1 /* $Id: marcread.c,v 1.21 2003-08-21 10:29:00 adam Exp $
2 Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
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 Zebra; see the file LICENSE.zebra. If not, write to the
19 Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
28 #include <yaz/yaz-util.h>
29 #include <yaz/marcdisp.h>
35 #define MARCOMP_DEBUG 0
37 static data1_node *grs_read_iso2709 (struct grs_read_info *p, int marc_xml)
43 int identifier_length;
45 int length_data_entry;
47 int length_implementation;
52 data1_node *res_root, *res_top;
54 data1_marctab *marctab;
56 if ((*p->readf)(p->fh, buf, 5) != 5)
58 record_length = atoi_n (buf, 5);
59 if (record_length < 25)
61 logf (LOG_WARN, "MARC record length < 25, is %d", record_length);
64 /* read remaining part - attempt to read one byte furhter... */
65 read_bytes = (*p->readf)(p->fh, buf+5, record_length-4);
66 if (read_bytes < record_length-5)
68 logf (LOG_WARN, "Couldn't read whole MARC record");
71 if (read_bytes == record_length - 4)
73 off_t cur_offset = (*p->tellf)(p->fh);
77 (*p->endf)(p->fh, cur_offset - 1);
80 res_root = data1_mk_root (p->dh, p->mem, absynName);
83 yaz_log (LOG_WARN, "cannot read MARC without an abstract syntax");
89 const char *attr[] = { "xmlns", "http://www.loc.gov/MARC21/slim", 0};
91 res_top = data1_mk_tag (p->dh, p->mem, "record", attr, res_root);
93 lead = data1_mk_tag(p->dh, p->mem, "leader", 0, res_top);
94 data1_mk_text_n(p->dh, p->mem, buf, 24, lead);
97 res_top = data1_mk_tag (p->dh, p->mem, absynName, 0, res_root);
99 if ((marctab = res_root->u.root.absyn->marc))
101 memcpy(marctab->leader, buf, 24);
102 memcpy(marctab->implementation_codes, buf+6, 4);
103 marctab->implementation_codes[4] = '\0';
104 memcpy(marctab->user_systems, buf+17, 3);
105 marctab->user_systems[3] = '\0';
108 if (marctab && marctab->force_indicator_length >= 0)
109 indicator_length = marctab->force_indicator_length;
111 indicator_length = atoi_n (buf+10, 1);
112 if (marctab && marctab->force_identifier_length >= 0)
113 identifier_length = marctab->force_identifier_length;
115 identifier_length = atoi_n (buf+11, 1);
116 base_address = atoi_n (buf+12, 4);
118 length_data_entry = atoi_n (buf+20, 1);
119 length_starting = atoi_n (buf+21, 1);
120 length_implementation = atoi_n (buf+22, 1);
122 for (entry_p = 24; buf[entry_p] != ISO2709_FS; )
123 entry_p += 3+length_data_entry+length_starting;
124 base_address = entry_p+1;
125 for (entry_p = 24; buf[entry_p] != ISO2709_FS; )
133 data1_node *parent = res_top;
135 memcpy (tag, buf+entry_p, 3);
142 res = data1_mk_tag_n (p->dh, p->mem, tag, 3, 0 /* attr */, parent);
145 fprintf (outf, "%s ", tag);
147 data_length = atoi_n (buf+entry_p, length_data_entry);
148 entry_p += length_data_entry;
149 data_offset = atoi_n (buf+entry_p, length_starting);
150 entry_p += length_starting;
151 i = data_offset + base_address;
152 end_offset = i+data_length-1;
154 if (memcmp (tag, "00", 2) && indicator_length)
156 /* generate indicator node */
159 const char *attr[10];
166 res = data1_mk_tag(p->dh, p->mem, "datafield", attr, res);
168 for (j = 0; j<indicator_length; j++)
170 char str1[18], str2[2];
171 sprintf (str1, "ind%d", j+1);
178 data1_tag_add_attr (p->dh, p->mem, res, attr);
186 res = data1_mk_tag_n (p->dh, p->mem,
187 buf+i, indicator_length, 0 /* attr */, res);
189 for (j = 0; j<indicator_length; j++)
190 fprintf (outf, "%c", buf[j+i]);
193 i += indicator_length;
199 const char *attr[10];
205 res = data1_mk_tag(p->dh, p->mem, "controlfield", attr, res);
209 /* traverse sub fields */
211 while (buf[i] != ISO2709_RS && buf[i] != ISO2709_FS && i < end_offset)
214 if (!memcmp(tag, "4", 1) && (!yaz_matchstr(absynName, "UNIMARC")||
215 !yaz_matchstr(absynName, "RUSMARC")))
219 data1_mk_tag_n (p->dh, p->mem,
220 buf+i+1, identifier_length-1,
221 0 /* attr */, parent);
222 i += identifier_length;
225 while (buf[i] != ISO2709_RS && buf[i] != ISO2709_IDFS &&
226 buf[i] != ISO2709_FS && i < end_offset)
230 if (!memcmp(buf+i+1, "1", 1) && i<end_offset)
238 } while (go && i < end_offset);
240 data1_mk_text_n (p->dh, p->mem, buf + i0, i - i0, res);
243 else if (memcmp (tag, "00", 2) && identifier_length)
252 for (j = 1; j<identifier_length && j < 9; j++)
253 code[j-1] = buf[i+j];
258 res = data1_mk_tag(p->dh, p->mem, "subfield",
263 res = data1_mk_tag_n (p->dh, p->mem,
264 buf+i+1, identifier_length-1,
265 0 /* attr */, parent);
268 fprintf (outf, " $");
269 for (j = 1; j<identifier_length; j++)
270 fprintf (outf, "%c", buf[j+i]);
273 i += identifier_length;
275 while (buf[i] != ISO2709_RS && buf[i] != ISO2709_IDFS &&
276 buf[i] != ISO2709_FS && i < end_offset)
279 fprintf (outf, "%c", buf[i]);
283 data1_mk_text_n (p->dh, p->mem, buf + i0, i - i0, res);
289 fprintf (outf, "%c", buf[i]);
296 data1_mk_text_n (p->dh, p->mem, buf + i0, i - i0, parent);
299 fprintf (outf, "\n");
301 fprintf (outf, "-- separator but not at end of field\n");
302 if (buf[i] != ISO2709_RS && buf[i] != ISO2709_FS)
303 fprintf (outf, "-- no separator at end of field\n");
309 * Locate some data under this node. This routine should handle variants
312 static char *get_data(data1_node *n, int *len)
318 if (n->which == DATA1N_data)
321 *len = n->u.data.len;
323 for (i = 0; i<*len; i++)
324 if (!d1_isspace(n->u.data.data[i]))
326 while (*len && d1_isspace(n->u.data.data[*len - 1]))
330 return n->u.data.data + i;
332 if (n->which == DATA1N_tag)
334 else if (n->which == DATA1N_data)
344 static data1_node *lookup_subfield(data1_node *node, const char *name)
348 for (p=node; p; p=p->next)
350 if (!yaz_matchstr(p->u.tag.tag, name))
355 static inline_subfield *lookup_inline_subfield(inline_subfield *pisf, const char *name)
359 for (p=pisf; p; p=p->next)
361 if (!yaz_matchstr(p->name, name))
366 static inline_subfield *cat_inline_subfield(mc_subfield *psf, char *buf, inline_subfield *pisf)
370 for (p = psf; p && pisf; p = p->next)
372 if (p->which == MC_SF)
374 inline_subfield *found = lookup_inline_subfield(pisf, p->name);
378 if (strcmp(p->prefix, "_"))
380 strcat(strcat(buf, " "), p->prefix);
382 if (p->interval.start == -1)
384 strcat(buf, found->data);
388 strncat(buf, found->data+p->interval.start,
389 p->interval.end-p->interval.start+1);
391 if (strcmp(p->suffix, "_"))
393 strcat(strcat(buf, p->suffix), " ");
396 logf(LOG_LOG, "cat_inline_subfield(): add subfield $%s", found->name);
401 else if (p->which == MC_SFVARIANT)
403 inline_subfield *next;
406 next = cat_inline_subfield(p->u.child, buf, pisf);
412 else if (p->which == MC_SFGROUP)
417 for (pp = p->u.child, found = 0; pp; pp = pp->next)
419 if (!yaz_matchstr(pisf->name, p->name))
428 pisf = cat_inline_subfield(p->u.child, buf, pisf);
435 static void cat_inline_field(mc_field *pf, char *buf, data1_node *subfield)
438 if (!pf || !subfield)
441 for (;subfield; subfield = subfield->next)
444 inline_field *pif = inline_parse(get_data(subfield,&len));
446 if (pif && !yaz_matchstr(pif->name, pf->name))
448 if (!pf->list && pif->list)
450 strcat(buf, pif->list->data);
460 ind1 = (pif->ind1[0] == ' ') ? '_':pif->ind1[0];
461 ind2 = (pif->ind2[0] == ' ') ? '_':pif->ind2[0];
463 if (((pf->ind1[0] == '.') || (ind1 == pf->ind1[0])) &&
464 ((pf->ind2[0] == '.') || (ind2 == pf->ind2[0])))
466 cat_inline_subfield(pf->list, buf, pif->list);
469 add separator for inline fields
478 logf(LOG_WARN, "In-line field %s missed -- indicators does not match", pif->name);
482 inline_destroy_field(pif);
485 logf(LOG_LOG, "cat_inline_field(): got buffer {%s}", buf);
488 static data1_node *cat_subfield(mc_subfield *psf, char *buf, data1_node *subfield)
492 for (p = psf; p && subfield; p = p->next)
494 if (p->which == MC_SF)
496 data1_node *found = lookup_subfield(subfield, p->name);
502 if (strcmp(p->prefix, "_"))
504 strcat(strcat(buf, " "), p->prefix);
509 cat_inline_field(p->u.in_line, buf, found);
511 else if (p->interval.start == -1)
513 strcat(buf, get_data(found, &len));
517 strncat(buf, get_data(found, &len)+p->interval.start,
518 p->interval.end-p->interval.start+1);
520 if (strcmp(p->suffix, "_"))
522 strcat(strcat(buf, p->suffix), " ");
525 logf(LOG_LOG, "cat_subfield(): add subfield $%s", found->u.tag.tag);
527 subfield = found->next;
530 else if (p->which == MC_SFVARIANT)
534 next = cat_subfield(p->u.child, buf, subfield);
535 if (next == subfield)
540 else if (p->which == MC_SFGROUP)
545 for (pp = p->u.child, found = 0; pp; pp = pp->next)
547 if (!yaz_matchstr(subfield->u.tag.tag, pp->name))
556 subfield = cat_subfield(p->u.child, buf, subfield);
563 static data1_node *cat_field(struct grs_read_info *p, mc_field *pf, char *buf, data1_node *field)
565 data1_node *subfield;
572 if (yaz_matchstr(field->u.tag.tag, pf->name))
575 subfield = field->child;
581 check subfield without indicators
584 if (!pf->list && subfield->which == DATA1N_data)
588 if (pf->interval.start == -1)
590 strcat(buf, get_data(field, &len));
594 strncat(buf, get_data(field, &len)+pf->interval.start,
595 pf->interval.end-pf->interval.start+1);
598 logf(LOG_LOG, "cat_field(): got buffer {%s}", buf);
607 ind1 = (subfield->u.tag.tag[0] == ' ') ? '_':subfield->u.tag.tag[0];
608 ind2 = (subfield->u.tag.tag[1] == ' ') ? '_':subfield->u.tag.tag[1];
611 ((pf->ind1[0] == '.') || (ind1 == pf->ind1[0])) &&
612 ((pf->ind2[0] == '.') || (ind2 == pf->ind2[0]))
616 logf(LOG_WARN, "Field %s missed -- does not match indicators", field->u.tag.tag);
621 subfield = subfield->child;
626 cat_subfield(pf->list, buf, subfield);
629 logf(LOG_LOG, "cat_field(): got buffer {%s}", buf);
634 static int is_empty(char *s)
645 static void parse_data1_tree(struct grs_read_info *p, const char *mc_stmnt, data1_node *root)
647 data1_marctab *marctab = root->u.root.absyn->marc;
648 data1_node *top = root->child;
654 c = mc_mk_context(mc_stmnt+3);
663 mc_destroy_context(c);
667 logf(LOG_LOG, "parse_data1_tree(): statement -{%s}", mc_stmnt);
669 if (!yaz_matchstr(pf->name, "ldr"))
673 logf(LOG_LOG,"parse_data1_tree(): try LEADER from {%d} to {%d} positions",
674 pf->interval.start, pf->interval.end);
676 new = data1_mk_tag_n(p->dh, p->mem, mc_stmnt, strlen(mc_stmnt), 0, top);
677 data1_mk_text_n(p->dh, p->mem, marctab->leader+pf->interval.start,
678 pf->interval.end-pf->interval.start+1, new);
686 if (!yaz_matchstr(field->u.tag.tag, pf->name))
691 logf(LOG_LOG, "parse_data1_tree(): try field {%s}", field->u.tag.tag);
694 field = cat_field(p, pf, buf, field);
696 for (pb = strtok(pb, "\n"); pb; pb = strtok(NULL, "\n"))
700 new = data1_mk_tag_n(p->dh, p->mem, mc_stmnt, strlen(mc_stmnt), 0, top);
701 data1_mk_text_n(p->dh, p->mem, pb, strlen(pb), new);
711 mc_destroy_field(pf);
712 mc_destroy_context(c);
715 data1_node *grs_read_marcxml(struct grs_read_info *p)
717 data1_node *root = grs_read_iso2709(p, 1);
723 for (e=root->u.root.absyn->main_elements; e; e=e->next)
725 data1_tag *tag = e->tag;
727 if (tag && tag->which == DATA1T_string &&
728 !yaz_matchstr(tag->value.string, "mc?"))
729 parse_data1_tree(p, tag->value.string, root);
735 data1_node *grs_read_marc(struct grs_read_info *p)
737 data1_node *root = grs_read_iso2709(p, 0);
743 for (e=root->u.root.absyn->main_elements; e; e=e->next)
745 data1_tag *tag = e->tag;
747 if (tag && tag->which == DATA1T_string &&
748 !yaz_matchstr(tag->value.string, "mc?"))
749 parse_data1_tree(p, tag->value.string, root);
753 static void *grs_init_marc(void)
758 static void grs_destroy_marc(void *clientData)
762 static struct recTypeGrs marc_type = {
769 RecTypeGrs recTypeGrs_marc = &marc_type;
771 static struct recTypeGrs marcxml_type = {
778 RecTypeGrs recTypeGrs_marcxml = &marcxml_type;