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 1.17 2007-12-16 11:08:51 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 #include <yaz/z-opac.h>
  25
  26 #if YAZ_HAVE_XML2
  27 #include <libxml/parser.h>
  28 #include <libxml/tree.h>
  29 #include <libxml/xinclude.h>
  30 #if YAZ_HAVE_XSLT
  31 #include <libxslt/xsltutils.h>
  32 #include <libxslt/transform.h>
  33 #endif
  34 #if YAZ_HAVE_EXSLT
  35 #include <libexslt/exslt.h>
  36 #endif
  37
  38 /** \brief The internal structure for yaz_record_conv_t */
  39 struct yaz_record_conv_struct {
  40     /** \brief memory for configuration */
  41     NMEM nmem;
  42
  43     /** \brief conversion rules (allocated using NMEM) */
  44     struct yaz_record_conv_rule *rules;
  45
  46     /** \brief pointer to last conversion rule pointer in chain */
  47     struct yaz_record_conv_rule **rules_p;
  48
  49     /** \brief string buffer for error messages */
  50     WRBUF wr_error;
  51
  52     /** \brief path for opening files  */
  53     char *path;
  54 };
  55
  56 /** \brief tranformation types (rule types) */
  57 enum YAZ_RECORD_CONV_RULE
  58 {
  59     YAZ_RECORD_CONV_RULE_XSLT,
  60     YAZ_RECORD_CONV_RULE_MARC
  61 };
  62
  63
  64 /** \brief tranformation info (rule info) */
  65 struct yaz_record_conv_rule {
  66     enum YAZ_RECORD_CONV_RULE which;
  67     union {
  68 #if YAZ_HAVE_XSLT
  69         struct {
  70             xsltStylesheetPtr xsp;
  71         } xslt;
  72 #endif
  73         struct {
  74             yaz_iconv_t iconv_t;
  75             int input_format;
  76             int output_format;
  77         } marc;
  78     } u;
  79     struct yaz_record_conv_rule *next;
  80 };
  81
  82 /** \brief reset rules+configuration */
  83 static void yaz_record_conv_reset(yaz_record_conv_t p)
  84 {
  85
  86     struct yaz_record_conv_rule *r;
  87     for (r = p->rules; r; r = r->next)
  88     {
  89         if (r->which == YAZ_RECORD_CONV_RULE_MARC)
  90         {
  91             if (r->u.marc.iconv_t)
  92                 yaz_iconv_close(r->u.marc.iconv_t);
  93         }
  94 #if YAZ_HAVE_XSLT
  95         else if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
  96         {
  97             xsltFreeStylesheet(r->u.xslt.xsp);
  98         }
  99 #endif
 100     }
 101     wrbuf_rewind(p->wr_error);
 102     nmem_reset(p->nmem);
 103
 104     p->rules = 0;
 105
 106     p->rules_p = &p->rules;
 107 }
 108
 109 yaz_record_conv_t yaz_record_conv_create()
 110 {
 111     yaz_record_conv_t p = (yaz_record_conv_t) xmalloc(sizeof(*p));
 112     p->nmem = nmem_create();
 113     p->wr_error = wrbuf_alloc();
 114     p->rules = 0;
 115     p->path = 0;
 116
 117 #if YAZ_HAVE_EXSLT
 118     exsltRegisterAll();
 119 #endif
 120     yaz_record_conv_reset(p);
 121     return p;
 122 }
 123
 124 void yaz_record_conv_destroy(yaz_record_conv_t p)
 125 {
 126     if (p)
 127     {
 128         yaz_record_conv_reset(p);
 129         nmem_destroy(p->nmem);
 130         wrbuf_destroy(p->wr_error);
 131         xfree(p->path);
 132         xfree(p);
 133     }
 134 }
 135
 136 /** \brief adds a rule */
 137 static struct yaz_record_conv_rule *add_rule(yaz_record_conv_t p,
 138                                              enum YAZ_RECORD_CONV_RULE type)
 139 {
 140     struct yaz_record_conv_rule *r = (struct yaz_record_conv_rule *)
 141         nmem_malloc(p->nmem, sizeof(*r));
 142     r->which = type;
 143     r->next = 0;
 144     *p->rules_p = r;
 145     p->rules_p = &r->next;
 146     return r;
 147 }
 148
 149 /** \brief parse 'xslt' conversion node */
 150 static int conv_xslt(yaz_record_conv_t p, const xmlNode *ptr)
 151 {
 152 #if YAZ_HAVE_XSLT
 153     struct _xmlAttr *attr;
 154     const char *stylesheet = 0;
 155
 156     for (attr = ptr->properties; attr; attr = attr->next)
 157     {
 158         if (!xmlStrcmp(attr->name, BAD_CAST "stylesheet") &&
 159             attr->children && attr->children->type == XML_TEXT_NODE)
 160             stylesheet = (const char *) attr->children->content;
 161         else
 162         {
 163             wrbuf_printf(p->wr_error, "Bad attribute '%s'"
 164                          "Expected stylesheet.", attr->name);
 165             return -1;
 166         }
 167     }
 168     if (!stylesheet)
 169     {
 170         wrbuf_printf(p->wr_error, "Element <xslt>: "
 171                      "attribute 'stylesheet' expected");
 172         return -1;
 173     }
 174     else
 175     {
 176         char fullpath[1024];
 177         xsltStylesheetPtr xsp;
 178         if (!yaz_filepath_resolve(stylesheet, p->path, 0, fullpath))
 179         {
 180             wrbuf_printf(p->wr_error, "Element <xslt stylesheet=\"%s\"/>:"
 181                          " could not locate stylesheet '%s' with path '%s'",
 182                          stylesheet, fullpath, p->path);
 183             return -1;
 184         }
 185         xsp = xsltParseStylesheetFile((xmlChar*) fullpath);
 186         if (!xsp)
 187         {
 188             wrbuf_printf(p->wr_error, "Element <xslt stylesheet=\"%s\"/>:"
 189                          " parsing stylesheet '%s' with path '%s' failed,"
 190 #if YAZ_HAVE_EXSLT
 191                          " EXSLT enabled",
 192 #else
 193                          " EXSLT not supported",
 194 #endif
 195                          stylesheet, fullpath, p->path);
 196             return -1;
 197         }
 198         else
 199         {
 200             struct yaz_record_conv_rule *r =
 201                 add_rule(p, YAZ_RECORD_CONV_RULE_XSLT);
 202             r->u.xslt.xsp = xsp;
 203         }
 204     }
 205     return 0;
 206 #else
 207     wrbuf_printf(p->wr_error, "xslt unsupported."
 208                  " YAZ compiled without XSLT support");
 209     return -1;
 210 #endif
 211 }
 212
 213 /** \brief parse 'marc' conversion node */
 214 static int conv_marc(yaz_record_conv_t p, const xmlNode *ptr)
 215 {
 216     struct _xmlAttr *attr;
 217     const char *input_charset = 0;
 218     const char *output_charset = 0;
 219     const char *input_format = 0;
 220     const char *output_format = 0;
 221     int input_format_mode = 0;
 222     int output_format_mode = 0;
 223     struct yaz_record_conv_rule *r;
 224     yaz_iconv_t cd = 0;
 225
 226     for (attr = ptr->properties; attr; attr = attr->next)
 227     {
 228         if (!xmlStrcmp(attr->name, BAD_CAST "inputcharset") &&
 229             attr->children && attr->children->type == XML_TEXT_NODE)
 230             input_charset = (const char *) attr->children->content;
 231         else if (!xmlStrcmp(attr->name, BAD_CAST "outputcharset") &&
 232             attr->children && attr->children->type == XML_TEXT_NODE)
 233             output_charset = (const char *) attr->children->content;
 234         else if (!xmlStrcmp(attr->name, BAD_CAST "inputformat") &&
 235             attr->children && attr->children->type == XML_TEXT_NODE)
 236             input_format = (const char *) attr->children->content;
 237         else if (!xmlStrcmp(attr->name, BAD_CAST "outputformat") &&
 238             attr->children && attr->children->type == XML_TEXT_NODE)
 239             output_format = (const char *) attr->children->content;
 240         else
 241         {
 242             wrbuf_printf(p->wr_error, "Element <marc>: expected attributes"
 243                          "'inputformat', 'inputcharset', 'outputformat' or"
 244                          " 'outputcharset', got attribute '%s'",
 245                          attr->name);
 246             return -1;
 247         }
 248     }
 249     if (!input_format)
 250     {
 251         wrbuf_printf(p->wr_error, "Element <marc>: "
 252                      "attribute 'inputformat' required");
 253         return -1;
 254     }
 255     else if (!strcmp(input_format, "marc"))
 256     {
 257         input_format_mode = YAZ_MARC_ISO2709;
 258     }
 259     else if (!strcmp(input_format, "xml"))
 260     {
 261         input_format_mode = YAZ_MARC_MARCXML;
 262         /** Libxml2 generates UTF-8 encoding by default .
 263             So we convert from UTF-8 to outputcharset (if defined)
 264         */
 265         if (!input_charset && output_charset)
 266             input_charset = "utf-8";
 267     }
 268     else
 269     {
 270         wrbuf_printf(p->wr_error, "Element <marc inputformat='%s'>: "
 271                      " Unsupported input format"
 272                      " defined by attribute value",
 273                      input_format);
 274         return -1;
 275     }
 276
 277     if (!output_format)
 278     {
 279         wrbuf_printf(p->wr_error,
 280                      "Element <marc>: attribute 'outputformat' required");
 281         return -1;
 282     }
 283     else if (!strcmp(output_format, "line"))
 284     {
 285         output_format_mode = YAZ_MARC_LINE;
 286     }
 287     else if (!strcmp(output_format, "marcxml"))
 288     {
 289         output_format_mode = YAZ_MARC_MARCXML;
 290         if (input_charset && !output_charset)
 291             output_charset = "utf-8";
 292     }
 293     else if (!strcmp(output_format, "marc"))
 294     {
 295         output_format_mode = YAZ_MARC_ISO2709;
 296     }
 297     else if (!strcmp(output_format, "marcxchange"))
 298     {
 299         output_format_mode = YAZ_MARC_XCHANGE;
 300         if (input_charset && !output_charset)
 301             output_charset = "utf-8";
 302     }
 303     else
 304     {
 305         wrbuf_printf(p->wr_error, "Element <marc outputformat='%s'>: "
 306                      " Unsupported output format"
 307                      " defined by attribute value",
 308                      output_format);
 309         return -1;
 310     }
 311     if (input_charset && output_charset)
 312     {
 313         cd = yaz_iconv_open(output_charset, input_charset);
 314         if (!cd)
 315         {
 316             wrbuf_printf(p->wr_error,
 317                          "Element <marc inputcharset='%s' outputcharset='%s'>:"
 318                          " Unsupported character set mapping"
 319                          " defined by attribute values",
 320                          input_charset, output_charset);
 321             return -1;
 322         }
 323     }
 324     else if (input_charset)
 325     {
 326         wrbuf_printf(p->wr_error, "Element <marc>: "
 327                      "attribute 'outputcharset' missing");
 328         return -1;
 329     }
 330     else if (output_charset)
 331     {
 332         wrbuf_printf(p->wr_error, "Element <marc>: "
 333                      "attribute 'inputcharset' missing");
 334         return -1;
 335     }
 336     r = add_rule(p, YAZ_RECORD_CONV_RULE_MARC);
 337     r->u.marc.iconv_t = cd;
 338
 339     r->u.marc.input_format = input_format_mode;
 340     r->u.marc.output_format = output_format_mode;
 341     return 0;
 342 }
 343
 344 int yaz_record_conv_configure(yaz_record_conv_t p, const xmlNode *ptr)
 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 static int yaz_record_conv_record_rule(yaz_record_conv_t p,
 375                                        struct yaz_record_conv_rule *r,
 376                                        const char *input_record_buf,
 377                                        size_t input_record_len,
 378                                        WRBUF output_record);
 379
 380 int yaz_record_conv_opac_record(yaz_record_conv_t p,
 381                                 Z_OPACRecord *input_record,
 382                                 WRBUF output_record)
 383 {
 384     int ret = 0;
 385     struct yaz_record_conv_rule *r = p->rules;
 386     WRBUF res = wrbuf_alloc();
 387     yaz_marc_t mt = yaz_marc_create();
 388
 389     wrbuf_rewind(p->wr_error);
 390     yaz_marc_xml(mt, r->u.marc.output_format);
 391     if (r->u.marc.iconv_t)
 392         yaz_marc_iconv(mt, r->u.marc.iconv_t);
 393     yaz_opac_decode_wrbuf(mt, input_record, res);
 394     if (ret != -1)
 395     {
 396         ret = yaz_record_conv_record_rule(p,
 397                                           r->next,
 398                                           wrbuf_buf(res), wrbuf_len(res),
 399                                           output_record);
 400     }
 401     yaz_marc_destroy(mt);
 402     wrbuf_destroy(res);
 403     return ret;
 404 }
 405
 406 int yaz_record_conv_record(yaz_record_conv_t p,
 407                            const char *input_record_buf,
 408                            size_t input_record_len,
 409                            WRBUF output_record)
 410 {
 411     return yaz_record_conv_record_rule(p, p->rules,
 412                                        input_record_buf,
 413                                        input_record_len, output_record);
 414 }
 415
 416 static int yaz_record_conv_record_rule(yaz_record_conv_t p,
 417                                        struct yaz_record_conv_rule *r,
 418                                        const char *input_record_buf,
 419                                        size_t input_record_len,
 420                                        WRBUF output_record)
 421 {
 422     int ret = 0;
 423     WRBUF record = output_record; /* pointer transfer */
 424     wrbuf_rewind(p->wr_error);
 425
 426     wrbuf_write(record, input_record_buf, input_record_len);
 427     for (; ret == 0 && r; r = r->next)
 428     {
 429         if (r->which == YAZ_RECORD_CONV_RULE_MARC)
 430         {
 431             yaz_marc_t mt = yaz_marc_create();
 432
 433             yaz_marc_xml(mt, r->u.marc.output_format);
 434
 435             if (r->u.marc.iconv_t)
 436                 yaz_marc_iconv(mt, r->u.marc.iconv_t);
 437             if (r->u.marc.input_format == YAZ_MARC_ISO2709)
 438             {
 439                 int sz = yaz_marc_read_iso2709(mt, wrbuf_buf(record),
 440                                                wrbuf_len(record));
 441                 if (sz > 0)
 442                     ret = 0;
 443                 else
 444                     ret = -1;
 445             }
 446             else if (r->u.marc.input_format == YAZ_MARC_MARCXML)
 447             {
 448                 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
 449                                                wrbuf_len(record));
 450                 if (!doc)
 451                 {
 452                     wrbuf_printf(p->wr_error, "xmlParseMemory failed");
 453                     ret = -1;
 454                 }
 455                 else
 456                 {
 457                     ret = yaz_marc_read_xml(mt, xmlDocGetRootElement(doc));
 458                     if (ret)
 459                         wrbuf_printf(p->wr_error, "yaz_marc_read_xml failed");
 460                 }
 461                 xmlFreeDoc(doc);
 462             }
 463             else
 464             {
 465                 wrbuf_printf(p->wr_error, "unsupported input format");
 466                 ret = -1;
 467             }
 468             if (ret == 0)
 469             {
 470                 wrbuf_rewind(record);
 471                 ret = yaz_marc_write_mode(mt, record);
 472                 if (ret)
 473                     wrbuf_printf(p->wr_error, "yaz_marc_write_mode failed");
 474             }
 475             yaz_marc_destroy(mt);
 476         }
 477 #if YAZ_HAVE_XSLT
 478         else if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
 479         {
 480             xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
 481                                            wrbuf_len(record));
 482             if (!doc)
 483             {
 484                 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
 485                 ret = -1;
 486             }
 487             else
 488             {
 489                 xmlDocPtr res = xsltApplyStylesheet(r->u.xslt.xsp, doc, 0);
 490                 if (res)
 491                 {
 492                     xmlChar *out_buf = 0;
 493                     int out_len;
 494
 495 #if YAZ_HAVE_XSLTSAVERESULTTOSTRING
 496                     xsltSaveResultToString(&out_buf, &out_len, res,
 497                                            r->u.xslt.xsp);
 498 #else
 499                     xmlDocDumpFormatMemory (res, &out_buf, &out_len, 1);
 500 #endif
 501                     if (!out_buf)
 502                     {
 503                         wrbuf_printf(p->wr_error,
 504                                      "xsltSaveResultToString failed");
 505                         ret = -1;
 506                     }
 507                     else
 508                     {
 509                         wrbuf_rewind(record);
 510                         wrbuf_write(record, (const char *) out_buf, out_len);
 511
 512                         xmlFree(out_buf);
 513                     }
 514                     xmlFreeDoc(res);
 515                 }
 516                 else
 517                 {
 518                     wrbuf_printf(p->wr_error, "xsltApplyStylesheet failed");
 519                     ret = -1;
 520                 }
 521                 xmlFreeDoc(doc);
 522             }
 523         }
 524 #endif
 525     }
 526     return ret;
 527 }
 528
 529 const char *yaz_record_conv_get_error(yaz_record_conv_t p)
 530 {
 531     return wrbuf_cstr(p->wr_error);
 532 }
 533
 534 void yaz_record_conv_set_path(yaz_record_conv_t p, const char *path)
 535 {
 536     xfree(p->path);
 537     p->path = 0;
 538     if (path)
 539         p->path = xstrdup(path);
 540 }
 541 #endif
 542
 543 /*
 544  * Local variables:
 545  * c-basic-offset: 4
 546  * indent-tabs-mode: nil
 547  * End:
 548  * vim: shiftwidth=4 tabstop=8 expandtab
 549  */
 550