src/record_conv.c

   1 /*
   2  * Copyright (C) 2005-2006, Index Data ApS
   3  * See the file LICENSE for details.
   4  *
   5  * $Id: record_conv.c,v 1.12 2006-12-12 10:41:38 marc Exp $
   6  */
   7 /**
   8  * \file record_conv.c
   9  * \brief Record Conversions utility
  10  */
  11
  12 #if HAVE_CONFIG_H
  13 #include <config.h>
  14 #endif
  15
  16 #include <string.h>
  17 #include <yaz/yaz-iconv.h>
  18 #include <yaz/marcdisp.h>
  19 #include <yaz/record_conv.h>
  20 #include <yaz/wrbuf.h>
  21 #include <yaz/xmalloc.h>
  22 #include <yaz/nmem.h>
  23 #include <yaz/tpath.h>
  24
  25 #if YAZ_HAVE_XML2
  26 #include <libxml/parser.h>
  27 #include <libxml/tree.h>
  28 #include <libxml/xinclude.h>
  29 #if YAZ_HAVE_XSLT
  30 #include <libxslt/xsltutils.h>
  31 #include <libxslt/transform.h>
  32 #endif
  33 #if YAZ_HAVE_EXSLT
  34 #include <libexslt/exslt.h>
  35 #endif
  36
  37 /** \brief The internal structure for yaz_record_conv_t */
  38 struct yaz_record_conv_struct {
  39     /** \brief memory for configuration */
  40     NMEM nmem;
  41
  42     /** \brief conversion rules (allocated using NMEM) */
  43     struct yaz_record_conv_rule *rules;
  44
  45     /** \brief pointer to last conversion rule pointer in chain */
  46     struct yaz_record_conv_rule **rules_p;
  47
  48     /** \brief string buffer for error messages */
  49     WRBUF wr_error;
  50
  51     /** \brief path for opening files  */
  52     char *path;
  53 };
  54
  55 /** \brief tranformation types (rule types) */
  56 enum YAZ_RECORD_CONV_RULE
  57 {
  58     YAZ_RECORD_CONV_RULE_XSLT,
  59     YAZ_RECORD_CONV_RULE_MARC
  60 };
  61
  62
  63 /** \brief tranformation info (rule info) */
  64 struct yaz_record_conv_rule {
  65     enum YAZ_RECORD_CONV_RULE which;
  66     union {
  67 #if YAZ_HAVE_XSLT
  68         struct {
  69             xsltStylesheetPtr xsp;
  70         } xslt;
  71 #endif
  72         struct {
  73             yaz_iconv_t iconv_t;
  74             int input_format;
  75             int output_format;
  76         } marc;
  77     } u;
  78     struct yaz_record_conv_rule *next;
  79 };
  80
  81 /** \brief reset rules+configuration */
  82 static void yaz_record_conv_reset(yaz_record_conv_t p)
  83 {
  84
  85     struct yaz_record_conv_rule *r;
  86     for (r = p->rules; r; r = r->next)
  87     {
  88         if (r->which == YAZ_RECORD_CONV_RULE_MARC)
  89         {
  90             if (r->u.marc.iconv_t)
  91                 yaz_iconv_close(r->u.marc.iconv_t);
  92         }
  93 #if YAZ_HAVE_XSLT
  94         else if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
  95         {
  96             xsltFreeStylesheet(r->u.xslt.xsp);
  97         }
  98 #endif
  99     }
 100     wrbuf_rewind(p->wr_error);
 101     nmem_reset(p->nmem);
 102
 103     p->rules = 0;
 104
 105     p->rules_p = &p->rules;
 106 }
 107
 108 yaz_record_conv_t yaz_record_conv_create()
 109 {
 110     yaz_record_conv_t p = xmalloc(sizeof(*p));
 111     p->nmem = nmem_create();
 112     p->wr_error = wrbuf_alloc();
 113     p->rules = 0;
 114     p->path = 0;
 115
 116 #if YAZ_HAVE_EXSLT
 117     exsltRegisterAll();
 118 #endif
 119     yaz_record_conv_reset(p);
 120     return p;
 121 }
 122
 123 void yaz_record_conv_destroy(yaz_record_conv_t p)
 124 {
 125     if (p)
 126     {
 127         yaz_record_conv_reset(p);
 128         nmem_destroy(p->nmem);
 129         wrbuf_free(p->wr_error, 1);
 130         xfree(p->path);
 131         xfree(p);
 132     }
 133 }
 134
 135 /** \brief adds a rule */
 136 static struct yaz_record_conv_rule *add_rule(yaz_record_conv_t p,
 137                                              enum YAZ_RECORD_CONV_RULE type)
 138 {
 139     struct yaz_record_conv_rule *r = nmem_malloc(p->nmem, sizeof(*r));
 140     r->which = type;
 141     r->next = 0;
 142     *p->rules_p = r;
 143     p->rules_p = &r->next;
 144     return r;
 145 }
 146
 147 /** \brief parse 'xslt' conversion node */
 148 static int conv_xslt(yaz_record_conv_t p, const xmlNode *ptr)
 149 {
 150 #if YAZ_HAVE_XSLT
 151     struct _xmlAttr *attr;
 152     const char *stylesheet = 0;
 153
 154     for (attr = ptr->properties; attr; attr = attr->next)
 155     {
 156         if (!xmlStrcmp(attr->name, BAD_CAST "stylesheet") &&
 157             attr->children && attr->children->type == XML_TEXT_NODE)
 158             stylesheet = (const char *) attr->children->content;
 159         else
 160         {
 161             wrbuf_printf(p->wr_error, "Bad attribute '%s'"
 162                          "Expected stylesheet.", attr->name);
 163             return -1;
 164         }
 165     }
 166     if (!stylesheet)
 167     {
 168         wrbuf_printf(p->wr_error, "Element <xslt>: "
 169                      "attribute 'stylesheet' expected");
 170         return -1;
 171     }
 172     else
 173     {
 174         char fullpath[1024];
 175         xsltStylesheetPtr xsp;
 176         if (!yaz_filepath_resolve(stylesheet, p->path, 0, fullpath))
 177         {
 178             wrbuf_printf(p->wr_error, "Element <xslt stylesheet=\"%s\"/>:"
 179                          " could not locate stylesheet '%s' with path '%s'",
 180                          stylesheet, fullpath, p->path);
 181             return -1;
 182         }
 183         xsp = xsltParseStylesheetFile((xmlChar*) fullpath);
 184         if (!xsp)
 185         {
 186             wrbuf_printf(p->wr_error, "Element <xslt stylesheet=\"%s\"/>:"
 187                          " parsing stylesheet '%s' with path '%s' failed,"
 188 #if YAZ_HAVE_EXSLT
 189                          " EXSLT enabled",
 190 #else
 191                          " EXSLT not supported",
 192 #endif
 193                          stylesheet, fullpath, p->path);
 194             return -1;
 195         }
 196         else
 197         {
 198             struct yaz_record_conv_rule *r =
 199                 add_rule(p, YAZ_RECORD_CONV_RULE_XSLT);
 200             r->u.xslt.xsp = xsp;
 201         }
 202     }
 203     return 0;
 204 #else
 205     wrbuf_printf(p->wr_error, "xslt unsupported."
 206                  " YAZ compiled without XSLT support");
 207     return -1;
 208 #endif
 209 }
 210
 211 /** \brief parse 'marc' conversion node */
 212 static int conv_marc(yaz_record_conv_t p, const xmlNode *ptr)
 213 {
 214     struct _xmlAttr *attr;
 215     const char *input_charset = 0;
 216     const char *output_charset = 0;
 217     const char *input_format = 0;
 218     const char *output_format = 0;
 219     int input_format_mode = 0;
 220     int output_format_mode = 0;
 221     struct yaz_record_conv_rule *r;
 222     yaz_iconv_t cd = 0;
 223
 224     for (attr = ptr->properties; attr; attr = attr->next)
 225     {
 226         if (!xmlStrcmp(attr->name, BAD_CAST "inputcharset") &&
 227             attr->children && attr->children->type == XML_TEXT_NODE)
 228             input_charset = (const char *) attr->children->content;
 229         else if (!xmlStrcmp(attr->name, BAD_CAST "outputcharset") &&
 230             attr->children && attr->children->type == XML_TEXT_NODE)
 231             output_charset = (const char *) attr->children->content;
 232         else if (!xmlStrcmp(attr->name, BAD_CAST "inputformat") &&
 233             attr->children && attr->children->type == XML_TEXT_NODE)
 234             input_format = (const char *) attr->children->content;
 235         else if (!xmlStrcmp(attr->name, BAD_CAST "outputformat") &&
 236             attr->children && attr->children->type == XML_TEXT_NODE)
 237             output_format = (const char *) attr->children->content;
 238         else
 239         {
 240             wrbuf_printf(p->wr_error, "Element <marc>: expected attributes"
 241                          "'inputformat', 'inputcharset', 'outputformat' or"
 242                          " 'outputcharset', got attribute '%s'",
 243                          attr->name);
 244             return -1;
 245         }
 246     }
 247     if (!input_format)
 248     {
 249         wrbuf_printf(p->wr_error, "Element <marc>: "
 250                      "attribute 'inputformat' required");
 251         return -1;
 252     }
 253     else if (!strcmp(input_format, "marc"))
 254     {
 255         input_format_mode = YAZ_MARC_ISO2709;
 256     }
 257     else if (!strcmp(input_format, "xml"))
 258     {
 259         input_format_mode = YAZ_MARC_MARCXML;
 260         /** Libxml2 generates UTF-8 encoding by default .
 261             So we convert from UTF-8 to outputcharset (if defined)
 262         */
 263         if (!input_charset && output_charset)
 264             input_charset = "utf-8";
 265     }
 266     else
 267     {
 268         wrbuf_printf(p->wr_error, "Element <marc inputformat='%s'>: "
 269                      " Unsupported input format"
 270                      " defined by attribute value",
 271                      input_format);
 272         return -1;
 273     }
 274
 275     if (!output_format)
 276     {
 277         wrbuf_printf(p->wr_error,
 278                      "Element <marc>: attribute 'outputformat' required");
 279         return -1;
 280     }
 281     else if (!strcmp(output_format, "line"))
 282     {
 283         output_format_mode = YAZ_MARC_LINE;
 284     }
 285     else if (!strcmp(output_format, "marcxml"))
 286     {
 287         output_format_mode = YAZ_MARC_MARCXML;
 288         if (input_charset && !output_charset)
 289             output_charset = "utf-8";
 290     }
 291     else if (!strcmp(output_format, "marc"))
 292     {
 293         output_format_mode = YAZ_MARC_ISO2709;
 294     }
 295     else if (!strcmp(output_format, "marcxchange"))
 296     {
 297         output_format_mode = YAZ_MARC_XCHANGE;
 298         if (input_charset && !output_charset)
 299             output_charset = "utf-8";
 300     }
 301     else
 302     {
 303         wrbuf_printf(p->wr_error, "Element <marc outputformat='%s'>: "
 304                      " Unsupported output format"
 305                      " defined by attribute value",
 306                      output_format);
 307         return -1;
 308     }
 309     if (input_charset && output_charset)
 310     {
 311         cd = yaz_iconv_open(output_charset, input_charset);
 312         if (!cd)
 313         {
 314             wrbuf_printf(p->wr_error,
 315                          "Element <marc inputcharset='%s' outputcharset='%s'>:"
 316                          " Unsupported character set mapping"
 317                          " defined by attribute values",
 318                          input_charset, output_charset);
 319             return -1;
 320         }
 321     }
 322     else if (input_charset)
 323     {
 324         wrbuf_printf(p->wr_error, "Element <marc>: "
 325                      "attribute 'outputcharset' missing");
 326         return -1;
 327     }
 328     else if (output_charset)
 329     {
 330         wrbuf_printf(p->wr_error, "Element <marc>: "
 331                      "attribute 'inputcharset' missing");
 332         return -1;
 333     }
 334     r = add_rule(p, YAZ_RECORD_CONV_RULE_MARC);
 335     r->u.marc.iconv_t = cd;
 336
 337     r->u.marc.input_format = input_format_mode;
 338     r->u.marc.output_format = output_format_mode;
 339     return 0;
 340 }
 341
 342 int yaz_record_conv_configure(yaz_record_conv_t p, const void *ptr_v)
 343 {
 344     const xmlNode *ptr = ptr_v;
 345
 346     yaz_record_conv_reset(p);
 347
 348     /* parsing element children */
 349     for (ptr = ptr->children; ptr; ptr = ptr->next)
 350         {
 351             if (ptr->type != XML_ELEMENT_NODE)
 352                 continue;
 353             if (!strcmp((const char *) ptr->name, "xslt"))
 354                 {
 355                     if (conv_xslt(p, ptr))
 356                         return -1;
 357                 }
 358             else if (!strcmp((const char *) ptr->name, "marc"))
 359                 {
 360                     if (conv_marc(p, ptr))
 361                         return -1;
 362                 }
 363             else
 364                 {
 365                     wrbuf_printf(p->wr_error, "Element <backend>: expected "
 366                                  "<marc> or <xslt> element, got <%s>"
 367                                  , ptr->name);
 368                     return -1;
 369                 }
 370         }
 371     return 0;
 372 }
 373
 374 int yaz_record_conv_record(yaz_record_conv_t p,
 375                            const char *input_record_buf,
 376                            size_t input_record_len,
 377                            WRBUF output_record)
 378 {
 379     int ret = 0;
 380     WRBUF record = output_record; /* pointer transfer */
 381     struct yaz_record_conv_rule *r = p->rules;
 382     wrbuf_rewind(p->wr_error);
 383
 384     wrbuf_write(record, input_record_buf, input_record_len);
 385     for (; ret == 0 && r; r = r->next)
 386     {
 387         if (r->which == YAZ_RECORD_CONV_RULE_MARC)
 388         {
 389             yaz_marc_t mt = yaz_marc_create();
 390
 391             yaz_marc_xml(mt, r->u.marc.output_format);
 392
 393             if (r->u.marc.iconv_t)
 394                 yaz_marc_iconv(mt, r->u.marc.iconv_t);
 395             if (r->u.marc.input_format == YAZ_MARC_ISO2709)
 396             {
 397                 int sz = yaz_marc_read_iso2709(mt, wrbuf_buf(record),
 398                                                wrbuf_len(record));
 399                 if (sz > 0)
 400                     ret = 0;
 401                 else
 402                     ret = -1;
 403             }
 404             else if (r->u.marc.input_format == YAZ_MARC_MARCXML)
 405             {
 406                 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
 407                                                wrbuf_len(record));
 408                 if (!doc)
 409                 {
 410                     wrbuf_printf(p->wr_error, "xmlParseMemory failed");
 411                     ret = -1;
 412                 }
 413                 else
 414                 {
 415                     ret = yaz_marc_read_xml(mt, xmlDocGetRootElement(doc));
 416                     if (ret)
 417                         wrbuf_printf(p->wr_error, "yaz_marc_read_xml failed");
 418                 }
 419                 xmlFreeDoc(doc);
 420             }
 421             else
 422             {
 423                 wrbuf_printf(p->wr_error, "unsupported input format");
 424                 ret = -1;
 425             }
 426             if (ret == 0)
 427             {
 428                 wrbuf_rewind(record);
 429                 ret = yaz_marc_write_mode(mt, record);
 430                 if (ret)
 431                     wrbuf_printf(p->wr_error, "yaz_marc_write_mode failed");
 432             }
 433             yaz_marc_destroy(mt);
 434         }
 435 #if YAZ_HAVE_XSLT
 436         else if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
 437         {
 438             xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
 439                                            wrbuf_len(record));
 440             if (!doc)
 441             {
 442                 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
 443                 ret = -1;
 444             }
 445             else
 446             {
 447                 xmlDocPtr res = xsltApplyStylesheet(r->u.xslt.xsp, doc, 0);
 448                 if (res)
 449                 {
 450                     xmlChar *out_buf = 0;
 451                     int out_len;
 452
 453 #if YAZ_HAVE_XSLTSAVERESULTTOSTRING
 454                     xsltSaveResultToString(&out_buf, &out_len, res,
 455                                            r->u.xslt.xsp);
 456 #else
 457                     xmlDocDumpFormatMemory (res, &out_buf, &out_len, 1);
 458 #endif
 459                     if (!out_buf)
 460                     {
 461                         wrbuf_printf(p->wr_error,
 462                                      "xsltSaveResultToString failed");
 463                         ret = -1;
 464                     }
 465                     else
 466                     {
 467                         wrbuf_rewind(record);
 468                         wrbuf_write(record, (const char *) out_buf, out_len);
 469
 470                         xmlFree(out_buf);
 471                     }
 472                     xmlFreeDoc(res);
 473                 }
 474                 else
 475                 {
 476                     wrbuf_printf(p->wr_error, "xsltApplyStylesheet failed");
 477                     ret = -1;
 478                 }
 479                 xmlFreeDoc(doc);
 480             }
 481         }
 482 #endif
 483     }
 484     return ret;
 485 }
 486
 487 const char *yaz_record_conv_get_error(yaz_record_conv_t p)
 488 {
 489     return wrbuf_buf(p->wr_error);
 490 }
 491
 492 void yaz_record_conv_set_path(yaz_record_conv_t p, const char *path)
 493 {
 494     xfree(p->path);
 495     p->path = 0;
 496     if (path)
 497         p->path = xstrdup(path);
 498 }
 499 #endif
 500
 501 /*
 502  * Local variables:
 503  * c-basic-offset: 4
 504  * indent-tabs-mode: nil
 505  * End:
 506  * vim: shiftwidth=4 tabstop=8 expandtab
 507  */
 508