2 * Copyright (c) 1995-1997, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.19 1997-12-09 16:17:09 adam
8 * Fix bug regarding variants. Tags with prefix "var" was incorrectly
9 * interpreted as "start of variants". Now, only "var" indicates such
11 * Cleaned up data1_read_node so tag names and variant names are
12 * copied and not pointed to by the generated data1 tree. Data nodes
13 * still point to old buffer.
15 * Revision 1.18 1997/11/18 09:51:09 adam
16 * Removed element num_children from data1_node. Minor changes in
19 * Revision 1.17 1997/11/05 09:20:51 adam
22 * Revision 1.16 1997/09/17 12:10:37 adam
25 * Revision 1.15 1997/09/05 09:50:57 adam
26 * Removed global data1_tabpath - uses data1_get_tabpath() instead.
28 * Revision 1.14 1997/05/14 06:54:04 adam
31 * Revision 1.13 1996/10/29 13:35:38 adam
32 * Implemented data1_set_tabpath and data1_get_tabpath.
34 * Revision 1.12 1996/10/11 10:35:38 adam
35 * Fixed a bug that caused data1_read_node to core dump when no abstract
36 * syntax was defined in a "sgml"-record.
38 * Revision 1.11 1996/07/06 19:58:35 quinn
39 * System headerfiles gathered in yconfig
41 * Revision 1.10 1996/01/19 15:41:47 quinn
42 * Fixed uninitialized boolean.
44 * Revision 1.9 1996/01/17 14:52:47 adam
45 * Changed prototype for reader function parsed to data1_read_record.
47 * Revision 1.8 1995/12/15 16:20:41 quinn
48 * Added formatted text.
50 * Revision 1.7 1995/12/13 13:44:32 quinn
51 * Modified Data1-system to use nmem
53 * Revision 1.6 1995/12/12 16:37:08 quinn
54 * Added destroy element to data1_node.
56 * Revision 1.5 1995/12/11 15:22:37 quinn
57 * Added last_child field to the node.
58 * Rewrote schema-mapping.
60 * Revision 1.4 1995/11/13 09:27:36 quinn
61 * Fiddling with the variant stuff.
63 * Revision 1.3 1995/11/01 16:34:57 quinn
64 * Making data1 look for tables in data1_tabpath
66 * Revision 1.2 1995/11/01 13:54:48 quinn
69 * Revision 1.1 1995/11/01 11:56:09 quinn
70 * Added Retrieval (data management) functions en masse.
72 * Revision 1.14 1995/10/30 12:40:55 quinn
73 * Fixed a couple of bugs.
75 * Revision 1.13 1995/10/25 16:00:47 quinn
76 * USMARC support is now almost operational
78 * Revision 1.12 1995/10/16 14:02:55 quinn
79 * Changes to support element set names and espec1
81 * Revision 1.11 1995/10/13 16:05:08 quinn
82 * Adding Espec1-processing
84 * Revision 1.10 1995/10/11 14:53:44 quinn
87 * Revision 1.9 1995/10/06 16:56:50 quinn
88 * Fixed ranked result.
90 * Revision 1.8 1995/10/06 16:44:13 quinn
91 * Work on attribute set mapping, etc.
93 * Revision 1.7 1995/10/06 12:58:35 quinn
96 * Revision 1.6 1995/10/04 09:29:49 quinn
97 * Adjustments to support USGS test data
99 * Revision 1.5 1995/10/03 17:56:43 quinn
102 * Revision 1.4 1995/10/02 15:53:19 quinn
105 * Revision 1.3 1995/10/02 14:55:21 quinn
106 * *** empty log message ***
108 * Revision 1.2 1995/09/14 15:18:13 quinn
111 * Revision 1.1 1995/09/12 11:24:30 quinn
112 * Beginning to add code for structured records.
127 * get the tag which is the immediate parent of this node (this may mean
128 * traversing intermediate things like variants and stuff.
130 data1_node *get_parent_tag (data1_handle dh, data1_node *n)
132 for (; n && n->which != DATA1N_root; n = n->parent)
133 if (n->which == DATA1N_tag)
138 data1_node *data1_mk_node (data1_handle dh, NMEM m)
142 r = nmem_malloc(m, sizeof(*r));
143 r->next = r->child = r->last_child = r->parent = 0;
148 void data1_free_tree (data1_handle dh, data1_node *t)
150 data1_node *p = t->child, *pn;
155 data1_free_tree (dh, p);
162 char *data1_insert_string (data1_handle dh, data1_node *res,
163 NMEM m, const char *str)
165 int len = strlen(str);
167 if (len > DATA1_LOCALDATA-1)
168 return nmem_strdup (m, str);
171 strcpy (res->lbuf, str);
177 * Insert a tagged node into the record root as first child of the node at
178 * which should be root or tag itself). Returns pointer to the data node,
179 * which can then be modified.
181 data1_node *data1_insert_taggeddata(data1_handle dh, data1_node *root,
183 char *tagname, NMEM m)
185 data1_node *tagn = data1_mk_node (dh, m);
188 tagn->which = DATA1N_tag;
191 tagn->u.tag.node_selected = 0;
192 tagn->u.tag.make_variantlist = 0;
193 tagn->u.tag.no_data_requested = 0;
194 tagn->u.tag.get_bytes = -1;
195 if (!(tagn->u.tag.element =
196 data1_getelementbytagname (dh, root->u.root.absyn, 0, tagname)))
198 tagn->child = datn = data1_mk_node (dh, m);
201 datn->which = DATA1N_data;
202 datn->u.data.formatted_text = 0;
203 tagn->next = at->child;
210 * Ugh. Sometimes functions just grow and grow on you. This one reads a
211 * 'node' and its children.
213 data1_node *data1_read_node (data1_handle dh, char **buf,
214 data1_node *parent, int *line,
215 data1_absyn *absyn, NMEM m)
219 while (**buf && isspace(**buf))
228 if (**buf == '<') /* beginning of tag */
233 char *t = (*buf) + 1;
235 data1_element *elem = 0;
237 for (i = 0; *t && *t != '>' && !isspace(*t); t++)
238 if (i < (sizeof(tag)-1))
243 for (i = 0; *t && *t != '>'; t++)
244 if (i < (sizeof(args)-1))
247 if (*t != '>' && !isspace(*t))
249 logf(LOG_WARN, "d1: %d: Malformed tag", *line);
253 * if end-tag, see if we terminate parent. If so, consume and return.
258 if (parent && (!*(tag +1) ||
259 (parent->which == DATA1N_root &&
260 !strcmp(tag + 1,parent->u.root.type)) ||
261 (parent->which == DATA1N_tag &&
262 !strcmp(tag + 1, parent->u.tag.tag))))
266 if (!absyn) /* parent node - what are we? */
268 if (!(absyn = data1_get_absyn (dh, tag)))
270 logf(LOG_WARN, "Unable to acquire abstract syntax for '%s'",
274 res = data1_mk_node (dh, m);
275 res->which = DATA1N_root;
276 res->u.root.type = data1_insert_string (dh, res, m, tag);
277 res->u.root.absyn = absyn;
281 else if (!strcmp(tag, "var"))
283 char tclass[DATA1_MAX_SYMBOL], type[DATA1_MAX_SYMBOL];
288 if (sscanf(args, "%s %s %n", tclass, type, &val_offset) != 2)
290 logf(LOG_WARN, "Malformed variant triple at '%s'", tag);
294 data1_getvartypebyct(dh, parent->root->u.root.absyn->varset,
299 * If we're the first variant in this group, create a parent var,
300 * and insert it before the current variant.
302 if (parent->which != DATA1N_variant)
304 res = data1_mk_node (dh, m);
305 res->which = DATA1N_variant;
306 res->u.variant.type = 0;
307 res->u.variant.value = 0;
308 res->root = parent->root;
313 * now determine if one of our ancestor triples is of same type.
314 * If so, we break here. This will make the parser unwind until
315 * we become a sibling (alternate variant) to the aforementioned
316 * triple. It stinks that we re-parse these tags on every
317 * iteration of this. This is a function in need of a rewrite.
319 for (p = parent; p->which == DATA1N_variant; p = p->parent)
320 if (p->u.variant.type == tp)
322 res = data1_mk_node (dh, m);
323 res->which = DATA1N_variant;
324 res->root = parent->root;
325 res->u.variant.type = tp;
326 res->u.variant.value =
327 data1_insert_string (dh, res, m, args + val_offset);
331 else /* tag.. acquire our element in the abstract syntax */
333 data1_node *partag = get_parent_tag (dh, parent);
334 data1_element *e = 0;
337 if (parent->which == DATA1N_variant)
340 if (!(e = partag->u.tag.element))
341 localtag = 1; /* our parent is a local tag */
343 elem = data1_getelementbytagname(dh, absyn, e, tag);
344 res = data1_mk_node (dh, m);
345 res->which = DATA1N_tag;
346 res->u.tag.tag = data1_insert_string (dh, res, m, tag);
347 res->u.tag.element = elem;
348 res->u.tag.node_selected = 0;
349 res->u.tag.make_variantlist = 0;
350 res->u.tag.no_data_requested = 0;
351 res->u.tag.get_bytes = -1;
352 res->root = parent->root;
356 res->parent = parent;
361 while ((*pp = data1_read_node(dh, buf, res, line, absyn, m)))
363 res->last_child = *pp;
367 else /* != '<'... this is a body of text */
370 char *data = *buf, *pp = *buf;
372 if (!parent) /* abort if abstract syntax is undefined */
375 /* Determine length and remove newlines/extra blanks */
376 while (**buf && **buf != '<')
384 while (isspace(**buf))
388 *(pp++) = *((*buf)++);
391 while (isspace(data[len-1]))
393 res = data1_mk_node(dh, m);
394 res->parent = parent;
395 res->which = DATA1N_data;
396 res->u.data.what = DATA1I_text;
398 res->u.data.len = len;
399 res->u.data.data = data;
400 res->u.data.formatted_text = 0;
401 res->root = parent->root;
407 * Read a record in the native syntax.
409 data1_node *data1_read_record(data1_handle dh,
410 int (*rf)(void *, char *, size_t), void *fh,
414 char **buf = data1_get_read_buf (dh, &size);
420 *buf = xmalloc(*size = 4096);
424 if (rd + 4096 > *size && !(*buf =xrealloc(*buf, *size *= 2)))
426 if ((res = (*rf)(fh, *buf + rd, 4096)) <= 0)
431 return data1_read_node(dh, &bp, 0, &line, 0, m);