src/marc_read_line.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_line.c
   8  * \brief Implements reading of MARC in line format
   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 <assert.h>
  20 #include <stdio.h>
  21 #include <string.h>
  22 #include <ctype.h>
  23
  24 #include <yaz/marcdisp.h>
  25 #include <yaz/wrbuf.h>
  26 #include <yaz/yaz-util.h>
  27
  28 int yaz_gets(int (*getbyte)(void *client_data),
  29              void (*ungetbyte)(int b, void *client_data),
  30              void *client_data,
  31              WRBUF w)
  32 {
  33     size_t sz = 0;
  34     int ch = getbyte(client_data);
  35
  36     while (ch != '\0' && ch != '\r' && ch != '\n')
  37     {
  38         wrbuf_putc(w, ch);
  39         sz++;
  40         ch = getbyte(client_data);
  41     }
  42     if (ch == '\r')
  43     {
  44         ch = getbyte(client_data);
  45         if (ch != '\n' && ch != '\0')
  46             ungetbyte(ch, client_data);
  47     }
  48     else if (ch == '\n')
  49     {
  50         ch = getbyte(client_data);
  51         if (ch != '\r' && ch != '\0')
  52             ungetbyte(ch, client_data);
  53     }
  54     if (sz)
  55     {
  56         return 1;
  57     }
  58     return 0;
  59 }
  60
  61 static int yaz_marc_line_gets(int (*getbyte)(void *client_data),
  62                               void (*ungetbyte)(int b, void *client_data),
  63                               void *client_data,
  64                               WRBUF w)
  65 {
  66     int more;
  67
  68     wrbuf_rewind(w);
  69     more = yaz_gets(getbyte, ungetbyte, client_data, w);
  70     if (!more)
  71         return 0;
  72
  73     while (more)
  74     {
  75         int i;
  76         for (i = 0; i<4; i++)
  77         {
  78             int ch = getbyte(client_data);
  79             if (ch != ' ')
  80             {
  81                 if (ch)
  82                     ungetbyte(ch, client_data);
  83                 return 1;
  84             }
  85         }
  86         if (wrbuf_len(w) > 60 && wrbuf_buf(w)[wrbuf_len(w)-1] == '=')
  87             wrbuf_cut_right(w, 1);
  88         else
  89             wrbuf_puts(w, " ");
  90         more = yaz_gets(getbyte, ungetbyte, client_data, w);
  91     }
  92     return 1;
  93 }
  94
  95
  96 int yaz_marc_read_line(yaz_marc_t mt,
  97                        int (*getbyte)(void *client_data),
  98                        void (*ungetbyte)(int b, void *client_data),
  99                        void *client_data)
 100 {
 101     int indicator_length;
 102     int identifier_length;
 103     int base_address;
 104     int length_data_entry;
 105     int length_starting;
 106     int length_implementation;
 107     int marker_ch = 0;
 108     int marker_skip = 0;
 109     int header_created = 0;
 110     WRBUF wrbuf_line = wrbuf_alloc();
 111
 112     yaz_marc_reset(mt);
 113
 114     while (yaz_marc_line_gets(getbyte, ungetbyte, client_data, wrbuf_line))
 115     {
 116         const char *line = wrbuf_cstr(wrbuf_line);
 117         int val;
 118         size_t line_len = strlen(line);
 119         if (line_len == 0)       /* empty line indicates end of record */
 120         {
 121             if (header_created)
 122                 break;
 123         }
 124         else if (line[0] == '$') /* indicates beginning/end of record */
 125         {
 126             if (header_created)
 127                 break;
 128         }
 129         else if (line[0] == '(') /* annotation, skip it */
 130             ;
 131         else if (line_len == 24 && atoi_n_check(line, 5, &val) && val >= 24)
 132         {
 133             /* deal with header lines:  00366nam  22001698a 4500
 134             */
 135
 136             if (header_created)
 137                 break;
 138             yaz_marc_set_leader(mt, line,
 139                                 &indicator_length,
 140                                 &identifier_length,
 141                                 &base_address,
 142                                 &length_data_entry,
 143                                 &length_starting,
 144                                 &length_implementation);
 145             header_created = 1;
 146         }
 147         else if (line_len > 4 && line[0] != ' ' && line[1] != ' '
 148                  && line[2] != ' ' && line[3] == ' ' )
 149         {
 150             /* deal with data/control lines: 245 12 ........ */
 151             char tag[4];
 152             const char *datafield_start = line+6;
 153             marker_ch = 0;
 154             marker_skip = 0;
 155
 156             memcpy(tag, line, 3);
 157             tag[3] = '\0';
 158             if (line_len >= 8) /* control - or datafield ? */
 159             {
 160                 if (*datafield_start == ' ')
 161                     datafield_start++;  /* skip blank after indicator */
 162
 163                 if (strchr("$_*", *datafield_start))
 164                 {
 165                     marker_ch = *datafield_start;
 166                     if (datafield_start[2] == ' ')
 167                         marker_skip = 1; /* subfields has blank before data */
 168                 }
 169             }
 170             if (!header_created)
 171             {
 172                 const char *leader = "01000cam  2200265 i 4500";
 173
 174                 yaz_marc_set_leader(mt, leader,
 175                                     &indicator_length,
 176                                     &identifier_length,
 177                                     &base_address,
 178                                     &length_data_entry,
 179                                     &length_starting,
 180                                     &length_implementation);
 181                 header_created = 1;
 182             }
 183
 184             if (marker_ch == 0)
 185             {   /* control field */
 186                 yaz_marc_add_controlfield(mt, tag, line+4, strlen(line+4));
 187             }
 188             else
 189             {   /* data field */
 190                 const char *indicator = line+4;
 191                 int indicator_len = 2;
 192                 const char *cp = datafield_start;
 193
 194                 yaz_marc_add_datafield(mt, tag, indicator, indicator_len);
 195                 for (;;)
 196                 {
 197                     const char *next;
 198                     size_t len;
 199
 200                     assert(cp[0] == marker_ch);
 201                     cp++;
 202                     next = cp;
 203                     while ((next = strchr(next, marker_ch)))
 204                     {
 205                         if ((next[1] >= 'A' && next[1] <= 'Z')
 206                             ||(next[1] >= 'a' && next[1] <= 'z'))
 207                         {
 208                             if (!marker_skip)
 209                                 break;
 210                             else if (next[2] == ' ')
 211                                 break;
 212                         }
 213                         next++;
 214                     }
 215                     len = strlen(cp);
 216                     if (next)
 217                         len = next - cp - marker_skip;
 218
 219                     if (marker_skip)
 220                     {
 221                         /* remove ' ' after subfield marker */
 222                         char *cp_blank = strchr(cp, ' ');
 223                         if (cp_blank)
 224                         {
 225                             len--;
 226                             while (cp_blank != cp)
 227                             {
 228                                 cp_blank[0] = cp_blank[-1];
 229                                 cp_blank--;
 230                             }
 231                             cp++;
 232                         }
 233                     }
 234                     yaz_marc_add_subfield(mt, cp, len);
 235                     if (!next)
 236                         break;
 237                     cp = next;
 238                 }
 239             }
 240         }
 241         else
 242         {
 243             yaz_marc_cprintf(mt, "Ignoring line: %s", line);
 244         }
 245     }
 246     wrbuf_destroy(wrbuf_line);
 247     if (!header_created)
 248         return -1;
 249     return 0;
 250 }
 251
 252 /*
 253  * Local variables:
 254  * c-basic-offset: 4
 255  * c-file-style: "Stroustrup"
 256  * indent-tabs-mode: nil
 257  * End:
 258  * vim: shiftwidth=4 tabstop=8 expandtab
 259  */
 260