2 * Copyright (C) 2005-2006, Index Data ApS
3 * See the file LICENSE for details.
5 * $Id: record_conv.c,v 1.4 2006-05-04 20:00:45 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>
29 #include <libxslt/xsltutils.h>
30 #include <libxslt/transform.h>
32 /** \brief The internal structure for yaz_record_conv_t */
33 struct yaz_record_conv_struct {
34 /** \brief memory for configuration */
37 /** \brief conversion rules (allocated using NMEM) */
38 struct yaz_record_conv_rule *rules;
40 /** \brief pointer to last conversion rule pointer in chain */
41 struct yaz_record_conv_rule **rules_p;
43 /** \brief string buffer for error messages */
46 /** \brief path for opening files */
50 /** \brief tranformation types (rule types) */
51 enum YAZ_RECORD_CONV_RULE
53 YAZ_RECORD_CONV_RULE_XSLT,
54 YAZ_RECORD_CONV_RULE_MARC
58 /** \brief tranformation info (rule info) */
59 struct yaz_record_conv_rule {
60 enum YAZ_RECORD_CONV_RULE which;
63 xsltStylesheetPtr xsp;
72 struct yaz_record_conv_rule *next;
75 /** \brief reset rules+configuration */
76 static void yaz_record_conv_reset(yaz_record_conv_t p)
78 struct yaz_record_conv_rule *r;
79 for (r = p->rules; r; r = r->next)
81 if (r->which == YAZ_RECORD_CONV_RULE_MARC)
83 if (r->u.marc.iconv_t)
84 yaz_iconv_close(r->u.marc.iconv_t);
86 else if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
88 xsltFreeStylesheet(r->u.xslt.xsp);
91 wrbuf_rewind(p->wr_error);
96 p->rules_p = &p->rules;
99 yaz_record_conv_t yaz_record_conv_create()
101 yaz_record_conv_t p = xmalloc(sizeof(*p));
102 p->nmem = nmem_create();
103 p->wr_error = wrbuf_alloc();
107 yaz_record_conv_reset(p);
111 void yaz_record_conv_destroy(yaz_record_conv_t p)
115 yaz_record_conv_reset(p);
116 nmem_destroy(p->nmem);
117 wrbuf_free(p->wr_error, 1);
123 /** \brief adds a rule */
124 static struct yaz_record_conv_rule *add_rule(yaz_record_conv_t p,
125 enum YAZ_RECORD_CONV_RULE type)
127 struct yaz_record_conv_rule *r = nmem_malloc(p->nmem, sizeof(*r));
131 p->rules_p = &r->next;
135 /** \brief parse 'xslt' conversion node */
136 static int conv_xslt(yaz_record_conv_t p, const xmlNode *ptr)
138 struct _xmlAttr *attr;
139 const char *stylesheet = 0;
141 for (attr = ptr->properties; attr; attr = attr->next)
143 if (!xmlStrcmp(attr->name, BAD_CAST "stylesheet") &&
144 attr->children && attr->children->type == XML_TEXT_NODE)
145 stylesheet = (const char *) attr->children->content;
148 wrbuf_printf(p->wr_error, "Bad attribute '%s'."
149 "Expected stylesheet.", attr->name);
155 wrbuf_printf(p->wr_error, "Missing attribute 'stylesheet'");
161 xsltStylesheetPtr xsp;
162 if (!yaz_filepath_resolve(stylesheet, p->path, 0, fullpath))
164 wrbuf_printf(p->wr_error, "could not locate '%s'. Path=%s",
165 stylesheet, p->path);
168 xsp = xsltParseStylesheetFile((xmlChar*) fullpath);
171 wrbuf_printf(p->wr_error, "xsltParseStylesheetFile failed'");
176 struct yaz_record_conv_rule *r =
177 add_rule(p, YAZ_RECORD_CONV_RULE_XSLT);
184 /** \brief parse 'marc' conversion node */
185 static int conv_marc(yaz_record_conv_t p, const xmlNode *ptr)
187 struct _xmlAttr *attr;
188 const char *input_charset = 0;
189 const char *output_charset = 0;
190 const char *input_format = 0;
191 const char *output_format = 0;
192 int input_format_mode = 0;
193 int output_format_mode = 0;
194 struct yaz_record_conv_rule *r;
197 for (attr = ptr->properties; attr; attr = attr->next)
199 if (!xmlStrcmp(attr->name, BAD_CAST "inputcharset") &&
200 attr->children && attr->children->type == XML_TEXT_NODE)
201 input_charset = (const char *) attr->children->content;
202 else if (!xmlStrcmp(attr->name, BAD_CAST "outputcharset") &&
203 attr->children && attr->children->type == XML_TEXT_NODE)
204 output_charset = (const char *) attr->children->content;
205 else if (!xmlStrcmp(attr->name, BAD_CAST "inputformat") &&
206 attr->children && attr->children->type == XML_TEXT_NODE)
207 input_format = (const char *) attr->children->content;
208 else if (!xmlStrcmp(attr->name, BAD_CAST "outputformat") &&
209 attr->children && attr->children->type == XML_TEXT_NODE)
210 output_format = (const char *) attr->children->content;
213 wrbuf_printf(p->wr_error, "Bad attribute '%s'.", attr->name);
219 wrbuf_printf(p->wr_error, "Attribute 'inputformat' required");
222 else if (!strcmp(input_format, "marc"))
224 input_format_mode = YAZ_MARC_ISO2709;
226 else if (!strcmp(input_format, "xml"))
228 input_format_mode = YAZ_MARC_MARCXML;
229 /** Libxml2 generates UTF-8 encoding by default .
230 So we convert from UTF-8 to outputcharset (if defined)
232 if (!input_charset && output_charset)
233 input_charset = "utf-8";
237 wrbuf_printf(p->wr_error, "Bad inputformat: '%s'", input_format);
243 wrbuf_printf(p->wr_error, "Attribute 'outputformat' required");
246 else if (!strcmp(output_format, "line"))
248 output_format_mode = YAZ_MARC_LINE;
250 else if (!strcmp(output_format, "marcxml"))
252 output_format_mode = YAZ_MARC_MARCXML;
253 if (input_charset && !output_charset)
254 output_charset = "utf-8";
256 else if (!strcmp(output_format, "marc"))
258 output_format_mode = YAZ_MARC_ISO2709;
260 else if (!strcmp(output_format, "marcxchange"))
262 output_format_mode = YAZ_MARC_XCHANGE;
263 if (input_charset && !output_charset)
264 output_charset = "utf-8";
268 wrbuf_printf(p->wr_error, "Bad outputformat: '%s'", input_format);
271 if (input_charset && output_charset)
273 cd = yaz_iconv_open(output_charset, input_charset);
276 wrbuf_printf(p->wr_error, "Unsupported character set mamping"
277 " inputcharset=%s outputcharset=%s",
278 input_charset, output_charset);
282 else if (input_charset)
284 wrbuf_printf(p->wr_error, "Attribute 'outputcharset' missing");
287 else if (output_charset)
289 wrbuf_printf(p->wr_error, "Attribute 'inputcharset' missing");
292 r = add_rule(p, YAZ_RECORD_CONV_RULE_MARC);
293 r->u.marc.iconv_t = cd;
295 r->u.marc.input_format = input_format_mode;
296 r->u.marc.output_format = output_format_mode;
300 int yaz_record_conv_configure(yaz_record_conv_t p, const void *ptr_v)
302 const xmlNode *ptr = ptr_v;
304 yaz_record_conv_reset(p);
306 if (ptr && ptr->type == XML_ELEMENT_NODE &&
307 !strcmp((const char *) ptr->name, "convert"))
309 for (ptr = ptr->children; ptr; ptr = ptr->next)
311 if (ptr->type != XML_ELEMENT_NODE)
313 if (!strcmp((const char *) ptr->name, "xslt"))
315 if (conv_xslt(p, ptr))
318 else if (!strcmp((const char *) ptr->name, "marc"))
320 if (conv_marc(p, ptr))
325 wrbuf_printf(p->wr_error, "Bad element '%s'."
326 "Expected marc, xslt, ..", ptr->name);
333 wrbuf_printf(p->wr_error, "Missing 'convert' element");
339 int yaz_record_conv_record(yaz_record_conv_t p, const char *input_record,
343 WRBUF record = output_record; /* pointer transfer */
344 struct yaz_record_conv_rule *r = p->rules;
345 wrbuf_rewind(p->wr_error);
347 wrbuf_puts(record, input_record);
348 for (; ret == 0 && r; r = r->next)
350 if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
352 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
356 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
361 xmlDocPtr res = xsltApplyStylesheet(r->u.xslt.xsp, doc, 0);
366 xmlDocDumpFormatMemory (res, &out_buf, &out_len, 1);
368 wrbuf_rewind(record);
369 wrbuf_write(record, (const char *) out_buf, out_len);
376 wrbuf_printf(p->wr_error, "xsltApplyStylesheet faailed");
382 else if (r->which == YAZ_RECORD_CONV_RULE_MARC)
384 yaz_marc_t mt = yaz_marc_create();
386 yaz_marc_xml(mt, r->u.marc.output_format);
388 if (r->u.marc.iconv_t)
389 yaz_marc_iconv(mt, r->u.marc.iconv_t);
390 if (r->u.marc.input_format == YAZ_MARC_ISO2709)
392 int sz = yaz_marc_read_iso2709(mt, wrbuf_buf(record),
399 else if (r->u.marc.input_format == YAZ_MARC_MARCXML)
401 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
405 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
410 ret = yaz_marc_read_xml(mt, xmlDocGetRootElement(doc));
412 wrbuf_printf(p->wr_error, "yaz_marc_read_xml failed");
418 wrbuf_printf(p->wr_error, "unsupported input format");
423 wrbuf_rewind(record);
424 ret = yaz_marc_write_mode(mt, record);
426 wrbuf_printf(p->wr_error, "yaz_marc_write_mode failed");
428 yaz_marc_destroy(mt);
434 const char *yaz_record_conv_get_error(yaz_record_conv_t p)
436 return wrbuf_buf(p->wr_error);
439 void yaz_record_conv_set_path(yaz_record_conv_t p, const char *path)
444 p->path = xstrdup(path);
451 * indent-tabs-mode: nil
453 * vim: shiftwidth=4 tabstop=8 expandtab