2 * Copyright (c) 1995, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.9 1996-01-17 14:52:47 adam
8 * Changed prototype for reader function parsed to data1_read_record.
10 * Revision 1.8 1995/12/15 16:20:41 quinn
11 * Added formatted text.
13 * Revision 1.7 1995/12/13 13:44:32 quinn
14 * Modified Data1-system to use nmem
16 * Revision 1.6 1995/12/12 16:37:08 quinn
17 * Added destroy element to data1_node.
19 * Revision 1.5 1995/12/11 15:22:37 quinn
20 * Added last_child field to the node.
21 * Rewrote schema-mapping.
23 * Revision 1.4 1995/11/13 09:27:36 quinn
24 * Fiddling with the variant stuff.
26 * Revision 1.3 1995/11/01 16:34:57 quinn
27 * Making data1 look for tables in data1_tabpath
29 * Revision 1.2 1995/11/01 13:54:48 quinn
32 * Revision 1.1 1995/11/01 11:56:09 quinn
33 * Added Retrieval (data management) functions en masse.
35 * Revision 1.14 1995/10/30 12:40:55 quinn
36 * Fixed a couple of bugs.
38 * Revision 1.13 1995/10/25 16:00:47 quinn
39 * USMARC support is now almost operational
41 * Revision 1.12 1995/10/16 14:02:55 quinn
42 * Changes to support element set names and espec1
44 * Revision 1.11 1995/10/13 16:05:08 quinn
45 * Adding Espec1-processing
47 * Revision 1.10 1995/10/11 14:53:44 quinn
50 * Revision 1.9 1995/10/06 16:56:50 quinn
51 * Fixed ranked result.
53 * Revision 1.8 1995/10/06 16:44:13 quinn
54 * Work on attribute set mapping, etc.
56 * Revision 1.7 1995/10/06 12:58:35 quinn
59 * Revision 1.6 1995/10/04 09:29:49 quinn
60 * Adjustments to support USGS test data
62 * Revision 1.5 1995/10/03 17:56:43 quinn
65 * Revision 1.4 1995/10/02 15:53:19 quinn
68 * Revision 1.3 1995/10/02 14:55:21 quinn
69 * *** empty log message ***
71 * Revision 1.2 1995/09/14 15:18:13 quinn
74 * Revision 1.1 1995/09/12 11:24:30 quinn
75 * Beginning to add code for structured records.
88 char *data1_tabpath = 0; /* global path for tables */
90 void data1_set_tabpath(char *p)
91 { data1_tabpath = p; }
94 static data1_node *freelist = 0;
98 * get the tag which is the immediate parent of this node (this may mean
99 * traversing intermediate things like variants and stuff.
101 data1_node *get_parent_tag(data1_node *n)
103 for (; n && n->which != DATA1N_root; n = n->parent)
104 if (n->which == DATA1N_tag)
109 data1_node *data1_mk_node(NMEM m)
117 if (!(r = xmalloc(sizeof(*r))))
120 r = nmem_malloc(m, sizeof(*r));
122 r->next = r->child = r->last_child = r->parent = 0;
129 static void fr_node(data1_node *n)
136 void data1_free_tree(data1_node *t)
138 data1_node *p = t->child, *pn;
154 * Insert a tagged node into the record root as first child of the node at
155 * which should be root or tag itself). Returns pointer to the data node,
156 * which can then be modified.
158 data1_node *data1_insert_taggeddata(data1_node *root, data1_node *at,
159 char *tagname, NMEM m)
161 data1_node *tagn = data1_mk_node(m);
164 tagn->which = DATA1N_tag;
167 tagn->u.tag.node_selected = 0;
168 tagn->u.tag.make_variantlist = 0;
169 tagn->u.tag.no_data_requested = 0;
170 if (!(tagn->u.tag.element = data1_getelementbytagname(root->u.root.absyn,
173 tagn->child = datn = data1_mk_node(m);
174 tagn->num_children = 1;
177 datn->which = DATA1N_data;
178 tagn->next = at->child;
186 * Ugh. Sometimes functions just grow and grow on you. This one reads a
187 * 'node' and its children.
189 data1_node *data1_read_node(char **buf, data1_node *parent, int *line,
190 data1_absyn *absyn, NMEM m)
194 while (**buf && isspace(**buf))
203 if (**buf == '<') /* beginning of tag */
205 char *tag = (*buf) + 1;
209 data1_element *elem = 0;
211 for (; *t && *t != '>' && !isspace(*t); t++);
212 if (*t != '>' && !isspace(*t))
214 logf(LOG_WARN, "d1: %d: Malformed tag", *line);
217 if (isspace(*t)) /* the tag has arguments */
224 for (; *t && *t != '>'; t++);
225 if (*t != '>' && !isspace(*t))
227 logf(LOG_WARN, "d1: %d: Malformed tag", *line);
234 * if end-tag, see if we terminate parent. If so, consume and return.
242 if (!*(tag +1) || (parent->which == DATA1N_root && !strcmp(tag + 1,
243 parent->u.root.type)) ||
244 (parent->which == DATA1N_tag && !strcmp(tag + 1,
257 if (!absyn) /* parent node - what are we? */
259 if (!(absyn = data1_get_absyn(tag)))
261 logf(LOG_WARN, "Unable to acquire abstract syntax for '%s'",
265 res = data1_mk_node(m);
266 res->which = DATA1N_root;
267 res->u.root.type = tag;
268 res->u.root.absyn = absyn;
272 else if (!strncmp(tag, "var", 3))
274 char class[DATA1_MAX_SYMBOL], type[DATA1_MAX_SYMBOL];
279 if (sscanf(args, "%s %s %n", class, type, &val_offset) != 2)
281 logf(LOG_WARN, "Malformed variant triple at '%s'", tag);
284 if (!(tp = data1_getvartypebyct(parent->root->u.root.absyn->varset,
289 * If we're the first variant in this group, create a parent var,
290 * and insert it before the current variant.
292 if (parent->which != DATA1N_variant)
294 res = data1_mk_node(m);
295 res->which = DATA1N_variant;
296 res->u.variant.type = 0;
297 res->u.variant.value = 0;
298 res->root = parent->root;
304 * now determine if one of our ancestor triples is of same type.
305 * If so, we break here. This will make the parser unwind until
306 * we become a sibling (alternate variant) to the aforementioned
307 * triple. It stinks that we re-parse these tags on every
308 * iteration of this. This is a function in need of a rewrite.
310 for (p = parent; p->which == DATA1N_variant; p = p->parent)
311 if (p->u.variant.type == tp)
317 res = data1_mk_node(m);
318 res->which = DATA1N_variant;
319 res->root = parent->root;
320 res->u.variant.type = tp;
321 res->u.variant.value = args + val_offset;
325 else /* tag.. acquire our element in the abstract syntax */
327 data1_node *partag = get_parent_tag(parent);
328 data1_element *e = 0;
331 if (parent->which == DATA1N_variant)
337 if (!(e = partag->u.tag.element))
338 localtag = 1; /* our parent is a local tag */
341 if (!localtag && !(elem = data1_getelementbytagname(absyn,
342 e, tag)) && (data1_gettagbyname(absyn->tagset, tag)))
344 if (parent->which == DATA1N_root)
345 logf(LOG_WARN, "Tag '%s' used out of context", tag);
350 elem = data1_getelementbytagname(absyn, e, tag);
352 res = data1_mk_node(m);
353 res->which = DATA1N_tag;
354 res->u.tag.element = elem;
355 res->u.tag.tag = tag;
356 res->u.tag.node_selected = 0;
357 res->u.tag.make_variantlist = 0;
358 res->u.tag.no_data_requested = 0;
359 res->root = parent->root;
363 res->parent = parent;
364 res->num_children = 0;
370 while ((*pp = data1_read_node(buf, res, line, absyn, m)))
372 res->last_child = *pp;
377 else /* != '<'... this is a body of text */
380 char *data = *buf, *pp = *buf;
382 data1_node *partag = get_parent_tag(parent);
385 /* Determine length and remove newlines/extra blanks */
386 while (**buf && **buf != '<')
394 while (isspace(**buf))
398 *(pp++) = *((*buf)++);
401 while (isspace(data[len-1]))
403 res = data1_mk_node(m);
404 res->parent = parent;
405 res->which = DATA1N_data;
406 res->u.data.what = DATA1I_text;
407 res->u.data.len = len;
408 res->u.data.data = data;
409 res->u.data.formatted_text = 0;
410 res->root = parent->root;
416 * Read a record in the native syntax.
418 data1_node *data1_read_record(int (*rf)(void *, char *, size_t), void *fh,
421 static char *buf = 0;
427 if (!buf && !(buf = xmalloc(size = 4096)))
432 if (rd + 4096 > size && !(buf =xrealloc(buf, size *= 2)))
434 if ((res = (*rf)(fh, buf + rd, 4096)) <= 0)
439 return data1_read_node(&bp, 0, &line, 0, m);