2 * Copyright (c) 1995-1999, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.28 1999-10-21 09:50:33 adam
8 * SGML reader uses own isspace - it doesn't do 8-bit on WIN32!
10 * Revision 1.27 1999/08/27 09:40:32 adam
11 * Renamed logf function to yaz_log. Removed VC++ project files.
13 * Revision 1.26 1999/07/13 13:23:48 adam
14 * Non-recursive version of data1_read_node. data1_read_nodex reads
15 * stream of bytes (instead of buffer in memory).
17 * Revision 1.25 1999/04/20 09:56:48 adam
18 * Added 'name' paramter to encoder/decoder routines (typedef Odr_fun).
19 * Modified all encoders/decoders to reflect this change.
21 * Revision 1.24 1998/10/28 15:10:09 adam
22 * Added --with-yc option to configure. For the data1_node in data1.h:
23 * decreased size of localdata and removed member "line" which wasn't useful.
25 * Revision 1.23 1998/03/12 11:28:45 adam
26 * Fix: didn't set root member of tagged node in function.
27 * data1_add_insert_taggeddata.
29 * Revision 1.22 1998/03/05 08:15:32 adam
30 * Implemented data1_add_insert_taggeddata utility which is more flexible
31 * than data1_insert_taggeddata.
33 * Revision 1.21 1998/02/27 14:08:05 adam
34 * Added const to some char pointer arguments.
35 * Reworked data1_read_node so that it doesn't create a tree with
36 * pointers to original "SGML"-buffer.
38 * Revision 1.20 1998/02/11 11:53:35 adam
39 * Changed code so that it compiles as C++.
41 * Revision 1.19 1997/12/09 16:17:09 adam
42 * Fix bug regarding variants. Tags with prefix "var" was incorrectly
43 * interpreted as "start of variants". Now, only "var" indicates such
45 * Cleaned up data1_read_node so tag names and variant names are
46 * copied and not pointed to by the generated data1 tree. Data nodes
47 * still point to old buffer.
49 * Revision 1.18 1997/11/18 09:51:09 adam
50 * Removed element num_children from data1_node. Minor changes in
53 * Revision 1.17 1997/11/05 09:20:51 adam
56 * Revision 1.16 1997/09/17 12:10:37 adam
59 * Revision 1.15 1997/09/05 09:50:57 adam
60 * Removed global data1_tabpath - uses data1_get_tabpath() instead.
62 * Revision 1.14 1997/05/14 06:54:04 adam
65 * Revision 1.13 1996/10/29 13:35:38 adam
66 * Implemented data1_set_tabpath and data1_get_tabpath.
68 * Revision 1.12 1996/10/11 10:35:38 adam
69 * Fixed a bug that caused data1_read_node to core dump when no abstract
70 * syntax was defined in a "sgml"-record.
72 * Revision 1.11 1996/07/06 19:58:35 quinn
73 * System headerfiles gathered in yconfig
75 * Revision 1.10 1996/01/19 15:41:47 quinn
76 * Fixed uninitialized boolean.
78 * Revision 1.9 1996/01/17 14:52:47 adam
79 * Changed prototype for reader function parsed to data1_read_record.
81 * Revision 1.8 1995/12/15 16:20:41 quinn
82 * Added formatted text.
84 * Revision 1.7 1995/12/13 13:44:32 quinn
85 * Modified Data1-system to use nmem
87 * Revision 1.6 1995/12/12 16:37:08 quinn
88 * Added destroy element to data1_node.
90 * Revision 1.5 1995/12/11 15:22:37 quinn
91 * Added last_child field to the node.
92 * Rewrote schema-mapping.
94 * Revision 1.4 1995/11/13 09:27:36 quinn
95 * Fiddling with the variant stuff.
97 * Revision 1.3 1995/11/01 16:34:57 quinn
98 * Making data1 look for tables in data1_tabpath
100 * Revision 1.2 1995/11/01 13:54:48 quinn
103 * Revision 1.1 1995/11/01 11:56:09 quinn
104 * Added Retrieval (data management) functions en masse.
106 * Revision 1.14 1995/10/30 12:40:55 quinn
107 * Fixed a couple of bugs.
109 * Revision 1.13 1995/10/25 16:00:47 quinn
110 * USMARC support is now almost operational
112 * Revision 1.12 1995/10/16 14:02:55 quinn
113 * Changes to support element set names and espec1
115 * Revision 1.11 1995/10/13 16:05:08 quinn
116 * Adding Espec1-processing
118 * Revision 1.10 1995/10/11 14:53:44 quinn
121 * Revision 1.9 1995/10/06 16:56:50 quinn
122 * Fixed ranked result.
124 * Revision 1.8 1995/10/06 16:44:13 quinn
125 * Work on attribute set mapping, etc.
127 * Revision 1.7 1995/10/06 12:58:35 quinn
130 * Revision 1.6 1995/10/04 09:29:49 quinn
131 * Adjustments to support USGS test data
133 * Revision 1.5 1995/10/03 17:56:43 quinn
136 * Revision 1.4 1995/10/02 15:53:19 quinn
139 * Revision 1.3 1995/10/02 14:55:21 quinn
140 * *** empty log message ***
142 * Revision 1.2 1995/09/14 15:18:13 quinn
145 * Revision 1.1 1995/09/12 11:24:30 quinn
146 * Beginning to add code for structured records.
159 #define d1_isspace(c) strchr(" \r\n\t\f", c)
161 * get the tag which is the immediate parent of this node (this may mean
162 * traversing intermediate things like variants and stuff.
164 data1_node *get_parent_tag (data1_handle dh, data1_node *n)
166 for (; n && n->which != DATA1N_root; n = n->parent)
167 if (n->which == DATA1N_tag)
172 data1_node *data1_mk_node (data1_handle dh, NMEM m)
176 r = (data1_node *)nmem_malloc(m, sizeof(*r));
177 r->next = r->child = r->last_child = r->parent = 0;
182 void data1_free_tree (data1_handle dh, data1_node *t)
184 data1_node *p = t->child, *pn;
189 data1_free_tree (dh, p);
196 char *data1_insert_string (data1_handle dh, data1_node *res,
197 NMEM m, const char *str)
199 int len = strlen(str);
201 if (len >= DATA1_LOCALDATA)
202 return nmem_strdup (m, str);
205 strcpy (res->lbuf, str);
210 data1_node *data1_add_insert_taggeddata(data1_handle dh, data1_node *root,
212 const char *tagname, NMEM m,
213 int first_flag, int local_allowed)
215 data1_node *partag = get_parent_tag (dh, at);
216 data1_node *tagn = data1_mk_node (dh, m);
217 data1_element *e = NULL;
220 tagn->which = DATA1N_tag;
221 tagn->u.tag.tag = data1_insert_string (dh, tagn, m, tagname);
222 tagn->u.tag.node_selected = 0;
223 tagn->u.tag.make_variantlist = 0;
224 tagn->u.tag.no_data_requested = 0;
225 tagn->u.tag.get_bytes = -1;
228 e = partag->u.tag.element;
229 tagn->u.tag.element =
230 data1_getelementbytagname (dh, root->u.root.absyn, e, tagname);
231 if (!local_allowed && !tagn->u.tag.element)
233 tagn->last_child = tagn->child = datn = data1_mk_node (dh, m);
237 datn->which = DATA1N_data;
238 datn->u.data.formatted_text = 0;
243 tagn->next = at->child;
245 at->last_child = tagn;
254 assert (at->last_child);
255 at->last_child->next = tagn;
257 at->last_child = tagn;
262 data1_node *data1_add_taggeddata(data1_handle dh, data1_node *root,
264 const char *tagname, NMEM m)
266 return data1_add_insert_taggeddata (dh, root, at, tagname, m, 0, 1);
271 * Insert a tagged node into the record root as first child of the node at
272 * which should be root or tag itself). Returns pointer to the data node,
273 * which can then be modified.
275 data1_node *data1_insert_taggeddata(data1_handle dh, data1_node *root,
277 const char *tagname, NMEM m)
279 return data1_add_insert_taggeddata (dh, root, at, tagname, m, 1, 0);
283 * Ugh. Sometimes functions just grow and grow on you. This one reads a
284 * 'node' and its children.
286 data1_node *data1_read_nodex (data1_handle dh, NMEM m,
287 int (*get_byte)(void *fh), void *fh, WRBUF wrbuf)
289 data1_absyn *absyn = 0;
290 data1_node *d1_stack[256];
300 data1_node *parent = level ? d1_stack[level-1] : 0;
301 while (c != '\0' && d1_isspace(c))
310 if (c == '<') /* beginning of tag */
316 for (i = 0; (c=(*get_byte)(fh)) && c != '>' && !d1_isspace(c);)
317 if (i < (sizeof(tag)-1))
320 while (d1_isspace(c))
322 for (i = 0; c && c != '>'; c = (*get_byte)(fh))
323 if (i < (sizeof(args)-1))
328 yaz_log(LOG_WARN, "d1: %d: Malformed tag", line);
344 parent = d1_stack[--i];
345 if ((parent->which == DATA1N_root &&
346 !strcmp(tag+1, parent->u.root.type)) ||
347 (parent->which == DATA1N_tag &&
348 !strcmp(tag+1, parent->u.tag.tag)))
356 yaz_log (LOG_WARN, "%d: no begin tag for %s",
365 if (level == 0) /* root ? */
367 if (!(absyn = data1_get_absyn (dh, tag)))
369 yaz_log(LOG_WARN, "Unable to acquire abstract syntax "
373 res = data1_mk_node (dh, m);
374 res->which = DATA1N_root;
375 res->u.root.type = data1_insert_string (dh, res, m, tag);
376 res->u.root.absyn = absyn;
379 else if (!strcmp(tag, "var"))
381 char tclass[DATA1_MAX_SYMBOL], type[DATA1_MAX_SYMBOL];
385 if (sscanf(args, "%s %s %n", tclass, type, &val_offset) != 2)
387 yaz_log(LOG_WARN, "Malformed variant triple at '%s'", tag);
391 data1_getvartypebyct(dh,
392 parent->root->u.root.absyn->varset,
396 * If we're the first variant in this group, create a parent
397 * variant, and insert it before the current variant.
399 if (parent->which != DATA1N_variant)
401 res = data1_mk_node (dh, m);
402 res->which = DATA1N_variant;
403 res->u.variant.type = 0;
404 res->u.variant.value = 0;
409 * now determine if one of our ancestor triples is of
410 * same type. If so, we break here.
413 for (i = level-1; d1_stack[i]->which==DATA1N_variant; --i)
414 if (d1_stack[i]->u.variant.type == tp)
419 res = data1_mk_node (dh, m);
420 res->which = DATA1N_variant;
421 res->u.variant.type = tp;
422 res->u.variant.value =
423 data1_insert_string (dh, res, m, args + val_offset);
426 else /* tag.. acquire our element in the abstract syntax */
428 data1_node *partag = get_parent_tag (dh, parent);
429 data1_element *elem, *e = 0;
432 if (parent->which == DATA1N_variant)
435 if (!(e = partag->u.tag.element))
436 localtag = 1; /* our parent is a local tag */
438 elem = data1_getelementbytagname(dh, absyn, e, tag);
439 res = data1_mk_node (dh, m);
440 res->which = DATA1N_tag;
441 res->u.tag.tag = data1_insert_string (dh, res, m, tag);
442 res->u.tag.element = elem;
443 res->u.tag.node_selected = 0;
444 res->u.tag.make_variantlist = 0;
445 res->u.tag.no_data_requested = 0;
446 res->u.tag.get_bytes = -1;
450 parent->last_child = res;
451 res->root = parent->root;
453 res->parent = parent;
455 d1_stack[level]->next = res;
458 d1_stack[level] = res;
459 d1_stack[++level] = 0;
461 else /* != '<'... this is a body of text */
465 int len, prev_char = 0;
472 res = data1_mk_node(dh, m);
473 res->parent = parent;
474 res->which = DATA1N_data;
475 res->u.data.what = DATA1I_text;
476 res->u.data.formatted_text = 0;
477 res->root = parent->root;
478 parent->last_child = res;
480 d1_stack[level]->next = res;
483 d1_stack[level] = res;
489 wrbuf_putc (wrbuf, c);
492 len = wrbuf_len(wrbuf);
494 /* use local buffer of nmem if too large */
495 if (len >= DATA1_LOCALDATA)
496 res->u.data.data = (char*) nmem_malloc (m, len);
498 res->u.data.data = res->lbuf;
500 /* read "data" and transfer while removing white space */
501 dst = res->u.data.data;
502 for (src = wrbuf_buf(wrbuf); --len >= 0; src++)
506 if (d1_isspace (*src))
518 res->u.data.len = dst - res->u.data.data;
524 int getc_mem (void *fh)
526 const char **p = (const char **) fh;
532 data1_node *data1_read_node (data1_handle dh, const char **buf, NMEM m)
534 WRBUF wrbuf = wrbuf_alloc();
537 node = data1_read_nodex(dh, m, getc_mem, buf, wrbuf);
538 wrbuf_free (wrbuf, 1);
543 * Read a record in the native syntax.
545 data1_node *data1_read_record(data1_handle dh,
546 int (*rf)(void *, char *, size_t), void *fh,
550 char **buf = data1_get_read_buf (dh, &size);
555 *buf = (char *)xmalloc(*size = 4096);
559 if (rd + 2048 >= *size && !(*buf =(char *)xrealloc(*buf, *size *= 2)))
561 if ((res = (*rf)(fh, *buf + rd, 2048)) <= 0)
567 return data1_read_node(dh, &bp, m);
576 data1_node *data1_read_sgml (data1_handle dh, NMEM m, const char *buf)
578 const char *bp = buf;
579 return data1_read_node (dh, &bp, m);