recctrl/xslt.c

   1 /* $Id: xslt.c,v 1.26 2006-05-29 13:48:43 marc Exp $
   2    Copyright (C) 1995-2005
   3    Index Data ApS
   4
   5 This file is part of the Zebra server.
   6
   7 Zebra is free software; you can redistribute it and/or modify it under
   8 the terms of the GNU General Public License as published by the Free
   9 Software Foundation; either version 2, or (at your option) any later
  10 version.
  11
  12 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
  13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  14 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  15 for more details.
  16
  17 You should have received a copy of the GNU General Public License
  18 along with Zebra; see the file LICENSE.zebra.  If not, write to the
  19 Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
  20 02111-1307, USA.
  21 */
  22
  23 #include <stdio.h>
  24 #include <assert.h>
  25 #include <ctype.h>
  26
  27 #include <yaz/diagbib1.h>
  28 #include <yaz/tpath.h>
  29
  30 #include <libxml/xmlversion.h>
  31 #include <libxml/parser.h>
  32 #include <libxml/tree.h>
  33 #include <libxml/xmlIO.h>
  34 #include <libxml/xmlreader.h>
  35 #include <libxslt/transform.h>
  36 /* #include <libxslt/xsltutils.h> */
  37
  38 #include <idzebra/util.h>
  39 #include <idzebra/recctrl.h>
  40
  41 struct filter_xslt_schema {
  42     const char *name;
  43     const char *identifier;
  44     const char *stylesheet;
  45     struct filter_xslt_schema *next;
  46     const char *default_schema;
  47     const char *include_snippet;
  48     xsltStylesheetPtr stylesheet_xsp;
  49 };
  50
  51 struct filter_xslt_info {
  52     xmlDocPtr doc;
  53     char *fname;
  54     char *full_name;
  55     const char *profile_path;
  56     const char *split_level;
  57     const char *split_path;
  58     ODR odr;
  59     struct filter_xslt_schema *schemas;
  60     xmlTextReaderPtr reader;
  61 };
  62
  63
  64 #define ZEBRA_SCHEMA_XSLT_NS "http://indexdata.dk/zebra/xslt/1"
  65
  66 #define XML_STRCMP(a,b)   strcmp((char*)a, b)
  67 #define XML_STRLEN(a) strlen((char*)a)
  68
  69 static const char *zebra_xslt_ns = ZEBRA_SCHEMA_XSLT_NS;
  70
  71 static void set_param_xml(const char **params, const char *name,
  72                           const char *value, ODR odr)
  73 {
  74     while (*params)
  75         params++;
  76     params[0] = name;
  77     params[1] = value;
  78     params[2] = 0;
  79 }
  80
  81 static void set_param_str(const char **params, const char *name,
  82                           const char *value, ODR odr)
  83 {
  84     char *quoted = odr_malloc(odr, 3 + strlen(value));
  85     sprintf(quoted, "'%s'", value);
  86     while (*params)
  87         params++;
  88     params[0] = name;
  89     params[1] = quoted;
  90     params[2] = 0;
  91 }
  92
  93 static void set_param_int(const char **params, const char *name,
  94                           zint value, ODR odr)
  95 {
  96     char *quoted = odr_malloc(odr, 30); /* 25 digits enough for 2^64 */
  97     while (*params)
  98         params++;
  99     sprintf(quoted, "'" ZINT_FORMAT "'", value);
 100     params[0] = name;
 101     params[1] = quoted;
 102     params[2] = 0;
 103 }
 104
 105 #define ENABLE_INPUT_CALLBACK 0
 106
 107 #if ENABLE_INPUT_CALLBACK
 108 static int zebra_xmlInputMatchCallback (char const *filename)
 109 {
 110     yaz_log(YLOG_LOG, "match %s", filename);
 111     return 0;
 112 }
 113
 114 static void * zebra_xmlInputOpenCallback (char const *filename)
 115 {
 116     return 0;
 117 }
 118
 119 static int zebra_xmlInputReadCallback (void * context, char * buffer, int len)
 120 {
 121     return 0;
 122 }
 123
 124 static int zebra_xmlInputCloseCallback (void * context)
 125 {
 126     return 0;
 127 }
 128 #endif
 129
 130 static void *filter_init(Res res, RecType recType)
 131 {
 132     struct filter_xslt_info *tinfo
 133       = (struct filter_xslt_info *) xmalloc(sizeof(*tinfo));
 134     tinfo->reader = 0;
 135     tinfo->fname = 0;
 136     tinfo->full_name = 0;
 137     tinfo->profile_path = 0;
 138     tinfo->split_level = 0;
 139     tinfo->split_path = 0;
 140     tinfo->odr = odr_createmem(ODR_ENCODE);
 141     tinfo->doc = 0;
 142     tinfo->schemas = 0;
 143
 144 #if ENABLE_INPUT_CALLBACK
 145     xmlRegisterDefaultInputCallbacks();
 146     xmlRegisterInputCallbacks(zebra_xmlInputMatchCallback,
 147                               zebra_xmlInputOpenCallback,
 148                               zebra_xmlInputReadCallback,
 149                               zebra_xmlInputCloseCallback);
 150 #endif
 151     return tinfo;
 152 }
 153
 154 static int attr_content(struct _xmlAttr *attr, const char *name,
 155                         const char **dst_content)
 156 {
 157     if (!XML_STRCMP(attr->name, name) && attr->children &&
 158         attr->children->type == XML_TEXT_NODE)
 159     {
 160         *dst_content = (const char *)(attr->children->content);
 161         return 1;
 162     }
 163     return 0;
 164 }
 165
 166 static void destroy_schemas(struct filter_xslt_info *tinfo)
 167 {
 168     struct filter_xslt_schema *schema = tinfo->schemas;
 169     while (schema)
 170     {
 171         struct filter_xslt_schema *schema_next = schema->next;
 172         if (schema->stylesheet_xsp)
 173             xsltFreeStylesheet(schema->stylesheet_xsp);
 174         xfree(schema);
 175         schema = schema_next;
 176     }
 177     tinfo->schemas = 0;
 178     xfree(tinfo->fname);
 179     if (tinfo->doc)
 180         xmlFreeDoc(tinfo->doc);
 181     tinfo->doc = 0;
 182 }
 183
 184 static ZEBRA_RES create_schemas(struct filter_xslt_info *tinfo,
 185                                 const char *fname)
 186 {
 187     char tmp_full_name[1024];
 188     xmlNodePtr ptr;
 189     tinfo->fname = xstrdup(fname);
 190
 191     if (yaz_filepath_resolve(tinfo->fname, tinfo->profile_path,
 192                              NULL, tmp_full_name))
 193       tinfo->full_name = xstrdup(tmp_full_name);
 194     else
 195       tinfo->full_name = xstrdup(tinfo->fname);
 196
 197     yaz_log(YLOG_LOG, "xslt filter: loading config file %s", tinfo->full_name);
 198
 199     tinfo->doc = xmlParseFile(tinfo->full_name);
 200     if (!tinfo->doc) {
 201       yaz_log(YLOG_WARN, "xslt filter: could not parse config file %s",
 202               tinfo->full_name);
 203       return ZEBRA_FAIL;
 204     }
 205
 206     ptr = xmlDocGetRootElement(tinfo->doc);
 207     if (!ptr || ptr->type != XML_ELEMENT_NODE ||
 208         XML_STRCMP(ptr->name, "schemaInfo")){
 209           yaz_log(YLOG_WARN,
 210                   "xslt filter:  config file %s :"
 211                   " expected root element <schemaInfo>",
 212               tinfo->full_name);
 213       return ZEBRA_FAIL;
 214     }
 215
 216     for (ptr = ptr->children; ptr; ptr = ptr->next)
 217     {
 218         if (ptr->type != XML_ELEMENT_NODE)
 219             continue;
 220         if (!XML_STRCMP(ptr->name, "schema"))
 221         {
 222             char tmp_xslt_full_name[1024];
 223             struct _xmlAttr *attr;
 224             struct filter_xslt_schema *schema = xmalloc(sizeof(*schema));
 225             schema->name = 0;
 226             schema->identifier = 0;
 227             schema->stylesheet = 0;
 228             schema->default_schema = 0;
 229             schema->next = tinfo->schemas;
 230             schema->stylesheet_xsp = 0;
 231             schema->include_snippet = 0;
 232             tinfo->schemas = schema;
 233             for (attr = ptr->properties; attr; attr = attr->next)
 234             {
 235                 attr_content(attr, "identifier", &schema->identifier);
 236                 attr_content(attr, "name", &schema->name);
 237                 attr_content(attr, "stylesheet", &schema->stylesheet);
 238                 attr_content(attr, "default", &schema->default_schema);
 239                 attr_content(attr, "snippet", &schema->include_snippet);
 240             }
 241             if (schema->stylesheet){
 242               yaz_filepath_resolve(schema->stylesheet, tinfo->profile_path,
 243                                    NULL, tmp_xslt_full_name);
 244               schema->stylesheet_xsp
 245                 = xsltParseStylesheetFile((const xmlChar*) tmp_xslt_full_name);
 246               if (!schema->stylesheet_xsp)
 247                 yaz_log(YLOG_WARN,
 248                         "xslt filter: could not parse xslt stylesheet %s",
 249                         tmp_xslt_full_name);
 250             }
 251
 252         }
 253         else if (!XML_STRCMP(ptr->name, "split"))
 254         {
 255             struct _xmlAttr *attr;
 256             for (attr = ptr->properties; attr; attr = attr->next)
 257             {
 258                 attr_content(attr, "level", &tinfo->split_level);
 259                 attr_content(attr, "path", &tinfo->split_path);
 260             }
 261         }
 262         else
 263         {
 264             yaz_log(YLOG_WARN, "Bad element %s in %s", ptr->name, fname);
 265             return ZEBRA_FAIL;
 266         }
 267     }
 268     return ZEBRA_OK;
 269 }
 270
 271 static struct filter_xslt_schema *lookup_schema(struct filter_xslt_info *tinfo,
 272                                            const char *est)
 273 {
 274     struct filter_xslt_schema *schema;
 275     for (schema = tinfo->schemas; schema; schema = schema->next)
 276     {
 277         /* find requested schema */
 278         if (est)
 279         {
 280             if (schema->identifier && !strcmp(schema->identifier, est))
 281                 return schema;
 282
 283             if (schema->name && !strcmp(schema->name, est))
 284                 return schema;
 285         }
 286         /* or return default schema if defined */
 287         else if (schema->default_schema)
 288             return schema;
 289     }
 290
 291     /* return first schema if no default schema defined */
 292     if (tinfo->schemas)
 293         return tinfo->schemas;
 294
 295     return 0;
 296 }
 297
 298 static ZEBRA_RES filter_config(void *clientData, Res res, const char *args)
 299 {
 300     struct filter_xslt_info *tinfo = clientData;
 301     if (!args || !*args){
 302       yaz_log(YLOG_WARN, "xslt filter: need config file");
 303       return ZEBRA_FAIL;
 304     }
 305
 306     if (tinfo->fname && !strcmp(args, tinfo->fname))
 307         return ZEBRA_OK;
 308
 309     tinfo->profile_path
 310       /* = res_get_def(res, "profilePath", DEFAULT_PROFILE_PATH); */
 311       = res_get(res, "profilePath");
 312     yaz_log(YLOG_LOG, "xslt filter: profilePath %s", tinfo->profile_path);
 313
 314     destroy_schemas(tinfo);
 315     create_schemas(tinfo, args);
 316     return ZEBRA_OK;
 317 }
 318
 319 static void filter_destroy(void *clientData)
 320 {
 321     struct filter_xslt_info *tinfo = clientData;
 322     destroy_schemas(tinfo);
 323     if (tinfo->reader)
 324         xmlFreeTextReader(tinfo->reader);
 325     odr_destroy(tinfo->odr);
 326     xfree(tinfo);
 327 }
 328
 329 static int ioread_ex(void *context, char *buffer, int len)
 330 {
 331     struct recExtractCtrl *p = context;
 332     return (*p->readf)(p->fh, buffer, len);
 333 }
 334
 335 static int ioclose_ex(void *context)
 336 {
 337     return 0;
 338 }
 339
 340 static void index_cdata(struct filter_xslt_info *tinfo, struct recExtractCtrl *ctrl,
 341                         xmlNodePtr ptr, RecWord *recWord)
 342 {
 343     for(; ptr; ptr = ptr->next)
 344     {
 345         index_cdata(tinfo, ctrl, ptr->children, recWord);
 346         if (ptr->type != XML_TEXT_NODE)
 347             continue;
 348         recWord->term_buf = (const char *)ptr->content;
 349         recWord->term_len = XML_STRLEN(ptr->content);
 350         (*ctrl->tokenAdd)(recWord);
 351     }
 352 }
 353
 354 static void index_node(struct filter_xslt_info *tinfo,  struct recExtractCtrl *ctrl,
 355                        xmlNodePtr ptr, RecWord *recWord)
 356 {
 357     for(; ptr; ptr = ptr->next)
 358     {
 359         index_node(tinfo, ctrl, ptr->children, recWord);
 360         if (ptr->type != XML_ELEMENT_NODE || !ptr->ns ||
 361             XML_STRCMP(ptr->ns->href, zebra_xslt_ns))
 362             continue;
 363         if (!XML_STRCMP(ptr->name, "index"))
 364         {
 365             const char *name_str = 0;
 366             const char *type_str = 0;
 367             const char *xpath_str = 0;
 368             struct _xmlAttr *attr;
 369             for (attr = ptr->properties; attr; attr = attr->next)
 370             {
 371                 attr_content(attr, "name", &name_str);
 372                 attr_content(attr, "xpath", &xpath_str);
 373                 attr_content(attr, "type", &type_str);
 374             }
 375             if (name_str)
 376             {
 377                 int prev_type = recWord->index_type; /* save default type */
 378
 379                 if (type_str && *type_str)
 380                     recWord->index_type = *type_str; /* type was given */
 381                 recWord->index_name = name_str;
 382                 index_cdata(tinfo, ctrl, ptr->children, recWord);
 383
 384                 recWord->index_type = prev_type;     /* restore it again */
 385             }
 386         }
 387     }
 388 }
 389
 390 static void index_record(struct filter_xslt_info *tinfo,struct recExtractCtrl *ctrl,
 391                          xmlNodePtr ptr, RecWord *recWord)
 392 {
 393     if (ptr && ptr->type == XML_ELEMENT_NODE && ptr->ns &&
 394         !XML_STRCMP(ptr->ns->href, zebra_xslt_ns)
 395         && !XML_STRCMP(ptr->name, "record"))
 396     {
 397         const char *type_str = "update";
 398         const char *id_str = 0;
 399         const char *rank_str = 0;
 400         struct _xmlAttr *attr;
 401         for (attr = ptr->properties; attr; attr = attr->next)
 402         {
 403             attr_content(attr, "type", &type_str);
 404             attr_content(attr, "id", &id_str);
 405             attr_content(attr, "rank", &rank_str);
 406         }
 407         if (id_str)
 408             sscanf(id_str, "%255s", ctrl->match_criteria);
 409         if (rank_str)
 410         {
 411             ctrl->staticrank = atoi(rank_str);
 412             yaz_log(YLOG_LOG, "rank=%d",ctrl->staticrank);
 413         }
 414         else
 415             yaz_log(YLOG_LOG, "no rank");
 416
 417         ptr = ptr->children;
 418     }
 419     index_node(tinfo, ctrl, ptr, recWord);
 420 }
 421
 422 static int extract_doc(struct filter_xslt_info *tinfo, struct recExtractCtrl *p,
 423                        xmlDocPtr doc)
 424 {
 425     RecWord recWord;
 426     const char *params[10];
 427     xmlChar *buf_out;
 428     int len_out;
 429
 430     struct filter_xslt_schema *schema = lookup_schema(tinfo, zebra_xslt_ns);
 431
 432     params[0] = 0;
 433     set_param_str(params, "schema", zebra_xslt_ns, tinfo->odr);
 434
 435     (*p->init)(p, &recWord);
 436
 437     if (schema && schema->stylesheet_xsp)
 438     {
 439         xmlNodePtr root_ptr;
 440         xmlDocPtr resDoc =
 441             xsltApplyStylesheet(schema->stylesheet_xsp,
 442                                 doc, params);
 443         if (p->flagShowRecords)
 444         {
 445             xmlDocDumpMemory(resDoc, &buf_out, &len_out);
 446             fwrite(buf_out, len_out, 1, stdout);
 447             xmlFree(buf_out);
 448         }
 449         root_ptr = xmlDocGetRootElement(resDoc);
 450         if (root_ptr)
 451             index_record(tinfo, p, root_ptr, &recWord);
 452         else
 453         {
 454             yaz_log(YLOG_WARN, "No root for index XML record."
 455                     " split_level=%s stylesheet=%s",
 456                     tinfo->split_level, schema->stylesheet);
 457         }
 458         xmlFreeDoc(resDoc);
 459     }
 460     xmlDocDumpMemory(doc, &buf_out, &len_out);
 461     if (p->flagShowRecords)
 462         fwrite(buf_out, len_out, 1, stdout);
 463     (*p->setStoreData)(p, buf_out, len_out);
 464     xmlFree(buf_out);
 465
 466     xmlFreeDoc(doc);
 467     return RECCTRL_EXTRACT_OK;
 468 }
 469
 470 static int extract_split(struct filter_xslt_info *tinfo, struct recExtractCtrl *p)
 471 {
 472     int ret;
 473     int split_depth = 0;
 474     if (p->first_record)
 475     {
 476         if (tinfo->reader)
 477             xmlFreeTextReader(tinfo->reader);
 478         tinfo->reader = xmlReaderForIO(ioread_ex, ioclose_ex,
 479                                        p /* I/O handler */,
 480                                        0 /* URL */,
 481                                        0 /* encoding */,
 482                                        XML_PARSE_XINCLUDE);
 483     }
 484     if (!tinfo->reader)
 485         return RECCTRL_EXTRACT_ERROR_GENERIC;
 486
 487     if (tinfo->split_level)
 488         split_depth = atoi(tinfo->split_level);
 489     ret = xmlTextReaderRead(tinfo->reader);
 490     while (ret == 1) {
 491         int type = xmlTextReaderNodeType(tinfo->reader);
 492         int depth = xmlTextReaderDepth(tinfo->reader);
 493         if (split_depth == 0 ||
 494             (split_depth > 0 &&
 495              type == XML_READER_TYPE_ELEMENT && split_depth == depth))
 496         {
 497             xmlNodePtr ptr = xmlTextReaderExpand(tinfo->reader);
 498             xmlNodePtr ptr2 = xmlCopyNode(ptr, 1);
 499             xmlDocPtr doc = xmlNewDoc((const xmlChar*) "1.0");
 500
 501             xmlDocSetRootElement(doc, ptr2);
 502
 503             return extract_doc(tinfo, p, doc);
 504         }
 505         ret = xmlTextReaderRead(tinfo->reader);
 506     }
 507     xmlFreeTextReader(tinfo->reader);
 508     tinfo->reader = 0;
 509     return RECCTRL_EXTRACT_EOF;
 510 }
 511
 512 static int extract_full(struct filter_xslt_info *tinfo, struct recExtractCtrl *p)
 513 {
 514     if (p->first_record) /* only one record per stream */
 515     {
 516         xmlDocPtr doc = xmlReadIO(ioread_ex, ioclose_ex, p /* I/O handler */,
 517                                   0 /* URL */,
 518                                   0 /* encoding */,
 519                                   XML_PARSE_XINCLUDE);
 520         if (!doc)
 521         {
 522             return RECCTRL_EXTRACT_ERROR_GENERIC;
 523         }
 524         return extract_doc(tinfo, p, doc);
 525     }
 526     else
 527         return RECCTRL_EXTRACT_EOF;
 528 }
 529
 530 static int filter_extract(void *clientData, struct recExtractCtrl *p)
 531 {
 532     struct filter_xslt_info *tinfo = clientData;
 533
 534     odr_reset(tinfo->odr);
 535
 536     if (tinfo->split_level == 0 && tinfo->split_path == 0)
 537         return extract_full(tinfo, p);
 538     else
 539     {
 540         return extract_split(tinfo, p);
 541     }
 542 }
 543
 544 static int ioread_ret(void *context, char *buffer, int len)
 545 {
 546     struct recRetrieveCtrl *p = context;
 547     return (*p->readf)(p->fh, buffer, len);
 548 }
 549
 550 static int ioclose_ret(void *context)
 551 {
 552     return 0;
 553 }
 554
 555
 556 static const char *snippet_doc(struct recRetrieveCtrl *p, int text_mode,
 557                                int window_size)
 558 {
 559     const char *xml_doc_str;
 560     int ord = 0;
 561     WRBUF wrbuf = wrbuf_alloc();
 562     zebra_snippets *res =
 563         zebra_snippets_window(p->doc_snippet, p->hit_snippet, window_size);
 564     zebra_snippet_word *w = zebra_snippets_list(res);
 565
 566     if (text_mode)
 567         wrbuf_printf(wrbuf, "\'");
 568     else
 569         wrbuf_printf(wrbuf, "<snippet xmlns='%s'>\n", zebra_xslt_ns);
 570     for (; w; w = w->next)
 571     {
 572         if (ord == 0)
 573             ord = w->ord;
 574         else if (ord != w->ord)
 575
 576             break;
 577         if (text_mode)
 578             wrbuf_printf(wrbuf, "%s%s%s ",
 579                          w->match ? "*" : "",
 580                          w->term,
 581                          w->match ? "*" : "");
 582         else
 583         {
 584             wrbuf_printf(wrbuf, " <term ord='%d' seqno='" ZINT_FORMAT "' %s>",
 585                          w->ord, w->seqno,
 586                          (w->match ? "match='1'" : ""));
 587             wrbuf_xmlputs(wrbuf, w->term);
 588             wrbuf_printf(wrbuf, "</term>\n");
 589         }
 590     }
 591     if (text_mode)
 592         wrbuf_printf(wrbuf, "\'");
 593     else
 594         wrbuf_printf(wrbuf, "</snippet>\n");
 595
 596     xml_doc_str = odr_strdup(p->odr, wrbuf_buf(wrbuf));
 597
 598     zebra_snippets_destroy(res);
 599     wrbuf_free(wrbuf, 1);
 600     return xml_doc_str;
 601 }
 602
 603 static int filter_retrieve (void *clientData, struct recRetrieveCtrl *p)
 604 {
 605     const char *esn = 0;
 606     const char *params[32];
 607     struct filter_xslt_info *tinfo = clientData;
 608     xmlDocPtr resDoc;
 609     xmlDocPtr doc;
 610     struct filter_xslt_schema *schema;
 611     int window_size = -1;
 612
 613     if (p->comp)
 614     {
 615         if (p->comp->which == Z_RecordComp_simple
 616             && p->comp->u.simple->which == Z_ElementSetNames_generic)
 617         {
 618             esn = p->comp->u.simple->u.generic;
 619         }
 620         else if (p->comp->which == Z_RecordComp_complex
 621                  && p->comp->u.complex->generic->elementSpec
 622                  && p->comp->u.complex->generic->elementSpec->which ==
 623                  Z_ElementSpec_elementSetName)
 624         {
 625             esn = p->comp->u.complex->generic->elementSpec->u.elementSetName;
 626         }
 627     }
 628     schema = lookup_schema(tinfo, esn);
 629     if (!schema)
 630     {
 631         p->diagnostic =
 632             YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 633         return 0;
 634     }
 635
 636     if (schema->include_snippet)
 637         window_size = atoi(schema->include_snippet);
 638
 639     params[0] = 0;
 640     set_param_int(params, "id", p->localno, p->odr);
 641     if (p->fname)
 642         set_param_str(params, "filename", p->fname, p->odr);
 643     if (p->staticrank >= 0)
 644         set_param_int(params, "rank", p->staticrank, p->odr);
 645
 646      if (esn)
 647         set_param_str(params, "schema", esn, p->odr);
 648     else
 649         if (schema->name)
 650             set_param_str(params, "schema", schema->name, p->odr);
 651         else if (schema->identifier)
 652             set_param_str(params, "schema", schema->identifier, p->odr);
 653         else
 654             set_param_str(params, "schema", "", p->odr);
 655
 656     if (p->score >= 0)
 657         set_param_int(params, "score", p->score, p->odr);
 658     set_param_int(params, "size", p->recordSize, p->odr);
 659
 660     if (window_size >= 0)
 661         set_param_xml(params, "snippet", snippet_doc(p, 1, window_size),
 662                       p->odr);
 663     doc = xmlReadIO(ioread_ret, ioclose_ret, p /* I/O handler */,
 664                     0 /* URL */,
 665                     0 /* encoding */,
 666                     XML_PARSE_XINCLUDE);
 667     if (!doc)
 668     {
 669         p->diagnostic = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 670         return 0;
 671     }
 672
 673     if (window_size >= 0)
 674     {
 675         xmlNodePtr node = xmlDocGetRootElement(doc);
 676         const char *snippet_str = snippet_doc(p, 0, window_size);
 677         xmlDocPtr snippet_doc = xmlParseMemory(snippet_str, strlen(snippet_str));
 678         xmlAddChild(node, xmlDocGetRootElement(snippet_doc));
 679     }
 680     if (!schema->stylesheet_xsp)
 681         resDoc = doc;
 682     else
 683     {
 684         resDoc = xsltApplyStylesheet(schema->stylesheet_xsp,
 685                                      doc, params);
 686         xmlFreeDoc(doc);
 687     }
 688     if (!resDoc)
 689     {
 690         p->diagnostic = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 691     }
 692     else if (p->input_format == VAL_NONE || p->input_format == VAL_TEXT_XML)
 693     {
 694         xmlChar *buf_out;
 695         int len_out;
 696
 697         xsltSaveResultToString(&buf_out, &len_out, resDoc,
 698                                schema->stylesheet_xsp);
 699
 700         p->output_format = VAL_TEXT_XML;
 701         p->rec_len = len_out;
 702         p->rec_buf = odr_malloc(p->odr, p->rec_len);
 703         memcpy(p->rec_buf, buf_out, p->rec_len);
 704
 705         xmlFree(buf_out);
 706     }
 707     else if (p->output_format == VAL_SUTRS)
 708     {
 709         xmlChar *buf_out;
 710         int len_out;
 711
 712         xsltSaveResultToString(&buf_out, &len_out, resDoc,
 713                                schema->stylesheet_xsp);
 714
 715         p->output_format = VAL_SUTRS;
 716         p->rec_len = len_out;
 717         p->rec_buf = odr_malloc(p->odr, p->rec_len);
 718         memcpy(p->rec_buf, buf_out, p->rec_len);
 719
 720         xmlFree(buf_out);
 721     }
 722     else
 723     {
 724         p->diagnostic = YAZ_BIB1_RECORD_SYNTAX_UNSUPP;
 725     }
 726     xmlFreeDoc(resDoc);
 727     return 0;
 728 }
 729
 730 static struct recType filter_type = {
 731     0,
 732     "xslt",
 733     filter_init,
 734     filter_config,
 735     filter_destroy,
 736     filter_extract,
 737     filter_retrieve
 738 };
 739
 740 RecType
 741 #ifdef IDZEBRA_STATIC_XSLT
 742 idzebra_filter_xslt
 743 #else
 744 idzebra_filter
 745 #endif
 746
 747 [] = {
 748     &filter_type,
 749     0,
 750 };
 751 /*
 752  * Local variables:
 753  * c-basic-offset: 4
 754  * indent-tabs-mode: nil
 755  * End:
 756  * vim: shiftwidth=4 tabstop=8 expandtab
 757  */
 758