2 * Copyright (c) 1995, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.6 1995-12-12 16:37:08 quinn
8 * Added destroy element to data1_node.
10 * Revision 1.5 1995/12/11 15:22:37 quinn
11 * Added last_child field to the node.
12 * Rewrote schema-mapping.
14 * Revision 1.4 1995/11/13 09:27:36 quinn
15 * Fiddling with the variant stuff.
17 * Revision 1.3 1995/11/01 16:34:57 quinn
18 * Making data1 look for tables in data1_tabpath
20 * Revision 1.2 1995/11/01 13:54:48 quinn
23 * Revision 1.1 1995/11/01 11:56:09 quinn
24 * Added Retrieval (data management) functions en masse.
26 * Revision 1.14 1995/10/30 12:40:55 quinn
27 * Fixed a couple of bugs.
29 * Revision 1.13 1995/10/25 16:00:47 quinn
30 * USMARC support is now almost operational
32 * Revision 1.12 1995/10/16 14:02:55 quinn
33 * Changes to support element set names and espec1
35 * Revision 1.11 1995/10/13 16:05:08 quinn
36 * Adding Espec1-processing
38 * Revision 1.10 1995/10/11 14:53:44 quinn
41 * Revision 1.9 1995/10/06 16:56:50 quinn
42 * Fixed ranked result.
44 * Revision 1.8 1995/10/06 16:44:13 quinn
45 * Work on attribute set mapping, etc.
47 * Revision 1.7 1995/10/06 12:58:35 quinn
50 * Revision 1.6 1995/10/04 09:29:49 quinn
51 * Adjustments to support USGS test data
53 * Revision 1.5 1995/10/03 17:56:43 quinn
56 * Revision 1.4 1995/10/02 15:53:19 quinn
59 * Revision 1.3 1995/10/02 14:55:21 quinn
60 * *** empty log message ***
62 * Revision 1.2 1995/09/14 15:18:13 quinn
65 * Revision 1.1 1995/09/12 11:24:30 quinn
66 * Beginning to add code for structured records.
79 char *data1_tabpath = 0; /* global path for tables */
81 void data1_set_tabpath(char *p)
82 { data1_tabpath = p; }
84 static data1_node *freelist = 0;
87 * get the tag which is the immediate parent of this node (this may mean
88 * traversing intermediate things like variants and stuff.
90 data1_node *get_parent_tag(data1_node *n)
92 for (; n && n->which != DATA1N_root; n = n->parent)
93 if (n->which == DATA1N_tag)
98 data1_node *data1_mk_node(void)
105 if (!(r = xmalloc(sizeof(*r))))
107 r->next = r->child = r->last_child = r->parent = 0;
113 static void fr_node(data1_node *n)
119 void data1_free_tree(data1_node *t)
121 data1_node *p = t->child, *pn;
135 * Insert a tagged node into the record root as first child of the node at
136 * which should be root or tag itself). Returns pointer to the data node,
137 * which can then be modified.
139 data1_node *data1_insert_taggeddata(data1_node *root, data1_node *at,
142 data1_node *tagn = data1_mk_node();
145 tagn->which = DATA1N_tag;
148 tagn->u.tag.node_selected = 0;
149 tagn->u.tag.make_variantlist = 0;
150 tagn->u.tag.no_data_requested = 0;
151 if (!(tagn->u.tag.element = data1_getelementbytagname(root->u.root.absyn,
157 tagn->child = datn = data1_mk_node();
158 tagn->num_children = 1;
161 datn->which = DATA1N_data;
162 tagn->next = at->child;
170 * Ugh. Sometimes functions just grow and grow on you. This one reads a
171 * 'node' and its children.
173 data1_node *data1_read_node(char **buf, data1_node *parent, int *line,
178 while (**buf && isspace(**buf))
187 if (**buf == '<') /* beginning of tag */
189 char *tag = (*buf) + 1;
193 data1_element *elem = 0;
195 for (; *t && *t != '>' && !isspace(*t); t++);
196 if (*t != '>' && !isspace(*t))
198 logf(LOG_WARN, "d1: %d: Malformed tag", *line);
201 if (isspace(*t)) /* the tag has arguments */
208 for (; *t && *t != '>'; t++);
209 if (*t != '>' && !isspace(*t))
211 logf(LOG_WARN, "d1: %d: Malformed tag", *line);
218 * if end-tag, see if we terminate parent. If so, consume and return.
226 if (!*(tag +1) || (parent->which == DATA1N_root && !strcmp(tag + 1,
227 parent->u.root.type)) ||
228 (parent->which == DATA1N_tag && !strcmp(tag + 1,
241 if (!absyn) /* parent node - what are we? */
243 if (!(absyn = data1_get_absyn(tag)))
245 logf(LOG_WARN, "Unable to acquire abstract syntax for '%s'",
249 res = data1_mk_node();
250 res->which = DATA1N_root;
251 res->u.root.type = tag;
252 res->u.root.absyn = absyn;
256 else if (!strncmp(tag, "var", 3))
258 char class[DATA1_MAX_SYMBOL], type[DATA1_MAX_SYMBOL];
263 if (sscanf(args, "%s %s %n", class, type, &val_offset) != 2)
265 logf(LOG_WARN, "Malformed variant triple at '%s'", tag);
268 if (!(tp = data1_getvartypebyct(parent->root->u.root.absyn->varset,
273 * If we're the first variant in this group, create a parent var,
274 * and insert it before the current variant.
276 if (parent->which != DATA1N_variant)
278 res = data1_mk_node();
279 res->which = DATA1N_variant;
280 res->u.variant.type = 0;
281 res->u.variant.value = 0;
282 res->root = parent->root;
288 * now determine if one of our ancestor triples is of same type.
289 * If so, we break here. This will make the parser unwind until
290 * we become a sibling (alternate variant) to the aforementioned
291 * triple. It stinks that we re-parse these tags on every
292 * iteration of this. This is a function in need of a rewrite.
294 for (p = parent; p->which == DATA1N_variant; p = p->parent)
295 if (p->u.variant.type == tp)
301 res = data1_mk_node();
302 res->which = DATA1N_variant;
303 res->root = parent->root;
304 res->u.variant.type = tp;
305 res->u.variant.value = args + val_offset;
309 else /* tag.. acquire our element in the abstract syntax */
311 data1_node *partag = get_parent_tag(parent);
312 data1_element *e = 0;
315 if (parent->which == DATA1N_variant)
321 if (!(e = partag->u.tag.element))
322 localtag = 1; /* our parent is a local tag */
325 if (!localtag && !(elem = data1_getelementbytagname(absyn,
326 e, tag)) && (data1_gettagbyname(absyn->tagset, tag)))
328 if (parent->which == DATA1N_root)
329 logf(LOG_WARN, "Tag '%s' used out of context", tag);
334 elem = data1_getelementbytagname(absyn, e, tag);
336 res = data1_mk_node();
337 res->which = DATA1N_tag;
338 res->u.tag.element = elem;
339 res->u.tag.tag = tag;
340 res->u.tag.node_selected = 0;
341 res->u.tag.make_variantlist = 0;
342 res->u.tag.no_data_requested = 0;
343 res->root = parent->root;
347 res->parent = parent;
348 res->num_children = 0;
354 while ((*pp = data1_read_node(buf, res, line, absyn)))
356 res->last_child = *pp;
361 else /* != '<'... this is a body of text */
364 char *data = *buf, *pp = *buf;
366 data1_node *partag = get_parent_tag(parent);
369 /* Determine length and remove newlines/extra blanks */
370 while (**buf && **buf != '<')
378 while (isspace(**buf))
382 *(pp++) = *((*buf)++);
385 while (isspace(data[len-1]))
387 res = data1_mk_node();
388 res->parent = parent;
389 res->which = DATA1N_data;
390 res->u.data.what = DATA1I_text;
391 res->u.data.len = len;
392 res->u.data.data = data;
393 res->root = parent->root;
396 * if the parent is structured, we'll insert a 'wellKnown' marker
397 * in front of the data.
400 if (partag->u.tag.element && partag->u.tag.element->tag->kind ==
403 data1_node *wk = mk_node();
404 static data1_element wk_element = { 0, 0, 0, 0, 0};
407 wk->root = partag->root;
408 wk->which = DATA1N_tag;
411 * get well-known tagdef if required.
413 if (!wk_element.tag && !(wk_element.tag =
414 data1_gettagbynum(wk->root->u.root.absyn->tagset, 1, 19)))
417 "Failed to initialize 'wellknown' tag from tagsetM");
420 wk->u.tag.element = &wk_element;
421 wk->child = partag->child;
423 wk->child->parent = wk;
432 * Read a record in the native syntax.
434 data1_node *data1_read_record(int (*rf)(int, char *, size_t), int fd)
436 static char *buf = 0;
442 if (!buf && !(buf = xmalloc(size = 4096)))
446 if (rd + 4096 > size && !(buf =xrealloc(buf, size *= 2)))
448 if ((res = (*rf)(fd, buf + rd, 4096)) <= 0)
453 return data1_read_node(&bp, 0, &line, 0);