2 * Copyright (C) 1995-2005, Index Data ApS
3 * See the file LICENSE for details.
5 * $Id: marcdisp.c,v 1.18 2005-02-25 17:04:45 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;
90 int length_data_entry;
92 int length_implementation;
94 int produce_warnings = 0;
98 if (mt->xml == YAZ_MARC_SIMPLEXML || mt->xml == YAZ_MARC_OAIMARC
99 || mt->xml == YAZ_MARC_MARCXML || mt->xml == YAZ_MARC_XCHANGE)
100 produce_warnings = 1;
102 record_length = atoi_n (buf, 5);
103 if (record_length < 25)
106 wrbuf_printf(wr, "<!-- Record length %d - aborting -->\n",
110 memcpy(lead, buf, 24); /* se can modify the header for output */
112 /* ballout if bsize is known and record_length is less than that */
113 if (bsize != -1 && record_length > bsize)
115 if (!atoi_n_check(buf+10, 1, &indicator_length))
117 if (produce_warnings)
118 wrbuf_printf(wr, "<!-- Indicator length at offset 10 should hold a digit. Assuming 2 -->\n");
120 indicator_length = 2;
122 if (!atoi_n_check(buf+11, 1, &identifier_length))
124 if (produce_warnings)
125 wrbuf_printf(wr, "<!-- Identifier length at offset 11 should hold a digit. Assuming 2 -->\n");
127 identifier_length = 2;
129 if (!atoi_n_check(buf+12, 5, &base_address))
131 if (produce_warnings)
132 wrbuf_printf(wr, "<!-- Base address at offsets 12..16 should hold a number. Assuming 0 -->\n");
135 if (!atoi_n_check(buf+20, 1, &length_data_entry))
137 if (produce_warnings)
138 wrbuf_printf(wr, "<!-- Length data entry at offset 20 should hold a digit. Assuming 4 -->\n");
139 length_data_entry = 4;
142 if (!atoi_n_check(buf+21, 1, &length_starting))
144 if (produce_warnings)
145 wrbuf_printf(wr, "<!-- Length starting at offset 21 should hold a digit. Assuming 5 -->\n");
149 if (!atoi_n_check(buf+22, 1, &length_implementation))
151 if (produce_warnings)
152 wrbuf_printf(wr, "<!-- Length implementation at offset 22 should hold a digit. Assuming 0 -->\n");
153 length_implementation = 0;
157 if (mt->xml != YAZ_MARC_LINE)
163 case YAZ_MARC_ISO2709:
165 case YAZ_MARC_SIMPLEXML:
166 wrbuf_puts (wr, "<iso2709\n");
167 sprintf (str, " RecordStatus=\"%c\"\n", buf[5]);
168 wrbuf_puts (wr, str);
169 sprintf (str, " TypeOfRecord=\"%c\"\n", buf[6]);
170 wrbuf_puts (wr, str);
171 for (i = 1; i<=19; i++)
173 sprintf (str, " ImplDefined%d=\"%c\"\n", i, buf[6+i]);
174 wrbuf_puts (wr, str);
176 wrbuf_puts (wr, ">\n");
178 case YAZ_MARC_OAIMARC:
181 "<oai_marc xmlns=\"http://www.openarchives.org/OIA/oai_marc\""
183 " xmlns:xsi=\"http://www.w3.org/2000/10/XMLSchema-instance\""
185 " xsi:schemaLocation=\"http://www.openarchives.org/OAI/oai_marc.xsd\""
189 sprintf (str, " status=\"%c\" type=\"%c\" catForm=\"%c\">\n",
190 buf[5], buf[6], buf[7]);
191 wrbuf_puts (wr, str);
193 case YAZ_MARC_MARCXML:
196 "<record xmlns=\"http://www.loc.gov/MARC21/slim\">\n"
198 lead[9] = 'a'; /* set leader to signal unicode */
199 marc_cdata(mt, lead, 24, wr);
200 wrbuf_printf(wr, "</leader>\n");
202 case YAZ_MARC_XCHANGE:
205 "<record xmlns=\"http://www.bs.dk/standards/MarcXchange\">\n"
207 marc_cdata(mt, lead, 24, wr);
208 wrbuf_printf(wr, "</leader>\n");
217 wrbuf_puts (wr, "<!--\n");
218 sprintf (str, "Record length %5d\n", record_length);
219 wrbuf_puts (wr, str);
220 sprintf (str, "Indicator length %5d\n", indicator_length);
221 wrbuf_puts (wr, str);
222 sprintf (str, "Identifier length %5d\n", identifier_length);
223 wrbuf_puts (wr, str);
224 sprintf (str, "Base address %5d\n", base_address);
225 wrbuf_puts (wr, str);
226 sprintf (str, "Length data entry %5d\n", length_data_entry);
227 wrbuf_puts (wr, str);
228 sprintf (str, "Length starting %5d\n", length_starting);
229 wrbuf_puts (wr, str);
230 sprintf (str, "Length implementation %5d\n", length_implementation);
231 wrbuf_puts (wr, str);
233 wrbuf_puts (wr, "-->\n");
236 /* first pass. determine length of directory & base of data */
237 for (entry_p = 24; buf[entry_p] != ISO2709_FS; )
239 entry_p += 3+length_data_entry+length_starting;
240 if (entry_p >= record_length)
243 if (base_address != entry_p+1)
245 if (produce_warnings)
246 wrbuf_printf (wr," <!-- Base address not at end of directory "
247 "base=%d end=%d -->\n", base_address, entry_p+1);
249 base_address = entry_p+1;
251 if (mt->xml == YAZ_MARC_ISO2709)
253 WRBUF wr_head = wrbuf_alloc();
254 WRBUF wr_dir = wrbuf_alloc();
255 WRBUF wr_tmp = wrbuf_alloc();
258 /* second pass. create directory for ISO2709 output */
259 for (entry_p = 24; buf[entry_p] != ISO2709_FS; )
261 int data_length, data_offset, end_offset;
264 wrbuf_write(wr_dir, buf+entry_p, 3);
267 data_length = atoi_n (buf+entry_p, length_data_entry);
268 entry_p += length_data_entry;
269 data_offset = atoi_n (buf+entry_p, length_starting);
270 entry_p += length_starting;
271 i = data_offset + base_address;
272 end_offset = i+data_length-1;
274 if (data_length <= 0 || data_offset < 0 || end_offset >= record_length)
277 while (i < end_offset &&
278 buf[i] != ISO2709_RS && buf[i] != ISO2709_FS)
280 sz1 = 1+i - (data_offset + base_address);
283 sz2 = wrbuf_iconv_write(wr_tmp, mt->iconv_cd,
284 buf + data_offset+base_address, sz1);
285 wrbuf_rewind(wr_tmp);
289 wrbuf_printf(wr_dir, "%0*d", length_data_entry, sz2);
290 wrbuf_printf(wr_dir, "%0*d", length_starting, data_p);
293 wrbuf_putc(wr_dir, ISO2709_FS);
294 wrbuf_printf(wr_head, "%05d", data_p+1 + base_address);
295 wrbuf_write(wr_head, lead+5, 7);
296 wrbuf_printf(wr_head, "%05d", base_address);
297 wrbuf_write(wr_head, lead+17, 7);
299 wrbuf_write(wr, wrbuf_buf(wr_head), 24);
300 wrbuf_write(wr, wrbuf_buf(wr_dir), wrbuf_len(wr_dir));
301 wrbuf_free(wr_head, 1);
302 wrbuf_free(wr_dir, 1);
303 wrbuf_free(wr_tmp, 1);
305 /* third pass. create data output */
306 for (entry_p = 24; buf[entry_p] != ISO2709_FS; )
313 int identifier_flag = 0;
316 memcpy (tag, buf+entry_p, 3);
320 data_length = atoi_n (buf+entry_p, length_data_entry);
321 entry_p += length_data_entry;
322 data_offset = atoi_n (buf+entry_p, length_starting);
323 entry_p += length_starting;
324 i = data_offset + base_address;
325 end_offset = i+data_length-1;
327 if (data_length <= 0 || data_offset < 0 || end_offset >= record_length)
329 if (produce_warnings)
330 wrbuf_printf (wr," <!-- Bad data-offset=%d or "
331 "data-length=%d -->\n",
332 data_length, data_offset);
338 wrbuf_printf(wr, "<!-- offset=%d data dlength=%d doffset=%d -->\n",
339 entry_p0, data_length, data_offset);
342 if (memcmp (tag, "00", 2))
343 identifier_flag = 1; /* if not 00X assume subfields */
344 else if (indicator_length < 4 && indicator_length > 0)
346 /* Danmarc 00X have subfields */
347 if (buf[i + indicator_length] == ISO2709_IDFS)
349 else if (buf[i + indicator_length + 1] == ISO2709_IDFS)
355 wrbuf_printf(wr, "<!-- identifier_flag = %d -->\n",
363 wrbuf_puts (wr, "Tag: ");
364 wrbuf_puts (wr, tag);
365 wrbuf_puts (wr, " ");
367 case YAZ_MARC_SIMPLEXML:
368 wrbuf_printf (wr, "<field tag=\"");
369 marc_cdata(mt, tag, strlen(tag), wr);
370 wrbuf_printf(wr, "\"");
372 case YAZ_MARC_OAIMARC:
374 wrbuf_printf (wr, " <varfield id=\"");
376 wrbuf_printf (wr, " <fixfield id=\"");
377 marc_cdata(mt, tag, strlen(tag), wr);
378 wrbuf_printf(wr, "\"");
380 case YAZ_MARC_MARCXML:
381 case YAZ_MARC_XCHANGE:
383 wrbuf_printf (wr, " <datafield tag=\"");
385 wrbuf_printf (wr, " <controlfield tag=\"");
386 marc_cdata(mt, tag, strlen(tag), wr);
387 wrbuf_printf(wr, "\"");
392 i += identifier_flag-1;
393 for (j = 0; j<indicator_length; j++, i++)
397 case YAZ_MARC_ISO2709:
398 wrbuf_putc(wr, buf[i]);
402 wrbuf_puts (wr, " Ind: ");
403 wrbuf_putc(wr, buf[i]);
405 case YAZ_MARC_SIMPLEXML:
406 wrbuf_printf(wr, " Indicator%d=\"", j+1);
407 marc_cdata(mt, buf+i, 1, wr);
408 wrbuf_printf(wr, "\"");
410 case YAZ_MARC_OAIMARC:
411 wrbuf_printf(wr, " i%d=\"", j+1);
412 marc_cdata(mt, buf+i, 1, wr);
413 wrbuf_printf(wr, "\"");
415 case YAZ_MARC_MARCXML:
416 case YAZ_MARC_XCHANGE:
417 wrbuf_printf(wr, " ind%d=\"", j+1);
418 marc_cdata(mt, buf+i, 1, wr);
419 wrbuf_printf(wr, "\"");
423 if (mt->xml == YAZ_MARC_SIMPLEXML || mt->xml == YAZ_MARC_MARCXML
424 || mt->xml == YAZ_MARC_OAIMARC || mt->xml == YAZ_MARC_XCHANGE)
426 wrbuf_puts (wr, ">");
428 wrbuf_puts (wr, "\n");
430 if (mt->xml == YAZ_MARC_LINE)
433 wrbuf_puts (wr, " Fields: ");
437 while (i < end_offset &&
438 buf[i] != ISO2709_RS && buf[i] != ISO2709_FS)
444 case YAZ_MARC_ISO2709:
446 wrbuf_iconv_write(wr, mt->iconv_cd,
447 buf+i, identifier_length);
448 i += identifier_length;
451 wrbuf_puts (wr, mt->subfield_str);
452 marc_cdata(mt, buf+i, identifier_length-1, wr);
453 i = i+identifier_length-1;
454 wrbuf_putc (wr, ' ');
456 case YAZ_MARC_SIMPLEXML:
457 wrbuf_puts (wr, " <subfield code=\"");
458 marc_cdata(mt, buf+i, identifier_length-1, wr);
459 i = i+identifier_length-1;
460 wrbuf_puts (wr, "\">");
462 case YAZ_MARC_OAIMARC:
463 wrbuf_puts (wr, " <subfield label=\"");
464 marc_cdata(mt, buf+i, identifier_length-1, wr);
465 i = i+identifier_length-1;
466 wrbuf_puts (wr, "\">");
468 case YAZ_MARC_MARCXML:
469 case YAZ_MARC_XCHANGE:
470 wrbuf_puts (wr, " <subfield code=\"");
471 marc_cdata(mt, buf+i, identifier_length-1, wr);
472 i = i+identifier_length-1;
473 wrbuf_puts (wr, "\">");
477 while (i < end_offset &&
478 buf[i] != ISO2709_RS && buf[i] != ISO2709_IDFS &&
479 buf[i] != ISO2709_FS)
481 marc_cdata(mt, buf + i0, i - i0, wr);
483 if (mt->xml == YAZ_MARC_ISO2709 && buf[i] != ISO2709_IDFS)
484 marc_cdata(mt, buf + i, 1, wr);
486 if (mt->xml == YAZ_MARC_SIMPLEXML ||
487 mt->xml == YAZ_MARC_MARCXML ||
488 mt->xml == YAZ_MARC_XCHANGE ||
489 mt->xml == YAZ_MARC_OAIMARC)
490 wrbuf_puts (wr, "</subfield>\n");
496 while (i < end_offset &&
497 buf[i] != ISO2709_RS && buf[i] != ISO2709_FS)
499 marc_cdata(mt, buf + i0, i - i0, wr);
500 if (mt->xml == YAZ_MARC_ISO2709)
501 marc_cdata(mt, buf + i, 1, wr);
503 if (mt->xml == YAZ_MARC_LINE)
504 wrbuf_puts (wr, mt->endline_str);
506 wrbuf_printf(wr, " <!-- separator but not at end of field length=%d-->\n", data_length);
507 if (buf[i] != ISO2709_RS && buf[i] != ISO2709_FS)
508 wrbuf_printf(wr, " <!-- no separator at end of field length=%d-->\n", data_length);
511 case YAZ_MARC_SIMPLEXML:
512 wrbuf_puts (wr, "</field>\n");
514 case YAZ_MARC_OAIMARC:
516 wrbuf_puts (wr, "</varfield>\n");
518 wrbuf_puts (wr, "</fixfield>\n");
520 case YAZ_MARC_MARCXML:
521 case YAZ_MARC_XCHANGE:
523 wrbuf_puts (wr, " </datafield>\n");
525 wrbuf_puts (wr, "</controlfield>\n");
534 case YAZ_MARC_SIMPLEXML:
535 wrbuf_puts (wr, "</iso2709>\n");
537 case YAZ_MARC_OAIMARC:
538 wrbuf_puts (wr, "</oai_marc>\n");
540 case YAZ_MARC_MARCXML:
541 case YAZ_MARC_XCHANGE:
542 wrbuf_puts (wr, "</record>\n");
544 case YAZ_MARC_ISO2709:
545 wrbuf_putc (wr, ISO2709_RS);
548 return record_length;
551 int yaz_marc_decode_buf (yaz_marc_t mt, const char *buf, int bsize,
552 char **result, int *rsize)
554 int r = yaz_marc_decode_wrbuf(mt, buf, bsize, mt->m_wr);
558 *result = wrbuf_buf(mt->m_wr);
560 *rsize = wrbuf_len(mt->m_wr);
565 void yaz_marc_xml(yaz_marc_t mt, int xmlmode)
571 void yaz_marc_debug(yaz_marc_t mt, int level)
577 void yaz_marc_iconv(yaz_marc_t mt, yaz_iconv_t cd)
583 int yaz_marc_decode(const char *buf, WRBUF wr, int debug, int bsize, int xml)
585 yaz_marc_t mt = yaz_marc_create();
590 r = yaz_marc_decode_wrbuf(mt, buf, bsize, wr);
591 yaz_marc_destroy(mt);
596 int marc_display_wrbuf (const char *buf, WRBUF wr, int debug, int bsize)
598 return yaz_marc_decode(buf, wr, debug, bsize, 0);
602 int marc_display_exl (const char *buf, FILE *outf, int debug, int bsize)
604 yaz_marc_t mt = yaz_marc_create();
608 r = yaz_marc_decode_wrbuf (mt, buf, bsize, mt->m_wr);
612 fwrite (wrbuf_buf(mt->m_wr), 1, wrbuf_len(mt->m_wr), outf);
613 yaz_marc_destroy(mt);
618 int marc_display_ex (const char *buf, FILE *outf, int debug)
620 return marc_display_exl (buf, outf, debug, -1);
624 int marc_display (const char *buf, FILE *outf)
626 return marc_display_ex (buf, outf, 0);