src/record_conv.c

   1 /* This file is part of the YAZ toolkit.
   2  * Copyright (C) 1995-2010 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 /** \brief tranformation info (rule info) */
  62 struct yaz_record_conv_rule {
  63     enum YAZ_RECORD_CONV_RULE which;
  64     union {
  65 #if YAZ_HAVE_XSLT
  66         struct {
  67             xmlDocPtr xsp_doc;
  68         } xslt;
  69 #endif
  70         struct {
  71             const char *input_charset;
  72             const char *output_charset;
  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             ;
  90         }
  91 #if YAZ_HAVE_XSLT
  92         else if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
  93         {
  94             xmlFreeDoc(r->u.xslt.xsp_doc);
  95         }
  96 #endif
  97     }
  98     wrbuf_rewind(p->wr_error);
  99     nmem_reset(p->nmem);
 100
 101     p->rules = 0;
 102
 103     p->rules_p = &p->rules;
 104 }
 105
 106 yaz_record_conv_t yaz_record_conv_create()
 107 {
 108     yaz_record_conv_t p = (yaz_record_conv_t) xmalloc(sizeof(*p));
 109     p->nmem = nmem_create();
 110     p->wr_error = wrbuf_alloc();
 111     p->rules = 0;
 112     p->path = 0;
 113
 114 #if YAZ_HAVE_EXSLT
 115     exsltRegisterAll();
 116 #endif
 117     yaz_record_conv_reset(p);
 118     return p;
 119 }
 120
 121 void yaz_record_conv_destroy(yaz_record_conv_t p)
 122 {
 123     if (p)
 124     {
 125         yaz_record_conv_reset(p);
 126         nmem_destroy(p->nmem);
 127         wrbuf_destroy(p->wr_error);
 128         xfree(p->path);
 129         xfree(p);
 130     }
 131 }
 132
 133 /** \brief adds a rule */
 134 static struct yaz_record_conv_rule *add_rule(yaz_record_conv_t p,
 135                                              enum YAZ_RECORD_CONV_RULE type)
 136 {
 137     struct yaz_record_conv_rule *r = (struct yaz_record_conv_rule *)
 138         nmem_malloc(p->nmem, sizeof(*r));
 139     r->which = type;
 140     r->next = 0;
 141     *p->rules_p = r;
 142     p->rules_p = &r->next;
 143     return r;
 144 }
 145
 146 /** \brief parse 'xslt' conversion node */
 147 static int conv_xslt(yaz_record_conv_t p, const xmlNode *ptr)
 148 {
 149 #if YAZ_HAVE_XSLT
 150     struct _xmlAttr *attr;
 151     const char *stylesheet = 0;
 152
 153     for (attr = ptr->properties; attr; attr = attr->next)
 154     {
 155         if (!xmlStrcmp(attr->name, BAD_CAST "stylesheet") &&
 156             attr->children && attr->children->type == XML_TEXT_NODE)
 157             stylesheet = (const char *) attr->children->content;
 158         else
 159         {
 160             wrbuf_printf(p->wr_error, "Bad attribute '%s'"
 161                          "Expected stylesheet.", attr->name);
 162             return -1;
 163         }
 164     }
 165     if (!stylesheet)
 166     {
 167         wrbuf_printf(p->wr_error, "Element <xslt>: "
 168                      "attribute 'stylesheet' expected");
 169         return -1;
 170     }
 171     else
 172     {
 173         char fullpath[1024];
 174         xsltStylesheetPtr xsp;
 175         xmlDocPtr xsp_doc;
 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_doc = xmlParseFile(fullpath);
 187         if (!xsp_doc)
 188         {
 189             wrbuf_printf(p->wr_error, "Element: <xslt stylesheet=\"%s\"/>:"
 190                          " xml parse failed: %s", stylesheet, fullpath);
 191             if (p->path)
 192                 wrbuf_printf(p->wr_error, " with path '%s'", p->path);
 193             return -1;
 194         }
 195         xsp = xsltParseStylesheetDoc(xsp_doc);
 196         if (!xsp)
 197         {
 198             wrbuf_printf(p->wr_error, "Element: <xslt stylesheet=\"%s\"/>:"
 199                          " xslt parse failed: %s", stylesheet, fullpath);
 200             if (p->path)
 201                 wrbuf_printf(p->wr_error, " with path '%s'", p->path);
 202             wrbuf_printf(p->wr_error, " ("
 203 #if YAZ_HAVE_EXSLT
 204
 205                          "EXSLT enabled"
 206 #else
 207                          "EXSLT not supported"
 208 #endif
 209                          ")");
 210             return -1;
 211         }
 212         else
 213         {
 214             struct yaz_record_conv_rule *r =
 215                 add_rule(p, YAZ_RECORD_CONV_RULE_XSLT);
 216             r->u.xslt.xsp_doc = xmlCopyDoc(xsp_doc, 1);
 217             xsltFreeStylesheet(xsp); /* will free xsp_doc */
 218         }
 219     }
 220     return 0;
 221 #else
 222     wrbuf_printf(p->wr_error, "xslt unsupported."
 223                  " YAZ compiled without XSLT support");
 224     return -1;
 225 #endif
 226 }
 227
 228 /** \brief parse 'marc' conversion node */
 229 static int conv_marc(yaz_record_conv_t p, const xmlNode *ptr)
 230 {
 231     struct _xmlAttr *attr;
 232     const char *input_charset = 0;
 233     const char *output_charset = 0;
 234     const char *input_format = 0;
 235     const char *output_format = 0;
 236     int input_format_mode = 0;
 237     int output_format_mode = 0;
 238     struct yaz_record_conv_rule *r;
 239
 240     for (attr = ptr->properties; attr; attr = attr->next)
 241     {
 242         if (!xmlStrcmp(attr->name, BAD_CAST "inputcharset") &&
 243             attr->children && attr->children->type == XML_TEXT_NODE)
 244             input_charset = (const char *) attr->children->content;
 245         else if (!xmlStrcmp(attr->name, BAD_CAST "outputcharset") &&
 246             attr->children && attr->children->type == XML_TEXT_NODE)
 247             output_charset = (const char *) attr->children->content;
 248         else if (!xmlStrcmp(attr->name, BAD_CAST "inputformat") &&
 249             attr->children && attr->children->type == XML_TEXT_NODE)
 250             input_format = (const char *) attr->children->content;
 251         else if (!xmlStrcmp(attr->name, BAD_CAST "outputformat") &&
 252             attr->children && attr->children->type == XML_TEXT_NODE)
 253             output_format = (const char *) attr->children->content;
 254         else
 255         {
 256             wrbuf_printf(p->wr_error, "Element <marc>: expected attributes"
 257                          "'inputformat', 'inputcharset', 'outputformat' or"
 258                          " 'outputcharset', got attribute '%s'",
 259                          attr->name);
 260             return -1;
 261         }
 262     }
 263     if (!input_format)
 264     {
 265         wrbuf_printf(p->wr_error, "Element <marc>: "
 266                      "attribute 'inputformat' required");
 267         return -1;
 268     }
 269     else if (!strcmp(input_format, "marc"))
 270     {
 271         input_format_mode = YAZ_MARC_ISO2709;
 272     }
 273     else if (!strcmp(input_format, "xml"))
 274     {
 275         input_format_mode = YAZ_MARC_MARCXML;
 276         /** Libxml2 generates UTF-8 encoding by default .
 277             So we convert from UTF-8 to outputcharset (if defined)
 278         */
 279         if (!input_charset && output_charset)
 280             input_charset = "utf-8";
 281     }
 282     else
 283     {
 284         wrbuf_printf(p->wr_error, "Element <marc inputformat='%s'>: "
 285                      " Unsupported input format"
 286                      " defined by attribute value",
 287                      input_format);
 288         return -1;
 289     }
 290
 291     if (!output_format)
 292     {
 293         wrbuf_printf(p->wr_error,
 294                      "Element <marc>: attribute 'outputformat' required");
 295         return -1;
 296     }
 297     else if (!strcmp(output_format, "line"))
 298     {
 299         output_format_mode = YAZ_MARC_LINE;
 300     }
 301     else if (!strcmp(output_format, "marcxml"))
 302     {
 303         output_format_mode = YAZ_MARC_MARCXML;
 304         if (input_charset && !output_charset)
 305             output_charset = "utf-8";
 306     }
 307     else if (!strcmp(output_format, "marc"))
 308     {
 309         output_format_mode = YAZ_MARC_ISO2709;
 310     }
 311     else if (!strcmp(output_format, "marcxchange"))
 312     {
 313         output_format_mode = YAZ_MARC_XCHANGE;
 314         if (input_charset && !output_charset)
 315             output_charset = "utf-8";
 316     }
 317     else
 318     {
 319         wrbuf_printf(p->wr_error, "Element <marc outputformat='%s'>: "
 320                      " Unsupported output format"
 321                      " defined by attribute value",
 322                      output_format);
 323         return -1;
 324     }
 325     if (input_charset && output_charset)
 326     {
 327         yaz_iconv_t cd = yaz_iconv_open(output_charset, input_charset);
 328         if (!cd)
 329         {
 330             wrbuf_printf(p->wr_error,
 331                          "Element <marc inputcharset='%s' outputcharset='%s'>:"
 332                          " Unsupported character set mapping"
 333                          " defined by attribute values",
 334                          input_charset, output_charset);
 335             return -1;
 336         }
 337         yaz_iconv_close(cd);
 338     }
 339     else if (input_charset)
 340     {
 341         wrbuf_printf(p->wr_error, "Element <marc>: "
 342                      "attribute 'outputcharset' missing");
 343         return -1;
 344     }
 345     else if (output_charset)
 346     {
 347         wrbuf_printf(p->wr_error, "Element <marc>: "
 348                      "attribute 'inputcharset' missing");
 349         return -1;
 350     }
 351     r = add_rule(p, YAZ_RECORD_CONV_RULE_MARC);
 352
 353     r->u.marc.input_charset = nmem_strdup(p->nmem, input_charset);
 354     r->u.marc.output_charset = nmem_strdup(p->nmem, output_charset);
 355     r->u.marc.input_format = input_format_mode;
 356     r->u.marc.output_format = output_format_mode;
 357     return 0;
 358 }
 359
 360 int yaz_record_conv_configure(yaz_record_conv_t p, const xmlNode *ptr)
 361 {
 362     yaz_record_conv_reset(p);
 363
 364     /* parsing element children */
 365     for (ptr = ptr->children; ptr; ptr = ptr->next)
 366         {
 367             if (ptr->type != XML_ELEMENT_NODE)
 368                 continue;
 369             if (!strcmp((const char *) ptr->name, "xslt"))
 370                 {
 371                     if (conv_xslt(p, ptr))
 372                         return -1;
 373                 }
 374             else if (!strcmp((const char *) ptr->name, "marc"))
 375                 {
 376                     if (conv_marc(p, ptr))
 377                         return -1;
 378                 }
 379             else
 380                 {
 381                     wrbuf_printf(p->wr_error, "Element <backend>: expected "
 382                                  "<marc> or <xslt> element, got <%s>"
 383                                  , ptr->name);
 384                     return -1;
 385                 }
 386         }
 387     return 0;
 388 }
 389
 390 static int yaz_record_conv_record_rule(yaz_record_conv_t p,
 391                                        struct yaz_record_conv_rule *r,
 392                                        const char *input_record_buf,
 393                                        size_t input_record_len,
 394                                        WRBUF output_record);
 395
 396 int yaz_record_conv_opac_record(yaz_record_conv_t p,
 397                                 Z_OPACRecord *input_record,
 398                                 WRBUF output_record)
 399 {
 400     int ret = 0;
 401     struct yaz_record_conv_rule *r = p->rules;
 402     if (!r || r->which != YAZ_RECORD_CONV_RULE_MARC)
 403         ret = -1; /* no marc rule so we can't do OPAC */
 404     else
 405     {
 406         WRBUF res = wrbuf_alloc();
 407         yaz_marc_t mt = yaz_marc_create();
 408         yaz_iconv_t cd = yaz_iconv_open(r->u.marc.output_charset,
 409                                         r->u.marc.input_charset);
 410
 411         wrbuf_rewind(p->wr_error);
 412         yaz_marc_xml(mt, r->u.marc.output_format);
 413
 414         yaz_marc_iconv(mt, cd);
 415
 416         yaz_opac_decode_wrbuf(mt, input_record, res);
 417         if (ret != -1)
 418         {
 419             ret = yaz_record_conv_record_rule(p,
 420                                               r->next,
 421                                               wrbuf_buf(res), wrbuf_len(res),
 422                                               output_record);
 423         }
 424         yaz_marc_destroy(mt);
 425         if (cd)
 426             yaz_iconv_close(cd);
 427         wrbuf_destroy(res);
 428     }
 429     return ret;
 430 }
 431
 432 int yaz_record_conv_record(yaz_record_conv_t p,
 433                            const char *input_record_buf,
 434                            size_t input_record_len,
 435                            WRBUF output_record)
 436 {
 437     return yaz_record_conv_record_rule(p, p->rules,
 438                                        input_record_buf,
 439                                        input_record_len, output_record);
 440 }
 441
 442 static int yaz_record_conv_record_rule(yaz_record_conv_t p,
 443                                        struct yaz_record_conv_rule *r,
 444                                        const char *input_record_buf,
 445                                        size_t input_record_len,
 446                                        WRBUF output_record)
 447 {
 448     int ret = 0;
 449     WRBUF record = output_record; /* pointer transfer */
 450     wrbuf_rewind(p->wr_error);
 451
 452     wrbuf_write(record, input_record_buf, input_record_len);
 453     for (; ret == 0 && r; r = r->next)
 454     {
 455         if (r->which == YAZ_RECORD_CONV_RULE_MARC)
 456         {
 457             yaz_iconv_t cd =
 458                 yaz_iconv_open(r->u.marc.output_charset,
 459                                r->u.marc.input_charset);
 460             yaz_marc_t mt = yaz_marc_create();
 461
 462             yaz_marc_xml(mt, r->u.marc.output_format);
 463
 464             if (cd)
 465                 yaz_marc_iconv(mt, cd);
 466             if (r->u.marc.input_format == YAZ_MARC_ISO2709)
 467             {
 468                 int sz = yaz_marc_read_iso2709(mt, wrbuf_buf(record),
 469                                                wrbuf_len(record));
 470                 if (sz > 0)
 471                     ret = 0;
 472                 else
 473                     ret = -1;
 474             }
 475             else if (r->u.marc.input_format == YAZ_MARC_MARCXML ||
 476                                          r->u.marc.input_format == YAZ_MARC_TMARCXML)
 477             {
 478                 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
 479                                                wrbuf_len(record));
 480                 if (!doc)
 481                 {
 482                     wrbuf_printf(p->wr_error, "xmlParseMemory failed");
 483                     ret = -1;
 484                 }
 485                 else
 486                 {
 487                                         ret = yaz_marc_read_xml(mt, xmlDocGetRootElement(doc));
 488                     if (ret)
 489                         wrbuf_printf(p->wr_error, "yaz_marc_read_xml failed");
 490                 }
 491                 xmlFreeDoc(doc);
 492             }
 493             else
 494             {
 495                 wrbuf_printf(p->wr_error, "unsupported input format");
 496                 ret = -1;
 497             }
 498             if (ret == 0)
 499             {
 500                 wrbuf_rewind(record);
 501                 ret = yaz_marc_write_mode(mt, record);
 502                 if (ret)
 503                     wrbuf_printf(p->wr_error, "yaz_marc_write_mode failed");
 504             }
 505             if (cd)
 506                 yaz_iconv_close(cd);
 507             yaz_marc_destroy(mt);
 508         }
 509 #if YAZ_HAVE_XSLT
 510         else if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
 511         {
 512             xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
 513                                            wrbuf_len(record));
 514             if (!doc)
 515             {
 516                 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
 517                 ret = -1;
 518             }
 519             else
 520             {
 521                 xmlDocPtr xsp_doc = xmlCopyDoc(r->u.xslt.xsp_doc, 1);
 522                 xsltStylesheetPtr xsp = xsltParseStylesheetDoc(xsp_doc);
 523                 xmlDocPtr res = xsltApplyStylesheet(xsp, doc, 0);
 524                 if (res)
 525                 {
 526                     xmlChar *out_buf = 0;
 527                     int out_len;
 528
 529 #if YAZ_HAVE_XSLTSAVERESULTTOSTRING
 530                     xsltSaveResultToString(&out_buf, &out_len, res, xsp);
 531 #else
 532                     xmlDocDumpFormatMemory (res, &out_buf, &out_len, 1);
 533 #endif
 534                     if (!out_buf)
 535                     {
 536                         wrbuf_printf(p->wr_error,
 537                                      "xsltSaveResultToString failed");
 538                         ret = -1;
 539                     }
 540                     else
 541                     {
 542                         wrbuf_rewind(record);
 543                         wrbuf_write(record, (const char *) out_buf, out_len);
 544
 545                         xmlFree(out_buf);
 546                     }
 547                     xmlFreeDoc(res);
 548                 }
 549                 else
 550                 {
 551                     wrbuf_printf(p->wr_error, "xsltApplyStylesheet failed");
 552                     ret = -1;
 553                 }
 554                 xmlFreeDoc(doc);
 555                 xsltFreeStylesheet(xsp); /* frees xsp_doc too */
 556             }
 557         }
 558 #endif
 559     }
 560     return ret;
 561 }
 562
 563 const char *yaz_record_conv_get_error(yaz_record_conv_t p)
 564 {
 565     return wrbuf_cstr(p->wr_error);
 566 }
 567
 568 void yaz_record_conv_set_path(yaz_record_conv_t p, const char *path)
 569 {
 570     xfree(p->path);
 571     p->path = 0;
 572     if (path)
 573         p->path = xstrdup(path);
 574 }
 575 #endif
 576
 577 /*
 578  * Local variables:
 579  * c-basic-offset: 4
 580  * c-file-style: "Stroustrup"
 581  * indent-tabs-mode: nil
 582  * End:
 583  * vim: shiftwidth=4 tabstop=8 expandtab
 584  */
 585