Network Working Group T. Howes Request for Comments: 1823 M. Smith Category: Informational University of Michigan August 1995
The LDAP Application Program Interface
Status of this Memo
This memo provides information for the Internet community. This memo does not specify an Internet standard of any kind. Distribution of this memo is unlimited.
This document defines a C language application program interface to the lightweight directory access protocol (LDAP). The LDAP API is designed to be powerful, yet simple to use. It defines compatible synchronous and asynchronous interfaces to LDAP to suit a wide variety of applications. This document gives a brief overview of the LDAP model, then an overview of how the API is used by an application program to obtain LDAP information. The API calls are described in detail, followed by an appendix that provides some example code demonstrating the use of the API.
LDAP is the lightweight directory access protocol, described in [2] and [7]. It can provide a lightweight frontend to the X.500 directory [1], or a stand-alone service. In either mode, LDAP is based on a client-server model in which a client makes a TCP connection to an LDAP server, over which it sends requests and receives responses.
The LDAP information model is based on the entry, which contains information about some object (e.g., a person). Entries are composed of attributes, which have a type and one or more values. Each attribute has a syntax that determines what kinds of values are allowed in the attribute (e.g., ASCII characters, a jpeg photograph, etc.) and how those values behave during directory operations (e.g., is case significant during comparisons).
Entries are organized in a tree structure, usually based on political, geographical, and organizational boundaries. Each entry is uniquely named relative to its sibling entries by its relative distinguished name (RDN) consisting of one or more distinguished attribute values from the entry. At most one value from each attribute may be used in the RDN. For example, the entry for the
Howes & Smith Informational [Page 1]
RFC 1823 LDAP API August 1995
person Babs Jensen might be named with the "Barbara Jensen" value from the commonName attribute. A globally unique name for an entry, called a distinguished name or DN, is constructed by concatenating the sequence of RDNs from the root of the tree down to the entry. For example, if Babs worked for the University of Michigan, the DN of her U-M entry might be "cn=Barbara Jensen, o=University of Michigan, c=US". The DN format used by LDAP is defined in [4].
Operations are provided to authenticate, search for and retrieve information, modify information, and add and delete entries from the tree. The next sections give an overview of how the API is used and detailed descriptions of the LDAP API calls that implement all of these functions.
An application generally uses the LDAP API in four simple steps.
o Open a connection to an LDAP server. The ldap_open() call returns a handle to the connection, allowing multiple connections to be open at once.
o Authenticate to the LDAP server and/or the X.500 DSA. The ldap_bind() call and friends support a variety of authentication methods.
o Perform some LDAP operations and obtain some results. ldap_search() and friends return results which can be parsed by ldap_result2error(), ldap_first_entry(), ldap_next_entry(), etc.
o Close the connection. The ldap_unbind() call closes the connection.
Operations can be performed either synchronously or asynchronously. Synchronous calls end in _s. For example, a synchronous search can be completed by calling ldap_search_s(). An asynchronous search can be initiated by calling ldap_search(). All synchronous routines return an indication of the outcome of the operation (e.g, the constant LDAP_SUCCESS or some other error code). The asynchronous routines return the message id of the operation initiated. This id can be used in subsequent calls to ldap_result() to obtain the result(s) of the operation. An asynchronous operation can be abandoned by calling ldap_abandon().
Howes & Smith Informational [Page 2]
RFC 1823 LDAP API August 1995
Results and errors are returned in an opaque structure called LDAPMessage. Routines are provided to parse this structure, step through entries and attributes returned, etc. Routines are also provided to interpret errors. The next sections describe these routines in more detail.
This section describes each LDAP operation API call in detail. All calls take a "connection handle", a pointer to an LDAP structure containing per-connection information. Many routines return results in an LDAPMessage structure. These structures and others are described as needed below.
ldap_open() opens a connection to the LDAP server.
typedef struct ldap { /* ... opaque parameters ... */ int ld_deref; int ld_timelimit; int ld_sizelimit; int ld_errno; char *ld_matched; char *ld_error; /* ... opaque parameters ... */ } LDAP;
LDAP *ldap_open( char *hostname, int portno );
Parameters are:
hostname Contains a space-separated list of hostnames or dotted strings representing the IP address of hosts running an LDAP server to connect to. The hosts are tried in the order listed, stopping with the first one to which a successful connection is made;
portno contains the TCP port number to which to connect. The default LDAP port can be obtained by supplying the constant LDAP_PORT.
ldap_open() returns a "connection handle", a pointer to an LDAP structure that should be passed to subsequent calls pertaining to the connection. It returns NULL if the connection cannot be opened. One of the ldap_bind calls described below must be completed before other operations can be performed on the connection.
Howes & Smith Informational [Page 3]
RFC 1823 LDAP API August 1995
The calling program should assume nothing about the order of the fields in the LDAP structure. There may be other fields in the structure for internal library use. The fields shown above are described as needed in the description of other calls below.
ldap_bind() and friends are used to authenticate to the directory.
int ldap_bind( LDAP *ld, char *dn, char *cred, int method );
int ldap_bind_s( LDAP *ld, char *dn, char *cred, int method );
int ldap_simple_bind( LDAP *ld, char *dn, char *passwd );
int ldap_simple_bind_s( LDAP *ld, char *dn, char *passwd );
int ldap_kerberos_bind( LDAP *ld, char *dn );
int ldap_kerberos_bind_s( LDAP *ld, char *dn );
Parameters are:
ld The connection handle;
dn The name of the entry to bind as;
cred The credentials with which to authenticate;
method One of LDAP_AUTH_SIMPLE, LDAP_AUTH_KRBV41, or LDAP_AUTH_KRBV42, indicating the authentication method to use;
passwd For ldap_simple_bind(), the password to compare to the entry's userPassword attribute;
There are three types of bind calls, providing simple authentication, kerberos authentication, and general routines to do either one. In the case of Kerberos version 4 authentication using the general ldap_bind() routines, the credentials are ignored, as the routines assume a valid ticket granting ticket already exists which can be used to retrieve the appropriate service tickets.
Synchronous versions of the routines have names that end in _s. These routines return the result of the bind operation, either the constant LDAP_SUCCESS if the operation was successful, or another LDAP error code if it was not. See the section below on error handling for more information about possible errors and how to interpret them.
Howes & Smith Informational [Page 4]
RFC 1823 LDAP API August 1995
Asynchronous versions of these routines return the message id of the bind operation initiated. A subsequent call to ldap_result(), described below, can be used to obtain the result of the bind. In case of error, these routines will return -1, setting the ld_errno field in the LDAP structure appropriately.
Note that no other operations over the connection should be attempted before a bind call has successfully completed. Subsequent bind calls can be used to re-authenticate over the same connection.
ldap_unbind() is used to unbind from the directory and close the connection.
int ldap_unbind( LDAP *ld );
Parameters are:
ld The connection handle.
ldap_unbind() works synchronously, unbinding from the directory, closing the connection, and freeing up the ld structure before returning. ldap_unbind() returns LDAP_SUCCESS (or another LDAP error code if the request cannot be sent to the LDAP server). After a call to ldap_unbind(), the ld connection handle is invalid.
ldap_search() and friends are used to search the LDAP directory, returning a requested set of attributes for each entry matched. There are three variations.
struct timeval { long tv_sec; long tv_usec; }; int ldap_search( LDAP *ld, char *base, int scope, char *filter, char *attrs[], int attrsonly ); int ldap_search_s( LDAP *ld, char *base,
Howes & Smith Informational [Page 5]
RFC 1823 LDAP API August 1995
int scope, char *filter, char *attrs[], int attrsonly, LDAPMessage **res ); int ldap_search_st( LDAP *ld, char *base, int scope, char *filter, char *attrs[], int attrsonly, struct timeval *timeout, LDAPMessage **res );
Parameters are:
ld The connection handle;
base The dn of the entry at which to start the search;
scope One of LDAP_SCOPE_BASE, LDAP_SCOPE_ONELEVEL, or LDAP_SCOPE_SUBTREE, indicating the scope of the search;
filter A character string as described in RFC 1558 [3], representing the search filter;
attrs A NULL-terminated array of strings indicating which attributes to return for each matching entry. Passing NULL for this parameter causes all available attributes to be retrieved;
attrsonly A boolean value that should be zero if both attribute types and values are to be returned, non-zero if only types are wanted;
timeout For the ldap_search_st() call, this specifies the local search timeout value;
res For the synchronous calls, this is a result parameter which will contain the results of the search upon completion of the call.
There are three fields in the ld connection handle which control how the search is performed. They are:
Howes & Smith Informational [Page 6]
RFC 1823 LDAP API August 1995
ld_sizelimit A limit on the number of entries to return from the search. A value of zero means no limit;
ld_timelimit A limit on the number of seconds to spend on the search. A value of zero means no limit;
ld_deref One of LDAP_DEREF_NEVER, LDAP_DEREF_SEARCHING, LDAP_DEREF_FINDING, or LDAP_DEREF_ALWAYS, specifying how aliases should be handled during the search. The LDAP_DEREF_SEARCHING value means aliases should be dereferenced during the search but not when locating the base object of the search. The LDAP_DEREF_FINDING value means aliases should be dereferenced when locating the base object but not during the search.
An asynchronous search is initiated by calling ldap_search(). It returns the message id of the initiated search. The results of the search can be obtained by a subsequent call to ldap_result(). The results can be parsed by the result parsing routines described in detail later. In case of error, -1 is returned and the ld_errno field in the LDAP structure is set appropriately.
A synchronous search is performed by calling ldap_search_s() or ldap_search_st(). The routines are identical, except that ldap_search_st() takes an additional parameter specifying a timeout for the search. Both routines return an indication of the result of the search, either LDAP_SUCCESS or some error indication (see Error Handling below). The entries returned from the search (if any) are contained in the res parameter. This parameter is opaque to the caller. Entries, attributes, values, etc., should be extracted by calling the parsing routines described below. The results contained in res should be freed when no longer in use by calling ldap_msgfree(), described later.
LDAP does not support a read operation directly. Instead, this operation is emulated by a search with base set to the DN of the entry to read, scope set to LDAP_SCOPE_BASE, and filter set to "(objectclass=*)". attrs contains the list of attributes to return.
LDAP does not support a list operation directly. Instead, this operation is emulated by a search with base set to the DN of the entry to list, scope set to LDAP_SCOPE_ONELEVEL, and filter set to "(objectclass=*)". attrs contains the list of attributes to return for each child entry.
int ldap_modify( LDAP *ld, char *dn, LDAPMod *mods[] );
int ldap_modify_s( LDAP *ld, char *dn, LDAPMod *mods[] );
Parameters are:
ld The connection handle;
dn The name of the entry to modify;
mods A NULL-terminated array of modifications to make to the entry.
The fields in the LDAPMod structure have the following meanings:
mod_op The modification operation to perform. It should be one of LDAP_MOD_ADD, LDAP_MOD_DELETE, or LDAP_MOD_REPLACE. This field also indicates the type of values included in the mod_vals union. It is ORed with LDAP_MOD_BVALUES to select the mod_bvalues form. Otherwise, the mod_values form is used;
mod_type The type of the attribute to modify;
mod_vals The values (if any) to add, delete, or replace. Only one of the mod_values or mod_bvalues variants should be used, selected by ORing the mod_op field with the constant LDAP_MOD_BVALUES. mod_values is a NULL-terminated array of zero-terminated strings and mod_bvalues is a NULL-terminated array of berval structures that can be used to pass binary values such as images.
Howes & Smith Informational [Page 8]
RFC 1823 LDAP API August 1995
For LDAP_MOD_ADD modifications, the given values are added to the entry, creating the attribute if necessary. For LDAP_MOD_DELETE modifications, the given values are deleted from the entry, removing the attribute if no values remain. If the entire attribute is to be deleted, the mod_vals field should be set to NULL. For LDAP_MOD_REPLACE modifications, the attribute will have the listed values after the modification, having been created if necessary. All modifications are performed in the order in which they are listed.
ldap_modify_s() returns the LDAP error code resulting from the modify operation. This code can be interpreted by ldap_perror() and friends.
ldap_modify() returns the message id of the request it initiates, or -1 on error. The result of the operation can be obtained by calling ldap_result().
The ldap_modrdn() and ldap_modrdn_s() routines are used to change the name of an LDAP entry.
int ldap_modrdn( LDAP *ld, char *dn, char *newrdn, int deleteoldrdn ); int ldap_modrdn_s( LDAP *ld, char *dn, char *newrdn, int deleteoldrdn );
Parameters are:
ld The connection handle;
dn The name of the entry whose RDN is to be changed;
newrdn The new RDN to give the entry;
deleteoldrdn A boolean value, if non-zero indicating that the old RDN value(s) should be removed, if zero indicating that the old RDN value(s) should be retained as non- distinguished values of the entry.
Howes & Smith Informational [Page 9]
RFC 1823 LDAP API August 1995
The ldap_modrdn_s() routine is synchronous, returning the LDAP error code indicating the outcome of the operation.
The ldap_modrdn() routine is asynchronous, returning the message id of the operation it initiates, or -1 in case of trouble. The result of the operation can be obtained by calling ldap_result().
ldap_add() and ldap_add_s() are used to add entries to the LDAP directory.
int ldap_add( LDAP *ld, char *dn, LDAPMod *attrs[] );
int ldap_add_s( LDAP *ld, char *dn, LDAPMod *attrs[] );
Parameters are:
ld The connection handle;
dn The name of the entry to add;
attrs The entry's attributes, specified using the LDAPMod structure defined for ldap_modify(). The mod_type and mod_vals fields should be filled in. The mod_op field is ignored unless ORed with the constant LDAP_MOD_BVALUES, used to select the mod_bvalues case of the mod_vals union.
Note that the parent of the entry must already exist.
ldap_add_s() is synchronous, returning the LDAP error code indicating the outcome of the operation.
ldap_add() is asynchronous, returning the message id of the operation it initiates, or -1 in case of trouble. The result of the operation can be obtained by calling ldap_result().
ldap_delete() and ldap_delete_s() are used to delete entries from the LDAP directory.
int ldap_delete( LDAP *ld, char *dn );
int ldap_delete_s( LDAP *ld, char *dn );
Howes & Smith Informational [Page 10]
RFC 1823 LDAP API August 1995
Parameters are:
ld The connection handle;
dn The name of the entry to delete.
Note that the entry to delete must be a leaf entry (i.e., it must have no children). Deletion of entire subtrees is not supported by LDAP.
ldap_delete_s() is synchronous, returning the LDAP error code indicating the outcome of the operation.
ldap_delete() is asynchronous, returning the message id of the operation it initiates, or -1 in case of trouble. The result of the operation can be obtained by calling ldap_result().
ldap_abandon() is used to abandon an operation in progress.
int ldap_abandon( LDAP *ld, int msgid );
ldap_abandon() abandons the operation with message id msgid. It returns zero if the abandon was successful, -1 otherwise. After a successful call to ldap_abandon(), results with the given message id are never returned from a call to ldap_result().
ldap_result() is used to obtain the result of a previous asynchronously initiated operation. ldap_msgfree() frees the results obtained from a previous call to ldap_result(), or a synchronous search routine.
int ldap_result( LDAP *ld, int msgid, int all, struct timeval *timeout, LDAPMessage **res );
int ldap_msgfree( LDAPMessage *res );
Howes & Smith Informational [Page 11]
RFC 1823 LDAP API August 1995
Parameters are:
ld The connection handle;
msgid The message id of the operation whose results are to be returned, or the constant LDAP_RES_ANY if any result is desired;
all A boolean parameter that only has meaning for search results. If non-zero it indicates that all results of a search should be retrieved before any are returned. If zero, search results (entries) will be returned one at a time as they arrive;
timeout A timeout specifying how long to wait for results to be returned. A NULL value causes ldap_result() to block until results are available. A timeout value of zero second specifies a polling behavior;
res For ldap_result(), a result parameter that will contain the result(s) of the operation. For ldap_msgfree(), the result chain to be freed, obtained from a previous call to ldap_result() or ldap_search_s() or ldap_search_st().
Upon successful completion, ldap_result() returns the type of the result returned in the res parameter. This will be one of the following constants.
ldap_result() returns 0 if the timeout expired and -1 if an error occurs, in which case the ld_errno field of the ld structure will be set accordingly.
ldap_msgfree() frees the result structure pointed to be res and returns the type of the message it freed.
The following calls are used to interpret errors returned by other LDAP API routines.
int ldap_result2error( LDAP *ld, LDAPMessage *res, int freeit );
char *ldap_err2string( int err );
void ldap_perror( LDAP *ld, char *msg );
Parameters are:
ld The connection handle;
res The result of an LDAP operation as returned by ldap_result() or one of the synchronous API operation calls;
freeit A boolean parameter indicating whether the res parameter should be freed (non-zero) or not (zero);
err An LDAP error code, as returned by ldap_result2error() or one of the synchronous API operation calls;
msg A message to be displayed before the LDAP error message.
ldap_result2error() is used to convert the LDAP result message obtained from ldap_result(), or the res parameter returned by one of the synchronous API operation calls, into a numeric LDAP error code. It also parses the ld_matched and ld_error portions of the result message and puts them into the connection handle information. All the synchronous operation routines call ldap_result2error() before returning, ensuring that these fields are set correctly. The relevant fields in the connection structue are:
ld_matched In the event of an LDAP_NO_SUCH_OBJECT error return, this parameter contains the extent of the DN matched;
ld_error This parameter contains the error message sent in the result by the LDAP server.
ld_errno The LDAP error code indicating the outcome of the operation. It is one of the following constants:
ldap_err2string() is used to convert a numeric LDAP error code, as returned by ldap_result2error() or one of the synchronous API operation calls, into an informative NULL-terminated character string message describing the error. It returns a pointer to static data.
ldap_perror() is used to print the message supplied in msg, followed by an indication of the error contained in the ld_errno field of the ld connection handle, to standard error.
The following calls are used to parse the entries returned by ldap_search() and friends. These entries are returned in an opaque structure that should only be accessed by calling the routines described below. Routines are provided to step through the entries returned, step through the attributes of an entry, retrieve the name of an entry, and retrieve the values associated with a given attribute in an entry.
The ldap_first_entry() and ldap_next_entry() routines are used to step through a set of entries in a search result. ldap_count_entries() is used to count the number of entries returned.
int ldap_count_entries( LDAP *ld, LDAPMessage *res );
Parameters are:
ld The connection handle;
res The search result, as obtained by a call to one of the syn- chronous search routines or ldap_result();
entry The entry returned by a previous call to ldap_first_entry() or ldap_next_entry().
ldap_first_entry() and ldap_next_entry() will return NULL when no more entries exist to be returned. NULL is also returned if an error occurs while stepping through the entries, in which case the ld_errno field of the ld connection handle will be set to indicate the error.
ldap_count_entries() returns the number of entries contained in a chain of entries. It can also be used to count the number of entries
Howes & Smith Informational [Page 15]
RFC 1823 LDAP API August 1995
that remain in a chain if called with an entry returned by ldap_first_entry() or ldap_next_entry().
entry The entry whose attributes are to be stepped through, as returned by ldap_first_entry() or ldap_next_entry();
ptr In ldap_first_attribute(), the address of a pointer used internally to keep track of the current position in the entry. In ldap_next_attribute(), the pointer returned by a previous call to ldap_first_attribute().
ldap_first_attribute() and ldap_next_attribute() will return NULL when the end of the attributes is reached, or if there is an error, in which case the ld_errno field in the ld connection handle will be set to indicate the error.
Both routines return a pointer to a per-connection buffer containing the current attribute name. This should be treated like static data. ldap_first_attribute() will allocate and return in ptr a pointer to a BerElement used to keep track of the current position. This pointer should be passed in subsequent calls to ldap_next_attribute() to step through the entry's attributes.
The attribute names returned are suitable for passing in a call to ldap_get_values() and friends to retrieve the associated values.
ldap_get_values() and ldap_get_values_len() are used to retrieve the values of a given attribute from an entry. ldap_count_values() and ldap_count_values_len() are used to count the returned values. ldap_value_free() and ldap_value_free_len() are used to free the values.
typedef struct berval { unsigned long bv_len; char *bv_val; };
int ldap_count_values_len( struct berval **vals );
int ldap_value_free( char **vals );
int ldap_value_free_len( struct berval **vals );
Parameters are:
ld The connection handle;
entry The entry from which to retrieve values, as returned by ldap_first_entry() or ldap_next_entry();
attr The attribute whose values are to be retrieved, as returned by ldap_first_attribute() or ldap_next_attribute(), or a caller- supplied string (e.g., "mail");
vals The values returned by a previous call to ldap_get_values() or ldap_get_values_len().
Howes & Smith Informational [Page 17]
RFC 1823 LDAP API August 1995
Two forms of the various calls are provided. The first form is only suitable for use with non-binary character string data only. The second _len form is used with any kind of data.
Note that the values returned are malloc'ed and should be freed by calling either ldap_value_free() or ldap_value_free_len() when no longer in use.
ldap_get_dn() is used to retrieve the name of an entry. ldap_explode_dn() is used to break up the name into its component parts. ldap_dn2ufn() is used to convert the name into a more "user friendly" format.
entry The entry whose name is to be retrieved, as returned by ldap_first_entry() or ldap_next_entry();
dn The dn to explode, as returned by ldap_get_dn();
notypes A boolean parameter, if non-zero indicating that the dn com- ponents should have their type information stripped off (i.e., "cn=Babs" would become "Babs").
ldap_get_dn() will return NULL if there is some error parsing the dn, setting ld_errno in the ld connection handle to indicate the error. It returns a pointer to malloc'ed space that the caller should free by calling free() when it is no longer in use. Note the format of the DNs returned is given by [4].
ldap_explode_dn() returns a char * array containing the RDN components of the DN supplied, with or without types as indicated by the notypes parameter. The array returned should be freed when it is no longer in use by calling ldap_value_free().
ldap_dn2ufn() converts the DN into the user friendly format described in [5]. The UFN returned is malloc'ed space that should be freed by a call to free() when no longer in use.
[1] The Directory: Selected Attribute Syntaxes. CCITT, Recommendation X.520.
[2] Howes, T., Kille, S., Yeong, W., and C. Robbins, "The String Representation of Standard Attribute Syntaxes", University of Michigan, ISODE Consortium, Performance Systems International, NeXor Ltd., RFC 1778, March 1995.
[3] Howes, T., "A String Representation of LDAP Search Filters", RFC 1558, University of Michigan, December 1993.
[4] Kille, S., "A String Representation of Distinguished Names", RFC 1779, ISODE Consortium, March 1995.
[5] Kille, S., "Using the OSI Directory to Achieve User Friendly Naming", RFC 1781, ISODE Consortium, March 1995.
[6] S.P. Miller, B.C. Neuman, J.I. Schiller, J.H. Saltzer, "Kerberos Authentication and Authorization System", MIT Project Athena Documentation Section E.2.1, December 1987
[7] Yeong, W., Howes, T., and S. Kille, "Lightweight Directory Access Protocol," RFC 1777, Performance Systems International, University of Michigan, ISODE Consortium, March 1995.
/* search for entries with cn of "Babs Jensen", return all attrs */ if ( ldap_search_s( ld, "o=University of Michigan, c=US", LDAP_SCOPE_SUBTREE, "(cn=Babs Jensen)", NULL, 0, &res ) != LDAP_SUCCESS ) { ldap_perror( ld, "ldap_search_s" ); exit( 1 ); }
/* step through each entry returned */ for ( e = ldap_first_entry( ld, res ); e != NULL; e = ldap_next_entry( ld, e ) ) { /* print its name */ dn = ldap_get_dn( ld, e ); printf( "dn: %s0, dn ); free( dn );
/* print each attribute */ for ( a = ldap_first_attribute( ld, e, &ptr ); a != NULL; a = ldap_next_attribute( ld, e, ptr ) ) { printf( "attribute: %s0, a );
/* print each value */
Howes & Smith Informational [Page 21]
RFC 1823 LDAP API August 1995
vals = ldap_get_values( ld, e, a ); for ( i = 0; vals[i] != NULL; i++ ) { printf( "value: %s0, vals[i] ); } ldap_value_free( vals ); } } /* free the search results */ ldap_msgfree( res );
/* close and free connection resources */ ldap_unbind( ld ); }