1 /* This file is part of the Zebra server.
2    Copyright (C) 2004-2013 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 #if HAVE_CONFIG_H
21 #include <config.h>
22 #endif
23 #include <stdio.h>
24 #include <assert.h>
25 #include <stdlib.h>
26 #include <string.h>
27 
28 #include <yaz/log.h>
29 #include <yaz/oid_db.h>
30 #include <idzebra/data1.h>
31 #include <idzebra/recctrl.h>
32 #include <zebra_xpath.h>
33 #include <d1_absyn.h>
34 
35 #define D1_MAX_NESTING  128
36 
37 struct data1_hash_table {
38     NMEM nmem;
39     int size;
40     struct data1_hash_entry **ar;
41 };
42 
43 struct data1_hash_entry {
44     void *clientData;
45     char *str;
46     struct data1_hash_entry *next;
47 };
48 
data1_hash_calc(struct data1_hash_table * ht,const char * str)49 unsigned data1_hash_calc(struct data1_hash_table *ht, const char *str)
50 {
51     unsigned v = 0;
52     assert(str);
53     while (*str)
54     {
55 	if (*str >= 'a' && *str <= 'z')
56 	    v = v*65509 + *str -'a'+10;
57 	else if (*str >= 'A' && *str <= 'Z')
58 	    v = v*65509 + *str -'A'+10;
59 	else if (*str >= '0' && *str <= '9')
60 	    v = v*65509 + *str -'0';
61 	str++;
62     }
63     return v % ht->size;
64 }
65 
data1_hash_open(int size,NMEM nmem)66 struct data1_hash_table *data1_hash_open(int size, NMEM nmem)
67 {
68     int i;
69     struct data1_hash_table *ht = nmem_malloc(nmem, sizeof(*ht));
70     ht->nmem = nmem;
71     ht->size = size;
72     if (ht->size <= 0)
73 	ht->size = 29;
74     ht->ar = nmem_malloc(nmem, sizeof(*ht->ar) * ht->size);
75     for (i = 0; i<ht->size; i++)
76 	ht->ar[i] = 0;
77     return ht;
78 }
79 
data1_hash_insert(struct data1_hash_table * ht,const char * str,void * clientData,int copy)80 void data1_hash_insert(struct data1_hash_table *ht, const char *str,
81 		       void *clientData, int copy)
82 {
83     char *dstr = copy ? nmem_strdup(ht->nmem, str) : (char*) str;
84     if (strchr(str, '?') || strchr(str, '.'))
85     {
86 	int i;
87 	for (i = 0; i<ht->size; i++)
88 	{
89 	    struct data1_hash_entry **he = &ht->ar[i];
90 	    for (; *he && strcmp(str, (*he)->str); he = &(*he)->next)
91 		;
92 	    if (!*he)
93 	    {
94 		*he = nmem_malloc(ht->nmem, sizeof(**he));
95 		(*he)->str = dstr;
96 		(*he)->next = 0;
97 	    }
98 	    (*he)->clientData = clientData;
99 	}
100     }
101     else
102     {
103 	struct data1_hash_entry **he = &ht->ar[data1_hash_calc(ht, str)];
104 	for (; *he && strcmp(str, (*he)->str); he = &(*he)->next)
105 	    ;
106 	if (!*he)
107 	{
108 	    *he = nmem_malloc(ht->nmem, sizeof(**he));
109 	    (*he)->str = dstr;
110 	    (*he)->next = 0;
111 	}
112 	(*he)->clientData = clientData;
113     }
114 }
115 
data1_hash_lookup(struct data1_hash_table * ht,const char * str)116 void *data1_hash_lookup(struct data1_hash_table *ht, const char *str)
117 {
118     struct data1_hash_entry **he = &ht->ar[data1_hash_calc(ht, str)];
119 
120     for (; *he && yaz_matchstr(str, (*he)->str); he = &(*he)->next)
121 	;
122     if (*he)
123 	return (*he)->clientData;
124     return 0;
125 }
126 
127 struct data1_systag {
128     char *name;
129     char *value;
130     struct data1_systag *next;
131 };
132 
133 struct data1_absyn_cache_info
134 {
135     char *name;
136     data1_absyn *absyn;
137     data1_absyn_cache next;
138 };
139 
140 struct data1_attset_cache_info
141 {
142     char *name;
143     data1_attset *attset;
144     data1_attset_cache next;
145 };
146 
data1_mk_element(data1_handle dh)147 data1_element *data1_mk_element(data1_handle dh)
148 {
149     data1_element *e = nmem_malloc(data1_nmem_get(dh), sizeof(*e));
150     e->name = 0;
151     e->tag = 0;
152     e->termlists = 0;
153     e->next = e->children = 0;
154     e->sub_name = 0;
155     e->hash = 0;
156     return e;
157 }
158 
data1_absyn_search(data1_handle dh,const char * name)159 data1_absyn *data1_absyn_search (data1_handle dh, const char *name)
160 {
161     data1_absyn_cache p = *data1_absyn_cache_get (dh);
162 
163     while (p)
164     {
165 	if (!yaz_matchstr (name, p->name))
166 	    return p->absyn;
167 	p = p->next;
168     }
169     return 0;
170 }
171 /* *ostrich*
172    We need to destroy DFAs, in xp_element (xelm) definitions
173    pop, 2002-12-13
174 */
175 
data1_absyn_destroy(data1_handle dh)176 void data1_absyn_destroy (data1_handle dh)
177 {
178     data1_absyn_cache p = *data1_absyn_cache_get (dh);
179 
180     while (p)
181     {
182         data1_absyn *abs = p->absyn;
183 	if (abs)
184 	{
185 	    data1_xpelement *xpe = abs->xp_elements;
186 	    while (xpe) {
187 		yaz_log (YLOG_DEBUG,"Destroy xp element %s",xpe->xpath_expr);
188 		if (xpe->dfa)
189                     dfa_delete (&xpe->dfa);
190 		xpe = xpe->next;
191 	    }
192 	}
193         p = p->next;
194     }
195 }
196 
197 
data1_absyn_trav(data1_handle dh,void * handle,void (* fh)(data1_handle dh,void * h,data1_absyn * a))198 void data1_absyn_trav (data1_handle dh, void *handle,
199 		       void (*fh)(data1_handle dh, void *h, data1_absyn *a))
200 {
201     data1_absyn_cache p = *data1_absyn_cache_get (dh);
202 
203     while (p)
204     {
205 	(*fh)(dh, handle, p->absyn);
206 	p = p->next;
207     }
208 }
209 
210 static data1_absyn *data1_read_absyn(data1_handle dh, const char *file,
211                                      enum DATA1_XPATH_INDEXING en);
212 
data1_absyn_add(data1_handle dh,const char * name,enum DATA1_XPATH_INDEXING en)213 static data1_absyn *data1_absyn_add(data1_handle dh, const char *name,
214                                     enum DATA1_XPATH_INDEXING en)
215 {
216     char fname[512];
217     NMEM mem = data1_nmem_get (dh);
218 
219     data1_absyn_cache p = (data1_absyn_cache)nmem_malloc (mem, sizeof(*p));
220     data1_absyn_cache *pp = data1_absyn_cache_get (dh);
221 
222     sprintf(fname, "%.500s.abs", name);
223     p->absyn = data1_read_absyn(dh, fname, en);
224     p->name = nmem_strdup(mem, name);
225     p->next = *pp;
226     *pp = p;
227     return p->absyn;
228 }
229 
data1_get_absyn(data1_handle dh,const char * name,enum DATA1_XPATH_INDEXING en)230 data1_absyn *data1_get_absyn (data1_handle dh, const char *name,
231                               enum DATA1_XPATH_INDEXING en)
232 {
233     data1_absyn *absyn;
234 
235     if (!(absyn = data1_absyn_search (dh, name)))
236 	absyn = data1_absyn_add (dh, name, en);
237     return absyn;
238 }
239 
data1_attset_search_name(data1_handle dh,const char * name)240 data1_attset *data1_attset_search_name (data1_handle dh, const char *name)
241 {
242     data1_attset_cache p = *data1_attset_cache_get (dh);
243 
244     while (p)
245     {
246 	if (!yaz_matchstr (name, p->name))
247 	    return p->attset;
248 	p = p->next;
249     }
250     return 0;
251 }
252 
data1_attset_search_id(data1_handle dh,const Odr_oid * oid)253 data1_attset *data1_attset_search_id(data1_handle dh, const Odr_oid *oid)
254 {
255     data1_attset_cache p = *data1_attset_cache_get (dh);
256 
257     while (p)
258     {
259 	if (p->attset->oid && !oid_oidcmp(oid, p->attset->oid))
260 	    return p->attset;
261 	p = p->next;
262     }
263     return 0;
264 }
265 
data1_attset_add(data1_handle dh,const char * name)266 data1_attset *data1_attset_add (data1_handle dh, const char *name)
267 {
268     NMEM mem = data1_nmem_get (dh);
269     data1_attset *attset;
270 
271     attset = data1_read_attset (dh, name);
272     if (!attset)
273 	yaz_log (YLOG_WARN|YLOG_ERRNO, "Couldn't load attribute set %s", name);
274     else
275     {
276 	data1_attset_cache p = (data1_attset_cache)
277 	    nmem_malloc (mem, sizeof(*p));
278 	data1_attset_cache *pp = data1_attset_cache_get (dh);
279 
280 	attset->name = p->name = nmem_strdup(mem, name);
281 	p->attset = attset;
282 	p->next = *pp;
283 	*pp = p;
284     }
285     return attset;
286 }
287 
data1_get_attset(data1_handle dh,const char * name)288 data1_attset *data1_get_attset (data1_handle dh, const char *name)
289 {
290     data1_attset *attset;
291 
292     if (!(attset = data1_attset_search_name (dh, name)))
293 	attset = data1_attset_add (dh, name);
294     return attset;
295 }
296 
data1_getesetbyname(data1_handle dh,data1_absyn * a,const char * name)297 data1_esetname *data1_getesetbyname(data1_handle dh, data1_absyn *a,
298 				    const char *name)
299 {
300     data1_esetname *r;
301 
302     for (r = a->esetnames; r; r = r->next)
303 	if (!data1_matchstr(r->name, name))
304 	    return r;
305     return 0;
306 }
307 
308 /* we have multiple versions of data1_getelementbyname */
309 #define DATA1_GETELEMENTBYTAGNAME_VERSION 1
310 
data1_getelementbytagname(data1_handle dh,data1_absyn * abs,data1_element * parent,const char * tagname)311 data1_element *data1_getelementbytagname (data1_handle dh, data1_absyn *abs,
312 					  data1_element *parent,
313 					  const char *tagname)
314 {
315     data1_element *r;
316     struct data1_hash_table *ht;
317 
318     /* It's now possible to have a data1 tree with no abstract syntax */
319     if ( !abs )
320         return 0;
321 
322     if (!parent)
323         r = abs->main_elements;
324     else
325 	r = parent->children;
326 
327 #if DATA1_GETELEMENTBYTAGNAME_VERSION==1
328     /* using hash search */
329     if (!r)
330 	return 0;
331 
332     ht = r->hash;
333     if (!ht)
334     {
335         /* build hash table (the first time) */
336 	ht = r->hash = data1_hash_open(29, data1_nmem_get(dh));
337 	for (; r; r = r->next)
338 	{
339 	    data1_name *n;
340 
341 	    for (n = r->tag->names; n; n = n->next)
342 		data1_hash_insert(ht, n->name, r, 0);
343 	}
344     }
345     return data1_hash_lookup(ht, tagname);
346 #else
347     /* using linear search */
348     for (; r; r = r->next)
349     {
350 	data1_name *n;
351 
352 	for (n = r->tag->names; n; n = n->next)
353 	    if (!data1_matchstr(tagname, n->name))
354 		return r;
355     }
356     return 0;
357 #endif
358 }
359 
data1_getelementbyname(data1_handle dh,data1_absyn * absyn,const char * name)360 data1_element *data1_getelementbyname (data1_handle dh, data1_absyn *absyn,
361 				       const char *name)
362 {
363     data1_element *r;
364 
365     /* It's now possible to have a data1 tree with no abstract syntax */
366     if ( !absyn )
367         return 0;
368     for (r = absyn->main_elements; r; r = r->next)
369 	if (!data1_matchstr(r->name, name))
370 	    return r;
371     return 0;
372 }
373 
374 
fix_element_ref(data1_handle dh,data1_absyn * absyn,data1_element * e)375 void fix_element_ref (data1_handle dh, data1_absyn *absyn, data1_element *e)
376 {
377     /* It's now possible to have a data1 tree with no abstract syntax */
378     if ( !absyn )
379         return;
380 
381     for (; e; e = e->next)
382     {
383 	if (!e->sub_name)
384 	{
385 	    if (e->children)
386 		fix_element_ref (dh, absyn, e->children);
387 	}
388 	else
389 	{
390 	    data1_sub_elements *sub_e = absyn->sub_elements;
391 	    while (sub_e && strcmp (e->sub_name, sub_e->name))
392 		sub_e = sub_e->next;
393 	    if (sub_e)
394 		e->children = sub_e->elements;
395 	    else
396 		yaz_log (YLOG_WARN, "Unresolved reference to sub-elements %s",
397 		      e->sub_name);
398 	}
399     }
400 }
401 /* *ostrich*
402 
403    New function, a bit dummy now... I've seen it in zrpn.c... We should build
404    more clever regexps...
405 
406 
407       //a    ->    ^a/.*$
408       //a/b  ->    ^b/a/.*$
409       /a     ->    ^a/$
410       /a/b   ->    ^b/a/$
411 
412       /      ->    none
413 
414    pop, 2002-12-13
415 
416    Now [] predicates are supported
417 
418    pop, 2003-01-17
419 
420  */
421 
mk_xpath_regexp(data1_handle dh,const char * expr)422 static const char * mk_xpath_regexp (data1_handle dh, const char *expr)
423 {
424     const char *p = expr;
425     int abs = 1;
426     int e = 0;
427     char *stack[32];
428     char *res_p, *res = 0;
429     size_t res_size = 1;
430 
431     if (*p != '/')
432 	return ("");
433     p++;
434     if (*p == '/')
435     {
436 	abs =0;
437 	p++;
438     }
439     while (*p)
440     {
441 	int is_predicate = 0;
442 	char *s;
443 	int i, j;
444         for (i = 0; *p && !strchr("/",*p); i++, p++)
445 	    ;
446 	res_size += (i+3); /* we'll add / between later .. */
447         stack[e] = (char *) nmem_malloc(data1_nmem_get(dh), i+1);
448 	s = stack[e];
449 	for (j = 0; j < i; j++)
450 	{
451 	    const char *pp = p-i+j;
452 	    if (*pp == '[')
453 		is_predicate=1;
454 	    else if (*pp == ']')
455 		is_predicate=0;
456 	    else
457 	    {
458 		if (!is_predicate) {
459 		    if (*pp == '*')
460 			*s++ = '.';
461 		    *s++ = *pp;
462 		}
463 	    }
464 	}
465 	*s = 0;
466         e++;
467         if (*p)
468 	    p++;
469     }
470     res_p = res = nmem_malloc(data1_nmem_get(dh), res_size + 10);
471 
472     if (stack[e-1][0] == '@')  /* path/@attr spec (leaf is attribute) */
473 	strcpy(res_p, "/");
474     else
475 	strcpy(res_p, "[^@]*/");  /* path .. (index all cdata below it) */
476     res_p = res_p + strlen(res_p);
477     while (--e >= 0) {
478 	sprintf(res_p, "%s/", stack[e]);
479 	res_p += strlen(stack[e]) + 1;
480     }
481     if (!abs)
482     {
483 	sprintf(res_p, ".*");
484 	res_p += 2;
485     }
486     sprintf (res_p, "$");
487     res_p++;
488     yaz_log(YLOG_DEBUG, "Got regexp: %s", res);
489     return res;
490 }
491 
parse_termlists(data1_handle dh,data1_termlist *** tpp,char * cp,const char * file,int lineno,const char * element_name,data1_absyn * res,int xpelement,data1_attset * attset)492 static int parse_termlists(data1_handle dh, data1_termlist ***tpp,
493                            char *cp, const char *file, int lineno,
494                            const char *element_name, data1_absyn *res,
495                            int xpelement,
496                            data1_attset *attset)
497 {
498     data1_termlist **tp = *tpp;
499     while(1)
500     {
501 	char attname[512], structure[512];
502 	char *source;
503 	int r, i;
504 	int level = 0;
505 	structure[0] = '\0';
506 	for (i = 0; cp[i] && i<sizeof(attname)-1; i++)
507 	    if (strchr(":,", cp[i]))
508 		break;
509 	    else
510 		attname[i] = cp[i];
511 	if (i == 0)
512 	{
513 	    if (*cp)
514 		yaz_log(YLOG_WARN,
515 			"%s:%d: Syntax error in termlistspec '%s'",
516 			file, lineno, cp);
517 	    break;
518 	}
519 	attname[i] = '\0';
520 	r = 1;
521 	cp += i;
522 	if (*cp == ':')
523 	    cp++;
524 
525 	for (i = 0; cp[i] && i<sizeof(structure)-1; i++)
526 	    if (level == 0 && strchr(",", cp[i]))
527 		break;
528 	    else
529 	    {
530 		structure[i] = cp[i];
531 		if (cp[i] == '(')
532 		    level++;
533 		else if (cp[i] == ')')
534 		    level--;
535 	    }
536 	structure[i] = '\0';
537 	if (i)
538 	    r = 2;
539 	cp += i;
540 	if (*cp)
541 	    cp++;  /* skip , */
542 
543 	*tp = (data1_termlist *)
544 	    nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
545 	(*tp)->next = 0;
546 
547         if (*attname == '!')
548         {
549             if (!xpelement && element_name)
550                 strcpy(attname, element_name);
551             else if (xpelement)
552                 strcpy(attname, ZEBRA_XPATH_CDATA);
553         }
554         if (attset)
555         {
556             if (!data1_getattbyname(dh, attset, attname))
557             {
558                 yaz_log(YLOG_WARN, "Index '%s' not found in attset(s)",
559                         attname);
560             }
561         }
562 
563         (*tp)->index_name = nmem_strdup(data1_nmem_get(dh), attname);
564         assert (*(*tp)->index_name != '!');
565 	if (r == 2 && (source = strchr(structure, ':')))
566 	    *source++ = '\0';   /* cut off structure .. */
567 	else
568 	    source = "data";    /* ok: default is leaf data */
569 	(*tp)->source = (char *)
570 	    nmem_strdup (data1_nmem_get (dh), source);
571 
572 	if (r < 2) /* is the structure qualified? */
573 	    (*tp)->structure = "w";
574 	else
575 	    (*tp)->structure = (char *)
576 		nmem_strdup (data1_nmem_get (dh), structure);
577 	tp = &(*tp)->next;
578     }
579 
580     *tpp = tp;
581     return 0;
582 }
583 
584 /* quinn
585  * Converts a 'melm' field[$subfield] pattern to a simple xpath
586  */
melm2xpath(char * melm,char * buf)587 static int melm2xpath(char *melm, char *buf)
588 {
589     char *dollar;
590     char *field = melm;
591     char *subfield;
592     char *fieldtype;
593     if ((dollar = strchr(melm, '$'))) {
594 	*dollar = '\0';
595 	subfield = ++dollar;
596     } else
597 	subfield = "";
598     if (field[0] == '0' && field[1] == '0')
599 	fieldtype = "controlfield";
600     else
601 	fieldtype = "datafield";
602     sprintf(buf, "/*/%s[@tag=\"%s\"]", fieldtype, field);
603     if (*subfield)
604 	sprintf(buf + strlen(buf), "/subfield[@code=\"%s\"]", subfield);
605     else if (field[0] != '0' || field[1] != '0')
606 	strcat(buf, "/subfield");
607     yaz_log(YLOG_DEBUG, "Created xpath: '%s'", buf);
608     return 0;
609 }
610 
data1_systag_lookup(data1_absyn * absyn,const char * tag,const char * default_value)611 const char *data1_systag_lookup(data1_absyn *absyn, const char *tag,
612                                 const char *default_value)
613 {
614     struct data1_systag *p = absyn->systags;
615     for (; p; p = p->next)
616         if (!strcmp(p->name, tag))
617             return p->value;
618     return default_value;
619 }
620 
621 #define l_isspace(c) ((c) == '\t' || (c) == ' ' || (c) == '\n' || (c) == '\r')
622 
read_absyn_line(FILE * f,int * lineno,char * line,int len,char * argv[],int num)623 int read_absyn_line(FILE *f, int *lineno, char *line, int len,
624 		    char *argv[], int num)
625 {
626     char *p;
627     int argc;
628     int quoted = 0;
629 
630     while ((p = fgets(line, len, f)))
631     {
632 	(*lineno)++;
633 	while (*p && l_isspace(*p))
634 	    p++;
635 	if (*p && *p != '#')
636 	    break;
637     }
638     if (!p)
639 	return 0;
640 
641     for (argc = 0; *p ; argc++)
642     {
643 	if (*p == '#')  /* trailing comment */
644 	    break;
645 	argv[argc] = p;
646 	while (*p && !(l_isspace(*p) && !quoted)) {
647   	  if (*p =='"') quoted = 1 - quoted;
648   	  if (*p =='[') quoted = 1;
649   	  if (*p ==']') quoted = 0;
650 	  p++;
651 	}
652 	if (*p)
653 	{
654 	    *(p++) = '\0';
655 	    while (*p && l_isspace(*p))
656 		p++;
657 	}
658     }
659     return argc;
660 }
661 
data1_absyn_getmarctab(data1_handle dh,data1_node * root)662 data1_marctab *data1_absyn_getmarctab(data1_handle dh, data1_node *root)
663 {
664     if (root->u.root.absyn)
665         return root->u.root.absyn->marc;
666     return 0;
667 }
668 
data1_absyn_getelements(data1_handle dh,data1_node * root)669 data1_element *data1_absyn_getelements(data1_handle dh,
670                                        data1_node *root)
671 {
672     if (root->u.root.absyn)
673         return root->u.root.absyn->main_elements;
674     return 0;
675 }
676 
data1_read_absyn(data1_handle dh,const char * file,enum DATA1_XPATH_INDEXING default_xpath)677 static data1_absyn *data1_read_absyn(data1_handle dh, const char *file,
678                                      enum DATA1_XPATH_INDEXING default_xpath)
679 {
680     data1_sub_elements *cur_elements = NULL;
681     data1_xpelement **cur_xpelement = NULL;
682     data1_attset *attset_list = data1_empty_attset(dh);
683     data1_attset_child **attset_childp = &attset_list->children;
684 
685     data1_absyn *res = 0;
686     FILE *f;
687     data1_element **ppl[D1_MAX_NESTING];
688     data1_esetname **esetpp;
689     data1_maptab **maptabp;
690     data1_marctab **marcp;
691     data1_termlist *all = 0;
692     data1_tagset **tagset_childp;
693     struct data1_systag **systagsp;
694     int level = 0;
695     int lineno = 0;
696     int argc;
697     char *argv[50], line[512];
698 
699     f = data1_path_fopen(dh, file, "r");
700 
701     res = (data1_absyn *) nmem_malloc(data1_nmem_get(dh), sizeof(*res));
702     res->name = 0;
703     res->oid = 0;
704     res->tagset = 0;
705     res->encoding = 0;
706     res->xpath_indexing =
707         (f ? DATA1_XPATH_INDEXING_DISABLE : default_xpath);
708     res->systags = 0;
709     systagsp = &res->systags;
710     tagset_childp = &res->tagset;
711 
712     res->varset = 0;
713     res->esetnames = 0;
714     esetpp = &res->esetnames;
715     res->maptabs = 0;
716     maptabp = &res->maptabs;
717     res->marc = 0;
718     marcp = &res->marc;
719     res->sub_elements = NULL;
720     res->main_elements = NULL;
721     res->xp_elements = NULL;
722     cur_xpelement = &res->xp_elements;
723 
724     while (f && (argc = read_absyn_line(f, &lineno, line, 512, argv, 50)))
725     {
726 	char *cmd = *argv;
727 	if (!strcmp(cmd, "elm") || !strcmp(cmd, "element"))
728 	{
729 	    data1_element *new_element;
730 	    int i;
731 	    char *p, *sub_p, *path, *name, *termlists;
732 	    int type, value;
733 	    data1_termlist **tp;
734 
735 	    if (argc < 4)
736 	    {
737 		yaz_log(YLOG_WARN, "%s:%d: Bad # of args to elm", file, lineno);
738 		continue;
739 	    }
740 	    path = argv[1];
741 	    name = argv[2];
742 	    termlists = argv[3];
743 
744 	    if (!cur_elements)
745 	    {
746                 cur_elements = (data1_sub_elements *)
747 		    nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
748 	        cur_elements->next = res->sub_elements;
749 		cur_elements->elements = NULL;
750 		cur_elements->name = "main";
751 		res->sub_elements = cur_elements;
752 
753 		level = 0;
754     		ppl[level] = &cur_elements->elements;
755             }
756 	    p = path;
757 	    for (i = 1;; i++)
758 	    {
759 		char *e;
760 
761 		if ((e = strchr(p, '/')))
762 		    p = e+1;
763 		else
764 		    break;
765 	    }
766 	    if (i > level+1)
767 	    {
768 		yaz_log(YLOG_WARN, "%s:%d: Bad level increase", file, lineno);
769 		fclose(f);
770 		return 0;
771 	    }
772 	    level = i;
773 	    new_element = *ppl[level-1] = data1_mk_element(dh);
774 
775 	    tp = &new_element->termlists;
776 	    ppl[level-1] = &new_element->next;
777 	    ppl[level] = &new_element->children;
778 
779 	    /* consider subtree (if any) ... */
780 	    if ((sub_p = strchr (p, ':')) && sub_p[1])
781 	    {
782 		*sub_p++ = '\0';
783 		new_element->sub_name =
784 		    nmem_strdup (data1_nmem_get(dh), sub_p);
785 	    }
786 	    /* well-defined tag */
787 	    if (sscanf(p, "(%d,%d)", &type, &value) == 2)
788 	    {
789 		if (!res->tagset)
790 		{
791 		    yaz_log(YLOG_WARN, "%s:%d: No tagset loaded", file, lineno);
792 		    fclose(f);
793 		    return 0;
794 		}
795 		if (!(new_element->tag = data1_gettagbynum (dh, res->tagset,
796 							    type, value)))
797 		{
798 		    yaz_log(YLOG_WARN, "%s:%d: Couldn't find tag %s in tagset",
799 			 file, lineno, p);
800 		    fclose(f);
801 		    return 0;
802 		}
803 	    }
804 	    /* private tag */
805 	    else if (*p)
806 	    {
807 		data1_tag *nt =
808 		    new_element->tag = (data1_tag *)
809 		    nmem_malloc(data1_nmem_get (dh),
810 				sizeof(*new_element->tag));
811 		nt->which = DATA1T_string;
812 		nt->value.string = nmem_strdup(data1_nmem_get (dh), p);
813 		nt->names = (data1_name *)
814 		    nmem_malloc(data1_nmem_get(dh),
815 				sizeof(*new_element->tag->names));
816 		nt->names->name = nt->value.string;
817 		nt->names->next = 0;
818 		nt->kind = DATA1K_string;
819 		nt->next = 0;
820 		nt->tagset = 0;
821 	    }
822 	    else
823 	    {
824 		yaz_log(YLOG_WARN, "%s:%d: Bad element", file, lineno);
825 		fclose(f);
826 		return 0;
827 	    }
828 	    /* parse termList definitions */
829 	    p = termlists;
830 	    if (*p != '-')
831 	    {
832 		if (parse_termlists (dh, &tp, p, file, lineno, name, res, 0,
833                                      attset_list))
834 		{
835 		    fclose (f);
836 		    return 0;
837 		}
838 	        *tp = all; /* append any ALL entries to the list */
839 	    }
840 	    new_element->name = nmem_strdup(data1_nmem_get (dh), name);
841 	}
842 	/* *ostrich*
843 	   New code to support xelm directive
844 	   for each xelm a dfa is built. xelms are stored in res->xp_elements
845 
846 	   maybe we should use a simple sscanf instead of dfa?
847 
848 	   pop, 2002-12-13
849 
850 	   Now [] predicates are supported. regexps and xpath structure is
851 	   a bit redundant, however it's comfortable later...
852 
853 	   pop, 2003-01-17
854 	*/
855 
856 	else if (!strcmp(cmd, "xelm") || !strcmp(cmd, "melm")) {
857 
858 	    int i;
859 	    char *p, *xpath_expr, *termlists;
860 	    const char *regexp;
861 	    struct DFA *dfa = 0;
862 	    data1_termlist **tp;
863 	    char melm_xpath[128];
864             data1_xpelement *xp_ele = 0;
865             data1_xpelement *last_match = 0;
866 
867 	    if (argc != 3)
868 	    {
869 		yaz_log(YLOG_WARN, "%s:%d: Bad # of args to %s",
870                         file, lineno, cmd);
871 		continue;
872 	    }
873 
874 	    if (!strcmp(cmd, "melm")) {
875 		if (melm2xpath(argv[1], melm_xpath) < 0)
876 		    continue;
877 		xpath_expr = melm_xpath;
878 	    } else {
879 		xpath_expr = argv[1];
880 	    }
881 	    termlists = argv[2];
882 	    regexp = mk_xpath_regexp(dh, xpath_expr);
883 
884 #if OPTIMIZE_MELM
885             /* get last of existing regulars with same regexp */
886             for (xp_ele = res->xp_elements; xp_ele; xp_ele = xp_ele->next)
887                 if (!strcmp(xp_ele->regexp, regexp))
888                     last_match = xp_ele;
889 #endif
890             if (!last_match)
891             {
892                 /* new regular expression . Parse + generate */
893                 const char *regexp_ptr = regexp;
894 
895                 dfa = dfa_init();
896                 i = dfa_parse (dfa, &regexp_ptr);
897                 if (i || *regexp_ptr) {
898                     yaz_log(YLOG_WARN, "%s:%d: Bad xpath to xelm", file, lineno);
899                     dfa_delete (&dfa);
900                     continue;
901                 }
902             }
903             *cur_xpelement = (data1_xpelement *)
904                 nmem_malloc(data1_nmem_get(dh), sizeof(**cur_xpelement));
905             (*cur_xpelement)->next = 0;
906             (*cur_xpelement)->match_next = 0;
907             if (last_match)
908                 last_match->match_next = *cur_xpelement;
909 #if OPTIMIZE_MELM
910             (*cur_xpelement)->regexp = regexp;
911 #endif
912 	    (*cur_xpelement)->xpath_expr = nmem_strdup(data1_nmem_get (dh),
913                                                        xpath_expr);
914 
915             if (dfa)
916                 dfa_mkstate (dfa);
917             (*cur_xpelement)->dfa = dfa;
918 
919 #ifdef ENHANCED_XELM
920             (*cur_xpelement)->xpath_len =
921                 zebra_parse_xpath_str(
922                     xpath_expr,
923                     (*cur_xpelement)->xpath, XPATH_STEP_COUNT,
924                     data1_nmem_get(dh));
925 #endif
926 	    (*cur_xpelement)->termlists = 0;
927 	    tp = &(*cur_xpelement)->termlists;
928 
929 	    /* parse termList definitions */
930 	    p = termlists;
931 	    if (*p != '-')
932 	    {
933 		if (parse_termlists (dh, &tp, p, file, lineno,
934                                      xpath_expr, res, 1, attset_list))
935 		{
936 		    fclose (f);
937 		    return 0;
938 		}
939 	        *tp = all; /* append any ALL entries to the list */
940 	    }
941             cur_xpelement = &(*cur_xpelement)->next;
942 	}
943  	else if (!strcmp(cmd, "section"))
944 	{
945 	    char *name;
946 
947 	    if (argc < 2)
948 	    {
949 		yaz_log(YLOG_WARN, "%s:%d: Bad # of args to section",
950                         file, lineno);
951 		continue;
952 	    }
953 	    name = argv[1];
954 
955             cur_elements = (data1_sub_elements *)
956 		nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
957 	    cur_elements->next = res->sub_elements;
958 	    cur_elements->elements = NULL;
959 	    cur_elements->name = nmem_strdup (data1_nmem_get(dh), name);
960 	    res->sub_elements = cur_elements;
961 
962 	    level = 0;
963     	    ppl[level] = &cur_elements->elements;
964 	}
965         else if (!strcmp(cmd, "xpath"))
966         {
967             if (argc != 2)
968             {
969 		yaz_log(YLOG_WARN, "%s:%d: Bad # of args to 'xpath' directive",
970 		     file, lineno);
971 		continue;
972             }
973             if (!strcmp(argv[1], "enable"))
974                 res->xpath_indexing = DATA1_XPATH_INDEXING_ENABLE;
975             else if (!strcmp (argv[1], "disable"))
976                 res->xpath_indexing = DATA1_XPATH_INDEXING_DISABLE;
977             else
978             {
979 		yaz_log(YLOG_WARN, "%s:%d: Expecting disable/enable "
980                         "after 'xpath' directive", file, lineno);
981             }
982         }
983 	else if (!strcmp(cmd, "all"))
984 	{
985 	    data1_termlist **tp = &all;
986 	    if (all)
987 	    {
988 		yaz_log(YLOG_WARN, "%s:%d: Too many 'all' directives - ignored",
989 		     file, lineno);
990 		continue;
991 	    }
992 	    if (argc != 2)
993 	    {
994 		yaz_log(YLOG_WARN, "%s:%d: Bad # of args to 'all' directive",
995 		     file, lineno);
996 		continue;
997 	    }
998 	    if (parse_termlists (dh, &tp, argv[1], file, lineno, 0, res, 0,
999                                  attset_list))
1000 	    {
1001 		fclose (f);
1002 		return 0;
1003 	    }
1004 	}
1005 	else if (!strcmp(cmd, "name"))
1006 	{
1007 	    if (argc != 2)
1008 	    {
1009 		yaz_log(YLOG_WARN, "%s:%d: Bad # of args to name directive",
1010 		     file, lineno);
1011 		continue;
1012 	    }
1013 	    res->name = nmem_strdup(data1_nmem_get(dh), argv[1]);
1014 	}
1015 	else if (!strcmp(cmd, "reference"))
1016 	{
1017 	    char *name;
1018 
1019 	    if (argc != 2)
1020 	    {
1021 		yaz_log(YLOG_WARN, "%s:%d: Bad # of args to reference",
1022 		     file, lineno);
1023 		continue;
1024 	    }
1025 	    name = argv[1];
1026             res->oid = yaz_string_to_oid_nmem(yaz_oid_std(),
1027                                               CLASS_SCHEMA, name,
1028                                               data1_nmem_get(dh));
1029             if (!res->oid)
1030 	    {
1031 		yaz_log(YLOG_WARN, "%s:%d: Unknown tagset ref '%s'",
1032 		     file, lineno, name);
1033 		continue;
1034 	    }
1035 	}
1036 	else if (!strcmp(cmd, "attset"))
1037 	{
1038            char *name;
1039            data1_attset *attset;
1040 
1041            if (argc != 2)
1042            {
1043                yaz_log(YLOG_WARN, "%s:%d: Bad # of args to attset",
1044                     file, lineno);
1045                continue;
1046            }
1047            name = argv[1];
1048            if (!(attset = data1_get_attset (dh, name)))
1049            {
1050                yaz_log(YLOG_WARN, "%s:%d: Couldn't find attset  %s",
1051                        file, lineno, name);
1052                continue;
1053            }
1054            *attset_childp = (data1_attset_child *)
1055                nmem_malloc (data1_nmem_get(dh), sizeof(**attset_childp));
1056            (*attset_childp)->child = attset;
1057            (*attset_childp)->next = 0;
1058            attset_childp = &(*attset_childp)->next;
1059 	}
1060 	else if (!strcmp(cmd, "tagset"))
1061 	{
1062 	    char *name;
1063 	    int type = 0;
1064 	    if (argc < 2)
1065 	    {
1066 		yaz_log(YLOG_WARN, "%s:%d: Bad # of args to tagset",
1067 		     file, lineno);
1068 		continue;
1069 	    }
1070 	    name = argv[1];
1071 	    if (argc == 3)
1072 		type = atoi(argv[2]);
1073 	    *tagset_childp = data1_read_tagset (dh, name, type);
1074 	    if (!(*tagset_childp))
1075 	    {
1076 		yaz_log(YLOG_WARN, "%s:%d: Couldn't load tagset %s",
1077 		     file, lineno, name);
1078 		continue;
1079 	    }
1080 	    tagset_childp = &(*tagset_childp)->next;
1081 	}
1082 	else if (!strcmp(cmd, "varset"))
1083 	{
1084 	    char *name;
1085 
1086 	    if (argc != 2)
1087 	    {
1088 		yaz_log(YLOG_WARN, "%s:%d: Bad # of args in varset",
1089 		     file, lineno);
1090 		continue;
1091 	    }
1092 	    name = argv[1];
1093 	    if (!(res->varset = data1_read_varset (dh, name)))
1094 	    {
1095 		yaz_log(YLOG_WARN, "%s:%d: Couldn't load Varset %s",
1096 		     file, lineno, name);
1097 		continue;
1098 	    }
1099 	}
1100 	else if (!strcmp(cmd, "esetname"))
1101 	{
1102 	    char *name, *fname;
1103 
1104 	    if (argc != 3)
1105 	    {
1106 		yaz_log(YLOG_WARN, "%s:%d: Bad # of args in esetname",
1107 		     file, lineno);
1108 		continue;
1109 	    }
1110 	    name = argv[1];
1111 	    fname = argv[2];
1112 
1113 	    *esetpp = (data1_esetname *)
1114 		nmem_malloc(data1_nmem_get(dh), sizeof(**esetpp));
1115 	    (*esetpp)->name = nmem_strdup(data1_nmem_get(dh), name);
1116 	    (*esetpp)->next = 0;
1117 	    if (*fname == '@')
1118 		(*esetpp)->spec = 0;
1119 	    else if (!((*esetpp)->spec = data1_read_espec1 (dh, fname)))
1120 	    {
1121 		yaz_log(YLOG_WARN, "%s:%d: Espec-1 read failed for %s",
1122 		     file, lineno, fname);
1123 		continue;
1124 	    }
1125 	    esetpp = &(*esetpp)->next;
1126 	}
1127 	else if (!strcmp(cmd, "maptab"))
1128 	{
1129 	    char *name;
1130 
1131 	    if (argc != 2)
1132 	    {
1133 		yaz_log(YLOG_WARN, "%s:%d: Bad # of args for maptab",
1134                      file, lineno);
1135 		continue;
1136 	    }
1137 	    name = argv[1];
1138 	    if (!(*maptabp = data1_read_maptab (dh, name)))
1139 	    {
1140 		yaz_log(YLOG_WARN, "%s:%d: Couldn't load maptab %s",
1141                      file, lineno, name);
1142 		continue;
1143 	    }
1144 	    maptabp = &(*maptabp)->next;
1145 	}
1146 	else if (!strcmp(cmd, "marc"))
1147 	{
1148 	    char *name;
1149 
1150 	    if (argc != 2)
1151 	    {
1152 		yaz_log(YLOG_WARN, "%s:%d: Bad # or args for marc",
1153 		     file, lineno);
1154 		continue;
1155 	    }
1156 	    name = argv[1];
1157 	    if (!(*marcp = data1_read_marctab (dh, name)))
1158 	    {
1159 		yaz_log(YLOG_WARN, "%s:%d: Couldn't read marctab %s",
1160                      file, lineno, name);
1161 		continue;
1162 	    }
1163 	    marcp = &(*marcp)->next;
1164 	}
1165 	else if (!strcmp(cmd, "encoding"))
1166 	{
1167 	    if (argc != 2)
1168 	    {
1169 		yaz_log(YLOG_WARN, "%s:%d: Bad # or args for encoding",
1170 		     file, lineno);
1171 		continue;
1172 	    }
1173             res->encoding = nmem_strdup (data1_nmem_get(dh), argv[1]);
1174 	}
1175         else if (!strcmp(cmd, "systag"))
1176         {
1177             if (argc != 3)
1178             {
1179 		yaz_log(YLOG_WARN, "%s:%d: Bad # or args for systag",
1180 		     file, lineno);
1181 		continue;
1182             }
1183             *systagsp = nmem_malloc (data1_nmem_get(dh), sizeof(**systagsp));
1184 
1185             (*systagsp)->name = nmem_strdup(data1_nmem_get(dh), argv[1]);
1186             (*systagsp)->value = nmem_strdup(data1_nmem_get(dh), argv[2]);
1187             systagsp = &(*systagsp)->next;
1188         }
1189 	else
1190 	{
1191 	    yaz_log(YLOG_WARN, "%s:%d: Unknown directive '%s'", file,
1192                     lineno, cmd);
1193 	    continue;
1194 	}
1195     }
1196     if (f)
1197         fclose(f);
1198 
1199     for (cur_elements = res->sub_elements; cur_elements;
1200 	 cur_elements = cur_elements->next)
1201     {
1202 	if (!strcmp (cur_elements->name, "main"))
1203 	    res->main_elements = cur_elements->elements;
1204 	fix_element_ref (dh, res, cur_elements->elements);
1205     }
1206     *systagsp = 0;
1207     return res;
1208 }
1209 
1210 /*
1211  * Local variables:
1212  * c-basic-offset: 4
1213  * c-file-style: "Stroustrup"
1214  * indent-tabs-mode: nil
1215  * End:
1216  * vim: shiftwidth=4 tabstop=8 expandtab
1217  */
1218 
1219