1 /* This file is part of the YAZ toolkit.
2 * Copyright (C) 1995-2008 Index Data
3 * See the file LICENSE for details.
7 * \brief Record Conversions utility
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>
21 #include <yaz/tpath.h>
22 #include <yaz/z-opac.h>
25 #include <libxml/parser.h>
26 #include <libxml/tree.h>
27 #include <libxml/xinclude.h>
29 #include <libxslt/xsltutils.h>
30 #include <libxslt/transform.h>
33 #include <libexslt/exslt.h>
36 /** \brief The internal structure for yaz_record_conv_t */
37 struct yaz_record_conv_struct {
38 /** \brief memory for configuration */
41 /** \brief conversion rules (allocated using NMEM) */
42 struct yaz_record_conv_rule *rules;
44 /** \brief pointer to last conversion rule pointer in chain */
45 struct yaz_record_conv_rule **rules_p;
47 /** \brief string buffer for error messages */
50 /** \brief path for opening files */
54 /** \brief tranformation types (rule types) */
55 enum YAZ_RECORD_CONV_RULE
57 YAZ_RECORD_CONV_RULE_XSLT,
58 YAZ_RECORD_CONV_RULE_MARC
62 /** \brief tranformation info (rule info) */
63 struct yaz_record_conv_rule {
64 enum YAZ_RECORD_CONV_RULE which;
68 xsltStylesheetPtr xsp;
77 struct yaz_record_conv_rule *next;
80 /** \brief reset rules+configuration */
81 static void yaz_record_conv_reset(yaz_record_conv_t p)
84 struct yaz_record_conv_rule *r;
85 for (r = p->rules; r; r = r->next)
87 if (r->which == YAZ_RECORD_CONV_RULE_MARC)
89 if (r->u.marc.iconv_t)
90 yaz_iconv_close(r->u.marc.iconv_t);
93 else if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
95 xsltFreeStylesheet(r->u.xslt.xsp);
99 wrbuf_rewind(p->wr_error);
104 p->rules_p = &p->rules;
107 yaz_record_conv_t yaz_record_conv_create()
109 yaz_record_conv_t p = (yaz_record_conv_t) xmalloc(sizeof(*p));
110 p->nmem = nmem_create();
111 p->wr_error = wrbuf_alloc();
118 yaz_record_conv_reset(p);
122 void yaz_record_conv_destroy(yaz_record_conv_t p)
126 yaz_record_conv_reset(p);
127 nmem_destroy(p->nmem);
128 wrbuf_destroy(p->wr_error);
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)
138 struct yaz_record_conv_rule *r = (struct yaz_record_conv_rule *)
139 nmem_malloc(p->nmem, sizeof(*r));
143 p->rules_p = &r->next;
147 /** \brief parse 'xslt' conversion node */
148 static int conv_xslt(yaz_record_conv_t p, const xmlNode *ptr)
151 struct _xmlAttr *attr;
152 const char *stylesheet = 0;
154 for (attr = ptr->properties; attr; attr = attr->next)
156 if (!xmlStrcmp(attr->name, BAD_CAST "stylesheet") &&
157 attr->children && attr->children->type == XML_TEXT_NODE)
158 stylesheet = (const char *) attr->children->content;
161 wrbuf_printf(p->wr_error, "Bad attribute '%s'"
162 "Expected stylesheet.", attr->name);
168 wrbuf_printf(p->wr_error, "Element <xslt>: "
169 "attribute 'stylesheet' expected");
175 xsltStylesheetPtr xsp;
176 if (!yaz_filepath_resolve(stylesheet, p->path, 0, fullpath))
178 wrbuf_printf(p->wr_error, "Element <xslt stylesheet=\"%s\"/>:"
179 " could not locate stylesheet '%s'",
180 stylesheet, fullpath);
182 wrbuf_printf(p->wr_error, " with path '%s'", p->path);
186 xsp = xsltParseStylesheetFile((xmlChar*) fullpath);
189 wrbuf_printf(p->wr_error, "Element: <xslt stylesheet=\"%s\"/>:"
190 " xslt parse failed: %s", stylesheet, fullpath);
192 wrbuf_printf(p->wr_error, " with path '%s'", p->path);
193 wrbuf_printf(p->wr_error, " ("
198 "EXSLT not supported"
205 struct yaz_record_conv_rule *r =
206 add_rule(p, YAZ_RECORD_CONV_RULE_XSLT);
212 wrbuf_printf(p->wr_error, "xslt unsupported."
213 " YAZ compiled without XSLT support");
218 /** \brief parse 'marc' conversion node */
219 static int conv_marc(yaz_record_conv_t p, const xmlNode *ptr)
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;
231 for (attr = ptr->properties; attr; attr = attr->next)
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;
247 wrbuf_printf(p->wr_error, "Element <marc>: expected attributes"
248 "'inputformat', 'inputcharset', 'outputformat' or"
249 " 'outputcharset', got attribute '%s'",
256 wrbuf_printf(p->wr_error, "Element <marc>: "
257 "attribute 'inputformat' required");
260 else if (!strcmp(input_format, "marc"))
262 input_format_mode = YAZ_MARC_ISO2709;
264 else if (!strcmp(input_format, "xml"))
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)
270 if (!input_charset && output_charset)
271 input_charset = "utf-8";
275 wrbuf_printf(p->wr_error, "Element <marc inputformat='%s'>: "
276 " Unsupported input format"
277 " defined by attribute value",
284 wrbuf_printf(p->wr_error,
285 "Element <marc>: attribute 'outputformat' required");
288 else if (!strcmp(output_format, "line"))
290 output_format_mode = YAZ_MARC_LINE;
292 else if (!strcmp(output_format, "marcxml"))
294 output_format_mode = YAZ_MARC_MARCXML;
295 if (input_charset && !output_charset)
296 output_charset = "utf-8";
298 else if (!strcmp(output_format, "marc"))
300 output_format_mode = YAZ_MARC_ISO2709;
302 else if (!strcmp(output_format, "marcxchange"))
304 output_format_mode = YAZ_MARC_XCHANGE;
305 if (input_charset && !output_charset)
306 output_charset = "utf-8";
310 wrbuf_printf(p->wr_error, "Element <marc outputformat='%s'>: "
311 " Unsupported output format"
312 " defined by attribute value",
316 if (input_charset && output_charset)
318 cd = yaz_iconv_open(output_charset, input_charset);
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);
329 else if (input_charset)
331 wrbuf_printf(p->wr_error, "Element <marc>: "
332 "attribute 'outputcharset' missing");
335 else if (output_charset)
337 wrbuf_printf(p->wr_error, "Element <marc>: "
338 "attribute 'inputcharset' missing");
341 r = add_rule(p, YAZ_RECORD_CONV_RULE_MARC);
342 r->u.marc.iconv_t = cd;
344 r->u.marc.input_format = input_format_mode;
345 r->u.marc.output_format = output_format_mode;
349 int yaz_record_conv_configure(yaz_record_conv_t p, const xmlNode *ptr)
351 yaz_record_conv_reset(p);
353 /* parsing element children */
354 for (ptr = ptr->children; ptr; ptr = ptr->next)
356 if (ptr->type != XML_ELEMENT_NODE)
358 if (!strcmp((const char *) ptr->name, "xslt"))
360 if (conv_xslt(p, ptr))
363 else if (!strcmp((const char *) ptr->name, "marc"))
365 if (conv_marc(p, ptr))
370 wrbuf_printf(p->wr_error, "Element <backend>: expected "
371 "<marc> or <xslt> element, got <%s>"
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);
385 int yaz_record_conv_opac_record(yaz_record_conv_t p,
386 Z_OPACRecord *input_record,
390 struct yaz_record_conv_rule *r = p->rules;
391 WRBUF res = wrbuf_alloc();
392 yaz_marc_t mt = yaz_marc_create();
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);
401 ret = yaz_record_conv_record_rule(p,
403 wrbuf_buf(res), wrbuf_len(res),
406 yaz_marc_destroy(mt);
411 int yaz_record_conv_record(yaz_record_conv_t p,
412 const char *input_record_buf,
413 size_t input_record_len,
416 return yaz_record_conv_record_rule(p, p->rules,
418 input_record_len, output_record);
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,
428 WRBUF record = output_record; /* pointer transfer */
429 wrbuf_rewind(p->wr_error);
431 wrbuf_write(record, input_record_buf, input_record_len);
432 for (; ret == 0 && r; r = r->next)
434 if (r->which == YAZ_RECORD_CONV_RULE_MARC)
436 yaz_marc_t mt = yaz_marc_create();
438 yaz_marc_xml(mt, r->u.marc.output_format);
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)
444 int sz = yaz_marc_read_iso2709(mt, wrbuf_buf(record),
451 else if (r->u.marc.input_format == YAZ_MARC_MARCXML)
453 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
457 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
462 ret = yaz_marc_read_xml(mt, xmlDocGetRootElement(doc));
464 wrbuf_printf(p->wr_error, "yaz_marc_read_xml failed");
470 wrbuf_printf(p->wr_error, "unsupported input format");
475 wrbuf_rewind(record);
476 ret = yaz_marc_write_mode(mt, record);
478 wrbuf_printf(p->wr_error, "yaz_marc_write_mode failed");
480 yaz_marc_destroy(mt);
483 else if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
485 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
489 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
494 xmlDocPtr res = xsltApplyStylesheet(r->u.xslt.xsp, doc, 0);
497 xmlChar *out_buf = 0;
500 #if YAZ_HAVE_XSLTSAVERESULTTOSTRING
501 xsltSaveResultToString(&out_buf, &out_len, res,
504 xmlDocDumpFormatMemory (res, &out_buf, &out_len, 1);
508 wrbuf_printf(p->wr_error,
509 "xsltSaveResultToString failed");
514 wrbuf_rewind(record);
515 wrbuf_write(record, (const char *) out_buf, out_len);
523 wrbuf_printf(p->wr_error, "xsltApplyStylesheet failed");
534 const char *yaz_record_conv_get_error(yaz_record_conv_t p)
536 return wrbuf_cstr(p->wr_error);
539 void yaz_record_conv_set_path(yaz_record_conv_t p, const char *path)
544 p->path = xstrdup(path);
551 * indent-tabs-mode: nil
553 * vim: shiftwidth=4 tabstop=8 expandtab