src/record_conv.c

   1 /* This file is part of the YAZ toolkit.
   2  * Copyright (C) 1995-2008 Index Data
   3  * See the file LICENSE for details.
   4  */
   5 /**
   6  * \file record_conv.c
   7  * \brief Record Conversions utility
   8  */
   9
  10 #if HAVE_CONFIG_H
  11 #include <config.h>
  12 #endif
  13
  14 #include <string.h>
  15 #include <yaz/yaz-iconv.h>
  16 #include <yaz/marcdisp.h>
  17 #include <yaz/record_conv.h>
  18 #include <yaz/wrbuf.h>
  19 #include <yaz/xmalloc.h>
  20 #include <yaz/nmem.h>
  21 #include <yaz/tpath.h>
  22 #include <yaz/z-opac.h>
  23
  24 #if YAZ_HAVE_XML2
  25 #include <libxml/parser.h>
  26 #include <libxml/tree.h>
  27 #include <libxml/xinclude.h>
  28 #if YAZ_HAVE_XSLT
  29 #include <libxslt/xsltutils.h>
  30 #include <libxslt/transform.h>
  31 #endif
  32 #if YAZ_HAVE_EXSLT
  33 #include <libexslt/exslt.h>
  34 #endif
  35
  36 /** \brief The internal structure for yaz_record_conv_t */
  37 struct yaz_record_conv_struct {
  38     /** \brief memory for configuration */
  39     NMEM nmem;
  40
  41     /** \brief conversion rules (allocated using NMEM) */
  42     struct yaz_record_conv_rule *rules;
  43
  44     /** \brief pointer to last conversion rule pointer in chain */
  45     struct yaz_record_conv_rule **rules_p;
  46
  47     /** \brief string buffer for error messages */
  48     WRBUF wr_error;
  49
  50     /** \brief path for opening files  */
  51     char *path;
  52 };
  53
  54 /** \brief tranformation types (rule types) */
  55 enum YAZ_RECORD_CONV_RULE
  56 {
  57     YAZ_RECORD_CONV_RULE_XSLT,
  58     YAZ_RECORD_CONV_RULE_MARC
  59 };
  60
  61
  62 /** \brief tranformation info (rule info) */
  63 struct yaz_record_conv_rule {
  64     enum YAZ_RECORD_CONV_RULE which;
  65     union {
  66 #if YAZ_HAVE_XSLT
  67         struct {
  68             xsltStylesheetPtr xsp;
  69         } xslt;
  70 #endif
  71         struct {
  72             const char *input_charset;
  73             const char *output_charset;
  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             ;
  91         }
  92 #if YAZ_HAVE_XSLT
  93         else if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
  94         {
  95             xsltFreeStylesheet(r->u.xslt.xsp);
  96         }
  97 #endif
  98     }
  99     wrbuf_rewind(p->wr_error);
 100     nmem_reset(p->nmem);
 101
 102     p->rules = 0;
 103
 104     p->rules_p = &p->rules;
 105 }
 106
 107 yaz_record_conv_t yaz_record_conv_create()
 108 {
 109     yaz_record_conv_t p = (yaz_record_conv_t) xmalloc(sizeof(*p));
 110     p->nmem = nmem_create();
 111     p->wr_error = wrbuf_alloc();
 112     p->rules = 0;
 113     p->path = 0;
 114
 115 #if YAZ_HAVE_EXSLT
 116     exsltRegisterAll();
 117 #endif
 118     yaz_record_conv_reset(p);
 119     return p;
 120 }
 121
 122 void yaz_record_conv_destroy(yaz_record_conv_t p)
 123 {
 124     if (p)
 125     {
 126         yaz_record_conv_reset(p);
 127         nmem_destroy(p->nmem);
 128         wrbuf_destroy(p->wr_error);
 129         xfree(p->path);
 130         xfree(p);
 131     }
 132 }
 133
 134 /** \brief adds a rule */
 135 static struct yaz_record_conv_rule *add_rule(yaz_record_conv_t p,
 136                                              enum YAZ_RECORD_CONV_RULE type)
 137 {
 138     struct yaz_record_conv_rule *r = (struct yaz_record_conv_rule *)
 139         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'",
 180                          stylesheet, fullpath);
 181             if (p->path)
 182                 wrbuf_printf(p->wr_error, " with path '%s'", p->path);
 183
 184             return -1;
 185         }
 186         xsp = xsltParseStylesheetFile((xmlChar*) fullpath);
 187         if (!xsp)
 188         {
 189             wrbuf_printf(p->wr_error, "Element: <xslt stylesheet=\"%s\"/>:"
 190                          " xslt parse failed: %s", stylesheet, fullpath);
 191             if (p->path)
 192                 wrbuf_printf(p->wr_error, " with path '%s'", p->path);
 193             wrbuf_printf(p->wr_error, " ("
 194 #if YAZ_HAVE_EXSLT
 195
 196                          "EXSLT enabled"
 197 #else
 198                          "EXSLT not supported"
 199 #endif
 200                          ")");
 201             return -1;
 202         }
 203         else
 204         {
 205             struct yaz_record_conv_rule *r =
 206                 add_rule(p, YAZ_RECORD_CONV_RULE_XSLT);
 207             r->u.xslt.xsp = xsp;
 208         }
 209     }
 210     return 0;
 211 #else
 212     wrbuf_printf(p->wr_error, "xslt unsupported."
 213                  " YAZ compiled without XSLT support");
 214     return -1;
 215 #endif
 216 }
 217
 218 /** \brief parse 'marc' conversion node */
 219 static int conv_marc(yaz_record_conv_t p, const xmlNode *ptr)
 220 {
 221     struct _xmlAttr *attr;
 222     const char *input_charset = 0;
 223     const char *output_charset = 0;
 224     const char *input_format = 0;
 225     const char *output_format = 0;
 226     int input_format_mode = 0;
 227     int output_format_mode = 0;
 228     struct yaz_record_conv_rule *r;
 229
 230     for (attr = ptr->properties; attr; attr = attr->next)
 231     {
 232         if (!xmlStrcmp(attr->name, BAD_CAST "inputcharset") &&
 233             attr->children && attr->children->type == XML_TEXT_NODE)
 234             input_charset = (const char *) attr->children->content;
 235         else if (!xmlStrcmp(attr->name, BAD_CAST "outputcharset") &&
 236             attr->children && attr->children->type == XML_TEXT_NODE)
 237             output_charset = (const char *) attr->children->content;
 238         else if (!xmlStrcmp(attr->name, BAD_CAST "inputformat") &&
 239             attr->children && attr->children->type == XML_TEXT_NODE)
 240             input_format = (const char *) attr->children->content;
 241         else if (!xmlStrcmp(attr->name, BAD_CAST "outputformat") &&
 242             attr->children && attr->children->type == XML_TEXT_NODE)
 243             output_format = (const char *) attr->children->content;
 244         else
 245         {
 246             wrbuf_printf(p->wr_error, "Element <marc>: expected attributes"
 247                          "'inputformat', 'inputcharset', 'outputformat' or"
 248                          " 'outputcharset', got attribute '%s'",
 249                          attr->name);
 250             return -1;
 251         }
 252     }
 253     if (!input_format)
 254     {
 255         wrbuf_printf(p->wr_error, "Element <marc>: "
 256                      "attribute 'inputformat' required");
 257         return -1;
 258     }
 259     else if (!strcmp(input_format, "marc"))
 260     {
 261         input_format_mode = YAZ_MARC_ISO2709;
 262     }
 263     else if (!strcmp(input_format, "xml"))
 264     {
 265         input_format_mode = YAZ_MARC_MARCXML;
 266         /** Libxml2 generates UTF-8 encoding by default .
 267             So we convert from UTF-8 to outputcharset (if defined)
 268         */
 269         if (!input_charset && output_charset)
 270             input_charset = "utf-8";
 271     }
 272     else
 273     {
 274         wrbuf_printf(p->wr_error, "Element <marc inputformat='%s'>: "
 275                      " Unsupported input format"
 276                      " defined by attribute value",
 277                      input_format);
 278         return -1;
 279     }
 280
 281     if (!output_format)
 282     {
 283         wrbuf_printf(p->wr_error,
 284                      "Element <marc>: attribute 'outputformat' required");
 285         return -1;
 286     }
 287     else if (!strcmp(output_format, "line"))
 288     {
 289         output_format_mode = YAZ_MARC_LINE;
 290     }
 291     else if (!strcmp(output_format, "marcxml"))
 292     {
 293         output_format_mode = YAZ_MARC_MARCXML;
 294         if (input_charset && !output_charset)
 295             output_charset = "utf-8";
 296     }
 297     else if (!strcmp(output_format, "marc"))
 298     {
 299         output_format_mode = YAZ_MARC_ISO2709;
 300     }
 301     else if (!strcmp(output_format, "marcxchange"))
 302     {
 303         output_format_mode = YAZ_MARC_XCHANGE;
 304         if (input_charset && !output_charset)
 305             output_charset = "utf-8";
 306     }
 307     else
 308     {
 309         wrbuf_printf(p->wr_error, "Element <marc outputformat='%s'>: "
 310                      " Unsupported output format"
 311                      " defined by attribute value",
 312                      output_format);
 313         return -1;
 314     }
 315     if (input_charset && output_charset)
 316     {
 317         yaz_iconv_t cd = yaz_iconv_open(output_charset, input_charset);
 318         if (!cd)
 319         {
 320             wrbuf_printf(p->wr_error,
 321                          "Element <marc inputcharset='%s' outputcharset='%s'>:"
 322                          " Unsupported character set mapping"
 323                          " defined by attribute values",
 324                          input_charset, output_charset);
 325             return -1;
 326         }
 327         yaz_iconv_close(cd);
 328     }
 329     else if (input_charset)
 330     {
 331         wrbuf_printf(p->wr_error, "Element <marc>: "
 332                      "attribute 'outputcharset' missing");
 333         return -1;
 334     }
 335     else if (output_charset)
 336     {
 337         wrbuf_printf(p->wr_error, "Element <marc>: "
 338                      "attribute 'inputcharset' missing");
 339         return -1;
 340     }
 341     r = add_rule(p, YAZ_RECORD_CONV_RULE_MARC);
 342
 343     r->u.marc.input_charset = nmem_strdup(p->nmem, input_charset);
 344     r->u.marc.output_charset = nmem_strdup(p->nmem, output_charset);
 345     r->u.marc.input_format = input_format_mode;
 346     r->u.marc.output_format = output_format_mode;
 347     return 0;
 348 }
 349
 350 int yaz_record_conv_configure(yaz_record_conv_t p, const xmlNode *ptr)
 351 {
 352     yaz_record_conv_reset(p);
 353
 354     /* parsing element children */
 355     for (ptr = ptr->children; ptr; ptr = ptr->next)
 356         {
 357             if (ptr->type != XML_ELEMENT_NODE)
 358                 continue;
 359             if (!strcmp((const char *) ptr->name, "xslt"))
 360                 {
 361                     if (conv_xslt(p, ptr))
 362                         return -1;
 363                 }
 364             else if (!strcmp((const char *) ptr->name, "marc"))
 365                 {
 366                     if (conv_marc(p, ptr))
 367                         return -1;
 368                 }
 369             else
 370                 {
 371                     wrbuf_printf(p->wr_error, "Element <backend>: expected "
 372                                  "<marc> or <xslt> element, got <%s>"
 373                                  , ptr->name);
 374                     return -1;
 375                 }
 376         }
 377     return 0;
 378 }
 379
 380 static int yaz_record_conv_record_rule(yaz_record_conv_t p,
 381                                        struct yaz_record_conv_rule *r,
 382                                        const char *input_record_buf,
 383                                        size_t input_record_len,
 384                                        WRBUF output_record);
 385
 386 int yaz_record_conv_opac_record(yaz_record_conv_t p,
 387                                 Z_OPACRecord *input_record,
 388                                 WRBUF output_record)
 389 {
 390     int ret = 0;
 391     struct yaz_record_conv_rule *r = p->rules;
 392     if (!r || r->which != YAZ_RECORD_CONV_RULE_MARC)
 393         ret = -1; /* no marc rule so we can't do OPAC */
 394     else
 395     {
 396         WRBUF res = wrbuf_alloc();
 397         yaz_marc_t mt = yaz_marc_create();
 398         yaz_iconv_t cd = yaz_iconv_open(r->u.marc.output_charset,
 399                                         r->u.marc.input_charset);
 400
 401         wrbuf_rewind(p->wr_error);
 402         yaz_marc_xml(mt, r->u.marc.output_format);
 403
 404         yaz_marc_iconv(mt, cd);
 405
 406         yaz_opac_decode_wrbuf(mt, input_record, res);
 407         if (ret != -1)
 408         {
 409             ret = yaz_record_conv_record_rule(p,
 410                                               r->next,
 411                                               wrbuf_buf(res), wrbuf_len(res),
 412                                               output_record);
 413         }
 414         yaz_marc_destroy(mt);
 415         if (cd)
 416             yaz_iconv_close(cd);
 417         wrbuf_destroy(res);
 418     }
 419     return ret;
 420 }
 421
 422 int yaz_record_conv_record(yaz_record_conv_t p,
 423                            const char *input_record_buf,
 424                            size_t input_record_len,
 425                            WRBUF output_record)
 426 {
 427     return yaz_record_conv_record_rule(p, p->rules,
 428                                        input_record_buf,
 429                                        input_record_len, output_record);
 430 }
 431
 432 static int yaz_record_conv_record_rule(yaz_record_conv_t p,
 433                                        struct yaz_record_conv_rule *r,
 434                                        const char *input_record_buf,
 435                                        size_t input_record_len,
 436                                        WRBUF output_record)
 437 {
 438     int ret = 0;
 439     WRBUF record = output_record; /* pointer transfer */
 440     wrbuf_rewind(p->wr_error);
 441
 442     wrbuf_write(record, input_record_buf, input_record_len);
 443     for (; ret == 0 && r; r = r->next)
 444     {
 445         if (r->which == YAZ_RECORD_CONV_RULE_MARC)
 446         {
 447             yaz_iconv_t cd =
 448                 yaz_iconv_open(r->u.marc.output_charset,
 449                                r->u.marc.input_charset);
 450             yaz_marc_t mt = yaz_marc_create();
 451
 452             yaz_marc_xml(mt, r->u.marc.output_format);
 453
 454             if (cd)
 455                 yaz_marc_iconv(mt, cd);
 456             if (r->u.marc.input_format == YAZ_MARC_ISO2709)
 457             {
 458                 int sz = yaz_marc_read_iso2709(mt, wrbuf_buf(record),
 459                                                wrbuf_len(record));
 460                 if (sz > 0)
 461                     ret = 0;
 462                 else
 463                     ret = -1;
 464             }
 465             else if (r->u.marc.input_format == YAZ_MARC_MARCXML)
 466             {
 467                 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
 468                                                wrbuf_len(record));
 469                 if (!doc)
 470                 {
 471                     wrbuf_printf(p->wr_error, "xmlParseMemory failed");
 472                     ret = -1;
 473                 }
 474                 else
 475                 {
 476                     ret = yaz_marc_read_xml(mt, xmlDocGetRootElement(doc));
 477                     if (ret)
 478                         wrbuf_printf(p->wr_error, "yaz_marc_read_xml failed");
 479                 }
 480                 xmlFreeDoc(doc);
 481             }
 482             else
 483             {
 484                 wrbuf_printf(p->wr_error, "unsupported input format");
 485                 ret = -1;
 486             }
 487             if (ret == 0)
 488             {
 489                 wrbuf_rewind(record);
 490                 ret = yaz_marc_write_mode(mt, record);
 491                 if (ret)
 492                     wrbuf_printf(p->wr_error, "yaz_marc_write_mode failed");
 493             }
 494             if (cd)
 495                 yaz_iconv_close(cd);
 496             yaz_marc_destroy(mt);
 497         }
 498 #if YAZ_HAVE_XSLT
 499         else if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
 500         {
 501             xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
 502                                            wrbuf_len(record));
 503             if (!doc)
 504             {
 505                 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
 506                 ret = -1;
 507             }
 508             else
 509             {
 510                 xmlDocPtr res = xsltApplyStylesheet(r->u.xslt.xsp, doc, 0);
 511                 if (res)
 512                 {
 513                     xmlChar *out_buf = 0;
 514                     int out_len;
 515
 516 #if YAZ_HAVE_XSLTSAVERESULTTOSTRING
 517                     xsltSaveResultToString(&out_buf, &out_len, res,
 518                                            r->u.xslt.xsp);
 519 #else
 520                     xmlDocDumpFormatMemory (res, &out_buf, &out_len, 1);
 521 #endif
 522                     if (!out_buf)
 523                     {
 524                         wrbuf_printf(p->wr_error,
 525                                      "xsltSaveResultToString failed");
 526                         ret = -1;
 527                     }
 528                     else
 529                     {
 530                         wrbuf_rewind(record);
 531                         wrbuf_write(record, (const char *) out_buf, out_len);
 532
 533                         xmlFree(out_buf);
 534                     }
 535                     xmlFreeDoc(res);
 536                 }
 537                 else
 538                 {
 539                     wrbuf_printf(p->wr_error, "xsltApplyStylesheet failed");
 540                     ret = -1;
 541                 }
 542                 xmlFreeDoc(doc);
 543             }
 544         }
 545 #endif
 546     }
 547     return ret;
 548 }
 549
 550 const char *yaz_record_conv_get_error(yaz_record_conv_t p)
 551 {
 552     return wrbuf_cstr(p->wr_error);
 553 }
 554
 555 void yaz_record_conv_set_path(yaz_record_conv_t p, const char *path)
 556 {
 557     xfree(p->path);
 558     p->path = 0;
 559     if (path)
 560         p->path = xstrdup(path);
 561 }
 562 #endif
 563
 564 /*
 565  * Local variables:
 566  * c-basic-offset: 4
 567  * indent-tabs-mode: nil
 568  * End:
 569  * vim: shiftwidth=4 tabstop=8 expandtab
 570  */
 571