2 * Copyright (c) 1995-2002, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
6 * $Id: d1_read.c,v 1.1 2002-10-22 12:53:33 adam Exp $
13 #include <yaz/xmalloc.h>
14 #include <yaz/yaz-util.h>
18 data1_node *data1_get_root_tag (data1_handle dh, data1_node *n)
22 if (data1_is_xmlmode(dh))
25 while (n && n->which != DATA1N_tag)
32 * get the tag which is the immediate parent of this node (this may mean
33 * traversing intermediate things like variants and stuff.
35 data1_node *get_parent_tag (data1_handle dh, data1_node *n)
37 if (data1_is_xmlmode(dh))
39 for (; n && n->which != DATA1N_root; n = n->parent)
40 if (n->which == DATA1N_tag && n->parent &&
41 n->parent->which != DATA1N_root)
46 for (; n && n->which != DATA1N_root; n = n->parent)
47 if (n->which == DATA1N_tag)
53 data1_node *data1_mk_node (data1_handle dh, NMEM m)
55 return data1_mk_node2 (dh, m, DATA1N_root, 0);
58 data1_node *data1_mk_node_type (data1_handle dh, NMEM m, int type)
60 return data1_mk_node2 (dh, m, type, 0);
63 static void data1_init_node (data1_handle dh, data1_node *r, int type)
71 r->u.tag.no_data_requested = 0;
72 r->u.tag.node_selected = 0;
73 r->u.tag.make_variantlist = 0;
74 r->u.tag.get_bytes = -1;
75 r->u.tag.attributes = 0;
85 r->u.data.formatted_text = 0;
91 r->u.data.formatted_text = 1;
94 r->u.variant.type = 0;
95 r->u.variant.value = 0;
97 case DATA1N_preprocess:
98 r->u.preprocess.target = 0;
99 r->u.preprocess.attributes = 0;
102 logf (LOG_WARN, "data_mk_node_type. bad type = %d\n", type);
106 data1_node *data1_append_node (data1_handle dh, NMEM m, int type,
109 data1_node *r = (data1_node *)nmem_malloc(m, sizeof(*r));
110 r->next = r->child = r->last_child = 0;
117 r->root = parent->root;
120 parent->child = parent->last_child = r;
122 parent->last_child->next = r;
123 parent->last_child = r;
125 data1_init_node(dh, r, type);
129 data1_node *data1_mk_node2 (data1_handle dh, NMEM m, int type,
132 return data1_append_node (dh, m, type, parent);
135 data1_node *data1_insert_node (data1_handle dh, NMEM m, int type,
138 data1_node *r = (data1_node *)nmem_malloc(m, sizeof(*r));
139 r->next = r->child = r->last_child = 0;
146 r->root = parent->root;
149 parent->last_child = r;
151 r->next = parent->child;
154 data1_init_node(dh, r, type);
158 void data1_free_tree (data1_handle dh, data1_node *t)
160 data1_node *p = t->child, *pn;
165 data1_free_tree (dh, p);
172 data1_node *data1_mk_root (data1_handle dh, NMEM nmem, const char *name)
174 data1_absyn *absyn = data1_get_absyn (dh, name);
178 yaz_log(LOG_WARN, "Unable to acquire abstract syntax " "for '%s'",
180 /* It's now OK for a record not to have an absyn */
182 res = data1_mk_node2 (dh, nmem, DATA1N_root, 0);
183 res->u.root.type = data1_insert_string (dh, res, nmem, name);
184 res->u.root.absyn = absyn;
188 void data1_set_root(data1_handle dh, data1_node *res,
189 NMEM nmem, const char *name)
191 data1_absyn *absyn = data1_get_absyn (dh, name);
193 res->u.root.type = data1_insert_string (dh, res, nmem, name);
194 res->u.root.absyn = absyn;
197 data1_node *data1_mk_preprocess (data1_handle dh, NMEM nmem,
199 const char **attr, data1_node *at)
201 return data1_mk_preprocess_n (dh, nmem, target, strlen(target),
205 data1_node *data1_mk_preprocess_n (data1_handle dh, NMEM nmem,
206 const char *target, size_t len,
207 const char **attr, data1_node *at)
210 data1_node *res = data1_mk_node2 (dh, nmem, DATA1N_preprocess, at);
211 res->u.preprocess.target = data1_insert_string_n (dh, res, nmem,
214 p = &res->u.preprocess.attributes;
215 while (attr && *attr)
217 *p = (data1_xattr*) nmem_malloc (nmem, sizeof(**p));
218 (*p)->name = nmem_strdup (nmem, *attr++);
219 (*p)->value = nmem_strdup (nmem, *attr++);
220 (*p)->what = DATA1I_text;
228 data1_node *data1_mk_tag_n (data1_handle dh, NMEM nmem,
229 const char *tag, size_t len, const char **attr,
232 data1_node *partag = get_parent_tag(dh, at);
233 data1_node *res = data1_mk_node2 (dh, nmem, DATA1N_tag, at);
235 data1_element *e = 0;
237 res->u.tag.tag = data1_insert_string_n (dh, res, nmem, tag, len);
239 if (!partag) /* top tag? */
240 e = data1_getelementbytagname (dh, at->root->u.root.absyn,
241 0 /* index as local */,
245 /* only set element for known tags */
246 e = partag->u.tag.element;
248 e = data1_getelementbytagname (dh, at->root->u.root.absyn,
251 res->u.tag.element = e;
252 p = &res->u.tag.attributes;
253 while (attr && *attr)
255 *p = (data1_xattr*) nmem_malloc (nmem, sizeof(**p));
256 (*p)->name = nmem_strdup (nmem, *attr++);
257 (*p)->value = nmem_strdup (nmem, *attr++);
258 (*p)->what = DATA1I_text;
265 void data1_tag_add_attr (data1_handle dh, NMEM nmem,
266 data1_node *res, const char **attr)
270 if (res->which != DATA1N_tag)
273 p = &res->u.tag.attributes;
277 while (attr && *attr)
279 *p = (data1_xattr*) nmem_malloc (nmem, sizeof(**p));
280 (*p)->name = nmem_strdup (nmem, *attr++);
281 (*p)->value = nmem_strdup (nmem, *attr++);
282 (*p)->what = DATA1I_text;
288 data1_node *data1_mk_tag (data1_handle dh, NMEM nmem,
289 const char *tag, const char **attr, data1_node *at)
291 return data1_mk_tag_n (dh, nmem, tag, strlen(tag), attr, at);
294 data1_node *data1_search_tag (data1_handle dh, data1_node *n,
299 n = data1_get_root_tag (dh, n);
304 for (; n; n = n->next)
305 if (n->which == DATA1N_tag && n->u.tag.tag &&
306 !yaz_matchstr (tag, n->u.tag.tag))
313 data1_node *data1_mk_tag_uni (data1_handle dh, NMEM nmem,
314 const char *tag, data1_node *at)
316 data1_node *node = data1_search_tag (dh, at->child, tag);
318 node = data1_mk_tag (dh, nmem, tag, 0 /* attr */, at);
320 node->child = node->last_child = 0;
324 data1_node *data1_mk_text_n (data1_handle dh, NMEM mem,
325 const char *buf, size_t len, data1_node *parent)
327 data1_node *res = data1_mk_node2 (dh, mem, DATA1N_data, parent);
328 res->u.data.what = DATA1I_text;
329 res->u.data.len = len;
331 res->u.data.data = data1_insert_string_n (dh, res, mem, buf, len);
335 data1_node *data1_mk_text_nf (data1_handle dh, NMEM mem,
336 const char *buf, size_t len, data1_node *parent)
338 data1_node *res = data1_mk_text_n (dh, mem, buf, len, parent);
339 res->u.data.formatted_text = 1;
343 data1_node *data1_mk_text (data1_handle dh, NMEM mem,
344 const char *buf, data1_node *parent)
346 return data1_mk_text_n (dh, mem, buf, strlen(buf), parent);
349 data1_node *data1_mk_comment_n (data1_handle dh, NMEM mem,
350 const char *buf, size_t len,
353 data1_node *res = data1_mk_node2 (dh, mem, DATA1N_comment, parent);
354 res->u.data.what = DATA1I_text;
355 res->u.data.len = len;
357 res->u.data.data = data1_insert_string_n (dh, res, mem, buf, len);
361 data1_node *data1_mk_comment (data1_handle dh, NMEM mem,
362 const char *buf, data1_node *parent)
364 return data1_mk_comment_n (dh, mem, buf, strlen(buf), parent);
367 char *data1_insert_string_n (data1_handle dh, data1_node *res,
368 NMEM m, const char *str, size_t len)
371 if (len >= DATA1_LOCALDATA)
372 b = (char *) nmem_malloc (m, len+1);
375 memcpy (b, str, len);
380 char *data1_insert_string (data1_handle dh, data1_node *res,
381 NMEM m, const char *str)
383 return data1_insert_string_n (dh, res, m, str, strlen(str));
386 static data1_node *data1_add_insert_taggeddata(data1_handle dh,
388 const char *tagname, NMEM m,
392 data1_node *root = at->root;
393 data1_node *partag = get_parent_tag (dh, at);
394 data1_element *e = NULL;
395 data1_node *datn = 0;
396 data1_node *tagn = 0;
399 e = data1_getelementbytagname (dh, root->u.root.absyn, 0, tagname);
402 e = partag->u.tag.element;
404 e = data1_getelementbytagname (dh, root->u.root.absyn, e, tagname);
406 if (local_allowed || e)
409 tagn = data1_insert_node (dh, m, DATA1N_tag, at);
411 tagn = data1_append_node (dh, m, DATA1N_tag, at);
412 tagn->u.tag.tag = data1_insert_string (dh, tagn, m, tagname);
413 tagn->u.tag.element = e;
414 datn = data1_mk_node2 (dh, m, DATA1N_data, tagn);
419 data1_node *data1_mk_tag_data(data1_handle dh, data1_node *at,
420 const char *tagname, NMEM m)
422 return data1_add_insert_taggeddata (dh, at, tagname, m, 1, 0);
427 * Insert a tagged node into the record root as first child of the node at
428 * which should be root or tag itself). Returns pointer to the data node,
429 * which can then be modified.
431 data1_node *data1_mk_tag_data_wd(data1_handle dh, data1_node *at,
432 const char *tagname, NMEM m)
434 return data1_add_insert_taggeddata (dh, at, tagname, m, 0, 1);
437 data1_node *data1_insert_taggeddata (data1_handle dh, data1_node *root,
438 data1_node *at, const char *tagname,
441 return data1_add_insert_taggeddata (dh, at, tagname, m, 0, 1);
444 data1_node *data1_add_taggeddata (data1_handle dh, data1_node *root,
445 data1_node *at, const char *tagname,
448 return data1_add_insert_taggeddata (dh, at, tagname, m, 1, 0);
451 data1_node *data1_mk_tag_data_int (data1_handle dh, data1_node *at,
452 const char *tag, int num,
455 data1_node *node_data;
457 node_data = data1_mk_tag_data (dh, at, tag, nmem);
460 node_data->u.data.what = DATA1I_num;
461 node_data->u.data.data = node_data->lbuf;
462 sprintf (node_data->u.data.data, "%d", num);
463 node_data->u.data.len = strlen (node_data->u.data.data);
467 data1_node *data1_mk_tag_data_oid (data1_handle dh, data1_node *at,
468 const char *tag, Odr_oid *oid,
471 data1_node *node_data;
472 char str[128], *p = str;
475 node_data = data1_mk_tag_data (dh, at, tag, nmem);
479 for (ii = oid; *ii >= 0; ii++)
483 sprintf (p, "%d", *ii);
486 node_data->u.data.what = DATA1I_oid;
487 node_data->u.data.len = strlen (str);
488 node_data->u.data.data = data1_insert_string (dh, node_data, nmem, str);
493 data1_node *data1_mk_tag_data_text (data1_handle dh, data1_node *at,
494 const char *tag, const char *str,
497 data1_node *node_data;
499 node_data = data1_mk_tag_data (dh, at, tag, nmem);
502 node_data->u.data.what = DATA1I_text;
503 node_data->u.data.len = strlen (str);
504 node_data->u.data.data = data1_insert_string (dh, node_data, nmem, str);
509 data1_node *data1_mk_tag_data_text_uni (data1_handle dh, data1_node *at,
510 const char *tag, const char *str,
513 data1_node *node = data1_search_tag (dh, at->child, tag);
515 return data1_mk_tag_data_text (dh, at, tag, str, nmem);
518 data1_node *node_data = node->child;
519 node_data->u.data.what = DATA1I_text;
520 node_data->u.data.len = strlen (str);
521 node_data->u.data.data = data1_insert_string (dh, node_data,
523 node_data->child = node_data->last_child = 0;
528 static int ampr (int (*get_byte)(void *fh), void *fh, int *amp)
531 int c = (*get_byte)(fh);
535 int c = (*get_byte)(fh);
550 if (!strcmp (ent, "quot"))
552 if (!strcmp (ent, "apos"))
554 if (!strcmp (ent, "gt"))
556 if (!strcmp (ent, "lt"))
558 if (!strcmp (ent, "amp"))
563 else if (c == 0 || d1_isspace(c))
573 data1_xattr *data1_read_xattr (data1_handle dh, NMEM m,
574 int (*get_byte)(void *fh), void *fh,
575 WRBUF wrbuf, int *ch, int *amp)
577 data1_xattr *p_first = 0;
578 data1_xattr **pp = &p_first;
584 while (*amp || (c && d1_isspace(c)))
585 c = ampr (get_byte, fh, amp);
586 if (*amp == 0 && (c == 0 || c == '>' || c == '/'))
588 *pp = p = (data1_xattr *) nmem_malloc (m, sizeof(*p));
592 p->what = DATA1I_xmltext;
595 while (c && c != '=' && c != '>' && c != '/' && !d1_isspace(c))
597 wrbuf_putc (wrbuf, c);
598 c = ampr (get_byte, fh, amp);
600 wrbuf_putc (wrbuf, '\0');
601 len = wrbuf_len(wrbuf);
602 p->name = (char*) nmem_malloc (m, len);
603 strcpy (p->name, wrbuf_buf(wrbuf));
606 c = ampr (get_byte, fh, amp);
607 if (*amp == 0 && c == '"')
609 c = ampr (get_byte, fh, amp);
611 while (*amp || (c && c != '"'))
613 wrbuf_putc (wrbuf, c);
614 c = ampr (get_byte, fh, amp);
617 c = ampr (get_byte, fh, amp);
619 else if (*amp == 0 && c == '\'')
621 c = ampr (get_byte, fh, amp);
623 while (*amp || (c && c != '\''))
625 wrbuf_putc (wrbuf, c);
626 c = ampr (get_byte, fh, amp);
629 c = ampr (get_byte, fh, amp);
634 while (*amp || (c && c != '>' && c != '/'))
636 wrbuf_putc (wrbuf, c);
637 c = ampr (get_byte, fh, amp);
640 wrbuf_putc (wrbuf, '\0');
641 len = wrbuf_len(wrbuf);
642 p->value = (char*) nmem_malloc (m, len);
643 strcpy (p->value, wrbuf_buf(wrbuf));
651 * Ugh. Sometimes functions just grow and grow on you. This one reads a
652 * 'node' and its children.
654 data1_node *data1_read_nodex (data1_handle dh, NMEM m,
655 int (*get_byte)(void *fh), void *fh, WRBUF wrbuf)
657 data1_node *d1_stack[256];
664 c = ampr (get_byte, fh, &);
667 data1_node *parent = level ? d1_stack[level-1] : 0;
669 if (amp == 0 && c == '<') /* beginning of tag */
679 c = ampr (get_byte, fh, &);
680 if (amp == 0 && c == '/')
683 c = ampr (get_byte, fh, &);
685 else if (amp == 0 && c == '!')
691 c0 = ampr (get_byte, fh, &0);
692 if (amp0 == 0 && c0 == '\0')
694 c = ampr (get_byte, fh, &);
696 if (amp0 == 0 && c0 == '-' && amp == 0 && c == '-')
698 /* COMMENT: <!-- ... --> */
701 c = ampr (get_byte, fh, &);
704 if (amp == 0 && c == '-')
706 else if (amp == 0 && c == '>' && no_dash >= 2)
712 wrbuf_buf(wrbuf), wrbuf_len(wrbuf)-2,
714 c = ampr (get_byte, fh, &); /* skip > */
719 wrbuf_putc (wrbuf, c);
720 c = ampr (get_byte, fh, &);
725 { /* DIRECTIVE: <! .. > */
730 if (amp == 0 && c == '>' && blevel == 0)
732 c = ampr (get_byte, fh, &);
735 if (amp == 0 && c == '[')
737 if (amp == 0 && c == ']' && blevel > 0)
739 c = ampr (get_byte, fh, &);
744 while (amp || (c && c != '>' && c != '/' && !d1_isspace(c)))
746 if (i < (sizeof(tag)-1))
748 c = ampr (get_byte, fh, &);
751 xattr = data1_read_xattr (dh, m, get_byte, fh, wrbuf, &c, &);
753 if (amp == 0 && c == '/')
754 { /* <tag attrs/> or <tag/> */
756 c = ampr (get_byte, fh, &);
760 yaz_log(LOG_WARN, "d1: %d: Malformed tag", line);
764 c = ampr (get_byte, fh, &);
776 parent = d1_stack[--i];
777 if ((parent->which == DATA1N_root &&
778 !strcmp(tag, parent->u.root.type)) ||
779 (parent->which == DATA1N_tag &&
780 !strcmp(tag, parent->u.tag.tag)))
788 yaz_log (LOG_WARN, "%d: no begin tag for %s",
793 if (data1_is_xmlmode(dh))
805 else if (!strcmp(tag, "var"))
807 char tclass[DATA1_MAX_SYMBOL], type[DATA1_MAX_SYMBOL];
811 if (sscanf(args, "%s %s %n", tclass, type, &val_offset) != 2)
813 yaz_log(LOG_WARN, "Malformed variant triple at '%s'", tag);
817 data1_getvartypebyct(dh,
818 parent->root->u.root.absyn->varset,
822 * If we're the first variant in this group, create a parent
823 * variant, and insert it before the current variant.
825 if (parent->which != DATA1N_variant)
827 res = data1_mk_node2 (dh, m, DATA1N_variant, parent);
832 * now determine if one of our ancestor triples is of
833 * same type. If so, we break here.
836 for (i = level-1; d1_stack[i]->which==DATA1N_variant; --i)
837 if (d1_stack[i]->u.variant.type == tp)
842 res = data1_mk_node2 (dh, m, DATA1N_variant, parent);
843 res->u.variant.type = tp;
844 res->u.variant.value =
845 data1_insert_string (dh, res, m, args + val_offset);
851 /* tag .. acquire our element in the abstract syntax */
854 parent = data1_mk_root (dh, m, tag);
855 res = d1_stack[level] = parent;
857 if (data1_is_xmlmode(dh))
860 res = data1_mk_tag (dh, m, tag, 0 /* attr */, parent);
861 res->u.tag.attributes = xattr;
866 res = data1_mk_tag (dh, m, tag, 0 /* attr */, parent);
867 res->u.tag.attributes = xattr;
870 d1_stack[level] = res;
871 d1_stack[level+1] = 0;
872 if (level < 250 && !null_tag)
875 else /* != '<'... this is a body of text */
881 c = ampr (get_byte, fh, &);
884 res = data1_mk_node2 (dh, m, DATA1N_data, parent);
885 res->u.data.what = DATA1I_xmltext;
886 res->u.data.formatted_text = 0;
887 d1_stack[level] = res;
891 while (amp || (c && c != '<'))
893 wrbuf_putc (wrbuf, c);
894 c = ampr (get_byte, fh, &);
896 len = wrbuf_len(wrbuf);
898 /* use local buffer of nmem if too large */
899 if (len >= DATA1_LOCALDATA)
900 res->u.data.data = (char*) nmem_malloc (m, len);
902 res->u.data.data = res->lbuf;
905 memcpy (res->u.data.data, wrbuf_buf(wrbuf), len);
907 res->u.data.data = 0;
908 res->u.data.len = len;
914 int getc_mem (void *fh)
916 const char **p = (const char **) fh;
922 data1_node *data1_read_node (data1_handle dh, const char **buf, NMEM m)
924 WRBUF wrbuf = wrbuf_alloc();
927 node = data1_read_nodex(dh, m, getc_mem, (void *) (buf), wrbuf);
928 wrbuf_free (wrbuf, 1);
933 * Read a record in the native syntax.
935 data1_node *data1_read_record(data1_handle dh,
936 int (*rf)(void *, char *, size_t), void *fh,
940 char **buf = data1_get_read_buf (dh, &size);
945 *buf = (char *)xmalloc(*size = 4096);
949 if (rd + 2048 >= *size && !(*buf =(char *)xrealloc(*buf, *size *= 2)))
951 if ((res = (*rf)(fh, *buf + rd, 2048)) <= 0)
957 return data1_read_node(dh, &bp, m);
966 data1_node *data1_read_sgml (data1_handle dh, NMEM m, const char *buf)
968 const char *bp = buf;
969 return data1_read_node (dh, &bp, m);
973 static int conv_item (NMEM m, yaz_iconv_t t,
974 WRBUF wrbuf, char *inbuf, size_t inlen)
976 wrbuf_rewind (wrbuf);
977 if (wrbuf->size < 10)
978 wrbuf_grow (wrbuf, 10);
981 char *outbuf = wrbuf->buf + wrbuf->pos;
982 size_t outlen = wrbuf->size - wrbuf->pos;
983 if (yaz_iconv (t, &inbuf, &inlen, &outbuf, &outlen) ==
984 (size_t)(-1) && yaz_iconv_error(t) != YAZ_ICONV_E2BIG)
986 /* bad data. stop and skip conversion entirely */
990 { /* finished converting */
991 wrbuf->pos = wrbuf->size - outlen;
996 /* buffer too small: make sure we expand buffer */
997 wrbuf->pos = wrbuf->size - outlen;
998 wrbuf_grow(wrbuf, 20);
1004 static void data1_iconv_s (data1_handle dh, NMEM m, data1_node *n,
1005 yaz_iconv_t t, WRBUF wrbuf, const char *tocode)
1007 for (; n; n = n->next)
1012 case DATA1N_comment:
1013 if (conv_item (m, t, wrbuf, n->u.data.data, n->u.data.len) == 0)
1016 data1_insert_string_n (dh, n, m, wrbuf->buf,
1018 n->u.data.len = wrbuf->pos;
1022 if (conv_item (m, t, wrbuf, n->u.tag.tag, strlen(n->u.tag.tag))
1026 data1_insert_string_n (dh, n, m,
1027 wrbuf->buf, wrbuf->pos);
1029 if (n->u.tag.attributes)
1032 for (p = n->u.tag.attributes; p; p = p->next)
1035 conv_item(m, t, wrbuf, p->value, strlen(p->value))
1038 wrbuf_puts (wrbuf, "");
1039 p->value = nmem_strdup (m, wrbuf->buf);
1044 case DATA1N_preprocess:
1045 if (strcmp(n->u.preprocess.target, "xml") == 0)
1047 data1_xattr *p = n->u.preprocess.attributes;
1048 for (; p; p = p->next)
1049 if (strcmp (p->name, "encoding") == 0)
1050 p->value = nmem_strdup (m, tocode);
1054 data1_iconv_s (dh, m, n->child, t, wrbuf, tocode);
1058 const char *data1_get_encoding (data1_handle dh, data1_node *n)
1060 /* see if we have an xml header that specifies encoding */
1061 if (n && n->child && n->child->which == DATA1N_preprocess &&
1062 strcmp (n->child->u.preprocess.target, "xml") == 0)
1064 data1_xattr *xp = n->child->u.preprocess.attributes;
1065 for (; xp; xp = xp->next)
1066 if (!strcmp (xp->name, "encoding") == 0)
1069 /* no encoding in header, so see if "encoding" was specified for abs */
1070 if (n && n->which == DATA1N_root &&
1071 n->u.root.absyn && n->u.root.absyn->encoding)
1072 return n->u.root.absyn->encoding;
1073 /* none of above, return a hard coded default */
1074 return "ISO-8859-1";
1077 int data1_iconv (data1_handle dh, NMEM m, data1_node *n,
1079 const char *fromcode)
1081 if (strcmp (tocode, fromcode))
1083 WRBUF wrbuf = wrbuf_alloc();
1084 yaz_iconv_t t = yaz_iconv_open (tocode, fromcode);
1087 data1_iconv_s (dh, m, n, t, wrbuf, tocode);
1088 yaz_iconv_close (t);
1089 wrbuf_free (wrbuf, 1);