src/marc_read_line.c

   1 /*
   2  * Copyright (C) 1995-2007, Index Data ApS
   3  * See the file LICENSE for details.
   4  *
   5  * $Id: marc_read_line.c,v 1.3 2007-01-03 08:42:15 adam Exp $
   6  */
   7
   8 /**
   9  * \file marc_read_line.c
  10  * \brief Implements reading of MARC in line format
  11  */
  12
  13 #if HAVE_CONFIG_H
  14 #include <config.h>
  15 #endif
  16
  17 #ifdef WIN32
  18 #include <windows.h>
  19 #endif
  20
  21 #include <assert.h>
  22 #include <stdio.h>
  23 #include <string.h>
  24 #include <ctype.h>
  25
  26 #include <yaz/marcdisp.h>
  27 #include <yaz/wrbuf.h>
  28 #include <yaz/yaz-util.h>
  29
  30 int yaz_gets(int (*getbyte)(void *client_data),
  31              void (*ungetbyte)(int b, void *client_data),
  32              void *client_data,
  33              char *buf, size_t len)
  34 {
  35     size_t sz = 0;
  36     int ch = getbyte(client_data);
  37     while (ch != '\0' && ch != '\r' && ch != '\n')
  38     {
  39         if (sz < len-1)
  40             buf[sz++] = ch;
  41         ch = getbyte(client_data);
  42     }
  43     if (ch == '\r')
  44     {
  45         ch = getbyte(client_data);
  46         if (ch != '\n' && ch != '\0')
  47             ungetbyte(ch, client_data);
  48     }
  49     else if (ch == '\n')
  50     {
  51         ch = getbyte(client_data);
  52         if (ch != '\r' && ch != '\0')
  53             ungetbyte(ch, client_data);
  54     }
  55     buf[sz] = '\0';
  56     if (sz)
  57         return 1;
  58     return 0;
  59 }
  60
  61 int yaz_marc_read_line(yaz_marc_t mt,
  62                        int (*getbyte)(void *client_data),
  63                        void (*ungetbyte)(int b, void *client_data),
  64                        void *client_data)
  65 {
  66     int indicator_length;
  67     int identifier_length;
  68     int base_address;
  69     int length_data_entry;
  70     int length_starting;
  71     int length_implementation;
  72     int marker_ch = 0;
  73     int marker_skip = 0;
  74     int header_created = 0;
  75     char line[4096];
  76
  77     yaz_marc_reset(mt);
  78
  79     while (yaz_gets(getbyte, ungetbyte, client_data, line, sizeof(line)))
  80     {
  81         int val;
  82         size_t line_len = strlen(line);
  83         /* see if have leader lines of the form:
  84            00366nam  22001698a 4500
  85         */
  86         if (line_len == 0)       /* empty line indicates end of record */
  87         {
  88             if (header_created)
  89                 break;
  90         }
  91         else if (line[0] == '$') /* indicates beginning/end of record */
  92         {
  93             if (header_created)
  94                 break;
  95         }
  96         else if (line[0] == '(') /* annotation, skip it */
  97             ;
  98         else if (line_len == 24 && atoi_n_check(line, 5, &val) && val >= 24)
  99         {
 100             if (header_created)
 101                 break;
 102             yaz_marc_set_leader(mt, line,
 103                                 &indicator_length,
 104                                 &identifier_length,
 105                                 &base_address,
 106                                 &length_data_entry,
 107                                 &length_starting,
 108                                 &length_implementation);
 109             header_created = 1;
 110         }
 111         else if (line_len > 5 && memcmp(line, "    ", 4) == 0)
 112         {  /* continuation line */
 113             ;
 114         }
 115         else if (line_len > 5 && line[3] == ' ')
 116         {
 117             char tag[4];
 118             char *datafield_start = line+6;
 119             marker_ch = 0;
 120             marker_skip = 0;
 121
 122             memcpy(tag, line, 3);
 123             tag[3] = '\0';
 124             if (line_len >= 8) /* control - or datafield ? */
 125             {
 126                 if (*datafield_start == ' ')
 127                     datafield_start++;  /* skip blank after indicator */
 128
 129                 if (strchr("$_*", *datafield_start))
 130                 {
 131                     marker_ch = *datafield_start;
 132                     if (datafield_start[2] == ' ')
 133                         marker_skip = 1; /* subfields has blank before data */
 134                 }
 135             }
 136             if (!header_created)
 137             {
 138                 const char *leader = "01000cam  2200265 i 4500";
 139
 140                 yaz_marc_set_leader(mt, leader,
 141                                     &indicator_length,
 142                                     &identifier_length,
 143                                     &base_address,
 144                                     &length_data_entry,
 145                                     &length_starting,
 146                                     &length_implementation);
 147                 header_created = 1;
 148             }
 149
 150             if (marker_ch == 0)
 151             {   /* control field */
 152                 yaz_marc_add_controlfield(mt, tag, line+4, strlen(line+4));
 153             }
 154             else
 155             {   /* data field */
 156                 const char *indicator = line+4;
 157                 int indicator_len = 2;
 158                 char *cp = datafield_start;
 159
 160                 yaz_marc_add_datafield(mt, tag, indicator, indicator_len);
 161                 for (;;)
 162                 {
 163                     char *next;
 164                     size_t len;
 165
 166                     assert(cp[0] == marker_ch);
 167                     cp++;
 168                     next = cp;
 169                     while ((next = strchr(next, marker_ch)))
 170                     {
 171                         if ((next[1] >= 'A' && next[1] <= 'Z')
 172                             ||(next[1] >= 'a' && next[1] <= 'z'))
 173                         {
 174                             if (!marker_skip)
 175                                 break;
 176                             else if (next[2] == ' ')
 177                                 break;
 178                         }
 179                         next++;
 180                     }
 181                     len = strlen(cp);
 182                     if (next)
 183                         len = next - cp - marker_skip;
 184
 185                     if (marker_skip)
 186                     {
 187                         /* remove ' ' after subfield marker */
 188                         char *cp_blank = strchr(cp, ' ');
 189                         if (cp_blank)
 190                         {
 191                             len--;
 192                             while (cp_blank != cp)
 193                             {
 194                                 cp_blank[0] = cp_blank[-1];
 195                                 cp_blank--;
 196                             }
 197                             cp++;
 198                         }
 199                     }
 200                     assert(len >= 0);
 201                     assert(len < 399);
 202                     yaz_marc_add_subfield(mt, cp, len);
 203                     if (!next)
 204                         break;
 205                     cp = next;
 206                 }
 207             }
 208         }
 209     }
 210     if (!header_created)
 211         return -1;
 212     return 0;
 213 }
 214
 215 /*
 216  * Local variables:
 217  * c-basic-offset: 4
 218  * indent-tabs-mode: nil
 219  * End:
 220  * vim: shiftwidth=4 tabstop=8 expandtab
 221  */
 222