2 * Copyright (C) 1995-2005, Index Data ApS
3 * See the file LICENSE for details.
5 * $Id: marcdisp.c,v 1.20 2005-03-07 06:07:33 adam Exp $
10 * \brief Implements MARC display - and conversion utilities
20 #include <yaz/marcdisp.h>
21 #include <yaz/wrbuf.h>
22 #include <yaz/yaz-util.h>
33 yaz_marc_t yaz_marc_create(void)
35 yaz_marc_t mt = (yaz_marc_t) xmalloc(sizeof(*mt));
36 mt->xml = YAZ_MARC_LINE;
38 mt->m_wr = wrbuf_alloc();
40 strcpy(mt->subfield_str, " $");
41 strcpy(mt->endline_str, "\n");
45 void yaz_marc_subfield_str(yaz_marc_t mt, const char *s)
47 strncpy(mt->subfield_str, s, sizeof(mt->subfield_str)-1);
48 mt->subfield_str[sizeof(mt->subfield_str)-1] = '\0';
51 void yaz_marc_endline_str(yaz_marc_t mt, const char *s)
53 strncpy(mt->endline_str, s, sizeof(mt->endline_str)-1);
54 mt->endline_str[sizeof(mt->endline_str)-1] = '\0';
57 void yaz_marc_destroy(yaz_marc_t mt)
61 wrbuf_free (mt->m_wr, 1);
65 static void marc_cdata (yaz_marc_t mt, const char *buf, size_t len, WRBUF wr)
67 if (mt->xml == YAZ_MARC_ISO2709)
68 wrbuf_iconv_write(wr, mt->iconv_cd, buf, len);
69 else if (mt->xml == YAZ_MARC_LINE)
70 wrbuf_iconv_write(wr, mt->iconv_cd, buf, len);
72 wrbuf_iconv_write_cdata(wr, mt->iconv_cd, buf, len);
75 static int atoi_n_check(const char *buf, int size, int *val)
77 if (!isdigit(*(const unsigned char *) buf))
79 *val = atoi_n(buf, size);
83 int yaz_marc_decode_wrbuf (yaz_marc_t mt, const char *buf, int bsize, WRBUF wr)
88 int identifier_length;
91 int length_data_entry;
93 int length_implementation;
95 int produce_warnings = 0;
99 if (mt->xml == YAZ_MARC_SIMPLEXML || mt->xml == YAZ_MARC_OAIMARC
100 || mt->xml == YAZ_MARC_MARCXML || mt->xml == YAZ_MARC_XCHANGE)
101 produce_warnings = 1;
103 record_length = atoi_n (buf, 5);
104 if (record_length < 25)
107 wrbuf_printf(wr, "<!-- Record length %d - aborting -->\n",
111 memcpy(lead, buf, 24); /* se can modify the header for output */
113 /* ballout if bsize is known and record_length is less than that */
114 if (bsize != -1 && record_length > bsize)
116 if (!atoi_n_check(buf+10, 1, &indicator_length))
118 if (produce_warnings)
119 wrbuf_printf(wr, "<!-- Indicator length at offset 10 should hold a digit. Assuming 2 -->\n");
121 indicator_length = 2;
123 if (!atoi_n_check(buf+11, 1, &identifier_length))
125 if (produce_warnings)
126 wrbuf_printf(wr, "<!-- Identifier length at offset 11 should hold a digit. Assuming 2 -->\n");
128 identifier_length = 2;
130 if (!atoi_n_check(buf+12, 5, &base_address))
132 if (produce_warnings)
133 wrbuf_printf(wr, "<!-- Base address at offsets 12..16 should hold a number. Assuming 0 -->\n");
136 if (!atoi_n_check(buf+20, 1, &length_data_entry))
138 if (produce_warnings)
139 wrbuf_printf(wr, "<!-- Length data entry at offset 20 should hold a digit. Assuming 4 -->\n");
140 length_data_entry = 4;
143 if (!atoi_n_check(buf+21, 1, &length_starting))
145 if (produce_warnings)
146 wrbuf_printf(wr, "<!-- Length starting at offset 21 should hold a digit. Assuming 5 -->\n");
150 if (!atoi_n_check(buf+22, 1, &length_implementation))
152 if (produce_warnings)
153 wrbuf_printf(wr, "<!-- Length implementation at offset 22 should hold a digit. Assuming 0 -->\n");
154 length_implementation = 0;
158 if (mt->xml != YAZ_MARC_LINE)
164 case YAZ_MARC_ISO2709:
166 case YAZ_MARC_SIMPLEXML:
167 wrbuf_puts (wr, "<iso2709\n");
168 sprintf (str, " RecordStatus=\"%c\"\n", buf[5]);
169 wrbuf_puts (wr, str);
170 sprintf (str, " TypeOfRecord=\"%c\"\n", buf[6]);
171 wrbuf_puts (wr, str);
172 for (i = 1; i<=19; i++)
174 sprintf (str, " ImplDefined%d=\"%c\"\n", i, buf[6+i]);
175 wrbuf_puts (wr, str);
177 wrbuf_puts (wr, ">\n");
179 case YAZ_MARC_OAIMARC:
182 "<oai_marc xmlns=\"http://www.openarchives.org/OIA/oai_marc\""
184 " xmlns:xsi=\"http://www.w3.org/2000/10/XMLSchema-instance\""
186 " xsi:schemaLocation=\"http://www.openarchives.org/OAI/oai_marc.xsd\""
190 sprintf (str, " status=\"%c\" type=\"%c\" catForm=\"%c\">\n",
191 buf[5], buf[6], buf[7]);
192 wrbuf_puts (wr, str);
194 case YAZ_MARC_MARCXML:
197 "<record xmlns=\"http://www.loc.gov/MARC21/slim\">\n"
199 lead[9] = 'a'; /* set leader to signal unicode */
200 marc_cdata(mt, lead, 24, wr);
201 wrbuf_printf(wr, "</leader>\n");
203 case YAZ_MARC_XCHANGE:
206 "<record xmlns=\"http://www.bs.dk/standards/MarcXchange\">\n"
208 marc_cdata(mt, lead, 24, wr);
209 wrbuf_printf(wr, "</leader>\n");
218 wrbuf_puts (wr, "<!--\n");
219 sprintf (str, "Record length %5d\n", record_length);
220 wrbuf_puts (wr, str);
221 sprintf (str, "Indicator length %5d\n", indicator_length);
222 wrbuf_puts (wr, str);
223 sprintf (str, "Identifier length %5d\n", identifier_length);
224 wrbuf_puts (wr, str);
225 sprintf (str, "Base address %5d\n", base_address);
226 wrbuf_puts (wr, str);
227 sprintf (str, "Length data entry %5d\n", length_data_entry);
228 wrbuf_puts (wr, str);
229 sprintf (str, "Length starting %5d\n", length_starting);
230 wrbuf_puts (wr, str);
231 sprintf (str, "Length implementation %5d\n", length_implementation);
232 wrbuf_puts (wr, str);
234 wrbuf_puts (wr, "-->\n");
237 /* first pass. determine length of directory & base of data */
238 for (entry_p = 24; buf[entry_p] != ISO2709_FS; )
240 /* length of directory entry */
241 int l = 3 + length_data_entry + length_starting;
242 if (entry_p + l >= record_length)
244 wrbuf_printf (wr, "<!-- Directory offset %d: end of record. "
245 "Missing FS char -->\n", entry_p);
249 wrbuf_printf (wr, "<!-- Directory offset %d: Tag %.3s -->\n",
250 entry_p, buf+entry_p);
251 /* check for digits in length info */
253 if (!isdigit(*(const unsigned char *) (buf + entry_p+l)))
257 /* not all digits, so stop directory scan */
258 wrbuf_printf (wr, "<!-- Directory offset %d: Bad data for data "
259 "length and/or length starting -->\n", entry_p);
262 entry_p += 3 + length_data_entry + length_starting;
264 end_of_directory = entry_p;
265 if (base_address != entry_p+1)
267 if (produce_warnings)
268 wrbuf_printf (wr,"<!-- Base address not at end of directory, "
269 "base %d, end %d -->\n", base_address, entry_p+1);
271 if (mt->xml == YAZ_MARC_ISO2709)
273 WRBUF wr_head = wrbuf_alloc();
274 WRBUF wr_dir = wrbuf_alloc();
275 WRBUF wr_tmp = wrbuf_alloc();
278 /* second pass. create directory for ISO2709 output */
279 for (entry_p = 24; entry_p != end_of_directory; )
281 int data_length, data_offset, end_offset;
284 wrbuf_write(wr_dir, buf+entry_p, 3);
287 data_length = atoi_n (buf+entry_p, length_data_entry);
288 entry_p += length_data_entry;
289 data_offset = atoi_n (buf+entry_p, length_starting);
290 entry_p += length_starting;
291 i = data_offset + base_address;
292 end_offset = i+data_length-1;
294 if (data_length <= 0 || data_offset < 0 || end_offset >= record_length)
297 while (i < end_offset &&
298 buf[i] != ISO2709_RS && buf[i] != ISO2709_FS)
300 sz1 = 1+i - (data_offset + base_address);
303 sz2 = wrbuf_iconv_write(wr_tmp, mt->iconv_cd,
304 buf + data_offset+base_address, sz1);
305 wrbuf_rewind(wr_tmp);
309 wrbuf_printf(wr_dir, "%0*d", length_data_entry, sz2);
310 wrbuf_printf(wr_dir, "%0*d", length_starting, data_p);
313 wrbuf_putc(wr_dir, ISO2709_FS);
314 wrbuf_printf(wr_head, "%05d", data_p+1 + base_address);
315 wrbuf_write(wr_head, lead+5, 7);
316 wrbuf_printf(wr_head, "%05d", base_address);
317 wrbuf_write(wr_head, lead+17, 7);
319 wrbuf_write(wr, wrbuf_buf(wr_head), 24);
320 wrbuf_write(wr, wrbuf_buf(wr_dir), wrbuf_len(wr_dir));
321 wrbuf_free(wr_head, 1);
322 wrbuf_free(wr_dir, 1);
323 wrbuf_free(wr_tmp, 1);
325 /* third pass. create data output */
326 for (entry_p = 24; entry_p != end_of_directory; )
333 int identifier_flag = 0;
334 int entry_p0 = entry_p;
336 memcpy (tag, buf+entry_p, 3);
339 data_length = atoi_n (buf+entry_p, length_data_entry);
340 entry_p += length_data_entry;
341 data_offset = atoi_n (buf+entry_p, length_starting);
342 entry_p += length_starting;
343 i = data_offset + base_address;
344 end_offset = i+data_length-1;
346 if (data_length <= 0 || data_offset < 0)
351 wrbuf_printf(wr, "<!-- Directory offset %d: data-length %d, "
352 "data-offset %d -->\n",
353 entry_p0, data_length, data_offset);
355 if (end_offset >= record_length)
357 wrbuf_printf (wr,"<!-- Directory offset %d: Data out of bounds "
359 entry_p0, end_offset, record_length);
363 if (memcmp (tag, "00", 2))
364 identifier_flag = 1; /* if not 00X assume subfields */
365 else if (indicator_length < 4 && indicator_length > 0)
367 /* Danmarc 00X have subfields */
368 if (buf[i + indicator_length] == ISO2709_IDFS)
370 else if (buf[i + indicator_length + 1] == ISO2709_IDFS)
376 wrbuf_printf(wr, "<!-- identifier_flag = %d -->\n",
384 wrbuf_puts (wr, "Tag: ");
385 wrbuf_puts (wr, tag);
386 wrbuf_puts (wr, " ");
388 case YAZ_MARC_SIMPLEXML:
389 wrbuf_printf (wr, "<field tag=\"");
390 marc_cdata(mt, tag, strlen(tag), wr);
391 wrbuf_printf(wr, "\"");
393 case YAZ_MARC_OAIMARC:
395 wrbuf_printf (wr, " <varfield id=\"");
397 wrbuf_printf (wr, " <fixfield id=\"");
398 marc_cdata(mt, tag, strlen(tag), wr);
399 wrbuf_printf(wr, "\"");
401 case YAZ_MARC_MARCXML:
402 case YAZ_MARC_XCHANGE:
404 wrbuf_printf (wr, " <datafield tag=\"");
406 wrbuf_printf (wr, " <controlfield tag=\"");
407 marc_cdata(mt, tag, strlen(tag), wr);
408 wrbuf_printf(wr, "\"");
413 i += identifier_flag-1;
414 for (j = 0; j<indicator_length; j++, i++)
418 case YAZ_MARC_ISO2709:
419 wrbuf_putc(wr, buf[i]);
423 wrbuf_puts (wr, " Ind: ");
424 wrbuf_putc(wr, buf[i]);
426 case YAZ_MARC_SIMPLEXML:
427 wrbuf_printf(wr, " Indicator%d=\"", j+1);
428 marc_cdata(mt, buf+i, 1, wr);
429 wrbuf_printf(wr, "\"");
431 case YAZ_MARC_OAIMARC:
432 wrbuf_printf(wr, " i%d=\"", j+1);
433 marc_cdata(mt, buf+i, 1, wr);
434 wrbuf_printf(wr, "\"");
436 case YAZ_MARC_MARCXML:
437 case YAZ_MARC_XCHANGE:
438 wrbuf_printf(wr, " ind%d=\"", j+1);
439 marc_cdata(mt, buf+i, 1, wr);
440 wrbuf_printf(wr, "\"");
444 if (mt->xml == YAZ_MARC_SIMPLEXML || mt->xml == YAZ_MARC_MARCXML
445 || mt->xml == YAZ_MARC_OAIMARC || mt->xml == YAZ_MARC_XCHANGE)
447 wrbuf_puts (wr, ">");
449 wrbuf_puts (wr, "\n");
451 if (mt->xml == YAZ_MARC_LINE)
454 wrbuf_puts (wr, " Fields: ");
458 while (i < end_offset &&
459 buf[i] != ISO2709_RS && buf[i] != ISO2709_FS)
465 case YAZ_MARC_ISO2709:
467 wrbuf_iconv_write(wr, mt->iconv_cd,
468 buf+i, identifier_length);
469 i += identifier_length;
472 wrbuf_puts (wr, mt->subfield_str);
473 marc_cdata(mt, buf+i, identifier_length-1, wr);
474 i = i+identifier_length-1;
475 wrbuf_putc (wr, ' ');
477 case YAZ_MARC_SIMPLEXML:
478 wrbuf_puts (wr, " <subfield code=\"");
479 marc_cdata(mt, buf+i, identifier_length-1, wr);
480 i = i+identifier_length-1;
481 wrbuf_puts (wr, "\">");
483 case YAZ_MARC_OAIMARC:
484 wrbuf_puts (wr, " <subfield label=\"");
485 marc_cdata(mt, buf+i, identifier_length-1, wr);
486 i = i+identifier_length-1;
487 wrbuf_puts (wr, "\">");
489 case YAZ_MARC_MARCXML:
490 case YAZ_MARC_XCHANGE:
491 wrbuf_puts (wr, " <subfield code=\"");
492 marc_cdata(mt, buf+i, identifier_length-1, wr);
493 i = i+identifier_length-1;
494 wrbuf_puts (wr, "\">");
498 while (i < end_offset &&
499 buf[i] != ISO2709_RS && buf[i] != ISO2709_IDFS &&
500 buf[i] != ISO2709_FS)
502 marc_cdata(mt, buf + i0, i - i0, wr);
504 if (mt->xml == YAZ_MARC_ISO2709 && buf[i] != ISO2709_IDFS)
505 marc_cdata(mt, buf + i, 1, wr);
507 if (mt->xml == YAZ_MARC_SIMPLEXML ||
508 mt->xml == YAZ_MARC_MARCXML ||
509 mt->xml == YAZ_MARC_XCHANGE ||
510 mt->xml == YAZ_MARC_OAIMARC)
511 wrbuf_puts (wr, "</subfield>\n");
517 while (i < end_offset &&
518 buf[i] != ISO2709_RS && buf[i] != ISO2709_FS)
520 marc_cdata(mt, buf + i0, i - i0, wr);
521 if (mt->xml == YAZ_MARC_ISO2709)
522 marc_cdata(mt, buf + i, 1, wr);
524 if (mt->xml == YAZ_MARC_LINE)
525 wrbuf_puts (wr, mt->endline_str);
527 wrbuf_printf(wr, "<!-- separator but not at end of field length=%d-->\n", data_length);
528 if (buf[i] != ISO2709_RS && buf[i] != ISO2709_FS)
529 wrbuf_printf(wr, "<!-- no separator at end of field length=%d-->\n", data_length);
532 case YAZ_MARC_SIMPLEXML:
533 wrbuf_puts (wr, "</field>\n");
535 case YAZ_MARC_OAIMARC:
537 wrbuf_puts (wr, "</varfield>\n");
539 wrbuf_puts (wr, "</fixfield>\n");
541 case YAZ_MARC_MARCXML:
542 case YAZ_MARC_XCHANGE:
544 wrbuf_puts (wr, " </datafield>\n");
546 wrbuf_puts (wr, "</controlfield>\n");
555 case YAZ_MARC_SIMPLEXML:
556 wrbuf_puts (wr, "</iso2709>\n");
558 case YAZ_MARC_OAIMARC:
559 wrbuf_puts (wr, "</oai_marc>\n");
561 case YAZ_MARC_MARCXML:
562 case YAZ_MARC_XCHANGE:
563 wrbuf_puts (wr, "</record>\n");
565 case YAZ_MARC_ISO2709:
566 wrbuf_putc (wr, ISO2709_RS);
569 return record_length;
572 int yaz_marc_decode_buf (yaz_marc_t mt, const char *buf, int bsize,
573 char **result, int *rsize)
575 int r = yaz_marc_decode_wrbuf(mt, buf, bsize, mt->m_wr);
577 *result = wrbuf_buf(mt->m_wr);
579 *rsize = wrbuf_len(mt->m_wr);
583 void yaz_marc_xml(yaz_marc_t mt, int xmlmode)
589 void yaz_marc_debug(yaz_marc_t mt, int level)
595 void yaz_marc_iconv(yaz_marc_t mt, yaz_iconv_t cd)
601 int yaz_marc_decode(const char *buf, WRBUF wr, int debug, int bsize, int xml)
603 yaz_marc_t mt = yaz_marc_create();
608 r = yaz_marc_decode_wrbuf(mt, buf, bsize, wr);
609 yaz_marc_destroy(mt);
614 int marc_display_wrbuf (const char *buf, WRBUF wr, int debug, int bsize)
616 return yaz_marc_decode(buf, wr, debug, bsize, 0);
620 int marc_display_exl (const char *buf, FILE *outf, int debug, int bsize)
622 yaz_marc_t mt = yaz_marc_create();
626 r = yaz_marc_decode_wrbuf (mt, buf, bsize, mt->m_wr);
630 fwrite (wrbuf_buf(mt->m_wr), 1, wrbuf_len(mt->m_wr), outf);
631 yaz_marc_destroy(mt);
636 int marc_display_ex (const char *buf, FILE *outf, int debug)
638 return marc_display_exl (buf, outf, debug, -1);
642 int marc_display (const char *buf, FILE *outf)
644 return marc_display_ex (buf, outf, 0);