2 * Copyright (c) 1995-1998, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.25 1999-04-20 09:56:48 adam
8 * Added 'name' paramter to encoder/decoder routines (typedef Odr_fun).
9 * Modified all encoders/decoders to reflect this change.
11 * Revision 1.24 1998/10/28 15:10:09 adam
12 * Added --with-yc option to configure. For the data1_node in data1.h:
13 * decreased size of localdata and removed member "line" which wasn't useful.
15 * Revision 1.23 1998/03/12 11:28:45 adam
16 * Fix: didn't set root member of tagged node in function.
17 * data1_add_insert_taggeddata.
19 * Revision 1.22 1998/03/05 08:15:32 adam
20 * Implemented data1_add_insert_taggeddata utility which is more flexible
21 * than data1_insert_taggeddata.
23 * Revision 1.21 1998/02/27 14:08:05 adam
24 * Added const to some char pointer arguments.
25 * Reworked data1_read_node so that it doesn't create a tree with
26 * pointers to original "SGML"-buffer.
28 * Revision 1.20 1998/02/11 11:53:35 adam
29 * Changed code so that it compiles as C++.
31 * Revision 1.19 1997/12/09 16:17:09 adam
32 * Fix bug regarding variants. Tags with prefix "var" was incorrectly
33 * interpreted as "start of variants". Now, only "var" indicates such
35 * Cleaned up data1_read_node so tag names and variant names are
36 * copied and not pointed to by the generated data1 tree. Data nodes
37 * still point to old buffer.
39 * Revision 1.18 1997/11/18 09:51:09 adam
40 * Removed element num_children from data1_node. Minor changes in
43 * Revision 1.17 1997/11/05 09:20:51 adam
46 * Revision 1.16 1997/09/17 12:10:37 adam
49 * Revision 1.15 1997/09/05 09:50:57 adam
50 * Removed global data1_tabpath - uses data1_get_tabpath() instead.
52 * Revision 1.14 1997/05/14 06:54:04 adam
55 * Revision 1.13 1996/10/29 13:35:38 adam
56 * Implemented data1_set_tabpath and data1_get_tabpath.
58 * Revision 1.12 1996/10/11 10:35:38 adam
59 * Fixed a bug that caused data1_read_node to core dump when no abstract
60 * syntax was defined in a "sgml"-record.
62 * Revision 1.11 1996/07/06 19:58:35 quinn
63 * System headerfiles gathered in yconfig
65 * Revision 1.10 1996/01/19 15:41:47 quinn
66 * Fixed uninitialized boolean.
68 * Revision 1.9 1996/01/17 14:52:47 adam
69 * Changed prototype for reader function parsed to data1_read_record.
71 * Revision 1.8 1995/12/15 16:20:41 quinn
72 * Added formatted text.
74 * Revision 1.7 1995/12/13 13:44:32 quinn
75 * Modified Data1-system to use nmem
77 * Revision 1.6 1995/12/12 16:37:08 quinn
78 * Added destroy element to data1_node.
80 * Revision 1.5 1995/12/11 15:22:37 quinn
81 * Added last_child field to the node.
82 * Rewrote schema-mapping.
84 * Revision 1.4 1995/11/13 09:27:36 quinn
85 * Fiddling with the variant stuff.
87 * Revision 1.3 1995/11/01 16:34:57 quinn
88 * Making data1 look for tables in data1_tabpath
90 * Revision 1.2 1995/11/01 13:54:48 quinn
93 * Revision 1.1 1995/11/01 11:56:09 quinn
94 * Added Retrieval (data management) functions en masse.
96 * Revision 1.14 1995/10/30 12:40:55 quinn
97 * Fixed a couple of bugs.
99 * Revision 1.13 1995/10/25 16:00:47 quinn
100 * USMARC support is now almost operational
102 * Revision 1.12 1995/10/16 14:02:55 quinn
103 * Changes to support element set names and espec1
105 * Revision 1.11 1995/10/13 16:05:08 quinn
106 * Adding Espec1-processing
108 * Revision 1.10 1995/10/11 14:53:44 quinn
111 * Revision 1.9 1995/10/06 16:56:50 quinn
112 * Fixed ranked result.
114 * Revision 1.8 1995/10/06 16:44:13 quinn
115 * Work on attribute set mapping, etc.
117 * Revision 1.7 1995/10/06 12:58:35 quinn
120 * Revision 1.6 1995/10/04 09:29:49 quinn
121 * Adjustments to support USGS test data
123 * Revision 1.5 1995/10/03 17:56:43 quinn
126 * Revision 1.4 1995/10/02 15:53:19 quinn
129 * Revision 1.3 1995/10/02 14:55:21 quinn
130 * *** empty log message ***
132 * Revision 1.2 1995/09/14 15:18:13 quinn
135 * Revision 1.1 1995/09/12 11:24:30 quinn
136 * Beginning to add code for structured records.
151 * get the tag which is the immediate parent of this node (this may mean
152 * traversing intermediate things like variants and stuff.
154 data1_node *get_parent_tag (data1_handle dh, data1_node *n)
156 for (; n && n->which != DATA1N_root; n = n->parent)
157 if (n->which == DATA1N_tag)
162 data1_node *data1_mk_node (data1_handle dh, NMEM m)
166 r = (data1_node *)nmem_malloc(m, sizeof(*r));
167 r->next = r->child = r->last_child = r->parent = 0;
172 void data1_free_tree (data1_handle dh, data1_node *t)
174 data1_node *p = t->child, *pn;
179 data1_free_tree (dh, p);
186 char *data1_insert_string (data1_handle dh, data1_node *res,
187 NMEM m, const char *str)
189 int len = strlen(str);
191 if (len >= DATA1_LOCALDATA)
192 return nmem_strdup (m, str);
195 strcpy (res->lbuf, str);
201 data1_node *data1_add_insert_taggeddata(data1_handle dh, data1_node *root,
203 const char *tagname, NMEM m,
204 int first_flag, int local_allowed)
206 data1_node *partag = get_parent_tag (dh, at);
207 data1_node *tagn = data1_mk_node (dh, m);
208 data1_element *e = NULL;
211 tagn->which = DATA1N_tag;
212 tagn->u.tag.tag = data1_insert_string (dh, tagn, m, tagname);
213 tagn->u.tag.node_selected = 0;
214 tagn->u.tag.make_variantlist = 0;
215 tagn->u.tag.no_data_requested = 0;
216 tagn->u.tag.get_bytes = -1;
219 e = partag->u.tag.element;
220 tagn->u.tag.element =
221 data1_getelementbytagname (dh, root->u.root.absyn, e, tagname);
222 if (!local_allowed && !tagn->u.tag.element)
224 tagn->last_child = tagn->child = datn = data1_mk_node (dh, m);
228 datn->which = DATA1N_data;
229 datn->u.data.formatted_text = 0;
234 tagn->next = at->child;
236 at->last_child = tagn;
245 assert (at->last_child);
246 at->last_child->next = tagn;
248 at->last_child = tagn;
253 data1_node *data1_add_taggeddata(data1_handle dh, data1_node *root,
255 const char *tagname, NMEM m)
257 return data1_add_insert_taggeddata (dh, root, at, tagname, m, 0, 1);
262 * Insert a tagged node into the record root as first child of the node at
263 * which should be root or tag itself). Returns pointer to the data node,
264 * which can then be modified.
266 data1_node *data1_insert_taggeddata(data1_handle dh, data1_node *root,
268 const char *tagname, NMEM m)
270 return data1_add_insert_taggeddata (dh, root, at, tagname, m, 1, 0);
274 * Ugh. Sometimes functions just grow and grow on you. This one reads a
275 * 'node' and its children.
277 data1_node *data1_read_node (data1_handle dh, const char **buf,
278 data1_node *parent, int *line,
279 data1_absyn *absyn, NMEM m)
283 while (**buf && isspace(**buf))
292 if (**buf == '<') /* beginning of tag */
297 const char *t = (*buf) + 1;
299 data1_element *elem = 0;
301 for (i = 0; *t && *t != '>' && !isspace(*t); t++)
302 if (i < (sizeof(tag)-1))
307 for (i = 0; *t && *t != '>'; t++)
308 if (i < (sizeof(args)-1))
311 if (*t != '>' && !isspace(*t))
313 logf(LOG_WARN, "d1: %d: Malformed tag", *line);
317 * if end-tag, see if we terminate parent. If so, consume and return.
322 if (parent && (!*(tag +1) ||
323 (parent->which == DATA1N_root &&
324 !strcmp(tag + 1,parent->u.root.type)) ||
325 (parent->which == DATA1N_tag &&
326 !strcmp(tag + 1, parent->u.tag.tag))))
330 if (!absyn) /* parent node - what are we? */
332 if (!(absyn = data1_get_absyn (dh, tag)))
334 logf(LOG_WARN, "Unable to acquire abstract syntax for '%s'",
338 res = data1_mk_node (dh, m);
339 res->which = DATA1N_root;
340 res->u.root.type = data1_insert_string (dh, res, m, tag);
341 res->u.root.absyn = absyn;
345 else if (!strcmp(tag, "var"))
347 char tclass[DATA1_MAX_SYMBOL], type[DATA1_MAX_SYMBOL];
352 if (sscanf(args, "%s %s %n", tclass, type, &val_offset) != 2)
354 logf(LOG_WARN, "Malformed variant triple at '%s'", tag);
358 data1_getvartypebyct(dh, parent->root->u.root.absyn->varset,
363 * If we're the first variant in this group, create a parent var,
364 * and insert it before the current variant.
366 if (parent->which != DATA1N_variant)
368 res = data1_mk_node (dh, m);
369 res->which = DATA1N_variant;
370 res->u.variant.type = 0;
371 res->u.variant.value = 0;
372 res->root = parent->root;
377 * now determine if one of our ancestor triples is of same type.
378 * If so, we break here. This will make the parser unwind until
379 * we become a sibling (alternate variant) to the aforementioned
380 * triple. It stinks that we re-parse these tags on every
381 * iteration of this. This is a function in need of a rewrite.
383 for (p = parent; p->which == DATA1N_variant; p = p->parent)
384 if (p->u.variant.type == tp)
386 res = data1_mk_node (dh, m);
387 res->which = DATA1N_variant;
388 res->root = parent->root;
389 res->u.variant.type = tp;
390 res->u.variant.value =
391 data1_insert_string (dh, res, m, args + val_offset);
395 else /* tag.. acquire our element in the abstract syntax */
397 data1_node *partag = get_parent_tag (dh, parent);
398 data1_element *e = 0;
401 if (parent->which == DATA1N_variant)
404 if (!(e = partag->u.tag.element))
405 localtag = 1; /* our parent is a local tag */
407 elem = data1_getelementbytagname(dh, absyn, e, tag);
408 res = data1_mk_node (dh, m);
409 res->which = DATA1N_tag;
410 res->u.tag.tag = data1_insert_string (dh, res, m, tag);
411 res->u.tag.element = elem;
412 res->u.tag.node_selected = 0;
413 res->u.tag.make_variantlist = 0;
414 res->u.tag.no_data_requested = 0;
415 res->u.tag.get_bytes = -1;
416 res->root = parent->root;
420 res->parent = parent;
425 while ((*pp = data1_read_node(dh, buf, res, line, absyn, m)))
427 res->last_child = *pp;
431 else /* != '<'... this is a body of text */
435 int len, prev_char = 0;
440 res = data1_mk_node(dh, m);
441 res->parent = parent;
442 res->which = DATA1N_data;
443 res->u.data.what = DATA1I_text;
444 res->u.data.formatted_text = 0;
445 res->root = parent->root;
447 /* determine length of "data" */
448 src = strchr (*buf, '<');
454 /* use local buffer of nmem if too large */
455 if (len >= DATA1_LOCALDATA)
456 res->u.data.data = (char*) nmem_malloc (m, len);
458 res->u.data.data = res->lbuf;
460 /* read "data" and transfer while removing white space */
461 dst = res->u.data.data;
462 for (src = *buf; --len >= 0; src++)
479 res->u.data.len = dst - res->u.data.data;
485 * Read a record in the native syntax.
487 data1_node *data1_read_record(data1_handle dh,
488 int (*rf)(void *, char *, size_t), void *fh,
492 char **buf = data1_get_read_buf (dh, &size);
498 *buf = (char *)xmalloc(*size = 4096);
502 if (rd + 4096 > *size && !(*buf =(char *)xrealloc(*buf, *size *= 2)))
504 if ((res = (*rf)(fh, *buf + rd, 4096)) <= 0)
509 return data1_read_node(dh, &bp, 0, &line, 0, m);
518 data1_node *data1_read_sgml (data1_handle dh, NMEM m, const char *buf)
520 const char *bp = buf;
522 return data1_read_node (dh, &bp, 0, &line, 0, m);