2 * Copyright (C) 2005-2007, Index Data ApS
3 * See the file LICENSE for details.
5 * $Id: record_conv.c,v 1.17 2007-12-16 11:08:51 adam Exp $
9 * \brief Record Conversions utility
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>
23 #include <yaz/tpath.h>
24 #include <yaz/z-opac.h>
27 #include <libxml/parser.h>
28 #include <libxml/tree.h>
29 #include <libxml/xinclude.h>
31 #include <libxslt/xsltutils.h>
32 #include <libxslt/transform.h>
35 #include <libexslt/exslt.h>
38 /** \brief The internal structure for yaz_record_conv_t */
39 struct yaz_record_conv_struct {
40 /** \brief memory for configuration */
43 /** \brief conversion rules (allocated using NMEM) */
44 struct yaz_record_conv_rule *rules;
46 /** \brief pointer to last conversion rule pointer in chain */
47 struct yaz_record_conv_rule **rules_p;
49 /** \brief string buffer for error messages */
52 /** \brief path for opening files */
56 /** \brief tranformation types (rule types) */
57 enum YAZ_RECORD_CONV_RULE
59 YAZ_RECORD_CONV_RULE_XSLT,
60 YAZ_RECORD_CONV_RULE_MARC
64 /** \brief tranformation info (rule info) */
65 struct yaz_record_conv_rule {
66 enum YAZ_RECORD_CONV_RULE which;
70 xsltStylesheetPtr xsp;
79 struct yaz_record_conv_rule *next;
82 /** \brief reset rules+configuration */
83 static void yaz_record_conv_reset(yaz_record_conv_t p)
86 struct yaz_record_conv_rule *r;
87 for (r = p->rules; r; r = r->next)
89 if (r->which == YAZ_RECORD_CONV_RULE_MARC)
91 if (r->u.marc.iconv_t)
92 yaz_iconv_close(r->u.marc.iconv_t);
95 else if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
97 xsltFreeStylesheet(r->u.xslt.xsp);
101 wrbuf_rewind(p->wr_error);
106 p->rules_p = &p->rules;
109 yaz_record_conv_t yaz_record_conv_create()
111 yaz_record_conv_t p = (yaz_record_conv_t) xmalloc(sizeof(*p));
112 p->nmem = nmem_create();
113 p->wr_error = wrbuf_alloc();
120 yaz_record_conv_reset(p);
124 void yaz_record_conv_destroy(yaz_record_conv_t p)
128 yaz_record_conv_reset(p);
129 nmem_destroy(p->nmem);
130 wrbuf_destroy(p->wr_error);
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)
140 struct yaz_record_conv_rule *r = (struct yaz_record_conv_rule *)
141 nmem_malloc(p->nmem, sizeof(*r));
145 p->rules_p = &r->next;
149 /** \brief parse 'xslt' conversion node */
150 static int conv_xslt(yaz_record_conv_t p, const xmlNode *ptr)
153 struct _xmlAttr *attr;
154 const char *stylesheet = 0;
156 for (attr = ptr->properties; attr; attr = attr->next)
158 if (!xmlStrcmp(attr->name, BAD_CAST "stylesheet") &&
159 attr->children && attr->children->type == XML_TEXT_NODE)
160 stylesheet = (const char *) attr->children->content;
163 wrbuf_printf(p->wr_error, "Bad attribute '%s'"
164 "Expected stylesheet.", attr->name);
170 wrbuf_printf(p->wr_error, "Element <xslt>: "
171 "attribute 'stylesheet' expected");
177 xsltStylesheetPtr xsp;
178 if (!yaz_filepath_resolve(stylesheet, p->path, 0, fullpath))
180 wrbuf_printf(p->wr_error, "Element <xslt stylesheet=\"%s\"/>:"
181 " could not locate stylesheet '%s' with path '%s'",
182 stylesheet, fullpath, p->path);
185 xsp = xsltParseStylesheetFile((xmlChar*) fullpath);
188 wrbuf_printf(p->wr_error, "Element <xslt stylesheet=\"%s\"/>:"
189 " parsing stylesheet '%s' with path '%s' failed,"
193 " EXSLT not supported",
195 stylesheet, fullpath, p->path);
200 struct yaz_record_conv_rule *r =
201 add_rule(p, YAZ_RECORD_CONV_RULE_XSLT);
207 wrbuf_printf(p->wr_error, "xslt unsupported."
208 " YAZ compiled without XSLT support");
213 /** \brief parse 'marc' conversion node */
214 static int conv_marc(yaz_record_conv_t p, const xmlNode *ptr)
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;
226 for (attr = ptr->properties; attr; attr = attr->next)
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;
242 wrbuf_printf(p->wr_error, "Element <marc>: expected attributes"
243 "'inputformat', 'inputcharset', 'outputformat' or"
244 " 'outputcharset', got attribute '%s'",
251 wrbuf_printf(p->wr_error, "Element <marc>: "
252 "attribute 'inputformat' required");
255 else if (!strcmp(input_format, "marc"))
257 input_format_mode = YAZ_MARC_ISO2709;
259 else if (!strcmp(input_format, "xml"))
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)
265 if (!input_charset && output_charset)
266 input_charset = "utf-8";
270 wrbuf_printf(p->wr_error, "Element <marc inputformat='%s'>: "
271 " Unsupported input format"
272 " defined by attribute value",
279 wrbuf_printf(p->wr_error,
280 "Element <marc>: attribute 'outputformat' required");
283 else if (!strcmp(output_format, "line"))
285 output_format_mode = YAZ_MARC_LINE;
287 else if (!strcmp(output_format, "marcxml"))
289 output_format_mode = YAZ_MARC_MARCXML;
290 if (input_charset && !output_charset)
291 output_charset = "utf-8";
293 else if (!strcmp(output_format, "marc"))
295 output_format_mode = YAZ_MARC_ISO2709;
297 else if (!strcmp(output_format, "marcxchange"))
299 output_format_mode = YAZ_MARC_XCHANGE;
300 if (input_charset && !output_charset)
301 output_charset = "utf-8";
305 wrbuf_printf(p->wr_error, "Element <marc outputformat='%s'>: "
306 " Unsupported output format"
307 " defined by attribute value",
311 if (input_charset && output_charset)
313 cd = yaz_iconv_open(output_charset, input_charset);
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);
324 else if (input_charset)
326 wrbuf_printf(p->wr_error, "Element <marc>: "
327 "attribute 'outputcharset' missing");
330 else if (output_charset)
332 wrbuf_printf(p->wr_error, "Element <marc>: "
333 "attribute 'inputcharset' missing");
336 r = add_rule(p, YAZ_RECORD_CONV_RULE_MARC);
337 r->u.marc.iconv_t = cd;
339 r->u.marc.input_format = input_format_mode;
340 r->u.marc.output_format = output_format_mode;
344 int yaz_record_conv_configure(yaz_record_conv_t p, const xmlNode *ptr)
346 yaz_record_conv_reset(p);
348 /* parsing element children */
349 for (ptr = ptr->children; ptr; ptr = ptr->next)
351 if (ptr->type != XML_ELEMENT_NODE)
353 if (!strcmp((const char *) ptr->name, "xslt"))
355 if (conv_xslt(p, ptr))
358 else if (!strcmp((const char *) ptr->name, "marc"))
360 if (conv_marc(p, ptr))
365 wrbuf_printf(p->wr_error, "Element <backend>: expected "
366 "<marc> or <xslt> element, got <%s>"
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);
380 int yaz_record_conv_opac_record(yaz_record_conv_t p,
381 Z_OPACRecord *input_record,
385 struct yaz_record_conv_rule *r = p->rules;
386 WRBUF res = wrbuf_alloc();
387 yaz_marc_t mt = yaz_marc_create();
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);
396 ret = yaz_record_conv_record_rule(p,
398 wrbuf_buf(res), wrbuf_len(res),
401 yaz_marc_destroy(mt);
406 int yaz_record_conv_record(yaz_record_conv_t p,
407 const char *input_record_buf,
408 size_t input_record_len,
411 return yaz_record_conv_record_rule(p, p->rules,
413 input_record_len, output_record);
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,
423 WRBUF record = output_record; /* pointer transfer */
424 wrbuf_rewind(p->wr_error);
426 wrbuf_write(record, input_record_buf, input_record_len);
427 for (; ret == 0 && r; r = r->next)
429 if (r->which == YAZ_RECORD_CONV_RULE_MARC)
431 yaz_marc_t mt = yaz_marc_create();
433 yaz_marc_xml(mt, r->u.marc.output_format);
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)
439 int sz = yaz_marc_read_iso2709(mt, wrbuf_buf(record),
446 else if (r->u.marc.input_format == YAZ_MARC_MARCXML)
448 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
452 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
457 ret = yaz_marc_read_xml(mt, xmlDocGetRootElement(doc));
459 wrbuf_printf(p->wr_error, "yaz_marc_read_xml failed");
465 wrbuf_printf(p->wr_error, "unsupported input format");
470 wrbuf_rewind(record);
471 ret = yaz_marc_write_mode(mt, record);
473 wrbuf_printf(p->wr_error, "yaz_marc_write_mode failed");
475 yaz_marc_destroy(mt);
478 else if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
480 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
484 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
489 xmlDocPtr res = xsltApplyStylesheet(r->u.xslt.xsp, doc, 0);
492 xmlChar *out_buf = 0;
495 #if YAZ_HAVE_XSLTSAVERESULTTOSTRING
496 xsltSaveResultToString(&out_buf, &out_len, res,
499 xmlDocDumpFormatMemory (res, &out_buf, &out_len, 1);
503 wrbuf_printf(p->wr_error,
504 "xsltSaveResultToString failed");
509 wrbuf_rewind(record);
510 wrbuf_write(record, (const char *) out_buf, out_len);
518 wrbuf_printf(p->wr_error, "xsltApplyStylesheet failed");
529 const char *yaz_record_conv_get_error(yaz_record_conv_t p)
531 return wrbuf_cstr(p->wr_error);
534 void yaz_record_conv_set_path(yaz_record_conv_t p, const char *path)
539 p->path = xstrdup(path);
546 * indent-tabs-mode: nil
548 * vim: shiftwidth=4 tabstop=8 expandtab