1 /* $Id: alvis.c,v 1.5 2005-08-19 21:40:17 adam Exp $
2 Copyright (C) 1995-2005
5 This file is part of the Zebra server.
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
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
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
27 #include <yaz/diagbib1.h>
28 #include <libxml/xmlversion.h>
29 #include <libxml/parser.h>
30 #include <libxml/tree.h>
31 #include <libxml/xmlIO.h>
32 #include <libxml/xmlreader.h>
33 #include <libxslt/transform.h>
35 #include <idzebra/util.h>
36 #include <idzebra/recctrl.h>
38 struct filter_schema {
40 const char *identifier;
41 const char *stylesheet;
42 struct filter_schema *next;
43 const char *default_schema;
44 const char *include_snippet;
45 xsltStylesheetPtr stylesheet_xsp;
51 const char *split_level;
52 const char *split_path;
54 struct filter_schema *schemas;
55 xmlTextReaderPtr reader;
58 #define ZEBRA_SCHEMA_XSLT_NS "http://indexdata.dk/zebra/xslt/1"
60 static const char *zebra_xslt_ns = ZEBRA_SCHEMA_XSLT_NS;
62 static void set_param_xml(const char **params, const char *name,
63 const char *value, ODR odr)
72 static void set_param_str(const char **params, const char *name,
73 const char *value, ODR odr)
75 char *quoted = odr_malloc(odr, 3 + strlen(value));
76 sprintf(quoted, "'%s'", value);
84 static void set_param_int(const char **params, const char *name,
87 char *quoted = odr_malloc(odr, 30); /* 25 digits enough for 2^64 */
90 sprintf(quoted, "'" ZINT_FORMAT "'", value);
96 static int zebra_xmlInputMatchCallback (char const *filename)
98 /* yaz_log(YLOG_LOG, "match %s", filename); */
102 static void * zebra_xmlInputOpenCallback (char const *filename)
107 static int zebra_xmlInputReadCallback (void * context, char * buffer, int len)
112 static int zebra_xmlInputCloseCallback (void * context)
117 static void *filter_init_xslt(Res res, RecType recType)
119 struct filter_info *tinfo = (struct filter_info *) xmalloc(sizeof(*tinfo));
122 tinfo->split_level = 0;
123 tinfo->split_path = 0;
124 tinfo->odr = odr_createmem(ODR_ENCODE);
129 xmlRegisterDefaultInputCallbacks();
130 xmlRegisterInputCallbacks(zebra_xmlInputMatchCallback,
131 zebra_xmlInputOpenCallback,
132 zebra_xmlInputReadCallback,
133 zebra_xmlInputCloseCallback);
138 static int attr_content(struct _xmlAttr *attr, const char *name,
139 const char **dst_content)
141 if (!strcmp(attr->name, name) && attr->children &&
142 attr->children->type == XML_TEXT_NODE)
144 *dst_content = attr->children->content;
150 static void destroy_schemas(struct filter_info *tinfo)
152 struct filter_schema *schema = tinfo->schemas;
155 struct filter_schema *schema_next = schema->next;
156 if (schema->stylesheet_xsp)
157 xsltFreeStylesheet(schema->stylesheet_xsp);
159 schema = schema_next;
164 xmlFreeDoc(tinfo->doc);
168 static ZEBRA_RES create_schemas(struct filter_info *tinfo, const char *fname)
171 tinfo->fname = xstrdup(fname);
172 tinfo->doc = xmlParseFile(tinfo->fname);
175 ptr = xmlDocGetRootElement(tinfo->doc);
176 if (!ptr || ptr->type != XML_ELEMENT_NODE ||
177 strcmp(ptr->name, "schemaInfo"))
179 for (ptr = ptr->children; ptr; ptr = ptr->next)
181 if (ptr->type != XML_ELEMENT_NODE)
183 if (!strcmp(ptr->name, "schema"))
185 struct _xmlAttr *attr;
186 struct filter_schema *schema = xmalloc(sizeof(*schema));
188 schema->identifier = 0;
189 schema->stylesheet = 0;
190 schema->default_schema = 0;
191 schema->next = tinfo->schemas;
192 schema->stylesheet_xsp = 0;
193 schema->include_snippet = 0;
194 tinfo->schemas = schema;
195 for (attr = ptr->properties; attr; attr = attr->next)
197 attr_content(attr, "identifier", &schema->identifier);
198 attr_content(attr, "name", &schema->name);
199 attr_content(attr, "stylesheet", &schema->stylesheet);
200 attr_content(attr, "default", &schema->default_schema);
201 attr_content(attr, "snippet", &schema->include_snippet);
203 if (schema->stylesheet)
204 schema->stylesheet_xsp =
205 xsltParseStylesheetFile(
206 (const xmlChar*) schema->stylesheet);
208 else if (!strcmp(ptr->name, "split"))
210 struct _xmlAttr *attr;
211 for (attr = ptr->properties; attr; attr = attr->next)
213 attr_content(attr, "level", &tinfo->split_level);
214 attr_content(attr, "path", &tinfo->split_path);
219 yaz_log(YLOG_WARN, "Bad element %s in %s", ptr->name, fname);
226 static struct filter_schema *lookup_schema(struct filter_info *tinfo,
229 struct filter_schema *schema;
230 for (schema = tinfo->schemas; schema; schema = schema->next)
234 if (schema->identifier && !strcmp(schema->identifier, est))
236 if (schema->name && !strcmp(schema->name, est))
239 if (schema->default_schema)
245 static void filter_config(void *clientData, Res res, const char *args)
247 struct filter_info *tinfo = clientData;
249 args = "xsltfilter.xml";
250 if (tinfo->fname && !strcmp(args, tinfo->fname))
252 destroy_schemas(tinfo);
253 create_schemas(tinfo, args);
256 static void filter_destroy(void *clientData)
258 struct filter_info *tinfo = clientData;
259 destroy_schemas(tinfo);
261 xmlFreeTextReader(tinfo->reader);
262 odr_destroy(tinfo->odr);
266 static int ioread_ex(void *context, char *buffer, int len)
268 struct recExtractCtrl *p = context;
269 return (*p->readf)(p->fh, buffer, len);
272 static int ioclose_ex(void *context)
277 static void index_cdata(struct filter_info *tinfo, struct recExtractCtrl *ctrl,
278 xmlNodePtr ptr, RecWord *recWord)
280 for(; ptr; ptr = ptr->next)
282 index_cdata(tinfo, ctrl, ptr->children, recWord);
283 if (ptr->type != XML_TEXT_NODE)
285 recWord->term_buf = ptr->content;
286 recWord->term_len = strlen(ptr->content);
287 (*ctrl->tokenAdd)(recWord);
291 static void index_node(struct filter_info *tinfo, struct recExtractCtrl *ctrl,
292 xmlNodePtr ptr, RecWord *recWord)
294 for(; ptr; ptr = ptr->next)
296 index_node(tinfo, ctrl, ptr->children, recWord);
297 if (ptr->type != XML_ELEMENT_NODE || !ptr->ns ||
298 strcmp(ptr->ns->href, zebra_xslt_ns))
300 if (!strcmp(ptr->name, "index"))
302 const char *name_str = 0;
303 const char *type_str = 0;
304 const char *xpath_str = 0;
305 struct _xmlAttr *attr;
306 for (attr = ptr->properties; attr; attr = attr->next)
308 attr_content(attr, "name", &name_str);
309 attr_content(attr, "xpath", &xpath_str);
310 attr_content(attr, "type", &type_str);
314 int prev_type = recWord->index_type; /* save default type */
316 if (type_str && *type_str)
317 recWord->index_type = *type_str; /* type was given */
318 recWord->index_name = name_str;
319 index_cdata(tinfo, ctrl, ptr->children, recWord);
321 recWord->index_type = prev_type; /* restore it again */
327 static void index_record(struct filter_info *tinfo,struct recExtractCtrl *ctrl,
328 xmlNodePtr ptr, RecWord *recWord)
330 if (ptr && ptr->type == XML_ELEMENT_NODE && ptr->ns &&
331 !strcmp(ptr->ns->href, zebra_xslt_ns)
332 && !strcmp(ptr->name, "record"))
334 const char *type_str = "update";
335 const char *id_str = 0;
336 const char *rank_str = 0;
337 struct _xmlAttr *attr;
338 for (attr = ptr->properties; attr; attr = attr->next)
340 attr_content(attr, "type", &type_str);
341 attr_content(attr, "id", &id_str);
342 attr_content(attr, "rank", &rank_str);
345 sscanf(id_str, "%255s", ctrl->match_criteria);
348 ctrl->staticrank = atoi(rank_str);
349 yaz_log(YLOG_LOG, "rank=%d",ctrl->staticrank);
352 yaz_log(YLOG_LOG, "no rank");
356 index_node(tinfo, ctrl, ptr, recWord);
359 static int extract_doc(struct filter_info *tinfo, struct recExtractCtrl *p,
363 const char *params[10];
367 struct filter_schema *schema = lookup_schema(tinfo, zebra_xslt_ns);
370 set_param_str(params, "schema", zebra_xslt_ns, tinfo->odr);
372 (*p->init)(p, &recWord);
374 if (schema && schema->stylesheet_xsp)
378 xsltApplyStylesheet(schema->stylesheet_xsp,
380 if (p->flagShowRecords)
382 xmlDocDumpMemory(resDoc, &buf_out, &len_out);
383 fwrite(buf_out, len_out, 1, stdout);
386 root_ptr = xmlDocGetRootElement(resDoc);
388 index_record(tinfo, p, root_ptr, &recWord);
391 yaz_log(YLOG_WARN, "No root for index XML record."
392 " split_level=%s stylesheet=%s",
393 tinfo->split_level, schema->stylesheet);
397 xmlDocDumpMemory(doc, &buf_out, &len_out);
398 if (p->flagShowRecords)
399 fwrite(buf_out, len_out, 1, stdout);
400 (*p->setStoreData)(p, buf_out, len_out);
404 return RECCTRL_EXTRACT_OK;
407 static int extract_split(struct filter_info *tinfo, struct recExtractCtrl *p)
414 xmlFreeTextReader(tinfo->reader);
415 tinfo->reader = xmlReaderForIO(ioread_ex, ioclose_ex,
422 return RECCTRL_EXTRACT_ERROR_GENERIC;
424 if (tinfo->split_level)
425 split_depth = atoi(tinfo->split_level);
426 ret = xmlTextReaderRead(tinfo->reader);
428 int type = xmlTextReaderNodeType(tinfo->reader);
429 int depth = xmlTextReaderDepth(tinfo->reader);
430 if (split_depth == 0 ||
432 type == XML_READER_TYPE_ELEMENT && split_depth == depth))
434 xmlNodePtr ptr = xmlTextReaderExpand(tinfo->reader);
435 xmlNodePtr ptr2 = xmlCopyNode(ptr, 1);
436 xmlDocPtr doc = xmlNewDoc("1.0");
438 xmlDocSetRootElement(doc, ptr2);
440 return extract_doc(tinfo, p, doc);
442 ret = xmlTextReaderRead(tinfo->reader);
444 xmlFreeTextReader(tinfo->reader);
446 return RECCTRL_EXTRACT_EOF;
449 static int extract_full(struct filter_info *tinfo, struct recExtractCtrl *p)
451 if (p->first_record) /* only one record per stream */
453 xmlDocPtr doc = xmlReadIO(ioread_ex, ioclose_ex, p /* I/O handler */,
459 return RECCTRL_EXTRACT_ERROR_GENERIC;
461 return extract_doc(tinfo, p, doc);
464 return RECCTRL_EXTRACT_EOF;
467 static int filter_extract(void *clientData, struct recExtractCtrl *p)
469 struct filter_info *tinfo = clientData;
471 odr_reset(tinfo->odr);
473 if (tinfo->split_level == 0 && tinfo->split_path == 0)
474 return extract_full(tinfo, p);
477 return extract_split(tinfo, p);
481 static int ioread_ret(void *context, char *buffer, int len)
483 struct recRetrieveCtrl *p = context;
484 return (*p->readf)(p->fh, buffer, len);
487 static int ioclose_ret(void *context)
493 static const char *snippet_doc(struct recRetrieveCtrl *p, int text_mode,
496 const char *xml_doc_str;
498 WRBUF wrbuf = wrbuf_alloc();
499 zebra_snippets *res =
500 zebra_snippets_window(p->doc_snippet, p->hit_snippet, window_size);
501 zebra_snippet_word *w = zebra_snippets_list(res);
504 wrbuf_printf(wrbuf, "\'");
506 wrbuf_printf(wrbuf, "<snippet xmlns='%s'>\n", zebra_xslt_ns);
507 for (; w; w = w->next)
511 else if (ord != w->ord)
515 wrbuf_printf(wrbuf, "%s%s%s ",
518 w->match ? "*" : "");
521 wrbuf_printf(wrbuf, " <term ord='%d' seqno='" ZINT_FORMAT "' %s>",
523 (w->match ? "match='1'" : ""));
524 wrbuf_xmlputs(wrbuf, w->term);
525 wrbuf_printf(wrbuf, "</term>\n");
529 wrbuf_printf(wrbuf, "\'");
531 wrbuf_printf(wrbuf, "</snippet>\n");
533 xml_doc_str = odr_strdup(p->odr, wrbuf_buf(wrbuf));
535 zebra_snippets_destroy(res);
536 wrbuf_free(wrbuf, 1);
540 static int filter_retrieve (void *clientData, struct recRetrieveCtrl *p)
542 const char *esn = zebra_xslt_ns;
543 const char *params[10];
544 struct filter_info *tinfo = clientData;
547 struct filter_schema *schema;
548 int window_size = -1;
552 if (p->comp->which != Z_RecordComp_simple
553 || p->comp->u.simple->which != Z_ElementSetNames_generic)
555 p->diagnostic = YAZ_BIB1_PRESENT_COMP_SPEC_PARAMETER_UNSUPP;
558 esn = p->comp->u.simple->u.generic;
560 schema = lookup_schema(tinfo, esn);
564 YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
568 if (schema->include_snippet)
569 window_size = atoi(schema->include_snippet);
572 set_param_str(params, "schema", esn, p->odr);
574 set_param_str(params, "filename", p->fname, p->odr);
576 set_param_int(params, "score", p->score, p->odr);
577 set_param_int(params, "size", p->recordSize, p->odr);
578 set_param_int(params, "id", p->localno, p->odr);
580 if (window_size >= 0)
581 set_param_xml(params, "snippet", snippet_doc(p, 1, window_size),
583 doc = xmlReadIO(ioread_ret, ioclose_ret, p /* I/O handler */,
589 p->diagnostic = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
593 if (window_size >= 0)
595 xmlNodePtr node = xmlDocGetRootElement(doc);
596 const char *snippet_str = snippet_doc(p, 0, window_size);
597 xmlDocPtr snippet_doc = xmlParseMemory(snippet_str, strlen(snippet_str));
598 xmlAddChild(node, xmlDocGetRootElement(snippet_doc));
600 if (!schema->stylesheet_xsp)
604 resDoc = xsltApplyStylesheet(schema->stylesheet_xsp,
610 p->diagnostic = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
612 else if (p->input_format == VAL_NONE || p->input_format == VAL_TEXT_XML)
616 xmlDocDumpMemory(resDoc, &buf_out, &len_out);
618 p->output_format = VAL_TEXT_XML;
619 p->rec_len = len_out;
620 p->rec_buf = odr_malloc(p->odr, p->rec_len);
621 memcpy(p->rec_buf, buf_out, p->rec_len);
625 else if (p->output_format == VAL_SUTRS)
629 xmlDocDumpMemory(resDoc, &buf_out, &len_out);
631 p->output_format = VAL_SUTRS;
632 p->rec_len = len_out;
633 p->rec_buf = odr_malloc(p->odr, p->rec_len);
634 memcpy(p->rec_buf, buf_out, p->rec_len);
640 p->diagnostic = YAZ_BIB1_RECORD_SYNTAX_UNSUPP;
646 static struct recType filter_type_alvis = {
657 #ifdef IDZEBRA_STATIC_ALVIS