Scroll to navigation

key(3elektra) Elektra key(3elektra)

NAME

key - Key

Key is an essential class that encapsulates key name , value and metainfo .

Modules


Meta Info Manipulation Methods
Methods to do various operations on Key metainfo. Methods for Making Tests
Methods to do various tests on Keys. Name Manipulation Methods
Methods to do various operations on Key names. Value Manipulation Methods
Methods to do various operations on Key values.

Enumerations

Functions


Key * keyNew (const char *name,...)
A practical way to fully create a Key object in one step. Key * keyDup (const Key *source)
Return a duplicate of a key. int keyCopy (Key *dest, const Key *source)
Copy or Clear a key. int keyDel (Key *key)
A destructor for Key objects. int keyClear (Key *key)
Key Object Cleaner. ssize_t keyIncRef (Key *key)
Increment the viability of a key object. ssize_t keyDecRef (Key *key)
Decrement the viability of a key object. ssize_t keyGetRef (const Key *key)
Return how many references the key has.

Detailed Description

Key is an essential class that encapsulates key name , value and metainfo .

To use it include:

#include <kdb.h>

Key properties are:

  • Key name
  • Key value
  • Key meta data , including but not limited to:
Key comment
Key owner
UID, GID and filesystem-like mode permissions
Mode, change and modification times

ABI

Due to ABI compatibility, the Key structure is not defined in kdb.h, only declared. So you can only declare pointers to Keys in your program, and allocate and free memory for them with keyNew() and keyDel() respectively.

Reference Counting

Every key has its reference counter (see keyGetRef() for longer explanation) that will be initialized with 0, that means a subsequent call of keyDel() will delete the key. If you append the key to a keyset the reference counter will be incremented by one (see keyIncRef()) and the key can't be be deleted by a keyDel().

As you can imagine this refcounting allows you to put the Key in your own datastructures. It can be a very powerful feature, e.g. if you need your own-defined ordering or different Models of your configuration.

Enumeration Type Documentation

enum keyswitch_t

Allows keyNew() to determine which information comes next.

See also:

keyNew()

Enumerator

KEY_NAME
Flag for the key name
KEY_VALUE
Flag for the key data
KEY_OWNER
Flag for the key user domain
KEY_COMMENT
Flag for the key comment
KEY_BINARY
Flag if the key is binary
KEY_UID
Flag for the key UID

Deprecated

do not use
KEY_GID
Flag for the key GID

Deprecated

do not use
KEY_MODE
Flag for the key permissions

Deprecated

do not use
KEY_ATIME
Flag for the key access time

Deprecated

do not use
KEY_MTIME
Flag for the key change time

Deprecated

do not use
KEY_CTIME
Flag for the key status change time

Deprecated

do not use
KEY_SIZE
Flag for maximum size to limit value
KEY_DIR
Flag for the key directories

Deprecated

do not use
KEY_META
Flag for meta data
KEY_END
Used as a parameter terminator to keyNew()

Function Documentation

int keyClear (Key * key)

Key Object Cleaner. Will reset all internal data.

After this call you will receive a fresh key.

The reference counter will stay unmodified.

Note:

that you might also clear() all aliases with this operation.

1 int f (Key *k)
2 {
3         keyClear (k);
4         // you have a fresh key k here
5         keySetString (k, "value");
6         // the caller will get an empty key k with an value
7 }

Return values:

returns 0 on success
-1 on null pointer

Parameters:

key the key object to work with

int keyCopy (Key * dest, const Key * source)

Copy or Clear a key. Most often you may prefer keyDup() which allocates a new key and returns a duplication of another key.

But when you need to copy into an existing key, e.g. because it was passed by a pointer in a function you can do so:

void h (Key *k)
{
        // receive key c
        keyCopy (k, c);
        // the caller will see the changed key k
}

The reference counter will not be changed for both keys. Affiliation to keysets are also not affected.

The meta data will be duplicated for the destination key. So it will not take much additional space, even with lots of metadata.

When you pass a NULL-pointer as source the data of dest will be cleaned completely (except reference counter, see keyClear()) and you get a fresh dest key:

void g (Key *k)
{
        keyCopy (k, 0);
        // k is now an empty and fresh key
}

If you want to copy everything, except e.g. the value you can use keyCopy() too:

void j (Key *k)
{
        size_t size = keyGetValueSize (k);
        char *value = malloc (size);
        int bstring = keyIsString (k);
        // receive key c
        memcpy (value, keyValue(k), size);
        keyCopy (k, c);
        if (bstring) keySetString (k, value);
        else keySetBinary (k, value, size);
        free (value);
        // the caller will see the changed key k
        // with the name and metadata from c (except
        // metadata "binary", which stayed the same)
}

Restrain from coping everything yourself, because it will lead to wrong metadata and is not able to copy empty or cascading names:

void i (Key *k)
{
        keySetName(k, keyName(c));
        keySetString(k, keyString(c));
        keyCopyAllMeta(k, c);
        // k is not a copy of c even if everything was successfully,
        // because it still contains meta data from k
}

Parameters:

dest the key which will be written to
source the key which should be copied or NULL to clean the destination key

Return values:

-1 on failure when a NULL pointer was passed for dest or a dynamic property could not be written. The content will be unmodified then.
0 when dest was cleaned
1 when source was successfully copied

See also:

keyDup() to get a duplication of a Key

ssize_t keyDecRef (Key * key)

Decrement the viability of a key object. The references will be decremented for ksPop() or successful calls of ksLookup() with the option KDB_O_POP. It will also be decremented with an following keyDel() in the case that an old key is replaced with another key with the same name.

The reference counter can't be decremented once it reached 0. In that situation nothing will happen and 0 will be returned.

Note:

keyDup() will reset the references for dupped key.

Returns:

the value of the new reference counter

Return values:

-1 on null pointer
0 when the key is ready to be freed

Parameters:

key the key object to work with

See also:

keyGetRef() for longer explanation, keyDel(), keyIncRef()

int keyDel (Key * key)

A destructor for Key objects. Every key created by keyNew() must be deleted with keyDel().

It is save to delete keys which are in a keyset, the number of references will be returned then.

It is save to delete a nullpointer, -1 will be returned then.

It is also save to delete a multiple referenced key, nothing will happen then and the reference counter will be returned.

Parameters:

key the key object to delete

See also:

keyNew(), keyIncRef(), keyGetRef()

Returns:

the value of the reference counter if the key is within keyset(s)

Return values:

0 when the key was freed
-1 on null pointers

Key* keyDup (const Key * source)

Return a duplicate of a key. Memory will be allocated as needed for dynamic properties.

The new key will not be member of any KeySet and will start with a new reference counter at 0. A subsequent keyDel() will delete the key.

1 int f (const Key * source)
2 {
3         Key * dup = keyDup (source);
4         // work with duplicate
5         keyDel (dup);
6         // everything related to dup is freed
7         // and source is unchanged
8 }

Like for a new key after keyNew() a subsequent ksAppend() makes a KeySet to take care of the lifecycle of the key.

1 int g (const Key * source, KeySet * ks)
2 {
3         Key * dup = keyDup (source);
4         // work with duplicate
5         ksAppendKey (ks, dup);
6         // ksDel(ks) will also free the duplicate
7         // source remains unchanged.
8 }

Duplication of keys should be preferred to keyNew(), because data like owner can be filled with a copy of the key instead of asking the environment. It can also be optimized in the checks, because the keyname is known to be valid.

Parameters:

source has to be an initialized source Key

Return values:

0 failure or on NULL pointer

Returns:

a fully copy of source on success

See also:

ksAppend(), keyDel(), keyNew()

ssize_t keyGetRef (const Key * key)

Return how many references the key has. The reference counting is the essential property of keys to make sure that they can be put safely into data structures. E.g. if you put a Key into a KeySet:

Key *k = keyNew("user/proper_name", KEY_END); // ref counter = 0
KeySet *ks = ksNew (1, k, KS_END);
keyDel(k); // key will not be deleted, because its in the keyset
ksDel(ks); // now the key will be deleted

You can even add the key to more KeySets:

Key *k = keyNew("user/proper_name", KEY_END); // ref counter 0
KeySet *ks1 = ksNew(1, k, KS_END); // ref counter of k 1
KeySet *ks2 = ksNew(1, k, KS_END); // ref counter of k 2
ksDel(ks1); // ref counter of k 1
ksDel(ks2); // k is now deleted

If you increment only by one with keyIncRef() the same as said above is valid:

Key *k = keyNew(0); // ref counter = 0
keyIncRef(k); // ref counter = 1
keyDel(k); // key will not be deleted
keyDecRef(k);
keyDel(k);

or use keyIncRef() more than once:

Key *k = keyNew(0); // ref counter 0
keyIncRef(k); // ref counter of key 1
keyDel (k);   // has no effect
keyIncRef(k); // ref counter of key 2
keyDel (k);   // has no effect
keyDecRef(k); // ref counter of key 1
keyDel (k);   // has no effect
keyDecRef(k); // ref counter is now 0
keyDel (k); // k is now deleted

The key won't be deleted by a keyDel() as long refcounter is not 0.

The references will be incremented on successful calls to ksAppendKey() or ksAppend().

Note:

keyDup() will reset the references for dupped key.

For your own applications you can use keyIncRef() and keyDecRef() for reference counting, too.

Parameters:

key the key object to work with

Returns:

the number of references

Return values:

-1 on null pointer

See also:

keyIncRef() and keyDecRef()

ssize_t keyIncRef (Key * key)

Increment the viability of a key object. This function is intended for applications using their own reference counter for key objects. With it you can increment the reference and thus avoid destruction of the object in a subsequent keyDel().

The reference counter can't be incremented once it reached SSIZE_MAX. In that situation nothing will happen and SSIZE_MAX will be returned.

Note:

keyDup() will reset the references for dupped key.

Returns:

the value of the new reference counter

Return values:

-1 on null pointer
SSIZE_MAX when maximum exceeded

Parameters:

key the key object to work with

See also:

keyGetRef() for longer explanation, keyDecRef(), keyDel()

Key* keyNew (const char * name, ...)

A practical way to fully create a Key object in one step. To just get a key object, simple do:

Key *k = keyNew(0);
// work with it
keyDel (k);

If you want the key object to contain a name, value, comment and other meta info read on.

Note:

When you already have a key with similar properties its easier to keyDup() the key.

You can call it in many different ways depending on the attribute tags you pass as parameters. Tags are represented as the keyswitch_t values, and tell keyNew() which Key attribute comes next.

We can also give an empty key name and a KEY_END tag with the same effect as before:

Key *k =keyNew("", KEY_END); // Has the same effect as above
// work with it
keyDel (k);

But we can also give the key a proper name right from the start:

// Create and initialize a key with a name and nothing else
Key *k=keyNew("user/some/example", KEY_END);
// work with it
keyDel (k);

So, keyNew() allocates memory for a key object and keyDel() cleans everything up.

keyNew() processes the given argument list even further. The Key attribute tags are the following:

KEY_VALUE Next parameter is a pointer to the value that will be used. If no KEY_BINARY was used before, a string is assumed.

// Create and initialize a key with a name and nothing else
Key *k=keyNew("user/tmp/ex0",
        KEY_VALUE, "some data",    // set a string value
        KEY_END);                  // end of args
KEY_SIZE Define a maximum length of the value. This is only used when setting a binary key.

// Create and initialize a key with a name and nothing else
Key *k=keyNew("user/tmp/ex1",
        KEY_SIZE, 4,               // has no effect on strings
        KEY_VALUE, "some data",    // set a string value
        KEY_END);                  // end of args
KEY_META Next two parameter is a meta name and a meta value. See keySetMeta().

Key *k=keyNew("user/tmp/ex3",
        KEY_META, "comment", "a comment",  // with a commet
        KEY_META, "owner", "root",         // and an owner
        KEY_META, "special", "yes",        // and any other meta data
        KEY_END);                  // end of args
  • KEY_END Must be the last parameter passed to keyNew(). It is always required, unless the keyName is 0.
  • ::KEY_FLAGS Bitwise disjunction of flags, which don't require one or more values. recommended way to set multiple flags. overrides previously defined flags.

Key *k=keyNew("user/tmp/ex3",
        KEY_FLAGS, KEY_BINARY | KEY_CASCADING_NAME, // flags
        KEY_SIZE, 7,                    // assume binary length 7
        KEY_VALUE, "some data",         // value that will be truncated in 7 bytes
        KEY_END);                       // end of args
KEY_BINARY Allows one to change the key to a binary key. Make sure that you also pass KEY_SIZE before you set the value. Otherwise it will be cut off with first \0 in the string. So this flag toggle from keySetString() to keySetBinary(). If no value (nor size) is given, it will be a NULL key.

// Create and initialize a key with a name and nothing else
Key *k=keyNew("user/tmp/ex2",
        KEY_BINARY,
        KEY_SIZE, 4,               // now the size is important
        KEY_VALUE, "some data",    // sets the binary value ("some")
        KEY_END);                  // end of args
  • ::KEY_CASCADING_NAME allow the name to start with / useful for ksLookup() and kdbGet() parent/lookup keys
  • ::KEY_META_NAME allow the name to start with arbitrary namespaces useful to compare with meta keys

Deprecated

These other flags deprecated and KEY_META should be preferred. They remain some time, however, for compatibility:
  • KEY_DIR Define that the key is a directory rather than a ordinary key. This means its executable bits in its mode are set. But even without this option the key can have subkeys. See keySetDir().
  • KEY_OWNER Next parameter is the owner. See keySetOwner().
  • KEY_UID, KEY_GID Next parameter is taken as the UID (uid_t) or GID (gid_t) that will be defined on the key. See keySetUID() and keySetGID().
  • KEY_MODE Next parameter is taken as mode permissions (int) to the key. See keySetMode().

Key *k=keyNew("user/tmp/ex3",
        KEY_VALUE, "some data",    // with a simple value
        KEY_MODE, 0777,            // permissions
        KEY_END);                  // end of args
KEY_COMMENT Next parameter is a comment. See keySetComment().

Key *k=keyNew("user/tmp/ex4",
        KEY_BINARY,                     // key type
        KEY_SIZE, 7,                    // assume binary length 7
        KEY_VALUE, "some data",         // value that will be truncated in 7 bytes
        KEY_COMMENT, "value is truncated",
        KEY_OWNER, "root",              // owner (not uid) is root
        KEY_UID, 0,                     // root uid
        KEY_END);                       // end of args

Parameters:

name a valid name to the key, or NULL to get a simple initialized, but really empty, object

See also:

keyDel()

Returns:

a pointer to a new allocated and initialized Key object.

Return values:

NULL on malloc error or if an invalid name was passed (see keySetName()).

Author

Generated automatically by Doxygen for Elektra from the source code.
Sun May 29 2016 Version 0.8.14