

In the above example, two subpatterns are used in the If expression, so $1 and $2 can be used as backreferences. In the example, the value of the first capturing subpattern is used within a NameTrans fn="rewrite" parameter. The value of the second capturing subpattern is ignored.

An If or ElseIf expression can contain backreferences to earlier regular expressions in that same If or ElseIf expression.

For example:

<If "foo" =~ "(.*)" and $1 eq "foo">
# Any contained directives will be executed
# since $1 will evaluate to "foo"
...
</If>

The contents of the above If expression are executed, because the given If expression always evaluates to true.

However, If and Elseif expressions, and contained directives, can't contain backreferences to regular expressions in parent containers. For example, the following obj.conf entry is invalid:

<If $path =~ '(.*)\.css'>
<If $browser = "*MSIE*">
# This example is invalid as $1 is not defined
AuthTrans fn="rewrite" path="$1-msie.css"
</If>
</If>

You can use $& to obtain the value that last successfully matched a regular expression. Use the following obj.conf entry to redirect requests for HTML files to another server:

<If $path =~ '\.html$' or $path =~ '\.htm$' >
NameTrans fn="redirect" url="http://docs.example.com$&amp;"
</If>