1 /* $Id: alvis.c,v 1.4 2005-08-19 14:06:31 marc 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);
97 int zebra_xmlInputMatchCallback (char const *filename)
99 /* yaz_log(YLOG_LOG, "match %s", filename); */
104 void * zebra_xmlInputOpenCallback (char const *filename)
109 int zebra_xmlInputReadCallback (void * context, char * buffer, int len)
114 int zebra_xmlInputCloseCallback (void * context)
119 static void *filter_init_xslt(Res res, RecType recType)
121 struct filter_info *tinfo = (struct filter_info *) xmalloc(sizeof(*tinfo));
124 tinfo->split_level = 0;
125 tinfo->split_path = 0;
126 tinfo->odr = odr_createmem(ODR_ENCODE);
131 xmlRegisterDefaultInputCallbacks();
132 xmlRegisterInputCallbacks(zebra_xmlInputMatchCallback,
133 zebra_xmlInputOpenCallback,
134 zebra_xmlInputReadCallback,
135 zebra_xmlInputCloseCallback);
140 static int attr_content(struct _xmlAttr *attr, const char *name,
141 const char **dst_content)
143 if (!strcmp(attr->name, name) && attr->children &&
144 attr->children->type == XML_TEXT_NODE)
146 *dst_content = attr->children->content;
152 static void destroy_schemas(struct filter_info *tinfo)
154 struct filter_schema *schema = tinfo->schemas;
157 struct filter_schema *schema_next = schema->next;
158 if (schema->stylesheet_xsp)
159 xsltFreeStylesheet(schema->stylesheet_xsp);
161 schema = schema_next;
166 xmlFreeDoc(tinfo->doc);
170 static ZEBRA_RES create_schemas(struct filter_info *tinfo, const char *fname)
173 tinfo->fname = xstrdup(fname);
174 tinfo->doc = xmlParseFile(tinfo->fname);
177 ptr = xmlDocGetRootElement(tinfo->doc);
178 if (!ptr || ptr->type != XML_ELEMENT_NODE ||
179 strcmp(ptr->name, "schemaInfo"))
181 for (ptr = ptr->children; ptr; ptr = ptr->next)
183 if (ptr->type != XML_ELEMENT_NODE)
185 if (!strcmp(ptr->name, "schema"))
187 struct _xmlAttr *attr;
188 struct filter_schema *schema = xmalloc(sizeof(*schema));
190 schema->identifier = 0;
191 schema->stylesheet = 0;
192 schema->default_schema = 0;
193 schema->next = tinfo->schemas;
194 schema->stylesheet_xsp = 0;
195 schema->include_snippet = 0;
196 tinfo->schemas = schema;
197 for (attr = ptr->properties; attr; attr = attr->next)
199 attr_content(attr, "identifier", &schema->identifier);
200 attr_content(attr, "name", &schema->name);
201 attr_content(attr, "stylesheet", &schema->stylesheet);
202 attr_content(attr, "default", &schema->default_schema);
203 attr_content(attr, "snippet", &schema->include_snippet);
205 if (schema->stylesheet)
206 schema->stylesheet_xsp =
207 xsltParseStylesheetFile(
208 (const xmlChar*) schema->stylesheet);
210 else if (!strcmp(ptr->name, "split"))
212 struct _xmlAttr *attr;
213 for (attr = ptr->properties; attr; attr = attr->next)
215 attr_content(attr, "level", &tinfo->split_level);
216 attr_content(attr, "path", &tinfo->split_path);
221 yaz_log(YLOG_WARN, "Bad element %s in %s", ptr->name, fname);
228 static struct filter_schema *lookup_schema(struct filter_info *tinfo,
231 struct filter_schema *schema;
232 for (schema = tinfo->schemas; schema; schema = schema->next)
236 if (schema->identifier && !strcmp(schema->identifier, est))
238 if (schema->name && !strcmp(schema->name, est))
241 if (schema->default_schema)
247 static void filter_config(void *clientData, Res res, const char *args)
249 struct filter_info *tinfo = clientData;
251 args = "xsltfilter.xml";
252 if (tinfo->fname && !strcmp(args, tinfo->fname))
254 destroy_schemas(tinfo);
255 create_schemas(tinfo, args);
258 static void filter_destroy(void *clientData)
260 struct filter_info *tinfo = clientData;
261 destroy_schemas(tinfo);
263 xmlFreeTextReader(tinfo->reader);
264 odr_destroy(tinfo->odr);
268 static int ioread_ex(void *context, char *buffer, int len)
270 struct recExtractCtrl *p = context;
271 return (*p->readf)(p->fh, buffer, len);
274 static int ioclose_ex(void *context)
279 static void index_cdata(struct filter_info *tinfo, struct recExtractCtrl *ctrl,
280 xmlNodePtr ptr, RecWord *recWord)
282 for(; ptr; ptr = ptr->next)
284 index_cdata(tinfo, ctrl, ptr->children, recWord);
285 if (ptr->type != XML_TEXT_NODE)
287 recWord->term_buf = ptr->content;
288 recWord->term_len = strlen(ptr->content);
289 (*ctrl->tokenAdd)(recWord);
293 static void index_node(struct filter_info *tinfo, struct recExtractCtrl *ctrl,
294 xmlNodePtr ptr, RecWord *recWord)
296 for(; ptr; ptr = ptr->next)
298 index_node(tinfo, ctrl, ptr->children, recWord);
299 if (ptr->type != XML_ELEMENT_NODE || !ptr->ns ||
300 strcmp(ptr->ns->href, zebra_xslt_ns))
302 if (!strcmp(ptr->name, "index"))
304 const char *name_str = 0;
305 const char *type_str = 0;
306 const char *xpath_str = 0;
307 struct _xmlAttr *attr;
308 for (attr = ptr->properties; attr; attr = attr->next)
310 attr_content(attr, "name", &name_str);
311 attr_content(attr, "xpath", &xpath_str);
312 attr_content(attr, "type", &type_str);
316 int prev_type = recWord->index_type; /* save default type */
318 if (type_str && *type_str)
319 recWord->index_type = *type_str; /* type was given */
320 recWord->index_name = name_str;
321 index_cdata(tinfo, ctrl, ptr->children, recWord);
323 recWord->index_type = prev_type; /* restore it again */
329 static void index_record(struct filter_info *tinfo,struct recExtractCtrl *ctrl,
330 xmlNodePtr ptr, RecWord *recWord)
332 if (ptr && ptr->type == XML_ELEMENT_NODE && ptr->ns &&
333 !strcmp(ptr->ns->href, zebra_xslt_ns)
334 && !strcmp(ptr->name, "record"))
336 const char *type_str = "update";
337 const char *id_str = 0;
338 const char *rank_str = 0;
339 struct _xmlAttr *attr;
340 for (attr = ptr->properties; attr; attr = attr->next)
342 attr_content(attr, "type", &type_str);
343 attr_content(attr, "id", &id_str);
344 attr_content(attr, "rank", &rank_str);
347 sscanf(id_str, "%255s", ctrl->match_criteria);
350 ctrl->staticrank = atoi(rank_str);
351 yaz_log(YLOG_LOG, "rank=%d",ctrl->staticrank);
354 yaz_log(YLOG_LOG, "no rank");
358 index_node(tinfo, ctrl, ptr, recWord);
361 static int extract_doc(struct filter_info *tinfo, struct recExtractCtrl *p,
365 const char *params[10];
369 struct filter_schema *schema = lookup_schema(tinfo, zebra_xslt_ns);
372 set_param_str(params, "schema", zebra_xslt_ns, tinfo->odr);
374 (*p->init)(p, &recWord);
376 if (schema && schema->stylesheet_xsp)
380 xsltApplyStylesheet(schema->stylesheet_xsp,
382 if (p->flagShowRecords)
384 xmlDocDumpMemory(resDoc, &buf_out, &len_out);
385 fwrite(buf_out, len_out, 1, stdout);
388 root_ptr = xmlDocGetRootElement(resDoc);
390 index_record(tinfo, p, root_ptr, &recWord);
393 yaz_log(YLOG_WARN, "No root for index XML record."
394 " split_level=%s stylesheet=%s",
395 tinfo->split_level, schema->stylesheet);
399 xmlDocDumpMemory(doc, &buf_out, &len_out);
400 if (p->flagShowRecords)
401 fwrite(buf_out, len_out, 1, stdout);
402 (*p->setStoreData)(p, buf_out, len_out);
406 return RECCTRL_EXTRACT_OK;
409 static int extract_split(struct filter_info *tinfo, struct recExtractCtrl *p)
416 xmlFreeTextReader(tinfo->reader);
417 tinfo->reader = xmlReaderForIO(ioread_ex, ioclose_ex,
424 return RECCTRL_EXTRACT_ERROR_GENERIC;
426 if (tinfo->split_level)
427 split_depth = atoi(tinfo->split_level);
428 ret = xmlTextReaderRead(tinfo->reader);
430 int type = xmlTextReaderNodeType(tinfo->reader);
431 int depth = xmlTextReaderDepth(tinfo->reader);
432 if (split_depth == 0 ||
434 type == XML_READER_TYPE_ELEMENT && split_depth == depth))
436 xmlNodePtr ptr = xmlTextReaderExpand(tinfo->reader);
437 xmlNodePtr ptr2 = xmlCopyNode(ptr, 1);
438 xmlDocPtr doc = xmlNewDoc("1.0");
440 xmlDocSetRootElement(doc, ptr2);
442 return extract_doc(tinfo, p, doc);
444 ret = xmlTextReaderRead(tinfo->reader);
446 xmlFreeTextReader(tinfo->reader);
448 return RECCTRL_EXTRACT_EOF;
451 static int extract_full(struct filter_info *tinfo, struct recExtractCtrl *p)
453 if (p->first_record) /* only one record per stream */
455 xmlDocPtr doc = xmlReadIO(ioread_ex, ioclose_ex, p /* I/O handler */,
461 return RECCTRL_EXTRACT_ERROR_GENERIC;
463 return extract_doc(tinfo, p, doc);
466 return RECCTRL_EXTRACT_EOF;
469 static int filter_extract(void *clientData, struct recExtractCtrl *p)
471 struct filter_info *tinfo = clientData;
473 odr_reset(tinfo->odr);
475 if (tinfo->split_level == 0 && tinfo->split_path == 0)
476 return extract_full(tinfo, p);
479 return extract_split(tinfo, p);
483 static int ioread_ret(void *context, char *buffer, int len)
485 struct recRetrieveCtrl *p = context;
486 return (*p->readf)(p->fh, buffer, len);
489 static int ioclose_ret(void *context)
495 static const char *snippet_doc(struct recRetrieveCtrl *p, int text_mode,
498 const char *xml_doc_str;
500 WRBUF wrbuf = wrbuf_alloc();
501 zebra_snippets *res =
502 zebra_snippets_window(p->doc_snippet, p->hit_snippet, window_size);
503 zebra_snippet_word *w = zebra_snippets_list(res);
506 wrbuf_printf(wrbuf, "\'");
508 wrbuf_printf(wrbuf, "<snippet xmlns='%s'>\n", zebra_xslt_ns);
509 for (; w; w = w->next)
513 else if (ord != w->ord)
517 wrbuf_printf(wrbuf, "%s%s%s ",
520 w->match ? "*" : "");
523 wrbuf_printf(wrbuf, " <term ord='%d' seqno='" ZINT_FORMAT "' %s>",
525 (w->match ? "match='1'" : ""));
526 wrbuf_xmlputs(wrbuf, w->term);
527 wrbuf_printf(wrbuf, "</term>\n");
531 wrbuf_printf(wrbuf, "\'");
533 wrbuf_printf(wrbuf, "</snippet>\n");
535 xml_doc_str = odr_strdup(p->odr, wrbuf_buf(wrbuf));
537 zebra_snippets_destroy(res);
538 wrbuf_free(wrbuf, 1);
542 static int filter_retrieve (void *clientData, struct recRetrieveCtrl *p)
544 const char *esn = zebra_xslt_ns;
545 const char *params[10];
546 struct filter_info *tinfo = clientData;
549 struct filter_schema *schema;
550 int window_size = -1;
554 if (p->comp->which != Z_RecordComp_simple
555 || p->comp->u.simple->which != Z_ElementSetNames_generic)
557 p->diagnostic = YAZ_BIB1_PRESENT_COMP_SPEC_PARAMETER_UNSUPP;
560 esn = p->comp->u.simple->u.generic;
562 schema = lookup_schema(tinfo, esn);
566 YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
570 if (schema->include_snippet)
571 window_size = atoi(schema->include_snippet);
574 set_param_str(params, "schema", esn, p->odr);
576 set_param_str(params, "filename", p->fname, p->odr);
578 set_param_int(params, "score", p->score, p->odr);
579 set_param_int(params, "size", p->recordSize, p->odr);
580 set_param_int(params, "id", p->localno, p->odr);
582 if (window_size >= 0)
583 set_param_xml(params, "snippet", snippet_doc(p, 1, window_size),
585 doc = xmlReadIO(ioread_ret, ioclose_ret, p /* I/O handler */,
591 p->diagnostic = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
595 if (window_size >= 0)
597 xmlNodePtr node = xmlDocGetRootElement(doc);
598 const char *snippet_str = snippet_doc(p, 0, window_size);
599 xmlDocPtr snippet_doc = xmlParseMemory(snippet_str, strlen(snippet_str));
600 xmlAddChild(node, xmlDocGetRootElement(snippet_doc));
602 if (!schema->stylesheet_xsp)
606 resDoc = xsltApplyStylesheet(schema->stylesheet_xsp,
612 p->diagnostic = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
614 else if (p->input_format == VAL_NONE || p->input_format == VAL_TEXT_XML)
618 xmlDocDumpMemory(resDoc, &buf_out, &len_out);
620 p->output_format = VAL_TEXT_XML;
621 p->rec_len = len_out;
622 p->rec_buf = odr_malloc(p->odr, p->rec_len);
623 memcpy(p->rec_buf, buf_out, p->rec_len);
627 else if (p->output_format == VAL_SUTRS)
631 xmlDocDumpMemory(resDoc, &buf_out, &len_out);
633 p->output_format = VAL_SUTRS;
634 p->rec_len = len_out;
635 p->rec_buf = odr_malloc(p->odr, p->rec_len);
636 memcpy(p->rec_buf, buf_out, p->rec_len);
642 p->diagnostic = YAZ_BIB1_RECORD_SYNTAX_UNSUPP;
648 static struct recType filter_type_alvis = {
659 #ifdef IDZEBRA_STATIC_ALVIS