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 1.6 2007-03-18 13:00:37 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     while (ch != '\0' && ch != '\r' && ch != '\n')
  39     {
  40         wrbuf_putc(w, ch);
  41         sz++;
  42         ch = getbyte(client_data);
  43     }
  44     if (ch == '\r')
  45     {
  46         ch = getbyte(client_data);
  47         if (ch != '\n' && ch != '\0')
  48             ungetbyte(ch, client_data);
  49     }
  50     else if (ch == '\n')
  51     {
  52         ch = getbyte(client_data);
  53         if (ch != '\r' && ch != '\0')
  54             ungetbyte(ch, client_data);
  55     }
  56     if (sz)
  57     {
  58         return 1;
  59     }
  60     return 0;
  61 }
  62
  63 static int yaz_marc_line_gets(int (*getbyte)(void *client_data),
  64                               void (*ungetbyte)(int b, void *client_data),
  65                               void *client_data,
  66                               WRBUF w)
  67 {
  68     int more;
  69
  70     wrbuf_rewind(w);
  71     more = yaz_gets(getbyte, ungetbyte, client_data, w);
  72     if (!more)
  73         return 0;
  74
  75     while (more)
  76     {
  77         int i;
  78         for (i = 0; i<4; i++)
  79         {
  80             int ch = getbyte(client_data);
  81             if (ch != ' ')
  82             {
  83                 if (ch)
  84                     ungetbyte(ch, client_data);
  85                 return 1;
  86             }
  87         }
  88         if (wrbuf_len(w) > 60 && wrbuf_buf(w)[wrbuf_len(w)-1] == '=')
  89             wrbuf_cut_right(w, 1);
  90         else
  91             wrbuf_puts(w, " ");
  92         more = yaz_gets(getbyte, ungetbyte, client_data, w);
  93     }
  94     return 1;
  95 }
  96
  97
  98 int yaz_marc_read_line(yaz_marc_t mt,
  99                        int (*getbyte)(void *client_data),
 100                        void (*ungetbyte)(int b, void *client_data),
 101                        void *client_data)
 102 {
 103     int indicator_length;
 104     int identifier_length;
 105     int base_address;
 106     int length_data_entry;
 107     int length_starting;
 108     int length_implementation;
 109     int marker_ch = 0;
 110     int marker_skip = 0;
 111     int header_created = 0;
 112     WRBUF wrbuf_line = wrbuf_alloc();
 113
 114     yaz_marc_reset(mt);
 115
 116     while (yaz_marc_line_gets(getbyte, ungetbyte, client_data, wrbuf_line))
 117     {
 118         const char *line = wrbuf_cstr(wrbuf_line);
 119         int val;
 120         size_t line_len = strlen(line);
 121         if (line_len == 0)       /* empty line indicates end of record */
 122         {
 123             if (header_created)
 124                 break;
 125         }
 126         else if (line[0] == '$') /* indicates beginning/end of record */
 127         {
 128             if (header_created)
 129                 break;
 130         }
 131         else if (line[0] == '(') /* annotation, skip it */
 132             ;
 133         else if (line_len == 24 && atoi_n_check(line, 5, &val) && val >= 24)
 134         {
 135             /* deal with header lines:  00366nam  22001698a 4500
 136             */
 137
 138             if (header_created)
 139                 break;
 140             yaz_marc_set_leader(mt, line,
 141                                 &indicator_length,
 142                                 &identifier_length,
 143                                 &base_address,
 144                                 &length_data_entry,
 145                                 &length_starting,
 146                                 &length_implementation);
 147             header_created = 1;
 148         }
 149         else if (line_len > 5 && line[0] != ' ' && line[1] != ' '
 150                  && line[2] != ' ' && line[3] == ' ' )
 151         {
 152             /* deal with data/control lines: 245 12 ........ */
 153             char tag[4];
 154             const char *datafield_start = line+6;
 155             marker_ch = 0;
 156             marker_skip = 0;
 157
 158             memcpy(tag, line, 3);
 159             tag[3] = '\0';
 160             if (line_len >= 8) /* control - or datafield ? */
 161             {
 162                 if (*datafield_start == ' ')
 163                     datafield_start++;  /* skip blank after indicator */
 164
 165                 if (strchr("$_*", *datafield_start))
 166                 {
 167                     marker_ch = *datafield_start;
 168                     if (datafield_start[2] == ' ')
 169                         marker_skip = 1; /* subfields has blank before data */
 170                 }
 171             }
 172             if (!header_created)
 173             {
 174                 const char *leader = "01000cam  2200265 i 4500";
 175
 176                 yaz_marc_set_leader(mt, leader,
 177                                     &indicator_length,
 178                                     &identifier_length,
 179                                     &base_address,
 180                                     &length_data_entry,
 181                                     &length_starting,
 182                                     &length_implementation);
 183                 header_created = 1;
 184             }
 185
 186             if (marker_ch == 0)
 187             {   /* control field */
 188                 yaz_marc_add_controlfield(mt, tag, line+4, strlen(line+4));
 189             }
 190             else
 191             {   /* data field */
 192                 const char *indicator = line+4;
 193                 int indicator_len = 2;
 194                 const char *cp = datafield_start;
 195
 196                 yaz_marc_add_datafield(mt, tag, indicator, indicator_len);
 197                 for (;;)
 198                 {
 199                     const char *next;
 200                     size_t len;
 201
 202                     assert(cp[0] == marker_ch);
 203                     cp++;
 204                     next = cp;
 205                     while ((next = strchr(next, marker_ch)))
 206                     {
 207                         if ((next[1] >= 'A' && next[1] <= 'Z')
 208                             ||(next[1] >= 'a' && next[1] <= 'z'))
 209                         {
 210                             if (!marker_skip)
 211                                 break;
 212                             else if (next[2] == ' ')
 213                                 break;
 214                         }
 215                         next++;
 216                     }
 217                     len = strlen(cp);
 218                     if (next)
 219                         len = next - cp - marker_skip;
 220
 221                     if (marker_skip)
 222                     {
 223                         /* remove ' ' after subfield marker */
 224                         char *cp_blank = strchr(cp, ' ');
 225                         if (cp_blank)
 226                         {
 227                             len--;
 228                             while (cp_blank != cp)
 229                             {
 230                                 cp_blank[0] = cp_blank[-1];
 231                                 cp_blank--;
 232                             }
 233                             cp++;
 234                         }
 235                     }
 236                     assert(len >= 0);
 237                     yaz_marc_add_subfield(mt, cp, len);
 238                     if (!next)
 239                         break;
 240                     cp = next;
 241                 }
 242             }
 243         }
 244         else
 245         {
 246             yaz_marc_cprintf(mt, "Ignoring line: %s", line);
 247         }
 248     }
 249     wrbuf_destroy(wrbuf_line);
 250     if (!header_created)
 251         return -1;
 252     return 0;
 253 }
 254
 255 /*
 256  * Local variables:
 257  * c-basic-offset: 4
 258  * indent-tabs-mode: nil
 259  * End:
 260  * vim: shiftwidth=4 tabstop=8 expandtab
 261  */
 262