index/alvis.c

   1 /* $Id: alvis.c,v 1.12 2007-02-12 10:33:51 adam Exp $
   2    Copyright (C) 1995-2007
   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 this program; if not, write to the Free Software
  19 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  20
  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 #if YAZ_HAVE_EXSLT
  39 #include <libexslt/exslt.h>
  40 #endif
  41
  42 #include <idzebra/util.h>
  43 #include <idzebra/recctrl.h>
  44
  45 struct filter_schema {
  46     const char *name;
  47     const char *identifier;
  48     const char *stylesheet;
  49     struct filter_schema *next;
  50     const char *default_schema;
  51     /* char default_schema; */
  52     const char *include_snippet;
  53     xsltStylesheetPtr stylesheet_xsp;
  54 };
  55
  56 struct filter_info {
  57     xmlDocPtr doc;
  58     char *fname;
  59     char *full_name;
  60     const char *profile_path;
  61     int split_level;
  62     const char *split_path;
  63     ODR odr;
  64     struct filter_schema *schemas;
  65     xmlTextReaderPtr reader;
  66 };
  67
  68 #define ZEBRA_SCHEMA_XSLT_NS "http://indexdata.dk/zebra/xslt/1"
  69
  70 #define XML_STRCMP(a,b)   strcmp((char*)a, b)
  71 #define XML_STRLEN(a) strlen((char*)a)
  72
  73 static const char *zebra_xslt_ns = ZEBRA_SCHEMA_XSLT_NS;
  74
  75 static void set_param_xml(const char **params, const char *name,
  76                           const char *value, ODR odr)
  77 {
  78     while (*params)
  79         params++;
  80     params[0] = name;
  81     params[1] = value;
  82     params[2] = 0;
  83 }
  84
  85 static void set_param_str(const char **params, const char *name,
  86                           const char *value, ODR odr)
  87 {
  88     char *quoted = odr_malloc(odr, 3 + strlen(value));
  89     sprintf(quoted, "'%s'", value);
  90     while (*params)
  91         params++;
  92     params[0] = name;
  93     params[1] = quoted;
  94     params[2] = 0;
  95 }
  96
  97 static void set_param_int(const char **params, const char *name,
  98                           zint value, ODR odr)
  99 {
 100     char *quoted = odr_malloc(odr, 30); /* 25 digits enough for 2^64 */
 101     while (*params)
 102         params++;
 103     sprintf(quoted, "'" ZINT_FORMAT "'", value);
 104     params[0] = name;
 105     params[1] = quoted;
 106     params[2] = 0;
 107 }
 108
 109 #define ENABLE_INPUT_CALLBACK 0
 110
 111 #if ENABLE_INPUT_CALLBACK
 112 static int zebra_xmlInputMatchCallback (char const *filename)
 113 {
 114     yaz_log(YLOG_LOG, "match %s", filename);
 115     return 0;
 116 }
 117
 118 static void * zebra_xmlInputOpenCallback (char const *filename)
 119 {
 120     return 0;
 121 }
 122
 123 static int zebra_xmlInputReadCallback (void * context, char * buffer, int len)
 124 {
 125     return 0;
 126 }
 127
 128 static int zebra_xmlInputCloseCallback (void * context)
 129 {
 130     return 0;
 131 }
 132 #endif
 133
 134 static void *filter_init(Res res, RecType recType)
 135 {
 136     struct filter_info *tinfo = (struct filter_info *) xmalloc(sizeof(*tinfo));
 137     tinfo->reader = 0;
 138     tinfo->fname = 0;
 139     tinfo->full_name = 0;
 140     tinfo->profile_path = 0;
 141     tinfo->split_level = 0;
 142     tinfo->split_path = 0;
 143     tinfo->odr = odr_createmem(ODR_ENCODE);
 144     tinfo->doc = 0;
 145     tinfo->schemas = 0;
 146
 147 #if YAZ_HAVE_EXSLT
 148     exsltRegisterAll();
 149 #endif
 150
 151 #if ENABLE_INPUT_CALLBACK
 152     xmlRegisterDefaultInputCallbacks();
 153     xmlRegisterInputCallbacks(zebra_xmlInputMatchCallback,
 154                               zebra_xmlInputOpenCallback,
 155                               zebra_xmlInputReadCallback,
 156                               zebra_xmlInputCloseCallback);
 157 #endif
 158     return tinfo;
 159 }
 160
 161 static int attr_content(struct _xmlAttr *attr, const char *name,
 162                         const char **dst_content)
 163 {
 164     if (!XML_STRCMP(attr->name, name) && attr->children
 165         && attr->children->type == XML_TEXT_NODE)
 166     {
 167         *dst_content = (const char *)(attr->children->content);
 168         return 1;
 169     }
 170     return 0;
 171 }
 172
 173 static void destroy_schemas(struct filter_info *tinfo)
 174 {
 175     struct filter_schema *schema = tinfo->schemas;
 176     while (schema)
 177     {
 178         struct filter_schema *schema_next = schema->next;
 179         if (schema->stylesheet_xsp)
 180             xsltFreeStylesheet(schema->stylesheet_xsp);
 181         xfree(schema);
 182         schema = schema_next;
 183     }
 184     tinfo->schemas = 0;
 185     xfree(tinfo->fname);
 186     if (tinfo->doc)
 187         xmlFreeDoc(tinfo->doc);
 188     tinfo->doc = 0;
 189 }
 190
 191 static ZEBRA_RES create_schemas(struct filter_info *tinfo, const char *fname)
 192 {
 193     char tmp_full_name[1024];
 194     xmlNodePtr ptr;
 195     tinfo->fname = xstrdup(fname);
 196
 197     if (yaz_filepath_resolve(tinfo->fname, tinfo->profile_path,
 198                              NULL, tmp_full_name))
 199         tinfo->full_name = xstrdup(tmp_full_name);
 200     else
 201         tinfo->full_name = xstrdup(tinfo->fname);
 202
 203     yaz_log(YLOG_LOG, "alvis filter: loading config file %s", tinfo->full_name);
 204
 205     tinfo->doc = xmlParseFile(tinfo->full_name);
 206
 207     if (!tinfo->doc)
 208     {
 209         yaz_log(YLOG_WARN, "alvis filter: could not parse config file %s",
 210                 tinfo->full_name);
 211
 212         return ZEBRA_FAIL;
 213     }
 214
 215     ptr = xmlDocGetRootElement(tinfo->doc);
 216     if (!ptr || ptr->type != XML_ELEMENT_NODE
 217         || XML_STRCMP(ptr->name, "schemaInfo"))
 218     {
 219         yaz_log(YLOG_WARN,
 220                 "alvis filter:  config file %s :"
 221                 " expected root element <schemaInfo>",
 222                 tinfo->full_name);
 223         return ZEBRA_FAIL;
 224     }
 225
 226     for (ptr = ptr->children; ptr; ptr = ptr->next)
 227     {
 228         if (ptr->type != XML_ELEMENT_NODE)
 229             continue;
 230         if (!XML_STRCMP(ptr->name, "schema"))
 231         {
 232             struct _xmlAttr *attr;
 233             struct filter_schema *schema = xmalloc(sizeof(*schema));
 234             schema->name = 0;
 235             schema->identifier = 0;
 236             schema->stylesheet = 0;
 237             schema->default_schema = 0;
 238             schema->next = tinfo->schemas;
 239             schema->stylesheet_xsp = 0;
 240             schema->include_snippet = 0;
 241             tinfo->schemas = schema;
 242             for (attr = ptr->properties; attr; attr = attr->next)
 243             {
 244                 attr_content(attr, "identifier", &schema->identifier);
 245                 attr_content(attr, "name", &schema->name);
 246                 attr_content(attr, "stylesheet", &schema->stylesheet);
 247                 attr_content(attr, "default", &schema->default_schema);
 248                 attr_content(attr, "snippet", &schema->include_snippet);
 249             }
 250             /*yaz_log(YLOG_LOG, "XSLT add %s %s %s",
 251               schema->name, schema->identifier, schema->stylesheet); */
 252
 253             /* find requested schema */
 254
 255             if (schema->stylesheet)
 256             {
 257                 char tmp_xslt_full_name[1024];
 258                 if (!yaz_filepath_resolve(schema->stylesheet, tinfo->profile_path,
 259                                           NULL, tmp_xslt_full_name))
 260                 {
 261                     yaz_log(YLOG_WARN,
 262                             "alvis filter: stylesheet %s not found in path %s",
 263                             schema->stylesheet, tinfo->profile_path);
 264                     return ZEBRA_FAIL;
 265                 }
 266                 schema->stylesheet_xsp
 267                     = xsltParseStylesheetFile((const xmlChar*) tmp_xslt_full_name);
 268                 if (!schema->stylesheet_xsp)
 269                 {
 270                     yaz_log(YLOG_WARN,
 271                             "alvis filter: could not parse xslt stylesheet %s",
 272                             tmp_xslt_full_name);
 273                     return ZEBRA_FAIL;
 274                 }
 275             }
 276         }
 277         else if (!XML_STRCMP(ptr->name, "split"))
 278         {
 279             struct _xmlAttr *attr;
 280             for (attr = ptr->properties; attr; attr = attr->next)
 281             {
 282                 const char *split_level_str = 0;
 283                 attr_content(attr, "level", &split_level_str);
 284                 tinfo->split_level =
 285                     split_level_str ? atoi(split_level_str) : 0;
 286             }
 287         }
 288         else
 289         {
 290             yaz_log(YLOG_WARN, "Bad element %s in %s", ptr->name, fname);
 291             return ZEBRA_FAIL;
 292         }
 293     }
 294     return ZEBRA_OK;
 295 }
 296
 297 static struct filter_schema *lookup_schema(struct filter_info *tinfo,
 298                                            const char *est)
 299 {
 300     struct filter_schema *schema;
 301
 302     for (schema = tinfo->schemas; schema; schema = schema->next)
 303     {
 304         /* find requested schema */
 305         if (est)
 306         {
 307             if (schema->identifier && !strcmp(schema->identifier, est))
 308                 return schema;
 309
 310             if (schema->name && !strcmp(schema->name, est))
 311                 return schema;
 312         }
 313         /* or return default schema if defined */
 314         else if (schema->default_schema)
 315             return schema;
 316     }
 317
 318     /* return first schema if no default schema defined */
 319     if (tinfo->schemas)
 320         return tinfo->schemas;
 321
 322     return 0;
 323 }
 324
 325 static ZEBRA_RES filter_config(void *clientData, Res res, const char *args)
 326 {
 327     struct filter_info *tinfo = clientData;
 328     if (!args || !*args)
 329     {
 330         yaz_log(YLOG_WARN, "alvis filter: need config file");
 331         return ZEBRA_FAIL;
 332     }
 333
 334     if (tinfo->fname && !strcmp(args, tinfo->fname))
 335         return ZEBRA_OK;
 336
 337     tinfo->profile_path = res_get(res, "profilePath");
 338     yaz_log(YLOG_LOG, "alvis filter: profilePath %s", tinfo->profile_path);
 339
 340     destroy_schemas(tinfo);
 341     return create_schemas(tinfo, args);
 342 }
 343
 344 static void filter_destroy(void *clientData)
 345 {
 346     struct filter_info *tinfo = clientData;
 347     destroy_schemas(tinfo);
 348     if (tinfo->reader)
 349         xmlFreeTextReader(tinfo->reader);
 350     odr_destroy(tinfo->odr);
 351     xfree(tinfo);
 352 }
 353
 354 static int ioread_ex(void *context, char *buffer, int len)
 355 {
 356     struct recExtractCtrl *p = context;
 357     return p->stream->readf(p->stream, buffer, len);
 358 }
 359
 360 static int ioclose_ex(void *context)
 361 {
 362     return 0;
 363 }
 364
 365 static void index_cdata(struct filter_info *tinfo, struct recExtractCtrl *ctrl,
 366                         xmlNodePtr ptr, RecWord *recWord)
 367 {
 368     for(; ptr; ptr = ptr->next)
 369     {
 370         index_cdata(tinfo, ctrl, ptr->children, recWord);
 371         if (ptr->type != XML_TEXT_NODE)
 372             continue;
 373         recWord->term_buf = (const char *)ptr->content;
 374         recWord->term_len = XML_STRLEN(ptr->content);
 375         (*ctrl->tokenAdd)(recWord);
 376     }
 377 }
 378
 379 static void index_node(struct filter_info *tinfo,  struct recExtractCtrl *ctrl,
 380                        xmlNodePtr ptr, RecWord *recWord)
 381 {
 382     for(; ptr; ptr = ptr->next)
 383     {
 384         index_node(tinfo, ctrl, ptr->children, recWord);
 385         if (ptr->type != XML_ELEMENT_NODE || !ptr->ns ||
 386             XML_STRCMP(ptr->ns->href, zebra_xslt_ns))
 387             continue;
 388         if (!XML_STRCMP(ptr->name, "index"))
 389         {
 390             const char *name_str = 0;
 391             const char *type_str = 0;
 392             const char *xpath_str = 0;
 393             struct _xmlAttr *attr;
 394             for (attr = ptr->properties; attr; attr = attr->next)
 395             {
 396                 attr_content(attr, "name", &name_str);
 397                 attr_content(attr, "xpath", &xpath_str);
 398                 attr_content(attr, "type", &type_str);
 399             }
 400             if (name_str)
 401             {
 402                 int prev_type = recWord->index_type; /* save default type */
 403
 404                 if (type_str && *type_str)
 405                     recWord->index_type = *type_str; /* type was given */
 406                 recWord->index_name = name_str;
 407                 index_cdata(tinfo, ctrl, ptr->children, recWord);
 408
 409                 recWord->index_type = prev_type;     /* restore it again */
 410             }
 411         }
 412     }
 413 }
 414
 415 static void index_record(struct filter_info *tinfo,struct recExtractCtrl *ctrl,
 416                          xmlNodePtr ptr, RecWord *recWord)
 417 {
 418     const char *type_str = "update";
 419
 420     if (ptr && ptr->type == XML_ELEMENT_NODE && ptr->ns &&
 421         !XML_STRCMP(ptr->ns->href, zebra_xslt_ns)
 422         && !XML_STRCMP(ptr->name, "record"))
 423     {
 424         const char *id_str = 0;
 425         const char *rank_str = 0;
 426         struct _xmlAttr *attr;
 427         for (attr = ptr->properties; attr; attr = attr->next)
 428         {
 429             attr_content(attr, "type", &type_str);
 430             attr_content(attr, "id", &id_str);
 431             attr_content(attr, "rank", &rank_str);
 432         }
 433         if (id_str)
 434             sscanf(id_str, "%255s", ctrl->match_criteria);
 435
 436         if (rank_str)
 437             ctrl->staticrank = atozint(rank_str);
 438         ptr = ptr->children;
 439     }
 440
 441     if (!strcmp("update", type_str))
 442         index_node(tinfo, ctrl, ptr, recWord);
 443     else if (!strcmp("delete", type_str))
 444          yaz_log(YLOG_WARN, "alvis filter delete: to be implemented");
 445     else
 446          yaz_log(YLOG_WARN, "alvis filter: unknown record type '%s'",
 447                  type_str);
 448 }
 449
 450 static int extract_doc(struct filter_info *tinfo, struct recExtractCtrl *p,
 451                        xmlDocPtr doc)
 452 {
 453     RecWord recWord;
 454     const char *params[10];
 455     xmlChar *buf_out;
 456     int len_out;
 457
 458     struct filter_schema *schema = lookup_schema(tinfo, zebra_xslt_ns);
 459
 460     params[0] = 0;
 461     set_param_str(params, "schema", zebra_xslt_ns, tinfo->odr);
 462
 463     (*p->init)(p, &recWord);
 464
 465     if (schema && schema->stylesheet_xsp)
 466     {
 467         xmlNodePtr root_ptr;
 468         xmlDocPtr resDoc =
 469             xsltApplyStylesheet(schema->stylesheet_xsp,
 470                                 doc, params);
 471         if (p->flagShowRecords)
 472         {
 473             xmlDocDumpMemory(resDoc, &buf_out, &len_out);
 474             fwrite(buf_out, len_out, 1, stdout);
 475             xmlFree(buf_out);
 476         }
 477         root_ptr = xmlDocGetRootElement(resDoc);
 478         if (root_ptr)
 479             index_record(tinfo, p, root_ptr, &recWord);
 480         else
 481         {
 482             yaz_log(YLOG_WARN, "No root for index XML record."
 483                     " split_level=%d stylesheet=%s",
 484                     tinfo->split_level, schema->stylesheet);
 485         }
 486         xmlFreeDoc(resDoc);
 487     }
 488     xmlDocDumpMemory(doc, &buf_out, &len_out);
 489     if (p->flagShowRecords)
 490         fwrite(buf_out, len_out, 1, stdout);
 491     (*p->setStoreData)(p, buf_out, len_out);
 492     xmlFree(buf_out);
 493
 494     xmlFreeDoc(doc);
 495     return RECCTRL_EXTRACT_OK;
 496 }
 497
 498 static int extract_split(struct filter_info *tinfo, struct recExtractCtrl *p)
 499 {
 500     int ret;
 501
 502     if (p->first_record)
 503     {
 504         if (tinfo->reader)
 505             xmlFreeTextReader(tinfo->reader);
 506         tinfo->reader = xmlReaderForIO(ioread_ex, ioclose_ex,
 507                                        p /* I/O handler */,
 508                                        0 /* URL */,
 509                                        0 /* encoding */,
 510                                        XML_PARSE_XINCLUDE|XML_PARSE_NOENT);
 511     }
 512     if (!tinfo->reader)
 513         return RECCTRL_EXTRACT_ERROR_GENERIC;
 514
 515     ret = xmlTextReaderRead(tinfo->reader);
 516     while (ret == 1)
 517     {
 518         int type = xmlTextReaderNodeType(tinfo->reader);
 519         int depth = xmlTextReaderDepth(tinfo->reader);
 520         if (type == XML_READER_TYPE_ELEMENT && tinfo->split_level == depth)
 521         {
 522             xmlNodePtr ptr = xmlTextReaderExpand(tinfo->reader);
 523             if (ptr)
 524             {
 525                 xmlNodePtr ptr2 = xmlCopyNode(ptr, 1);
 526                 xmlDocPtr doc = xmlNewDoc((const xmlChar*) "1.0");
 527
 528                 xmlDocSetRootElement(doc, ptr2);
 529
 530                 return extract_doc(tinfo, p, doc);
 531             }
 532             else
 533             {
 534                 xmlFreeTextReader(tinfo->reader);
 535                 tinfo->reader = 0;
 536                 return RECCTRL_EXTRACT_ERROR_GENERIC;
 537             }
 538         }
 539         ret = xmlTextReaderRead(tinfo->reader);
 540     }
 541     xmlFreeTextReader(tinfo->reader);
 542     tinfo->reader = 0;
 543     return RECCTRL_EXTRACT_EOF;
 544 }
 545
 546 static int extract_full(struct filter_info *tinfo, struct recExtractCtrl *p)
 547 {
 548     if (p->first_record) /* only one record per stream */
 549     {
 550        xmlDocPtr doc = xmlReadIO(ioread_ex, ioclose_ex, p /* I/O handler */,
 551                                  0 /* URL */,
 552                                  0 /* encoding */,
 553                                  XML_PARSE_XINCLUDE|XML_PARSE_NOENT);
 554        if (!doc)
 555        {
 556            return RECCTRL_EXTRACT_ERROR_GENERIC;
 557        }
 558        return extract_doc(tinfo, p, doc);
 559     }
 560     else
 561        return RECCTRL_EXTRACT_EOF;
 562 }
 563
 564 static int filter_extract(void *clientData, struct recExtractCtrl *p)
 565 {
 566     struct filter_info *tinfo = clientData;
 567
 568     odr_reset(tinfo->odr);
 569     if (tinfo->split_level == 0)
 570         return extract_full(tinfo, p);
 571     else
 572         return extract_split(tinfo, p);
 573 }
 574
 575 static int ioread_ret(void *context, char *buffer, int len)
 576 {
 577     struct recRetrieveCtrl *p = context;
 578     return p->stream->readf(p->stream, buffer, len);
 579 }
 580
 581 static int ioclose_ret(void *context)
 582 {
 583     return 0;
 584 }
 585
 586 static const char *snippet_doc(struct recRetrieveCtrl *p, int text_mode,
 587                                int window_size)
 588 {
 589     const char *xml_doc_str;
 590     int ord = 0;
 591     WRBUF wrbuf = wrbuf_alloc();
 592     zebra_snippets *res =
 593         zebra_snippets_window(p->doc_snippet, p->hit_snippet, window_size);
 594     zebra_snippet_word *w = zebra_snippets_list(res);
 595
 596     if (text_mode)
 597         wrbuf_printf(wrbuf, "\'");
 598     else
 599         wrbuf_printf(wrbuf, "<snippet xmlns='%s'>\n", zebra_xslt_ns);
 600     for (; w; w = w->next)
 601     {
 602         if (ord == 0)
 603             ord = w->ord;
 604         else if (ord != w->ord)
 605
 606             break;
 607         if (text_mode)
 608             wrbuf_printf(wrbuf, "%s%s%s ",
 609                          w->match ? "*" : "",
 610                          w->term,
 611                          w->match ? "*" : "");
 612         else
 613         {
 614             wrbuf_printf(wrbuf, " <term ord='%d' seqno='" ZINT_FORMAT "' %s>",
 615                          w->ord, w->seqno,
 616                          (w->match ? "match='1'" : ""));
 617             wrbuf_xmlputs(wrbuf, w->term);
 618             wrbuf_printf(wrbuf, "</term>\n");
 619         }
 620     }
 621     if (text_mode)
 622         wrbuf_printf(wrbuf, "\'");
 623     else
 624         wrbuf_printf(wrbuf, "</snippet>\n");
 625
 626     xml_doc_str = odr_strdup(p->odr, wrbuf_buf(wrbuf));
 627
 628     zebra_snippets_destroy(res);
 629     wrbuf_free(wrbuf, 1);
 630     return xml_doc_str;
 631 }
 632
 633 static int filter_retrieve (void *clientData, struct recRetrieveCtrl *p)
 634 {
 635     /* const char *esn = zebra_xslt_ns; */
 636     const char *esn = 0;
 637     const char *params[32];
 638     struct filter_info *tinfo = clientData;
 639     xmlDocPtr resDoc;
 640     xmlDocPtr doc;
 641     struct filter_schema *schema;
 642     int window_size = -1;
 643
 644     if (p->comp)
 645     {
 646         if (p->comp->which == Z_RecordComp_simple
 647             && p->comp->u.simple->which == Z_ElementSetNames_generic)
 648         {
 649             esn = p->comp->u.simple->u.generic;
 650         }
 651         else if (p->comp->which == Z_RecordComp_complex
 652                  && p->comp->u.complex->generic->elementSpec
 653                  && p->comp->u.complex->generic->elementSpec->which ==
 654                  Z_ElementSpec_elementSetName)
 655         {
 656             esn = p->comp->u.complex->generic->elementSpec->u.elementSetName;
 657         }
 658     }
 659     schema = lookup_schema(tinfo, esn);
 660     if (!schema)
 661     {
 662         p->diagnostic =
 663             YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 664         return 0;
 665     }
 666
 667     if (schema->include_snippet)
 668         window_size = atoi(schema->include_snippet);
 669
 670     params[0] = 0;
 671     set_param_int(params, "id", p->localno, p->odr);
 672     if (p->fname)
 673         set_param_str(params, "filename", p->fname, p->odr);
 674     if (p->staticrank >= 0)
 675         set_param_int(params, "rank", p->staticrank, p->odr);
 676
 677     if (esn)
 678         set_param_str(params, "schema", esn, p->odr);
 679     else
 680         if (schema->name)
 681             set_param_str(params, "schema", schema->name, p->odr);
 682         else if (schema->identifier)
 683             set_param_str(params, "schema", schema->identifier, p->odr);
 684         else
 685             set_param_str(params, "schema", "", p->odr);
 686
 687     if (p->score >= 0)
 688         set_param_int(params, "score", p->score, p->odr);
 689     set_param_int(params, "size", p->recordSize, p->odr);
 690
 691     if (window_size >= 0)
 692         set_param_xml(params, "snippet", snippet_doc(p, 1, window_size),
 693                       p->odr);
 694     doc = xmlReadIO(ioread_ret, ioclose_ret, p /* I/O handler */,
 695                     0 /* URL */,
 696                     0 /* encoding */,
 697                     XML_PARSE_XINCLUDE|XML_PARSE_NOENT);
 698     if (!doc)
 699     {
 700         p->diagnostic = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 701         return 0;
 702     }
 703
 704     if (window_size >= 0)
 705     {
 706         xmlNodePtr node = xmlDocGetRootElement(doc);
 707         const char *snippet_str = snippet_doc(p, 0, window_size);
 708         xmlDocPtr snippet_doc = xmlParseMemory(snippet_str, strlen(snippet_str));
 709         xmlAddChild(node, xmlDocGetRootElement(snippet_doc));
 710     }
 711     if (!schema->stylesheet_xsp)
 712         resDoc = doc;
 713     else
 714     {
 715         resDoc = xsltApplyStylesheet(schema->stylesheet_xsp,
 716                                      doc, params);
 717         xmlFreeDoc(doc);
 718     }
 719     if (!resDoc)
 720     {
 721         p->diagnostic = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
 722     }
 723     else if (p->input_format == VAL_NONE || p->input_format == VAL_TEXT_XML)
 724     {
 725         xmlChar *buf_out;
 726         int len_out;
 727
 728         if (schema->stylesheet_xsp)
 729             xsltSaveResultToString(&buf_out, &len_out, resDoc,
 730                                    schema->stylesheet_xsp);
 731         else
 732             xmlDocDumpMemory(resDoc, &buf_out, &len_out);
 733
 734         p->output_format = VAL_TEXT_XML;
 735         p->rec_len = len_out;
 736         p->rec_buf = odr_malloc(p->odr, p->rec_len);
 737         memcpy(p->rec_buf, buf_out, p->rec_len);
 738         xmlFree(buf_out);
 739     }
 740     else if (p->output_format == VAL_SUTRS)
 741     {
 742         xmlChar *buf_out;
 743         int len_out;
 744
 745         if (schema->stylesheet_xsp)
 746             xsltSaveResultToString(&buf_out, &len_out, resDoc,
 747                                    schema->stylesheet_xsp);
 748         else
 749             xmlDocDumpMemory(resDoc, &buf_out, &len_out);
 750
 751         p->output_format = VAL_SUTRS;
 752         p->rec_len = len_out;
 753         p->rec_buf = odr_malloc(p->odr, p->rec_len);
 754         memcpy(p->rec_buf, buf_out, p->rec_len);
 755
 756         xmlFree(buf_out);
 757     }
 758     else
 759     {
 760         p->diagnostic = YAZ_BIB1_RECORD_SYNTAX_UNSUPP;
 761     }
 762     xmlFreeDoc(resDoc);
 763     return 0;
 764 }
 765
 766 static struct recType filter_type = {
 767     0,
 768     "alvis",
 769     filter_init,
 770     filter_config,
 771     filter_destroy,
 772     filter_extract,
 773     filter_retrieve
 774 };
 775
 776 RecType
 777 #ifdef IDZEBRA_STATIC_ALVIS
 778 idzebra_filter_alvis
 779 #else
 780 idzebra_filter
 781 #endif
 782
 783 [] = {
 784     &filter_type,
 785     0,
 786 };
 787 /*
 788  * Local variables:
 789  * c-basic-offset: 4
 790  * indent-tabs-mode: nil
 791  * End:
 792  * vim: shiftwidth=4 tabstop=8 expandtab
 793  */
 794