src/record_conv.c

   1 /*
   2  * Copyright (C) 2005-2007, Index Data ApS
   3  * See the file LICENSE for details.
   4  *
   5  * $Id: record_conv.c,v 1.16 2007-05-06 20:12:20 adam 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 = (yaz_record_conv_t) 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_destroy(p->wr_error);
 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 = (struct yaz_record_conv_rule *)
 140         nmem_malloc(p->nmem, sizeof(*r));
 141     r->which = type;
 142     r->next = 0;
 143     *p->rules_p = r;
 144     p->rules_p = &r->next;
 145     return r;
 146 }
 147
 148 /** \brief parse 'xslt' conversion node */
 149 static int conv_xslt(yaz_record_conv_t p, const xmlNode *ptr)
 150 {
 151 #if YAZ_HAVE_XSLT
 152     struct _xmlAttr *attr;
 153     const char *stylesheet = 0;
 154
 155     for (attr = ptr->properties; attr; attr = attr->next)
 156     {
 157         if (!xmlStrcmp(attr->name, BAD_CAST "stylesheet") &&
 158             attr->children && attr->children->type == XML_TEXT_NODE)
 159             stylesheet = (const char *) attr->children->content;
 160         else
 161         {
 162             wrbuf_printf(p->wr_error, "Bad attribute '%s'"
 163                          "Expected stylesheet.", attr->name);
 164             return -1;
 165         }
 166     }
 167     if (!stylesheet)
 168     {
 169         wrbuf_printf(p->wr_error, "Element <xslt>: "
 170                      "attribute 'stylesheet' expected");
 171         return -1;
 172     }
 173     else
 174     {
 175         char fullpath[1024];
 176         xsltStylesheetPtr xsp;
 177         if (!yaz_filepath_resolve(stylesheet, p->path, 0, fullpath))
 178         {
 179             wrbuf_printf(p->wr_error, "Element <xslt stylesheet=\"%s\"/>:"
 180                          " could not locate stylesheet '%s' with path '%s'",
 181                          stylesheet, fullpath, p->path);
 182             return -1;
 183         }
 184         xsp = xsltParseStylesheetFile((xmlChar*) fullpath);
 185         if (!xsp)
 186         {
 187             wrbuf_printf(p->wr_error, "Element <xslt stylesheet=\"%s\"/>:"
 188                          " parsing stylesheet '%s' with path '%s' failed,"
 189 #if YAZ_HAVE_EXSLT
 190                          " EXSLT enabled",
 191 #else
 192                          " EXSLT not supported",
 193 #endif
 194                          stylesheet, fullpath, p->path);
 195             return -1;
 196         }
 197         else
 198         {
 199             struct yaz_record_conv_rule *r =
 200                 add_rule(p, YAZ_RECORD_CONV_RULE_XSLT);
 201             r->u.xslt.xsp = xsp;
 202         }
 203     }
 204     return 0;
 205 #else
 206     wrbuf_printf(p->wr_error, "xslt unsupported."
 207                  " YAZ compiled without XSLT support");
 208     return -1;
 209 #endif
 210 }
 211
 212 /** \brief parse 'marc' conversion node */
 213 static int conv_marc(yaz_record_conv_t p, const xmlNode *ptr)
 214 {
 215     struct _xmlAttr *attr;
 216     const char *input_charset = 0;
 217     const char *output_charset = 0;
 218     const char *input_format = 0;
 219     const char *output_format = 0;
 220     int input_format_mode = 0;
 221     int output_format_mode = 0;
 222     struct yaz_record_conv_rule *r;
 223     yaz_iconv_t cd = 0;
 224
 225     for (attr = ptr->properties; attr; attr = attr->next)
 226     {
 227         if (!xmlStrcmp(attr->name, BAD_CAST "inputcharset") &&
 228             attr->children && attr->children->type == XML_TEXT_NODE)
 229             input_charset = (const char *) attr->children->content;
 230         else if (!xmlStrcmp(attr->name, BAD_CAST "outputcharset") &&
 231             attr->children && attr->children->type == XML_TEXT_NODE)
 232             output_charset = (const char *) attr->children->content;
 233         else if (!xmlStrcmp(attr->name, BAD_CAST "inputformat") &&
 234             attr->children && attr->children->type == XML_TEXT_NODE)
 235             input_format = (const char *) attr->children->content;
 236         else if (!xmlStrcmp(attr->name, BAD_CAST "outputformat") &&
 237             attr->children && attr->children->type == XML_TEXT_NODE)
 238             output_format = (const char *) attr->children->content;
 239         else
 240         {
 241             wrbuf_printf(p->wr_error, "Element <marc>: expected attributes"
 242                          "'inputformat', 'inputcharset', 'outputformat' or"
 243                          " 'outputcharset', got attribute '%s'",
 244                          attr->name);
 245             return -1;
 246         }
 247     }
 248     if (!input_format)
 249     {
 250         wrbuf_printf(p->wr_error, "Element <marc>: "
 251                      "attribute 'inputformat' required");
 252         return -1;
 253     }
 254     else if (!strcmp(input_format, "marc"))
 255     {
 256         input_format_mode = YAZ_MARC_ISO2709;
 257     }
 258     else if (!strcmp(input_format, "xml"))
 259     {
 260         input_format_mode = YAZ_MARC_MARCXML;
 261         /** Libxml2 generates UTF-8 encoding by default .
 262             So we convert from UTF-8 to outputcharset (if defined)
 263         */
 264         if (!input_charset && output_charset)
 265             input_charset = "utf-8";
 266     }
 267     else
 268     {
 269         wrbuf_printf(p->wr_error, "Element <marc inputformat='%s'>: "
 270                      " Unsupported input format"
 271                      " defined by attribute value",
 272                      input_format);
 273         return -1;
 274     }
 275
 276     if (!output_format)
 277     {
 278         wrbuf_printf(p->wr_error,
 279                      "Element <marc>: attribute 'outputformat' required");
 280         return -1;
 281     }
 282     else if (!strcmp(output_format, "line"))
 283     {
 284         output_format_mode = YAZ_MARC_LINE;
 285     }
 286     else if (!strcmp(output_format, "marcxml"))
 287     {
 288         output_format_mode = YAZ_MARC_MARCXML;
 289         if (input_charset && !output_charset)
 290             output_charset = "utf-8";
 291     }
 292     else if (!strcmp(output_format, "marc"))
 293     {
 294         output_format_mode = YAZ_MARC_ISO2709;
 295     }
 296     else if (!strcmp(output_format, "marcxchange"))
 297     {
 298         output_format_mode = YAZ_MARC_XCHANGE;
 299         if (input_charset && !output_charset)
 300             output_charset = "utf-8";
 301     }
 302     else
 303     {
 304         wrbuf_printf(p->wr_error, "Element <marc outputformat='%s'>: "
 305                      " Unsupported output format"
 306                      " defined by attribute value",
 307                      output_format);
 308         return -1;
 309     }
 310     if (input_charset && output_charset)
 311     {
 312         cd = yaz_iconv_open(output_charset, input_charset);
 313         if (!cd)
 314         {
 315             wrbuf_printf(p->wr_error,
 316                          "Element <marc inputcharset='%s' outputcharset='%s'>:"
 317                          " Unsupported character set mapping"
 318                          " defined by attribute values",
 319                          input_charset, output_charset);
 320             return -1;
 321         }
 322     }
 323     else if (input_charset)
 324     {
 325         wrbuf_printf(p->wr_error, "Element <marc>: "
 326                      "attribute 'outputcharset' missing");
 327         return -1;
 328     }
 329     else if (output_charset)
 330     {
 331         wrbuf_printf(p->wr_error, "Element <marc>: "
 332                      "attribute 'inputcharset' missing");
 333         return -1;
 334     }
 335     r = add_rule(p, YAZ_RECORD_CONV_RULE_MARC);
 336     r->u.marc.iconv_t = cd;
 337
 338     r->u.marc.input_format = input_format_mode;
 339     r->u.marc.output_format = output_format_mode;
 340     return 0;
 341 }
 342
 343 int yaz_record_conv_configure(yaz_record_conv_t p, const xmlNode *ptr)
 344 {
 345     yaz_record_conv_reset(p);
 346
 347     /* parsing element children */
 348     for (ptr = ptr->children; ptr; ptr = ptr->next)
 349         {
 350             if (ptr->type != XML_ELEMENT_NODE)
 351                 continue;
 352             if (!strcmp((const char *) ptr->name, "xslt"))
 353                 {
 354                     if (conv_xslt(p, ptr))
 355                         return -1;
 356                 }
 357             else if (!strcmp((const char *) ptr->name, "marc"))
 358                 {
 359                     if (conv_marc(p, ptr))
 360                         return -1;
 361                 }
 362             else
 363                 {
 364                     wrbuf_printf(p->wr_error, "Element <backend>: expected "
 365                                  "<marc> or <xslt> element, got <%s>"
 366                                  , ptr->name);
 367                     return -1;
 368                 }
 369         }
 370     return 0;
 371 }
 372
 373 int yaz_record_conv_record(yaz_record_conv_t p,
 374                            const char *input_record_buf,
 375                            size_t input_record_len,
 376                            WRBUF output_record)
 377 {
 378     int ret = 0;
 379     WRBUF record = output_record; /* pointer transfer */
 380     struct yaz_record_conv_rule *r = p->rules;
 381     wrbuf_rewind(p->wr_error);
 382
 383     wrbuf_write(record, input_record_buf, input_record_len);
 384     for (; ret == 0 && r; r = r->next)
 385     {
 386         if (r->which == YAZ_RECORD_CONV_RULE_MARC)
 387         {
 388             yaz_marc_t mt = yaz_marc_create();
 389
 390             yaz_marc_xml(mt, r->u.marc.output_format);
 391
 392             if (r->u.marc.iconv_t)
 393                 yaz_marc_iconv(mt, r->u.marc.iconv_t);
 394             if (r->u.marc.input_format == YAZ_MARC_ISO2709)
 395             {
 396                 int sz = yaz_marc_read_iso2709(mt, wrbuf_buf(record),
 397                                                wrbuf_len(record));
 398                 if (sz > 0)
 399                     ret = 0;
 400                 else
 401                     ret = -1;
 402             }
 403             else if (r->u.marc.input_format == YAZ_MARC_MARCXML)
 404             {
 405                 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
 406                                                wrbuf_len(record));
 407                 if (!doc)
 408                 {
 409                     wrbuf_printf(p->wr_error, "xmlParseMemory failed");
 410                     ret = -1;
 411                 }
 412                 else
 413                 {
 414                     ret = yaz_marc_read_xml(mt, xmlDocGetRootElement(doc));
 415                     if (ret)
 416                         wrbuf_printf(p->wr_error, "yaz_marc_read_xml failed");
 417                 }
 418                 xmlFreeDoc(doc);
 419             }
 420             else
 421             {
 422                 wrbuf_printf(p->wr_error, "unsupported input format");
 423                 ret = -1;
 424             }
 425             if (ret == 0)
 426             {
 427                 wrbuf_rewind(record);
 428                 ret = yaz_marc_write_mode(mt, record);
 429                 if (ret)
 430                     wrbuf_printf(p->wr_error, "yaz_marc_write_mode failed");
 431             }
 432             yaz_marc_destroy(mt);
 433         }
 434 #if YAZ_HAVE_XSLT
 435         else if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
 436         {
 437             xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
 438                                            wrbuf_len(record));
 439             if (!doc)
 440             {
 441                 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
 442                 ret = -1;
 443             }
 444             else
 445             {
 446                 xmlDocPtr res = xsltApplyStylesheet(r->u.xslt.xsp, doc, 0);
 447                 if (res)
 448                 {
 449                     xmlChar *out_buf = 0;
 450                     int out_len;
 451
 452 #if YAZ_HAVE_XSLTSAVERESULTTOSTRING
 453                     xsltSaveResultToString(&out_buf, &out_len, res,
 454                                            r->u.xslt.xsp);
 455 #else
 456                     xmlDocDumpFormatMemory (res, &out_buf, &out_len, 1);
 457 #endif
 458                     if (!out_buf)
 459                     {
 460                         wrbuf_printf(p->wr_error,
 461                                      "xsltSaveResultToString failed");
 462                         ret = -1;
 463                     }
 464                     else
 465                     {
 466                         wrbuf_rewind(record);
 467                         wrbuf_write(record, (const char *) out_buf, out_len);
 468
 469                         xmlFree(out_buf);
 470                     }
 471                     xmlFreeDoc(res);
 472                 }
 473                 else
 474                 {
 475                     wrbuf_printf(p->wr_error, "xsltApplyStylesheet failed");
 476                     ret = -1;
 477                 }
 478                 xmlFreeDoc(doc);
 479             }
 480         }
 481 #endif
 482     }
 483     return ret;
 484 }
 485
 486 const char *yaz_record_conv_get_error(yaz_record_conv_t p)
 487 {
 488     return wrbuf_cstr(p->wr_error);
 489 }
 490
 491 void yaz_record_conv_set_path(yaz_record_conv_t p, const char *path)
 492 {
 493     xfree(p->path);
 494     p->path = 0;
 495     if (path)
 496         p->path = xstrdup(path);
 497 }
 498 #endif
 499
 500 /*
 501  * Local variables:
 502  * c-basic-offset: 4
 503  * indent-tabs-mode: nil
 504  * End:
 505  * vim: shiftwidth=4 tabstop=8 expandtab
 506  */
 507