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             yaz_iconv_t iconv_t;
  73             int input_format;
  74             int output_format;
  75         } marc;
  76     } u;
  77     struct yaz_record_conv_rule *next;
  78 };
  79
  80 /** \brief reset rules+configuration */
  81 static void yaz_record_conv_reset(yaz_record_conv_t p)
  82 {
  83
  84     struct yaz_record_conv_rule *r;
  85     for (r = p->rules; r; r = r->next)
  86     {
  87         if (r->which == YAZ_RECORD_CONV_RULE_MARC)
  88         {
  89             if (r->u.marc.iconv_t)
  90                 yaz_iconv_close(r->u.marc.iconv_t);
  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     yaz_iconv_t cd = 0;
 230
 231     for (attr = ptr->properties; attr; attr = attr->next)
 232     {
 233         if (!xmlStrcmp(attr->name, BAD_CAST "inputcharset") &&
 234             attr->children && attr->children->type == XML_TEXT_NODE)
 235             input_charset = (const char *) attr->children->content;
 236         else if (!xmlStrcmp(attr->name, BAD_CAST "outputcharset") &&
 237             attr->children && attr->children->type == XML_TEXT_NODE)
 238             output_charset = (const char *) attr->children->content;
 239         else if (!xmlStrcmp(attr->name, BAD_CAST "inputformat") &&
 240             attr->children && attr->children->type == XML_TEXT_NODE)
 241             input_format = (const char *) attr->children->content;
 242         else if (!xmlStrcmp(attr->name, BAD_CAST "outputformat") &&
 243             attr->children && attr->children->type == XML_TEXT_NODE)
 244             output_format = (const char *) attr->children->content;
 245         else
 246         {
 247             wrbuf_printf(p->wr_error, "Element <marc>: expected attributes"
 248                          "'inputformat', 'inputcharset', 'outputformat' or"
 249                          " 'outputcharset', got attribute '%s'",
 250                          attr->name);
 251             return -1;
 252         }
 253     }
 254     if (!input_format)
 255     {
 256         wrbuf_printf(p->wr_error, "Element <marc>: "
 257                      "attribute 'inputformat' required");
 258         return -1;
 259     }
 260     else if (!strcmp(input_format, "marc"))
 261     {
 262         input_format_mode = YAZ_MARC_ISO2709;
 263     }
 264     else if (!strcmp(input_format, "xml"))
 265     {
 266         input_format_mode = YAZ_MARC_MARCXML;
 267         /** Libxml2 generates UTF-8 encoding by default .
 268             So we convert from UTF-8 to outputcharset (if defined)
 269         */
 270         if (!input_charset && output_charset)
 271             input_charset = "utf-8";
 272     }
 273     else
 274     {
 275         wrbuf_printf(p->wr_error, "Element <marc inputformat='%s'>: "
 276                      " Unsupported input format"
 277                      " defined by attribute value",
 278                      input_format);
 279         return -1;
 280     }
 281
 282     if (!output_format)
 283     {
 284         wrbuf_printf(p->wr_error,
 285                      "Element <marc>: attribute 'outputformat' required");
 286         return -1;
 287     }
 288     else if (!strcmp(output_format, "line"))
 289     {
 290         output_format_mode = YAZ_MARC_LINE;
 291     }
 292     else if (!strcmp(output_format, "marcxml"))
 293     {
 294         output_format_mode = YAZ_MARC_MARCXML;
 295         if (input_charset && !output_charset)
 296             output_charset = "utf-8";
 297     }
 298     else if (!strcmp(output_format, "marc"))
 299     {
 300         output_format_mode = YAZ_MARC_ISO2709;
 301     }
 302     else if (!strcmp(output_format, "marcxchange"))
 303     {
 304         output_format_mode = YAZ_MARC_XCHANGE;
 305         if (input_charset && !output_charset)
 306             output_charset = "utf-8";
 307     }
 308     else
 309     {
 310         wrbuf_printf(p->wr_error, "Element <marc outputformat='%s'>: "
 311                      " Unsupported output format"
 312                      " defined by attribute value",
 313                      output_format);
 314         return -1;
 315     }
 316     if (input_charset && output_charset)
 317     {
 318         cd = yaz_iconv_open(output_charset, input_charset);
 319         if (!cd)
 320         {
 321             wrbuf_printf(p->wr_error,
 322                          "Element <marc inputcharset='%s' outputcharset='%s'>:"
 323                          " Unsupported character set mapping"
 324                          " defined by attribute values",
 325                          input_charset, output_charset);
 326             return -1;
 327         }
 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     r->u.marc.iconv_t = cd;
 343
 344     r->u.marc.input_format = input_format_mode;
 345     r->u.marc.output_format = output_format_mode;
 346     return 0;
 347 }
 348
 349 int yaz_record_conv_configure(yaz_record_conv_t p, const xmlNode *ptr)
 350 {
 351     yaz_record_conv_reset(p);
 352
 353     /* parsing element children */
 354     for (ptr = ptr->children; ptr; ptr = ptr->next)
 355         {
 356             if (ptr->type != XML_ELEMENT_NODE)
 357                 continue;
 358             if (!strcmp((const char *) ptr->name, "xslt"))
 359                 {
 360                     if (conv_xslt(p, ptr))
 361                         return -1;
 362                 }
 363             else if (!strcmp((const char *) ptr->name, "marc"))
 364                 {
 365                     if (conv_marc(p, ptr))
 366                         return -1;
 367                 }
 368             else
 369                 {
 370                     wrbuf_printf(p->wr_error, "Element <backend>: expected "
 371                                  "<marc> or <xslt> element, got <%s>"
 372                                  , ptr->name);
 373                     return -1;
 374                 }
 375         }
 376     return 0;
 377 }
 378
 379 static int yaz_record_conv_record_rule(yaz_record_conv_t p,
 380                                        struct yaz_record_conv_rule *r,
 381                                        const char *input_record_buf,
 382                                        size_t input_record_len,
 383                                        WRBUF output_record);
 384
 385 int yaz_record_conv_opac_record(yaz_record_conv_t p,
 386                                 Z_OPACRecord *input_record,
 387                                 WRBUF output_record)
 388 {
 389     int ret = 0;
 390     struct yaz_record_conv_rule *r = p->rules;
 391     WRBUF res = wrbuf_alloc();
 392     yaz_marc_t mt = yaz_marc_create();
 393
 394     wrbuf_rewind(p->wr_error);
 395     yaz_marc_xml(mt, r->u.marc.output_format);
 396     if (r->u.marc.iconv_t)
 397         yaz_marc_iconv(mt, r->u.marc.iconv_t);
 398     yaz_opac_decode_wrbuf(mt, input_record, res);
 399     if (ret != -1)
 400     {
 401         ret = yaz_record_conv_record_rule(p,
 402                                           r->next,
 403                                           wrbuf_buf(res), wrbuf_len(res),
 404                                           output_record);
 405     }
 406     yaz_marc_destroy(mt);
 407     wrbuf_destroy(res);
 408     return ret;
 409 }
 410
 411 int yaz_record_conv_record(yaz_record_conv_t p,
 412                            const char *input_record_buf,
 413                            size_t input_record_len,
 414                            WRBUF output_record)
 415 {
 416     return yaz_record_conv_record_rule(p, p->rules,
 417                                        input_record_buf,
 418                                        input_record_len, output_record);
 419 }
 420
 421 static int yaz_record_conv_record_rule(yaz_record_conv_t p,
 422                                        struct yaz_record_conv_rule *r,
 423                                        const char *input_record_buf,
 424                                        size_t input_record_len,
 425                                        WRBUF output_record)
 426 {
 427     int ret = 0;
 428     WRBUF record = output_record; /* pointer transfer */
 429     wrbuf_rewind(p->wr_error);
 430
 431     wrbuf_write(record, input_record_buf, input_record_len);
 432     for (; ret == 0 && r; r = r->next)
 433     {
 434         if (r->which == YAZ_RECORD_CONV_RULE_MARC)
 435         {
 436             yaz_marc_t mt = yaz_marc_create();
 437
 438             yaz_marc_xml(mt, r->u.marc.output_format);
 439
 440             if (r->u.marc.iconv_t)
 441                 yaz_marc_iconv(mt, r->u.marc.iconv_t);
 442             if (r->u.marc.input_format == YAZ_MARC_ISO2709)
 443             {
 444                 int sz = yaz_marc_read_iso2709(mt, wrbuf_buf(record),
 445                                                wrbuf_len(record));
 446                 if (sz > 0)
 447                     ret = 0;
 448                 else
 449                     ret = -1;
 450             }
 451             else if (r->u.marc.input_format == YAZ_MARC_MARCXML)
 452             {
 453                 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
 454                                                wrbuf_len(record));
 455                 if (!doc)
 456                 {
 457                     wrbuf_printf(p->wr_error, "xmlParseMemory failed");
 458                     ret = -1;
 459                 }
 460                 else
 461                 {
 462                     ret = yaz_marc_read_xml(mt, xmlDocGetRootElement(doc));
 463                     if (ret)
 464                         wrbuf_printf(p->wr_error, "yaz_marc_read_xml failed");
 465                 }
 466                 xmlFreeDoc(doc);
 467             }
 468             else
 469             {
 470                 wrbuf_printf(p->wr_error, "unsupported input format");
 471                 ret = -1;
 472             }
 473             if (ret == 0)
 474             {
 475                 wrbuf_rewind(record);
 476                 ret = yaz_marc_write_mode(mt, record);
 477                 if (ret)
 478                     wrbuf_printf(p->wr_error, "yaz_marc_write_mode failed");
 479             }
 480             yaz_marc_destroy(mt);
 481         }
 482 #if YAZ_HAVE_XSLT
 483         else if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
 484         {
 485             xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
 486                                            wrbuf_len(record));
 487             if (!doc)
 488             {
 489                 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
 490                 ret = -1;
 491             }
 492             else
 493             {
 494                 xmlDocPtr res = xsltApplyStylesheet(r->u.xslt.xsp, doc, 0);
 495                 if (res)
 496                 {
 497                     xmlChar *out_buf = 0;
 498                     int out_len;
 499
 500 #if YAZ_HAVE_XSLTSAVERESULTTOSTRING
 501                     xsltSaveResultToString(&out_buf, &out_len, res,
 502                                            r->u.xslt.xsp);
 503 #else
 504                     xmlDocDumpFormatMemory (res, &out_buf, &out_len, 1);
 505 #endif
 506                     if (!out_buf)
 507                     {
 508                         wrbuf_printf(p->wr_error,
 509                                      "xsltSaveResultToString failed");
 510                         ret = -1;
 511                     }
 512                     else
 513                     {
 514                         wrbuf_rewind(record);
 515                         wrbuf_write(record, (const char *) out_buf, out_len);
 516
 517                         xmlFree(out_buf);
 518                     }
 519                     xmlFreeDoc(res);
 520                 }
 521                 else
 522                 {
 523                     wrbuf_printf(p->wr_error, "xsltApplyStylesheet failed");
 524                     ret = -1;
 525                 }
 526                 xmlFreeDoc(doc);
 527             }
 528         }
 529 #endif
 530     }
 531     return ret;
 532 }
 533
 534 const char *yaz_record_conv_get_error(yaz_record_conv_t p)
 535 {
 536     return wrbuf_cstr(p->wr_error);
 537 }
 538
 539 void yaz_record_conv_set_path(yaz_record_conv_t p, const char *path)
 540 {
 541     xfree(p->path);
 542     p->path = 0;
 543     if (path)
 544         p->path = xstrdup(path);
 545 }
 546 #endif
 547
 548 /*
 549  * Local variables:
 550  * c-basic-offset: 4
 551  * indent-tabs-mode: nil
 552  * End:
 553  * vim: shiftwidth=4 tabstop=8 expandtab
 554  */
 555