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;
72 const char *input_charset;
73 const char *output_charset;
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)
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;
230 for (attr = ptr->properties; attr; attr = attr->next)
232 if (!xmlStrcmp(attr->name, BAD_CAST "inputcharset") &&
233 attr->children && attr->children->type == XML_TEXT_NODE)
234 input_charset = (const char *) attr->children->content;
235 else if (!xmlStrcmp(attr->name, BAD_CAST "outputcharset") &&
236 attr->children && attr->children->type == XML_TEXT_NODE)
237 output_charset = (const char *) attr->children->content;
238 else if (!xmlStrcmp(attr->name, BAD_CAST "inputformat") &&
239 attr->children && attr->children->type == XML_TEXT_NODE)
240 input_format = (const char *) attr->children->content;
241 else if (!xmlStrcmp(attr->name, BAD_CAST "outputformat") &&
242 attr->children && attr->children->type == XML_TEXT_NODE)
243 output_format = (const char *) attr->children->content;
246 wrbuf_printf(p->wr_error, "Element <marc>: expected attributes"
247 "'inputformat', 'inputcharset', 'outputformat' or"
248 " 'outputcharset', got attribute '%s'",
255 wrbuf_printf(p->wr_error, "Element <marc>: "
256 "attribute 'inputformat' required");
259 else if (!strcmp(input_format, "marc"))
261 input_format_mode = YAZ_MARC_ISO2709;
263 else if (!strcmp(input_format, "xml"))
265 input_format_mode = YAZ_MARC_MARCXML;
266 /** Libxml2 generates UTF-8 encoding by default .
267 So we convert from UTF-8 to outputcharset (if defined)
269 if (!input_charset && output_charset)
270 input_charset = "utf-8";
274 wrbuf_printf(p->wr_error, "Element <marc inputformat='%s'>: "
275 " Unsupported input format"
276 " defined by attribute value",
283 wrbuf_printf(p->wr_error,
284 "Element <marc>: attribute 'outputformat' required");
287 else if (!strcmp(output_format, "line"))
289 output_format_mode = YAZ_MARC_LINE;
291 else if (!strcmp(output_format, "marcxml"))
293 output_format_mode = YAZ_MARC_MARCXML;
294 if (input_charset && !output_charset)
295 output_charset = "utf-8";
297 else if (!strcmp(output_format, "marc"))
299 output_format_mode = YAZ_MARC_ISO2709;
301 else if (!strcmp(output_format, "marcxchange"))
303 output_format_mode = YAZ_MARC_XCHANGE;
304 if (input_charset && !output_charset)
305 output_charset = "utf-8";
309 wrbuf_printf(p->wr_error, "Element <marc outputformat='%s'>: "
310 " Unsupported output format"
311 " defined by attribute value",
315 if (input_charset && output_charset)
317 yaz_iconv_t cd = yaz_iconv_open(output_charset, input_charset);
320 wrbuf_printf(p->wr_error,
321 "Element <marc inputcharset='%s' outputcharset='%s'>:"
322 " Unsupported character set mapping"
323 " defined by attribute values",
324 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);
343 r->u.marc.input_charset = nmem_strdup(p->nmem, input_charset);
344 r->u.marc.output_charset = nmem_strdup(p->nmem, output_charset);
345 r->u.marc.input_format = input_format_mode;
346 r->u.marc.output_format = output_format_mode;
350 int yaz_record_conv_configure(yaz_record_conv_t p, const xmlNode *ptr)
352 yaz_record_conv_reset(p);
354 /* parsing element children */
355 for (ptr = ptr->children; ptr; ptr = ptr->next)
357 if (ptr->type != XML_ELEMENT_NODE)
359 if (!strcmp((const char *) ptr->name, "xslt"))
361 if (conv_xslt(p, ptr))
364 else if (!strcmp((const char *) ptr->name, "marc"))
366 if (conv_marc(p, ptr))
371 wrbuf_printf(p->wr_error, "Element <backend>: expected "
372 "<marc> or <xslt> element, got <%s>"
380 static int yaz_record_conv_record_rule(yaz_record_conv_t p,
381 struct yaz_record_conv_rule *r,
382 const char *input_record_buf,
383 size_t input_record_len,
384 WRBUF output_record);
386 int yaz_record_conv_opac_record(yaz_record_conv_t p,
387 Z_OPACRecord *input_record,
391 struct yaz_record_conv_rule *r = p->rules;
392 if (!r || r->which != YAZ_RECORD_CONV_RULE_MARC)
393 ret = -1; /* no marc rule so we can't do OPAC */
396 WRBUF res = wrbuf_alloc();
397 yaz_marc_t mt = yaz_marc_create();
398 yaz_iconv_t cd = yaz_iconv_open(r->u.marc.output_charset,
399 r->u.marc.input_charset);
401 wrbuf_rewind(p->wr_error);
402 yaz_marc_xml(mt, r->u.marc.output_format);
404 yaz_marc_iconv(mt, cd);
406 yaz_opac_decode_wrbuf(mt, input_record, res);
409 ret = yaz_record_conv_record_rule(p,
411 wrbuf_buf(res), wrbuf_len(res),
414 yaz_marc_destroy(mt);
422 int yaz_record_conv_record(yaz_record_conv_t p,
423 const char *input_record_buf,
424 size_t input_record_len,
427 return yaz_record_conv_record_rule(p, p->rules,
429 input_record_len, output_record);
432 static int yaz_record_conv_record_rule(yaz_record_conv_t p,
433 struct yaz_record_conv_rule *r,
434 const char *input_record_buf,
435 size_t input_record_len,
439 WRBUF record = output_record; /* pointer transfer */
440 wrbuf_rewind(p->wr_error);
442 wrbuf_write(record, input_record_buf, input_record_len);
443 for (; ret == 0 && r; r = r->next)
445 if (r->which == YAZ_RECORD_CONV_RULE_MARC)
448 yaz_iconv_open(r->u.marc.output_charset,
449 r->u.marc.input_charset);
450 yaz_marc_t mt = yaz_marc_create();
452 yaz_marc_xml(mt, r->u.marc.output_format);
455 yaz_marc_iconv(mt, cd);
456 if (r->u.marc.input_format == YAZ_MARC_ISO2709)
458 int sz = yaz_marc_read_iso2709(mt, wrbuf_buf(record),
465 else if (r->u.marc.input_format == YAZ_MARC_MARCXML)
467 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
471 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
476 ret = yaz_marc_read_xml(mt, xmlDocGetRootElement(doc));
478 wrbuf_printf(p->wr_error, "yaz_marc_read_xml failed");
484 wrbuf_printf(p->wr_error, "unsupported input format");
489 wrbuf_rewind(record);
490 ret = yaz_marc_write_mode(mt, record);
492 wrbuf_printf(p->wr_error, "yaz_marc_write_mode failed");
496 yaz_marc_destroy(mt);
499 else if (r->which == YAZ_RECORD_CONV_RULE_XSLT)
501 xmlDocPtr doc = xmlParseMemory(wrbuf_buf(record),
505 wrbuf_printf(p->wr_error, "xmlParseMemory failed");
510 xmlDocPtr res = xsltApplyStylesheet(r->u.xslt.xsp, doc, 0);
513 xmlChar *out_buf = 0;
516 #if YAZ_HAVE_XSLTSAVERESULTTOSTRING
517 xsltSaveResultToString(&out_buf, &out_len, res,
520 xmlDocDumpFormatMemory (res, &out_buf, &out_len, 1);
524 wrbuf_printf(p->wr_error,
525 "xsltSaveResultToString failed");
530 wrbuf_rewind(record);
531 wrbuf_write(record, (const char *) out_buf, out_len);
539 wrbuf_printf(p->wr_error, "xsltApplyStylesheet failed");
550 const char *yaz_record_conv_get_error(yaz_record_conv_t p)
552 return wrbuf_cstr(p->wr_error);
555 void yaz_record_conv_set_path(yaz_record_conv_t p, const char *path)
560 p->path = xstrdup(path);
567 * indent-tabs-mode: nil
569 * vim: shiftwidth=4 tabstop=8 expandtab