src/marc_read_xml.c

   1 /* This file is part of the YAZ toolkit.
   2  * Copyright (C) 1995-2013 Index Data
   3  * See the file LICENSE for details.
   4  */
   5
   6 /**
   7  * \file marc_read_xml.c
   8  * \brief Implements reading of MARC as XML
   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 <yaz/marcdisp.h>
  22 #include <yaz/wrbuf.h>
  23 #include <yaz/yaz-util.h>
  24 #include <yaz/nmem_xml.h>
  25
  26 #if YAZ_HAVE_XML2
  27 #include <libxml/tree.h>
  28 #endif
  29
  30 #if YAZ_HAVE_XML2
  31 int yaz_marc_read_xml_subfields(yaz_marc_t mt, const xmlNode *ptr)
  32 {
  33     NMEM nmem = yaz_marc_get_nmem(mt);
  34     for (; ptr; ptr = ptr->next)
  35     {
  36         if (ptr->type == XML_ELEMENT_NODE)
  37         {
  38             if (!strcmp((const char *) ptr->name, "subfield"))
  39             {
  40                 size_t ctrl_data_len = 0;
  41                 char *ctrl_data_buf = 0;
  42                 const xmlNode *p = 0, *ptr_code = 0;
  43                 struct _xmlAttr *attr;
  44                 for (attr = ptr->properties; attr; attr = attr->next)
  45                     if (!strcmp((const char *)attr->name, "code"))
  46                         ptr_code = attr->children;
  47                     else
  48                     {
  49                         yaz_marc_cprintf(
  50                             mt, "Bad attribute '%.80s' for 'subfield'",
  51                             attr->name);
  52                         return -1;
  53                     }
  54                 if (!ptr_code)
  55                 {
  56                     yaz_marc_cprintf(
  57                         mt, "Missing attribute 'code' for 'subfield'" );
  58                     return -1;
  59                 }
  60                 if (ptr_code->type == XML_TEXT_NODE)
  61                 {
  62                     ctrl_data_len =
  63                         strlen((const char *)ptr_code->content);
  64                 }
  65                 else
  66                 {
  67                     yaz_marc_cprintf(
  68                         mt, "Missing value for 'code' in 'subfield'" );
  69                     return -1;
  70                 }
  71                 for (p = ptr->children; p ; p = p->next)
  72                     if (p->type == XML_TEXT_NODE)
  73                         ctrl_data_len += strlen((const char *)p->content);
  74                 ctrl_data_buf = (char *) nmem_malloc(nmem, ctrl_data_len+1);
  75                 strcpy(ctrl_data_buf, (const char *)ptr_code->content);
  76                 for (p = ptr->children; p ; p = p->next)
  77                     if (p->type == XML_TEXT_NODE)
  78                         strcat(ctrl_data_buf, (const char *)p->content);
  79                 yaz_marc_add_subfield(mt, ctrl_data_buf, ctrl_data_len);
  80             }
  81             else
  82             {
  83                 yaz_marc_cprintf(
  84                     mt, "Expected element 'subfield', got '%.80s'", ptr->name);
  85                 return -1;
  86             }
  87         }
  88     }
  89     return 0;
  90 }
  91
  92 const char *tag_value_extract(const char *name, char tag_buffer[5])
  93 {
  94     size_t length = strlen(name);
  95     if (length == 3)
  96     {
  97         strcpy(tag_buffer, name);
  98         return tag_buffer;
  99     }
 100     return 0;
 101 }
 102
 103 // Given a xmlNode ptr,  extract a value from either a element name or from a given attribute
 104 char *element_attribute_value_extract(const xmlNode *ptr,
 105                                       const char *attribute_name,
 106                                       NMEM nmem)
 107 {
 108     const char *name = (const char *) ptr->name;
 109     size_t length = strlen(name);
 110     xmlAttr *attr;
 111     if (length > 1 )
 112         return nmem_strdup(nmem, name+1);
 113     // TODO Extract from attribute where matches attribute_name
 114     for (attr = ptr->properties; attr; attr = attr->next)
 115         if (!strcmp((const char *)attr->name, attribute_name))
 116             return nmem_text_node_cdata(attr->children, nmem);
 117     return 0;
 118 }
 119
 120
 121 int yaz_marc_read_turbo_xml_subfields(yaz_marc_t mt, const xmlNode *ptr)
 122 {
 123     for (; ptr; ptr = ptr->next)
 124     {
 125         if (ptr->type == XML_ELEMENT_NODE)
 126         {
 127             if (!strncmp((const char *) ptr->name, "s", 1))
 128             {
 129                 NMEM nmem = yaz_marc_get_nmem(mt);
 130                 xmlNode *p;
 131                 size_t ctrl_data_len = 0;
 132                 char *ctrl_data_buf = 0;
 133                 const char *tag_value = element_attribute_value_extract(ptr, "code", nmem);
 134                 if (!tag_value)
 135                 {
 136                     yaz_marc_cprintf(
 137                         mt, "Missing 'code' value for 'subfield'" );
 138                     return -1;
 139                 }
 140
 141                 ctrl_data_len = strlen((const char *) tag_value);
 142                 // Extract (length) from CDATA
 143                 for (p = ptr->children; p ; p = p->next)
 144                     if (p->type == XML_TEXT_NODE)
 145                         ctrl_data_len += strlen((const char *)p->content);
 146                 // Allocate memory for code value (1 character (can be multi-byte) and data
 147                 ctrl_data_buf = (char *) nmem_malloc(nmem, ctrl_data_len+1);
 148                 // Build a string with "<Code><data>"
 149                 strcpy(ctrl_data_buf, (const char *) tag_value);
 150                 for (p = ptr->children; p ; p = p->next)
 151                     if (p->type == XML_TEXT_NODE)
 152                         strcat(ctrl_data_buf, (const char *)p->content);
 153                 yaz_marc_add_subfield(mt, ctrl_data_buf, ctrl_data_len);
 154             }
 155             else
 156             {
 157                 yaz_marc_cprintf(
 158                     mt, "Expected element 'subfield', got '%.80s'", ptr->name);
 159                 return -1;
 160             }
 161         }
 162     }
 163     return 0;
 164 }
 165
 166
 167 static int yaz_marc_read_xml_leader(yaz_marc_t mt, const xmlNode **ptr_p,
 168                                     int *indicator_length)
 169 {
 170     int identifier_length;
 171     int base_address;
 172     int length_data_entry;
 173     int length_starting;
 174     int length_implementation;
 175     const char *leader = 0;
 176     const xmlNode *ptr = *ptr_p;
 177
 178     for(; ptr; ptr = ptr->next)
 179         if (ptr->type == XML_ELEMENT_NODE)
 180         {
 181             if ( !strcmp( (const char *) ptr->name, "leader") ||
 182                  (!strncmp((const char *) ptr->name, "l", 1) ))
 183             {
 184                 xmlNode *p = ptr->children;
 185                 for(; p; p = p->next)
 186                     if (p->type == XML_TEXT_NODE)
 187                         leader = (const char *) p->content;
 188             }
 189             break;
 190         }
 191     if (!leader)
 192     {
 193         yaz_marc_cprintf(mt, "Missing leader. Inserting fake leader");
 194         leader = "00000nam a22000000a 4500";
 195     }
 196     if (strlen(leader) != 24)
 197     {
 198         yaz_marc_cprintf(mt, "Bad length %d of leader data."
 199                          " Must have length of 24 characters", strlen(leader));
 200         return -1;
 201     }
 202     yaz_marc_set_leader(mt, leader,
 203                         indicator_length,
 204                         &identifier_length,
 205                         &base_address,
 206                         &length_data_entry,
 207                         &length_starting,
 208                         &length_implementation);
 209     *ptr_p = ptr;
 210     return 0;
 211 }
 212
 213 static int yaz_marc_read_xml_fields(yaz_marc_t mt, const xmlNode *ptr,
 214                                     int indicator_length)
 215 {
 216     for(; ptr; ptr = ptr->next)
 217         if (ptr->type == XML_ELEMENT_NODE)
 218         {
 219             if (!strcmp( (const char *) ptr->name, "controlfield"))
 220             {
 221                 const xmlNode *ptr_tag = 0;
 222                 struct _xmlAttr *attr;
 223                 for (attr = ptr->properties; attr; attr = attr->next)
 224                     if (!strcmp((const char *)attr->name, "tag"))
 225                         ptr_tag = attr->children;
 226                     else
 227                     {
 228                         yaz_marc_cprintf(
 229                             mt, "Bad attribute '%.80s' for 'controlfield'",
 230                             attr->name);
 231                         return -1;
 232                     }
 233                 if (!ptr_tag)
 234                 {
 235                     yaz_marc_cprintf(
 236                         mt, "Missing attribute 'tag' for 'controlfield'" );
 237                     return -1;
 238                 }
 239                 yaz_marc_add_controlfield_xml(mt, ptr_tag, ptr->children);
 240             }
 241             else if (!strcmp((const char *) ptr->name, "datafield"))
 242             {
 243                 char indstr[11]; /* 0(unused), 1,....9, + zero term */
 244                 const xmlNode *ptr_tag = 0;
 245                 struct _xmlAttr *attr;
 246                 int i;
 247                 for (i = 0; i < indicator_length; i++)
 248                     indstr[i] = ' ';
 249                 indstr[i] = '\0';
 250                 for (attr = ptr->properties; attr; attr = attr->next)
 251                     if (!strcmp((const char *)attr->name, "tag"))
 252                         ptr_tag = attr->children;
 253                     else if (strlen((const char *)attr->name) == 4 &&
 254                              !memcmp(attr->name, "ind", 3))
 255                     {
 256                         int no = atoi((const char *)attr->name + 3);
 257                         if (attr->children &&
 258                             attr->children->type == XML_TEXT_NODE &&
 259                             no <= indicator_length && no > 0 &&
 260                             attr->children->content[0])
 261                         {
 262                             indstr[no - 1] = attr->children->content[0];
 263                         }
 264                         else
 265                         {
 266                             yaz_marc_cprintf(
 267                                 mt, "Bad attribute '%.80s' for 'datafield'",
 268                                 attr->name);
 269                         }
 270                     }
 271                     else
 272                     {
 273                         yaz_marc_cprintf(
 274                             mt, "Bad attribute '%.80s' for 'datafield'",
 275                             attr->name);
 276                     }
 277                 if (!ptr_tag)
 278                 {
 279                     yaz_marc_cprintf(
 280                         mt, "Missing attribute 'tag' for 'datafield'" );
 281                     return -1;
 282                 }
 283                 yaz_marc_add_datafield_xml(mt, ptr_tag,
 284                                            indstr, indicator_length);
 285                 if (yaz_marc_read_xml_subfields(mt, ptr->children))
 286                     return -1;
 287             }
 288             else
 289             {
 290                 yaz_marc_cprintf(mt,
 291                                  "Expected element controlfield or datafield,"
 292                                  " got %.80s", ptr->name);
 293                 return -1;
 294             }
 295         }
 296     return 0;
 297 }
 298
 299
 300 static int yaz_marc_read_turbo_xml_fields(yaz_marc_t mt, const xmlNode *ptr,
 301                                           int indicator_length)
 302 {
 303     for(; ptr; ptr = ptr->next)
 304         if (ptr->type == XML_ELEMENT_NODE)
 305         {
 306             if (!strncmp( (const char *) ptr->name, "c", 1))
 307             {
 308                 NMEM nmem = yaz_marc_get_nmem(mt);
 309                 char *tag_value = element_attribute_value_extract(ptr, "tag", nmem);
 310                 if (!tag_value)
 311                 {
 312                     yaz_marc_cprintf(
 313                         mt, "Missing attribute 'tag' for 'controlfield'" );
 314                     return -1;
 315                 }
 316                 yaz_marc_add_controlfield_xml2(mt, tag_value, ptr->children);
 317             }
 318             else if (!strncmp((const char *) ptr->name, "d",1))
 319             {
 320                 struct _xmlAttr *attr;
 321                 NMEM nmem = yaz_marc_get_nmem(mt);
 322                 char *tag_value;
 323                 char *indstr = nmem_malloc(nmem, indicator_length + 1);
 324                 int i = 0;
 325                 for (i = 0; i < indicator_length; i++)
 326                     indstr[i] = ' ';
 327                 indstr[i] = '\0';
 328                 tag_value = element_attribute_value_extract(ptr, "tag", nmem);
 329                 if (!tag_value)
 330                 {
 331                     yaz_marc_cprintf(
 332                         mt, "Missing attribute 'tag' for 'datafield'" );
 333                     return -1;
 334                 }
 335                 for (attr = ptr->properties; attr; attr = attr->next)
 336                     if (strlen((const char *)attr->name) == 2 &&
 337                         attr->name[0] == 'i')
 338                     {
 339                         //extract indicator attribute from i#="Y" pattern
 340                         int no = atoi((const char *)attr->name + 1);
 341                         if (attr->children &&
 342                             attr->children->type == XML_TEXT_NODE &&
 343                             no <= indicator_length && no > 0 &&
 344                             attr->children->content[0])
 345                         {
 346                             indstr[no - 1] = attr->children->content[0];
 347                         }
 348                         else
 349                         {
 350                             yaz_marc_cprintf(
 351                                 mt, "Bad attribute '%.80s' for 'd'",attr->name);
 352                         }
 353                     }
 354                     else
 355                     {
 356                         yaz_marc_cprintf(
 357                             mt, "Bad attribute '%.80s' for 'd'", attr->name);
 358                     }
 359                 yaz_marc_add_datafield_xml2(mt, tag_value, indstr);
 360                 if (yaz_marc_read_turbo_xml_subfields(mt, ptr->children /*, indstr */))
 361                     return -1;
 362             }
 363             else
 364             {
 365                 yaz_marc_cprintf(mt,
 366                                  "Expected element controlfield or datafield,"
 367                                  " got %.80s", ptr->name);
 368                 return -1;
 369             }
 370         }
 371     return 0;
 372 }
 373
 374
 375 #endif
 376
 377 #if YAZ_HAVE_XML2
 378 int yaz_marc_read_xml(yaz_marc_t mt, const xmlNode *ptr)
 379 {
 380     int indicator_length = 0;
 381     int format = 0;
 382     yaz_marc_reset(mt);
 383
 384     for(; ptr; ptr = ptr->next)
 385         if (ptr->type == XML_ELEMENT_NODE)
 386         {
 387             if (!strcmp((const char *) ptr->name, "record"))
 388             {
 389                 format = YAZ_MARC_MARCXML;
 390                 break;
 391             }
 392             else if (!strcmp((const char *) ptr->name, "r"))
 393             {
 394                 format = YAZ_MARC_TURBOMARC;
 395                 break;
 396             }
 397             else
 398             {
 399                 yaz_marc_cprintf(
 400                     mt, "Unknown element '%.80s' in MARC XML reader",
 401                     ptr->name);
 402                 return -1;
 403             }
 404         }
 405     if (!ptr)
 406     {
 407         yaz_marc_cprintf(mt, "Missing element 'record' in MARC XML record");
 408         return -1;
 409     }
 410     /* ptr points to record node now */
 411     ptr = ptr->children;
 412     if (yaz_marc_read_xml_leader(mt, &ptr, &indicator_length))
 413         return -1;
 414
 415     switch (format)
 416     {
 417     case YAZ_MARC_MARCXML:
 418         return yaz_marc_read_xml_fields(mt, ptr->next, indicator_length);
 419     case YAZ_MARC_TURBOMARC:
 420         return yaz_marc_read_turbo_xml_fields(mt, ptr->next, indicator_length);
 421     }
 422     return -1;
 423 }
 424 #endif
 425
 426
 427 /*
 428  * Local variables:
 429  * c-basic-offset: 4
 430  * c-file-style: "Stroustrup"
 431  * indent-tabs-mode: nil
 432  * End:
 433  * vim: shiftwidth=4 tabstop=8 expandtab
 434  */
 435