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'",
182 stylesheet, fullpath);
184 wrbuf_printf(p->wr_error, " with path '%s'", p->path);
188 xsp = xsltParseStylesheetFile((xmlChar*) fullpath);
191 wrbuf_printf(p->wr_error, "Element: <xslt stylesheet=\"%s\"/>:"
192 " xslt parse failed: %s", stylesheet, fullpath);
194 wrbuf_printf(p->wr_error, " with path '%s'", p->path);
195 wrbuf_printf(p->wr_error, " ("
200 "EXSLT not supported"
207 struct yaz_record_conv_rule *r =
208 add_rule(p, YAZ_RECORD_CONV_RULE_XSLT);
214 wrbuf_printf(p->wr_error, "xslt unsupported."
215 " YAZ compiled without XSLT support");
220 /** \brief parse 'marc' conversion node */
221 static int conv_marc(yaz_record_conv_t p, const xmlNode *ptr)
223 struct _xmlAttr *attr;
224 const char *input_charset = 0;
225 const char *output_charset = 0;
226 const char *input_format = 0;
227 const char *output_format = 0;
228 int input_format_mode = 0;
229 int output_format_mode = 0;
230 struct yaz_record_conv_rule *r;
233 for (attr = ptr->properties; attr; attr = attr->next)
235 if (!xmlStrcmp(attr->name, BAD_CAST "inputcharset") &&
236 attr->children && attr->children->type == XML_TEXT_NODE)
237 input_charset = (const char *) attr->children->content;
238 else if (!xmlStrcmp(attr->name, BAD_CAST "outputcharset") &&
239 attr->children && attr->children->type == XML_TEXT_NODE)
240 output_charset = (const char *) attr->children->content;
241 else if (!xmlStrcmp(attr->name, BAD_CAST "inputformat") &&
242 attr->children && attr->children->type == XML_TEXT_NODE)
243 input_format = (const char *) attr->children->content;
244 else if (!xmlStrcmp(attr->name, BAD_CAST "outputformat") &&
245 attr->children && attr->children->type == XML_TEXT_NODE)
246 output_format = (const char *) attr->children->content;
249 wrbuf_printf(p->wr_error, "Element <marc>: expected attributes"
250 "'inputformat', 'inputcharset', 'outputformat' or"
251 " 'outputcharset', got attribute '%s'",
258 wrbuf_printf(p->wr_error, "Element <marc>: "
259 "attribute 'inputformat' required");
262 else if (!strcmp(input_format, "marc"))
264 input_format_mode = YAZ_MARC_ISO2709;
266 else if (!strcmp(input_format, "xml"))
268 input_format_mode = YAZ_MARC_MARCXML;
269 /** Libxml2 generates UTF-8 encoding by default .
270 So we convert from UTF-8 to outputcharset (if defined)
272 if (!input_charset && output_charset)
273 input_charset = "utf-8";
277 wrbuf_printf(p->wr_error, "Element <marc inputformat='%s'>: "
278 " Unsupported input format"
279 " defined by attribute value",
286 wrbuf_printf(p->wr_error,
287 "Element <marc>: attribute 'outputformat' required");
290 else if (!strcmp(output_format, "line"))
292 output_format_mode = YAZ_MARC_LINE;
294 else if (!strcmp(output_format, "marcxml"))
296 output_format_mode = YAZ_MARC_MARCXML;
297 if (input_charset && !output_charset)
298 output_charset = "utf-8";
300 else if (!strcmp(output_format, "marc"))
302 output_format_mode = YAZ_MARC_ISO2709;
304 else if (!strcmp(output_format, "marcxchange"))
306 output_format_mode = YAZ_MARC_XCHANGE;
307 if (input_charset && !output_charset)
308 output_charset = "utf-8";
312 wrbuf_printf(p->wr_error, "Element <marc outputformat='%s'>: "
313 " Unsupported output format"
314 " defined by attribute value",
318 if (input_charset && output_charset)
320 cd = yaz_iconv_open(output_charset, input_charset);
323 wrbuf_printf(p->wr_error,
324 "Element <marc inputcharset='%s' outputcharset='%s'>:"
325 " Unsupported character set mapping"
326 " defined by attribute values",
327 input_charset, output_charset);
331 else if (input_charset)
333 wrbuf_printf(p->wr_error, "Element <marc>: "
334 "attribute 'outputcharset' missing");
337 else if (output_charset)
339 wrbuf_printf(p->wr_error, "Element <marc>: "
340 "attribute 'inputcharset' missing");
343 r = add_rule(p, YAZ_RECORD_CONV_RULE_MARC);
344 r->u.marc.iconv_t = cd;
346 r->u.marc.input_format = input_format_mode;
347 r->u.marc.output_format = output_format_mode;
351 int yaz_record_conv_configure(yaz_record_conv_t p, const xmlNode *ptr)
353 yaz_record_conv_reset(p);
355 /* parsing element children */
356 for (ptr = ptr->children; ptr; ptr = ptr->next)
358 if (ptr->type != XML_ELEMENT_NODE)
360 if (!strcmp((const char *) ptr->name, "xslt"))
362 if (conv_xslt(p, ptr))
365 else if (!strcmp((const char *) ptr->name, "marc"))
367 if (conv_marc(p, ptr))
372 wrbuf_printf(p->wr_error, "Element <backend>: expected "
373 "<marc> or <xslt> element, got <%s>"
381 static int yaz_record_conv_record_rule(yaz_record_conv_t p,
382 struct yaz_record_conv_rule *r,
383 const char *input_record_buf,
384 size_t input_record_len,
385 WRBUF output_record);
387 int yaz_record_conv_opac_record(yaz_record_conv_t p,
388 Z_OPACRecord *input_record,
392 struct yaz_record_conv_rule *r = p->rules;
393 WRBUF res = wrbuf_alloc();
394 yaz_marc_t mt = yaz_marc_create();
396 wrbuf_rewind(p->wr_error);
397 yaz_marc_xml(mt, r->u.marc.output_format);
398 if (r->u.marc.iconv_t)
399 yaz_marc_iconv(mt, r->u.marc.iconv_t);
400 yaz_opac_decode_wrbuf(mt, input_record, res);
403 ret = yaz_record_conv_record_rule(p,
405 wrbuf_buf(res), wrbuf_len(res),
408 yaz_marc_destroy(mt);
413 int yaz_record_conv_record(yaz_record_conv_t p,
414 const char *input_record_buf,
415 size_t input_record_len,
418 return yaz_record_conv_record_rule(p, p->rules,
420 input_record_len, output_record);
423 static int yaz_record_conv_record_rule(yaz_record_conv_t p,
424 struct yaz_record_conv_rule *r,
425 const char *input_record_buf,
426 size_t input_record_len,
430 WRBUF record = output_record; /* pointer transfer */
431 wrbuf_rewind(p->wr_error);
433 wrbuf_write(record, input_record_buf, input_record_len);
434 for (; ret == 0 && r; r = r->next)
436 if (r->which == YAZ_RECORD_CONV_RULE_MARC)
438 yaz_marc_t mt = yaz_marc_create();
440 yaz_marc_xml(mt, r->u.marc.output_format);
442 if (r->u.marc.iconv_t)
443 yaz_marc_iconv(mt, r->u.marc.iconv_t);
444 if (r->u.marc.input_format == YAZ_MARC_ISO2709)
446 int sz = yaz_marc_read_iso2709(mt, wrbuf_buf(record),
453 else if (r->u.marc.input_format == YAZ_MARC_MARCXML)
455 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
459 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
464 ret = yaz_marc_read_xml(mt, xmlDocGetRootElement(doc));
466 wrbuf_printf(p->wr_error, "yaz_marc_read_xml failed");
472 wrbuf_printf(p->wr_error, "unsupported input format");
477 wrbuf_rewind(record);
478 ret = yaz_marc_write_mode(mt, record);
480 wrbuf_printf(p->wr_error, "yaz_marc_write_mode failed");
482 yaz_marc_destroy(mt);
485 else if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
487 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
491 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
496 xmlDocPtr res = xsltApplyStylesheet(r->u.xslt.xsp, doc, 0);
499 xmlChar *out_buf = 0;
502 #if YAZ_HAVE_XSLTSAVERESULTTOSTRING
503 xsltSaveResultToString(&out_buf, &out_len, res,
506 xmlDocDumpFormatMemory (res, &out_buf, &out_len, 1);
510 wrbuf_printf(p->wr_error,
511 "xsltSaveResultToString failed");
516 wrbuf_rewind(record);
517 wrbuf_write(record, (const char *) out_buf, out_len);
525 wrbuf_printf(p->wr_error, "xsltApplyStylesheet failed");
536 const char *yaz_record_conv_get_error(yaz_record_conv_t p)
538 return wrbuf_cstr(p->wr_error);
541 void yaz_record_conv_set_path(yaz_record_conv_t p, const char *path)
546 p->path = xstrdup(path);
553 * indent-tabs-mode: nil
555 * vim: shiftwidth=4 tabstop=8 expandtab