2 * Copyright (c) 1995-1998, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.22 1998-03-05 08:15:32 adam
8 * Implemented data1_add_insert_taggeddata utility which is more flexible
9 * than data1_insert_taggeddata.
11 * Revision 1.21 1998/02/27 14:08:05 adam
12 * Added const to some char pointer arguments.
13 * Reworked data1_read_node so that it doesn't create a tree with
14 * pointers to original "SGML"-buffer.
16 * Revision 1.20 1998/02/11 11:53:35 adam
17 * Changed code so that it compiles as C++.
19 * Revision 1.19 1997/12/09 16:17:09 adam
20 * Fix bug regarding variants. Tags with prefix "var" was incorrectly
21 * interpreted as "start of variants". Now, only "var" indicates such
23 * Cleaned up data1_read_node so tag names and variant names are
24 * copied and not pointed to by the generated data1 tree. Data nodes
25 * still point to old buffer.
27 * Revision 1.18 1997/11/18 09:51:09 adam
28 * Removed element num_children from data1_node. Minor changes in
31 * Revision 1.17 1997/11/05 09:20:51 adam
34 * Revision 1.16 1997/09/17 12:10:37 adam
37 * Revision 1.15 1997/09/05 09:50:57 adam
38 * Removed global data1_tabpath - uses data1_get_tabpath() instead.
40 * Revision 1.14 1997/05/14 06:54:04 adam
43 * Revision 1.13 1996/10/29 13:35:38 adam
44 * Implemented data1_set_tabpath and data1_get_tabpath.
46 * Revision 1.12 1996/10/11 10:35:38 adam
47 * Fixed a bug that caused data1_read_node to core dump when no abstract
48 * syntax was defined in a "sgml"-record.
50 * Revision 1.11 1996/07/06 19:58:35 quinn
51 * System headerfiles gathered in yconfig
53 * Revision 1.10 1996/01/19 15:41:47 quinn
54 * Fixed uninitialized boolean.
56 * Revision 1.9 1996/01/17 14:52:47 adam
57 * Changed prototype for reader function parsed to data1_read_record.
59 * Revision 1.8 1995/12/15 16:20:41 quinn
60 * Added formatted text.
62 * Revision 1.7 1995/12/13 13:44:32 quinn
63 * Modified Data1-system to use nmem
65 * Revision 1.6 1995/12/12 16:37:08 quinn
66 * Added destroy element to data1_node.
68 * Revision 1.5 1995/12/11 15:22:37 quinn
69 * Added last_child field to the node.
70 * Rewrote schema-mapping.
72 * Revision 1.4 1995/11/13 09:27:36 quinn
73 * Fiddling with the variant stuff.
75 * Revision 1.3 1995/11/01 16:34:57 quinn
76 * Making data1 look for tables in data1_tabpath
78 * Revision 1.2 1995/11/01 13:54:48 quinn
81 * Revision 1.1 1995/11/01 11:56:09 quinn
82 * Added Retrieval (data management) functions en masse.
84 * Revision 1.14 1995/10/30 12:40:55 quinn
85 * Fixed a couple of bugs.
87 * Revision 1.13 1995/10/25 16:00:47 quinn
88 * USMARC support is now almost operational
90 * Revision 1.12 1995/10/16 14:02:55 quinn
91 * Changes to support element set names and espec1
93 * Revision 1.11 1995/10/13 16:05:08 quinn
94 * Adding Espec1-processing
96 * Revision 1.10 1995/10/11 14:53:44 quinn
99 * Revision 1.9 1995/10/06 16:56:50 quinn
100 * Fixed ranked result.
102 * Revision 1.8 1995/10/06 16:44:13 quinn
103 * Work on attribute set mapping, etc.
105 * Revision 1.7 1995/10/06 12:58:35 quinn
108 * Revision 1.6 1995/10/04 09:29:49 quinn
109 * Adjustments to support USGS test data
111 * Revision 1.5 1995/10/03 17:56:43 quinn
114 * Revision 1.4 1995/10/02 15:53:19 quinn
117 * Revision 1.3 1995/10/02 14:55:21 quinn
118 * *** empty log message ***
120 * Revision 1.2 1995/09/14 15:18:13 quinn
123 * Revision 1.1 1995/09/12 11:24:30 quinn
124 * Beginning to add code for structured records.
139 * get the tag which is the immediate parent of this node (this may mean
140 * traversing intermediate things like variants and stuff.
142 data1_node *get_parent_tag (data1_handle dh, data1_node *n)
144 for (; n && n->which != DATA1N_root; n = n->parent)
145 if (n->which == DATA1N_tag)
150 data1_node *data1_mk_node (data1_handle dh, NMEM m)
154 r = (data1_node *)nmem_malloc(m, sizeof(*r));
155 r->next = r->child = r->last_child = r->parent = 0;
160 void data1_free_tree (data1_handle dh, data1_node *t)
162 data1_node *p = t->child, *pn;
167 data1_free_tree (dh, p);
174 char *data1_insert_string (data1_handle dh, data1_node *res,
175 NMEM m, const char *str)
177 int len = strlen(str);
179 if (len >= DATA1_LOCALDATA)
180 return nmem_strdup (m, str);
183 strcpy (res->lbuf, str);
189 data1_node *data1_add_insert_taggeddata(data1_handle dh, data1_node *root,
191 const char *tagname, NMEM m,
192 int first_flag, int local_allowed)
194 data1_node *partag = get_parent_tag (dh, at);
195 data1_node *tagn = data1_mk_node (dh, m);
196 data1_element *e = NULL;
199 tagn->which = DATA1N_tag;
201 tagn->u.tag.tag = data1_insert_string (dh, tagn, m, tagname);
202 tagn->u.tag.node_selected = 0;
203 tagn->u.tag.make_variantlist = 0;
204 tagn->u.tag.no_data_requested = 0;
205 tagn->u.tag.get_bytes = -1;
208 e = partag->u.tag.element;
209 tagn->u.tag.element =
210 data1_getelementbytagname (dh, root->u.root.absyn, e, tagname);
211 if (!local_allowed && !tagn->u.tag.element)
213 tagn->last_child = tagn->child = datn = data1_mk_node (dh, m);
216 datn->which = DATA1N_data;
217 datn->u.data.formatted_text = 0;
222 tagn->next = at->child;
224 at->last_child = tagn;
233 assert (at->last_child);
234 at->last_child->next = tagn;
236 at->last_child = tagn;
241 data1_node *data1_add_taggeddata(data1_handle dh, data1_node *root,
243 const char *tagname, NMEM m)
245 return data1_add_insert_taggeddata (dh, root, at, tagname, m, 0, 1);
250 * Insert a tagged node into the record root as first child of the node at
251 * which should be root or tag itself). Returns pointer to the data node,
252 * which can then be modified.
254 data1_node *data1_insert_taggeddata(data1_handle dh, data1_node *root,
256 const char *tagname, NMEM m)
258 return data1_add_insert_taggeddata (dh, root, at, tagname, m, 1, 0);
262 * Ugh. Sometimes functions just grow and grow on you. This one reads a
263 * 'node' and its children.
265 data1_node *data1_read_node (data1_handle dh, const char **buf,
266 data1_node *parent, int *line,
267 data1_absyn *absyn, NMEM m)
271 while (**buf && isspace(**buf))
280 if (**buf == '<') /* beginning of tag */
285 const char *t = (*buf) + 1;
287 data1_element *elem = 0;
289 for (i = 0; *t && *t != '>' && !isspace(*t); t++)
290 if (i < (sizeof(tag)-1))
295 for (i = 0; *t && *t != '>'; t++)
296 if (i < (sizeof(args)-1))
299 if (*t != '>' && !isspace(*t))
301 logf(LOG_WARN, "d1: %d: Malformed tag", *line);
305 * if end-tag, see if we terminate parent. If so, consume and return.
310 if (parent && (!*(tag +1) ||
311 (parent->which == DATA1N_root &&
312 !strcmp(tag + 1,parent->u.root.type)) ||
313 (parent->which == DATA1N_tag &&
314 !strcmp(tag + 1, parent->u.tag.tag))))
318 if (!absyn) /* parent node - what are we? */
320 if (!(absyn = data1_get_absyn (dh, tag)))
322 logf(LOG_WARN, "Unable to acquire abstract syntax for '%s'",
326 res = data1_mk_node (dh, m);
327 res->which = DATA1N_root;
328 res->u.root.type = data1_insert_string (dh, res, m, tag);
329 res->u.root.absyn = absyn;
333 else if (!strcmp(tag, "var"))
335 char tclass[DATA1_MAX_SYMBOL], type[DATA1_MAX_SYMBOL];
340 if (sscanf(args, "%s %s %n", tclass, type, &val_offset) != 2)
342 logf(LOG_WARN, "Malformed variant triple at '%s'", tag);
346 data1_getvartypebyct(dh, parent->root->u.root.absyn->varset,
351 * If we're the first variant in this group, create a parent var,
352 * and insert it before the current variant.
354 if (parent->which != DATA1N_variant)
356 res = data1_mk_node (dh, m);
357 res->which = DATA1N_variant;
358 res->u.variant.type = 0;
359 res->u.variant.value = 0;
360 res->root = parent->root;
365 * now determine if one of our ancestor triples is of same type.
366 * If so, we break here. This will make the parser unwind until
367 * we become a sibling (alternate variant) to the aforementioned
368 * triple. It stinks that we re-parse these tags on every
369 * iteration of this. This is a function in need of a rewrite.
371 for (p = parent; p->which == DATA1N_variant; p = p->parent)
372 if (p->u.variant.type == tp)
374 res = data1_mk_node (dh, m);
375 res->which = DATA1N_variant;
376 res->root = parent->root;
377 res->u.variant.type = tp;
378 res->u.variant.value =
379 data1_insert_string (dh, res, m, args + val_offset);
383 else /* tag.. acquire our element in the abstract syntax */
385 data1_node *partag = get_parent_tag (dh, parent);
386 data1_element *e = 0;
389 if (parent->which == DATA1N_variant)
392 if (!(e = partag->u.tag.element))
393 localtag = 1; /* our parent is a local tag */
395 elem = data1_getelementbytagname(dh, absyn, e, tag);
396 res = data1_mk_node (dh, m);
397 res->which = DATA1N_tag;
398 res->u.tag.tag = data1_insert_string (dh, res, m, tag);
399 res->u.tag.element = elem;
400 res->u.tag.node_selected = 0;
401 res->u.tag.make_variantlist = 0;
402 res->u.tag.no_data_requested = 0;
403 res->u.tag.get_bytes = -1;
404 res->root = parent->root;
408 res->parent = parent;
413 while ((*pp = data1_read_node(dh, buf, res, line, absyn, m)))
415 res->last_child = *pp;
419 else /* != '<'... this is a body of text */
423 int len, prev_char = 0;
428 res = data1_mk_node(dh, m);
429 res->parent = parent;
430 res->which = DATA1N_data;
431 res->u.data.what = DATA1I_text;
432 res->u.data.formatted_text = 0;
433 res->root = parent->root;
435 /* determine length of "data" */
436 src = strchr (*buf, '<');
442 /* use local buffer of nmem if too large */
443 if (len >= DATA1_LOCALDATA)
444 res->u.data.data = (char*) nmem_malloc (m, len);
446 res->u.data.data = res->lbuf;
448 /* read "data" and transfer while removing white space */
449 dst = res->u.data.data;
450 for (src = *buf; --len >= 0; src++)
467 res->u.data.len = dst - res->u.data.data;
473 * Read a record in the native syntax.
475 data1_node *data1_read_record(data1_handle dh,
476 int (*rf)(void *, char *, size_t), void *fh,
480 char **buf = data1_get_read_buf (dh, &size);
486 *buf = (char *)xmalloc(*size = 4096);
490 if (rd + 4096 > *size && !(*buf =(char *)xrealloc(*buf, *size *= 2)))
492 if ((res = (*rf)(fh, *buf + rd, 4096)) <= 0)
497 return data1_read_node(dh, &bp, 0, &line, 0, m);
506 data1_node *data1_read_sgml (data1_handle dh, NMEM m, const char *buf)
508 const char *bp = buf;
510 return data1_read_node (dh, &bp, 0, &line, 0, m);