changelog

harvest是一个下载html网页得机器人

Possible compatibility problems with earlier versions marked with '*'.

--- 2.0.4 2003/09/04

Allow any CQL relation (not just all,any,exact,scr).

Fixes for OCLC UI ASN.1 to make it work with SilverPlatter targets
that features the same UI.

Implemented command update0 in YAZ client which is equivalent to
update but uses old of Ext Update ASN.1.

Use OID 1.2.840.10003.15.1000.81.1 for privately defined charsets
in charset negotiation. Added VAL_ID_CHARSET in oid.h.
Previously no OID was used for this info.

ZOOM uses smallSetUpperBound=1 instead of 0 for piggyback searches.
It apparantely upsets EBSCO Publishing IR Z39.50 Server which
returns 'malformed search term' otherwise.

SortResponse member resultCount now set to NULL in sort handler of
frontend server.

OPAC support ZOOM. ZOOM_record_get returns OPAC info (as XML) if
type is "opac".

Add three new utility functions for translating OIDs between various
formats -- symbolic name such as "Usmarc", minus-1-terminated int
arrays and dotted strings such as "1.2.840.10003.9.5.1":
 int *oid_name_to_oid(oid_class oclass, const char *name, int *oid);
 char *oid_to_dotstring(const int *oid, char *oidbuf);
 char *oid_name_to_dotstring(oid_class oclass, const char *name, char *oidbuf);

Reverted modifications for ES Update definition for z_IU0Update 
(OID 1.2.840.10003.9.5.1 and 1.2.840.10003.9.5). The definition
is back to YAZ 2.0.2.

String value attributes for CCL parser.

Support for OCLC's locally registered "OCLC-UserInformation" PDU,
which their FirstSearch server returns in Init responses, sometimes
carrying useful diagnostic information when Init fails.  This has the
OID 1.2.840.10003.10.1000.17.1 and is defined in the file
"z39.50/oclcui.asn", where its definition is somewhat different from
what OCLC document on their web-site, as required to interoperate with
their servers.

Support for anonymous authentication in yaz-client.  Use the command
"auth -".

* Fix the OID of the Index Data-local "admin" Extended Service from 
  1.2.840.10003.9.81.1 to 1.2.840.10003.9.1000.81.1.  The omission of
  the 1000, which indicates a locally-defined OID, was a mistake.

--- 2.0.3 2003/06/20

Fix CCL directive @case handling so that it affects string match for
both CCL keywords and qualifiers.

* ESFormat-Update updates. Packages with OID 1.2.840.10003.9.5.1 and
1.2.840.10003.9.5 are now handled by the same decoder z_IU0Update .
Decoder has been updated with ASN.1 changes as listed in
http://lcweb.loc.gov/z3950/agency/defns/updateES.html
Note that the "latest" DB Update, OID 1.2.840.10003.9.5.1.1 , handler
z_IUUpdate, remains unchanged. 

ZOOM_record_get may now return a record in a specific character set.
For type use "form; charset=from[,to]" where form (render, xml, ..) ,
"from" is the character set of record as returned by target. "to" is
the character set to be returned. If omitted, "to" is UTF-8.

YAZ ASN.1 compiler renamed from yaz-comp to yaz-asncomp

New ODR utility, odr_getelement, which returns name of element for
which encoding/decoding failed.

Fixed ODR so that it returns error code OREQUIRED rather than ONONE
in cases where a required element was omitted.

Bug fix: some MARC8 sequences were not converted.

New ZOOM option "step" which specifies number of records to be
retrieved in one chunk. Used in conjunction with "start" and "count".

SRW support for yaz-client. Use scheme http: to use it, e.g.
 http://host:port/db

yaz-client no longer does (un)intelligent character set conversions
by default. Can be enabled with charset and marc_charset commands.

xmalloc trace fix for 64-bit systems such as DEC alpha. 

Solaris 2.6 compile fix.

New function, cql_strerror() translates SRW error codes, as returned
by cql_transform_error(), into human-readable English strings.  Used
in yaz-client diagnostics when appropriate.

New CQL2RPN query-type for the command-line yaz-client, parses CQL and
converts it into a type-1 query which is sent to the server.  This
makes CQL available for use with all Z-servers, as opposed to the tiny
minority that can handle CQL queries themselves.

--- 2.0.2 2003/04/28

New Debian package layout similar to the Redhat Package layout.

New ZOOM events ZOOM_EVENT_RECV_{RECORD,SEARCH} for receiving a record
and search result respectively.

--- 2.0.1 2003/04/25

Redhat package is now several packages: libyaz (runtime libraries),
libyaz-devel (development and documentation, and yaz (utility programs).
Package spec file by Morten B鴊eskov.

New member 'schema' in struct bend_fetch_rr which the name of requested
schema (SRW/SRU) for record (or NULL if none was given). The fetch
handler MAY set this to reflect the schema of the returned record.

New member 'display_term' in struct scan_entry. GFS now sets member entries
and allocates scan entries to be filled by user scan handler. In previous
version, entries member was allocated by the scan handler. That still works,
but the GFS will ignore member display_term - assuming it was NOT set by
the handler. The fact that the GFS now allocates the entries both allows
for new members and makes a scan handler easier to write.

Fix CQL lex buffer overflow.

SRW/SRU recordPacking. For SRW default recordPacking is string. For
SRU default recordPacking is string.

SRU protocol support for frontend server.

Fix compile bug for systems that have nl_langinfo but CODESET undefined.

Added missing PQF transform rules for <= and >= . Thanks to Peter Popovics.

Added scan for the ZOOM shell (zoomsh).

--- 2.0 2003/02/23

* String Identifers for Schemas (Amendment 5).
http://lcweb.loc.gov/z3950/agency/amend/am5.html
Old definition for schema
  Odr_oid *schema; /* OPT */
New:
  int which;
  union {
     Odr_oid *oid;
     Z_InternationalString *uri;
  #define Z_Schema_oid 1
  #define Z_Schema_uri 2
  } schema; /* OPT */

* resultCount parameter to Sort Response (Amendment 1).
http://lcweb.loc.gov/z3950/agency/amend/am1.html

Support for SRW 1.0 over HTTP. This is an optional feature and
requires libxml2 to operate. Enable SOAP by specifying --with-xml2
for configure.

Generic frontend server supports HTTP/SOAP/SRW and Z39.50/BER
on the same port.  SRW SearchRetrieveRequests are mapped
to bend_init,bend_search,bend_fetch,bend_close. 

Z39.50 Query Type-104 added - to facilitate CQL within Z39.50.

CQL support. Source is directory 'cql'. CQL is supported in ZOOM
(both SRW and Z39.50) and the YAZ client (Z39.50 only).

ZOOM connections are SRW based if schem http is used for hostname
in connect, e.g.  z = ZOOM_connection_new("http://myserver");
ZOOM_query may be of type CQL, e.g. ZOOM_query_cql(q, "dc.title=x");

CCL proximity operators !n, %n converts to PQF @prox 0 n 1 2 k 2
and @prox 0 n 0 1 k 2 respectively, meaning: exlusion=false,
distance=n, order=true/false, relation=le, prox unit=word. If
n is omitted, distance 1 is used.

Fixed problem with C compiler include path in wrong order.

For TCP/IP COMSTACK, set recv buffer to an appropriate "large" value
on Solaris. Patch from Ko van der Sloot.

New MARC decode API. All new functions operate on a yaz_marc_t
handle. The most important new functions are: yaz_marc_create,
yaz_marc_decode_{buf,wrbuf}, yaz_marc_destroy, .. to create handler,
decode and destroy respectively. Decoder can produce formats
MARC line, simple XML, OAI MARC and MARC XML (LoC). See
include/yaz/marcdisp.h.

YAZ Iconv utility now supports MARC8 decoding (marc8.c). Converts
to UTF-8, UCS-32, wchar_t or Latin-1.

* Prototypes for yaz_iconv_-functions moved to separate header 
include/yaz/yaz-iconv.h.

Make a few private functions 'static' in unix.c. This prevents
duplicate unix_close in PHP with YAZ and imap. PHP Bug 20977.

For ZOOM connection, the options targetImplementation{Id,Name,Version}
are set when Init Response is received.

New function ZOOM_connection_error_x similar to ZOOM_connection_error
but returns diagnostic set as well.

New function yaz_strerror which is a portable wrapper for
strerror/strerror_r/GetLastMessage.

* ZOOM_record_get supports type "xml" in which case MARC XML (from LOC)
is returned for MARC. If type is "oai", then OAI MARC is returned.

Fix creation of lib/yaz-config so it works if srcdir != objdir. Patch
from Kang-Jin Lee.

--- 1.9.2 2002/11/26

yaz_marc_decode uses OAI MARC as XML format for ISO2709 records.

Fix bad race conditions on SIGTERM and terminate gracefully in
frontend server.

Move retrieval (data1) module to Zebra since nobody seems to be
using it.

Add GRS-1 render for ZOOM using same format as yaz-client.

--- 1.9.1 2002/10/05

Added man pages: yaz-client(1), yaz-ztest(8), yaz(7), zoomsh(1).

Fix broken ESpec handling in data1 module.

SGML reader parses comments and makes them part of the data1 tree.

Added several type casts so that YAZ compiles as C++ code.

ZOOM didn't handle PDU encoding failures properly.

Added more Bib-1 diagnostics (approved extensions).

GNU configure: better check for iconv.

More portable use of accept, getpeername. Check for socklen_t. Use
int/size_t if socklen_t is not defined.

--- 1.9 2002/09/06

PQF parser rejects bad queries - including those with extra
characters in them. PQF parser used to silently ignore that.
Cleaner API for PQF added, which allows you to get detailed
error information for bad queries (see yaz/pquery.h).

Fixed bug in ZOOM_record_get for type "database".

Implemented mini iconv library supporting conversions between
UTF-8, UCS4, UCS4LE and ISO-8859-1. Implemented in util/siconv.c.

* Removed XML reader (d1_expat.c). It's part of Zebra instead.

--- 1.8.9 2002/08/20

Fixed ZOOM C record cache so that preferredRecordSyntax and
elementSetName is used to check for already fetched records.

--- 1.8.8 2002/08/02

Added three members in bend_initrequest structure to facilitate
character set negotiation.

Fixed a bug in frontend server that could cause indefinite
loops under rare conditions. 

Revised character set utilities (charneg.h) so that encoding
UCS-4, UCSA-4, UTF-16 and UTF-8 are encoded non-privately.

For PQF, directive @term <type> sets term type for Scan/Search.
Here <type> is general, string, numeric, null. The term type
is inherited - just like @attr.

New call odr_set_charset that sets character set conversion for
international-strings on a ODR stream. Prototype is:
 int odr_set_charset(ODR o, const char *to, const char *from);
The InternationalString is still represented as C string in YAZ 
so native strings of type UTF-8, ASCII, etc. will work (UTF-16
won't). On the protocol level, you can use any encoding.

Support for UNIX sockets in ZOOM.

Solaris recv sometimes returns errno == ENOENT. It is treated as
EINPROGRESS. COMSTACK patch by Ko van der Sloot.

New COMSTACK of type UNIX Socket. Use "unix:/path" as address for
both client and server. Code by Morten B鴊eskov.

--- 1.8.7 2002/05/22

On Windows ZOOM functions are of type __stdcall rather than __cdecl, so
that they can be used directly from Visual Basic. You must recompile
your windows APP if you're using ZOOM.

Character Set and Language Negotiation(3) for ZOOM API and YAZ client.
Commands "lang" and "charset" sets language and character set respectively.
Options "lang" and "charset" is interpreted by the ZOOM API. Note, these
are not part of the official ZOOM spec) All is due to Oleg Kolobov. 

Attributes from multiple CCL fields may be OR'ed rather than merged
using directive "@field or". Fields are still merged by default -
equivalent to "@field merge".

Fixed bug in ZOOM connect that occurred on OpenBSD (maybe others). Patch
by Oleg Kolobov.

Fixed bug in zget for parameter ExtendedServicesResponse.

Fix OID for ILL.

--- 1.8.6 2002/03/25

Added PQF feature: '@attr type=value' may be written as '@type=value'.

Client displays USR.1:SearchResult-1 (hits per term).

Additional Search Info for search in Generic Frontend Server.

Fixed CCL parsing of "field=s".

WIN32 installer - using the excellent NSIS software from nullsoft.

Fixed CCL parsing of "field = - <number>".

Added function yaz_marc_decode that decodes MARC for line mode
display or XML (similar to that used by JZKit).

Fixed BER decoding of OPTIONAL CHOICE. The bug didn't affect the
Z39.50. But ISO/ILL was, however.

Added patch by Rustam T. Usmanov that implements Z39.50 duplicate
detection service (Z39.50-1995 Amendment 2) and Language Negotiation (3).
See:
 http://lcweb.loc.gov/z3950/agency/amend/am2.html
 http://lcweb.loc.gov/z3950/agency/defns/charneg-3.html

--- 1.8.5 2002/02/03

New YAZ client features: Command completion and new commands:
push_command, set_apdufile, set_marcdump, set_cclfields, register_oid.
Code provided by Jacob Poulsen.

New call oid_trav, that visits all registered OIDs via callback.

Function ZOOM_record_get now returns const char pointer rather than
void pointer. It saves a lot of type casts.

For YAZ client, command "auth" now allows you to set authentication
in IdPass style. It is enabled when three tokens are given - order
is user, group and password.

Fix memory leaks in MARC record handling for ZOOM.

New option for YAZ client, -k <kilobytes> that sets