2 * Copyright (c) 1995-1998, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.24 1998-10-28 15:10:09 adam
8 * Added --with-yc option to configure. For the data1_node in data1.h:
9 * decreased size of localdata and removed member "line" which wasn't useful.
11 * Revision 1.23 1998/03/12 11:28:45 adam
12 * Fix: didn't set root member of tagged node in function.
13 * data1_add_insert_taggeddata.
15 * Revision 1.22 1998/03/05 08:15:32 adam
16 * Implemented data1_add_insert_taggeddata utility which is more flexible
17 * than data1_insert_taggeddata.
19 * Revision 1.21 1998/02/27 14:08:05 adam
20 * Added const to some char pointer arguments.
21 * Reworked data1_read_node so that it doesn't create a tree with
22 * pointers to original "SGML"-buffer.
24 * Revision 1.20 1998/02/11 11:53:35 adam
25 * Changed code so that it compiles as C++.
27 * Revision 1.19 1997/12/09 16:17:09 adam
28 * Fix bug regarding variants. Tags with prefix "var" was incorrectly
29 * interpreted as "start of variants". Now, only "var" indicates such
31 * Cleaned up data1_read_node so tag names and variant names are
32 * copied and not pointed to by the generated data1 tree. Data nodes
33 * still point to old buffer.
35 * Revision 1.18 1997/11/18 09:51:09 adam
36 * Removed element num_children from data1_node. Minor changes in
39 * Revision 1.17 1997/11/05 09:20:51 adam
42 * Revision 1.16 1997/09/17 12:10:37 adam
45 * Revision 1.15 1997/09/05 09:50:57 adam
46 * Removed global data1_tabpath - uses data1_get_tabpath() instead.
48 * Revision 1.14 1997/05/14 06:54:04 adam
51 * Revision 1.13 1996/10/29 13:35:38 adam
52 * Implemented data1_set_tabpath and data1_get_tabpath.
54 * Revision 1.12 1996/10/11 10:35:38 adam
55 * Fixed a bug that caused data1_read_node to core dump when no abstract
56 * syntax was defined in a "sgml"-record.
58 * Revision 1.11 1996/07/06 19:58:35 quinn
59 * System headerfiles gathered in yconfig
61 * Revision 1.10 1996/01/19 15:41:47 quinn
62 * Fixed uninitialized boolean.
64 * Revision 1.9 1996/01/17 14:52:47 adam
65 * Changed prototype for reader function parsed to data1_read_record.
67 * Revision 1.8 1995/12/15 16:20:41 quinn
68 * Added formatted text.
70 * Revision 1.7 1995/12/13 13:44:32 quinn
71 * Modified Data1-system to use nmem
73 * Revision 1.6 1995/12/12 16:37:08 quinn
74 * Added destroy element to data1_node.
76 * Revision 1.5 1995/12/11 15:22:37 quinn
77 * Added last_child field to the node.
78 * Rewrote schema-mapping.
80 * Revision 1.4 1995/11/13 09:27:36 quinn
81 * Fiddling with the variant stuff.
83 * Revision 1.3 1995/11/01 16:34:57 quinn
84 * Making data1 look for tables in data1_tabpath
86 * Revision 1.2 1995/11/01 13:54:48 quinn
89 * Revision 1.1 1995/11/01 11:56:09 quinn
90 * Added Retrieval (data management) functions en masse.
92 * Revision 1.14 1995/10/30 12:40:55 quinn
93 * Fixed a couple of bugs.
95 * Revision 1.13 1995/10/25 16:00:47 quinn
96 * USMARC support is now almost operational
98 * Revision 1.12 1995/10/16 14:02:55 quinn
99 * Changes to support element set names and espec1
101 * Revision 1.11 1995/10/13 16:05:08 quinn
102 * Adding Espec1-processing
104 * Revision 1.10 1995/10/11 14:53:44 quinn
107 * Revision 1.9 1995/10/06 16:56:50 quinn
108 * Fixed ranked result.
110 * Revision 1.8 1995/10/06 16:44:13 quinn
111 * Work on attribute set mapping, etc.
113 * Revision 1.7 1995/10/06 12:58:35 quinn
116 * Revision 1.6 1995/10/04 09:29:49 quinn
117 * Adjustments to support USGS test data
119 * Revision 1.5 1995/10/03 17:56:43 quinn
122 * Revision 1.4 1995/10/02 15:53:19 quinn
125 * Revision 1.3 1995/10/02 14:55:21 quinn
126 * *** empty log message ***
128 * Revision 1.2 1995/09/14 15:18:13 quinn
131 * Revision 1.1 1995/09/12 11:24:30 quinn
132 * Beginning to add code for structured records.
147 * get the tag which is the immediate parent of this node (this may mean
148 * traversing intermediate things like variants and stuff.
150 data1_node *get_parent_tag (data1_handle dh, data1_node *n)
152 for (; n && n->which != DATA1N_root; n = n->parent)
153 if (n->which == DATA1N_tag)
158 data1_node *data1_mk_node (data1_handle dh, NMEM m)
162 r = (data1_node *)nmem_malloc(m, sizeof(*r));
163 r->next = r->child = r->last_child = r->parent = 0;
168 void data1_free_tree (data1_handle dh, data1_node *t)
170 data1_node *p = t->child, *pn;
175 data1_free_tree (dh, p);
182 char *data1_insert_string (data1_handle dh, data1_node *res,
183 NMEM m, const char *str)
185 int len = strlen(str);
187 if (len >= DATA1_LOCALDATA)
188 return nmem_strdup (m, str);
191 strcpy (res->lbuf, str);
197 data1_node *data1_add_insert_taggeddata(data1_handle dh, data1_node *root,
199 const char *tagname, NMEM m,
200 int first_flag, int local_allowed)
202 data1_node *partag = get_parent_tag (dh, at);
203 data1_node *tagn = data1_mk_node (dh, m);
204 data1_element *e = NULL;
207 tagn->which = DATA1N_tag;
208 tagn->u.tag.tag = data1_insert_string (dh, tagn, m, tagname);
209 tagn->u.tag.node_selected = 0;
210 tagn->u.tag.make_variantlist = 0;
211 tagn->u.tag.no_data_requested = 0;
212 tagn->u.tag.get_bytes = -1;
215 e = partag->u.tag.element;
216 tagn->u.tag.element =
217 data1_getelementbytagname (dh, root->u.root.absyn, e, tagname);
218 if (!local_allowed && !tagn->u.tag.element)
220 tagn->last_child = tagn->child = datn = data1_mk_node (dh, m);
224 datn->which = DATA1N_data;
225 datn->u.data.formatted_text = 0;
230 tagn->next = at->child;
232 at->last_child = tagn;
241 assert (at->last_child);
242 at->last_child->next = tagn;
244 at->last_child = tagn;
249 data1_node *data1_add_taggeddata(data1_handle dh, data1_node *root,
251 const char *tagname, NMEM m)
253 return data1_add_insert_taggeddata (dh, root, at, tagname, m, 0, 1);
258 * Insert a tagged node into the record root as first child of the node at
259 * which should be root or tag itself). Returns pointer to the data node,
260 * which can then be modified.
262 data1_node *data1_insert_taggeddata(data1_handle dh, data1_node *root,
264 const char *tagname, NMEM m)
266 return data1_add_insert_taggeddata (dh, root, at, tagname, m, 1, 0);
270 * Ugh. Sometimes functions just grow and grow on you. This one reads a
271 * 'node' and its children.
273 data1_node *data1_read_node (data1_handle dh, const char **buf,
274 data1_node *parent, int *line,
275 data1_absyn *absyn, NMEM m)
279 while (**buf && isspace(**buf))
288 if (**buf == '<') /* beginning of tag */
293 const char *t = (*buf) + 1;
295 data1_element *elem = 0;
297 for (i = 0; *t && *t != '>' && !isspace(*t); t++)
298 if (i < (sizeof(tag)-1))
303 for (i = 0; *t && *t != '>'; t++)
304 if (i < (sizeof(args)-1))
307 if (*t != '>' && !isspace(*t))
309 logf(LOG_WARN, "d1: %d: Malformed tag", *line);
313 * if end-tag, see if we terminate parent. If so, consume and return.
318 if (parent && (!*(tag +1) ||
319 (parent->which == DATA1N_root &&
320 !strcmp(tag + 1,parent->u.root.type)) ||
321 (parent->which == DATA1N_tag &&
322 !strcmp(tag + 1, parent->u.tag.tag))))
326 if (!absyn) /* parent node - what are we? */
328 if (!(absyn = data1_get_absyn (dh, tag)))
330 logf(LOG_WARN, "Unable to acquire abstract syntax for '%s'",
334 res = data1_mk_node (dh, m);
335 res->which = DATA1N_root;
336 res->u.root.type = data1_insert_string (dh, res, m, tag);
337 res->u.root.absyn = absyn;
341 else if (!strcmp(tag, "var"))
343 char tclass[DATA1_MAX_SYMBOL], type[DATA1_MAX_SYMBOL];
348 if (sscanf(args, "%s %s %n", tclass, type, &val_offset) != 2)
350 logf(LOG_WARN, "Malformed variant triple at '%s'", tag);
354 data1_getvartypebyct(dh, parent->root->u.root.absyn->varset,
359 * If we're the first variant in this group, create a parent var,
360 * and insert it before the current variant.
362 if (parent->which != DATA1N_variant)
364 res = data1_mk_node (dh, m);
365 res->which = DATA1N_variant;
366 res->u.variant.type = 0;
367 res->u.variant.value = 0;
368 res->root = parent->root;
373 * now determine if one of our ancestor triples is of same type.
374 * If so, we break here. This will make the parser unwind until
375 * we become a sibling (alternate variant) to the aforementioned
376 * triple. It stinks that we re-parse these tags on every
377 * iteration of this. This is a function in need of a rewrite.
379 for (p = parent; p->which == DATA1N_variant; p = p->parent)
380 if (p->u.variant.type == tp)
382 res = data1_mk_node (dh, m);
383 res->which = DATA1N_variant;
384 res->root = parent->root;
385 res->u.variant.type = tp;
386 res->u.variant.value =
387 data1_insert_string (dh, res, m, args + val_offset);
391 else /* tag.. acquire our element in the abstract syntax */
393 data1_node *partag = get_parent_tag (dh, parent);
394 data1_element *e = 0;
397 if (parent->which == DATA1N_variant)
400 if (!(e = partag->u.tag.element))
401 localtag = 1; /* our parent is a local tag */
403 elem = data1_getelementbytagname(dh, absyn, e, tag);
404 res = data1_mk_node (dh, m);
405 res->which = DATA1N_tag;
406 res->u.tag.tag = data1_insert_string (dh, res, m, tag);
407 res->u.tag.element = elem;
408 res->u.tag.node_selected = 0;
409 res->u.tag.make_variantlist = 0;
410 res->u.tag.no_data_requested = 0;
411 res->u.tag.get_bytes = -1;
412 res->root = parent->root;
416 res->parent = parent;
421 while ((*pp = data1_read_node(dh, buf, res, line, absyn, m)))
423 res->last_child = *pp;
427 else /* != '<'... this is a body of text */
431 int len, prev_char = 0;
436 res = data1_mk_node(dh, m);
437 res->parent = parent;
438 res->which = DATA1N_data;
439 res->u.data.what = DATA1I_text;
440 res->u.data.formatted_text = 0;
441 res->root = parent->root;
443 /* determine length of "data" */
444 src = strchr (*buf, '<');
450 /* use local buffer of nmem if too large */
451 if (len >= DATA1_LOCALDATA)
452 res->u.data.data = (char*) nmem_malloc (m, len);
454 res->u.data.data = res->lbuf;
456 /* read "data" and transfer while removing white space */
457 dst = res->u.data.data;
458 for (src = *buf; --len >= 0; src++)
475 res->u.data.len = dst - res->u.data.data;
481 * Read a record in the native syntax.
483 data1_node *data1_read_record(data1_handle dh,
484 int (*rf)(void *, char *, size_t), void *fh,
488 char **buf = data1_get_read_buf (dh, &size);
494 *buf = (char *)xmalloc(*size = 4096);
498 if (rd + 4096 > *size && !(*buf =(char *)xrealloc(*buf, *size *= 2)))
500 if ((res = (*rf)(fh, *buf + rd, 4096)) <= 0)
505 return data1_read_node(dh, &bp, 0, &line, 0, m);
514 data1_node *data1_read_sgml (data1_handle dh, NMEM m, const char *buf)
516 const char *bp = buf;
518 return data1_read_node (dh, &bp, 0, &line, 0, m);