| DACS.EXPRS(5) | DACS Formats Manual | DACS.EXPRS(5) |
NAME¶
dacs.exprs - DACS expression languageDESCRIPTION¶
These files are part of the DACS suite. DPL (the DACS programming language) is used in access control rules, its revocation list, and in configuration files. This programmability gives DACS maximum run-time configurability and flexibility. A DPL expression - or even a small program - may appear within predicate, allow, and deny elements of an access control rule, for example. DPL is also accessible using the dacsexpr(1)[1] command, which can be used for writing scripts even for non- DACS applications. DPL, which is gradually evolving in mostly backward-compatible ways, is similar in many ways to Perl[2], PHP[3], Tcl[4] and its expressions look and behave much like C/C++ expressions. The calling signatures for functions are reminiscent of those of Tcl, with literal or string arguments used to select a particular mode of operation or specify options. The syntaxes used for strings and variables have been influenced by various Unix shells. Our intent is for the language to feel familiar and be easy to use for the typical tasks at hand. We have tried not to be gratuitously different.Expression Syntax¶
Expression evaluation consists of a lexical analysis stage, in which the expression is broken into a sequence of tokens, followed by evaluation of the tokens. Expression syntax is checked before an expression is evaluated. Any syntactic or run-time evaluation error immediately terminates evaluation of the top-level expression and returns a False result.
Three comment styles are recognized:
Additionally, when expressions are parsed in the context of an XML document
(such as in an access control rule), the XML comment syntax can be used
(<!-- A comment -->). Such comments can span multiple lines.
•The /* ... */ C style comment syntax,
which does not nest;
•The // syntax of C++, where the
remainder of the line following the token is ignored; and
•The # syntax of shells and many
scripting languages, provided the # is either at the beginning of a line or
appears after whitespace, where the remainder of the line following the token
is ignored. Note that escaping the # by preceding it with a backslash prevents
the text that follows from being interpreted as a comment. For example, this
will result in a syntax error if the backslash is omitted:
Here are examples of all three styles:
> ${foo:? \#xxx}
" #xxx"
/*
* This is a comment
*/
// This is another comment
${x} = 17; # And one last comment
<!-- Comment out this clause for now... <Auth id="authx"> STYLE "expr" CONTROL "sufficient" </Auth> -->
Basic Data Types¶
The following basic data types are supported: integer, Integers are represented internally as a
C/C++ long int. Maximum and minimum values are platform dependent.
Integers are written in the C-style syntax; for example, -1958, 0377 (octal),
and 0xABC (hexadecimal, upper or lower case).
real
, Reals are represented internally as a C/C++
double. Maximum and minimum values are platform dependent. A real
constant is an optional sequence of decimal digits (possibly signed) followed
by a period and 1) at least one digit or 2) an 'e' or 'E' followed by at least
one digit.
string
A string is a sequence of characters enclosed
between matching single or double quotes (e.g., 'Hello world'). Interpolation
of variables occurs within double quotes but not single quotes. C-style
character escape codes and octal numeric escape codes are understood (e.g.,
"\t", "\010") and either quote character (e.g., 'It\'s
here') and the backslash character (e.g., "\\") can be quoted. An
unrecognized quoted character is mapped to that character (e.g.,
"\x" is "x"). Character strings are limited in length by
available memory and are represented internally as a null-terminated vector.
Note
Variable references may occur within a (double-quoted) string; the value of the
variable reference is interpolated at that point. If ${Foo::bar} is
"hello", then the value of "${Foo::bar}, world" is
"hello, world".
Note
The first expression is invalid and must be written as the second:
binary
•Because a string is null-terminated, it
cannot contain a NUL character. Also, functions that deal with strings usually
do not expect (most) ASCII control characters to appear in a string. Therefore
a string that contains an unprintable character (a character that is not a
tab, newline, carriage return, and that does not satisfy isprint(3)[5])
automatically becomes a bstring (see below).
•Because DACS configuration files
are XML documents, characters special to XML must be properly escaped within
them. In particular, an ampersand character must always be written as
& and a < character must be written as <. For example, the
query string a=1&b=2 might be used as
${Foo::QUERY_STRING} = "a=1&b=2"
foo"baz" foo."baz"
, A binary string is a sequence of bytes,
limited in length by available memory. Most language operators cannot be
applied to data of this type without converting it to another type (e.g., two
bstring values cannot be added using the + operator). A binary string is not
necessarily portable across systems.
bareword
> "\0\1\2" "000102"
This type is a "literal word" much
like Perl's barewords. A bareword consists of an initial alphabetic
character, followed by any number of alphanumerics and underscores. The
resulting lexical token must have no other interpretation in the language and
is treated as if it were a quoted string. This syntactic convenience makes
these two function calls equivalent:
These two expressions are equivalent and yield "foobaz":
bool
file(test, "-e", foo)
file("test", "-e", "foo")
foo."baz" foo.baz
The boolean values True and
False are either the result of evaluating certain expressions or are
implicit argument values. This is really a pseudo-type because it is
represented internally as an integer. In the former case, the integer 1
is the canonical "true" value and 0 is considered
"false". In the latter case, there are several possibilities. If the
argument is an integer or real, any non-zero value is considered True
and 0 is considered False. For the string data type, both the
empty string (i.e., "") and the string "0" are considered
False and anything else is considered True. A binary string is
equivalent to False if and only if its length is zero. An empty list of
either variety ("[]" or "{}") is False, while any
non-empty list or alist is True.
Automatic type conversion is performed when necessary and possible. In general,
a "lower" type is promoted to a "higher" type (e.g., an
integer is converted to a real when it is added to a real) and the result is
of the higher type. Arguments to function calls are automatically coerced to
the required types. A printable binary string (one not containing any
"troublesome" control characters) can be converted into a string
without loss; other binary strings are converted into a hexadecimal string
representation for assignment or display.
The C/C++ unary cast operation is available for explicit type conversion. Not
all conversions are supported (e.g., integer to binary and binary to string).
These type names are case sensitive.
The language includes the concept of the void type, which cannot be stored in a
variable, used as an operand, or printed. Some functions are void,
print()[6] for example. A value can be cast[7] to void.
Variables and Namespaces¶
Every variable exists within a namespace. Namespaces exist so that the same variable name can exist safely and without ambiguity in different contexts. They also serve to group together and name a set of closely related variables, and they make it easy for all variables in the set to be assigned a characteristic (such as being read-only). For example, CGI parameter values are automatically put in the Args namespace and variables automatically created by DACS are put in the DACS namespace. Namespaces address the problem of a parameter name that happens to have the same name as a variable created by DACS, for example. They also allow intermediate results to be stored in their own namespace, also avoiding the problem of clashing variable names. Variables are not declared in advance. The value of an uninitialized variable is the empty string, which is invalid in a numerical context, but variables should always be initialized before being used. Some variables are created automatically by DACS from the execution context (e.g., the value of a CGI parameter value, the identity of the client, an environment variable), as a side-effect of function evaluation, or by an assignment operator. The interpreter tries to maintain the natural type of a variable when possible, to avoid conversions to and from the string type.
A variable reference may have either of the following syntaxes:
For instance, the following refers to the value of a variable called
JURISDICTION_NAME within the namespace called Conf:
A variable called JURISDICTION_NAME within a different namespace could
exist and would be completely distinct.
A namespace must begin with an alphabetic character and can be followed by any
number of alphabetics, digits, dashes, and underscores. By convention,
predefined namespaces begin with an upper case letter.
If the namespace is omitted from a variable reference, a default namespace is
implied (see below).
A variable name consists of any number of alphanumeric characters (upper and
lower case), and characters from this set:
Additionally, a "%" character that is followed by two hexadecimal
characters (upper and lower case) is acceptable.
If instead of a variable name the character "#" appears, the number of
variables in the namespace is returned. If the namespace does not exist, 0 is
returned. For example, the value of this variable reference is the number of
variables in the Conf namespace:
When the syntax with braces is used, a variable name may be followed by a colon
and then one or more modifier flags that affect the processing of the
variable. Referencing an invalid variable name or unknown namespace, or using
an undefined modifier flag is an error. Referencing an undefined variable
yields the empty string.
Variable names are case sensitive by default; namespaces are always case
sensitive.
User-defined variables and namespaces are not persistent. They disappear when
their execution context terminates.
Tip
A variable reference may not contain any whitespace except when it appears after
a ? or + modifier flag.
Tip
Because many variable references do not include flags or use punctuation
characters in the variable name, as a convenience the braces that surround a
variable reference may be omitted in certain cases. This is only possible if
the variable name begins with an alphabetic or an underscore, which can be
followed by alphanumerics and underscores. A namespace may be specified, but
flags are not permitted, although the special "#" construct is also
allowed. The variable name ends with the first invalid character. For example,
these pairs of variable references are equivalent:
Note that the variable reference ${foo-17} has a valid but different
interpretation if the braces are omitted.
${[ namespace::]variable-name[:flags]}
$[ namespace::]variable-name
${Conf::JURISDICTION_NAME}
-_.!~*'()
${Conf::#}
${myvar}
$myvar
${foo::baz}
$foo::baz
A variable reference may include one or more modifier flags that control how the
reference is to be interpreted.
The following modifier flags are recognized:
e
In the first example, if ${Args::SCALE} is undefined or empty, the value
of the variable reference is "17" instead of the value of
${Args::SCALE}. In the second example, if ${Foo::bar} (case
insensitive) is defined, the result is its value, otherwise the result is the
value of the string "${Bar::baz}baz". In the third example, if
${DACS::QUERY} is defined and not empty, the value of the expression
will be a question mark followed by the value of ${DACS::QUERY}. If
${DACS::QUERY} is undefined or empty, the value will be the empty
string.
Exists: The "e" modifier flag
is used to test whether the variable exists (has been defined). Instead of
returning the value of the variable or causing an error, the value of the
variable reference is the string "1" if the variable is defined, the
empty string otherwise (equivalent to False).
i
Insensitive: When looking up the name
of a variable, the default is to use a case-sensitive comparison for the
variable name. To use a case-insensitive comparison instead, an "i"
flag is used (e.g., ${FOO::i}). The namespace lookup is always case
sensitive.
n
Non-empty: The "n" modifier
flag tests whether the variable exists (has been defined) and is not
the empty string (i.e., has zero length). Instead of returning the value of
the variable or causing an error, the value of the variable reference is the
string "1" if the variable is defined and is not the empty string,
otherwise it is the empty string (equivalent to False).
z
Zero: The opposite of the "n"
flag, instead of returning the value of the variable or causing an error, the
value of the variable reference is the string "1" if the variable is
undefined or the empty string, otherwise it is the empty string (equivalent to
False).
?
Default: The "?" modifier
flag must appear last if it is used. The flag is immediately followed by zero
or more characters. Its purpose is to associate a default value with the
variable reference. If the variable is defined and is not empty, then the
result of the variable reference is the value of the variable; otherwise, the
result is the evaluation of the characters that follow the "?" flag.
If no character follows the "?" flag, the empty string is indicated.
The default may itself contain variable references, embedded spaces, etc., and
is evaluated left to right. Any "}" characters appearing in the
string before the last closing brace must be escaped by being preceded by a
backslash.
+
Substitute: The "+" modifier
flag must appear last if it is used. The flag is immediately followed by zero
or more characters. Its purpose is to associate a substitute value with a
variable reference. If the variable is defined and is not the empty string,
then the result of the variable reference is the evaluation of the characters
that follow the "+" flag; if the variable is undefined or is the
empty string, the value of the variable reference is the empty string. If no
character follows the "+" flag, the empty string is indicated. The
substitute may itself contain variable references, embedded spaces, etc., and
is evaluated left to right. Any "}" characters appearing in the
string before the last closing brace must be escaped by being preceded by a
backslash.
The i can be combined with any other flag, but it must appear first. All other
flags are mutually exclusive. Repetitions of a flag are ignored. An
unrecognized flag raises an error condition.
Consider these examples:
${Args::SCALE:?17}
${Foo::bar:i?${Bar::baz\}baz}
"${DACS::QUERY:+?}${DACS::QUERY:?}"
The following namespaces are predefined by DACS and reserved for
particular uses. Some are read-only, which means that only DACS can
create a variable or change the value of a variable in the namespace, except
in certain contexts.
Args
Instantiated from query string arguments and
the POST data stream (if the content type is application/x-www-form-urlencoded
or multipart/form-data). This namespace is read-only.
Argv
Instantiated by dacsexpr from the
command line flags passed to the script. The value of ${Argv::0} is the
name of the file being processed, with - signifying the standard input. The
next argument, if any, will be ${Argv::1}, and so on. This namespace is
read-only.
Auth
Used by dacs_authenticate(8)[8] during
authentication processing.
Conf
Instantiated with configuration directive
variables, this namespace is made read-only after configuration processing.
See dacs.conf(5)[9].
Cookies
This namespace is instantiated with HTTP
cookies that were submitted with a request. For security reasons, those
associated with DACS credentials are excluded. This is a read-only
namespace. If a cookie named foo is sent by a user agent, an access control
rule can access the cookie value as ${Cookies::foo}.
DACS
Instantiated with DACS-specific
variables. It is read-only. See dacs_acs(8)[10].
Env
For web services, instantiated with the
standard Apache environment variables; for other programs, instantiated
from the execution environment ( environ(7)[11]). It is
read-only.
ExecEnv
Used by exec()[12].
LDAP
Used by local_ldap_authenticate.
Temp
Unless disabled or redefined at build-time,
variable references that do not include a namespace are associated with this
namespace as a convenience. The following three expressions are therefore
equivalent:
In a future release, this mechanism may be generalized to provide a run-time
means of selecting the default namespace.
${foo} = 17
${Temp::foo} = 17
$foo = 17
Lists, Alists, and Arrays¶
DPL supports more complicated data structures based on lists and associative lists. These types may also be combined and composed so that programmers can create lists of lists, and so on.
A list is composed of zero or more basic data types or sub-lists. A list is
created using the following syntax:
A list can also be created through the list()[13] function.
Here is a list consisting of four elements:
The length()[14] function returns the number of elements in a list.
A list can be assigned to a variable:
Note
These two statements are equivalent:
And so are these two:
Modifier flags therefore do not apply to list elements, only the list variable.
A list or element can be appended to another list using the "."
("dot") concatenation operator. List elements can be rotated using
the ">>" ("shift left") or "<<"
("shift right") operators. The compound assignment operator versions
of these operators may also be used.
A list element can be referenced using a subscript between zero and one less
than the number of elements in the list:
It is an error to reference a non-existent list element using a subscript.
(Note: additional syntax may be introduced to provide a way to declare lists
and arrays.)
The values of one or more list elements are selected by a list reference, which
includes the simple subscript case just described. The value of a list
reference is either a basic data type or a list.
An EXP must evaluate to a non-negative integer value. The ".."
("dotdot") range operator specifies a sequence of subscripts between
the value to its left and the value to its right, inclusive. The left value
must not be greater than the right value. If "#" appears to the
right of the ".." operator, the number of elements in the list
variable or the intermediate list computation is implied. A "#" may
not appear to the left of ".." and may not be used in an expression
(e.g., "#-2" is invalid). As in a function's argument list, a comma
is not treated as the comma operator in this context. Note that it is not an
error to specify non-existent elements in a slice; therefore it is possible
for the value of a list reference to be the empty list.
The dotdot operator can also be used to construct an element of a list or alist:
Whether a "[" ... "]" sequence introduces a list constructor
or list reference depends on the context; if it appears to the right of a list
variable, list constructor, a function that returns a list, or another list
reference, it is treated as a list reference.
List references can be composed as a right-associative operation. For example:
Tip
Individual characters and sequences of characters of a string-valued expression
can be selected using strchars()[15], which uses a similar syntax.
Note
LIST -> "[" "]" | "[" LIST-ELS "]" LIST-ELS -> EL | EL "," LIST-ELS EL -> BASIC-DATA-TYPE | LIST
[1, "one", 1.000, ["one sublist"]]
$mylist = [1, 2, 3, 4, 5, 6] $mylist_copy = $mylist
$mylist = ["one", "two"]
${mylist} = ["one", "two"]
$mylist[0]
${mylist}[0]
> $mylist=[orange, apple, grape] [orange,apple,grape] > $mylist . banana [orange,apple,grape,banana] > $mylist .= [prune,plum] [orange,apple,grape,prune,plum] > $mylist .= [[lime]] [orange,apple,grape,prune,plum,[lime]] >$mylist << 1 [apple,grape,banana,prune,plum,[lime],orange]
> $mylist = [1, 2, 3, 4, 5, 6]; length($mylist) 6 > $mylist[0] 1
LIST-REFERENCE -> "[" LIST-REFERENCE-ELS "]" LIST-REFERENCE-ELS -> EMPTY | LIST-REFERENCE-EL | LIST-REFERENCE-EL "," LIST-REFERENCE-ELS LIST-REFERENCE-EL -> EXP | LIST-REFERENCE-SLICE LIST-REFERENCE-SLICE -> EXP ".." EXP LIST-REFERENCE-SEQ -> LIST-REFERENCE | LIST-REFERENCE LIST-REFERENCE-SEQ
> $i=1, $mylist[$i] 2 > $mylist[1,3,5] [2,4,6] > $mylist[0..2,4] [1,2,3,5] > $mylist[2..#] [3,4,5,6] > $mylist[0..3] [1,2,3,4]
> $a = [1, 4..8, 10, 12, 13] [1,4..8,10,12,13] > length($a) 5 > $b = [0..2,4]; listref($a, $b) [1,4..8,10,13]
> $a = [[1,2,3], [4,5,6], [7,8,9]] [[1,2,3], [4,5,6], [7,8,9]] > $a[1][1] 5 > $a[0..1][1..2] [[4,5,6]] > $a[0..1][1..2][0][2] 6
•The list constructor and list reference
syntax has not yet been integrated with the expression
grammar[16].
•A list value can also be assigned to a
subscripted variable; only a single subscript is allowed, however, and the
referenced element must already exist:
> $a = [1, 2, 3] [1,2,3] > $a[2] = 17 17 > $i = 1 1 > $a[$i] = [10, 11] [10,11] > $a [1,[10,11],17]
DPL's associative list, or "alist", is similar to Perl's
hashes. An alist is composed of zero or more pairs. The first element of each
pair is a case-sensitive key, unique within the alist, that is used to index
the element. The second element of a pair is its value, which may be any data
type. The key element of a pair, or all the keys in an alist, can be obtained
using keysof()[17]. Similarly, valuesof()[18] yields the value
element or a list of value elements.
Unlike a regular list, elements within an alist are not ordered. Two alists can
only be compared for equality (or inequality); they are equal if they contain
exactly the same pairs.
An alist has the following syntax:
An alist can also be created through the alist()[19] function.
Here is an alist consisting of four elements:
The length()[14] function returns the number of pairs of elements in an
alist.
An alist can be assigned to a variable:
An alist can be appended to another alist using the "."
("dot") concatenation operator. The compound assignment operator
version of this operator may also be used.
An alist element or pair is referenced using a string subscript. A sequence of
string subscripts can be used to select multiple pairs. If the subscript (or
subscripts) are within brackets, then a successful result will be a basic data
type or a list. If the subscript (or subscripts) are within braces, then a
successful result will always be an alist. Note that because an alist
subscript is not automatically converted to the string type, a numeric
subscript is illegal.
It is an error to reference a non-existent alist element. (Note: additional
syntax may be introduced to provide a way to declare lists and arrays.)
Like regular lists, alist references can be composed as a right-associative
operation:
It is possible to convert an alist to a regular list, or vice versa; see the
cast[7] operator.
Note
There is currently no way to delete an alist pair.
ALIST -> "{" "}" | "{" ALIST-PAIRS "}"
ALIST-PAIRS -> ALIST-PAIR | ALIST-PAIRS "," ALIST-PAIR
ALIST-PAIR -> KEY-EL "," VALUE-EL
KEY-EL -> STRING
VALUE-EL -> BASIC-DATA-TYPE | LIST | ALIST
{"red", 0, "blue", 2, "green", 5, "black", 7}
$myalist = {1, 2, 3, 4, 5, 6}
$myalist_copy = $myalist
> $myalist={sunny, 3}
{"sunny", 3}
> $myalist . {rainy, 11}
{"sunny", 3, "rainy", 11}
> $myalist .= {"snowy", 13}
{"sunny", 3, "snowy", 13}
> $myalist = {a, 2, b, 4, c, 6}; length($myalist)
3
> $myalist["a"]
2
> $myalist{"b"}
{"b", 4}
> $myalist{"c", "a"}
{"c", 6, "a", 2}
> $myalist = {a, [1, 2], b, [3, 4], c, [5, 6]}; length($myalist)
3
> $myalist["a"]
[1, 2]
> $myalist{"b"}
{"b", [3, 4]}
> $myalist{"b"}[1]
4
Expression Grammar¶
The following grammar is used to construct an expression (EXP) or sequence (S) of expressions.S -> E | E ";" | E ";" S
E -> E2 | E2 "," E
E2 -> E3 | VAR ASSIGN_OP E2 | IF_ELSEIF_ELSE
E3 -> E4 | E4 "?" E ":" E
E4 -> E5 | E5 OR E5
E5 -> E6 | E6 AND E5
E6 -> E7 | E7 "|" E7
E7 -> E8 | E8 "^" E8
E8 -> E9 | E9 "&" E9
E9 -> E10 | E10 EQ_OP E10
E10 -> E11 | E11 REL_OP E11
E11 -> E12 | E12 "." E12
E12 -> E13 | E13 "<<" E13 | E13 ">>" E13
E13 -> E14 | E14 "+" E14 | E14 "-" E14
E14 -> E15 | E15 "*" E15 | E15 "/" E15 | E15 "%" E15
E15 -> E16 | E16 "^" E14 | E16 "**" E14
E16 -> E17 | NOT E16 | "~" E16 | "++" VAR | "--" VAR
| "+" E | "-" E | "(" type ")"
E17 -> "(" E ")" | VAR "++" | VAR "--" | FUNCTION_CALL | PRIMARY
ASSIGN_OP -> "=" | "+=" | "-=" | "*=" | "/=" | "%=" | ">>="
| "<<=" | "&=" | "^=" | "|=" | ".="
PRIMARY -> a number | a string | VAR
OR -> "||" | "or"
AND -> "&&" | "and"
NOT -> "!" | "not"
EQ_OP -> "==" | "!=" | "eq" | "ne"
REL_OP -> "<" | "<=" | ">" | ">=" | "lt" | "le" | "gt" | "ge"
VAR -> a variable reference
FUNCTION_CALL -> FUNCTION_NAME "(" ARG_LIST ")"
ARG_LIST -> EMPTY | E2 | ARG_LIST "," E2
EMPTY ->
${a} += 17
if ( expression) {
sequence
}
[elseif ( expression) {
sequence
}] ...
[else {
sequence
}]
${a} = if (${b} > 8) {${b}++; 33;} else {${b} - 1}
"hello, " . (if (0) {b . y . e} else {"world"})
Operators¶
The operators that appear in the grammar have the following semantics. They are listed in order of increasing precedence (which is very close to ISO C's), with operators in the same section having equal precedence. The result of applying an operator is one of the supported data types[20], or an error. Parentheses can be applied to subexpressions in the usual way. Whenever it makes sense, intermediate values are automatically converted to an appropriate type by an operator. So, for example, adding an integer and a real will cause the integer to automatically be converted to a real, yielding a real value. Adding a string and a number will work only if the string can be successfully converted to a number. In situations where an integer is required, a real value (including a string that represents a valid real number) will be truncated to an integer. For logical comparison operators, the operands will both be converted to integers, reals, or strings as necessary. A string value that is an illegal number will always be treated as a string.id="note11" xreflabel="dacsexpr prompts".PP In the examples that follow, the '>' character at the beginning of an input line is a prompt from dacsexpr(1)[1].
This is the C/C++ comma operator. A pair of
expressions separated by a comma is evaluated left to right, and the type and
value of the result are the type and value of the right operand.
=, +=, -=, *=, /=, %=, >>=, <<=, &=, ^=, |=, .=
Assignment is done using a simple or compound
assignment operator, each of which has right to left associativity. In the
case of a compound assignment operator, the left hand side is evaluated only
once. The type and value of an assignment is that of its right hand side. A
variable reference is expected on the left side of the operator. Modifier
flags are not permitted. The variable, which is created if it does not exist.
The syntax of the variable reference includes the initial "${" and
terminating "}" character (so it's similar to Perl's syntax).
?:
> ${foo::bar} = "hello"
"hello"
> ${foo::bar} .= ", world"
"hello, world"
> ${a} = [1, 2]
[1,2]
> ${a} .= [3, 4]
[1,2,3,4]
This is equivalent to the C/C++ conditional
expression, which has right to left associativity. If the first expression is
True, the result is the value of the second expression (the third is
not evaluated). If the first expression is False, the result is the
value of the third expression (the second is not evaluated).
or, ||
This is the C/C++ logical OR operator, which
yields 1 ( True) if either operand is True, otherwise it
yields 0 ( False). Evaluation is from left to right and and
stops as soon as the truth or falsehood of the result is known. The two tokens
are synonymous.
and, &&
This is the C/C++ logical AND operator, which
yields 1 ( True) if both operands are True, otherwise it
yields 0 ( False). Evaluation is from left to right and and
stops as soon as the truth or falsehood of the result is known. The two tokens
are synonymous.
Note
When expressions are parsed as XML attribute values, an '&' character must
be encoded as the five characters '&'.
|
This is the C/C++ bitwise inclusive OR
operator. Both operands must be integers.
^
This is the C/C++ bitwise exclusive OR
operator. Both operands must be integers.
&
This is the C/C++ bitwise AND operator. Both
operands must be integers.
Note
When expressions are parsed as XML attribute values, an '&' character must
be encoded as the five characters '&'.
==, !=, eq, ne, eq:i, ne:i
These operators compare their arguments and
return 1 if the relation is true, 0 otherwise. If both arguments are lists,
corresponding elements of both lists are compared, recursively. If both
arguments are alists, the number of pairs in both lists is compared and, if
necessary, pairs in the first list are looked up in the second list for
matching values (note that the case-insensitive variant applies only to the
value component of a pair, not the key component). For other valid arguments
an attempt is first made to coerce both arguments to numbers and do a numeric
comparison. If that fails, a lexicographic comparison is performed. Operators
having a :i modifier are like their counterparts without the modifier except
they do case-insensitive string comparisons.
If either argument is of type bstring, however, the comparison is done
differently than explained above. Two bstring arguments are equal if and only
if they are byte-wise identical. If one argument is a bstring and the other is
a string, the latter is treated as a bstring of length( string) bytes.
The case flag is ignored if at least one argument is a bstring.
<, <=, >, >=, lt, le, lt:i, le:i, gt, ge, gt:i, ge:i
These operators compare their arguments and
return 1 if the relation is true, 0 otherwise. An attempt is first made to
coerce both arguments to numbers and do a numeric comparison. If that fails, a
lexicographic comparison is performed. Operators having a :i modifier are like
their counterparts without the modifier except they do a case-insensitive
comparison.
Note
When expressions are parsed as XML attribute values, the '<' character must
be encoded as the four characters '<'; the same applies to the
"greater than" symbol.
If either argument is of type bstring, however, the comparison is done
differently than explained above. If two bstring arguments are compared, the
shorter bstring is "less than" the other argument and they are equal
if and only if they are byte-wise identical. If one argument is a bstring and
the other is a string, the latter is treated as a bstring of length(
string) bytes. The case flag is ignored if at least one argument is a
bstring.
.
The "dot" operator (not in ISO C)
concatenates its right operand to its left operand. If both arguments are of
type bstring, the result is also of type bstring. If the left operand is a
list and the right operand is a basic data type, the right operand is appended
to the list. If the left operand is a list and the right operand is also a
list, the elements of the right operand are appended to the left operand. A
list may not appear as the right operand if the left operand is not a list. In
all other cases, both arguments are coerced to string (an error occurs if this
cannot be done) before the left operand is appended to the right.
Note
A period will be recognized as a decimal point in a real number context rather
than as the dot operator, so the input:
will be scanned as a number whereas, for example, the input:
will evaluate to the string "45".
<<, >>
> "hello" . ", world" "hello, world" > "hello" . (16 + 1) "hello17" > 17 . (16 + 1) "1717" > [1, 2, 3] . 4 [1,2,3,4] > [1, 2, 3] . [4, 5, 6] [1,2,3,4,5,6] > [1, 2, 3] . [[4]] [1,2,3,[4]]
4.5
"4".5
These are the C/C++ bitwise left shift and
right shift operators, respectively. The first operand may be an integer or a
list, the second operand must be an integer. When shifting an integer, these
operators are implemented using the corresponding C/C++ operators. In the case
of right shifting, the behaviour with respect to arithmetic vs. logical shifts
will be platform dependent.
+, -
These are the (binary) addition and
subtraction operators, respectively. Both arguments are coerced to numbers. An
error occurs if this cannot be done. Also, unary + and - operators may precede
an arithmetic-valued expression.
*, /, %
These are the multiplication, division, and
remainder operators, respectively. Both arguments are coerced to numbers. An
error occurs if this cannot be done, such as attempting to divide by zero. For
the remainder operator, both operands must be integers.
**
This is the exponentiation operator (not in
ISO C). Both arguments are coerced to numbers (either both integers or both
reals). An error occurs if this cannot be done, such as attempting to raise to
a negative power.
+, -, not, !, ~, ++ VAR, --VAR, (type)
> 2**10 1024
The + and - operators are the (unary)
arithmetic plus and minus operators, respectively. These may precede an
arithmetic-valued expression. Both arguments are coerced to numbers. An error
occurs if this cannot be done.
The logical NOT operator (not, or equivalently, !) yields a result of zero when
applied to a non-zero numeric value and non-zero when applied to an operand of
zero. The result of applying this operator to a non-empty string is zero and
it is non-zero when applied to an empty string string. These two tokens are
synonymous.
The ~ operator is the one's complement (bitwise not) unary operator.
The ++ VAR and -- VAR operators are the prefix increment and
decrement operators, respectively. These operators are followed by a variable
reference. The variable must have an integer value.
An explicit type conversion can be forced by using a cast. The syntax for this
type coercion is:
The type must be a recognized data type name: integer or int (for an
integer), real or double (for a real), bool (for a boolean value as a long
integer), string (for a character string), bstring or binary (for a binary
string), list, alist, or void.
A list can be cast to an alist, provided it has no elements or an even number of
elements and if no key would appear more than once in the alist. A namespace
can be cast to an alist; the operand specifies the namespace, either as a
literal or a string. An alist can be cast to an list; the ordering of the
pairs in the resulting list is unspecified. A void type can only be cast to
void, which is a no-op. Here are some examples:
VAR++, VAR--, primary
> ${foo} = 17, ++${foo}
18
( type) expression
> (int) 3.4
3
> (int) "3.6"
3
> (bool) 17
1
> (bool) ""
0
> (string) (4 * 3)
"12"
> ${x} = "17"; (int) ((real) ${x} + (bool) 1965)
18
> (bstring) "abc"
"abc"
> (bstring) 4.4
"4.400000"
> (bstring) "\0\1\2"
""
> bstring("\0\1\2",3) . bstring("\3\4", 3)
"0001020304"
> (void) ($b=$x)
>
> (alist) [a, 1, "b", 2, 3, 3]
{"a", 1, "b", 2, "3", 3}
> (list) { red, first, blue, second, white, third }
["blue", "second", "white", "third", "red", "first"]
> $env = (alist) Env; $env["HOME"]
"/home/bobo"
> $env{HOME}
{"HOME","/home/bobo"}
The VAR++ and VAR-- operators
are the postfix increment and decrement operators, respectively. These
operators are preceded by a variable reference. The variable must have an
integer value.
A primary is a basic data type[20] (i.e., an integer or real number,
string, bareword, or binary string), or a variable reference[21].
Functions¶
A function call is written as a function name, optionally followed by whitespace, a left parenthesis, zero or more comma-separated arguments, and a right parenthesis. A function name begins with either an alphabetic character or an underscore, followed by any number of alphanumerics and underscores. Additionally, a pair of colons may appear exactly once within the name (except at the beginning or end of the name). The number of arguments and their expected types depends on the particular function being called. The order in which the arguments to a function are evaluated is undefined. There is no mechanism for creating user-defined functions yet (they will eventually be available on some platforms through dynamically linked libraries). The result of a function call is one of the supported data types[20], or an error. An invalid function call, including those that fail during execution, yields a False result. Function Index:•ack: notice acknowledgement
processing
•alist: create an alist
•alistref: create an alist
reference
•bstring: convert a string to
binary
•contains_any: count elements
common to two lists
•counter: persistent integer
counters
•dacs_admin: test if user is an
administrator
•dacs_approval: create or test a
signed authorization
•dacs_meta: get or update
metadata
•dacsauth: perform authentication
tests
•dacscheck: perform authorization
tests
•debug: control debugging
output
•decode: convert from a text
representation
•digest: cryptographic hash
functions
•encode: convert to a text
representation
•eval: evaluate a string
•exec: execute a program
•exit: terminate current
evaluation
•expand: variable
interpolation
•file: perform an operation on a
file
•file_group: test if user is
associated with file's group
•file_owner: test if user is
associated with file's owner
•from: test where the current
request comes from
•get: read the contents of a file
or VFS object
•hash: fast hashes
•hmac: secure keyed-hashes
•http: invoke an HTTP
request
•index: search a string or
list
•info: information about
namespaces and variables
•keysof: extract keys from an
alist
•ldap: extract a component from
an LDAP name
•length: string length
•list: create a list
•listref: dereference a
list
•on_success: evaluate an
expression if authentication or authorization succeeds
•password: compute or check a
password hash
•pathname: filename-based string
interpolation
•pbkdf2: password-based key
derivation
•print: display a string
•printf: display a formatted
string
•random: generate random
values
•redirect: redirect user after
access is denied
•regmatch: string matching
•regsub: string
substitution
•request_match: compare the
current request to a URI
•return: terminate current
evaluation
•rule: recursive authorization
checking
•setvar: operations on
namespaces
•sizeof: basic data type
sizes
•sleep: suspend execution
temporarily
•source: read and evaluate
external expressions
•sprintf: format a string
•strchars: select characters from
a string
•strftime: format the current
date and time
•strptime: parse a date and
time
•strrstr: locate the last
instance of a substring
•strstr: locate the first
instance of a substring
•strtolower: map uppercase to
lowercase
•strtoupper: map lowercase to
uppercase
•strtr: character
transliteration
•subset: test if one set is a
subset of another
•substr: extract a
substring
•syntax: perform a syntax check
on a string
•time: local time and date
•transform: filter text through
rule-based transformations
•transform_config: set options
for transform
•trim: delete trailing
characters
•typeof: get or test data
type
•undef: an undefined value
•user: test current user's
identity
•ustamp: generate a unique
stamp
•valuesof: extract values from an
alist
•var: operations on individual
variables
•vfs: perform a VFS
operation
ack(notice-uri[, ...][, EXACT_MATCH | ALL_MATCH])
This function is associated with notice
acknowledgement processing. The function indicates that the current service
request has one or more notices associated with it (identified by a sequence
of notice-uri arguments), each one represented by a URI that will
return the text of a notice that must be acknowledged by the user. Following
the last URI is an optional mode argument. The EXACT_MATCH mode is the default
mode and requires a single acknowledgement to address all of the specified
notices. The ALL_MATCH argument specifies a less stringent matching mode and
requires any set of acknowledgements to collectively address all of the
specified notices. See dacs_notices(8)[22].
alist([key, value [, ...])
This function is equivalent to the alist
construction operator[23]. There must be an even number of arguments, or
no arguments. If the first argument of each pair (the key) is not a string or
literal, it will be converted to a string, if possible.
is equivalent to the expression:
And the call:
yields:
alistref(list)
alist(cars, 2, bikes, 5)
{"cars", 2, "bikes", 5}
alist(2, xx, [0, 1], yy)
{"2", xx, "[0,1]", yy}
This function creates a new list that is
equivalent to that of the special "brace syntax" subscript used to
dereference an alist. This is currently useful only in conjunction with
listref()[24].
is equivalent to the expression:
the value of which is:
bstring(string, length)
listref({"a", 1, "b", 2, "c", 3}, alistref(["b"]))
{"a", 1, "b", 2, "c", 3}{"b"}
{"b", 2}
This function converts the first length
characters of string (which may also be a bstring and which is
converted to a string if necessary) into the binary type. The length
argument may be less than the actual length of string; if it is zero,
then actual length is computed, and if length is greater than the
actual length, the actual length is used. The implicit null character on the
end of string is not considered part of it.
contains_any(format, test-set, target-set[, nocase])
> bstring("\0\1\2", 4)
"000102"
> bstring("\0\1\2", 2)
"0001"
This function returns a count of the number of
elements of test-set that appear in target-set at least once.
Duplicate elements may appear in test-set and are considered to be
distinct. The format indicates how to parse the set arguments. It can
be the space, tab, or newline character, or any punctuation character. For
both sets, it is currently interpreted as the character that separates
elements. If the optional nocase literal argument is given, then set elements
are compared case-insensitively. The greatest possible return value is the
number of distinct elements in the third parameter.
The first expression returns 3 if every element in the third parameter
appears at least once (case insensitive) in the second parameter, otherwise
the value of the expression is 0. The second expression returns
4.
counter(op, vfs-ref, counter_name [,value])
contains_any(",", ${Args::LAYERS:i}, "Nests,Secret_roads,Heritage")
contains_any(",", "a,a,b,z", "a,a,a,b,b,b,a,z,z")
This function is used to manage persistent
integer counters, which can be useful for a variety of purposes, such as
counting the number of logins for a particular identity, limiting the number
of logins, or restricting the number of times a resource can be accessed.
Internally, counter values are integers[20].
The first argument specifies an operation and is case-insensitive. The second
argument identifies a filestore (typically a file or database). It must be an
indexed filestore scheme, such as dacs-kwv-fs or dacs-db (see VFS[25]).
The third argument is the name of the counter, which acts as a key. The
meaning of the fourth argument depends on the operation, but if present it
must be an integer.
Note
The current implementation has a limitation; a counter name (key) can be any
printable string but cannot contain a space character. You can work around
this limitation by encoding all keys every time they are used in a filestore
operation.
Operations that set or change the counter value return the new value.
For filestores that support locking, read-only operations obtain a shared lock
while the other operations obtain an exclusive lock.
It is an error to reference a counter that does not exist unless the operation
is set or exists.
Note
To some extent, this function is a poor substitute for a more general Perl-like
tie() function. Such a function is being considered.
Modifications to counters are not atomic. Amongst other things, this means that
a crash may cause counter updates to be lost.
A counter would typically be created by running dacsexpr(1)[1]:
The counter's value might then be tested in the revocation list[26] or by
an access control rule[27], for instance:
The counter might be conditionally updated using the on_success()[28]
function, or the AUTH_SUCCESS[29] or ACS_SUCCESS[30] directives,
using an expression like:
dacs_admin()
1.counter(set, vfs-ref,
counter_name, new-value)
This is used to create a new counter or reset an existing counter. The counter's
value will be new-value, which must be an integer, and is the return
value.
2.counter(create, vfs-ref,
counter_name, initial-value)
This is used to create a new counter if it does not already exist. The new
counter's value will be initial-value, which must be an integer. If the
counter exists, its value will not be changed and is returned.
3.counter(del[ete], vfs-ref,
counter_name)
This operation deletes an existing counter. The operation can be del or
delete.
4.counter(exists, vfs-ref,
counter_name)
This operation returns 1 if a counter exists, 0 otherwise.
5.counter(get, vfs-ref,
counter_name)
This operation returns the current counter value.
6.counter(inc|dec, vfs-ref,
counter_name[, amount])
This operation increments or decrements an existing counter by amount,
which must be an integer. If amount is not given, 1 is used. The
updated counter value is returned.
7.counter(decdel, vfs-ref,
counter_name[, amount])
This operation decrements an existing counter by amount, which must be an
integer. If amount is not given, 1 is used. If the resulting value is
zero or negative, the counter is deleted and zero is returned. If the counter
is not deleted, its updated value is returned.
8.counter(list, vfs-ref)
This operation returns a list of counters as a string, newline separated, each
with its current value.
% dacsexpr -e 'counter(set, "dacs-kwv-fs:/usr/local/dacs/counters/logins", "EXAMPLE::EX:bob", 1)'
counter(exists, "dacs-kwv-fs:/usr/local/dacs/counters/logins", ${DACS::IDENTITY})
counter(decdel, "dacs-kwv-fs:/usr/local/dacs/counters/logins", ${DACS::IDENTITY})
This predicate returns True if the user
making a service request has any credentials that match any specified by the
ADMIN_IDENTITY[31] configuration directive.
dacs_approval(op[, ...])
This function is used to create an approval
stamp[32] or inspect or validate one.
The following operations are available:
dacs_approval(approval, dacs64-approval-message, namespace)
dacs_meta(op[, ...])
This operation parses the
dacs64-approval-message (the value of DACS_APPROVAL), setting
variables in namespace, after first dacs64 decoding[33] the
argument. If namespace exists, its contents are deleted. Variables set
are: j (jurisdiction name), h (hash/digest name), s
(stamp), u (URI), m (HTTP method), and i (user identity).
See dacs_acs(8)[32]. The signature is not checked. The function
returns True (1) if the approval message is syntactically correct,
otherwise False (0).
dacs_approval(check, dacs64-approval-message)
The dacs64-approval-message is decoded
and parsed, and the signature is validated. The function returns True
(1) only if the signature is correct, otherwise False (0).
In the current implementation, the signature can only be validated by the
jurisdiction that signed the message. This deficiency will be addressed in a
future release and a web service will also supply this functionality. Ideally,
for maximum convenience, availability, efficiency, and simplicity, the
recipient of an approval message should be able to validate it directly if it
has the appropriate public key, invoke a web service at any jurisdiction in
the federation if public keys are distributed and kept current, or at the
jurisdiction that signed the message.
dacs_approval(create, uri, method, ident,
digest-name)
Create and return a
dacs64-approval-message (as described above and in
dacs_acs(8)[32]), formed from the given arguments and signed by the
current jurisdiction.
This function returns information associated
with the current federation, current jurisdiction, or other jurisdictions in
the current federation. See dacs_list_jurisdictions(8)[34] for
additional information.
The following operations are available:
dacs_meta(federation, namespace)
dacsauth(dacsauth-flags)
Return metadata for the current federation,
setting variables in namespace. If namespace exists, its
contents are deleted. Variables set are: federation, domain,
fed_id (if available), and fed_public_key (if available, in PEM
format).
dacs_meta(jname, jurisdiction-name, namespace)
Return metadata for the jurisdiction named
jurisdiction-name in the current federation. If namespace
exists, its contents are deleted. Variables set are: jname,
name, alt_name, dacs_url, authenticates,
prompts, auxiliary (if available), and public_key (if
available, in PEM format).
dacs_meta(jurisdiction, namespace)
This is equivalent to the jname operation with
jurisdiction_name set to the name of the current jurisdiction.
dacs_meta(list_jurisdictions)
Return a newline-separated list of all
jurisdiction names in the current federation. A local copy of the metadata is
used.
dacs_meta(update_jurisdiction, jname [,url])
Not implemented. Intended to update the
local metadata for the jurisdiction name jname. If url is
absent, then the current jurisdiction must already have the correct dacs_url
attribute in its entry for jname. If url is given, it is assumed
to be the URL for dacs_list_jurisdictions and it is used instead of one
formed from dacs_url for the jurisdiction.
dacs_meta(update_jurisdictions, jname)
Not implemented. Intended to update the
local metadata for all of the jurisdictions. If jname looks like a URL
(i.e., it begins with either "http" or "https", then it is
assumed to be the URL for dacs_list_jurisdictions and it is used to
obtain a fresh copy of the metadata; otherwise, jname is assumed to be
a jurisdiction name for which the current jurisdiction already has a correct
dacs_url attribute and metadata is retrieved from that jurisdiction.
This function provides an interface to
dacsauth(1)[35]. In the first usage, the single string argument is
parsed into space or tab separated flags. Single or double quotes are allowed.
In the second usage, each flag is a separate string or literal argument and is
not parsed.
An alist is returned that has the following three elements:
result
Important
This function should be considered experimental. Use it with caution. In version
1.4.25 and earlier, this function returned an integer value (the result).
Security
Like dacsauth and dacs_authenticate, if a built-in module is used
to perform authentication, this function must be run by a setuid or setgid
process to obtain sufficient privileges to access the required files; this is
true for Unix password authentication, for example.
Examples: .sp .if n { .RS 4 . .nf > dacsauth("-m unix suff -user
bobo -p apassword")
{"result",0,"identity","","roles",""}
> dacsauth("-m", "unix", "suff",
"-user", "bobo", "-p", "bpassword")
{"result",1,"identity","EXAMPLE::FEDROOT:bobo","roles",""}
> dacsauth("-r unix
-DVFS='[federation_keys]dacs-fs:/usr/local/dacs/federations/federation_keys'
-u bobo")
{"result",0,"identity","","roles","bobo,wheel,www,users"}
.fi .if n { .RE . .sp
dacscheck(dacscheck-flags)
An integer: 1 if authentication succeeded, 0
if it failed or was not requested, and -1 if an error occured.
identity
A string: if authentication was requested and
succeeded, this is the corresponding identity, otherwise it is the empty
string.
roles
A string: if roles were requested (and
authentication succeeded, if requested), this is the role descriptor string,
otherwise it is the empty string.
This function provides an interface to
dacscheck(1)[36], returning 1 if access is granted, 0 if access is
denied, and -1 if an error occurs. In the first usage, the single string
argument is parsed into space or tab separated flags. Single or double quotes
are allowed. In the second usage, each flag is a separate string or literal
argument and is not parsed.
Important
This function should be considered experimental. Use it with caution.
debug(type, value)
This function enables, disables, or adjusts
the amount of debugging output produced by the interpreter. Output type
type is set to value, which may be "on",
"off", or a non-negative integer level (the meaning of which depends
on type.
The following type names are recognized: TBD
decode(encoding-type, string)
This function performs the inverse of
encode()[37] for the same encoding-type. The result is a
bstring. The function will fail if its argument is not properly encoded.
For the hex encoding type, alphabetic characters may be upper case or lower
case.
digest(msg, msg-len [, digest-name])
This function computes a cryptographic
hash[38] of msg (a string or bstring). The msg-len is the
length of msg in bytes; if it is 0, its length is computed. The hash
algorithm can be any function provided by OpenSSL and may be specified
as digest-name, case insensitively, otherwise SHA1 is used. The list of
available digest algorithms is subject to change, but is likely to include
md5, sha, sha1, sha224, sha256, sha384, and sha512. The function value is a
bstring. If cryptographic strength is not required, see hash()[39].
encode(encoding-type, arg)
> digest("foo", 0, "md5")
"acbd18db4cc2f85cedef654fccc4a4d8"
> digest("Hello, world", 0, "SHA256")
"4ae7c3b6ac0beff671efa8cf57386151c06e58ca53a78d83f36107316cec125f"
This function converts arg, a string or
bstring, into a printable text representation that depends on
encoding-type. Applying decode()[40] with the same
encoding-type to the output of this function will produce a value
equivalent to the original arg. The result is a string.
Note that encoding is only a representational or formatting change. If secrecy,
authentication, or verification of integrity are required, use a cryptographic
method.
The following encoding types are recognized:
encode(ascii85, arg)
eval(expression)
This encoding, also known as
radix-85[41], uses nearly every printable character to obtain a compact
encoding. But note that the resulting strings may be problematic in many
contexts without additional encoding, which can largely defeat the reason for
selecting this encoding in the first place. The start-of-data
("<~") and end-of-data ("~>") indicators that are
sometimes used with this encoding are not included.
encode(cescape, arg)
> encode(ascii85, decode(hex, "123456789a")) "&i<X6RK"
This encoding converts its argument into a
C-style escaped string. Character escape codes are used when possible, numeric
escape codes are used for other non-printable characters, and all other
characters map to themselves.
encode(dacs64, arg)
> encode(cescape, bstring("hi\0\1\2\3\012", 7))
"hi\0\001\002\003\n"
This encoding type produces a base-64 encoding
of arg using upper- and lower-case alphabetics, digits, '-', and '_'.
It is similar to the mime encoding except that '-' and '_' are used in the
encoding character set instead of '+' and '/'. This encoding is better suited
for use in paths and URIs, for example, and is used extensively within
DACS. It is sometimes referred to as "the dacs64 encoding" or
just "dacs64" in the DACS documentation.
encode(hex, arg)
> encode(dacs64, bstring("\0\0\0\1", 4))
"_____-"
This encoding converts each byte in arg
into a hexadecimal character pair.
encode(mime, arg)
> encode(hex, "Hello") "48656c6c6f"
This encoding applies the MIME base-64
encoding function ( RFC 2045[42], Section 6.8) to its argument and
returns the result.
encode(url, arg)
> encode(mime, bstring("\0\0\0\1", 4))
"AAAAAQ=="
This returns the URL-encoding of the argument
( RFC 1738[43], RFC 2396[44] (Section 2.4), and RFC
3986[45]).
> encode(url, bstring("a\0b", 3))
"a%00b"
This function evaluates its string argument
and returns the result.
The call:
exec(prog, ...)
> eval("length(\"abc\")")
3
The exec function executes prog,
waits (indefinitely) for it to terminate, and returns the program's standard
output. A trailing newline in the output is deleted. Optionally, command line
arguments to prog may be given; they are automatically converted to
strings. By default, no environment variables are passed to the program; if
the namespace ExecEnv exists, however, its contents are used as the
executed program's environment variables. The exit status of prog is
made available as the value of ${DACS::status}. The program is executed
using the execv(3)[46] function, not a command shell.
On POSIX systems, this call returns the string "1\n" on Thursdays,
"0\n" on any other day:
Security
The program is executed as the same user and group IDs as the DACS
program that calls exec(). Take appropriate precautions to prevent
unauthorized users from modifying or replacing DACS configuration
files, access control rules, and so on.
exit(result)
> exec("/bin/sh", "-c", "date | grep -c ^Thu")
"0"
> ${ExecEnv::PATH} = "/usr/bin";
"/usr/bin"
> exec("/bin/sh", "-c", "printenv");
"PATH=/usr/bin"
Equivalent to return, this function
causes evaluation of the expression, block, or program being evaluated to
terminate and returns result as the value of the expression or the
program's exit status.
expand(string)
The argument, a string, is returned with
variable references expanded. An undefined variable expands to the empty
string.
file(op [,arg-list])
> ${a} = 17
17
> "${a}"
"17"
> '${a}'
"${a}"
> expand('${a}')
"17"
> ${b} = 1999, ${c} = expand('${a}, \${b}')
"17, ${b}"
> expand(${c})
"17, 1999"
This function performs various operations on
files and filenames according to op, which is one of the following
operation names, followed by command-specific arguments. All arguments must
either be strings or literal words.
file_group([path])
1.file(basename, string
[,suffix])
This is used to extract the last component of a pathname and is equivalent to
the basename(1)[47] command. It deletes any prefix that ends with the
last slash character in string, after first stripping trailing slashes,
and a suffix, if present. The suffix is not stripped,
however, if it is identical to the remaining characters in string. A
non-existent suffix is ignored. The value is the resulting string.
> file(basename,"/a/b/c") "c" > file(basename,"/a/b/c.c") "c.c" > file(basename,"/a/b/c.c", ".c") "c" > file(basename,"/a/b/c.c", "c") "c." > file(basename,"/a/b/c.c", "c.c") "c.c" > file(basename,"/a/b/c.c//", "c.c") "c.c"
2.file(chmod, abs-mode, file)
Change the mode of file to abs-mode, which is an absolute (octal)
file mode (note, however, that DACS always set the process umask
to 07).
file(chmod, "0755", "/usr/local/dacs/tmp/foofile")
3.file(dirname, string)
Equivalent to the dirname(1)[48] command, its value is the string that
remains after deleting the filename portion of string (a pathname),
beginning with the last slash character to the end of string, after
first stripping trailing slashes.
> file(dirname,"/usr/local/dacs/bin/dacsexpr") "/usr/local/dacs/bin" > file(dirname,"/usr/local/dacs///") "/usr/local"
4.file(extension, pathname)
The returned value is all of the characters in pathname after and
including the last dot in the last element. If there is no dot in the last
element of pathname, the value is the empty string.
> file(extension,"acl-myapp.0") ".0"
5.file(lstat, fmt, file)
This is like the stat[49] operation, except in the case where the named
file is a symbolic link, in which case lstat returns information about
the link, while stat returns information about the file the link
references.
6.file(mkdir, directory [,abs-mode])
Create directory. If an absolute (octal) mode is given, the new directory
will have that mode (note, however, that DACS always set the process
umask to 07).
7.file(readlink, file)
If file is a symbolic link, print its contents.
8.file(remove, file)
Remove (delete) file.
9.file(rename, source-file,
target-file)
Rename (mv) source-file to target-file.
10.file(rmdir, directory)
Remove (delete) directory, which must be empty.
11.file(stat, fmt, file)
Similar to the stat(1)[50] command available on some systems, this makes
the functionality of the stat(2)[51] system call available. The
fmt argument is a printf(3)[52]-type descriptor that indicates
what file status information is wanted and how it is to be printed.
Non-formatting characters, including \n, \t, and \\, are copied to the output
verbatim.
The following format specifiers are understood:
This excerpt from an access control rule limits access to authenticated users
for every file greater than 999 bytes in length that it DACS-wraps:
•%d
The value of st_dev.
•%i
The value of st_ino.
•%m
The value of st_mode in octal.
•%M
The value of st_mode as text.
•%l
The value of st_nlink.
•%u
The value of st_uid in decimal.
•%U
The value of st_uid as text.
•%g
The value of st_gid in decimal.
•%G
The value of st_gid as text.
•%r
The value of st_rdev.
•%s
The value of st_size.
•%b
The value of st_blksize.
•%n
The value of the file argument.
•%N
If the argument is a symbolic link, print the contents of the link, otherwise
print the file argument.
•%ta
The value of st_atime in decimal.
•%tA
The value of st_atime as text.
•%tm
The value of st_mtime in decimal.
•%tM
The value of st_mtime as text.
•%tc
The value of st_ctime in decimal.
•%tC
The value of st_ctime as text.
•%f
The name of the host (fileserver) where the file is stored.
•%%
A literal '%' character.
<allow>
user("auth")
</allow>
<allow>
user("any") and file(stat, "%s", ${DACS::FILENAME}) lt 1000
</allow>
12.file(test, op [, args])
Most of the file-testing predicates of the test(1)[53] command are
available.
•-b file
True if file exists and is a block special file.
•-c file
True if file exists and is a character special file.
•-d file
True if file exists and is a directory.
•-e file
True if file exists, regardless of its type.
•-f file
True if file exists and is a regular file.
•-g file
True if file exists and its set group ID flag is set.
•-k file
True if file exists and its sticky bit is set.
•-p file
True if file exists and is a named pipe (FIFO).
•-r file
True if file exists and is readable (access(file, R_OK) ==
0).
•-s file
True if file exists and has a size greater than zero bytes.
•-u file
True if file exists and its set user ID flag is set.
•-w file
True if file exists and is writable (access(file, W_OK) ==
0).
•-x file
True if file exists and is executable (access(file, X_OK) ==
0).
•-L file
True if file exists and is a symbolic link.
•-O file
True if file exists and its owner matches the effective user id of
this process.
•-G file
True if file exists and its group matches the effective group id
of this process.
•-S file
True if file exists and is a socket.
•-nt file1 file2
True if file1 and file2 exist and the former is newer than
the latter.
•-ot file1 file2
True if file1 and file2 exist and the former is older than
the latter.
•-ef file1 file2
True if file1 and file2 exist and refer to the same
file.
13.file(touch, file [,
abs-mode])
If file does not exist, it is created; if an absolute (octal) mode is
given, the new file will have that mode (note, however, that DACS
always set the process umask to 07). If the file exists, its
modification time will be set to the current date and time.
Test if path (defaults to
${DACS::FILENAME}, which is equivalent to Apache's
SCRIPT_FILENAME or REQUEST_FILENAME variables) has a group
ownership with which the user making the request is associated. This is
effectively the same as:
This predicate provides a simple way of limiting access to a file to its group
membership with respect to file system permissions:
For example, if the user requesting access has been assigned the following roles
by the current jurisdiction (e.g., through local_unix_roles):
and the resource being requested is the file:
then this predicate would return True because the file has group
ownership www and the user is associated with that role.
There is an implicit assumption that the file in question is associated with the
current jurisdiction; this might be problematic if more than one jurisdiction
can claim this association.
file_owner([path])
file(test, "-e", ${DACS::FILENAME})
and
user("%" . ${Conf::JURISDICTION_NAME} . ":" \
. file(stat, "%G", ${DACS::FILENAME}))
<allow> file_group() </allow>
wheel,www,users
-rw-r--r-- 1 bobo www 75 Apr 11 12:41 htdocs/foo.html
Test if path (defaults to
${DACS::FILENAME}, which is equivalent to Apache's
SCRIPT_FILENAME or REQUEST_FILENAME variables) is owned by the
user making the request. This is effectively the same as:
This predicate provides a simple way of limiting access to a file to its owner
with respect to file system permissions:
There is an implicit assumption that the file in question is associated with the
current jurisdiction; this might be problematic if more than one jurisdiction
can claim this association.
from(string)
file(test, "-e", ${DACS::FILENAME})
and
user(${Conf::JURISDICTION_NAME} . ":" . file(stat, "%U", ${DACS::FILENAME}))
<allow> file_owner() </allow>
This predicate is used to test where a request
comes from, based on the values of REMOTE_ADDR and REMOTE_HOST.
These environment variables are passed to DACS from Apache. The
supported argument types are similar to those recognized by the Apache
mod_access[54] module's allow and deny directives. If either
REMOTE_HOST or REMOTE_ADDR are needed to evaluate the argument
but are not available, the result will be False.
The string argument may be:
An alternative method is to perform a regular expression match against
${DACS::REMOTE_ADDR} using regmatch()[59].
Tip
To test where a client authenticated from, which is not necessarily the same as
the place from which a request is sent, use the user()[60] function.
get(vfs-ref [,key])
1.a full or partially matching domain name:
Here, the function yields True if the given domain name matches
REMOTE_HOST or is a subdomain of REMOTE_HOST. Case-insensitive
matching is performed ( RFC 1035[55]). Only complete components are
matched, so the above example will match foo.metalogic.example.com but not
foonmetalogic.example.com. If REMOTE_ADDR is available but not
REMOTE_HOST, a reverse DNS lookup will be performed on the domain name
and all IP addresses that result will be tested against REMOTE_ADDR; if
this lookup results in an error (i.e., it fails), then the function raises an
error condition.
from("metalogic.example.com")
2.a full IPv4 address in standard dot
notation:
from("10.0.0.123")
3.a partial IPv4 address (the first one, two,
or three bytes) in standard dot notation:
from("10.0")
4.a network/netmask pair:
from("10.0.0.0/255.255.0.0")
5.a network/nnn pair using CIDR
notation[56] ( RFC 1338[57]):
from("10.0.0.0/8")
6.a full or partial IPv4 address in standard
dot notation where any address element can be a decimal number (0 through 255)
or a range specification[58], similar to that used with
strchars()[15]; note that the range separator in this context is
":" instead of "..": :
In the example above, the two high-order octets of ${DACS::REMOTE_ADDR}
must be 10 and 0, the value of the next octet must be between 0 and 100
(inclusive) or be 255 (decimal), and the value of the fourth octet is
unimportant. The following expressions are equivalent:
from("10.0.[0:100,255]")
from("10")
from("10.")
from("[10]")
from("[10].")
from("10.0.0.0/8")
from("10.0.0.0/255.0.0.0")
7."all" (always yields True
and is included for compatibility with Apache):
from("all")
The file or item specified by vfs-ref,
which may be followed by a key if it is an indexed filestore, is read
and returned. The vfs-ref may be an absolute pathname, an item type, or
a vfs_uri[61], except if called from a standalone application without a
key argument, in which case vfs-ref may also be a relative
pathname.
Note
A proper I/O subsystem does not exist yet, but until then you may use the
special item type stdin to read the standard input until end of file. This
function will probably not work if a special file is used (e.g., /dev/stdin).
hash(msg, msg-len [,hash-name])
This function computes a fast hash of
msg, a string or bstring. The msg-len is the length of
msg in bytes; if it is 0, its length is computed. The hash-name
can be the 32-bit hash "hash32" (the default) or the 64-bit hash
"hash64". The result is a string. Although the algorithms have been
used extensively with very good results, they should not be used for
cryptographic purposes; see digest()[62].
hmac(msg, msg-len, key, key-len [,
digest-name])
> hash("Hello, world", 0)
"3696529580"
> hash("Hello, world", 0, hash64)
"462009511995194717"
This function computes a cryptographic
message authentication code[63] - specifically, the Keyed-Hash
Message Authentication Code (HMAC)[64] - of msg (a string or
bstring), using key (a string or bstring). The msg-len is the
length of msg in bytes; if it is 0, its length is computed. Similarly,
key-len is the length of key in bytes and if it is 0, its length
is computed. Any of the available Secure Hash Standard functions[65],
as well as MD5 (RFC 2104)[66], may be specified as digest-name
(case insensitively), otherwise SHA1 is used. The list of available digest
algorithms is subject to change, but is likely to include md5, sha1, sha224,
sha256, sha384, and sha512. The function value is a bstring. Note that the
function is not commutative; the key is the third argument, not the first.
Although the MD5 hash function is deprecated for some purposes, it is still
considered adequate in some applications and is required by many older
protocols that are still in widespread use.
http(url, [method [,arglist]])
> hmac("Sample #2", 0,
decode(hex, "303132333435363738393a3b3c3d3e3f40414243"), 0)
"0922d3405faa3d194f82a45830737d5cc6c75d24"
This function sends an HTTP request to
url, using a given method (GET, POST, HEAD, PUT, DELETE, or OPTIONS,
case insensitively), and optionally passing parameters. If no method is given
(and no arguments), GET is assumed. The value of the function is the message
returned by the request. The url is in the usual syntax and must use
either the http or https scheme (case insensitive). The argument list, if
present, consists of some number of pairs, the first being the name of the
parameter and the second the value of the parameter.
The first statement sends an HTTP request to example.com and sets the variable
to the message body (if any) that is returned. The second statement makes a
GET request to port 8443 of example.com over SSL, passing it two parameters,
FOO=17 and FOO=2:
index(string, character-class [, nocase])
> ${x} = http("http://example.com")
> http("https://example.com:8443/cgi-bin/dacs_prenv.cgi", "GET", "FOO", 17,
"BAZ", 1+1)
If the first argument is a string, this
function returns the first position in string (counting from 1) where
the first character in character-class was found, or 0. Case-sensitive
character comparison is used unless the optional nocase literal
argument is present.
If the first argument is a list, the position of element search_operand
(counting from 1) in list is returned, or 0 if it is not found. During
comparison, types are automatically converted as necessary. Case-sensitive
character comparison is used unless the optional nocase literal
argument is present.
Examples:
info(namespaces)
> index("abcdef", "abc")
1
> index("abcdef", "e")
5
> index("zzz", "abc")
0
> index([a, b, c, d, e], d)
4
> index(["hello", world, 2009, qUAKe], "quake", nocase)
4
> index([1.0, 2.2, 3.3, 4.4, 5.0, 6.6], "1")
1
> index(["apple", ["orange", "banana"], ["peach", "mango"]], "orange")
0
> index(["apple", ["orange", "banana"], ["peach", "mango"]], ["orange", "banana"])
2
Return a string containing information about
variables and namespaces. The first form returns a comma-separated list of
known namespaces. The second form returns a list containing all variables in
the given namespace and their values, one per line. This can be useful for
debugging.
Examples:
keysof(alist)
info(namespaces) info(namespace, "Conf")
If its argument is a single pair, the pair's
key is returned. If there is more than one pair in the argument, a list of
keys is returned. To get the value component of a pair or set of pairs, use
valuesof()[18].
Examples:
ldap(dn_length, dn-string)
> keysof({red, 17})
"red"
> keysof({red, 17, blue, 100})
["red", "blue"]
The ldap function is used to extract
components of LDAP names. Its first argument, a literal, determines the
operation mode to be used and the semantics of the following arguments.
Distinguished Name (DN) and Relative Distinguished Name (RDN) strings are as
defined in RFC 2253[67].
The dn_length mode returns the number of RDN components in its DN argument; -1
is returned if the argument is not a valid DN. The dn_index mode returns the
nth RDN component of the DN, where nth is an integer greater
than zero. If nth is greater than the number of components, the last
component is returned.
The rdn_length mode returns the number of AttributeTypeAndValue elements in its
RDN argument; -1 is returned if the argument is not a valid RDN. The rdn_index
mode returns the nth AttributeTypeAndValue component of the RDN, where
nth is an integer greater than zero. If nth is greater than the
number of components, the last component is returned.
The rdn_attrtype mode returns the AttributeType of the nth
AttributeTypeAndValue component of the RDN, where nth is an integer
greater than zero. If nth is missing, it is taken to be 1. If
nth is greater than the number of components, the last component is
selected. The rdn_attrvalue mode is similar except that it returns the
AttributeValue.
The first and second expressions below return 2, the third expression returns
Administrator:
length(string)
ldap(dn_length, "dc=example,dc=com")
ldap(rdn_length, "foo=bar+bar=baz")
ldap(rdn_attrvalue, ldap(dn_index, \
"CN=Administrator,CN=Users,DC=example,DC=com", 1))
This function returns the length, in
characters, of string, the number of bytes in binary string
bstring, the number of elements in list, or the number of pairs
in alist.
list([value [, ...])
This function is equivalent to the list
construction operator[68].
is equivalent to the expression:
listref(list, list-ref [, ...])
list(1, 2, [hello, world], 5)
[1, 2, [hello, world], 5]
This function provides an alternate syntax to
the language's list/array notation. For example, the function call:
is equivalent to the expression:
Note that a list reference may follow a list-valued expression (e.g., a list
constructor, a list-valued variable, a function that returns a list) this
syntax is valid:
The parentheses are necessary here because the subscript binds more tightly than
the concatenation operator. This expression can also be written as:
on_success(list-name [, expr])
listref([1, 2, [3, 4], 5], 2, 1)
[1, 2, [3, 4], 5][2][1]
($a . $b)[0]
listref($a . $b, 0)
The list-name argument must be either
acs or auth (case insensitive) to select the post-authorization list or the
post-authentication list, respectively. For the former case, if authorization
is successful, the expr argument (a string) will be evaluated by
dacs_acs immediately after any ACS_SUCCESS[30] directive, and
just prior to program termination. These expressions are not evaluated if
authorization is denied, an authorization processing error occurs, or a
DACS_ACS argument prevents execution of the request. For the latter
case, if authentication is successful, the expr argument (a string)
will be evaluated by dacs_authenticate(8)[8] immediately after any
AUTH_SUCCESS[29] directive, and just prior to program termination.
These expressions are not evaluated if authentication fails or an
authentication processing error occurs.
Once added to either list, an entry cannot be removed. The expressions are
evaluated in the order in which on_success() was called. The values
returned by the expressions are discarded and errors are ignored.
If no expr is given, the current list of expressions is returned, one per
line, in order of evaluation. With an expression argument, it returns the
number of expressions in the list after any addition.
password(op [, op-args])
This function performs a variety of read-only
operations on DACS accounts and their passwords. See
dacspasswd(1)[69] and directives PASSWORD_DIGEST[70] and
PASSWORD_SALT_PREFIX[71] for additional information.
The following operations are available:
password(check, given-password, password-digest
[,alg-name])
Examples:
pathname(path, hostname, port)
With the check operation, the digest of
given-password is computed (as computed by the hash operation) and
compared to password-digest, which was previously generated by the hash
operation of this function, retrieved by the getdigest operation, or obtained
using dacspasswd. This algorithm is identical to the one used by
local_passwd_authenticate[72] to validate passwords. If alg-name
is given, it names the digest algorithm to use instead of the one specified
within password-digest. If given-password is correct (i.e., the
same passwords were used to generate the two digest values), True (1)
is returned, otherwise False (0) is returned.
password(getdata, username [,vfs-ref])
The getdata operation returns the private data
associated with the account for username. The result is a bstring. If
there is no private data, the length of the result will be zero (the length of
the empty string is one). If a vfs-ref is given, it identifies the
virtual filestore to use, otherwise the item type passwds is used. It is an
error if the account does not exist, so a test operation will often be
performed first.
password(getdigest, username [,vfs-ref])
The getdigest operation is similar to getdata
except that the digest string for the account is returned; this digest string
can be used with the check operation.
password(hash, plain-password [,alg-name])
With the hash operation, a digest of the
string plain-password is returned as a printable string. The password
hashing algorithm is identical to the one used by dacspasswd(1)[69]. If
alg-name is given (see digest()[62]), it names the digest
algorithm to use instead of the configured default.
password(list [, vfs-ref])
The list operation returns a list of account
names, one per line. An empty string is returned if there are no accounts. If
a vfs-ref is given, it identifies the virtual filestore to use,
otherwise the item type passwds is used. To test if a password file exists,
use vfs()[73].
password(syntax, password [,constraints])
The syntax operation tests if password
satisfies the constraints argument, if provided, otherwise the value of
the PASSWORD_CONSTRAINTS[74] directive. The constraints are
specified in the same syntax as the PASSWORD_CONSTRAINTS directive. The
function returns True (1) if the constraints are satisfied, otherwise
False (0).
password(test, test-op, username [,vfs-ref])
The test operation applies test-op to
the account entry for username in the virtual filestore vfs-ref
(or item type passwds). It is an error if the password file does not exist or
is unreadable. It returns True if the test is successful, otherwise the
result is False. The recognized values of test-op are (case
insensitively): data (to test if the account exists and has private data),
disabled (to test if the account exists and is disabled), enabled (to test if
the account exists and is enabled), or exists (to test if the account
exists).
> password(hash, "bobo") "2|XYZZYxBhU/7VgJAt2lc.G|HL4RQ2vo0uNoXlXnv.GcY3Vlf9." > password(check, "bobo", "2|XYZZYxBhU/7VgJAt2lc.G|HL4RQ2vo0uNoXlXnv.GcY3Vlf9.") 1
Perform string interpolation on path
based on the other arguments. For details, please see dacs.conf(5)[75]
(where hostname is SERVER_NAME).
pbkdf2(password, salt, count, dklen)
Apply a pseudorandom function (HMAC-SHA-1) to
password and salt (both binary strings, or converted as
required), modified by count iterations, returning a binary string of
length dklen bytes. For details, please see RFC 2898[76] and
RFC 3962[77].
print(...)
> pbkdf2("password", "ATHENA.MIT.EDUraeburn", 1200, 32)
"5c08eb61fdf71e4e4ec3cf6ba1f5512ba7e52ddbc5e5142f708a31e2e62b1e13"
> pbkdf2("password", decode(hex,"1234567878563412"), 5, 16)
"d1daa78615f287e6a1c8b120d7062a49"
Each argument is converted to a string, the
strings are concatenated, a newline is appended, and the result is printed.
The return type is void. If called from dacsexpr(1)[1], the string is
printed to the standard output; otherwise, it is printed to the DACS
log file (or stderr), which can be useful for debugging purposes. These log
messages are associated with the user class (see the LOG_FILTER[78]
directive).
printf(fmt, ...)
This is a slightly scaled-down version of the
printf(3)[52] library function. If called from dacsexpr(1)[1],
the string is printed to the standard output; otherwise, it is printed to the
DACS log file (or stderr), which can be useful for debugging purposes.
These log messages are associated with the user class (see the
LOG_FILTER[78] directive). This can be useful for debugging purposes.
If necessary and possible, arguments are converted to the type requested by a
formatting specification. The return type is void.
random(bytes, nbytes)
The various forms of this function,
distinguished by the first argument, return cryptographically strong
pseudo-random values[79] in various formats. The starting point (seed
value) for the pseudo-random sequence cannot be set, meaning that the sequence
cannot be (intentionally) reproduced.
The bytes operation requests nbytes bytes of random material. The result
is a bstring of that length.
The uint operation requests an unsigned random integer between lo and
hi (both unsigned integers), inclusive. It is an error if lo is
not greater than hi. The result is an (unsigned) integer.
The string operation requests nbytes of random material, returned as a
hex-encoded string. If a spec argument is present, it uses the
character specification syntax of strtr()[80] to indicate the
characters that can be used to encode the result. Only printable characters,
excluding the space, are allowed in the result, regardless of the spec
argument. Example:
The functionality of the stringc operation is identical to that of the
three-argument instance of the string operation except that the sense of the
spec argument is complemented to indicate those characters that may
not be used in the encoding of the result.
redirect(error-code, target)
> random(string,12,"a-zA-Z0-9") "LgROshy6SMMH" > random(string,12,"a-z") "kehhvwydhhbk"
Permitted only within the context of an access
control rule's deny clause, this function causes expression evaluation and
rule processing to stop immediately, access to be denied, and the client to be
redirected to target, a URL that may contain a query component. If the
error-code is present, it must be an ACS error name or number (see the
ACS_ERROR_HANDLER[81] directive), otherwise "BY_REDIRECT" is
used.
Note
The URL must be properly escaped if it appears within an XML document, such as
an access control rule; for example, if an ampersand occurs in the query
component in a context where it must be escaped, it must appear as the five
characters "&".
The target string is expected to have one of the syntaxes of the document
component of Apache's ErrorDocument directive[82]. In essence,
this function causes an ACS_ERROR_HANDLER directive to be created and
triggered. The function returns the target string, although because of
the function's run time behaviour the value cannot be used.
Tip
One application of this function is to create a short link, which is a
relatively concise URL that acts as an "alias" for another, usually
much longer URL (here, the target). The short link is made public. It
must be DACS-wrapped; the target does not need to be. Any
attempt to access the short link is denied by its rule, but the rule uses the
redirect() function, probably with BY_SIMPLE_REDIRECT as the
error-code (see dacs.conf(5)[81]), to redirect the user agent to
the target.
The following rule demonstrates how this can be done:
With this rule in place, a request like:
would result in a redirect to this target:
The target URL can depend on contextual elements, and it is straightforward to
do things like make the target URL depend on the time of day, identity of the
user, and so on. The technique can also be used with Rlinks[83].
Because the rule associated with the short link can be changed at any time, this
feature can be used to implement smart permalinks[84].
regmatch(string, regex [, namespace] [, nocase])
<acl_rule status="enabled">
<services>
<service url_pattern="/id/*"/>
</services>
<rule order="allow,deny">
<deny>
setvar(split, "X", ${Env::REQUEST_URI}, "/");
${x} = var(get, X, ${X::#} - 1);
redirect(BY_SIMPLE_REDIRECT, "https://example.com/docs/${x}.html");
</deny>
</rule>
</acl_rule>
This is a pattern matching function. The first
two arguments are coerced to strings, with the second one taken to be the
regular expression, with a "^" (the start-of-string anchor)
implicitly prepended. The string argument is then matched against the
regular expression, which may contain subexpressions enclosed between '(' and
')' (or '\(' and '\)'). If the match fails, the result is 0. If the match
succeeds there are several possibilities:
If the optional nocase literal argument is given, then matching is done
case-insensitively. Only one parenthesized pair can be used. IEEE Std 1003.2
("POSIX.2") "extended" regular expressions are supported (
regex(3)[85], re_format(7)[86]).
Examples:
regsub(string, regex, replacement [,nocase]
[,repeat])
•if there are no subexpressions in
regex, the result is an integer that is the number of characters
matched.
•if there is at least one subexpression
in regex but no namespace (a string argument) is given, the
result is the substring of string that was matched by the entire
regular expression.
•if there is at least one subexpression
in regex and a namespace argument is given, the result is an integer
that is the number of characters matched by the entire regular expression. The
value of the first matching subexpression is assigned to the variable named
" 1" in the namespace, the value of the second subexpression
is assigned to a variable named " 2" in the namespace, and so
on up to the ninth subexpression. The variable named " 0" in
the namespace is assigned the substring of string that was matched by the
entire regular expression. Following function evaluation in the context of ACL
rule processing, namespace is accessible only within the predicate,
allow, or deny element in which it appears.
> ${X} = "abfoo"
"abfoo"
> regmatch(${X}, ".*foo", nocase)
5
> regmatch("abcdefgzz", "(.*)g")
"abcdefg"
> regmatch("foo", "(bar)|(baz)|(foo)")
"foo"
> regmatch("abcdefgzz", "ab(.*)efg(.*)", "x")
9
> ${x::0}
"abcdefgzz"
> ${x::1}
"cd"
> ${x::2}
"zz"
> $addr = "192.168.7.3"
"192.168.7.3"
> regmatch($addr, "192\\.168\\.(.*)\\..*", "X")
11
> ${X::1}
"7"
This function matches regex against
string, like regmatch()[59] does, and returns the string that
results when the substitution specified by replacement is applied to
the matched text. This is similar to the ed/vi/sed
command "s/regex/replacement/" applied to string. If no match
is found, the empty string is returned.
The optional repeat literal argument causes the replacement to be applied
to all matches; i.e., like the ed/vi/sed command
"s/regex/replacement/g".
Examples:
request_match(uri-string)
> regsub("hello world", "world", "auggie")
"hello auggie"
> regsub("hello world", "auggie", "world")
""
> regsub("hello", ".*", "& &")
"hello hello"
> regsub("one two three", "(.*) (.*) (.*)", "\${3} \${2} \${1}")
"three two one"
> regsub("one two three", "(.*) (.*) (.*)", '${3} ${2} ${1}')
"three two one"
> strtr(regsub("https://BOB.Example.com",
"\([^:]*\)://\([^.]*\)\\.\(.*\)", '${1}-${2}@${3}'),
"A-Z", "a-z")
"https-bob@example.com"
> regsub("one, bone, cone, hone", "one", "two", repeat)
"two, btwo, ctwo, htwo"
This function is used to inspect the current
request. The argument is either a valid URI or a path component that begins
with a slash. In the latter case, the scheme and authority components of the
current request are effectively prepended to the given path. The path
component is like the url_pattern attribute used in access control
rules[87] in that it can either specify an exact match or, by ending in
"/*", a wildcard match. A query component is allowed but ignored.
The function returns 0 if uri-string does not match the current
request, otherwise it returns the number of path components of
uri-string that match the current request. If the scheme and authority
components are given in uri-string, they count as one naming component.
Assuming that the current request is http://example.com:18123/a/b/c, we get:
return(result)
> request_match("http://example.com:18123/a/b/c")
4
> request_match("https://example.com:18123/a/b/c")
0
> request_match("http://example.com:18123/a/b/c/d")
0
> request_match("http://example.com:18123/a/b")
0
> request_match("http://example.com:18123/a/b/*")
4
> request_match("http://example.com:18123/*")
2
> request_match("http://example.com:18123")
0
> request_match("http://example.com")
0
> request_match("http://example.com/*")
2
> request_match("/*")
1
> request_match("/a/b/c")
3
> request_match("/a/b/*")
3
> request_match("/")
0
Equivalent to exit, this function
causes evaluation of the expression to terminate and returns result as
the value of the expression.
rule(object, ruleset_vfs)
The rule predicate is an interface to
the DACS rule processing engine. It is used to test if the rule set
ruleset_vfs authorizes object, much as dacscheck(1)[36]
does. The object argument is the name to match against the services
specified in access control rules and can either be a URI or an absolute
pathname (one that begins with a slash character). It can have an optional
query string component attached. An absolute pathname path is mapped
internally to a URI as file:// path; e.g., /myapp is interpreted as
file:///myapp (see RFC 1738[43]).
One application of this predicate is for a rule associated with a program to
check that the user requesting access is entitled to use a data file needed by
the program.
Note
Only the path component of the URI is considered when DACS matches
an object's name against the url_pattern of an access control rule. At
present, the object name is not automatically canonicalized or resolved (see
RFC 3986[45]), as is usually done by a web server, so relative path
components should be avoided.
The ruleset_vfs is a URI in the syntax of the VFS[25]
configuration directive.
The various components of the URI that names the object are available as
DACS variables and environment variables (see below). If a query string
is given, it is parsed and the individual arguments are made available to
rules through the Args namespace, just as for DACS-wrapped web
services.
Many variables normally set by a web server are instantiated based on the object
name and the execution environment. These variables are available in the
DACS namespace. For example, if the object name is
https://example.com:8443/myapp/edit-menu?entry=item1, the following variables
will be set as indicated:
The value of ${Args::entry} will be item1. The request method is always
GET. The variable ${DACS::REMOTE_USER} will be set if credentials are
available in the execution environment.
For example, assuming that the file /usr/local/exams/acls/acl-exams.17 contains:
The following call would only return True ( 1) any day between
5:00pm and 5:59pm:
Note
setvar(op, dst-namespace [, args ...])
${DACS::HTTPS}=on
${DACS::SERVER_NAME}=example.com
${DACS::SERVER_ADDR}=142.179.101.118
${DACS::HTTP_HOST}=example.com:8443
${DACS::SERVER_PORT}=8443
${DACS::REQUEST_URI}=/myapp/edit-menu
${DACS::DOCUMENT_ROOT}=/
${DACS::REQUEST_METHOD}=GET
${DACS::SERVER_SOFTWARE}=dacsexpr-1.4.14
${DACS::QUERY_STRING}=entry=item1
${DACS::ARG_COUNT}=1
${DACS::CURRENT_URI}=/myapp/edit-menu?entry=item1
${DACS::CURRENT_URI_NO_QUERY}=/myapp/edit-menu
<acl_rule status="enabled">
<services>
<service url_pattern="/exam1.html"/>
</services>
<rule order="allow,deny">
<precondition><predicate>
${Args::user} eq "teacher"
</predicate></precondition>
<allow>
time(hour) eq 17
</allow>
</rule>
</acl_rule>
rule("/exam1.html?user=teacher", "dacs-fs:/usr/local/exams/acls");
•Since any rule can call the rule
function, take care to avoid infinite recursion.
•Although this function is similar in
concept to the dacscheck(1)[36] command, there are some significant
differences, particularly with respect to the context available during rule
evaluation.
•The Env namespace is not
reinitialized or altered during evaluation of rules processed by rule.
That is, the Env namespace is the same as the outer-most one.
This function, which performs various
operations on namespaces, has several different syntaxes. The first argument
always specifies the operation (case insensitively) and determines the meaning
of the arguments that follow it. The second argument always specifies a
namespace that is created or modified. If successful, the function returns the
number of variables created (or replaced) in dst-namespace.
The dst-namespace cannot be a read-only namespace[88]. Unless
otherwise specified, if dst-namespace exists, variables are added to
it, with any existing variable assigned its new value.
The following operations are recognized:
setvar(authorization, dst-namespace, auth-str)
setvar(load, dst-namespace, item_type, key)
setvar(load, dst-namespace, vfs-ref, key)
sizeof(typename)
The auth-str argument, which is the
value of an Authorization HTTP request header, is parsed into its component
fields and assigned to variables in the destination namespace
dst-namespace. If dst-namespace exists, its contents are deleted
first. Corresponding to the field names used in RFC 2617[89] Section
3.2.2, the following variables are created: AUTH_SCHEME,
USERNAME, PASSWORD, REALM, NONCE,
DIGEST_URI, RESPONSE, ALGORITHM, CNONCE,
OPAQUE, MESSAGE_QOP, NONCE_COUNT, and AUTH_PARAM.
Any variable that corresponds to a non-existent field is assigned the empty
string.
The following call sets ${Foo::AUTH_SCHEME} to Basic,
${Foo::USERNAME} to Bobo, and ${Foo::PASSWORD} to myPassWord.
setvar(copy, dst-namespace, src-namespace)
setvar(authorization, Foo, "Basic Qm9ibzpteVBhc3NXb3Jk")
With the copy operation, all variables in an
existing namespace src-namespace are copied to dst-namespace. If
the latter exists, its contents are deleted, otherwise the namespace is
created.
setvar(delete, dst-namespace)
The delete operation is used to delete
dst-namespace and its contents.
setvar(kwv, dst-namespace, assign-char, sep-chars,
string)
For the kwv operation, string is
parsed, creating (or replacing) variables in dst-namespace. The
string consists of zero or more keyword/value pairs. The keyword, which
is used as the variable name, is separated by the value by the character
assign-char. A keyword/value pair is separated from the next by any
character that appears in sep-chars. Here is an example:
The value of this call is 3 and it sets ${Foo::a} to "b",
${Foo::c} to "d", and ${Foo::e} to
"f".
setvar(load, dst-namespace, filename)
setvar(kwv, "Foo", "=", ", ", "a=b, c=d, e=f")
The contents of a text object, which can be
specified using a filename, as an item_type, or using a vfs-ref
(see vfs()[73]), is split into newline separated lines. The first line
is assigned index 0 in dst-namespace, the second index 1, and so on.
Consider the following example:
Here, there are 25 lines in /etc/passwd and the third line is printed.
setvar(loadi, dst-namespace, vfs-ref)
> setvar(load, PASSWD, "/etc/passwd")
25
> ${PASSWD::2}
"root:*:0:0:Charlie &:/root:/bin/csh"
Each item in the indexed text object specified
by vfs-ref (an absolute pathname, an item type, or a VFS
URI[61]) is copied to dst-namespace, with the same index. The index
must be a valid variable-name.
Here, 23 items are copied into the PASSWD namespace (the first two lines
in this particular /etc/passwd are ignored because they are comments that are
not recognized as items). The lines indexed by the keys root and bobo are
printed.
setvar(merge, dst-namespace, src-namespace)
> setvar(loadi, PASSWD, "dacs-kwv-fs:/etc/passwd")
23
> ${PASSWD::root}
"*:0:0:Charlie &:/root:/bin/csh"
> ${PASSWD::bobo}
"bobo:*:1001:1001:Bobo &:/home/bobo:/bin/tcsh"
The merge operation is similar to
copy[90] except that if dst-namespace exists its contents are
not deleted.
setvar(post, dst-namespace [, content-type, string])
Like query[91], this operation parses
its input into arguments in dst-namespace. The function reads its
standard input, unless a string argument is given. The input is
expected to be a correctly formatted application/x-www-form-urlencoded or
multipart/form-data content type. If the standard input is read, both the
CONTENT_TYPE and CONTENT_LENGTH environment variables must be
set (as they are when Apache runs a script that is passed an entity-body).
The form that takes string is not yet implemented.
setvar(query, dst-namespace, query-string)
For the query operation, query-string
is parsed, creating variables in dst-namespace. This uses the same
parsing algorithm employed by cgiparse(8)[92]. In the case of a
malformed query string, like "a&b", variables will be created
but will have the empty string as their value. If successful, the function
returns the number of variables created. The following call returns 3 and sets
${Foo::a} to "b", ${Foo::c} to "d", and
${Foo::e} to "f":
One application of this function it to distinguish query arguments (which are
part of the requested resource's URI and made available through the
environment variable QUERY_STRING) from arguments supplied in the body
of a POST method (or other such method). For example:
setvar(regsplit, dst-namespace, string,
delimiter-regex [, limit])
setvar(query, "Foo", "a=b&c=d&e=f")
setvar(query, "Qargs", "${Env::QUERY_STRING}")
if (${Qargs::foo:e}) {
/* "foo" is a query argument */
}
else {
/* "foo" is not a query argument */
}
if (${Args::foo:e} and not ${Qargs::foo:e}) {
/* "foo" is a POST argument */
}
else {
/* "foo" is not a POST argument */
}
The regsplit operation is similar to
split[93] except that substrings are separated by the regular
expression delimiter-regex. IEEE Std 1003.2 ("POSIX.2")
"extended" regular expressions are used (
regex(3)[85]).
setvar(rename, dst-namespace, src-namespace)
The rename operation deletes
dst-namespace, if it exists, and changes the name of
src-namespace to dst-namespace. The two namespace arguments must
be different.
setvar(split, dst-namespace, string, delimiter
[,limit [, dflag]])
The split operation extracts substrings from
string. Substrings are separated by the string delimiter. For
example, this call separates a composite role string into individual basic
roles:
If the variable reference ${DACS::ROLES} has the value
"root,wheel,www,users", then the example would return 4 and set
${ROLES::0} to "root", ${ROLES::1} to
"wheel", and so on.
If a limit is given, it is an integer that specifies the maximum number
of substrings to extract. Once the maximum has been reached, the remainder of
string that has not been split will be assigned to the last element. A
limit of zero is equivalent to the default, which is for there to be no
maximum. For instance, setvar(split, X, "a,b,c,d", ",", 2)
will assign "a" to ${X::0} and "b,c,d" to
${X::1}.
Here is another example:
This function can be used to break a pathname into its individual components.
For instance, the following call results in ${X::0} set to the empty
string, ${X::1} set to "a", ${X::2} set to
"long", and ${X::3} set to "path":
(You may need to first remove redundant slashes in string using
strtr()[80].)
A dflag argument may follow the limit argument to indicate whether
delimiter should not be included in substrings ( dflag == 0,
which is the default behavior), whether it should be included at the start of
substrings with the possible exception of the first one ( dflag >
0), or whether it should be included at the end of substrings with the
possible exception of the last one ( dflag < 0).
setvar(uri, dst-namespace, uri)
setvar(split, "ROLES", ${DACS::ROLES}, ",")
> setvar(split, "X", "a\nb\nc\n", "\n")
"3"
> ${X::0}
"a"
> ${X::#}
"3"
> setvar(split, "X", "/a/long/path", "/") 4
> setvar(split, P, "/a/long/path", "/", 0, 1)
3
> ${P::0}
"/a"
> ${P::1}
"/long"
> ${P::2}
"/path"
> setvar(split, P, "/a/long/path", "/", 0, -1)
4
> ${P::0}
"/"
> ${P::1}
"a/"
> ${P::2}
"long/"
> ${P::3}
"path"
The given uri, a URI conforming to
RFC 2396[44] or RFC 3986[45], is parsed into its components.
Variables in dst-namespace are set accordingly: SCHEME (mapped
to lower case), HOST (mapped to lower case), AUTHORITY,
PORT, SERVER, USERINFO, PATH, QUERY, and
FRAGMENT. If a component is absent from uri, the corresponding
variable will not be defined.
Security
It is possible for USERINFO to include a plaintext password.
In addition, the URI's path component is split into its slash-delimited pieces.
The variable PATH_LENGTH is set to the number of such pieces (it will
be zero if there are none), and variables PATH_0, PATH_1, and so
on are set to the first, second, and successive pieces. An "empty"
path component is treated as a piece consisting of the empty string.
> setvar(uri, "X", "https://bar@foo.example.com:8443/cgi-bin/prog?a=17") 11 > info(namespace,X) "SCHEME="https" AUTHORITY="bar@foo.example.com:8443" HOST="foo.example.com" PORT="8443" SERVER="foo.example.com:8443" USERINFO="bar" PATH="/cgi-bin/prog" PATH_COUNT="2" PATH_0="cgi-bin" PATH_1="prog" QUERY="a=17" "
This function returns the amount of memory in
bytes, as an integer, used by typename, the name of a basic data
type[20]. For the string and binary types, the returned value is the
number of bytes used by each element of that type (1, typically). To find the
number of elements in string or binary data, use length()[14].
sleep(seconds)
> sizeof(real) 8
The process is suspended for approximately
seconds seconds, or until a signal is received and caught or the
process terminated. It returns the "unslept" number of seconds,
which will be zero if the process slept for the requested interval. This is an
interface to sleep(3)[94]. It can be useful for inserting delays in
conjunction with error handlers, for instance.
source(vfs-ref [,key])
The expressions in the file or item specified
by vfs-ref, which may be followed by a key if it is an indexed
filestore, are read and evaluated as a block. The vfs-ref can be an
absolute pathname, an item type, or a VFS URI[61]. The value returned
is that of the evaluated block. The following two expressions are essentially
equivalent:
This function is handy when a lengthy expression is needed but one does not want
to clutter a configuration file or a rule.
sprintf(fmt, ...)
source("/usr/local/dacs/scripts/script17")
eval(get("/usr/local/dacs/scripts/script17"))
This is a slightly scaled-down version of the
sprintf(3)[95] library function. If necessary and possible, arguments
are converted to the type requested by a formatting specification. The
formatted string is returned.
strchars(str, range-spec [,...])
${a} = sprintf("Hello") . ", world."
"Hello, world."
length(sprintf("Hello") . ", world.")
13
This function returns a new string by
selecting characters from str according to a sequence of one or more
range specifications (each one a range-spec). A range-spec is a
string argument that determines the indexes of characters to select within
str. Indexes start at zero. The result of each successive range
specification is appended to the previous result.
A range-spec is an unordered set of one or more comma-separated elements,
each of which is either an index or a range. An index may
either be a non-negative integer or "#", which means "all
indexes". A range represents a sequence of indexes and has the
syntax:
A range-start may be a non-negative integer, the character "#"
(which means "from the beginning"), or may be elided (also meaning
"from the beginning"). A range-end may be a non-negative
integer (not less than range-start, if it is also a non-negative
integer), the character "#" (which means "to the end"), or
may be omitted (also meaning "to the end").
strftime(format)
range-start ".." range-end
> $a = "abcdef" "abcdef" > strchars($a, 2) "c" > strchars($a, "1..4", "0") "bcdea" > strchars($a . $a, "5..#") "fabcdef" > strchars($a, "#") "abcdef" > strchars($a, "#..#") "abcdef" > strchars($a, "#..3") "abcd" > strchars($a, "..3") "abcd" > strchars($a, "..3", "#") "abcdabcdef"
This function is an interface to the
strftime(3)[96] function. It is applied to the current date and
time.
strptime(date-str, date-format, namespace)
This function is an interface to the
strptime(3)[97] function. The date-str argument is a string
representation of a date and/or time, with date-format describing its
syntax. If the parse of date-str succeeds, the following elements of
namespace are set from the corresponding fields of struct tm:
tm_sec, tm_min, tm_hour, tm_mday, tm_mon,
tm_year, tm_wday, tm_yday, tm_isdst,
tm_zone, and tm_gmtoff. Additionally, a variable named
clock is set to the Unix time that corresponds to the parsed date and
time. Any existing elements of namespace are not modified. If
date-str does not fully describe a date and time, it is taken to be
relative to the current date and time (e.g., if only a time is given,
"today's date" is used).
In the single-argument usage, the current date and time are parsed and
namespace is assigned values as previously described.
The return value is the "Unix time" equivalent of the resulting time
and date.
strrstr(string, substring)
> strptime("6 Dec 2001 12:33:45", "%d %b %Y %H:%M:%S", tm)
1007670825
> "${tm::tm_mon} ${tm::tm_mday} ${tm::tm_hour} ${tm::tm_min}"
"11 6 12 33"
> ${tm::clock}
1007670825
Return the start of the last occurrence of
substring within string. The empty string is returned if
string is empty or if no occurrence of substring is found. If
substring is empty, string is returned.
strstr(string, substring)
Return the start of the first occurrence of
substring within string. The empty string is returned if
string is empty or if no occurrence of substring is found. If
substring is empty, string is returned.
strtolower(string)
> strstr("foobazbar", "baz")
"bazbar"
> strstr("foobazbar", "")
"foobazbar"
> strstr("foobazbar", "zzz")
""
> strstr("", "zzz")
""
> strstr("afoofoofooz", "foo")
"foofoofooz"
> strrstr("afoofoofooz", "foo")
"fooz"
A new string is returned where each uppercase
character in string is mapped to lowercase and all other characters are
mapped to themselves. These two expressions are equivalent and have the value
"hello, world 2008":
strtoupper(string)
strtolower("Hello, World 2008")
strtr("Hello, World 2008", "A-Z", "a-z")
A new string is returned where each lowercase
character in string is mapped to uppercase and all other characters are
mapped to themselves. These two expressions are equivalent and have the value
"HELLO, WORLD 2008":
strtr(input-string, string1, [string2 [,cds]])
strtoupper("Hello, World 2008")
strtr("Hello, World 2008", "a-z", "A-Z")
This function performs string transliteration,
like the tr(1)[98] command and Perl's tr and y operators. The
result is the transliterated string. The first argument is the input string to
be transliterated (stdin in the tr command). The second argument is the
search list ("string1" in the tr command). The third argument
is the (possibly empty) replacement list ("string2" in the tr
command); it may be omitted if no flag string argument follows.
The fourth, optional argument is a literal flag string made of the characters
'c', 'd', and 's' (in any order), which correspond to the flags of the same
name in the tr command:
c
subset(format, purported-subset, superset [,
nocase])
Complement the set of values in
string1.
d
Delete characters in string1 from the
input string.
s
Squeeze multiple occurrences of the characters
listed in the last operand (either string1 or string2) in the
input into a single instance of the character. This occurs after all deletion
and translation is completed.
> strtr("AbCdEf", "A-Z", "a-z")
"abcdef"
> strtr("/a//b///c", "/", "", "s")
"/a/b/c"
This function returns True if every
element of the purported-subset appears in superset. The format
indicates how to parse the set arguments. It can be the space, tab, or newline
character, or any punctuation character. It is currently interpreted as the
character that separates elements. If the optional nocase literal argument is
given, then set elements are compared case-insensitively.
Example:
This call returns True if every element of the LAYERS parameter (case
insensitive) appears in the given list, otherwise the expression is
False.
substr(string, start-position, length)
subset(",", ${Args::LAYERS:i}, "RELIEF:Foundation,GTOPO30:Foundation")
This function returns the substring of
string beginning at start-position with length at most
length characters. The first character is in position one. If
start-position is negative, the position is relative to the end of
string (-1 specifies the last character in string). If the
effective starting position is outside of string, an empty string is
returned. If length is negative, it means "the remainder of the
string". It is an error if either numeric argument is zero. It is not an
error if length exceeds the actual number of characters returned.
syntax(type, name [, flag])
> substr("foozle", 3, 4)
"ozle"
> substr("foobar", -3, 2)
"ba"
> substr("foobar", -5, -1)
"oobar"
> substr("foobar", 10, -1)
""
> substr("foobar", -10, 3)
""
This function performs a syntax test,
specified by type, on name. It returns 0 if the test fails, 1 or
a type-dependent, non-zero value if the test is successful. It can be
useful for testing, catching errors, recognizing when a string must be mapped,
and for learning about DACS. Note that these are purely syntactical
checks. They do not test whether an object called name exists or is
configured.
The following tests are recognized:
syntax(charset, name, charset_spec)
time(format [, timeval])
Test if each of the characters in name
is specified by charset_spec, which is a character set specification as
used by strtr()[80] ("the search list").
syntax(dacsname, name)
Test if name is valid as a
DACS name[99]. If the string is recognized, one of the following
values is returned to classify it:
syntax(emailaddr, name)
•1 if it is a DACS identity
•2 if it is a group name
•3 if it is a jurisdiction name
•4 if it is a federation name
•5 if it is an IP address in numeric dot
notation
Test if name is a syntactically valid
RFC 822[100] email address. A successful test does not imply that a
message can be delivered to the address.
Note
The implementation does not currently recognize valid addresses where the
local-part (the substring to the left of the '@' character) contains a
quoted-string component.
syntax(expr, name)
Test if name is a syntactically valid
expression. The expression is not actually evaluated. A successful test does
not imply that evaluation of the expression will necessarily be successful or
error-free.
syntax(domainname, name)
Test if name is a syntactically valid
domain name ( RFC 952[101]). A successful test does not imply that
name exists or has a DNS entry.
syntax(federation, name)
Test if name is valid as a federation
name (e.g., as the value of FEDERATION_NAME[102]).
syntax(group, name)
Test if name is valid as a group
name.
syntax(hostname, name)
Test if name is valid as a host name
(an alphanumeric, followed by any number of alphanumerics and hyphens, but not
ending with a hyphen; see RFC 952[101] and RFC 1123[103]).
syntax(ipaddr, name)
Test if name is a valid Class C IPv4
address ( RFC 790[104]).
syntax(jurisdiction, name)
Test if name is valid as a jurisdiction
name (e.g., as the value of JURISDICTION_NAME[105]).
syntax(namespace, name)
Test if name is valid as the name of a
namespace[21].
syntax(role, name)
Test if name is valid as a role
descriptor string[99].
syntax(uri, name)
Test if name is a valid URI ( RFC
2396[44], but partially RFC 3986[45]). It must consist of a scheme,
authority component, path component, and optional query and fragment
components.
syntax(username, name)
Test if name is valid as a username
(e.g., as the value of the USERNAME argument to many DACS web
services).
syntax(variable, name)
Test if name is valid as a variable
reference[21]. This does not test if the named variable exists.
syntax(varname, name)
Test if name is a syntactically correct
variable name[21], with or without a namespace. This does not test if
the named variable exists.
> syntax(federation, "FOO")
1
> syntax(dacsname, "FOO::BAZ:bar")
1
> syntax(dacsname, "FOO::")
4
> syntax(charset, "bobo17+", "a-z0-9")
0
> syntax(expr, '1 + 1 + 1')
1
> syntax(variable, '${1$}')
0
> syntax(variable, '${Foo::baz:z}')
0
> syntax(varname, 'Foo::baz')
1
> syntax(varname, "17")
1
> syntax(username, "/bobo/")
0
> syntax(group, "blop")
1
> syntax(group, "%blop")
0
> syntax(dacsname, "%blop:flop")
1
> syntax(uri,"https://foo.example.com:8443/cgi-bin/prog?a=17")
1
This function returns time and date
information, as specified by the first argument. The second argument, if
present, either specifies the "Unix time" from which to obtain the
time and date or a namespace that was returned by strptime()[106]. If
the second argument is absent, the result is the same as if a second argument
were given as time("now"). The localtime(3)[107] library
function is used internally to perform the date calculations.
The format argument, which is treated case-insensitively, can be any of
the following:
Note
A more powerful function is planned to test whether the current time and date
satisfy a predicate. It might, for example, understand arguments such as
"Tuesday" ( True on any Tuesday), "last day of the
month", "between midnight and 8:30am", "January 30, 2004
at 1:23pm", "between March 2 and April 1", "the second
Tuesday of the month", or "within 15 days of April 30".
transform(input,name,rules,docs
[,idents])
•If the argument is "now", the
function's value is the current "Unix time" (the value of time in
seconds since 0 hours, 0 minutes, 0 seconds, January 1, 1970, Coordinated
Universal Time). If the second argument is present, however, it is the
function's value.
•If the argument is "sec" or
"secs" or "seconds", the function's value is the system
clock's seconds reading.
•If the argument is "min" or
"mins" or "minutes", the function's value is the system
clock's minutes reading.
•If the argument is "hour",
the function's value is the system clock's hour reading (0 - 23).
•If the argument is "mday",
the function's value is the day of the month (1 - 31).
•If the argument is
"ismdaylast", the function's value is non-zero if this is the last
day of the month.
•If the argument is "mon" or
"month", the function's value is the month of the year (0 -
11).
•If the argument is "year",
the function's value is the year (from 1900 onward).
•If the argument is
"isleapyear", the function's value is non-zero if this is a leap
year.
•If the argument is "wday",
the function's value is the day of the week (Sunday is 0).
•If the argument is "yday",
the function's value is the day of the year (0 - 365).
•If the argument is "isdst",
the function's value is non-zero if daylight saving time is in effect.
•If the argument is "zone",
the function's value is system clock's time zone, abbreviated. If the time
zone is not known, the value will be the empty string.
•If the argument is "gmtoff",
the function's value is the offset (in seconds) of the system clock's time
represented from UTC, with positive values indicating east of the Prime
Meridian.
This function provides a simplified API for
dacstransform(1)[108] - refer to its description for additional
details. The first form of the function uses compile-time defaults, unless
they are overridden by configuration variables (e.g.,
${Conf::transform_prefix}). The second form passes a configuration
object returned by transform_config()[109]. The input argument
is the text to be passed through the function. The name argument is
equivalent to the value of the dacstransform -name flag,
rules is equivalent to the value of the -r flag, docs is
equivalent to the value of the -docs flag, and the optional
idents argument is a whitespace-separated list of identities in the
concise user syntax[110]. The function returns the transformed
input.
transform_config(flags)
This function returns a configuration object
that is passed to subsequent calls to transform()[111] so that defaults
can be overridden. The single string argument is parsed into
whitespace-separated words. If a flag is repeated, the right-most occurrence
is used.
The following flags are recognized:
trim(string, delete-set [,limit])
•-prefix
prefix-string: The string used to introduce a directive,
which must appear at the beginning of a line.
•-suffix
suffix-string: The string used to end a directive.
•-rprefix
regex-prefix: A line whose beginning matches the specified
regular expression introduces a directive.
•-rsuffix
regex-suffix: The end of a directive is found by matching
the specified regular expression.
Delete each character in delete-set
that appears at the end of string, up to limit characters. The
delete-set is a search list specification as used by
strtr()[80]. If limit is missing or zero, all of the characters
in string can potentially be deleted (leaving the empty string). The
new string is returned.
typeof([typename,] expression)
> trim("abceffff", "f")
"abce"
> trim("abceffff", abf)
"abce"
> trim("a\n\n\n", "\n")
"a"
> trim("a", "a-z")
""
If there are two arguments and the first is a
recognized data type name[20], the return value is 1 ( True) if
expression has that type and 0 ( False) otherwise. If there is
one argument, the function yields a string that is the data type name of the
evaluated expression.
undef()
> typeof(4.5) "real" > typeof(integer, 4.5) 0
This function returns a special value that
represents the "undefined" value. It is used in certain
circumstances to undefine a symbol. See dacs.conf(5)[9].
user(string)
This function compares its argument against
each set of current credentials and returns the number of credentials that
match. The argument is a user filter expression that must evaluate to
True for a set of credentials for those credentials to match. See
dacs(1)[99] for information about naming.
Note
In typical usage, each user will have only one set of credentials or will be
unauthenticated. One should keep in mind, however, that multiple concurrent
identities are allowed, subject to ACS_CREDENTIALS_LIMIT[112].
The string argument ( EXP) has the following syntax:
Figure 2. User Filter Expression Grammar
Whitespace (spaces and tabs) is permitted before and after lexical elements.
Keywords are case sensitive except when otherwise stated.
A primary, which evaluates to True or False, is one of the
following:
username
By default, an exact string comparison (case sensitive) is used to match name
components other than the special names; this default behaviour can be
overridden using the NAME_COMPARE configuration directive (
dacs.conf(5)[9]). The method used to compare federation names,
jurisdiction names, and usernames can also be specified by following the
primary with a mode. If the value of mode (which is
itself case insensitive) is case, then case-sensitive comparisons are used, if
its value is nocase, then case-insensitive comparisons are used, and if its
value is default, then the value of the NAME_COMPARE directive will be used if
present, otherwise the application default is used (either case or the value
selected by the application).
Important
Keep in mind that user() can return False because no credentials
matched the user filter expression and because there are no credentials at all
(i.e., the user is unauthenticated). For example,
will return True if the user's identity is not METALOGIC:rmorriso, even
if the user is not authenticated. It may therefore be necessary to explicitly
test for an authenticated user:
Here are examples of the user() function. Note that any non-zero
expression value implies True.
Tip
The following two tests are not equivalent:
The first is True if the user making the request has been authenticated;
it does matter which jurisdiction authenticated the user or what the username
is. The second test requires the user making the request to have a specific
identity; she must have been authenticated by the jurisdiction DSS as the
username auth.
ustamp(op, vfs-ref [,args ...])
EXP -> E1
E1 -> E2 | E2 OR E2
E2 -> E3 | E3 AND E2
E3 -> E4 | NOT E3
E4 -> primary | "(" E1 ")"
OR -> "or" | "||"
AND -> "and" | "&&"
NOT -> "not" | "!"
True if the DACS identity
username matches.
If the jurisdiction name or federation name components are omitted, the
current federation and jurisdiction[99] are implied. The jurisdiction
name component may be specified as "*" (e.g., *: username),
in which case it will match any jurisdiction name in the current
federation. In addition, both the federation name and the jurisdiction name
components may be specified as "*" (e.g., *::*: username), in
which case it will match any federation name and any
jurisdiction name.
jurisdiction
user("METALOGIC:auggie")
user(":bobo")
True if jurisdiction matches the
name of the jurisdiction that created the credentials.
federation
user("METALOGIC:")
user("DEMO::METALOGIC:")
True if federation matches the
name of the federation that created the credentials.
address
user("DEMO::")
Given an argument acceptable to the
from()[113] predicate, the result is True if the credentials
were generated by a user apparently located at address.
group
user("10.0.0.123")
user("10.0.0.0/24")
user("example.com")
True if the identity is a member of
group, which is a DACS group.
A group name may reference an explicit group membership list or a role-based
group. Also, it is possible for an explicit group membership list to have the
same name as a role-based group; if the name is referenced in a rule, the rule
processing engine will first check if the user is associated with the role. If
he's not, it will go on to check for an explicit group membership list with
the same name. This allows an administrator to easily supplement the
membership associated with a role-based group. Refer to
dacs.groups(5)[114].
namespace ns
user("%METALOGIC:admin")
The value of each element in ns (a
namespace) is evaluated as a primary. The order in which the list is
evaluated is unspecified. Processing of the list terminates with the first
primary that evaluates to True or when the list is exhausted.
This primary can appear in an element (so that one list can reference
other lists) but beware of infinite recursion.
For example, if /usr/local/dacs/app_users consists of usernames, one per line,
an access control rule can grant permission to any of the users by having an
allow element containing the statements:
style style-list
setvar(load, APP_USERS, "/usr/local/dacs/app_users");
user("namespace APP_USERS")
The keyword style is followed by a list of one
or more comma-separated, case-insensitive style keywords, described below.
Each style keyword may be abbreviated up to the indicated minimum number of
initial characters. Every set of credentials has one or more
styles[115] associated with it that indicate which authentication
method or methods were successfully applied and how (by what means) the
credentials were generated within DACS. A primary is True if the
tested credentials satisfy all of the keywords in the
style-list.
For example, this expression tests if both the passwd and certificate styles are
associated with it:
This is equivalent to the following expression, which tests if the user was
authenticated via a username/password style of authentication and a valid
X.509 client certificate was presented:
The following style keywords are understood:
acs
passwd
This test can be used as part of a risk-based authentication configuration; a
user with credentials obtained through an authentication style deemed not to
be sufficiently secure with respect to a resource could be forced to
reauthenticate using a stronger authentication method. See
dacs_authenticate(8)[8] for additional information.
importedby jurisdiction
user("style passwd,cert")
user("style passwd") and user("style CERT")
True if the credentials were created
during an authorization check by dacs_acs
admin
True if the credentials were created
for use internal to DACS.
alien
True if the credentials were imported
by dacs_auth_agent(8)[116] in its "alien" mode, or by
dacs_auth_transfer(8)[117].
cas
True if the user was authenticated
using CAS.
cert[ificate]
True if the user authenticated using an
X.509 certificate.
digest
True if the user authenticated using
RFC 2617[89] Digest authentication.
expr
True if the user was authenticated
using an expression.
gen[erated]
True if the credentials were generated
by a DACS utility (e.g., dacscookie(1)[118]).
import[ed]
True if the credentials were imported
by dacs_auth_agent(8)[116] or dacs_auth_transfer(8)[117].
infocard
True if the user was authenticated
using an InfoCard.
nat[ive]
True if the user was authenticated
using the native authentication style.
managed_infocard
True if the user was authenticated
using a managed InfoCard.
pass[word]
True if the user authenticated using a
password.
prompt[ed]
True if the user was authenticated
using the prompted authentication style.
rlink
True if the user was authenticated
using an Rlink[83].
selfissued_infocard
True if the user was authenticated
using a self-issued InfoCard.
simple
True if the user authenticated without
using a password.
The keyword importedby is followed by the name
of a jurisdiction within the current federation; the result is True if
the credentials were imported using dacs_auth_transfer(8)[117] at that
jurisdiction.
version protocol-version
user("importedby METALOGIC")
The keyword version is followed by a
DACS protocol version number (every release of DACS defines this
as the value of the compile-time symbol DACS_VERSION_NUMBER); the result is
True if the credentials match that protocol version number.
authenticated, unauthenticated
user("version 1.4")
Either of two keywords: authenticated (or
simply auth) or unauthenticated (or simply unauth). The former is True
if the user is authenticated, while the latter is True if the user is
not authenticated. A case-insensitive string comparison is used to match these
special names.
mine
user("auth")
user("unauth")
The keyword "mine" (case
insensitive) is True if the user was authenticated by the current
jurisdiction.
any
user("mine")
The keyword "any" (case insensitive)
is always True.
none
user("any")
The keyword "none" (case
insensitive) is always False.
user("none")
user("not METALOGIC:rmorriso")
user("not METALOGIC:rmorriso and auth")
1.
Return True if the client was authenticated by the jurisdiction METALOGIC
in this federation
user("METALOGIC:")
2.
Return True if the client was authenticated as the user
METALOGIC:rmorriso
user("METALOGIC:rmorriso")
3.
Return True if the client was authenticated by the given federation and
jurisdiction as rmorriso
user("DEMO::METALOGIC:rmorriso")
4.
Return True if the client is a member of the group METALOGIC:admin
user("%METALOGIC:admin")
5.
Return True if the client was authenticated as the username rmorriso by
any jurisdiction in this federation
user("*:rmorriso")
6.
Return True if the client was authenticated anywhere
user("auth")
7.
Return True if the client is not authenticated
user("UnAuthenticated")
8.
Return True if the client was authenticated through a request from a host
having the IP address 10.0.0.123
user("10.0.0.123")
9.
Return True if the client is unauthenticated or was not authenticated
through a request from a host having the IP address 10.0.0.123 (use
user("auth and not 10.0.0.123") to remove the unauthenticated
case)
user("not 10.0.0.123")
10.
Always return True
user("ANY")
11.
Return True if the client has more than one set of current credentials
(i.e., has authenticated as two or more identities)
user("any") gt 1
12.
Return True if the client was authenticated as rmorriso by this
jurisdiction
user(":rmorriso")
13.
Return True if the client was authenticated as rmorriso,
case-insensitively, by this jurisdiction
user(":rmorriso nocase")
14.
Return True if the client was authenticated as the user
metalogic:RMORRISO, but comparing the jurisdiction name, username, and implied
federation name case-insensitively
user("metalogic:RMORRISO nocase")
15.
Equivalent to user("METALOGIC:rmorriso"), return True if the
client was authenticated as the user METALOGIC:rmorriso, but comparing the
jurisdiction name, username, and implied federation name according to the
NAME_COMPARE directive, otherwise using the application's default
user("METALOGIC:rmorriso default")
user("auth")
user("DSS:auth")
This function generates a string called a
stamp that is globally unique and sequenced, with high probability. It
has the following syntax:
A hostid consists of one or more characters from the same set used for a
DACS username[119]. A seqno consists of two elements, separated
by a colon, each of which is an unsigned decimal value.
The first component of a stamp, the hostid, is intended to be uniquely
associated with the host that generates the stamp. By default, it is a
128-bit, cryptographically strong pseudo-random value. This value is stored in
vfs-ref, which may be an absolute pathname, an item type, or a VFS
URI[61]. If vfs-ref does not exist, it is created and a new value
is stored in it.
Note that by default, hostid identifies a host, not a jurisdiction. If
required, it is possible to configure unique stamps for each jurisdiction on a
host.
The second component ( seqno) is a sequence number string relative to
hostid. Sequence numbers should never repeat with respect to a host and
always increase in value so that any two sequence numbers created by the same
host must be different. Successive sequence numbers need not increase by
uniform steps. If stamp1 compares less than stamp2, then
stamp1 was created before stamp2. Comparison of sequence numbers
is performed on matching elements numerically, left to right. Two
hostid components are compared case insensitively. No ordering is
necessarily implied by stamps created by different hosts.
Sequence number state information is stored in a file that must be specified
using the configuration variable ${Conf::ustamp_seqno}; e.g.,
The variable must be set to the absolute pathname of a file that is readable and
writable by any process that needs to generate a stamp. If this file is
deleted, the sequence will be reinitialized. Note that updates to the state
information are unlikely to be atomic, which means that in the event of a
system crash the state information should be deleted so that a new stream of
sequence numbers is generated.
One application of these stamps is to provide an efficient way to detect
replayed messages. A recipient may only need to keep track of the stamp sent
with the last message received from a jurisdiction to detect an invalid stamp
in any subsequent message. Cryptographic techniques can be employed to prevent
a stamp from being altered or forged.
The following operations are recognized:
ustamp(clock, vfs-uri)
Examples:
valuesof(alist)
h= hostid, s=seqno
EVAL ${Conf::ustamp_seqno} = "${Conf::DACS_HOME}/seqno"
The host's system clock is used for the
stamp's sequence number. Its first element is the number of seconds since the
start of the epoch and the second is a counter value. Note that if the system
clock is reset to an earlier time, sequence numbers may repeat with
unpredictable consequences; a future version of this function may detect a
reset clock.
ustamp(ntpclock, vfs-uri, ntp_host)
This operation is not implemented.
Rather than using the system clock, this operation obtains the current time
from ntp_host, which is assumed to be more reliable than the system
clock in that it will never be reset to an earlier time. The ntp_host
argument is a hostname or IP address. The default port number (123) may be
overridden by appending a colon and the port number to use.
ustamp(user, vfs-uri, seqno)
Instead of incorporating the current time into
the stamp's sequence number, this operation uses a user-supplied string that
is assumed to have the necessary syntax and characteristics.
> ustamp(clock, "${Conf::DACS_HOME}/hostid")
"h=2fbae312ddc1d2ae388cea1b57a47c66, s=1185565675:9"
If its argument is a single pair, the pair's
value is returned. If there is more than one pair in the argument, a list of
values is returned. To get the key component of a pair or set of pairs, use
keysof()[17].
Examples:
var(op, namespace, variable-name [, args
...])
> valuesof({red, 17})
17
> valuesof({red, 17, blue, 100})
[17, 100]
This function performs various operations on a
variable, some of which are awkward or impossible to do using the more concise
variable reference syntax. For example, the namespace or variable name
argument to var() can be specified by an expression.
The following operations are available:
var(delete, namespace, variable-name)
Examples:
vfs(op, vfs-ref [, argument ...])
Delete (undefine) the variable named
variable-name within namespace. If the variable is deleted, 1 is
returned, and if it does not exist, 0 is returned.
var(exists, namespace, variable-name)
Test if the variable named
variable-name within namespace exists, returning 1 if so and 0
if not.
var(get, namespace, variable-name [, altval])
Return the value of the variable named
variable-name within namespace. If the variable does not exist,
altval is returned if given, otherwise the empty string is returned
(which could potentially be confused with a legitimate value).
var(set, namespace, variable-name, value)
Set the value of the variable named
variable-name within namespace to value. If
namespace or variable-name do not exist they are created. If the
variable already exists, its value is replaced. The function returns
value.
> ${Y::foo} = 17
17
> setvar(split, X, "/a/b/c/Y", "/")
5
> var(get, X, 4)
"Y"
> var(get, X, ${X::#} - 1)
"Y"
> var(get, var(get, X, "4"), "foo")
"17"
> var(set, Y, "f" . "o" . "o", 2007)
2007
> ${Y::foo}
2007
This function is an interface to the
DACS virtual filestore subsystem, described in dacs.vfs(5)[120].
Please refer to dacs.conf(5)[25] and dacsvfs(1)[121] for details
and examples.
The first argument specifies the operation to be performed. The second argument
identifies a filestore (typically a file or database); it can be an absolute
pathname, an item_type that has been configured through a VFS[25]
directive, or a VFS URI[61]. Zero or more arguments may follow,
depending on op. For most operations, the third argument will be the
key that identifies the object of interest. The underlying filestore is
implicitly opened and closed.
An operation that fails abnormally triggers a fatal error.
The following operations ( op) are available:
vfs(control, vfs-ref, c_op [, argument]
vfs(put, vfs-ref, key, value)
This statement sets a variable to the contents of the file /tmp/somefile:
As do this equivalent statements:
This expression lists the files in the /tmp directory:
These expressions 1) add a key/value pair to a Berkeley DB database
(/tmp/mydb.db), creating the database file if necessary, 2) retrieve the
value, 3) rename the key, and 4) list the keys in the database:
This rule fragment denies access if the user has already been granted access
five times:
The item type counter_db would be configured in dacs.conf; e.g.,
Perform a configuration operation on the
underlying storage scheme. Returns True.
vfs(defined, item-type)
Test if the specified item-type has
been defined by a VFS[61] directive.
vfs(delete, vfs-ref [, key])
vfs(defined, "passwds")
Delete the referenced object.
vfs(enabled [, store-name])
With an argument, test if the specified
store-name can be used. With no argument, return a list[68] of
enabled store names.
vfs(exists, vfs-ref [, key])
vfs(enabled, "db") ? print("yes") : print("no");
Test whether the referenced object exists,
returning True or False.
vfs(get, vfs-ref [, key])
vfs(exists, "/usr/local/dacs/conf/passwd") vfs(exists, "file:///usr/local/dacs/conf/passwd")
Retrieve the referenced object.
vfs(getsize, vfs-ref [, key])
Return the length, in bytes, of the referenced
object.
vfs(list, vfs-ref)
List the keys of all objects in the
store.
vfs(put, vfs-ref, value)
Store an item under the given key, replacing
any existing instance. The value is null-terminated.
vfs(rename, vfs-ref, oldkey, newkey)
Change the key associated with an existing
item from oldkey to newkey.
vfs(uri, item-type)
If item-type has been defined by a
VFS[61] directive, return its URI, otherwise the empty string.
> vfs(uri, "passwds") "[passwds]dacs-kwv-fs:/usr/local/dacs/conf/passwd?field_sep=:"
${somefile} = vfs(get, "file:///tmp/somefile")
${somefile} = vfs(get, "/tmp/somefile")
${somefile} = get("/tmp/somefile")
vfs(list,"dacs-fs:/tmp")
vfs(put, "dacs-db:/tmp/mydb.db", "foo", "baz"); vfs(get, "dacs-db:/tmp/mydb.db", "foo"); vfs(rename, "dacs-db:/tmp/mydb.db", "foo", "bar"); vfs(list, "dacs-db:/tmp/mydb.db");
<deny>
if (user("auth")) {
if (vfs(exists, counter_db, ${DACS::IDENTITY})) {
${count} = vfs(get, counter_db, ${DACS::IDENTITY});
}
else {
${count} = 0;
}
if (${count} gt 5) {
return(1);
}
vfs(put, counter_db, ${DACS::IDENTITY}, ++${count});
return(0);
}
</deny>
VFS "[counter_db]dacs-db:/usr/local/dacs/federations/counters.db"
SEE ALSO¶
dacsexpr(1)[1]BUGS¶
Assorted clunky aspects of the language are likely to be replaced by simplified or more general approaches once requirements are clearer. The list and alist data types have not been fully developed and integrated. Assignment to a namespace would be a useful extension. A way to handle errors and exceptions (such as with try/catch/throw statements) would be nice. A switch statement and dynamically loaded functions are planned. A foreach statement might be useful, although the language has so far successfully avoided loop constructs as a way to limit its complexity. Various aspects of variables and namespaces are not implemented. A namespace cannot be copied by assignment; use setvar(). Input and output processing is still rather limited. Having to use ":" instead of ".." when matching octet ranges with from()[113] is unfortunate but avoids pesky period proliferation. Some of the more esoteric functions and modes of operation exist primarily to expose DACS core code for testing purposes.AUTHOR¶
Distributed Systems Software ( www.dss.ca[122])COPYING¶
Copyright2003-2012 Distributed Systems Software. See the LICENSE[123] file that accompanies the distribution for licensing information.NOTES¶
- 1.
- dacsexpr(1)
- 2.
- Perl
- 3.
- PHP
- 4.
- Tcl
- 5.
- isprint(3)
- 6.
- print()
- 7.
- cast
- 9.
- dacs.conf(5)
- 10.
- dacs_acs(8)
- 11.
- environ(7)
- 12.
- exec()
- 13.
- list()
- 14.
- length()
- 15.
- strchars()
- 16.
- expression grammar
- 17.
- keysof()
- 18.
- valuesof()
- 19.
- alist()
- 20.
- supported data types
- 21.
- variable reference
- 22.
- dacs_notices(8)
- 23.
- alist construction operator
- 24.
- listref()
- 25.
- VFS
- 26.
- revocation list
- 27.
- access control rule
- 28.
- on_success()
- 29.
- AUTH_SUCCESS
- 30.
- ACS_SUCCESS
- 31.
- ADMIN_IDENTITY
- 32.
- approval stamp
- 33.
- dacs64 decoding
- 35.
- dacsauth(1)
- 36.
- dacscheck(1)
- 37.
- encode()
- 38.
- cryptographic hash
- 39.
- hash()
- 40.
- decode()
- 41.
- radix-85
- 42.
- RFC 2045
- 43.
- RFC 1738
- 44.
- RFC 2396
- 45.
- RFC 3986
- 46.
- execv(3)
- 47.
- basename(1)
- 48.
- dirname(1)
- 49.
- stat
- 50.
- stat(1)
- 51.
- stat(2)
- 52.
- printf(3)
- 53.
- test(1)
- 54.
- mod_access
- 55.
- RFC 1035
- 56.
- CIDR notation
- 57.
- RFC 1338
- 58.
- range specification
- 59.
- regmatch()
- 60.
- user()
- 61.
- vfs_uri
- 62.
- digest()
- 63.
- message authentication code
- 64.
- Keyed-Hash Message Authentication Code (HMAC)
- 65.
- Secure Hash Standard functions
- 66.
- MD5 (RFC 2104)
- 67.
- RFC 2253
- 68.
- list construction operator
- 69.
- dacspasswd(1)
- 70.
- PASSWORD_DIGEST
- 71.
- PASSWORD_SALT_PREFIX
- 72.
- local_passwd_authenticate
- 73.
- vfs()
- 74.
- PASSWORD_CONSTRAINTS
- 75.
- dacs.conf(5)
- 76.
- RFC 2898
- 77.
- RFC 3962
- 78.
- LOG_FILTER
- 79.
- cryptographically strong pseudo-random values
- 80.
- strtr()
- 81.
- ACS_ERROR_HANDLER
- 82.
- ErrorDocument directive
- 83.
- Rlinks
- 84.
- permalinks
- 85.
- regex(3)
- 86.
- re_format(7)
- 87.
- access control rules
- 88.
- read-only namespace
- 89.
- RFC 2617
- 90.
- copy
- 91.
- query
- 92.
- cgiparse(8)
- 93.
- split
- 94.
- sleep(3)
- 95.
- sprintf(3)
- 96.
- strftime(3)
- 97.
- strptime(3)
- 98.
- tr(1)
- 99.
- DACS name
- 00.
- RFC 822
- 01.
- RFC 952
- 02.
- FEDERATION_NAME
- 03.
- RFC 1123
- 04.
- RFC 790
- 05.
- JURISDICTION_NAME
- 06.
- strptime()
- 07.
- localtime(3)
- 08.
- dacstransform(1)
- 09.
- transform_config()
- 10.
- concise user syntax
- 11.
- transform()
- 12.
- ACS_CREDENTIALS_LIMIT
- 13.
- from()
- 14.
- dacs.groups(5)
- 15.
- styles
- 18.
- dacscookie(1)
- 19.
- DACS username
- 20.
- dacs.vfs(5)
- 21.
- dacsvfs(1)
- 22.
- www.dss.ca
- 23.
- LICENSE
| 10/22/2012 | DACS 1.4.27b |