6acb1d72c25d5d2c992dcfd5f2878f978e28cde7
[yaz-moved-to-github.git] / src / record_render.c
1 /* This file is part of the YAZ toolkit.
2  * Copyright (C) 1995-2013 Index Data
3  * See the file LICENSE for details.
4  */
5 /**
6  * \file record_render.c
7  * \brief Render Z39.50 records (NamePlusRecord)
8  */
9 #if HAVE_CONFIG_H
10 #include <config.h>
11 #endif
12
13 #include <assert.h>
14 #include <string.h>
15 #include <errno.h>
16
17 #include <yaz/marcdisp.h>
18 #include <yaz/record_render.h>
19 #include <yaz/yaz-iconv.h>
20 #include <yaz/proto.h>
21 #include <yaz/oid_db.h>
22 #include <yaz/nmem_xml.h>
23 #include <yaz/base64.h>
24
25 #if YAZ_HAVE_XML2
26 #include <libxml/xpath.h>
27 #include <libxml/xpathInternals.h>
28 #endif
29
30 static yaz_iconv_t iconv_create_charset(const char *record_charset,
31                                         yaz_iconv_t *cd2)
32 {
33     char charset_buf[40];
34     yaz_iconv_t cd = 0;
35     char *from_set1 = 0;
36     char *from_set2 = 0;
37     char *to_set = "utf-8";
38     if (record_charset && *record_charset)
39     {
40         char *cp = charset_buf;
41
42         strncpy(charset_buf, record_charset, sizeof(charset_buf)-1);
43         charset_buf[sizeof(charset_buf)-1] = '\0';
44
45         from_set1 = cp;
46         while (*cp && *cp != ',' && *cp != '/')
47             cp++;
48         if (*cp == '/')
49         {
50             *cp++ = '\0'; /* terminate from_set1 */
51             from_set2 = cp;
52             while (*cp && *cp != ',')
53                 cp++;
54         }
55         if (*cp == ',')
56         {
57             *cp++ = '\0';  /* terminate from_set1 or from_set2 */
58             to_set = cp;
59             while (*cp)
60                 cp++;
61         }
62     }
63
64     if (from_set1)
65         cd = yaz_iconv_open(to_set, from_set1);
66     if (cd2)
67     {
68         if (from_set2)
69             *cd2 = yaz_iconv_open(to_set, from_set2);
70         else
71             *cd2 = 0;
72     }
73     return cd;
74 }
75
76 static const char *return_marc_record(WRBUF wrbuf,
77                                       int marc_type,
78                                       int *len,
79                                       const char *buf, int sz,
80                                       const char *record_charset)
81 {
82     yaz_iconv_t cd = iconv_create_charset(record_charset, 0);
83     yaz_marc_t mt = yaz_marc_create();
84     const char *ret_string = 0;
85
86     if (cd)
87         yaz_marc_iconv(mt, cd);
88     yaz_marc_xml(mt, marc_type);
89     if (yaz_marc_decode_wrbuf(mt, buf, sz, wrbuf) > 0)
90     {
91         *len = wrbuf_len(wrbuf);
92         ret_string = wrbuf_cstr(wrbuf);
93     }
94     yaz_marc_destroy(mt);
95     if (cd)
96         yaz_iconv_close(cd);
97     return ret_string;
98 }
99
100 static const char *return_opac_record(WRBUF wrbuf,
101                                       int marc_type,
102                                       int *len,
103                                       Z_OPACRecord *opac_rec,
104                                       const char *record_charset)
105 {
106     yaz_iconv_t cd2;
107     yaz_iconv_t cd = iconv_create_charset(record_charset, &cd2);
108     yaz_marc_t mt = yaz_marc_create();
109
110     if (cd)
111         yaz_marc_iconv(mt, cd);
112     yaz_marc_xml(mt, marc_type);
113
114     if (cd2)
115         yaz_opac_decode_wrbuf2(mt, opac_rec, wrbuf, cd2);
116     else
117         yaz_opac_decode_wrbuf(mt, opac_rec, wrbuf);
118
119     yaz_marc_destroy(mt);
120
121     if (cd)
122         yaz_iconv_close(cd);
123     if (cd2)
124         yaz_iconv_close(cd2);
125     *len = wrbuf_len(wrbuf);
126     return wrbuf_cstr(wrbuf);
127 }
128
129 static const char *return_string_record(WRBUF wrbuf,
130                                         int *len,
131                                         const char *buf, int sz,
132                                         const char *record_charset)
133 {
134     yaz_iconv_t cd = iconv_create_charset(record_charset, 0);
135
136     if (cd)
137     {
138         wrbuf_iconv_write(wrbuf, cd, buf, sz);
139         wrbuf_iconv_reset(wrbuf, cd);
140
141         buf = wrbuf_cstr(wrbuf);
142         sz = wrbuf_len(wrbuf);
143         yaz_iconv_close(cd);
144     }
145     *len = sz;
146     return buf;
147 }
148
149 static const char *return_record_wrbuf(WRBUF wrbuf, int *len,
150                                        Z_NamePlusRecord *npr,
151                                        int marctype, const char *charset)
152 {
153     Z_External *r = (Z_External *) npr->u.databaseRecord;
154     const Odr_oid *oid = r->direct_reference;
155
156     wrbuf_rewind(wrbuf);
157     /* render bibliographic record .. */
158     if (r->which == Z_External_OPAC)
159     {
160         return return_opac_record(wrbuf, marctype, len,
161                                   r->u.opac, charset);
162     }
163     if (r->which == Z_External_sutrs)
164         return return_string_record(wrbuf, len,
165                                     (char*) r->u.sutrs->buf,
166                                     r->u.sutrs->len,
167                                     charset);
168     else if (r->which == Z_External_octet)
169     {
170         const char *ret_buf = return_marc_record(
171             wrbuf, marctype, len,
172             (const char *) r->u.octet_aligned->buf,
173             r->u.octet_aligned->len,
174             charset);
175         if (ret_buf)
176             return ret_buf;
177         /* not ISO2709. Return fail unless raw (ISO2709) is wanted */
178         if (yaz_oid_is_iso2709(oid) && marctype != YAZ_MARC_ISO2709)
179             return 0;
180         return return_string_record(wrbuf, len,
181                                     (const char *) r->u.octet_aligned->buf,
182                                     r->u.octet_aligned->len,
183                                     charset);
184     }
185     else if (r->which == Z_External_grs1)
186     {
187         yaz_display_grs1(wrbuf, r->u.grs1, 0);
188         return return_string_record(wrbuf, len,
189                                     wrbuf_buf(wrbuf),
190                                     wrbuf_len(wrbuf),
191                                     charset);
192     }
193     return 0;
194 }
195
196 static const char *get_record_format(WRBUF wrbuf, int *len,
197                                      Z_NamePlusRecord *npr,
198                                      int marctype, const char *charset,
199                                      const char *format)
200 {
201     const char *res = return_record_wrbuf(wrbuf, len, npr, marctype, charset);
202 #if YAZ_HAVE_XML2
203     if (*format == '1')
204     {
205         /* try to XML format res */
206         xmlDocPtr doc;
207         xmlKeepBlanksDefault(0); /* get get xmlDocFormatMemory to work! */
208         doc = xmlParseMemory(res, *len);
209         if (doc)
210         {
211             xmlChar *xml_mem;
212             int xml_size;
213             xmlDocDumpFormatMemory(doc, &xml_mem, &xml_size, 1);
214             wrbuf_rewind(wrbuf);
215             wrbuf_write(wrbuf, (const char *) xml_mem, xml_size);
216             xmlFree(xml_mem);
217             xmlFreeDoc(doc);
218             res = wrbuf_cstr(wrbuf);
219             *len = wrbuf_len(wrbuf);
220         }
221     }
222 #endif
223     return res;
224 }
225
226 #if YAZ_HAVE_XML2
227 static int replace_node(NMEM nmem, xmlNode *ptr,
228                         const char *type_spec, char *record_buf)
229 {
230     int ret = -1;
231     const char *res;
232     int len;
233     int m_len;
234     WRBUF wrbuf = wrbuf_alloc();
235     ODR odr = odr_createmem(ODR_ENCODE);
236     Z_NamePlusRecord *npr = odr_malloc(odr, sizeof(*npr));
237     npr->which = Z_NamePlusRecord_databaseRecord;
238
239     if (atoi_n_check(record_buf, 5, &m_len))
240         npr->u.databaseRecord =
241             z_ext_record_usmarc(odr, record_buf, strlen(record_buf));
242     else
243         npr->u.databaseRecord =
244             z_ext_record_xml(odr, record_buf, strlen(record_buf));
245     res = yaz_record_render(npr, 0, wrbuf, type_spec, &len);
246     if (res)
247     {
248         xmlDoc *doc = xmlParseMemory(res, strlen(res));
249         if (doc)
250         {
251             xmlNode *nptr = xmlCopyNode(xmlDocGetRootElement(doc), 1);
252             xmlReplaceNode(ptr, nptr);
253             xmlFreeDoc(doc);
254         }
255         else
256         {
257             xmlNode *nptr = xmlNewText(BAD_CAST res);
258             xmlReplaceNode(ptr, nptr);
259         }
260         ret = 0;
261     }
262     wrbuf_destroy(wrbuf);
263     odr_destroy(odr);
264     return ret;
265 }
266 #endif
267
268 static const char *base64_render(NMEM nmem, WRBUF wrbuf,
269                                  const char *buf, int *len,
270                                  const char *expr, const char *type_spec)
271 {
272 #if YAZ_HAVE_XML2
273     xmlDocPtr doc = xmlParseMemory(buf, *len);
274     if (doc)
275     {
276         xmlChar *buf_out;
277         int len_out;
278         xmlXPathContextPtr xpathCtx = xmlXPathNewContext(doc);
279         if (xpathCtx)
280         {
281             xmlXPathObjectPtr xpathObj =
282                 xmlXPathEvalExpression((const xmlChar *) expr, xpathCtx);
283             if (xpathObj)
284             {
285                 xmlNodeSetPtr nodes = xpathObj->nodesetval;
286                 if (nodes)
287                 {
288                     int i;
289                     for (i = 0; i < nodes->nodeNr; i++)
290                     {
291                         xmlNode *ptr = nodes->nodeTab[i];
292                         if (ptr->type == XML_TEXT_NODE)
293                         {
294                             const char *input =
295                                 nmem_text_node_cdata(ptr, nmem);
296                             char *output = nmem_malloc(
297                                 nmem, strlen(input) + 1);
298                             if (yaz_base64decode(input, output) == 0)
299                             {
300                                 if (!replace_node(nmem, ptr, type_spec, output))
301                                 {
302                                     /* replacement OK */
303                                     xmlFreeNode(ptr);
304                                     /* unset below to avoid a bad reference in
305                                        xmlXPathFreeObject below */
306                                     nodes->nodeTab[i] = 0;
307                                 }
308                             }
309                         }
310                     }
311                 }
312                 xmlXPathFreeObject(xpathObj);
313             }
314             xmlXPathFreeContext(xpathCtx);
315         }
316         xmlDocDumpMemory(doc, &buf_out, &len_out);
317         if (buf_out)
318         {
319             wrbuf_rewind(wrbuf);
320             wrbuf_write(wrbuf, (const char *) buf_out, len_out);
321             buf = wrbuf_cstr(wrbuf);
322             *len = len_out;
323         }
324         xmlFreeDoc(doc);
325         xmlFree(buf_out);
326     }
327 #endif
328     return buf;
329 }
330
331 const char *yaz_record_render(Z_NamePlusRecord *npr, const char *schema,
332                               WRBUF wrbuf,
333                               const char *type_spec, int *len)
334 {
335     const char *ret = 0;
336     NMEM nmem = 0;
337     char *base64_xpath = 0;
338     size_t i;
339     char type[40];
340     char charset[40];
341     char format[3];
342     const char *cp = type_spec;
343     int len0;
344
345     if (!len)
346         len = &len0;
347
348     for (i = 0; cp[i] && cp[i] != ';' && cp[i] != ' ' && i < sizeof(type)-1;
349          i++)
350         type[i] = cp[i];
351     type[i] = '\0';
352     charset[0] = '\0';
353     format[0] = '\0';
354     while (1)
355     {
356         while (cp[i] == ' ')
357             i++;
358         if (cp[i] != ';')
359             break;
360         i++;
361         while (cp[i] == ' ')
362             i++;
363         if (!strncmp(cp + i, "charset=", 8))
364         {
365             size_t j = 0;
366             i = i + 8; /* skip charset= */
367             while (cp[i] == ' ')
368                 i++;
369             for (j = 0; cp[i] && cp[i] != ';' && cp[i] != ' '; i++)
370             {
371                 if (j < sizeof(charset)-1)
372                     charset[j++] = cp[i];
373             }
374             charset[j] = '\0';
375         }
376         else if (!strncmp(cp + i, "format=", 7))
377         {
378             size_t j = 0;
379             i = i + 7;
380             while (cp[i] == ' ')
381                 i++;
382             for (j = 0; cp[i] && cp[i] != ';' && cp[i] != ' '; i++)
383             {
384                 if (j < sizeof(format)-1)
385                     format[j++] = cp[i];
386             }
387             format[j] = '\0';
388         }
389         else if (!strncmp(cp + i, "base64=", 7))
390         {
391             size_t i0;
392             i = i + 7;
393             while (cp[i] == ' ')
394                 i++;
395             i0 = i;
396             while (cp[i] && cp[i] != ';')
397                 i++;
398
399             nmem = nmem_create();
400             base64_xpath = nmem_strdupn(nmem, cp + i0, i - i0);
401         }
402     }
403     if (!strcmp(type, "database"))
404     {
405         *len = (npr->databaseName ? strlen(npr->databaseName) : 0);
406         ret = npr->databaseName;
407     }
408     else if (!strcmp(type, "schema"))
409     {
410         *len = schema ? strlen(schema) : 0;
411         ret = schema;
412     }
413     else if (!strcmp(type, "syntax"))
414     {
415         const char *desc = 0;
416         if (npr->which == Z_NamePlusRecord_databaseRecord)
417         {
418             Z_External *r = (Z_External *) npr->u.databaseRecord;
419             desc = yaz_oid_to_string(yaz_oid_std(), r->direct_reference, 0);
420         }
421         if (!desc)
422             desc = "none";
423         *len = strlen(desc);
424         ret = desc;
425     }
426     if (npr->which != Z_NamePlusRecord_databaseRecord)
427         ;
428     else if (!strcmp(type, "render"))
429     {
430         ret = get_record_format(wrbuf, len, npr, YAZ_MARC_LINE, charset, format);
431     }
432     else if (!strcmp(type, "xml"))
433     {
434         ret = get_record_format(wrbuf, len, npr, YAZ_MARC_MARCXML, charset,
435                                 format);
436     }
437     else if (!strcmp(type, "txml"))
438     {
439         ret = get_record_format(wrbuf, len, npr, YAZ_MARC_TURBOMARC, charset,
440                                 format);
441     }
442     else if (!strcmp(type, "raw"))
443     {
444         ret = get_record_format(wrbuf, len, npr, YAZ_MARC_ISO2709, charset,
445                                 format);
446     }
447     else if (!strcmp(type, "ext"))
448     {
449         *len = -1;
450         ret = (const char *) npr->u.databaseRecord;
451     }
452     else if (!strcmp(type, "opac"))
453     {
454         if (npr->u.databaseRecord->which == Z_External_OPAC)
455             ret = get_record_format(wrbuf, len, npr, YAZ_MARC_MARCXML, charset,
456                                     format);
457     }
458
459     if (base64_xpath && *len != -1)
460     {
461         char *type_spec = nmem_malloc(nmem,
462                                       strlen(type) + strlen(charset) + 11);
463         strcpy(type_spec, type);
464         if (*charset)
465         {
466             strcat(type_spec, "; charset=");
467             strcat(type_spec, charset);
468         }
469         ret = base64_render(nmem, wrbuf, ret, len, base64_xpath, type_spec);
470     }
471     nmem_destroy(nmem);
472     return ret;
473 }
474
475 /*
476  * Local variables:
477  * c-basic-offset: 4
478  * c-file-style: "Stroustrup"
479  * indent-tabs-mode: nil
480  * End:
481  * vim: shiftwidth=4 tabstop=8 expandtab
482  */
483