Start work on record caching - does not work!
[yaz-moved-to-github.git] / src / record_render.c
1 /* This file is part of the YAZ toolkit.
2  * Copyright (C) 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         if (oid_oidcmp(oid, yaz_oid_recsyn_xml)
171             && oid_oidcmp(oid, yaz_oid_recsyn_application_xml)
172             && oid_oidcmp(oid, yaz_oid_recsyn_mab)
173             && oid_oidcmp(oid, yaz_oid_recsyn_html))
174         {
175             const char *ret_buf = return_marc_record(
176                 wrbuf, marctype, len,
177                 (const char *) r->u.octet_aligned->buf,
178                 r->u.octet_aligned->len,
179                 charset);
180             if (ret_buf)
181                 return ret_buf;
182             /* not ISO2709. Return fail unless raw (ISO2709) is wanted */
183             if (yaz_oid_is_iso2709(oid) && marctype != YAZ_MARC_ISO2709)
184                 return 0;
185         }
186         return return_string_record(wrbuf, len,
187                                     (const char *) r->u.octet_aligned->buf,
188                                     r->u.octet_aligned->len,
189                                     charset);
190     }
191     else if (r->which == Z_External_grs1)
192     {
193         yaz_display_grs1(wrbuf, r->u.grs1, 0);
194         return return_string_record(wrbuf, len,
195                                     wrbuf_buf(wrbuf),
196                                     wrbuf_len(wrbuf),
197                                     charset);
198     }
199     return 0;
200 }
201
202 static const char *get_record_format(WRBUF wrbuf, int *len,
203                                      Z_NamePlusRecord *npr,
204                                      int marctype, const char *charset,
205                                      const char *format)
206 {
207     const char *res = return_record_wrbuf(wrbuf, len, npr, marctype, charset);
208 #if YAZ_HAVE_XML2
209     if (*format == '1')
210     {
211         /* try to XML format res */
212         xmlDocPtr doc;
213         xmlKeepBlanksDefault(0); /* get get xmlDocFormatMemory to work! */
214         doc = xmlParseMemory(res, *len);
215         if (doc)
216         {
217             xmlChar *xml_mem;
218             int xml_size;
219             xmlDocDumpFormatMemory(doc, &xml_mem, &xml_size, 1);
220             wrbuf_rewind(wrbuf);
221             wrbuf_write(wrbuf, (const char *) xml_mem, xml_size);
222             xmlFree(xml_mem);
223             xmlFreeDoc(doc);
224             res = wrbuf_cstr(wrbuf);
225             *len = wrbuf_len(wrbuf);
226         }
227     }
228 #endif
229     return res;
230 }
231
232 #if YAZ_HAVE_XML2
233 static int replace_node(NMEM nmem, xmlNode *ptr,
234                         const char *type_spec, char *record_buf)
235 {
236     int ret = -1;
237     const char *res;
238     int len;
239     int m_len;
240     WRBUF wrbuf = wrbuf_alloc();
241     ODR odr = odr_createmem(ODR_ENCODE);
242     Z_NamePlusRecord *npr = odr_malloc(odr, sizeof(*npr));
243     npr->which = Z_NamePlusRecord_databaseRecord;
244
245     if (atoi_n_check(record_buf, 5, &m_len))
246         npr->u.databaseRecord =
247             z_ext_record_usmarc(odr, record_buf, strlen(record_buf));
248     else
249         npr->u.databaseRecord =
250             z_ext_record_xml(odr, record_buf, strlen(record_buf));
251     res = yaz_record_render(npr, 0, wrbuf, type_spec, &len);
252     if (res)
253     {
254         xmlDoc *doc = xmlParseMemory(res, strlen(res));
255         if (doc)
256         {
257             xmlNode *nptr = xmlCopyNode(xmlDocGetRootElement(doc), 1);
258             xmlReplaceNode(ptr, nptr);
259             xmlFreeDoc(doc);
260         }
261         else
262         {
263             xmlNode *nptr = xmlNewText(BAD_CAST res);
264             xmlReplaceNode(ptr, nptr);
265         }
266         ret = 0;
267     }
268     wrbuf_destroy(wrbuf);
269     odr_destroy(odr);
270     return ret;
271 }
272 #endif
273
274 static const char *base64_render(NMEM nmem, WRBUF wrbuf,
275                                  const char *buf, int *len,
276                                  const char *expr, const char *type_spec)
277 {
278 #if YAZ_HAVE_XML2
279     xmlDocPtr doc = xmlParseMemory(buf, *len);
280     if (doc)
281     {
282         xmlChar *buf_out;
283         int len_out;
284         xmlXPathContextPtr xpathCtx = xmlXPathNewContext(doc);
285         if (xpathCtx)
286         {
287             xmlXPathObjectPtr xpathObj =
288                 xmlXPathEvalExpression((const xmlChar *) expr, xpathCtx);
289             if (xpathObj)
290             {
291                 xmlNodeSetPtr nodes = xpathObj->nodesetval;
292                 if (nodes)
293                 {
294                     int i;
295                     for (i = 0; i < nodes->nodeNr; i++)
296                     {
297                         xmlNode *ptr = nodes->nodeTab[i];
298                         if (ptr->type == XML_TEXT_NODE)
299                         {
300                             const char *input =
301                                 nmem_text_node_cdata(ptr, nmem);
302                             char *output = nmem_malloc(
303                                 nmem, strlen(input) + 1);
304                             if (yaz_base64decode(input, output) == 0)
305                             {
306                                 if (!replace_node(nmem, ptr, type_spec, output))
307                                 {
308                                     /* replacement OK */
309                                     xmlFreeNode(ptr);
310                                     /* unset below to avoid a bad reference in
311                                        xmlXPathFreeObject below */
312                                     nodes->nodeTab[i] = 0;
313                                 }
314                             }
315                         }
316                     }
317                 }
318                 xmlXPathFreeObject(xpathObj);
319             }
320             xmlXPathFreeContext(xpathCtx);
321         }
322         xmlDocDumpMemory(doc, &buf_out, &len_out);
323         if (buf_out)
324         {
325             wrbuf_rewind(wrbuf);
326             wrbuf_write(wrbuf, (const char *) buf_out, len_out);
327             buf = wrbuf_cstr(wrbuf);
328             *len = len_out;
329         }
330         xmlFreeDoc(doc);
331         xmlFree(buf_out);
332     }
333 #endif
334     return buf;
335 }
336
337 const char *yaz_record_render(Z_NamePlusRecord *npr, const char *schema,
338                               WRBUF wrbuf,
339                               const char *type_spec, int *len)
340 {
341     const char *ret = 0;
342     NMEM nmem = 0;
343     char *base64_xpath = 0;
344     size_t i;
345     char type[40];
346     char charset[40];
347     char format[3];
348     const char *cp = type_spec;
349     int len0;
350
351     if (!len)
352         len = &len0;
353
354     for (i = 0; cp[i] && cp[i] != ';' && cp[i] != ' ' && i < sizeof(type)-1;
355          i++)
356         type[i] = cp[i];
357     type[i] = '\0';
358     charset[0] = '\0';
359     format[0] = '\0';
360     while (1)
361     {
362         while (cp[i] == ' ')
363             i++;
364         if (cp[i] != ';')
365             break;
366         i++;
367         while (cp[i] == ' ')
368             i++;
369         if (!strncmp(cp + i, "charset=", 8))
370         {
371             size_t j = 0;
372             i = i + 8; /* skip charset= */
373             while (cp[i] == ' ')
374                 i++;
375             for (j = 0; cp[i] && cp[i] != ';' && cp[i] != ' '; i++)
376             {
377                 if (j < sizeof(charset)-1)
378                     charset[j++] = cp[i];
379             }
380             charset[j] = '\0';
381         }
382         else if (!strncmp(cp + i, "format=", 7))
383         {
384             size_t j = 0;
385             i = i + 7;
386             while (cp[i] == ' ')
387                 i++;
388             for (j = 0; cp[i] && cp[i] != ';' && cp[i] != ' '; i++)
389             {
390                 if (j < sizeof(format)-1)
391                     format[j++] = cp[i];
392             }
393             format[j] = '\0';
394         }
395         else if (!strncmp(cp + i, "base64=", 7))
396         {
397             size_t i0;
398             i = i + 7;
399             while (cp[i] == ' ')
400                 i++;
401             i0 = i;
402             while (cp[i] && cp[i] != ';')
403                 i++;
404
405             nmem = nmem_create();
406             base64_xpath = nmem_strdupn(nmem, cp + i0, i - i0);
407         }
408     }
409     if (!strcmp(type, "database"))
410     {
411         *len = (npr->databaseName ? strlen(npr->databaseName) : 0);
412         ret = npr->databaseName;
413     }
414     else if (!strcmp(type, "schema"))
415     {
416         *len = schema ? strlen(schema) : 0;
417         ret = schema;
418     }
419     else if (!strcmp(type, "syntax"))
420     {
421         const char *desc = 0;
422         if (npr->which == Z_NamePlusRecord_databaseRecord)
423         {
424             Z_External *r = (Z_External *) npr->u.databaseRecord;
425             desc = yaz_oid_to_string(yaz_oid_std(), r->direct_reference, 0);
426         }
427         if (!desc)
428             desc = "none";
429         *len = strlen(desc);
430         ret = desc;
431     }
432     if (npr->which != Z_NamePlusRecord_databaseRecord)
433         ;
434     else if (!strcmp(type, "render"))
435     {
436         ret = get_record_format(wrbuf, len, npr, YAZ_MARC_LINE, charset, format);
437     }
438     else if (!strcmp(type, "xml"))
439     {
440         ret = get_record_format(wrbuf, len, npr, YAZ_MARC_MARCXML, charset,
441                                 format);
442     }
443     else if (!strcmp(type, "txml"))
444     {
445         ret = get_record_format(wrbuf, len, npr, YAZ_MARC_TURBOMARC, charset,
446                                 format);
447     }
448     else if (!strcmp(type, "json"))
449     {
450         ret = get_record_format(wrbuf, len, npr, YAZ_MARC_JSON, charset,
451                                 format);
452     }
453     else if (!strcmp(type, "raw"))
454     {
455         ret = get_record_format(wrbuf, len, npr, YAZ_MARC_ISO2709, charset,
456                                 format);
457     }
458     else if (!strcmp(type, "ext"))
459     {
460         *len = -1;
461         ret = (const char *) npr->u.databaseRecord;
462     }
463     else if (!strcmp(type, "opac"))
464     {
465         if (npr->u.databaseRecord->which == Z_External_OPAC)
466             ret = get_record_format(wrbuf, len, npr, YAZ_MARC_MARCXML, charset,
467                                     format);
468     }
469
470     if (base64_xpath && *len != -1)
471     {
472         char *type_spec = nmem_malloc(nmem,
473                                       strlen(type) + strlen(charset) + 11);
474         strcpy(type_spec, type);
475         if (*charset)
476         {
477             strcat(type_spec, "; charset=");
478             strcat(type_spec, charset);
479         }
480         ret = base64_render(nmem, wrbuf, ret, len, base64_xpath, type_spec);
481     }
482     nmem_destroy(nmem);
483     return ret;
484 }
485
486 /*
487  * Local variables:
488  * c-basic-offset: 4
489  * c-file-style: "Stroustrup"
490  * indent-tabs-mode: nil
491  * End:
492  * vim: shiftwidth=4 tabstop=8 expandtab
493  */
494