oid.c

harvest是一个下载html网页得机器人

/*
 * Copyright (c) 1995-2003, Index Data
 * See the file LICENSE for details.
 *
 * $Id: oid.c,v 1.67 2003/09/02 12:12:13 adam Exp $
 */

/*
 * More or less protocol-transparent OID database.
 * We could (and should?) extend this so that the user app can add new
 * entries to the list at initialization.
 */
#if HAVE_CONFIG_H
#include <config.h>
#endif

#include <stdlib.h>
#include <string.h>
#include <ctype.h>

#include <yaz/oid.h>
#include <yaz/yaz-util.h>

static int z3950_prefix[] = { 1, 2, 840, 10003, -1 };
static int sr_prefix[]    = { 1, 0, 10163, -1 };

struct oident_list {
    struct oident oident;
    struct oident_list *next;
};

static struct oident_list *oident_table = NULL;
static int oid_value_dynamic = VAL_DYNAMIC;
static int oid_init_flag = 0;
static NMEM_MUTEX oid_mutex = 0;
static NMEM oid_nmem = 0;

/*
 * OID database
 */
static oident standard_oids[] =
{
    /* General definitions */
    {PROTO_GENERAL, CLASS_TRANSYN, VAL_BER,          {2,1,1,-1},
     "BER" },
    {PROTO_GENERAL, CLASS_TRANSYN, VAL_ISO2709,      {1,0,2709,1,1,-1},
     "ISO2709"},
    {PROTO_GENERAL, CLASS_GENERAL, VAL_ISO_ILL_1,    {1,0,10161,2,1,-1},
     "ISOILL-1"},
    /* Z39.50v3 definitions */
    {PROTO_Z3950,   CLASS_ABSYN,   VAL_APDU,         {2,1,-1},
     "Z-APDU"},    
    {PROTO_Z3950,   CLASS_APPCTX,  VAL_BASIC_CTX,    {1,1,-1},
     "Z-BASIC"},
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_BIB1,         {3,1,-1},
     "Bib-1"},
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_EXP1,         {3,2,-1},
     "Exp-1"},
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_EXT1,         {3,3,-1},
     "Ext-1"},
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_CCL1,         {3,4,-1},
     "CCL-1"},
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_GILS,         {3,5,-1},
     "GILS-attset"},
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_GILS,         {3,5,-1},
     "GILS"},
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_STAS,         {3,6,-1},
     "STAS-attset"},
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_COLLECT1,     {3,7,-1},
     "Collections-attset"},
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_CIMI1,        {3,8,-1},
     "CIMI-attset"},
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_GEO,          {3,9,-1},
     "Geo-attset"},

    {PROTO_Z3950,   CLASS_ATTSET,  VAL_ZBIG,         {3,10,-1},
     "ZBIG"},
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_UTIL,         {3,11,-1},
     "Util"},
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_XD1,          {3,12,-1},
     "XD-1"},
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_ZTHES,        {3,13,-1},
     "Zthes"},
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_FIN1,         {3,14,-1},
     "Fin-1"},
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_DAN1,         {3,15,-1},
     "Dan-1"},
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_HOLDINGS,     {3,16,-1},
     "Holdings"},
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_USMARC,       {3,17,-1},
     "MARC"},
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_BIB2,         {3,18,-1},
     "Bib-2"},
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_ZEEREX,       {3,19,-1},
     "ZeeRex"},
#if 0  /* This is the spawn of Satan.  Use Zthes-1 instead */
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_THESAURUS,    {3,1000,81,1,-1},     
     "Thesaurus-attset"},
#endif /*0*/
    {PROTO_Z3950,   CLASS_ATTSET,  VAL_IDXPATH,      {3,1000,81,2,-1},
     "IDXPATH"},
    {PROTO_Z3950,   CLASS_DIAGSET, VAL_BIB1,         {4,1,-1},
     "Bib-1"},
    {PROTO_Z3950,   CLASS_DIAGSET, VAL_DIAG1,        {4,2,-1},
     "Diag-1"},
    {PROTO_Z3950,   CLASS_DIAGSET, VAL_DIAG_ES,      {4,3,-1},
     "Diag-ES"},
    {PROTO_Z3950,   CLASS_DIAGSET, VAL_DIAG_GENERAL, {4,3,-1},
     "Diag-General"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_UNIMARC,      {5,1,-1},
     "Unimarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_INTERMARC,    {5,2,-1},
     "Intermarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_CCF,          {5,3,-1},
     "CCF"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_USMARC,       {5,10,-1},
     "USmarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_UKMARC,       {5,11,-1},
     "UKmarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_NORMARC,      {5,12,-1},
     "Normarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_LIBRISMARC,   {5,13,-1},
     "Librismarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_DANMARC,      {5,14,-1},
     "Danmarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_FINMARC,      {5,15,-1},
     "Finmarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_MAB,          {5,16,-1},
     "MAB"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_CANMARC,      {5,17,-1},
     "Canmarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_SBN,          {5,18,-1},
     "SBN"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_PICAMARC,     {5,19,-1},
     "Picamarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_AUSMARC,      {5,20,-1},
     "Ausmarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_IBERMARC,     {5,21,-1},
     "Ibermarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_CATMARC,      {5,22,-1},
     "Carmarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_MALMARC,      {5,23,-1},
     "Malmarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_JPMARC,       {5,24,-1},
     "JPmarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_SWEMARC,      {5,25,-1},
     "SWEmarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_SIGLEMARC,    {5,26,-1},
     "SIGLEmarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_ISDSMARC,     {5,27,-1},
     "ISDSmarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_RUSMARC,      {5,28,-1},
     "RUSmarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_HUNMARC,      {5,29,-1},
     "Hunmarc"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_EXPLAIN,      {5,100,-1},
     "Explain"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_SUTRS,        {5,101,-1},
     "SUTRS"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_OPAC,         {5,102,-1},
     "OPAC"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_SUMMARY,      {5,103,-1},
     "Summary"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_GRS0,         {5,104,-1},
     "GRS-0"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_GRS1,         {5,105,-1},
     "GRS-1"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_EXTENDED,     {5,106,-1},
     "Extended"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_FRAGMENT,     {5,107,-1},
     "Fragment"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_PDF,          {5,109,1,-1},
     "pdf"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_POSTSCRIPT,   {5,109,2,-1},
     "postscript"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_HTML,         {5,109,3,-1},
     "html"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_TIFF,         {5,109,4,-1},
     "tiff"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_GIF,          {5,109,5,-1},
     "gif"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_JPEG,         {5,109,6,-1},
     "jpeg"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_PNG,          {5,109,7,-1},
     "png"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_MPEG,         {5,109,8,-1},
     "mpeg"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_SGML,         {5,109,9,-1},
     "sgml"},

    {PROTO_Z3950,   CLASS_RECSYN,  VAL_TIFFB,        {5,110,1,-1},
     "tiff-b"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_WAV,          {5,110,2,-1},
     "wav"},

    {PROTO_Z3950,   CLASS_RECSYN,  VAL_SQLRS,        {5,111,-1},
     "SQL-RS"},
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_SOIF,         {5,1000,81,2,-1},
     "SOIF" },
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_TEXT_XML,     {5,109,10,-1},
     "text-XML" },
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_TEXT_XML,     {5,109,10,-1},
     "XML" },
    {PROTO_Z3950,   CLASS_RECSYN,  VAL_APPLICATION_XML, {5,109,11,-1},
     "application-XML" },
    {PROTO_Z3950,   CLASS_RESFORM, VAL_RESOURCE1,    {7,1,-1},
     "Resource-1"},
    {PROTO_Z3950,   CLASS_RESFORM, VAL_RESOURCE2,    {7,2,-1},
     "Resource-2"},
    {PROTO_Z3950,   CLASS_RESFORM, VAL_UNIVERSE_REPORT, {7,1000,81,1,-1},
     "UNIverse-Resource-Report"},

    {PROTO_Z3950,   CLASS_ACCFORM, VAL_PROMPT1,      {8,1,-1},
     "Prompt-1"},
    {PROTO_Z3950,   CLASS_ACCFORM, VAL_DES1,         {8,2,-1},
     "Des-1"},
    {PROTO_Z3950,   CLASS_ACCFORM, VAL_KRB1,         {8,3,-1},
     "Krb-1"},
    {PROTO_Z3950,   CLASS_EXTSERV, VAL_PRESSET,      {9,1,-1},
     "Pers. set"},
    {PROTO_Z3950,   CLASS_EXTSERV, VAL_PQUERY,       {9,2,-1},
     "Pers. query"},
    {PROTO_Z3950,   CLASS_EXTSERV, VAL_PCQUERY,      {9,3,-1},
     "Per'd query"},
    {PROTO_Z3950,   CLASS_EXTSERV, VAL_ITEMORDER,    {9,4,-1},
     "Item order"},
    {PROTO_Z3950,   CLASS_EXTSERV, VAL_DBUPDATE0,    {9,5,-1},
     "DB. Update (first version)"},
    {PROTO_Z3950,   CLASS_EXTSERV, VAL_DBUPDATE1,    {9,5,1,-1},
     "DB. Update (second version)"},
    {PROTO_Z3950,   CLASS_EXTSERV, VAL_DBUPDATE,     {9,5,1,1,-1},
     "DB. Update"},
    {PROTO_Z3950,   CLASS_EXTSERV, VAL_EXPORTSPEC,   {9,6,-1},
     "exp. spec."},
    {PROTO_Z3950,   CLASS_EXTSERV, VAL_EXPORTINV,    {9,7,-1},
     "exp. inv."},
    {PROTO_Z3950,   CLASS_EXTSERV, VAL_ADMINSERVICE, {9,1000,81,1,-1},
     "Admin"},
    {PROTO_Z3950,   CLASS_USERINFO,VAL_SEARCHRES1,   {10,1,-1},
     "searchResult-1"},
    {PROTO_Z3950,   CLASS_USERINFO,VAL_CHARLANG,     {10,2,-1},
     "CharSetandLanguageNegotiation"},
    {PROTO_Z3950,   CLASS_USERINFO,VAL_USERINFO1,    {10,3,-1},
     "UserInfo-1"},
    {PROTO_Z3950,   CLASS_USERINFO,VAL_MULTISRCH1,   {10,4,-1},
     "MultipleSearchTerms-1"},
    {PROTO_Z3950,   CLASS_USERINFO,VAL_MULTISRCH2,   {10,5,-1},
     "MultipleSearchTerms-2"},
    {PROTO_Z3950,   CLASS_USERINFO,VAL_DATETIME,     {10,6,-1},
     "DateTime"},
    {PROTO_Z3950,   CLASS_USERINFO,VAL_PROXY,        {10,1000,81,1,-1},
     "Proxy" },
    {PROTO_Z3950,   CLASS_USERINFO,VAL_COOKIE,       {10,1000,81,2,-1},
     "Cookie" },
    {PROTO_Z3950,   CLASS_USERINFO,VAL_CLIENT_IP,    {10,1000,81,3,-1},
     "Client-IP" },
    {PROTO_Z3950,   CLASS_ELEMSPEC,VAL_ESPEC1,       {11,1,-1},
     "Espec-1"},
    {PROTO_Z3950,   CLASS_VARSET,  VAL_VAR1,         {12,1,-1},
     "Variant-1"},
    {PROTO_Z3950,   CLASS_SCHEMA,  VAL_WAIS,         {13,1,-1},
     "WAIS-schema"},
    {PROTO_Z3950,   CLASS_SCHEMA,  VAL_GILS,         {13,2,-1},
     "GILS-schema"},
    {PROTO_Z3950,   CLASS_SCHEMA,  VAL_COLLECT1,     {13,3,-1},
     "Collections-schema"},
    {PROTO_Z3950,   CLASS_SCHEMA,  VAL_GEO,          {13,4,-1},
     "Geo-schema"},
    {PROTO_Z3950,   CLASS_SCHEMA,  VAL_CIMI1,        {13,5,-1},
     "CIMI-schema"},
    {PROTO_Z3950,   CLASS_SCHEMA,  VAL_UPDATEES,     {13,6,-1},
     "Update ES"},
    {PROTO_Z3950,   CLASS_SCHEMA,  VAL_HOLDINGS,     {13,7,-1},
     "Holdings"},
    {PROTO_Z3950,   CLASS_SCHEMA,  VAL_ZTHES,        {13,8,-1},
     "Zthes"},
    {PROTO_Z3950,   CLASS_SCHEMA,  VAL_THESAURUS,    {13,1000,81,1,-1},
     "thesaurus-schema"},
    {PROTO_Z3950,   CLASS_SCHEMA,  VAL_EXPLAIN,      {13,1000,81,2,-1},
     "Explain-schema"},
    {PROTO_Z3950,   CLASS_TAGSET,  VAL_SETM,         {14,1,-1},
     "TagsetM"},
    {PROTO_Z3950,   CLASS_TAGSET,  VAL_SETG,         {14,2,-1},
     "TagsetG"},
    {PROTO_Z3950,   CLASS_TAGSET,  VAL_STAS,         {14,3,-1},
     "STAS-tagset"},
    {PROTO_Z3950,   CLASS_TAGSET,  VAL_GILS,         {14,4,-1},
     "GILS-tagset"},
    {PROTO_Z3950,   CLASS_TAGSET,  VAL_COLLECT1,     {14,5,-1},
     "Collections-tagset"},
    {PROTO_Z3950,   CLASS_TAGSET,  VAL_CIMI1,        {14,6,-1},
     "CIMI-tagset"},
    {PROTO_Z3950,   CLASS_TAGSET,  VAL_THESAURUS,    {14,1000,81,1,-1},
     "thesaurus-tagset"},	/* What is this Satan-spawn doing here? */
    {PROTO_Z3950,   CLASS_TAGSET,  VAL_EXPLAIN,      {14,1000,81,2,-1},
     "Explain-tagset"},
    {PROTO_Z3950,   CLASS_TAGSET,  VAL_ZTHES,        {14,8,-1},
     "Zthes-tagset"},
    {PROTO_Z3950,   CLASS_NEGOT,   VAL_CHARNEG3,     {15,3,-1},
     "CharSetandLanguageNegotiation-3"},
    {PROTO_Z3950,   CLASS_NEGOT,   VAL_ID_CHARSET,   {15,1000,81,1,-1},
     "ID-Charset" },
    {PROTO_Z3950,   CLASS_USERINFO,VAL_CQL,          {16, 2, -1},
     "CQL"},
    {PROTO_GENERAL, CLASS_GENERAL, VAL_UCS2,    {1,0,10646,1,0,2,-1},
     "UCS-2"},
    {PROTO_GENERAL, CLASS_GENERAL, VAL_UCS4,    {1,0,10646,1,0,4,-1},
     "UCS-4"},