src/marc_read_iso2709.c

   1 /* This file is part of the YAZ toolkit.
   2  * Copyright (C) 1995-2010 Index Data
   3  * See the file LICENSE for details.
   4  */
   5
   6 /**
   7  * \file marc_read_iso2709.c
   8  * \brief Implements reading of MARC as ISO2709
   9  */
  10
  11 #if HAVE_CONFIG_H
  12 #include <config.h>
  13 #endif
  14
  15 #ifdef WIN32
  16 #include <windows.h>
  17 #endif
  18
  19 #include <stdio.h>
  20 #include <string.h>
  21 #include <ctype.h>
  22 #include <yaz/marcdisp.h>
  23 #include <yaz/wrbuf.h>
  24 #include <yaz/yaz-util.h>
  25
  26 int yaz_marc_read_iso2709(yaz_marc_t mt, const char *buf, int bsize)
  27 {
  28     int entry_p;
  29     int record_length;
  30     int indicator_length;
  31     int identifier_length;
  32     int end_of_directory;
  33     int base_address;
  34     int length_data_entry;
  35     int length_starting;
  36     int length_implementation;
  37
  38     yaz_marc_reset(mt);
  39
  40     record_length = atoi_n (buf, 5);
  41     if (record_length < 25)
  42     {
  43         yaz_marc_cprintf(mt, "Record length %d < 24", record_length);
  44         return -1;
  45     }
  46     /* ballout if bsize is known and record_length is less than that */
  47     if (bsize != -1 && record_length > bsize)
  48     {
  49         yaz_marc_cprintf(mt, "Record appears to be larger than buffer %d < %d",
  50                          record_length, bsize);
  51         return -1;
  52     }
  53     if (yaz_marc_get_debug(mt))
  54         yaz_marc_cprintf(mt, "Record length         %5d", record_length);
  55
  56     yaz_marc_set_leader(mt, buf,
  57                         &indicator_length,
  58                         &identifier_length,
  59                         &base_address,
  60                         &length_data_entry,
  61                         &length_starting,
  62                         &length_implementation);
  63
  64     /* First pass. determine length of directory & base of data */
  65     for (entry_p = 24; buf[entry_p] != ISO2709_FS; )
  66     {
  67         /* length of directory entry */
  68         int l = 3 + length_data_entry + length_starting;
  69         if (entry_p + l >= record_length)
  70         {
  71             yaz_marc_cprintf(mt, "Directory offset %d: end of record."
  72                              " Missing FS char", entry_p);
  73             return -1;
  74         }
  75         if (yaz_marc_get_debug(mt))
  76         {
  77             WRBUF hex = wrbuf_alloc();
  78
  79             wrbuf_puts(hex, "Tag ");
  80             wrbuf_write_escaped(hex, buf + entry_p, 3);
  81             wrbuf_puts(hex, ", length ");
  82             wrbuf_write_escaped(hex, buf + entry_p + 3,
  83                                 length_data_entry);
  84             wrbuf_puts(hex, ", starting ");
  85             wrbuf_write_escaped(hex, buf + entry_p + 3 + length_data_entry,
  86                                 length_starting);
  87             yaz_marc_cprintf(mt, "Directory offset %d: %s",
  88                              entry_p, wrbuf_cstr(hex));
  89             wrbuf_destroy(hex);
  90         }
  91         /* Check for digits in length+starting info */
  92         while (--l >= 3)
  93             if (!isdigit(*(const unsigned char *) (buf + entry_p+l)))
  94                 break;
  95         if (l >= 3)
  96         {
  97             WRBUF hex = wrbuf_alloc();
  98             /* Not all digits, so stop directory scan */
  99             wrbuf_write_escaped(hex, buf + entry_p,
 100                                 length_data_entry + length_starting + 3);
 101             yaz_marc_cprintf(mt, "Directory offset %d: Bad value for data"
 102                              " length and/or length starting (%s)", entry_p,
 103                              wrbuf_cstr(hex));
 104             wrbuf_destroy(hex);
 105             break;
 106         }
 107         entry_p += 3 + length_data_entry + length_starting;
 108     }
 109     end_of_directory = entry_p;
 110     if (base_address != entry_p+1)
 111     {
 112         yaz_marc_cprintf(mt, "Base address not at end of directory,"
 113                          " base %d, end %d", base_address, entry_p+1);
 114     }
 115
 116     /* Second pass. parse control - and datafields */
 117     for (entry_p = 24; entry_p != end_of_directory; )
 118     {
 119         int data_length;
 120         int data_offset;
 121         int end_offset;
 122         int i;
 123         char tag[4];
 124         int identifier_flag = 0;
 125         int entry_p0 = entry_p;
 126
 127         memcpy (tag, buf+entry_p, 3);
 128         entry_p += 3;
 129         tag[3] = '\0';
 130         data_length = atoi_n(buf+entry_p, length_data_entry);
 131         entry_p += length_data_entry;
 132         data_offset = atoi_n(buf+entry_p, length_starting);
 133         entry_p += length_starting;
 134         i = data_offset + base_address;
 135         end_offset = i+data_length-1;
 136
 137         if (data_length <= 0 || data_offset < 0)
 138             break;
 139
 140         if (yaz_marc_get_debug(mt))
 141         {
 142             yaz_marc_cprintf(mt, "Tag: %s. Directory offset %d: data-length %d,"
 143                              " data-offset %d",
 144                              tag, entry_p0, data_length, data_offset);
 145         }
 146         if (end_offset >= record_length)
 147         {
 148             yaz_marc_cprintf(mt, "Directory offset %d: Data out of bounds %d >= %d",
 149                              entry_p0, end_offset, record_length);
 150             break;
 151         }
 152
 153         if (memcmp (tag, "00", 2))
 154             identifier_flag = 1;  /* if not 00X assume subfields */
 155         else if (indicator_length < 4 && indicator_length > 0)
 156         {
 157             /* Danmarc 00X have subfields */
 158             if (buf[i + indicator_length] == ISO2709_IDFS)
 159                 identifier_flag = 1;
 160             else if (buf[i + indicator_length + 1] == ISO2709_IDFS)
 161                 identifier_flag = 2;
 162         }
 163
 164         if (identifier_flag)
 165         {
 166             /* datafield */
 167             i += identifier_flag-1;
 168             if (indicator_length)
 169             {
 170                 /* skip RS/FS bytes in indicator. They are not allowed there */
 171                 int j;
 172                 for (j = indicator_length; --j >= 0; )
 173                     if (buf[j+i] < ' ')
 174                     {
 175                         j++;
 176                         i += j;
 177                         end_offset += j;
 178                         yaz_marc_cprintf(mt, "Bad indicator data. "
 179                                          "Skipping %d bytes", j);
 180                         break;
 181                     }
 182                 yaz_marc_add_datafield(mt, tag, buf+i, indicator_length);
 183                 i += indicator_length;
 184             }
 185
 186             while (i < end_offset &&
 187                     buf[i] != ISO2709_RS && buf[i] != ISO2709_FS)
 188             {
 189                 int code_offset = i+1;
 190
 191                 i ++;
 192                 while (i < end_offset &&
 193                         buf[i] != ISO2709_RS && buf[i] != ISO2709_IDFS &&
 194                        buf[i] != ISO2709_FS)
 195                     i++;
 196                 if (i > code_offset)
 197                     yaz_marc_add_subfield(mt, buf+code_offset, i - code_offset);
 198             }
 199         }
 200         else
 201         {
 202             /* controlfield */
 203             int i0 = i;
 204             while (i < end_offset &&
 205                 buf[i] != ISO2709_RS && buf[i] != ISO2709_FS)
 206                 i++;
 207             yaz_marc_add_controlfield(mt, tag, buf+i0, i-i0);
 208         }
 209         if (i < end_offset)
 210         {
 211             yaz_marc_cprintf(mt, "Separator but not at end of field length=%d",
 212                     data_length);
 213         }
 214         if (buf[i] != ISO2709_RS && buf[i] != ISO2709_FS)
 215         {
 216             yaz_marc_cprintf(mt, "No separator at end of field length=%d",
 217                              data_length);
 218         }
 219     }
 220     return record_length;
 221 }
 222
 223 /*
 224  * Local variables:
 225  * c-basic-offset: 4
 226  * c-file-style: "Stroustrup"
 227  * indent-tabs-mode: nil
 228  * End:
 229  * vim: shiftwidth=4 tabstop=8 expandtab
 230  */
 231