2 * Copyright (C) 2005-2007, Index Data ApS
3 * See the file LICENSE for details.
5 * $Id: record_conv.c,v 1.14 2007-01-03 08:42:15 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>
26 #include <libxml/parser.h>
27 #include <libxml/tree.h>
28 #include <libxml/xinclude.h>
30 #include <libxslt/xsltutils.h>
31 #include <libxslt/transform.h>
34 #include <libexslt/exslt.h>
37 /** \brief The internal structure for yaz_record_conv_t */
38 struct yaz_record_conv_struct {
39 /** \brief memory for configuration */
42 /** \brief conversion rules (allocated using NMEM) */
43 struct yaz_record_conv_rule *rules;
45 /** \brief pointer to last conversion rule pointer in chain */
46 struct yaz_record_conv_rule **rules_p;
48 /** \brief string buffer for error messages */
51 /** \brief path for opening files */
55 /** \brief tranformation types (rule types) */
56 enum YAZ_RECORD_CONV_RULE
58 YAZ_RECORD_CONV_RULE_XSLT,
59 YAZ_RECORD_CONV_RULE_MARC
63 /** \brief tranformation info (rule info) */
64 struct yaz_record_conv_rule {
65 enum YAZ_RECORD_CONV_RULE which;
69 xsltStylesheetPtr xsp;
78 struct yaz_record_conv_rule *next;
81 /** \brief reset rules+configuration */
82 static void yaz_record_conv_reset(yaz_record_conv_t p)
85 struct yaz_record_conv_rule *r;
86 for (r = p->rules; r; r = r->next)
88 if (r->which == YAZ_RECORD_CONV_RULE_MARC)
90 if (r->u.marc.iconv_t)
91 yaz_iconv_close(r->u.marc.iconv_t);
94 else if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
96 xsltFreeStylesheet(r->u.xslt.xsp);
100 wrbuf_rewind(p->wr_error);
105 p->rules_p = &p->rules;
108 yaz_record_conv_t yaz_record_conv_create()
110 yaz_record_conv_t p = xmalloc(sizeof(*p));
111 p->nmem = nmem_create();
112 p->wr_error = wrbuf_alloc();
119 yaz_record_conv_reset(p);
123 void yaz_record_conv_destroy(yaz_record_conv_t p)
127 yaz_record_conv_reset(p);
128 nmem_destroy(p->nmem);
129 wrbuf_free(p->wr_error, 1);
135 /** \brief adds a rule */
136 static struct yaz_record_conv_rule *add_rule(yaz_record_conv_t p,
137 enum YAZ_RECORD_CONV_RULE type)
139 struct yaz_record_conv_rule *r = 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' with path '%s'",
180 stylesheet, fullpath, p->path);
183 xsp = xsltParseStylesheetFile((xmlChar*) fullpath);
186 wrbuf_printf(p->wr_error, "Element <xslt stylesheet=\"%s\"/>:"
187 " parsing stylesheet '%s' with path '%s' failed,"
191 " EXSLT not supported",
193 stylesheet, fullpath, p->path);
198 struct yaz_record_conv_rule *r =
199 add_rule(p, YAZ_RECORD_CONV_RULE_XSLT);
205 wrbuf_printf(p->wr_error, "xslt unsupported."
206 " YAZ compiled without XSLT support");
211 /** \brief parse 'marc' conversion node */
212 static int conv_marc(yaz_record_conv_t p, const xmlNode *ptr)
214 struct _xmlAttr *attr;
215 const char *input_charset = 0;
216 const char *output_charset = 0;
217 const char *input_format = 0;
218 const char *output_format = 0;
219 int input_format_mode = 0;
220 int output_format_mode = 0;
221 struct yaz_record_conv_rule *r;
224 for (attr = ptr->properties; attr; attr = attr->next)
226 if (!xmlStrcmp(attr->name, BAD_CAST "inputcharset") &&
227 attr->children && attr->children->type == XML_TEXT_NODE)
228 input_charset = (const char *) attr->children->content;
229 else if (!xmlStrcmp(attr->name, BAD_CAST "outputcharset") &&
230 attr->children && attr->children->type == XML_TEXT_NODE)
231 output_charset = (const char *) attr->children->content;
232 else if (!xmlStrcmp(attr->name, BAD_CAST "inputformat") &&
233 attr->children && attr->children->type == XML_TEXT_NODE)
234 input_format = (const char *) attr->children->content;
235 else if (!xmlStrcmp(attr->name, BAD_CAST "outputformat") &&
236 attr->children && attr->children->type == XML_TEXT_NODE)
237 output_format = (const char *) attr->children->content;
240 wrbuf_printf(p->wr_error, "Element <marc>: expected attributes"
241 "'inputformat', 'inputcharset', 'outputformat' or"
242 " 'outputcharset', got attribute '%s'",
249 wrbuf_printf(p->wr_error, "Element <marc>: "
250 "attribute 'inputformat' required");
253 else if (!strcmp(input_format, "marc"))
255 input_format_mode = YAZ_MARC_ISO2709;
257 else if (!strcmp(input_format, "xml"))
259 input_format_mode = YAZ_MARC_MARCXML;
260 /** Libxml2 generates UTF-8 encoding by default .
261 So we convert from UTF-8 to outputcharset (if defined)
263 if (!input_charset && output_charset)
264 input_charset = "utf-8";
268 wrbuf_printf(p->wr_error, "Element <marc inputformat='%s'>: "
269 " Unsupported input format"
270 " defined by attribute value",
277 wrbuf_printf(p->wr_error,
278 "Element <marc>: attribute 'outputformat' required");
281 else if (!strcmp(output_format, "line"))
283 output_format_mode = YAZ_MARC_LINE;
285 else if (!strcmp(output_format, "marcxml"))
287 output_format_mode = YAZ_MARC_MARCXML;
288 if (input_charset && !output_charset)
289 output_charset = "utf-8";
291 else if (!strcmp(output_format, "marc"))
293 output_format_mode = YAZ_MARC_ISO2709;
295 else if (!strcmp(output_format, "marcxchange"))
297 output_format_mode = YAZ_MARC_XCHANGE;
298 if (input_charset && !output_charset)
299 output_charset = "utf-8";
303 wrbuf_printf(p->wr_error, "Element <marc outputformat='%s'>: "
304 " Unsupported output format"
305 " defined by attribute value",
309 if (input_charset && output_charset)
311 cd = yaz_iconv_open(output_charset, input_charset);
314 wrbuf_printf(p->wr_error,
315 "Element <marc inputcharset='%s' outputcharset='%s'>:"
316 " Unsupported character set mapping"
317 " defined by attribute values",
318 input_charset, output_charset);
322 else if (input_charset)
324 wrbuf_printf(p->wr_error, "Element <marc>: "
325 "attribute 'outputcharset' missing");
328 else if (output_charset)
330 wrbuf_printf(p->wr_error, "Element <marc>: "
331 "attribute 'inputcharset' missing");
334 r = add_rule(p, YAZ_RECORD_CONV_RULE_MARC);
335 r->u.marc.iconv_t = cd;
337 r->u.marc.input_format = input_format_mode;
338 r->u.marc.output_format = output_format_mode;
342 int yaz_record_conv_configure(yaz_record_conv_t p, const xmlNode *ptr)
344 yaz_record_conv_reset(p);
346 /* parsing element children */
347 for (ptr = ptr->children; ptr; ptr = ptr->next)
349 if (ptr->type != XML_ELEMENT_NODE)
351 if (!strcmp((const char *) ptr->name, "xslt"))
353 if (conv_xslt(p, ptr))
356 else if (!strcmp((const char *) ptr->name, "marc"))
358 if (conv_marc(p, ptr))
363 wrbuf_printf(p->wr_error, "Element <backend>: expected "
364 "<marc> or <xslt> element, got <%s>"
372 int yaz_record_conv_record(yaz_record_conv_t p,
373 const char *input_record_buf,
374 size_t input_record_len,
378 WRBUF record = output_record; /* pointer transfer */
379 struct yaz_record_conv_rule *r = p->rules;
380 wrbuf_rewind(p->wr_error);
382 wrbuf_write(record, input_record_buf, input_record_len);
383 for (; ret == 0 && r; r = r->next)
385 if (r->which == YAZ_RECORD_CONV_RULE_MARC)
387 yaz_marc_t mt = yaz_marc_create();
389 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 if (r->u.marc.input_format == YAZ_MARC_ISO2709)
395 int sz = yaz_marc_read_iso2709(mt, wrbuf_buf(record),
402 else if (r->u.marc.input_format == YAZ_MARC_MARCXML)
404 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
408 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
413 ret = yaz_marc_read_xml(mt, xmlDocGetRootElement(doc));
415 wrbuf_printf(p->wr_error, "yaz_marc_read_xml failed");
421 wrbuf_printf(p->wr_error, "unsupported input format");
426 wrbuf_rewind(record);
427 ret = yaz_marc_write_mode(mt, record);
429 wrbuf_printf(p->wr_error, "yaz_marc_write_mode failed");
431 yaz_marc_destroy(mt);
434 else if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
436 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
440 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
445 xmlDocPtr res = xsltApplyStylesheet(r->u.xslt.xsp, doc, 0);
448 xmlChar *out_buf = 0;
451 #if YAZ_HAVE_XSLTSAVERESULTTOSTRING
452 xsltSaveResultToString(&out_buf, &out_len, res,
455 xmlDocDumpFormatMemory (res, &out_buf, &out_len, 1);
459 wrbuf_printf(p->wr_error,
460 "xsltSaveResultToString failed");
465 wrbuf_rewind(record);
466 wrbuf_write(record, (const char *) out_buf, out_len);
474 wrbuf_printf(p->wr_error, "xsltApplyStylesheet failed");
485 const char *yaz_record_conv_get_error(yaz_record_conv_t p)
487 return wrbuf_buf(p->wr_error);
490 void yaz_record_conv_set_path(yaz_record_conv_t p, const char *path)
495 p->path = xstrdup(path);
502 * indent-tabs-mode: nil
504 * vim: shiftwidth=4 tabstop=8 expandtab