2 * Copyright (C) 2005-2006, Index Data ApS
3 * See the file LICENSE for details.
5 * $Id: record_conv.c,v 1.3 2006-05-04 15:31:04 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 /** memory for configuration */
37 /** conversion rules (allocated using NMEM) */
38 struct yaz_record_conv_rule *rules;
40 /** pointer to last conversion rule pointer in chain */
41 struct yaz_record_conv_rule **rules_p;
43 /** string buffer for error messages */
46 /** 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 /** 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 static struct yaz_record_conv_rule *add_rule(yaz_record_conv_t p,
124 enum YAZ_RECORD_CONV_RULE type)
126 struct yaz_record_conv_rule *r = nmem_malloc(p->nmem, sizeof(*r));
130 p->rules_p = &r->next;
134 static int conv_xslt(yaz_record_conv_t p, const xmlNode *ptr)
136 struct _xmlAttr *attr;
137 const char *stylesheet = 0;
139 for (attr = ptr->properties; attr; attr = attr->next)
141 if (!xmlStrcmp(attr->name, BAD_CAST "stylesheet") &&
142 attr->children && attr->children->type == XML_TEXT_NODE)
143 stylesheet = (const char *) attr->children->content;
146 wrbuf_printf(p->wr_error, "Bad attribute '%s'."
147 "Expected stylesheet.", attr->name);
153 wrbuf_printf(p->wr_error, "Missing attribute 'stylesheet'");
159 xsltStylesheetPtr xsp;
160 if (!yaz_filepath_resolve(stylesheet, p->path, 0, fullpath))
162 wrbuf_printf(p->wr_error, "could not locate '%s'. Path=%s",
163 stylesheet, p->path);
166 xsp = xsltParseStylesheetFile((xmlChar*) fullpath);
169 wrbuf_printf(p->wr_error, "xsltParseStylesheetFile failed'");
174 struct yaz_record_conv_rule *r =
175 add_rule(p, YAZ_RECORD_CONV_RULE_XSLT);
182 static int conv_marc(yaz_record_conv_t p, const xmlNode *ptr)
184 struct _xmlAttr *attr;
185 const char *input_charset = 0;
186 const char *output_charset = 0;
187 const char *input_format = 0;
188 const char *output_format = 0;
189 int input_format_mode = 0;
190 int output_format_mode = 0;
191 struct yaz_record_conv_rule *r;
194 for (attr = ptr->properties; attr; attr = attr->next)
196 if (!xmlStrcmp(attr->name, BAD_CAST "inputcharset") &&
197 attr->children && attr->children->type == XML_TEXT_NODE)
198 input_charset = (const char *) attr->children->content;
199 else if (!xmlStrcmp(attr->name, BAD_CAST "outputcharset") &&
200 attr->children && attr->children->type == XML_TEXT_NODE)
201 output_charset = (const char *) attr->children->content;
202 else if (!xmlStrcmp(attr->name, BAD_CAST "inputformat") &&
203 attr->children && attr->children->type == XML_TEXT_NODE)
204 input_format = (const char *) attr->children->content;
205 else if (!xmlStrcmp(attr->name, BAD_CAST "outputformat") &&
206 attr->children && attr->children->type == XML_TEXT_NODE)
207 output_format = (const char *) attr->children->content;
210 wrbuf_printf(p->wr_error, "Bad attribute '%s'.", attr->name);
216 wrbuf_printf(p->wr_error, "Attribute 'inputformat' required");
219 else if (!strcmp(input_format, "marc"))
221 input_format_mode = YAZ_MARC_ISO2709;
223 else if (!strcmp(input_format, "xml"))
225 input_format_mode = YAZ_MARC_MARCXML;
226 /** Libxml2 generates UTF-8 encoding by default .
227 So we convert from UTF-8 to outputcharset (if defined)
229 if (!input_charset && output_charset)
230 input_charset = "utf-8";
234 wrbuf_printf(p->wr_error, "Bad inputformat: '%s'", input_format);
240 wrbuf_printf(p->wr_error, "Attribute 'outputformat' required");
243 else if (!strcmp(output_format, "line"))
245 output_format_mode = YAZ_MARC_LINE;
247 else if (!strcmp(output_format, "marcxml"))
249 output_format_mode = YAZ_MARC_MARCXML;
250 if (input_charset && !output_charset)
251 output_charset = "utf-8";
253 else if (!strcmp(output_format, "marc"))
255 output_format_mode = YAZ_MARC_ISO2709;
257 else if (!strcmp(output_format, "marcxchange"))
259 output_format_mode = YAZ_MARC_XCHANGE;
260 if (input_charset && !output_charset)
261 output_charset = "utf-8";
265 wrbuf_printf(p->wr_error, "Bad outputformat: '%s'", input_format);
268 if (input_charset && output_charset)
270 cd = yaz_iconv_open(output_charset, input_charset);
273 wrbuf_printf(p->wr_error, "Unsupported character set mamping"
274 " inputcharset=%s outputcharset=%s",
275 input_charset, output_charset);
279 else if (input_charset)
281 wrbuf_printf(p->wr_error, "Attribute 'outputcharset' missing");
284 else if (output_charset)
286 wrbuf_printf(p->wr_error, "Attribute 'inputcharset' missing");
289 r = add_rule(p, YAZ_RECORD_CONV_RULE_MARC);
290 r->u.marc.iconv_t = cd;
292 r->u.marc.input_format = input_format_mode;
293 r->u.marc.output_format = output_format_mode;
297 int yaz_record_conv_configure(yaz_record_conv_t p, const void *ptr_v)
299 const xmlNode *ptr = ptr_v;
301 yaz_record_conv_reset(p);
303 if (ptr && ptr->type == XML_ELEMENT_NODE &&
304 !strcmp((const char *) ptr->name, "convert"))
306 for (ptr = ptr->children; ptr; ptr = ptr->next)
308 if (ptr->type != XML_ELEMENT_NODE)
310 if (!strcmp((const char *) ptr->name, "xslt"))
312 if (conv_xslt(p, ptr))
315 else if (!strcmp((const char *) ptr->name, "marc"))
317 if (conv_marc(p, ptr))
322 wrbuf_printf(p->wr_error, "Bad element '%s'."
323 "Expected marc, xslt, ..", ptr->name);
330 wrbuf_printf(p->wr_error, "Missing 'convert' element");
336 int yaz_record_conv_record(yaz_record_conv_t p, const char *input_record,
340 WRBUF record = output_record; /* pointer transfer */
341 struct yaz_record_conv_rule *r = p->rules;
342 wrbuf_rewind(p->wr_error);
344 wrbuf_puts(record, input_record);
345 for (; ret == 0 && r; r = r->next)
347 if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
349 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
353 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
358 xmlDocPtr res = xsltApplyStylesheet(r->u.xslt.xsp, doc, 0);
363 xmlDocDumpFormatMemory (res, &out_buf, &out_len, 1);
365 wrbuf_rewind(record);
366 wrbuf_write(record, (const char *) out_buf, out_len);
373 wrbuf_printf(p->wr_error, "xsltApplyStylesheet faailed");
379 else if (r->which == YAZ_RECORD_CONV_RULE_MARC)
381 yaz_marc_t mt = yaz_marc_create();
383 yaz_marc_xml(mt, r->u.marc.output_format);
385 if (r->u.marc.iconv_t)
386 yaz_marc_iconv(mt, r->u.marc.iconv_t);
387 if (r->u.marc.input_format == YAZ_MARC_ISO2709)
389 int sz = yaz_marc_read_iso2709(mt, wrbuf_buf(record),
396 else if (r->u.marc.input_format == YAZ_MARC_MARCXML)
398 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
402 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
407 ret = yaz_marc_read_xml(mt, xmlDocGetRootElement(doc));
409 wrbuf_printf(p->wr_error, "yaz_marc_read_xml failed");
415 wrbuf_printf(p->wr_error, "unsupported input format");
420 wrbuf_rewind(record);
421 ret = yaz_marc_write_mode(mt, record);
423 wrbuf_printf(p->wr_error, "yaz_marc_write_mode failed");
425 yaz_marc_destroy(mt);
431 const char *yaz_record_conv_get_error(yaz_record_conv_t p)
433 return wrbuf_buf(p->wr_error);
436 void yaz_record_conv_set_path(yaz_record_conv_t p, const char *path)
443 p->path = xstrdup(path);
451 * indent-tabs-mode: nil
453 * vim: shiftwidth=4 tabstop=8 expandtab