1 <!-- $Id: frontend.xml,v 1.16 2003-02-23 20:40:01 adam Exp $ -->
2 <chapter id="server"><title>Generic server</title>
3 <sect1><title>Introduction</title>
6 If you aren't into documentation, a good way to learn how the
7 back end interface works is to look at the <filename>backend.h</filename>
8 file. Then, look at the small dummy-server in
9 <filename>ztest/ztest.c</filename>. The <filename>backend.h</filename>
10 file also makes a good reference, once you've chewed your way through
11 the prose of this file.
15 If you have a database system that you would like to make available by
16 means of Z39.50 or SRW, &yaz; basically offers your two options. You
17 can use the APIs provided by the &asn;, &odr;, and &comstack;
19 create and decode PDUs, and exchange them with a client.
20 Using this low-level interface gives you access to all fields and
21 options of the protocol, and you can construct your server as close
22 to your existing database as you like.
23 It is also a fairly involved process, requiring
24 you to set up an event-handling mechanism, protocol state machine,
25 etc. To simplify server implementation, we have implemented a compact
26 and simple, but reasonably full-functioned server-frontend that will
27 handle most of the protocol mechanics, while leaving you to
28 concentrate on your database interface.
33 The backend interface was designed in anticipation of a specific
34 integration task, while still attempting to achieve some degree of
35 generality. We realize fully that there are points where the
36 interface can be improved significantly. If you have specific
37 functions or parameters that you think could be useful, send us a
38 mail (or better, sign on to the mailing list referred to in the
39 top-level README file). We will try to fit good suggestions into future
40 releases, to the extent that it can be done without requiring
41 too many structural changes in existing applications.
47 The &yaz; server does not provide full SRW functionality.
48 However, it provides an adapter for SRW (to Z39.50).
53 <sect1 id="server.frontend"><title>The Database Frontend</title>
56 We refer to this software as a generic database frontend. Your
57 database system is the <emphasis>backend database</emphasis>, and the
58 interface between the two is called the <emphasis>backend API</emphasis>.
59 The backend API consists of a small number of function handlers and
60 structure definitions. You are required to provide the
61 <function>main()</function> routine for the server (which can be
62 quite simple), as well as a set of handlers to match each of the
64 The interface functions that you write can use any mechanism you like
65 to communicate with your database system: You might link the whole
66 thing together with your database application and access it by
67 function calls; you might use IPC to talk to a database server
68 somewhere; or you might link with third-party software that handles
69 the communication for you (like a commercial database client library).
70 At any rate, the handlers will perform the tasks of:
88 Scanning the database index (optional - if you wish to implement SCAN).
92 Extended Services (optional).
96 Result-Set Delete (optional).
100 Result-Set Sort (optional).
106 (more functions will be added in time to support as much of
107 Z39.50-1995 as possible).
111 <sect1 id="server.backend"><title>The Backend API</title>
114 The header file that you need to use the interface are in the
115 <filename>include/yaz</filename> directory. It's called
116 <filename>backend.h</filename>. It will include other files from
117 the <filename>include/yaz</filename> directory, so you'll
118 probably want to use the -I option of your compiler to tell it
119 where to find the files. When you run
120 <literal>make</literal> in the top-level &yaz; directory,
121 everything you need to create your server is to link with the
122 <filename>lib/libyaz.la</filename> library.
126 <sect1 id="server.main"><title>Your main() Routine</title>
129 As mentioned, your <function>main()</function> routine can be quite brief.
130 If you want to initialize global parameters, or read global configuration
131 tables, this is the place to do it. At the end of the routine, you should
136 int statserv_main(int argc, char **argv,
137 bend_initresult *(*bend_init)(bend_initrequest *r),
138 void (*bend_close)(void *handle));
142 The third and fourth arguments are pointers to handlers. Handler
143 <function>bend_init</function> is called whenever the server receives
144 an Initialize Request, so it serves as a Z39.50 session initializer. The
145 <function>bend_close</function> handler is called when the session is
150 <function>statserv_main</function> will establish listening sockets
151 according to the parameters given. When connection requests are received,
152 the event handler will typically <function>fork()</function> and
153 create a sub-process to handle a new connection.
154 Alternatively the server may be setup to create threads for each
156 If you do use global variables and forking, you should be aware, then,
157 that these cannot be shared between associations, unless you explicitly
158 disable forking by command line parameters.
162 The server provides a mechanism for controlling some of its behavior
163 without using command-line options. The function
167 statserv_options_block *statserv_getcontrol(void);
171 will return a pointer to a <literal>struct statserv_options_block</literal>
172 describing the current default settings of the server. The structure
173 contains these elements:
177 <literal>int dynamic</literal></term><listitem><para>
178 A boolean value, which determines whether the server
179 will fork on each incoming request (TRUE), or not (FALSE). Default is
180 TRUE. This flag is only read by UNIX-based servers (WIN32 based servers
182 </para></listitem></varlistentry>
185 <literal>int threads</literal></term><listitem><para>
186 A boolean value, which determines whether the server
187 will create a thread on each incoming request (TRUE), or not (FALSE).
188 Default is FALSE. This flag is only read by UNIX-based servers
189 that offer POSIX Threads support.
190 WIN32-based servers always operate in threaded mode.
191 </para></listitem></varlistentry>
194 <literal>int inetd</literal></term><listitem><para>
195 A boolean value, which determines whether the server
196 will operates under a UNIX INET daemon (inetd). Default is FALSE.
197 </para></listitem></varlistentry>
200 <literal>int loglevel</literal></term><listitem><para>
201 Set this by ORing the constants defined in
202 <filename>include/yaz/yaz-log.h</filename>.
203 </para></listitem></varlistentry>
206 <literal>char logfile[ODR_MAXNAME+1]</literal></term>
207 <listitem><para>File for diagnostic output ("": stderr).
208 </para></listitem></varlistentry>
211 <literal>char apdufile[ODR_MAXNAME+1]</literal></term>
213 Name of file for logging incoming and outgoing APDUs
214 ("": don't log APDUs, "-":
215 <literal>stderr</literal>).
216 </para></listitem></varlistentry>
219 <literal>char default_listen[1024]</literal></term>
220 <listitem><para>Same form as the command-line specification of
221 listener address. "": no default listener address.
222 Default is to listen at "tcp:@:9999". You can only
223 specify one default listener address in this fashion.
224 </para></listitem></varlistentry>
227 <literal>enum oid_proto default_proto;</literal></term>
228 <listitem><para>Either <literal>PROTO_Z3950</literal> or
229 <literal>PROTO_SR</literal>.
230 Default is <literal>PROTO_Z39_50</literal>.
231 </para></listitem></varlistentry>
234 <literal>int idle_timeout;</literal></term>
235 <listitem><para>Maximum session idle-time, in minutes. Zero indicates
236 no (infinite) timeout. Default is 120 minutes.
237 </para></listitem></varlistentry>
240 <literal>int maxrecordsize;</literal></term>
241 <listitem><para>Maximum permissible record (message) size. Default
242 is 1Mb. This amount of memory will only be allocated if a
243 client requests a very large amount of records in one operation
245 Set it to a lower number if you are worried about resource
246 consumption on your host system.
247 </para></listitem></varlistentry>
250 <literal>char configname[ODR_MAXNAME+1]</literal></term>
251 <listitem><para>Passed to the backend when a new connection is received.
252 </para></listitem></varlistentry>
255 <literal>char setuid[ODR_MAXNAME+1]</literal></term>
256 <listitem><para>Set user id to the user specified, after binding
257 the listener addresses.
258 </para></listitem></varlistentry>
261 <literal>void (*bend_start)(struct statserv_options_block *p)</literal>
263 <listitem><para>Pointer to function which is called after the
264 command line options have been parsed - but before the server
266 For forked UNIX servers this handler is called in the mother
267 process; for threaded servers this handler is called in the
269 The default value of this pointer is NULL in which case it
270 isn't invoked by the frontend server.
271 When the server operates as an NT service this handler is called
272 whenever the service is started.
273 </para></listitem></varlistentry>
276 <literal>void (*bend_stop)(struct statserv_options_block *p)</literal>
278 <listitem><para>Pointer to function which is called whenever the server
279 has stopped listening for incoming connections. This function pointer
280 has a default value of NULL in which case it isn't called.
281 When the server operates as an NT service this handler is called
282 whenever the service is stopped.
283 </para></listitem></varlistentry>
286 <literal>void *handle</literal></term>
287 <listitem><para>User defined pointer (default value NULL).
288 This is a per-server handle that can be used to specify "user-data".
289 Do not confuse this with the session-handle as returned by bend_init.
290 </para></listitem></varlistentry>
296 The pointer returned by <literal>statserv_getcontrol</literal> points to
297 a static area. You are allowed to change the contents of the structure,
298 but the changes will not take effect before you call
302 void statserv_setcontrol(statserv_options_block *block);
307 that you should generally update this structure before calling
308 <function>statserv_main()</function>.
313 <sect1 id="server.backendfunctions"><title>The Backend Functions</title>
316 For each service of the protocol, the backend interface declares one or
317 two functions. You are required to provide implementations of the
318 functions representing the services that you wish to implement.
321 <sect2><title>Init</title>
324 bend_initresult (*bend_init)(bend_initrequest *r);
328 This handler is called once for each new connection request, after
329 a new process/thread has been created, and an Initialize Request has
330 been received from the client. The pointer to the
331 <function>bend_init</function> handler is passed in the call to
332 <function>statserv_start</function>.
336 This handler is also called when operating in SRW mode - when
337 a connection has been made (even though SRW does not offer
342 Unlike previous versions of YAZ, the <function>bend_init</function> also
343 serves as a handler that defines the Z39.50 services that the backend
344 wish to support. Pointers to <emphasis>all</emphasis> service handlers,
345 including search - and fetch must be specified here in this handler.
348 The request - and result structures are defined as
352 typedef struct bend_initrequest
354 Z_IdAuthentication *auth;
355 ODR stream; /* encoding stream */
356 ODR print; /* printing stream */
357 Z_ReferenceId *referenceId;/* reference ID */
358 char *peer_name; /* dns host of peer (client) */
360 char *implementation_id;
361 char *implementation_name;
362 char *implementation_version;
363 int (*bend_sort) (void *handle, bend_sort_rr *rr);
364 int (*bend_search) (void *handle, bend_search_rr *rr);
365 int (*bend_fetch) (void *handle, bend_fetch_rr *rr);
366 int (*bend_present) (void *handle, bend_present_rr *rr);
367 int (*bend_esrequest) (void *handle, bend_esrequest_rr *rr);
368 int (*bend_delete)(void *handle, bend_delete_rr *rr);
369 int (*bend_scan)(void *handle, bend_scan_rr *rr);
370 int (*bend_segment)(void *handle, bend_segment_rr *rr);
372 ODR decode; /* decoding stream */
373 /* character set and language negotiation - see include/yaz/z-charneg.h */
374 Z_CharSetandLanguageNegotiation *charneg_request;
375 Z_External *charneg_response;
378 typedef struct bend_initresult
380 int errcode; /* 0==OK */
381 char *errstring; /* system error string or NULL */
382 void *handle; /* private handle to the backend module */
387 In general, the server frontend expects that the
388 <literal>bend_*result</literal> pointer that you return is valid at
389 least until the next call to a <literal>bend_* function</literal>.
390 This applies to all of the functions described herein. The parameter
391 structure passed to you in the call belongs to the server frontend, and
392 you should not make assumptions about its contents after the current
393 function call has completed. In other words, if you want to retain any
394 of the contents of a request structure, you should copy them.
398 The <literal>errcode</literal> should be zero if the initialization of
399 the backend went well. Any other value will be interpreted as an error.
400 The <literal>errstring</literal> isn't used in the current version, but
401 one option would be to stick it in the initResponse as a VisibleString.
402 The <literal>handle</literal> is the most important parameter. It should
403 be set to some value that uniquely identifies the current session to
404 the backend implementation. It is used by the frontend server in any
405 future calls to a backend function.
406 The typical use is to set it to point to a dynamically allocated state
407 structure that is private to your backend module.
411 The <literal>auth</literal> member holds the authentication information
412 part of the Z39.50 Initialize Request. Interpret this if your serves
413 requires authentication.
417 The members <literal>peer_name</literal>,
418 <literal>implementation_id</literal>,
419 <literal>implementation_name</literal> and
420 <literal>implementation_version</literal> holds
421 DNS of client, ID of implementor, name
422 of client (Z39.50) implementation - and version.
426 The <literal>bend_</literal> - members are set to NULL when
427 <function>bend_init</function> is called. Modify the pointers by
428 setting them to point to backend functions.
433 <sect2><title>Search and retrieve</title>
435 <para>We now describe the handlers that are required to support search -
436 and retrieve. You must support two functions - one for search - and one
437 for fetch (retrieval of one record). If desirable you can provide a
438 third handler which is called when a present request is received which
439 allows you to optimize retrieval of multiple-records.
443 int (*bend_search) (void *handle, bend_search_rr *rr);
446 char *setname; /* name to give to this set */
447 int replace_set; /* replace set, if it already exists */
448 int num_bases; /* number of databases in list */
449 char **basenames; /* databases to search */
450 Z_ReferenceId *referenceId;/* reference ID */
451 Z_Query *query; /* query structure */
452 ODR stream; /* encode stream */
453 ODR decode; /* decode stream */
454 ODR print; /* print stream */
456 bend_request request;
457 bend_association association;
459 int hits; /* number of hits */
460 int errcode; /* 0==OK */
461 char *errstring; /* system error string or NULL */
466 The <function>bend_search</function> handler is a fairly close
467 approximation of a protocol Z39.50 Search Request - and Response PDUs
468 The <literal>setname</literal> is the resultSetName from the protocol.
469 You are required to establish a mapping between the set name and whatever
470 your backend database likes to use.
471 Similarly, the <literal>replace_set</literal> is a boolean value
472 corresponding to the resultSetIndicator field in the protocol.
473 <literal>num_bases/basenames</literal> is a length of/array of character
474 pointers to the database names provided by the client.
475 The <literal>query</literal> is the full query structure as defined in
476 the protocol ASN.1 specification.
477 It can be either of the possible query types, and it's up to you to
478 determine if you can handle the provided query type.
479 Rather than reproduce the C interface here, we'll refer you to the
480 structure definitions in the file
481 <filename>include/yaz/z-core.h</filename>. If you want to look at the
482 attributeSetId OID of the RPN query, you can either match it against
483 your own internal tables, or you can use the
484 <literal>oid_getentbyoid</literal> function provided by &yaz;.
488 The structure contains a number of hits, and an
489 <literal>errcode/errstring</literal> pair. If an error occurs
490 during the search, or if you're unhappy with the request, you should
491 set the errcode to a value from the BIB-1 diagnostic set. The value
492 will then be returned to the user in a nonsurrogate diagnostic record
493 in the response. The <literal>errstring</literal>, if provided, will
494 go in the addinfo field. Look at the protocol definition for the
495 defined error codes, and the suggested uses of the addinfo field.
499 The <function>bend_search</function> handler is also called when
500 the frontend server receives a SRW SearchRetrieveRequest.
501 For SRW, a CQL query is usually provided by the client.
502 The CQL query is available as part of <literal>Z_Query</literal>
503 structure (note that CQL is now part of Z39.50 via an external).
504 To support CQL in existing implementations that only do Type-1,
505 we refer to the CQL-to-PQF tool described
506 <link linkend="tools.cql.pqf">here</link>.
510 To maintain backwards compatibility, the frontend server
511 of yaz always assume that error codes are BIB-1 diagnostics.
512 For SRW operation, a Bib-1 diagnostic code is mapped to
517 int (*bend_fetch) (void *handle, bend_fetch_rr *rr);
519 typedef struct bend_fetch_rr {
520 char *setname; /* set name */
521 int number; /* record number */
522 Z_ReferenceId *referenceId;/* reference ID */
523 oid_value request_format; /* One of the CLASS_RECSYN members */
524 int *request_format_raw; /* same as above (raw OID) */
525 Z_RecordComposition *comp; /* Formatting instructions */
526 ODR stream; /* encoding stream - memory source if req */
527 ODR print; /* printing stream */
529 char *basename; /* name of database that provided record */
530 int len; /* length of record or -1 if structured */
531 char *record; /* record */
532 int last_in_set; /* is it? */
533 oid_value output_format; /* format */
534 int *output_format_raw; /* used instead of above if not-null */
535 int errcode; /* 0==success */
536 char *errstring; /* system error string or NULL */
537 int surrogate_flag; /* surrogate diagnostic */
542 The frontend server calls the <function>bend_fetch</function> handler
543 when it needs database records to fulfill a Z39.50 Search Request, a
544 Z39.50 Present Request or a SRW SearchRetrieveRequest.
545 The <literal>setname</literal> is simply the name of the result set
546 that holds the reference to the desired record.
547 The <literal>number</literal> is the offset into the set (with 1
548 being the first record in the set). The <literal>format</literal> field
549 is the record format requested by the client (See section
550 <link linkend="oid">Object Identifiers</link>). The value
551 <literal>VAL_NONE</literal> indicates that the client did not
552 request a specific format. The <literal>stream</literal> argument
553 is an &odr; stream which should be used for
554 allocating space for structured data records.
555 The stream will be reset when all records have been assembled, and
556 the response package has been transmitted.
557 For unstructured data, the backend is responsible for maintaining a
558 static or dynamic buffer for the record between calls.
562 If a SRW SearchRetrieveRequest is received by the frontend server,
563 the <literal>referenceId</literal> is NULL, the <literal>request_format
564 </literal> (transfer syntax) is XML (OID name
565 <literal>VAL_TEXT_XML</literal>).
566 The schema for SRW is stored in the
567 <literal>Z_RecordComposition</literal>
572 In the structure, the <literal>basename</literal> is the name of the
573 database that holds the
574 record. <literal>len</literal> is the length of the record returned, in
575 bytes, and <literal>record</literal> is a pointer to the record.
576 <literal>Last_in_set</literal> should be nonzero only if the record
577 returned is the last one in the given result set.
578 <literal>errcode</literal> and <literal>errstring</literal>, if
579 given, will be interpreted as a global error pertaining to the
580 set, and will be returned in a non-surrogate-diagnostic.
581 If you wish to return the error as a surrogate-diagnostic
582 (local error) you can do this by setting
583 <literal>surrogate_flag</literal> to 1 also.
587 If the <literal>len</literal> field has the value -1, then
588 <literal>record</literal> is assumed to point to a constructed data
589 type. The <literal>format</literal> field will be used to determine
590 which encoder should be used to serialize the data.
595 If your backend generates structured records, it should use
596 <function>odr_malloc()</function> on the provided stream for allocating
597 data: This allows the frontend server to keep track of the record sizes.
602 The <literal>format</literal> field is mapped to an object identifier
603 in the direct reference of the resulting EXTERNAL representation
609 The current version of &yaz; only supports the direct reference mode.
614 int (*bend_present) (void *handle, bend_present_rr *rr);
617 char *setname; /* set name */
619 int number; /* record number */
620 oid_value format; /* One of the CLASS_RECSYN members */
621 Z_ReferenceId *referenceId;/* reference ID */
622 Z_RecordComposition *comp; /* Formatting instructions */
623 ODR stream; /* encoding stream */
624 ODR print; /* printing stream */
625 bend_request request;
626 bend_association association;
628 int hits; /* number of hits */
629 int errcode; /* 0==OK */
630 char *errstring; /* system error string or NULL */
635 The <function>bend_present</function> handler is called when
636 the server receives a Z39.50 Present Request.
637 The <literal>setname</literal>,
638 <literal>start</literal> and <literal>number</literal> is the
639 name of the result set - start position - and number of records to
640 be retrieved respectively. <literal>format</literal> and
641 <literal>comp</literal> is the preferred transfer syntax and element
642 specifications of the present request.
645 Note that this is handler serves as a supplement for
646 <function>bend_fetch</function> and need not to be defined in order to
647 support search - and retrieve.
652 <sect2><title>Delete</title>
655 For back-ends that supports delete of a result set only one handler
660 int (*bend_delete)(void *handle, bend_delete_rr *rr);
662 typedef struct bend_delete_rr {
666 Z_ReferenceId *referenceId;
667 int delete_status; /* status for the whole operation */
668 int *statuses; /* status each set - indexed as setnames */
676 The delete set function definition is rather primitive, mostly because
677 we have had no practical need for it as of yet. If someone wants
678 to provide a full delete service, we'd be happy to add the
679 extra parameters that are required. Are there clients out there
680 that will actually delete sets they no longer need?
686 <sect2><title>scan</title>
689 For servers that wish to offer the scan service one handler
694 int (*bend_delete)(void *handle, bend_delete_rr *rr);
697 BEND_SCAN_SUCCESS, /* ok */
698 BEND_SCAN_PARTIAL /* not all entries could be found */
701 typedef struct bend_scan_rr {
702 int num_bases; /* number of elements in database list */
703 char **basenames; /* databases to search */
704 oid_value attributeset;
705 Z_ReferenceId *referenceId; /* reference ID */
706 Z_AttributesPlusTerm *term;
707 ODR stream; /* encoding stream - memory source if required */
708 ODR print; /* printing stream */
710 int *step_size; /* step size */
711 int term_position; /* desired index of term in result list/returned */
712 int num_entries; /* number of entries requested/returned */
714 struct scan_entry *entries;
715 bend_scan_status status;
723 <sect1 id="server.invocation"><title>Application Invocation</title>
726 The finished application has the following
727 invocation syntax (by way of <function>statserv_main()</function>):
731 <command>appname</command>
732 <arg choice="opt"><option>-a <replaceable>file</replaceable></option></arg>
733 <arg choice="opt"><option>-v <replaceable>level</replaceable></option></arg>
734 <arg choice="opt"><option>-l <replaceable>file</replaceable></option></arg>
735 <arg choice="opt"><option>-u <replaceable>uid</replaceable></option></arg>
736 <arg choice="opt"><option>-c <replaceable>config</replaceable></option></arg>
737 <arg choice="opt"><option>-t <replaceable>minutes</replaceable></option></arg>
739 <arg choice="opt"><option>-k <replaceable>kilobytes</replaceable></option></arg>
740 <arg choice="opt"><option>-d <replaceable>daemon</replaceable></option></arg>
741 <arg choice="opt"><option>-w <replaceable>dir</replaceable></option></arg>
742 <arg choice="opt"><option>-ziST1</option></arg>
743 <arg choice="opt" rep="repeat">listener-spec</arg>
754 A listener specification consists of a transport mode followed by a
755 colon (:) followed by a listener address. The transport mode is
756 either <literal>tcp</literal>, <literal>unix:</literal> or
757 <literal>ssl</literal>.
761 For TCP and SSL, an address has the form
765 hostname | IP-number [: portnumber]
769 The port number defaults to 210 (standard Z39.50 port).
773 For UNIX, the address is the filename of socket.
777 For TCP/IP and SSL, the special hostname "@" is mapped to
778 the address <literal>INADDR_ANY</literal>, which causes the
779 server to listen on any local interface.
796 <!-- Keep this comment at the end of the file
801 sgml-minimize-attributes:nil
802 sgml-always-quote-attributes:t
805 sgml-parent-document: "yaz.xml"
806 sgml-local-catalogs: nil
807 sgml-namecase-general:t