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 1.5 2007-03-17 09:13:29 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              WRBUF w)
  34 {
  35     size_t sz = 0;
  36     int ch = getbyte(client_data);
  37
  38     wrbuf_rewind(w);
  39     while (ch != '\0' && ch != '\r' && ch != '\n')
  40     {
  41         wrbuf_putc(w, ch);
  42         sz++;
  43         ch = getbyte(client_data);
  44     }
  45     if (ch == '\r')
  46     {
  47         ch = getbyte(client_data);
  48         if (ch != '\n' && ch != '\0')
  49             ungetbyte(ch, client_data);
  50     }
  51     else if (ch == '\n')
  52     {
  53         ch = getbyte(client_data);
  54         if (ch != '\r' && ch != '\0')
  55             ungetbyte(ch, client_data);
  56     }
  57     if (sz)
  58     {
  59         return 1;
  60     }
  61     return 0;
  62 }
  63
  64 int yaz_marc_read_line(yaz_marc_t mt,
  65                        int (*getbyte)(void *client_data),
  66                        void (*ungetbyte)(int b, void *client_data),
  67                        void *client_data)
  68 {
  69     int indicator_length;
  70     int identifier_length;
  71     int base_address;
  72     int length_data_entry;
  73     int length_starting;
  74     int length_implementation;
  75     int marker_ch = 0;
  76     int marker_skip = 0;
  77     int header_created = 0;
  78     WRBUF wrbuf_line = wrbuf_alloc();
  79
  80     yaz_marc_reset(mt);
  81
  82     while (yaz_gets(getbyte, ungetbyte, client_data, wrbuf_line))
  83     {
  84         const char *line = wrbuf_cstr(wrbuf_line);
  85         int val;
  86         size_t line_len = strlen(line);
  87         if (line_len == 0)       /* empty line indicates end of record */
  88         {
  89             if (header_created)
  90                 break;
  91         }
  92         else if (line[0] == '$') /* indicates beginning/end of record */
  93         {
  94             if (header_created)
  95                 break;
  96         }
  97         else if (line[0] == '(') /* annotation, skip it */
  98             ;
  99         else if (line_len == 24 && atoi_n_check(line, 5, &val) && val >= 24)
 100         {
 101             /* deal with header lines:  00366nam  22001698a 4500
 102             */
 103
 104             if (header_created)
 105                 break;
 106             yaz_marc_set_leader(mt, line,
 107                                 &indicator_length,
 108                                 &identifier_length,
 109                                 &base_address,
 110                                 &length_data_entry,
 111                                 &length_starting,
 112                                 &length_implementation);
 113             header_created = 1;
 114         }
 115         else if (line_len > 5 && memcmp(line, "    ", 4) == 0)
 116         {  /* continuation line */
 117             ;
 118         }
 119         else if (line_len > 5 && line[3] == ' ')
 120         {
 121             /* deal with data/control lines: 245 12 ........ */
 122             char tag[4];
 123             const char *datafield_start = line+6;
 124             marker_ch = 0;
 125             marker_skip = 0;
 126
 127             memcpy(tag, line, 3);
 128             tag[3] = '\0';
 129             if (line_len >= 8) /* control - or datafield ? */
 130             {
 131                 if (*datafield_start == ' ')
 132                     datafield_start++;  /* skip blank after indicator */
 133
 134                 if (strchr("$_*", *datafield_start))
 135                 {
 136                     marker_ch = *datafield_start;
 137                     if (datafield_start[2] == ' ')
 138                         marker_skip = 1; /* subfields has blank before data */
 139                 }
 140             }
 141             if (!header_created)
 142             {
 143                 const char *leader = "01000cam  2200265 i 4500";
 144
 145                 yaz_marc_set_leader(mt, leader,
 146                                     &indicator_length,
 147                                     &identifier_length,
 148                                     &base_address,
 149                                     &length_data_entry,
 150                                     &length_starting,
 151                                     &length_implementation);
 152                 header_created = 1;
 153             }
 154
 155             if (marker_ch == 0)
 156             {   /* control field */
 157                 yaz_marc_add_controlfield(mt, tag, line+4, strlen(line+4));
 158             }
 159             else
 160             {   /* data field */
 161                 const char *indicator = line+4;
 162                 int indicator_len = 2;
 163                 const char *cp = datafield_start;
 164
 165                 yaz_marc_add_datafield(mt, tag, indicator, indicator_len);
 166                 for (;;)
 167                 {
 168                     const char *next;
 169                     size_t len;
 170
 171                     assert(cp[0] == marker_ch);
 172                     cp++;
 173                     next = cp;
 174                     while ((next = strchr(next, marker_ch)))
 175                     {
 176                         if ((next[1] >= 'A' && next[1] <= 'Z')
 177                             ||(next[1] >= 'a' && next[1] <= 'z'))
 178                         {
 179                             if (!marker_skip)
 180                                 break;
 181                             else if (next[2] == ' ')
 182                                 break;
 183                         }
 184                         next++;
 185                     }
 186                     len = strlen(cp);
 187                     if (next)
 188                         len = next - cp - marker_skip;
 189
 190                     if (marker_skip)
 191                     {
 192                         /* remove ' ' after subfield marker */
 193                         char *cp_blank = strchr(cp, ' ');
 194                         if (cp_blank)
 195                         {
 196                             len--;
 197                             while (cp_blank != cp)
 198                             {
 199                                 cp_blank[0] = cp_blank[-1];
 200                                 cp_blank--;
 201                             }
 202                             cp++;
 203                         }
 204                     }
 205                     assert(len >= 0);
 206                     yaz_marc_add_subfield(mt, cp, len);
 207                     if (!next)
 208                         break;
 209                     cp = next;
 210                 }
 211             }
 212         }
 213     }
 214     wrbuf_destroy(wrbuf_line);
 215     if (!header_created)
 216         return -1;
 217     return 0;
 218 }
 219
 220 /*
 221  * Local variables:
 222  * c-basic-offset: 4
 223  * indent-tabs-mode: nil
 224  * End:
 225  * vim: shiftwidth=4 tabstop=8 expandtab
 226  */
 227