1 /* $Id: d1_read.c,v 1.24 2007-03-20 22:07:35 adam Exp $
2 Copyright (C) 1995-2007
5 This file is part of the Zebra server.
7 Zebra is free software; you can redistribute it and/or modify it under
8 the terms of the GNU General Public License as published by the Free
9 Software Foundation; either version 2, or (at your option) any later
12 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
14 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
17 You should have received a copy of the GNU General Public License
18 along with this program; if not, write to the Free Software
19 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
25 * This module reads "loose" SGML and converts it to data1 tree
32 #include <yaz/yaz-util.h>
35 data1_node *data1_get_root_tag (data1_handle dh, data1_node *n)
39 if (data1_is_xmlmode(dh))
42 while (n && n->which != DATA1N_tag)
49 * get the tag which is the immediate parent of this node (this may mean
50 * traversing intermediate things like variants and stuff.
52 data1_node *get_parent_tag (data1_handle dh, data1_node *n)
54 if (data1_is_xmlmode(dh))
56 for (; n && n->which != DATA1N_root; n = n->parent)
57 if (n->which == DATA1N_tag && n->parent &&
58 n->parent->which != DATA1N_root)
63 for (; n && n->which != DATA1N_root; n = n->parent)
64 if (n->which == DATA1N_tag)
70 data1_node *data1_mk_node (data1_handle dh, NMEM m)
72 return data1_mk_node2 (dh, m, DATA1N_root, 0);
75 data1_node *data1_mk_node_type (data1_handle dh, NMEM m, int type)
77 return data1_mk_node2 (dh, m, type, 0);
80 static void data1_init_node (data1_handle dh, data1_node *r, int type)
88 r->u.tag.no_data_requested = 0;
89 r->u.tag.node_selected = 0;
90 r->u.tag.make_variantlist = 0;
91 r->u.tag.get_bytes = -1;
92 r->u.tag.attributes = 0;
102 r->u.data.formatted_text = 0;
108 r->u.data.formatted_text = 1;
111 r->u.variant.type = 0;
112 r->u.variant.value = 0;
114 case DATA1N_preprocess:
115 r->u.preprocess.target = 0;
116 r->u.preprocess.attributes = 0;
119 yaz_log (YLOG_WARN, "data_mk_node_type. bad type = %d\n", type);
123 data1_node *data1_append_node (data1_handle dh, NMEM m, int type,
126 data1_node *r = (data1_node *)nmem_malloc(m, sizeof(*r));
127 r->next = r->child = r->last_child = 0;
134 r->root = parent->root;
136 parent->child = parent->last_child = r;
138 parent->last_child->next = r;
139 parent->last_child = r;
141 data1_init_node(dh, r, type);
145 data1_node *data1_mk_node2 (data1_handle dh, NMEM m, int type,
148 return data1_append_node (dh, m, type, parent);
151 data1_node *data1_insert_node (data1_handle dh, NMEM m, int type,
154 data1_node *r = (data1_node *)nmem_malloc(m, sizeof(*r));
155 r->next = r->child = r->last_child = 0;
161 r->root = parent->root;
164 parent->last_child = r;
166 r->next = parent->child;
169 data1_init_node(dh, r, type);
173 data1_node *data1_mk_root (data1_handle dh, NMEM nmem, const char *name)
175 data1_absyn *absyn = data1_get_absyn(dh, name, 1);
179 yaz_log(YLOG_WARN, "Unable to acquire abstract syntax " "for '%s'",
181 /* It's now OK for a record not to have an absyn */
183 res = data1_mk_node2 (dh, nmem, DATA1N_root, 0);
184 res->u.root.type = data1_insert_string (dh, res, nmem, name);
185 res->u.root.absyn = absyn;
189 void data1_set_root(data1_handle dh, data1_node *res,
190 NMEM nmem, const char *name)
192 data1_absyn *absyn = data1_get_absyn(
193 dh, name, DATA1_XPATH_INDEXING_ENABLE);
195 res->u.root.type = data1_insert_string (dh, res, nmem, name);
196 res->u.root.absyn = absyn;
199 void data1_add_attrs(data1_handle dh, NMEM nmem, const char **attr,
205 while (attr && *attr)
207 *p = (data1_xattr*) nmem_malloc (nmem, sizeof(**p));
208 (*p)->name = nmem_strdup (nmem, *attr++);
209 (*p)->value = nmem_strdup (nmem, *attr++);
210 (*p)->what = DATA1I_text;
217 data1_node *data1_mk_preprocess (data1_handle dh, NMEM nmem,
219 const char **attr, data1_node *at)
221 return data1_mk_preprocess_n (dh, nmem, target, strlen(target),
225 data1_node *data1_mk_preprocess_n (data1_handle dh, NMEM nmem,
226 const char *target, size_t len,
227 const char **attr, data1_node *at)
229 data1_node *res = data1_mk_node2 (dh, nmem, DATA1N_preprocess, at);
230 res->u.preprocess.target = data1_insert_string_n (dh, res, nmem,
233 data1_add_attrs(dh, nmem, attr, &res->u.preprocess.attributes);
237 data1_node *data1_insert_preprocess (data1_handle dh, NMEM nmem,
239 const char **attr, data1_node *at)
241 return data1_insert_preprocess_n (dh, nmem, target, strlen(target),
245 data1_node *data1_insert_preprocess_n (data1_handle dh, NMEM nmem,
246 const char *target, size_t len,
247 const char **attr, data1_node *at)
249 data1_node *res = data1_insert_node (dh, nmem, DATA1N_preprocess, at);
250 res->u.preprocess.target = data1_insert_string_n (dh, res, nmem,
253 data1_add_attrs(dh, nmem, attr, &res->u.preprocess.attributes);
257 data1_node *data1_mk_tag_n (data1_handle dh, NMEM nmem,
258 const char *tag, size_t len, const char **attr,
261 data1_node *partag = get_parent_tag(dh, at);
262 data1_node *res = data1_mk_node2 (dh, nmem, DATA1N_tag, at);
263 data1_element *e = 0;
265 res->u.tag.tag = data1_insert_string_n (dh, res, nmem, tag, len);
267 if (!partag) /* top tag? */
268 e = data1_getelementbytagname (dh, at->root->u.root.absyn,
269 0 /* index as local */,
273 /* only set element for known tags */
274 e = partag->u.tag.element;
276 e = data1_getelementbytagname (dh, at->root->u.root.absyn,
279 res->u.tag.element = e;
280 data1_add_attrs(dh, nmem, attr, &res->u.tag.attributes);
284 void data1_tag_add_attr (data1_handle dh, NMEM nmem,
285 data1_node *res, const char **attr)
287 if (res->which != DATA1N_tag)
290 data1_add_attrs(dh, nmem, attr, &res->u.tag.attributes);
293 data1_node *data1_mk_tag (data1_handle dh, NMEM nmem,
294 const char *tag, const char **attr, data1_node *at)
296 return data1_mk_tag_n (dh, nmem, tag, strlen(tag), attr, at);
299 data1_node *data1_search_tag (data1_handle dh, data1_node *n,
304 n = data1_get_root_tag (dh, n);
309 for (; n; n = n->next)
310 if (n->which == DATA1N_tag && n->u.tag.tag &&
311 !yaz_matchstr (n->u.tag.tag, tag))
318 data1_node *data1_mk_tag_uni (data1_handle dh, NMEM nmem,
319 const char *tag, data1_node *at)
321 data1_node *node = data1_search_tag (dh, at->child, tag);
323 node = data1_mk_tag (dh, nmem, tag, 0 /* attr */, at);
325 node->child = node->last_child = 0;
329 data1_node *data1_mk_text_n (data1_handle dh, NMEM mem,
330 const char *buf, size_t len, data1_node *parent)
332 data1_node *res = data1_mk_node2 (dh, mem, DATA1N_data, parent);
333 res->u.data.what = DATA1I_text;
334 res->u.data.len = len;
336 res->u.data.data = data1_insert_string_n (dh, res, mem, buf, len);
340 data1_node *data1_mk_text_nf (data1_handle dh, NMEM mem,
341 const char *buf, size_t len, data1_node *parent)
343 data1_node *res = data1_mk_text_n (dh, mem, buf, len, parent);
344 res->u.data.formatted_text = 1;
348 data1_node *data1_mk_text (data1_handle dh, NMEM mem,
349 const char *buf, data1_node *parent)
351 return data1_mk_text_n (dh, mem, buf, strlen(buf), parent);
354 data1_node *data1_mk_comment_n (data1_handle dh, NMEM mem,
355 const char *buf, size_t len,
358 data1_node *res = data1_mk_node2 (dh, mem, DATA1N_comment, parent);
359 res->u.data.what = DATA1I_text;
360 res->u.data.len = len;
362 res->u.data.data = data1_insert_string_n (dh, res, mem, buf, len);
366 data1_node *data1_mk_comment (data1_handle dh, NMEM mem,
367 const char *buf, data1_node *parent)
369 return data1_mk_comment_n (dh, mem, buf, strlen(buf), parent);
372 char *data1_insert_string_n (data1_handle dh, data1_node *res,
373 NMEM m, const char *str, size_t len)
376 if (len >= DATA1_LOCALDATA)
377 b = (char *) nmem_malloc (m, len+1);
380 memcpy (b, str, len);
385 char *data1_insert_string (data1_handle dh, data1_node *res,
386 NMEM m, const char *str)
388 return data1_insert_string_n (dh, res, m, str, strlen(str));
391 static data1_node *data1_add_insert_taggeddata(data1_handle dh,
393 const char *tagname, NMEM m,
397 data1_node *root = at->root;
398 data1_node *partag = get_parent_tag (dh, at);
399 data1_element *e = NULL;
400 data1_node *datn = 0;
401 data1_node *tagn = 0;
404 e = data1_getelementbytagname (dh, root->u.root.absyn, 0, tagname);
407 e = partag->u.tag.element;
409 e = data1_getelementbytagname (dh, root->u.root.absyn, e, tagname);
411 if (local_allowed || e)
414 tagn = data1_insert_node (dh, m, DATA1N_tag, at);
416 tagn = data1_append_node (dh, m, DATA1N_tag, at);
417 tagn->u.tag.tag = data1_insert_string (dh, tagn, m, tagname);
418 tagn->u.tag.element = e;
419 datn = data1_mk_node2 (dh, m, DATA1N_data, tagn);
424 data1_node *data1_mk_tag_data(data1_handle dh, data1_node *at,
425 const char *tagname, NMEM m)
427 return data1_add_insert_taggeddata (dh, at, tagname, m, 1, 0);
432 * Insert a tagged node into the record root as first child of the node at
433 * which should be root or tag itself). Returns pointer to the data node,
434 * which can then be modified.
436 data1_node *data1_mk_tag_data_wd(data1_handle dh, data1_node *at,
437 const char *tagname, NMEM m)
439 return data1_add_insert_taggeddata (dh, at, tagname, m, 0, 1);
442 data1_node *data1_insert_taggeddata (data1_handle dh, data1_node *root,
443 data1_node *at, const char *tagname,
446 return data1_add_insert_taggeddata (dh, at, tagname, m, 0, 1);
449 data1_node *data1_add_taggeddata (data1_handle dh, data1_node *root,
450 data1_node *at, const char *tagname,
453 return data1_add_insert_taggeddata (dh, at, tagname, m, 1, 0);
456 data1_node *data1_mk_tag_data_zint (data1_handle dh, data1_node *at,
457 const char *tag, zint num,
460 data1_node *node_data;
462 node_data = data1_mk_tag_data (dh, at, tag, nmem);
465 node_data->u.data.what = DATA1I_num;
466 node_data->u.data.data = node_data->lbuf;
467 sprintf (node_data->u.data.data, ZINT_FORMAT, num);
468 node_data->u.data.len = strlen (node_data->u.data.data);
472 data1_node *data1_mk_tag_data_int (data1_handle dh, data1_node *at,
473 const char *tag, int num,
476 return data1_mk_tag_data_zint(dh, at, tag, num, nmem);
479 data1_node *data1_mk_tag_data_oid (data1_handle dh, data1_node *at,
480 const char *tag, Odr_oid *oid,
483 data1_node *node_data;
484 char str[128], *p = str;
487 node_data = data1_mk_tag_data (dh, at, tag, nmem);
491 for (ii = oid; *ii >= 0; ii++)
495 sprintf (p, "%d", *ii);
498 node_data->u.data.what = DATA1I_oid;
499 node_data->u.data.len = strlen (str);
500 node_data->u.data.data = data1_insert_string (dh, node_data, nmem, str);
505 data1_node *data1_mk_tag_data_text (data1_handle dh, data1_node *at,
506 const char *tag, const char *str,
509 data1_node *node_data;
511 node_data = data1_mk_tag_data (dh, at, tag, nmem);
514 node_data->u.data.what = DATA1I_text;
515 node_data->u.data.len = strlen (str);
516 node_data->u.data.data = data1_insert_string (dh, node_data, nmem, str);
521 data1_node *data1_mk_tag_data_text_uni (data1_handle dh, data1_node *at,
522 const char *tag, const char *str,
525 data1_node *node = data1_search_tag (dh, at->child, tag);
527 return data1_mk_tag_data_text (dh, at, tag, str, nmem);
530 data1_node *node_data = node->child;
531 node_data->u.data.what = DATA1I_text;
532 node_data->u.data.len = strlen (str);
533 node_data->u.data.data = data1_insert_string (dh, node_data,
535 node_data->child = node_data->last_child = 0;
540 static int ampr (int (*get_byte)(void *fh), void *fh, int *amp)
543 int c = (*get_byte)(fh);
547 int c = (*get_byte)(fh);
562 if (!strcmp (ent, "quot"))
564 if (!strcmp (ent, "apos"))
566 if (!strcmp (ent, "gt"))
568 if (!strcmp (ent, "lt"))
570 if (!strcmp (ent, "amp"))
575 else if (c == 0 || d1_isspace(c))
585 data1_xattr *data1_read_xattr (data1_handle dh, NMEM m,
586 int (*get_byte)(void *fh), void *fh,
587 WRBUF wrbuf, int *ch, int *amp)
589 data1_xattr *p_first = 0;
590 data1_xattr **pp = &p_first;
595 while (*amp || (c && d1_isspace(c)))
596 c = ampr (get_byte, fh, amp);
597 if (*amp == 0 && (c == 0 || c == '>' || c == '/'))
599 *pp = p = (data1_xattr *) nmem_malloc (m, sizeof(*p));
603 p->what = DATA1I_xmltext;
606 while (c && c != '=' && c != '>' && c != '/' && !d1_isspace(c))
608 wrbuf_putc (wrbuf, c);
609 c = ampr (get_byte, fh, amp);
611 p->name = nmem_strdup (m, wrbuf_cstr(wrbuf));
614 c = ampr (get_byte, fh, amp);
615 if (*amp == 0 && c == '"')
617 c = ampr (get_byte, fh, amp);
619 while (*amp || (c && c != '"'))
621 wrbuf_putc (wrbuf, c);
622 c = ampr (get_byte, fh, amp);
625 c = ampr (get_byte, fh, amp);
627 else if (*amp == 0 && c == '\'')
629 c = ampr (get_byte, fh, amp);
631 while (*amp || (c && c != '\''))
633 wrbuf_putc (wrbuf, c);
634 c = ampr (get_byte, fh, amp);
637 c = ampr (get_byte, fh, amp);
642 while (*amp || (c && c != '>' && c != '/'))
644 wrbuf_putc (wrbuf, c);
645 c = ampr (get_byte, fh, amp);
648 p->value = nmem_strdup(m, wrbuf_cstr(wrbuf));
656 * Ugh. Sometimes functions just grow and grow on you. This one reads a
657 * 'node' and its children.
659 data1_node *data1_read_nodex (data1_handle dh, NMEM m,
660 int (*get_byte)(void *fh), void *fh, WRBUF wrbuf)
662 data1_node *d1_stack[256];
669 c = ampr (get_byte, fh, &);
672 data1_node *parent = level ? d1_stack[level-1] : 0;
674 if (amp == 0 && c == '<') /* beginning of tag */
683 c = ampr (get_byte, fh, &);
684 if (amp == 0 && c == '/')
687 c = ampr (get_byte, fh, &);
689 else if (amp == 0 && c == '?')
692 while ((c = ampr(get_byte, fh, &)))
704 c = ampr(get_byte, fh, &);
710 if (amp == 0 && c == quote_mode)
716 else if (amp == 0 && c == '!')
722 c0 = ampr (get_byte, fh, &0);
723 if (amp0 == 0 && c0 == '\0')
725 c = ampr (get_byte, fh, &);
727 if (amp0 == 0 && c0 == '-' && amp == 0 && c == '-')
729 /* COMMENT: <!-- ... --> */
732 c = ampr (get_byte, fh, &);
735 if (amp == 0 && c == '-')
737 else if (amp == 0 && c == '>' && no_dash >= 2)
743 wrbuf_buf(wrbuf), wrbuf_len(wrbuf)-2,
745 c = ampr (get_byte, fh, &); /* skip > */
750 wrbuf_putc (wrbuf, c);
751 c = ampr (get_byte, fh, &);
756 { /* DIRECTIVE: <! .. > */
761 if (amp == 0 && c == '>' && blevel == 0)
763 c = ampr (get_byte, fh, &);
766 if (amp == 0 && c == '[')
768 if (amp == 0 && c == ']' && blevel > 0)
770 c = ampr (get_byte, fh, &);
775 while (amp || (c && c != '>' && c != '/' && !d1_isspace(c)))
777 if (i < (sizeof(tag)-1))
779 c = ampr (get_byte, fh, &);
782 xattr = data1_read_xattr (dh, m, get_byte, fh, wrbuf, &c, &);
783 if (amp == 0 && c == '/')
784 { /* <tag attrs/> or <tag/> */
786 c = ampr (get_byte, fh, &);
790 yaz_log(YLOG_WARN, "d1: %d: Malformed tag", line);
794 c = ampr (get_byte, fh, &);
806 parent = d1_stack[--i];
807 if ((parent->which == DATA1N_root &&
808 !strcmp(tag, parent->u.root.type)) ||
809 (parent->which == DATA1N_tag &&
810 !strcmp(tag, parent->u.tag.tag)))
818 yaz_log (YLOG_WARN, "%d: no begin tag for %s",
823 if (data1_is_xmlmode(dh))
835 else if (!strcmp(tag, "var")
836 && xattr && xattr->next && xattr->next->next
838 && xattr->next->value == 0
839 && xattr->next->next->value == 0)
841 /* <var class type value> */
842 const char *tclass = xattr->name;
843 const char *type = xattr->next->name;
844 const char *value = xattr->next->name;
847 yaz_log(YLOG_LOG, "Variant class=%s type=%s value=%s",
848 tclass, type, value);
850 data1_getvartypebyct(dh,
851 parent->root->u.root.absyn->varset,
855 * If we're the first variant in this group, create a parent
856 * variant, and insert it before the current variant.
858 if (parent->which != DATA1N_variant)
860 res = data1_mk_node2 (dh, m, DATA1N_variant, parent);
865 * now determine if one of our ancestor triples is of
866 * same type. If so, we break here.
869 for (i = level-1; d1_stack[i]->which==DATA1N_variant; --i)
870 if (d1_stack[i]->u.variant.type == tp)
875 res = data1_mk_node2 (dh, m, DATA1N_variant, parent);
876 res->u.variant.type = tp;
877 res->u.variant.value =
878 data1_insert_string (dh, res, m, value);
884 /* tag .. acquire our element in the abstract syntax */
887 parent = data1_mk_root (dh, m, tag);
888 res = d1_stack[level] = parent;
890 if (data1_is_xmlmode(dh))
893 res = data1_mk_tag (dh, m, tag, 0 /* attr */, parent);
894 res->u.tag.attributes = xattr;
899 res = data1_mk_tag (dh, m, tag, 0 /* attr */, parent);
900 res->u.tag.attributes = xattr;
903 d1_stack[level] = res;
904 d1_stack[level+1] = 0;
905 if (level < 250 && !null_tag)
908 else /* != '<'... this is a body of text */
914 c = ampr (get_byte, fh, &);
917 res = data1_mk_node2 (dh, m, DATA1N_data, parent);
918 res->u.data.what = DATA1I_xmltext;
919 res->u.data.formatted_text = 0;
920 d1_stack[level] = res;
924 while (amp || (c && c != '<'))
926 wrbuf_putc (wrbuf, c);
927 c = ampr (get_byte, fh, &);
929 len = wrbuf_len(wrbuf);
931 /* use local buffer of nmem if too large */
932 if (len >= DATA1_LOCALDATA)
933 res->u.data.data = (char*) nmem_malloc (m, len);
935 res->u.data.data = res->lbuf;
938 memcpy (res->u.data.data, wrbuf_buf(wrbuf), len);
940 res->u.data.data = 0;
941 res->u.data.len = len;
947 int getc_mem (void *fh)
949 const char **p = (const char **) fh;
955 data1_node *data1_read_node (data1_handle dh, const char **buf, NMEM m)
957 WRBUF wrbuf = wrbuf_alloc();
960 node = data1_read_nodex(dh, m, getc_mem, (void *) (buf), wrbuf);
961 wrbuf_destroy(wrbuf);
966 * Read a record in the native syntax.
968 data1_node *data1_read_record(data1_handle dh,
969 int (*rf)(void *, char *, size_t), void *fh,
973 char **buf = data1_get_read_buf (dh, &size);
978 *buf = (char *)xmalloc(*size = 4096);
982 if (rd + 2048 >= *size && !(*buf =(char *)xrealloc(*buf, *size *= 2)))
984 if ((res = (*rf)(fh, *buf + rd, 2048)) <= 0)
990 return data1_read_node(dh, &bp, m);
999 data1_node *data1_read_sgml (data1_handle dh, NMEM m, const char *buf)
1001 const char *bp = buf;
1002 return data1_read_node (dh, &bp, m);
1006 static int conv_item (NMEM m, yaz_iconv_t t,
1007 WRBUF wrbuf, char *inbuf, size_t inlen)
1009 wrbuf_rewind (wrbuf);
1010 if (wrbuf->size < 10)
1011 wrbuf_grow (wrbuf, 10);
1014 char *outbuf = wrbuf->buf + wrbuf->pos;
1015 size_t outlen = wrbuf->size - wrbuf->pos;
1016 if (yaz_iconv(t, &inbuf, &inlen, &outbuf, &outlen) ==
1017 (size_t)(-1) && yaz_iconv_error(t) != YAZ_ICONV_E2BIG)
1019 /* bad data. stop and skip conversion entirely */
1022 else if (inlen == 0)
1023 { /* finished converting, flush it */
1024 yaz_iconv(t, 0, 0, &outbuf, &outlen);
1025 wrbuf->pos = wrbuf->size - outlen;
1030 /* buffer too small: make sure we expand buffer */
1031 wrbuf->pos = wrbuf->size - outlen;
1032 wrbuf_grow(wrbuf, 20);
1038 static void data1_iconv_s (data1_handle dh, NMEM m, data1_node *n,
1039 yaz_iconv_t t, WRBUF wrbuf, const char *tocode)
1041 for (; n; n = n->next)
1046 case DATA1N_comment:
1047 if (conv_item (m, t, wrbuf, n->u.data.data, n->u.data.len) == 0)
1050 data1_insert_string_n (dh, n, m, wrbuf->buf,
1052 n->u.data.len = wrbuf->pos;
1056 if (conv_item (m, t, wrbuf, n->u.tag.tag, strlen(n->u.tag.tag))
1060 data1_insert_string_n (dh, n, m,
1061 wrbuf->buf, wrbuf->pos);
1063 if (n->u.tag.attributes)
1066 for (p = n->u.tag.attributes; p; p = p->next)
1069 conv_item(m, t, wrbuf, p->value, strlen(p->value))
1072 p->value = nmem_strdup(m, wrbuf_cstr(wrbuf));
1077 case DATA1N_preprocess:
1078 if (strcmp(n->u.preprocess.target, "xml") == 0)
1080 data1_xattr *p = n->u.preprocess.attributes;
1081 for (; p; p = p->next)
1082 if (strcmp (p->name, "encoding") == 0)
1083 p->value = nmem_strdup (m, tocode);
1087 data1_iconv_s (dh, m, n->child, t, wrbuf, tocode);
1091 const char *data1_get_encoding (data1_handle dh, data1_node *n)
1093 /* see if we have an xml header that specifies encoding */
1094 if (n && n->child && n->child->which == DATA1N_preprocess &&
1095 strcmp (n->child->u.preprocess.target, "xml") == 0)
1097 data1_xattr *xp = n->child->u.preprocess.attributes;
1098 for (; xp; xp = xp->next)
1099 if (!strcmp (xp->name, "encoding") == 0)
1102 /* no encoding in header, so see if "encoding" was specified for abs */
1103 if (n && n->which == DATA1N_root &&
1104 n->u.root.absyn && n->u.root.absyn->encoding)
1105 return n->u.root.absyn->encoding;
1106 /* none of above, return a hard coded default */
1107 return "ISO-8859-1";
1110 int data1_iconv (data1_handle dh, NMEM m, data1_node *n,
1112 const char *fromcode)
1114 if (yaz_matchstr (tocode, fromcode))
1116 WRBUF wrbuf = wrbuf_alloc();
1117 yaz_iconv_t t = yaz_iconv_open(tocode, fromcode);
1120 wrbuf_destroy(wrbuf);
1123 data1_iconv_s(dh, m, n, t, wrbuf, tocode);
1125 wrbuf_destroy(wrbuf);
1130 void data1_concat_text(data1_handle dh, NMEM m, data1_node *n)
1132 for (; n; n = n->next)
1134 if (n->which == DATA1N_data && n->next &&
1135 n->next->which == DATA1N_data)
1141 for (np = n; np && np->which == DATA1N_data; np=np->next)
1142 sz += np->u.data.len;
1143 ndata = nmem_malloc(m, sz);
1144 for (np = n; np && np->which == DATA1N_data; np=np->next)
1146 memcpy(ndata+off, np->u.data.data, np->u.data.len);
1147 off += np->u.data.len;
1149 n->u.data.data = ndata;
1152 if (!np && n->parent)
1153 n->parent->last_child = n;
1156 data1_concat_text(dh, m, n->child);
1162 * indent-tabs-mode: nil
1164 * vim: shiftwidth=4 tabstop=8 expandtab