Name Manipulation Methods

Methods to do various operations on Key names. More...

Collaboration diagram for Name Manipulation Methods:

Functions
const char *	keyName (const Key *key)
	Returns a pointer to the abbreviated real internal key name.
ssize_t	keyGetNameSize (const Key *key)
	Bytes needed to store the key name without owner.
ssize_t	keyGetName (const Key *key, char *returnedName, size_t maxSize)
	Get abbreviated key name (without owner name).
ssize_t	keySetName (Key *key, const char *newName)
	Set a new name to a key.
ssize_t	keyGetFullNameSize (const Key *key)
	Bytes needed to store the key name including user domain and ending NULL.
ssize_t	keyGetFullName (const Key *key, char *returnedName, size_t maxSize)
	Get key full name, including the user domain name.
const char *	keyBaseName (const Key *key)
	Returns a pointer to the internal unescaped key name where the basename starts.
ssize_t	keyGetBaseNameSize (const Key *key)
	Calculates number of bytes needed to store basename of key.
ssize_t	keyGetBaseName (const Key *key, char *returned, size_t maxSize)
	Calculate the basename of a key name and put it in returned finalizing the string with NULL.
ssize_t	keyAddBaseName (Key *key, const char *baseName)
	Adds baseName (that will be escaped) to the current key name.
ssize_t	keyAddName (Key *key, const char *newName)
	Add a already escaped name to the keyname.
ssize_t	keySetBaseName (Key *key, const char *baseName)
	Sets baseName as the new basename for key.

Detailed Description

Methods to do various operations on Key names.

To use them:

#include <kdb.h>

These functions make it easier for C programmers to work with key names.

Terminology of Key Names

• A key name (see keySetName() and keyName()) defines the place of a key within the key database. To be unique, it is always absolute and canonical.
• Key names are composed out of many key name parts split by a separator. These key name parts do not contain a unescaped separator.
• A key base name (see keySetBaseName() and keyAddBaseName()) is the last part of the key name.
• A C-String is a null terminated sequence of characters. So \0 (null-character) must not occur within a C-String.

Namespaces

A namespace denotes the place the key comes from:

• spec/something for specification of other keys
• proc/something for in-memory keys (e.g. command-line)
• dir/something for dir keys in current working directory
• system/something for system keys in /etc or /
• user/something for user keys in home directory
• user:username/something for other users (deprecated: kdbGet() + kdbSet() currently unsupported)
• /something for cascading keys (actually refers to one of the above, see also ksLookup())

Note

The rules are currently not formally specified and are subject of change in the next major release. So, always prefer:

• To use keySetName() and keyAddName() to get the canonified version of the keyname
• To use keySetBaseName() and keyAddBaseName() to get an escaped key name part.
• Not to escape or canonify with your own algorithms!
• To use keyUnescapedName() and keyBaseName() to have access to the key name without escape sequences (key name parts are null terminated)
• Not to unescape the strings yourself!

Syntax for Key Names

Key names and key name parts have following goals:

• The C-String passed to keySetName() and keyAddName() may be any C-String.
• The key name parts (e.g. keySetBaseName(), keyBaseName()) may be any C-String. Escaping is needed to achieve both goals.

Semantics for Key Name Parts

• % denotes an empty key name part.

Canonicalization for Key Names

• / (slash) is the separator between key name parts.
• // is shortened to /
• trailing / (slashes) are removed
• . (dot) and .. (dot-dot) is removed in an canonical key name, with following rules:
  • /./ is shortened to /
  • _/../ is shortened to _

Conventions for key names

• Key name parts starting with # are array elements. Then only _ (underscore) followed by 0-9 is allowed. So we have the regular expression #[_]*[0-9]+ with the further limitation that the number of _ is defined by the number of digits-1.
• Key name parts starting with _ are reserved for special purposes (if you use this within a plugin you still have to make sure _ is escaped properly)
• Key name parts starting with @ are reserved for special purposes (if you use this within a plugin you still have to make sure @ is escaped properly)
• If any key name part starts with . (dot) it means the key is inactive, see keyIsInactive().

Escaping rules

• \ (backslash) is the escape character for the situations as described here (and only these). The \ character must only be escaped, when one of the following rules apply.
• Stray escape characters are only possible in the end of the string.
• \/ allows to escape / (any uneven number of \). Does not introduce a new part.
• Any uneven number N of \ before / allows you to escape / with the N/2 of \ prefixed. Does not introduce a new part.
• \\/ allows to use \ as character before / and introduces a new part.
• Any even number N of \ before / allows you to have N/2 of \ prefixed before a / which introduces a new part.
• Use and . if you want your key name part to represent . and ..
• \ and \. allows us to use \ as character before . and .. (and so on)
• Use \% if you want your key name part to start with % (and does not represent an empty name)
• Use \\% allows to use \ as character before % (and so on)

Semantics for Key Name Specifications

• _ denotes that the key name part is arbitrary (syntax as described above).
• # denotes that the key name part has array syntax.
• names surrounded by % (e.g. %profile%) denotes a placeholder.

Usage of Key Names

When using Elektra to store your application's configuration and state, please keep in mind the following rules:

• Avoid to have your applications root right under system or user. (rationale: it would make the hierarchy too flat.)
• Avoid the usage of characters other then a-z, 0-9 and _. (rationale: it would allow too many similar, confusing names.) (exceptions: if the user or a technology, decide about parts of the key name, this restriction does not apply, e.g. if the wlan essid is used as part of the key name)
• It is suggested to make your application look for default keys under /sw/myapp/#/%/ where # is a major version number, e.g. #3 for the 4th version and % is a profile (% for default profile). This way, from a sysadmin perspective, it will be possible to copy the system/sw/myapp/#3/%/ tree to something like system/sw/myapp/#3/old/ and keep system clean and organized. Additionally, it is possible to start the old version of the app, using /sw/myapp/#2.

Function Documentation

ssize_t keyAddBaseName	(	Key *	key,
		const char *	baseName
	)

Adds baseName (that will be escaped) to the current key name.

A new baseName will be added, no other part of the key name will be affected.

Assumes that key is a directory and will append baseName to it. The function adds the path separator for concatenating.

So if key has name "system/dir1/dir2" and this method is called with baseName "mykey", the resulting key will have the name "system/dir1/dir2/mykey".

When baseName is 0 nothing will happen and the size of the name is returned.

The escaping rules apply as in above .

A simple example is:

Key * k = keyNew("user/my/long", KEY_END);
keyAddBaseName(k, "myname");
printf ("%s\n", keyName(k)); // will print user/my/long/myname
keyDel(k);

E.g. if you add . it will be escaped:

keySetName (k, "system/valid");
succeed_if (keyAddBaseName (k, ".") >= 0, "could not add a base name");
succeed_if_same_string(keyName(k), "system/valid/\\.");
succeed_if_same_string(keyBaseName(k), ".");

See Also

keySetBaseName() to set a base name

keySetName() to set a new name.

Parameters

key	the key object to work with
baseName	the string to append to the name

Returns

the size in bytes of the new key name including the ending NULL

-1 if the key had no name

-1 on NULL pointers

Return values

-1	if key was inserted to a keyset before

ssize_t keyAddName	(	Key *	key,
		const char *	newName
	)

Add a already escaped name to the keyname.

The same way as in keySetName() this method finds the canonical pathname:

• it will ignore /./
• it will remove a level when /../ is used
• it will remove multiple slashes ////

For example:

Key *k = keyNew("user/x/r", KEY_END);
keyAddName(k, "../y/a//././z");
assert(!strcmp(keyName(k), "user/x/a/z"));
keyDel(k);

Unlike keySetName() it adds relative to the previous name and cannot change the namespace of a key. For example:

Key *n = keyNew("user/away", KEY_END);
keyAddName(n, "../../../new/name");
assert(!strcmp(keyName(n), "user/new/name"));
keyDel(n);

The passed name needs to be valid according the key name rules . It is not allowed to:

• be empty
• end with unequal number of \

Parameters

key	the key where a name should be added
newName	the new name to append

Since

0.8.11

Return values

size	of the new key
-1	if key is a null pointer or did not have a valid name before
-1	if newName is not a valid escaped name
-1	on allocation errors
-1	if key was inserted to a keyset before
0	if nothing was done because newName had only slashes, is too short, is empty or is null

const char* keyBaseName

(

const Key *

key

)

Returns a pointer to the internal unescaped key name where the basename starts.

This is a much more efficient version of keyGetBaseName() and you should use it if you are responsible enough to not mess up things. The name might change or even point to a wrong place after a keySetName(). So make sure to copy the memory before the name changes.

keyBaseName() returns "" when there is no keyBaseName. The reason is

keySetName(k,"");
succeed_if_same_string(keyBaseName(k), "");
keySetName(k,"user");
succeed_if_same_string(keyBaseName(k), "");

And there is also support for really empty basenames:

keySetName (k, "system/valid");
succeed_if (keyAddBaseName (k, "") >= 0, "could not add a base name");
succeed_if_same_string(keyName(k), "system/valid/%");
succeed_if_same_string(keyBaseName(k), "");

Note

You must never use the pointer returned by keyBaseName() method to change the name, but you should use keySetBaseName() instead.

Do not assume that keyBaseName() points to the same region as keyName() does.

Parameters

key	the object to obtain the basename from

Returns

a pointer to the basename

Return values

""	when the key has no (base)name
0	on NULL pointer

See Also

keyGetBaseName(), keyGetBaseNameSize()

keyName() to get a pointer to the name

keyOwner() to get a pointer to the owner

ssize_t keyGetBaseName	(	const Key *	key,
		char *	returned,
		size_t	maxSize
	)

Calculate the basename of a key name and put it in returned finalizing the string with NULL.

Some examples:

• basename of system/some/keyname is keyname
• basename of "user/tmp/some key" is "some key"

Parameters

key	the key to extract basename from
returned	a pre-allocated buffer to store the basename
maxSize	size of the returned buffer

Returns

number of bytes copied to returned

1 on empty name

-1 on NULL pointers

-1 when maxSize is 0 or larger than SSIZE_MAX

See Also

keyBaseName(), keyGetBaseNameSize()

keyName(), keyGetName(), keySetName()

ssize_t keyGetBaseNameSize

(

const Key *

key

)

Calculates number of bytes needed to store basename of key.

Key names that have only root names (e.g. "system" or "user" or "user:domain" ) does not have basenames, thus the function will return 1 bytes to store "".

Basenames are denoted as:

• system/some/thing/basename -> basename
• user:domain/some/thing/base\/name > base\/name

Parameters

key	the key object to work with

Returns

size in bytes of key's basename including ending NULL

See Also

keyBaseName(), keyGetBaseName()

keyName(), keyGetName(), keySetName()

ssize_t keyGetFullName	(	const Key *	key,
		char *	returnedName,
		size_t	maxSize
	)

Get key full name, including the user domain name.

Returns

number of bytes written

1 on empty name

-1 on NULL pointers

-1 if maxSize is 0 or larger than SSIZE_MAX

Parameters

key	the key object
returnedName	pre-allocated memory to write the key name
maxSize	maximum number of bytes that will fit in returnedName, including the final NULL

ssize_t keyGetFullNameSize

(

const Key *

key

)

Bytes needed to store the key name including user domain and ending NULL.

Parameters

key	the key object to work with

Returns

number of bytes needed to store key name including user domain

1 on empty name

-1 on NULL pointer

See Also

keyGetFullName(), keyGetNameSize()

ssize_t keyGetName	(	const Key *	key,
		char *	returnedName,
		size_t	maxSize
	)

Get abbreviated key name (without owner name).

When there is not enough space to write the name, nothing will be written and -1 will be returned.

maxSize is limited to SSIZE_MAX. When this value is exceeded -1 will be returned. The reason for that is that any value higher is just a negative return value passed by accident. Of course malloc is not as failure tolerant and will try to allocate.

char *getBack = malloc (keyGetNameSize(key));
keyGetName(key, getBack, keyGetNameSize(key));

Returns

number of bytes written to returnedName

1 when only a null was written

-1 when keyname is longer then maxSize or 0 or any NULL pointer

Parameters

key	the key object to work with
returnedName	pre-allocated memory to write the key name
maxSize	maximum number of bytes that will fit in returnedName, including the final NULL

See Also

keyGetNameSize(), keyGetFullName(), keyGetFullNameSize()

ssize_t keyGetNameSize

(

const Key *

key

)

Bytes needed to store the key name without owner.

For an empty key name you need one byte to store the ending NULL. For that reason 1 is returned.

Parameters

key	the key object to work with

Returns

number of bytes needed, including ending NULL, to store key name without owner

1 if there is is no key Name

-1 on NULL pointer

See Also

keyGetName(), keyGetFullNameSize()

const char* keyName

(

const Key *

key

)

Returns a pointer to the abbreviated real internal key name.

This is a much more efficient version of keyGetName() and can use it if you are responsible enough to not mess up things. You are not allowed to change anything in the returned array. The content of that string may change after keySetName() and similar functions. If you need a copy of the name, consider using keyGetName().

The name will be without owner, see keyGetFullName() if you need the name with its owner.

keyName() returns "" when there is no keyName. The reason is

key=keyNew(0);
keySetName(key,"");
keyName(key); // you would expect "" here
keyDel(key);

Valid key names are:

• spec/something for specification of other keys
• proc/something for in-memory keys (e.g. command-line)
• dir/something for dir keys in current working directory
• system/something for system keys in /etc or /
• user/something for user keys in home directory
• user:username/something for other users (deprecated: kdbGet() + kdbSet() currently unsupported)
• /something for cascading keys (actually refers to one of the above, see also ksLookup())

Note

Note that the Key structure keeps its own size field that is calculated by library internal calls, so to avoid inconsistencies, you must never use the pointer returned by keyName() method to set a new value. Use keySetName() instead.

Parameters

key	the key object to work with

Returns

a pointer to the keyname which must not be changed.

"" when there is no (a empty) keyname

0 on NULL pointer

See Also

keyGetNameSize() for the string length

keyGetFullName(), keyGetFullNameSize() to get the full name

keyGetName() as alternative to get a copy

keyOwner() to get a pointer to owner

ssize_t keySetBaseName	(	Key *	key,
		const char *	baseName
	)

Sets baseName as the new basename for key.

Only the baseName will be affected and no other part of the key.

All text after the last '/' in the key keyname is erased and baseName is appended.

So let us suppose key has name "system/dir1/dir2/key1". If baseName is "key2", the resulting key name will be "system/dir1/dir2/key2". If baseName is empty or NULL, the resulting key name will be "system/dir1/dir2".

This function does proper escaping on the supplied name argument.

You can use all names to set as basename (e.g. . (dot), .. (dot-dot), % and "" (empty)). They will be properly escaped.

A simple example is:

Key * k = keyNew("user/my/long/name", KEY_END);
keySetBaseName(k, "myname");
printf ("%s\n", keyName(k)); // will print user/my/long/myname
keyDel(k);

If you want to add and not change the basename, use keyAddBaseName() instead. If you do not want escaping, use keyAddName() instead.

To add an inactive key name, use:

        keySetName (k, "system/valid");
        keySetBaseName(k, ".hiddenkey");
        succeed_if_same_string(keyName(k), "system/.hiddenkey");
        succeed_if_same_string(keyBaseName(k), ".hiddenkey");

When you want to add an array item, use:

        keySetName (k, "system/valid");
        keySetBaseName(k, "");
        succeed_if_same_string(keyName(k), "system/%");
        succeed_if_same_string(keyBaseName(k), "");

See Also

Name Manipulation Methods for more details on special names

Parameters

key	the key object to work with
baseName	the string used to overwrite the basename of the key

Returns

the size in bytes of the new key name

-1 on NULL pointers

Return values

-1	if key was inserted to a keyset before

See Also

keyAddBaseName()

keySetName() to set a new name

ssize_t keySetName	(	Key *	key,
		const char *	newName
	)

Set a new name to a key.

A valid name is one of the forms:

• spec/something for specification of other keys
• proc/something for in-memory keys (e.g. command-line)
• dir/something for dir keys in current working directory
• system/something for system keys in /etc or /
• user/something for user keys in home directory
• user:username/something for other users (deprecated: kdbGet() + kdbSet() currently unsupported)
• /something for cascading keys (actually refers to one of the above, see also ksLookup())

An invalid name either has an invalid namespace or a wrongly escaped \ at the end of the name.

See key names for the exact rules.

The last form has explicitly set the owner, to let the library know in which user folder to save the key. A owner is a user name. If it is not defined (the second form) current user is used.

You should always follow the guidelines for key tree structure creation.

A private copy of the key name will be stored, and the newName parameter can be freed after this call.

.., . and / will be handled as in filesystem pathes. A valid name will be build out of the (valid) name what you pass, e.g. user///sw/../sw//././MyApp -> user/sw/MyApp

On invalid names, NULL or "" the name will be "" afterwards.

Return values

size	in bytes of this new key name including ending NULL
0	if newName is an empty string or a NULL pointer (name will be empty afterwards)
-1	if newName is invalid (name will be empty afterwards)
-1	if key was inserted to a keyset before

Parameters

key	the key object to work with
newName	the new key name

See Also

keyNew(), keySetOwner()

keyGetName(), keyGetFullName(), keyName()

keySetBaseName(), keyAddBaseName() to manipulate a name