index/mod_dom.c

   1 /* This file is part of the Zebra server.
   2    Copyright (C) 1995-2008 Index Data
   3
   4 Zebra is free software; you can redistribute it and/or modify it under
   5 the terms of the GNU General Public License as published by the Free
   6 Software Foundation; either version 2, or (at your option) any later
   7 version.
   8
   9 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
  10 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  11 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  12 for more details.
  13
  14 You should have received a copy of the GNU General Public License
  15 along with this program; if not, write to the Free Software
  16 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  17
  18 */
  19
  20 #include <stdio.h>
  21 #include <assert.h>
  22 #include <ctype.h>
  23 #include <stdarg.h>
  24
  25 #include <yaz/diagbib1.h>
  26 #include <yaz/tpath.h>
  27 #include <yaz/snprintf.h>
  28
  29 #include <libxml/xmlversion.h>
  30 #include <libxml/parser.h>
  31 #include <libxml/tree.h>
  32 #include <libxml/xmlIO.h>
  33 #include <libxml/xmlreader.h>
  34 #include <libxslt/transform.h>
  35 #include <libxslt/xsltutils.h>
  36
  37 #if YAZ_HAVE_EXSLT
  38 #include <libexslt/exslt.h>
  39 #endif
  40
  41 #include <idzebra/util.h>
  42 #include <idzebra/recctrl.h>
  43 #include <yaz/oid_db.h>
  44
  45 /* DOM filter style indexing */
  46 #define ZEBRA_DOM_NS "http://indexdata.com/zebra-2.0"
  47 static const char *zebra_dom_ns = ZEBRA_DOM_NS;
  48
  49 /* DOM filter style indexing */
  50 #define ZEBRA_PI_NAME "zebra-2.0"
  51 static const char *zebra_pi_name = ZEBRA_PI_NAME;
  52
  53 enum convert_type {
  54     convert_xslt_type,
  55     convert_meta_type
  56 };
  57
  58 struct convert_xslt {
  59     const char *stylesheet;
  60     xsltStylesheetPtr stylesheet_xsp;
  61 };
  62
  63 struct convert_meta {
  64     int dummy;
  65 };
  66
  67 struct convert_s {
  68     enum convert_type which;
  69     union {
  70         struct convert_xslt xslt;
  71         struct convert_meta meta;
  72     } u;
  73     struct convert_s *next;
  74 };
  75
  76 struct filter_extract {
  77     const char *name;
  78     struct convert_s *convert;
  79 };
  80
  81 struct filter_store {
  82     struct convert_s *convert;
  83 };
  84
  85 struct filter_retrieve {
  86     const char *name;
  87     const char *identifier;
  88     struct convert_s *convert;
  89     struct filter_retrieve *next;
  90 };
  91
  92 #define DOM_INPUT_XMLREADER 1
  93 #define DOM_INPUT_MARC 2
  94 struct filter_input {
  95     const char *syntax;
  96     const char *name;
  97     struct convert_s *convert;
  98     int type;
  99     union {
 100         struct {
 101             xmlTextReaderPtr reader;
 102             int split_level;
 103         } xmlreader;
 104         struct {
 105             const char *input_charset;
 106             yaz_marc_t handle;
 107             yaz_iconv_t iconv;
 108         } marc;
 109     } u;
 110     struct filter_input *next;
 111 };
 112
 113 struct filter_info {
 114     char *fname;
 115     char *full_name;
 116     const char *profile_path;
 117     NMEM nmem_record;
 118     NMEM nmem_config;
 119     xmlDocPtr doc_config;
 120     struct filter_extract *extract;
 121     struct filter_retrieve *retrieve_list;
 122     struct filter_input *input_list;
 123     struct filter_store *store;
 124     int record_info_invoked;
 125 };
 126
 127
 128
 129 #define XML_STRCMP(a,b)   strcmp((char*)a, b)
 130 #define XML_STRLEN(a) strlen((char*)a)
 131
 132
 133 #define FOR_EACH_ELEMENT(ptr) for (; ptr; ptr = ptr->next) if (ptr->type == XML_ELEMENT_NODE)
 134
 135 static void dom_log(int level, struct filter_info *tinfo, xmlNodePtr ptr,
 136                     const char *fmt, ...)
 137 #ifdef __GNUC__
 138     __attribute__ ((format (printf, 4, 5)))
 139 #endif
 140     ;
 141
 142 static void dom_log(int level, struct filter_info *tinfo, xmlNodePtr ptr,
 143                     const char *fmt, ...)
 144 {
 145     va_list ap;
 146     char buf[4096];
 147
 148     va_start(ap, fmt);
 149     yaz_vsnprintf(buf, sizeof(buf)-1, fmt, ap);
 150     if (ptr)
 151     {
 152         yaz_log(level, "%s:%ld: %s", tinfo->fname ? tinfo->fname : "none",
 153                 xmlGetLineNo(ptr), buf);
 154     }
 155     else
 156     {
 157         yaz_log(level, "%s: %s", tinfo->fname ? tinfo->fname : "none", buf);
 158     }
 159     va_end(ap);
 160 }
 161
 162
 163 static void set_param_str(const char **params, const char *name,
 164                           const char *value, NMEM nmem)
 165 {
 166     char *quoted = nmem_malloc(nmem, 3 + strlen(value));
 167     sprintf(quoted, "'%s'", value);
 168     while (*params)
 169         params++;
 170     params[0] = name;
 171     params[1] = quoted;
 172     params[2] = 0;
 173 }
 174
 175 static void set_param_int(const char **params, const char *name,
 176                           zint value, NMEM nmem)
 177 {
 178     char *quoted = nmem_malloc(nmem, 30); /* 25 digits enough for 2^64 */
 179     while (*params)
 180         params++;
 181     sprintf(quoted, "'" ZINT_FORMAT "'", value);
 182     params[0] = name;
 183     params[1] = quoted;
 184     params[2] = 0;
 185 }
 186
 187 static void *filter_init(Res res, RecType recType)
 188 {
 189     struct filter_info *tinfo = (struct filter_info *) xmalloc(sizeof(*tinfo));
 190     tinfo->fname = 0;
 191     tinfo->full_name = 0;
 192     tinfo->profile_path = 0;
 193     tinfo->nmem_record = nmem_create();
 194     tinfo->nmem_config = nmem_create();
 195     tinfo->extract = 0;
 196     tinfo->retrieve_list = 0;
 197     tinfo->input_list = 0;
 198     tinfo->store = 0;
 199     tinfo->doc_config = 0;
 200     tinfo->record_info_invoked = 0;
 201
 202 #if YAZ_HAVE_EXSLT
 203     exsltRegisterAll();
 204 #endif
 205
 206     return tinfo;
 207 }
 208
 209 static int attr_content(struct _xmlAttr *attr, const char *name,
 210                         const char **dst_content)
 211 {
 212     if (!XML_STRCMP(attr->name, name) && attr->children
 213         && attr->children->type == XML_TEXT_NODE)
 214     {
 215         *dst_content = (const char *)(attr->children->content);
 216         return 1;
 217     }
 218     return 0;
 219 }
 220
 221 static void destroy_xsp(struct convert_s *c)
 222 {
 223     while (c)
 224     {
 225         if (c->which == convert_xslt_type)
 226         {
 227             if (c->u.xslt.stylesheet_xsp)
 228                 xsltFreeStylesheet(c->u.xslt.stylesheet_xsp);
 229         }
 230         c = c->next;
 231     }
 232 }
 233
 234 static void destroy_dom(struct filter_info *tinfo)
 235 {
 236     if (tinfo->extract)
 237     {
 238         destroy_xsp(tinfo->extract->convert);
 239         tinfo->extract = 0;
 240     }
 241     if (tinfo->store)
 242     {
 243         destroy_xsp(tinfo->store->convert);
 244         tinfo->store = 0;
 245     }
 246     if (tinfo->input_list)
 247     {
 248         struct filter_input *i_ptr;
 249         for (i_ptr = tinfo->input_list; i_ptr; i_ptr = i_ptr->next)
 250         {
 251             switch(i_ptr->type)
 252             {
 253             case DOM_INPUT_XMLREADER:
 254                 if (i_ptr->u.xmlreader.reader)
 255                     xmlFreeTextReader(i_ptr->u.xmlreader.reader);
 256                 break;
 257             case DOM_INPUT_MARC:
 258                 yaz_iconv_close(i_ptr->u.marc.iconv);
 259                 yaz_marc_destroy(i_ptr->u.marc.handle);
 260                 break;
 261             }
 262             destroy_xsp(i_ptr->convert);
 263         }
 264         tinfo->input_list = 0;
 265     }
 266     if (tinfo->retrieve_list)
 267     {
 268         struct filter_retrieve *r_ptr;
 269         for (r_ptr = tinfo->retrieve_list; r_ptr; r_ptr = r_ptr->next)
 270             destroy_xsp(r_ptr->convert);
 271         tinfo->retrieve_list = 0;
 272     }
 273
 274     if (tinfo->doc_config)
 275     {
 276         xmlFreeDoc(tinfo->doc_config);
 277         tinfo->doc_config = 0;
 278     }
 279     nmem_reset(tinfo->nmem_config);
 280 }
 281
 282 static ZEBRA_RES parse_convert(struct filter_info *tinfo, xmlNodePtr ptr,
 283                                struct convert_s **l)
 284 {
 285     *l = 0;
 286     FOR_EACH_ELEMENT(ptr) {
 287         if (!XML_STRCMP(ptr->name, "xslt"))
 288         {
 289             struct _xmlAttr *attr;
 290             struct convert_s *p = nmem_malloc(tinfo->nmem_config, sizeof(*p));
 291
 292             p->next = 0;
 293             p->which = convert_xslt_type;
 294             p->u.xslt.stylesheet = 0;
 295             p->u.xslt.stylesheet_xsp = 0;
 296
 297             for (attr = ptr->properties; attr; attr = attr->next)
 298                 if (attr_content(attr, "stylesheet", &p->u.xslt.stylesheet))
 299                     ;
 300                 else
 301                 {
 302                     dom_log(YLOG_WARN, tinfo, ptr,
 303                             "bad attribute @%s", attr->name);
 304                 }
 305             if (p->u.xslt.stylesheet)
 306             {
 307                 char tmp_xslt_full_name[1024];
 308                 if (!yaz_filepath_resolve(p->u.xslt.stylesheet,
 309                                           tinfo->profile_path,
 310                                           NULL,
 311                                           tmp_xslt_full_name))
 312                 {
 313                     dom_log(YLOG_WARN, tinfo, 0,
 314                             "stylesheet %s not found in "
 315                             "path %s",
 316                             p->u.xslt.stylesheet,
 317                             tinfo->profile_path);
 318                     return ZEBRA_FAIL;
 319                 }
 320
 321                 p->u.xslt.stylesheet_xsp
 322                     = xsltParseStylesheetFile((const xmlChar*)
 323                                               tmp_xslt_full_name);
 324                 if (!p->u.xslt.stylesheet_xsp)
 325                 {
 326                     dom_log(YLOG_WARN, tinfo, 0,
 327                             "could not parse xslt stylesheet %s",
 328                             tmp_xslt_full_name);
 329                     return ZEBRA_FAIL;
 330                 }
 331             }
 332             else
 333             {
 334                 dom_log(YLOG_WARN, tinfo, ptr,
 335                         "missing attribute 'stylesheet'");
 336                 return ZEBRA_FAIL;
 337             }
 338             *l = p;
 339             l = &p->next;
 340         }
 341         else if (!XML_STRCMP(ptr->name, "process-meta"))
 342         {
 343             struct _xmlAttr *attr;
 344             struct convert_s *p = nmem_malloc(tinfo->nmem_config, sizeof(*p));
 345
 346             p->next = 0;
 347             p->which = convert_meta_type;
 348
 349             for (attr = ptr->properties; attr; attr = attr->next)
 350                 dom_log(YLOG_WARN, tinfo, ptr,
 351                         "bad attribute @%s", attr->name);
 352             *l = p;
 353             l = &p->next;
 354         }
 355         else
 356         {
 357             dom_log(YLOG_WARN, tinfo, ptr,
 358                     "bad element '%s', expected <xslt>", ptr->name);
 359             return ZEBRA_FAIL;
 360         }
 361     }
 362     return ZEBRA_OK;
 363 }
 364
 365 static int process_meta(struct filter_info *tinfo, xmlDocPtr doc, xmlNodePtr node,
 366                         struct recRetrieveCtrl *retctr)
 367 {
 368
 369     if (node->type == XML_ELEMENT_NODE && node->ns && node->ns->href &&
 370         0 == XML_STRCMP(node->ns->href, zebra_dom_ns))
 371     {
 372         if (0 == XML_STRCMP(node->name, "meta"))
 373         {
 374             const char *element_set_name = 0;
 375
 376             struct _xmlAttr *attr;
 377             for (attr = node->properties; attr; attr = attr->next)
 378             {
 379                 if (attr_content(attr, "name", &element_set_name))
 380                     ;
 381                 else
 382                 {
 383                     dom_log(YLOG_WARN, tinfo, node,
 384                             "bad attribute @%s, expected @name", attr->name);
 385                 }
 386             }
 387             if (element_set_name)
 388             {
 389                 WRBUF result = wrbuf_alloc();
 390                 WRBUF addinfo = wrbuf_alloc();
 391                 const Odr_oid *input_format = yaz_oid_recsyn_xml;
 392                 const Odr_oid *output_format = 0;
 393                 int ret;
 394
 395                 ret = retctr->special_fetch(retctr->handle,
 396                                             element_set_name,
 397                                             input_format, &output_format,
 398                                             result, addinfo);
 399                 if (ret == 0)
 400                 {
 401                     xmlDocPtr sub_doc =
 402                         xmlParseMemory(    wrbuf_buf(result), wrbuf_len(result));
 403                     if (sub_doc)
 404                     {
 405                         xmlNodePtr t = xmlDocGetRootElement(sub_doc);
 406                         xmlAddChild(node, xmlCopyNode(t, 1));
 407                         xmlFreeDoc(sub_doc);
 408                     }
 409                 }
 410                 wrbuf_destroy(result);
 411                 wrbuf_destroy(addinfo);
 412             }
 413         }
 414     }
 415     for (node = node->children; node; node = node->next)
 416         process_meta(tinfo, doc, node, retctr);
 417     return 0;
 418 }
 419
 420 static ZEBRA_RES perform_convert(struct filter_info *tinfo,
 421                                  struct recExtractCtrl *extctr,
 422                                  struct recRetrieveCtrl *retctr,
 423                                  struct convert_s *convert,
 424                                  const char **params,
 425                                  xmlDocPtr *doc,
 426                                  xsltStylesheetPtr *last_xsp)
 427 {
 428     for (; convert; convert = convert->next)
 429     {
 430         if (convert->which == convert_xslt_type)
 431         {
 432             xmlChar *buf_out = 0;
 433             int len_out = 0;
 434             xmlDocPtr res_doc = xsltApplyStylesheet(convert->u.xslt.stylesheet_xsp,
 435                                                     *doc, params);
 436             if (last_xsp)
 437                 *last_xsp = convert->u.xslt.stylesheet_xsp;
 438
 439             if (!res_doc)
 440                 break;
 441
 442             /* now saving into buffer and re-reading into DOM to avoid annoing
 443                XSLT problem with thrown-out indentation text nodes */
 444             xsltSaveResultToString(&buf_out, &len_out, res_doc,
 445                                    convert->u.xslt.stylesheet_xsp);
 446             xmlFreeDoc(res_doc);
 447
 448             xmlFreeDoc(*doc);
 449
 450             *doc = xmlParseMemory((const char *) buf_out, len_out);
 451
 452             /* writing debug info out */
 453             if (extctr && extctr->flagShowRecords)
 454                 yaz_log(YLOG_LOG, "%s: XSLT %s\n %.*s",
 455                         tinfo->fname ? tinfo->fname : "(none)",
 456                         convert->u.xslt.stylesheet,
 457                         len_out, buf_out);
 458
 459             xmlFree(buf_out);
 460         }
 461         else if (convert->which == convert_meta_type)
 462         {
 463             if (retctr) /* only execute meta on retrieval */
 464             {
 465                 process_meta(tinfo, *doc, xmlDocGetRootElement(*doc), retctr);
 466
 467                 /* last stylesheet absent */
 468                 if (last_xsp)
 469                     *last_xsp = 0;
 470             }
 471         }
 472     }
 473     return ZEBRA_OK;
 474 }
 475
 476 static struct filter_input *new_input(struct filter_info *tinfo, int type)
 477 {
 478     struct filter_input *p;
 479     struct filter_input **np = &tinfo->input_list;
 480     for (;*np; np = &(*np)->next)
 481         ;
 482     p = *np = nmem_malloc(tinfo->nmem_config, sizeof(*p));
 483     p->next = 0;
 484     p->syntax = 0;
 485     p->name = 0;
 486     p->convert = 0;
 487     p->type = type;
 488     return p;
 489 }
 490
 491 static ZEBRA_RES parse_input(struct filter_info *tinfo, xmlNodePtr ptr,
 492                              const char *syntax, const char *name)
 493 {
 494     FOR_EACH_ELEMENT(ptr) {
 495         if (!XML_STRCMP(ptr->name, "marc"))
 496         {
 497             yaz_iconv_t iconv = 0;
 498             const char *input_charset = "marc-8";
 499             struct _xmlAttr *attr;
 500
 501             for (attr = ptr->properties; attr; attr = attr->next)
 502             {
 503                 if (attr_content(attr, "inputcharset", &input_charset))
 504                     ;
 505                 else
 506                 {
 507                     dom_log(YLOG_WARN, tinfo, ptr,
 508                             "bad attribute @%s, expected @inputcharset",
 509                             attr->name);
 510                 }
 511             }
 512             iconv = yaz_iconv_open("utf-8", input_charset);
 513             if (!iconv)
 514             {
 515                 dom_log(YLOG_WARN, tinfo, ptr,
 516                         "unsupported @charset '%s'", input_charset);
 517                 return ZEBRA_FAIL;
 518             }
 519             else
 520             {
 521                 struct filter_input *p
 522                     = new_input(tinfo, DOM_INPUT_MARC);
 523                 p->u.marc.handle = yaz_marc_create();
 524                 p->u.marc.iconv = iconv;
 525
 526                 yaz_marc_iconv(p->u.marc.handle, p->u.marc.iconv);
 527
 528                 ptr = ptr->next;
 529
 530                 parse_convert(tinfo, ptr, &p->convert);
 531             }
 532             break;
 533
 534         }
 535         else if (!XML_STRCMP(ptr->name, "xmlreader"))
 536         {
 537             struct filter_input *p
 538                 = new_input(tinfo, DOM_INPUT_XMLREADER);
 539             struct _xmlAttr *attr;
 540             const char *level_str = 0;
 541
 542             p->u.xmlreader.split_level = 0;
 543             p->u.xmlreader.reader = 0;
 544
 545             for (attr = ptr->properties; attr; attr = attr->next)
 546             {
 547                 if (attr_content(attr, "level", &level_str))
 548                     ;
 549                 else
 550                 {
 551                     dom_log(YLOG_WARN, tinfo, ptr,
 552                             "bad attribute @%s, expected @level",
 553                             attr->name);
 554                 }
 555             }
 556             if (level_str)
 557                 p->u.xmlreader.split_level = atoi(level_str);
 558
 559             ptr = ptr->next;
 560
 561             parse_convert(tinfo, ptr, &p->convert);
 562             break;
 563         }
 564         else
 565         {
 566             dom_log(YLOG_WARN, tinfo, ptr,
 567                     "bad element <%s>, expected <marc>|<xmlreader>",
 568                     ptr->name);
 569             return ZEBRA_FAIL;
 570         }
 571     }
 572     return ZEBRA_OK;
 573 }
 574
 575 static ZEBRA_RES parse_dom(struct filter_info *tinfo, const char *fname)
 576 {
 577     char tmp_full_name[1024];
 578     xmlNodePtr ptr;
 579     xmlDocPtr doc;
 580
 581     tinfo->fname = nmem_strdup(tinfo->nmem_config, fname);
 582
 583     if (yaz_filepath_resolve(tinfo->fname, tinfo->profile_path,
 584                              NULL, tmp_full_name))
 585         tinfo->full_name = nmem_strdup(tinfo->nmem_config, tmp_full_name);
 586     else
 587         tinfo->full_name = nmem_strdup(tinfo->nmem_config, tinfo->fname);
 588
 589     yaz_log(YLOG_LOG, "%s dom filter: "
 590             "loading config file %s", tinfo->fname, tinfo->full_name);
 591
 592     doc = xmlParseFile(tinfo->full_name);
 593     if (!doc)
 594     {
 595         yaz_log(YLOG_WARN, "%s: dom filter: "
 596                 "failed to parse config file %s",
 597                 tinfo->fname, tinfo->full_name);
 598         return ZEBRA_FAIL;
 599     }
 600     /* save because we store ptrs to the content */
 601     tinfo->doc_config = doc;
 602
 603     ptr = xmlDocGetRootElement(doc);
 604     if (!ptr || ptr->type != XML_ELEMENT_NODE
 605         || XML_STRCMP(ptr->name, "dom"))
 606     {
 607         dom_log(YLOG_WARN, tinfo, ptr,
 608                 "bad root element <%s>, expected root element <dom>",
 609                 ptr->name);
 610         return ZEBRA_FAIL;
 611     }
 612
 613     ptr = ptr->children;
 614     FOR_EACH_ELEMENT(ptr) {
 615         if (!XML_STRCMP(ptr->name, "extract"))
 616         {
 617             /*
 618               <extract name="index">
 619               <xslt stylesheet="first.xsl"/>
 620               <xslt stylesheet="second.xsl"/>
 621               </extract>
 622             */
 623             struct _xmlAttr *attr;
 624             struct filter_extract *f =
 625                 nmem_malloc(tinfo->nmem_config, sizeof(*f));
 626
 627             tinfo->extract = f;
 628             f->name = 0;
 629             f->convert = 0;
 630             for (attr = ptr->properties; attr; attr = attr->next)
 631             {
 632                 if (attr_content(attr, "name", &f->name))
 633                     ;
 634                 else
 635                 {
 636                     dom_log(YLOG_WARN, tinfo, ptr,
 637                             "bad attribute @%s, expected @name",
 638                             attr->name);
 639                 }
 640             }
 641             parse_convert(tinfo, ptr->children, &f->convert);
 642         }
 643         else if (!XML_STRCMP(ptr->name, "retrieve"))
 644         {
 645             /*
 646                <retrieve name="F">
 647                <xslt stylesheet="some.xsl"/>
 648                <xslt stylesheet="some.xsl"/>
 649                </retrieve>
 650             */
 651             struct _xmlAttr *attr;
 652             struct filter_retrieve **fp = &tinfo->retrieve_list;
 653             struct filter_retrieve *f =
 654                 nmem_malloc(tinfo->nmem_config, sizeof(*f));
 655
 656             while (*fp)
 657                 fp = &(*fp)->next;
 658
 659             *fp = f;
 660             f->name = 0;
 661             f->identifier = 0;
 662             f->convert = 0;
 663             f->next = 0;
 664
 665             for (attr = ptr->properties; attr; attr = attr->next)
 666             {
 667                 if (attr_content(attr, "identifier",
 668                                  &f->identifier))
 669                     ;
 670                 else if (attr_content(attr, "name", &f->name))
 671                     ;
 672                 else
 673                 {
 674                     dom_log(YLOG_WARN, tinfo, ptr,
 675                             "bad attribute @%s,  expected @identifier|@name",
 676                             attr->name);
 677                 }
 678             }
 679             parse_convert(tinfo, ptr->children, &f->convert);
 680         }
 681         else if (!XML_STRCMP(ptr->name, "store"))
 682         {
 683             /*
 684               <store name="F">
 685               <xslt stylesheet="some.xsl"/>
 686               <xslt stylesheet="some.xsl"/>
 687               </retrieve>
 688             */
 689             struct filter_store *f =
 690                 nmem_malloc(tinfo->nmem_config, sizeof(*f));
 691
 692             tinfo->store = f;
 693             f->convert = 0;
 694             parse_convert(tinfo, ptr->children, &f->convert);
 695         }
 696         else if (!XML_STRCMP(ptr->name, "input"))
 697         {
 698             /*
 699               <input syntax="xml">
 700               <xmlreader level="1"/>
 701               </input>
 702               <input syntax="usmarc">
 703               <marc inputcharset="marc-8"/>
 704               </input>
 705             */
 706             struct _xmlAttr *attr;
 707             const char  *syntax = 0;
 708             const char *name = 0;
 709             for (attr = ptr->properties; attr; attr = attr->next)
 710             {
 711                 if (attr_content(attr, "syntax", &syntax))
 712                     ;
 713                 else if (attr_content(attr, "name", &name))
 714                     ;
 715                 else
 716                 {
 717                     dom_log(YLOG_WARN, tinfo, ptr,
 718                             "bad attribute @%s,  expected @syntax|@name",
 719                             attr->name);
 720                 }
 721             }
 722             parse_input(tinfo, ptr->children, syntax, name);
 723         }
 724         else
 725         {
 726             dom_log(YLOG_WARN, tinfo, ptr,
 727                     "bad element <%s>, "
 728                     "expected <extract>|<input>|<retrieve>|<store>",
 729                     ptr->name);
 730             return ZEBRA_FAIL;
 731         }
 732     }
 733     if (!tinfo->input_list)
 734     {
 735         struct filter_input *p
 736             = new_input(tinfo, DOM_INPUT_XMLREADER);
 737         p->u.xmlreader.split_level = 0;
 738         p->u.xmlreader.reader = 0;
 739     }
 740     return ZEBRA_OK;
 741 }
 742
 743 static struct filter_retrieve *lookup_retrieve(struct filter_info *tinfo,
 744                                                const char *est)
 745 {
 746     struct filter_retrieve *f = tinfo->retrieve_list;
 747
 748     /* return first schema if no est is provided */
 749     if (!est)
 750         return f;
 751     for (; f; f = f->next)
 752     {
 753         /* find requested schema */
 754         if (est)
 755         {
 756             if (f->identifier && !strcmp(f->identifier, est))
 757                 return f;
 758             if (f->name && !strcmp(f->name, est))
 759                 return f;
 760         }
 761     }
 762     return 0;
 763 }
 764
 765 static ZEBRA_RES filter_config(void *clientData, Res res, const char *args)
 766 {
 767     struct filter_info *tinfo = clientData;
 768     if (!args || !*args)
 769     {
 770         yaz_log(YLOG_WARN, "dom filter: need config file");
 771         return ZEBRA_FAIL;
 772     }
 773
 774     if (tinfo->fname && !strcmp(args, tinfo->fname))
 775         return ZEBRA_OK;
 776
 777     tinfo->profile_path = res_get(res, "profilePath");
 778
 779     destroy_dom(tinfo);
 780     return parse_dom(tinfo, args);
 781 }
 782
 783 static void filter_destroy(void *clientData)
 784 {
 785     struct filter_info *tinfo = clientData;
 786     destroy_dom(tinfo);
 787     nmem_destroy(tinfo->nmem_config);
 788     nmem_destroy(tinfo->nmem_record);
 789     xfree(tinfo);
 790 }
 791
 792 static int ioread_ex(void *context, char *buffer, int len)
 793 {
 794     struct recExtractCtrl *p = context;
 795     return p->stream->readf(p->stream, buffer, len);
 796 }
 797
 798 static int ioclose_ex(void *context)
 799 {
 800     return 0;
 801 }
 802
 803
 804
 805 /* DOM filter style indexing */
 806 static void index_value_of(struct filter_info *tinfo,
 807                            struct recExtractCtrl *extctr,
 808                            RecWord* recword,
 809                            xmlNodePtr node,
 810                            const char *index_p)
 811 {
 812     if (tinfo->record_info_invoked == 1)
 813     {
 814         xmlChar *text = xmlNodeGetContent(node);
 815         size_t text_len = strlen((const char *)text);
 816
 817         /* if there is no text, we do not need to proceed */
 818         if (text_len)
 819         {
 820             const char *look = index_p;
 821             const char *bval;
 822             const char *eval;
 823
 824             xmlChar index[256];
 825             xmlChar type[256];
 826
 827             /* assingning text to be indexed */
 828             recword->term_buf = (const char *)text;
 829             recword->term_len = text_len;
 830
 831             /* parsing all index name/type pairs */
 832             /* may not start with ' ' or ':' */
 833             while (*look && ' ' != *look && ':' != *look)
 834             {
 835                 /* setting name and type to zero */
 836                 *index = '\0';
 837                 *type = '\0';
 838
 839                 /* parsing one index name */
 840                 bval = look;
 841                 while (*look && ':' != *look && ' ' != *look)
 842                 {
 843                     look++;
 844                 }
 845                 eval = look;
 846                 strncpy((char *)index, (const char *)bval, eval - bval);
 847                 index[eval - bval] = '\0';
 848
 849
 850                 /* parsing one index type, if existing */
 851                 if (':' == *look)
 852                 {
 853                     look++;
 854
 855                     bval = look;
 856                     while (*look && ' ' != *look)
 857                     {
 858                         look++;
 859                     }
 860                     eval = look;
 861                     strncpy((char *)type, (const char *)bval, eval - bval);
 862                     type[eval - bval] = '\0';
 863                 }
 864
 865                 /* actually indexing the text given */
 866
 867                 recword->index_name = (const char *)index;
 868                 if (*type)
 869                     recword->index_type = (const char *) type;
 870
 871                 /* writing debug out */
 872                 if (extctr->flagShowRecords)
 873                     dom_log(YLOG_LOG, tinfo, 0,
 874                             "INDEX '%s:%s' '%s'",
 875                             (const char *) index,
 876                             (const char *) type,
 877                             (const char *) text);
 878
 879                 (extctr->tokenAdd)(recword);
 880
 881                 /* eat whitespaces */
 882                 if (*look && ' ' == *look)
 883                 {
 884                     look++;
 885                 }
 886             }
 887         }
 888         xmlFree(text);
 889     }
 890 }
 891
 892
 893 /* DOM filter style indexing */
 894 static void set_record_info(struct filter_info *tinfo,
 895                             struct recExtractCtrl *extctr,
 896                             xmlNodePtr node,
 897                             const char * id_p,
 898                             const char * rank_p,
 899                             const char * type_p)
 900 {
 901     /* writing debug info out */
 902     if (extctr && extctr->flagShowRecords)
 903         dom_log(YLOG_LOG, tinfo, node,
 904                 "RECORD id=%s rank=%s type=%s",
 905                 id_p ? (const char *) id_p : "(null)",
 906                 rank_p ? (const char *) rank_p : "(null)",
 907                 type_p ? (const char *) type_p : "(null)");
 908
 909
 910     if (id_p && *id_p)
 911         sscanf((const char *)id_p, "%255s", extctr->match_criteria);
 912
 913     if (rank_p && *rank_p)
 914         extctr->staticrank = atozint((const char *)rank_p);
 915
 916     if (type_p && *type_p)
 917     {
 918         enum zebra_recctrl_action_t action = action_update;
 919         if (!strcmp(type_p, "insert"))
 920             action = action_insert;
 921         else if (!strcmp(type_p, "delete"))
 922             action = action_delete;
 923         else if (!strcmp(type_p, "replace"))
 924             action = action_replace;
 925         else if (!strcmp(type_p, "update"))
 926             action = action_update;
 927         else
 928             dom_log(YLOG_WARN, tinfo, node, "bad @type value: %s", type_p);
 929         extctr->action = action;
 930     }
 931
 932     if (tinfo->record_info_invoked == 1)
 933     {
 934         /* warn about multiple only once */
 935         dom_log(YLOG_WARN, tinfo, node, "multiple record elements");
 936     }
 937     tinfo->record_info_invoked++;
 938
 939 }
 940
 941
 942 /* DOM filter style indexing */
 943 static void process_xml_element_zebra_node(struct filter_info *tinfo,
 944                                            struct recExtractCtrl *extctr,
 945                                            RecWord* recword,
 946                                            xmlNodePtr node)
 947 {
 948     if (node->type == XML_ELEMENT_NODE && node->ns && node->ns->href
 949         && 0 == XML_STRCMP(node->ns->href, zebra_dom_ns))
 950     {
 951         if (0 == XML_STRCMP(node->name, "index"))
 952         {
 953             const char *index_p = 0;
 954
 955             struct _xmlAttr *attr;
 956             for (attr = node->properties; attr; attr = attr->next)
 957             {
 958                 if (attr_content(attr, "name", &index_p))
 959                 {
 960                     index_value_of(tinfo, extctr, recword, node, index_p);
 961                 }
 962                 else
 963                 {
 964                     dom_log(YLOG_WARN, tinfo, node,
 965                             "bad attribute @%s, expected @name",
 966                             attr->name);
 967                 }
 968             }
 969         }
 970         else if (0 == XML_STRCMP(node->name, "record"))
 971         {
 972             const char *id_p = 0;
 973             const char *rank_p = 0;
 974             const char *type_p = 0;
 975
 976             struct _xmlAttr *attr;
 977             for (attr = node->properties; attr; attr = attr->next)
 978             {
 979                 if (attr_content(attr, "id", &id_p))
 980                     ;
 981                 else if (attr_content(attr, "rank", &rank_p))
 982                     ;
 983                 else if (attr_content(attr, "type", &type_p))
 984                     ;
 985                 else
 986                 {
 987                     dom_log(YLOG_WARN, tinfo, node,
 988                             "bad attribute @%s, expected @id|@rank|@type",
 989                             attr->name);
 990                 }
 991             }
 992             set_record_info(tinfo, extctr, node, id_p, rank_p, type_p);
 993         }
 994         else
 995         {
 996             dom_log(YLOG_WARN, tinfo, node,
 997                     "bad element <%s>,"
 998                     " expected <record>|<index> in namespace '%s'",
 999                     node->name, zebra_dom_ns);
1000         }
1001     }
1002 }
1003
1004 static int attr_content_pi(const char **c_ptr, const char *name,
1005                            char *value, size_t value_max)
1006 {
1007     size_t name_len = strlen(name);
1008     const char *look = *c_ptr;
1009     int ret = 0;
1010
1011     *value = '\0';
1012     while (*look && ' ' == *look)
1013         look++;
1014     if (strlen(look) > name_len)
1015     {
1016         if (look[name_len] == '=' && !memcmp(look, name, name_len))
1017         {
1018             size_t i = 0;
1019             look += name_len+1;
1020             while (*look && ' ' != *look)
1021             {
1022                 if (i < value_max-1)
1023                     value[i++] = *look;
1024                 look++;
1025             }
1026             value[i] = '\0';
1027             ret = 1;
1028         }
1029     }
1030     while (*look && ' ' == *look)
1031         look++;
1032     *c_ptr = look;
1033     return ret;
1034 }
1035
1036 /* DOM filter style indexing */
1037 static void process_xml_pi_node(struct filter_info *tinfo,
1038                                 struct recExtractCtrl *extctr,
1039                                 xmlNodePtr node,
1040                                 const char **index_pp)
1041 {
1042     /* if right PI name, continue parsing PI */
1043     if (0 == strcmp(zebra_pi_name, (const char *)node->name))
1044     {
1045         xmlChar *pi_p =  node->content;
1046         const char *look = (const char *) node->content;
1047
1048         /* parsing PI record instructions */
1049         if (0 == strncmp((const char *)look, "record", 6))
1050         {
1051             char id[256];
1052             char rank[256];
1053             char type[256];
1054
1055             *id = '\0';
1056             *rank = '\0';
1057             *type = '\0';
1058             look += 6;
1059             while (*look)
1060                 if (attr_content_pi(&look, "id", id, sizeof(id)))
1061                     ;
1062                 else if (attr_content_pi(&look, "rank", rank, sizeof(rank)))
1063                     ;
1064                 else if (attr_content_pi(&look, "type", type, sizeof(type)))
1065                 {
1066                     dom_log(YLOG_WARN, tinfo, node,
1067                             "content '%s', can not parse '%s'",
1068                             pi_p, look);
1069                     break;
1070                 }
1071             set_record_info(tinfo, extctr, node, id, rank, type);
1072         }
1073         /* parsing index instruction */
1074         else if (0 == strncmp((const char *)look, "index", 5))
1075         {
1076             look += 5;
1077
1078             /* eat whitespace */
1079             while (*look && ' ' == *look)
1080                 look++;
1081
1082             /* export index instructions to outside */
1083             *index_pp = look;
1084         }
1085         else
1086         {
1087             dom_log(YLOG_WARN, tinfo, node,
1088                     "content '%s', can not parse '%s'",
1089                     pi_p, look);
1090         }
1091     }
1092 }
1093
1094 /* DOM filter style indexing */
1095 static void process_xml_element_node(struct filter_info *tinfo,
1096                                      struct recExtractCtrl *extctr,
1097                                      RecWord* recword,
1098                                      xmlNodePtr node)
1099 {
1100     /* remember indexing instruction from PI to next element node */
1101     const char *index_p = 0;
1102
1103     /* check if we are an element node in the special zebra namespace
1104        and either set record data or index value-of node content*/
1105     process_xml_element_zebra_node(tinfo, extctr, recword, node);
1106
1107     /* loop through kid nodes */
1108     for (node = node->children; node; node = node->next)
1109     {
1110         /* check and set PI record and index index instructions */
1111         if (node->type == XML_PI_NODE)
1112         {
1113             process_xml_pi_node(tinfo, extctr, node, &index_p);
1114         }
1115         else if (node->type == XML_ELEMENT_NODE)
1116         {
1117             /* if there was a PI index instruction before this element */
1118             if (index_p)
1119             {
1120                 index_value_of(tinfo, extctr, recword, node, index_p);
1121                 index_p = 0;
1122             }
1123             process_xml_element_node(tinfo, extctr, recword,node);
1124         }
1125         else
1126             continue;
1127     }
1128 }
1129
1130
1131 /* DOM filter style indexing */
1132 static void extract_dom_doc_node(struct filter_info *tinfo,
1133                                  struct recExtractCtrl *extctr,
1134                                  xmlDocPtr doc)
1135 {
1136     /* only need to do the initialization once, reuse recword for all terms */
1137     RecWord recword;
1138     (*extctr->init)(extctr, &recword);
1139
1140     process_xml_element_node(tinfo, extctr, &recword, (xmlNodePtr)doc);
1141 }
1142
1143
1144 static int convert_extract_doc(struct filter_info *tinfo,
1145                                struct filter_input *input,
1146                                struct recExtractCtrl *p,
1147                                xmlDocPtr doc)
1148 {
1149     xmlChar *buf_out;
1150     int len_out;
1151     const char *params[10];
1152     xsltStylesheetPtr last_xsp = 0;
1153
1154     /* per default do not ingest record */
1155     tinfo->record_info_invoked = 0;
1156
1157     /* exit if empty document given */
1158     if (!doc)
1159         return RECCTRL_EXTRACT_SKIP;
1160
1161     /* we actuallu have a document which needs to be processed further */
1162     params[0] = 0;
1163     set_param_str(params, "schema", zebra_dom_ns, tinfo->nmem_record);
1164
1165     if (p && p->flagShowRecords)
1166     {
1167         xmlChar *buf_out;
1168         int len_out;
1169 #if 0
1170         FILE *outf = fopen("extract.xml", "w");
1171         xmlDocDumpMemory(doc, &buf_out, &len_out);
1172         fwrite(buf_out, 1, len_out, outf);
1173 #endif
1174         yaz_log(YLOG_LOG, "Extract Doc: %.*s", len_out, buf_out);
1175 #if 0
1176         fclose(outf);
1177 #endif
1178     }
1179
1180     if (p->setStoreData)
1181     {
1182         xmlDocPtr store_doc = 0;
1183
1184         /* input conversion */
1185         perform_convert(tinfo, p, 0, input->convert, params, &doc, 0);
1186
1187         if (tinfo->store)
1188         {
1189             /* store conversion */
1190             store_doc = xmlCopyDoc(doc, 1);
1191             perform_convert(tinfo, p, 0, tinfo->store->convert,
1192                             params, &store_doc, &last_xsp);
1193         }
1194
1195         /* saving either store doc or original doc in case no store doc exists */
1196         if (last_xsp)
1197             xsltSaveResultToString(&buf_out, &len_out,
1198                                    store_doc ? store_doc : doc, last_xsp);
1199         else
1200             xmlDocDumpMemory(store_doc ? store_doc : doc, &buf_out, &len_out);
1201
1202         if (p->setStoreData)
1203             (*p->setStoreData)(p, buf_out, len_out);
1204         xmlFree(buf_out);
1205         if (store_doc)
1206             xmlFreeDoc(store_doc);
1207     }
1208
1209
1210     /* extract conversion */
1211     perform_convert(tinfo, p, 0, tinfo->extract->convert, params, &doc, 0);
1212
1213
1214     /* finally, do the indexing */
1215     if (doc){
1216         extract_dom_doc_node(tinfo, p, doc);
1217         xmlFreeDoc(doc);
1218     }
1219
1220     /* there was nothing to index, so there is no inserted/updated record */
1221     if (tinfo->record_info_invoked == 0)
1222         return RECCTRL_EXTRACT_SKIP;
1223
1224     return RECCTRL_EXTRACT_OK;
1225 }
1226
1227 static int extract_xml_split(struct filter_info *tinfo,
1228                              struct filter_input *input,
1229                              struct recExtractCtrl *p)
1230 {
1231     int ret;
1232
1233     if (p->first_record)
1234     {
1235         if (input->u.xmlreader.reader)
1236             xmlFreeTextReader(input->u.xmlreader.reader);
1237         input->u.xmlreader.reader = xmlReaderForIO(ioread_ex, ioclose_ex,
1238                                                    p /* I/O handler */,
1239                                                    0 /* URL */,
1240                                                    0 /* encoding */,
1241                                                    XML_PARSE_XINCLUDE
1242                                                    | XML_PARSE_NOENT
1243                                                    | XML_PARSE_NONET);
1244     }
1245     if (!input->u.xmlreader.reader)
1246         return RECCTRL_EXTRACT_ERROR_GENERIC;
1247
1248     ret = xmlTextReaderRead(input->u.xmlreader.reader);
1249     while (ret == 1)
1250     {
1251         int type = xmlTextReaderNodeType(input->u.xmlreader.reader);
1252         int depth = xmlTextReaderDepth(input->u.xmlreader.reader);
1253
1254         if (type == XML_READER_TYPE_ELEMENT &&
1255             input->u.xmlreader.split_level == depth)
1256         {
1257             xmlNodePtr ptr;
1258
1259             /* per default do not ingest record */
1260             tinfo->record_info_invoked = 0;
1261
1262             ptr = xmlTextReaderExpand(input->u.xmlreader.reader);
1263             if (ptr)
1264             {
1265                 /* we have a new document */
1266
1267                 xmlNodePtr ptr2 = xmlCopyNode(ptr, 1);
1268                 xmlDocPtr doc = xmlNewDoc((const xmlChar*) "1.0");
1269
1270                 xmlDocSetRootElement(doc, ptr2);
1271
1272                 /* writing debug info out */
1273                 if (p->flagShowRecords)
1274                 {
1275                     xmlChar *buf_out = 0;
1276                     int len_out = 0;
1277                     xmlDocDumpMemory(doc, &buf_out, &len_out);
1278                     yaz_log(YLOG_LOG, "%s: XMLREADER level: %i\n%.*s",
1279                             tinfo->fname ? tinfo->fname : "(none)",
1280                             depth, len_out, buf_out);
1281                     xmlFree(buf_out);
1282                 }
1283
1284                 return convert_extract_doc(tinfo, input, p, doc);
1285             }
1286             else
1287             {
1288                 xmlFreeTextReader(input->u.xmlreader.reader);
1289                 input->u.xmlreader.reader = 0;
1290                 return RECCTRL_EXTRACT_ERROR_GENERIC;
1291             }
1292         }
1293         ret = xmlTextReaderRead(input->u.xmlreader.reader);
1294     }
1295     xmlFreeTextReader(input->u.xmlreader.reader);
1296     input->u.xmlreader.reader = 0;
1297     return RECCTRL_EXTRACT_EOF;
1298 }
1299
1300 static int extract_xml_full(struct filter_info *tinfo,
1301                             struct filter_input *input,
1302                             struct recExtractCtrl *p)
1303 {
1304     if (p->first_record) /* only one record per stream */
1305     {
1306         xmlDocPtr doc = xmlReadIO(ioread_ex, ioclose_ex,
1307                                   p /* I/O handler */,
1308                                   0 /* URL */,
1309                                   0 /* encoding */,
1310                                   XML_PARSE_XINCLUDE
1311                                   | XML_PARSE_NOENT
1312                                   | XML_PARSE_NONET);
1313         if (!doc)
1314         {
1315             return RECCTRL_EXTRACT_ERROR_GENERIC;
1316         }
1317         return convert_extract_doc(tinfo, input, p, doc);
1318     }
1319     else
1320         return RECCTRL_EXTRACT_EOF;
1321 }
1322
1323 static int extract_iso2709(struct filter_info *tinfo,
1324                            struct filter_input *input,
1325                            struct recExtractCtrl *p)
1326 {
1327     char buf[100000];
1328     int record_length;
1329     int read_bytes, r;
1330
1331     if (p->stream->readf(p->stream, buf, 5) != 5)
1332         return RECCTRL_EXTRACT_EOF;
1333     while (*buf < '0' || *buf > '9')
1334     {
1335         int i;
1336
1337         dom_log(YLOG_WARN, tinfo, 0,
1338                 "MARC: Skipping bad byte %d (0x%02X)",
1339                 *buf & 0xff, *buf & 0xff);
1340         for (i = 0; i<4; i++)
1341             buf[i] = buf[i+1];
1342
1343         if (p->stream->readf(p->stream, buf+4, 1) != 1)
1344             return RECCTRL_EXTRACT_EOF;
1345     }
1346     record_length = atoi_n (buf, 5);
1347     if (record_length < 25)
1348     {
1349         dom_log(YLOG_WARN, tinfo, 0,
1350                 "MARC record length < 25, is %d",  record_length);
1351         return RECCTRL_EXTRACT_ERROR_GENERIC;
1352     }
1353     read_bytes = p->stream->readf(p->stream, buf+5, record_length-5);
1354     if (read_bytes < record_length-5)
1355     {
1356         dom_log(YLOG_WARN, tinfo, 0,
1357                 "couldn't read whole MARC record");
1358         return RECCTRL_EXTRACT_ERROR_GENERIC;
1359     }
1360     r = yaz_marc_read_iso2709(input->u.marc.handle,  buf, record_length);
1361     if (r < record_length)
1362     {
1363         dom_log (YLOG_WARN, tinfo, 0,
1364                  "parsing of MARC record failed r=%d length=%d",
1365                  r, record_length);
1366         return RECCTRL_EXTRACT_ERROR_GENERIC;
1367     }
1368     else
1369     {
1370         xmlDocPtr rdoc;
1371         xmlNode *root_ptr;
1372         yaz_marc_write_xml(input->u.marc.handle, &root_ptr,
1373                            "http://www.loc.gov/MARC21/slim", 0, 0);
1374         rdoc = xmlNewDoc((const xmlChar*) "1.0");
1375         xmlDocSetRootElement(rdoc, root_ptr);
1376         return convert_extract_doc(tinfo, input, p, rdoc);
1377     }
1378     return RECCTRL_EXTRACT_OK;
1379 }
1380
1381 static int filter_extract(void *clientData, struct recExtractCtrl *p)
1382 {
1383     struct filter_info *tinfo = clientData;
1384     struct filter_input *input = tinfo->input_list;
1385
1386     if (!input)
1387         return RECCTRL_EXTRACT_ERROR_GENERIC;
1388
1389     nmem_reset(tinfo->nmem_record);
1390
1391     if (p->setStoreData == 0)
1392         return extract_xml_full(tinfo, input, p);
1393     switch(input->type)
1394     {
1395     case DOM_INPUT_XMLREADER:
1396         if (input->u.xmlreader.split_level == 0)
1397             return extract_xml_full(tinfo, input, p);
1398         else
1399             return extract_xml_split(tinfo, input, p);
1400         break;
1401     case DOM_INPUT_MARC:
1402         return extract_iso2709(tinfo, input, p);
1403     }
1404     return RECCTRL_EXTRACT_ERROR_GENERIC;
1405 }
1406
1407 static int ioread_ret(void *context, char *buffer, int len)
1408 {
1409     struct recRetrieveCtrl *p = context;
1410     return p->stream->readf(p->stream, buffer, len);
1411 }
1412
1413 static int ioclose_ret(void *context)
1414 {
1415     return 0;
1416 }
1417
1418 static int filter_retrieve(void *clientData, struct recRetrieveCtrl *p)
1419 {
1420     /* const char *esn = zebra_dom_ns; */
1421     const char *esn = 0;
1422     const char *params[32];
1423     struct filter_info *tinfo = clientData;
1424     xmlDocPtr doc;
1425     struct filter_retrieve *retrieve;
1426     xsltStylesheetPtr last_xsp = 0;
1427
1428     if (p->comp)
1429     {
1430         if (p->comp->which == Z_RecordComp_simple
1431             && p->comp->u.simple->which == Z_ElementSetNames_generic)
1432         {
1433             esn = p->comp->u.simple->u.generic;
1434         }
1435         else if (p->comp->which == Z_RecordComp_complex
1436                  && p->comp->u.complex->generic->elementSpec
1437                  && p->comp->u.complex->generic->elementSpec->which ==
1438                  Z_ElementSpec_elementSetName)
1439         {
1440             esn = p->comp->u.complex->generic->elementSpec->u.elementSetName;
1441         }
1442     }
1443     retrieve = lookup_retrieve(tinfo, esn);
1444     if (!retrieve)
1445     {
1446         p->diagnostic =
1447             YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
1448         p->addinfo = odr_strdup(p->odr, esn);
1449         return 0;
1450     }
1451
1452     params[0] = 0;
1453     set_param_int(params, "id", p->localno, p->odr->mem);
1454     if (p->fname)
1455         set_param_str(params, "filename", p->fname, p->odr->mem);
1456     if (p->staticrank >= 0)
1457         set_param_int(params, "rank", p->staticrank, p->odr->mem);
1458
1459     if (esn)
1460         set_param_str(params, "schema", esn, p->odr->mem);
1461     else
1462         if (retrieve->name)
1463             set_param_str(params, "schema", retrieve->name, p->odr->mem);
1464         else if (retrieve->identifier)
1465             set_param_str(params, "schema", retrieve->identifier, p->odr->mem);
1466         else
1467             set_param_str(params, "schema", "", p->odr->mem);
1468
1469     if (p->score >= 0)
1470         set_param_int(params, "score", p->score, p->odr->mem);
1471     set_param_int(params, "size", p->recordSize, p->odr->mem);
1472
1473     doc = xmlReadIO(ioread_ret, ioclose_ret, p /* I/O handler */,
1474                     0 /* URL */,
1475                     0 /* encoding */,
1476                     XML_PARSE_XINCLUDE | XML_PARSE_NOENT | XML_PARSE_NONET);
1477     if (!doc)
1478     {
1479         p->diagnostic = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1480         return 0;
1481     }
1482
1483     /* retrieve conversion */
1484     perform_convert(tinfo, 0, p, retrieve->convert, params, &doc, &last_xsp);
1485     if (!doc)
1486     {
1487         p->diagnostic = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
1488     }
1489     else if (!p->input_format
1490              || !oid_oidcmp(p->input_format, yaz_oid_recsyn_xml))
1491     {
1492         xmlChar *buf_out;
1493         int len_out;
1494
1495         if (last_xsp)
1496             xsltSaveResultToString(&buf_out, &len_out, doc, last_xsp);
1497         else
1498             xmlDocDumpMemory(doc, &buf_out, &len_out);
1499
1500         p->output_format = yaz_oid_recsyn_xml;
1501         p->rec_len = len_out;
1502         p->rec_buf = odr_malloc(p->odr, p->rec_len);
1503         memcpy(p->rec_buf, buf_out, p->rec_len);
1504         xmlFree(buf_out);
1505     }
1506     else if (!oid_oidcmp(p->output_format, yaz_oid_recsyn_sutrs))
1507     {
1508         xmlChar *buf_out;
1509         int len_out;
1510
1511         if (last_xsp)
1512             xsltSaveResultToString(&buf_out, &len_out, doc, last_xsp);
1513         else
1514             xmlDocDumpMemory(doc, &buf_out, &len_out);
1515
1516         p->output_format = yaz_oid_recsyn_sutrs;
1517         p->rec_len = len_out;
1518         p->rec_buf = odr_malloc(p->odr, p->rec_len);
1519         memcpy(p->rec_buf, buf_out, p->rec_len);
1520
1521         xmlFree(buf_out);
1522     }
1523     else
1524     {
1525         p->diagnostic = YAZ_BIB1_RECORD_SYNTAX_UNSUPP;
1526     }
1527     xmlFreeDoc(doc);
1528     return 0;
1529 }
1530
1531 static struct recType filter_type = {
1532     0,
1533     "dom",
1534     filter_init,
1535     filter_config,
1536     filter_destroy,
1537     filter_extract,
1538     filter_retrieve
1539 };
1540
1541 RecType
1542 #ifdef IDZEBRA_STATIC_DOM
1543 idzebra_filter_dom
1544 #else
1545 idzebra_filter
1546 #endif
1547
1548 [] = {
1549     &filter_type,
1550     0,
1551 };
1552 /*
1553  * Local variables:
1554  * c-basic-offset: 4
1555  * indent-tabs-mode: nil
1556  * End:
1557  * vim: shiftwidth=4 tabstop=8 expandtab
1558  */
1559