-/* $Id: retrieve.c,v 1.66 2007-03-06 12:40:18 adam Exp $
+/* $Id: retrieve.c,v 1.72 2007-08-21 13:27:04 adam Exp $
Copyright (C) 1995-2007
Index Data ApS
#include "index.h"
#include <yaz/diagbib1.h>
+#include <yaz/snprintf.h>
#include <direntz.h>
-
+#include <yaz/oid_db.h>
#define ZEBRA_XML_HEADER_STR "<record xmlns=\"http://www.indexdata.com/zebra/\""
static int zebra_create_record_stream(ZebraHandle zh,
- Record *rec,
- struct ZebraRecStream *stream)
+ Record *rec,
+ struct ZebraRecStream *stream)
{
RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, *rec);
int zebra_special_sort_fetch(ZebraHandle zh, zint sysno, ODR odr,
const char *elemsetname,
- oid_value input_format,
- oid_value *output_format,
+ const Odr_oid *input_format,
+ const Odr_oid **output_format,
char **rec_bufp, int *rec_lenp)
{
const char *retrieval_index;
int ord;
/* only accept XML and SUTRS requests */
- if (input_format != VAL_TEXT_XML && input_format != VAL_SUTRS)
+ if (oid_oidcmp(input_format, yaz_oid_recsyn_xml)
+ && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
{
yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
elemsetname);
- *output_format = VAL_NONE;
+ *output_format = 0;
return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
}
zebra_term_untrans(zh, index_type, dst_buf, str);
- if (input_format == VAL_TEXT_XML)
+ if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
{
- *output_format = VAL_TEXT_XML;
+ *output_format = yaz_oid_recsyn_xml;
wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
" sysno=\"" ZINT_FORMAT "\""
" set=\"zebra::index%s/\">\n",
wrbuf_printf(wrbuf, "</index>\n");
wrbuf_printf(wrbuf, "</record>\n");
}
- else if (input_format == VAL_SUTRS)
+ else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
{
- *output_format = VAL_SUTRS;
+ *output_format = yaz_oid_recsyn_sutrs;
wrbuf_printf(wrbuf, "%s %c %s\n", string_index, index_type,
dst_buf);
*rec_lenp = wrbuf_len(wrbuf);
*rec_bufp = odr_malloc(odr, *rec_lenp);
memcpy(*rec_bufp, wrbuf_buf(wrbuf), *rec_lenp);
- wrbuf_free(wrbuf, 1);
+ wrbuf_destroy(wrbuf);
return 0;
}
}
int zebra_special_index_fetch(ZebraHandle zh, zint sysno, ODR odr,
Record rec,
const char *elemsetname,
- oid_value input_format,
- oid_value *output_format,
+ const Odr_oid *input_format,
+ const Odr_oid **output_format,
char **rec_bufp, int *rec_lenp)
{
const char *retrieval_index;
/* *rec_lenp = 0; */
/* only accept XML and SUTRS requests */
- if (input_format != VAL_TEXT_XML && input_format != VAL_SUTRS)
+ if (oid_oidcmp(input_format, yaz_oid_recsyn_xml)
+ && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
{
yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
elemsetname);
- *output_format = VAL_NONE;
+ *output_format = 0;
return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
}
if (!zebra_rec_keys_rewind(keys))
{
- ret_code =
- YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
+ ret_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
}
else
{
struct it_key key_in;
WRBUF wrbuf = wrbuf_alloc();
- if (input_format == VAL_TEXT_XML)
+ if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
{
- *output_format = VAL_TEXT_XML;
+ *output_format = input_format;
wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
" sysno=\"" ZINT_FORMAT "\""
" set=\"zebra::index%s/\">\n",
sysno, elemsetname);
}
- else if (input_format == VAL_SUTRS)
- *output_format = VAL_SUTRS;
+ else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
+ *output_format = input_format;
while (zebra_rec_keys_read(keys, &str, &slen, &key_in))
{
zebra_term_untrans(zh, index_type, dst_buf, str);
if (strlen(dst_buf))
{
- if (input_format == VAL_TEXT_XML){
+ if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
+ {
wrbuf_printf(wrbuf, " <index name=\"%s\"",
string_index);
wrbuf_xmlputs(wrbuf, dst_buf);
wrbuf_printf(wrbuf, "</index>\n");
}
- else if (input_format == VAL_SUTRS){
+ else
+ {
wrbuf_printf(wrbuf, "%s ", string_index);
wrbuf_printf(wrbuf, "%c", index_type);
for (i = 1; i < key_in.len; i++)
wrbuf_printf(wrbuf, " " ZINT_FORMAT,
key_in.mem[i]);
-
- /* zebra_term_untrans(zh, index_type, dst_buf, str); */
+
+ /* zebra_term_untrans(zh, index_type, dst_buf, str); */
wrbuf_printf(wrbuf, " %s", dst_buf);
wrbuf_printf(wrbuf, "\n");
}
}
}
- if (input_format == VAL_TEXT_XML)
+ if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
wrbuf_printf(wrbuf, "</record>\n");
*rec_lenp = wrbuf_len(wrbuf);
*rec_bufp = odr_malloc(odr, *rec_lenp);
memcpy(*rec_bufp, wrbuf_buf(wrbuf), *rec_lenp);
- wrbuf_free(wrbuf, 1);
+ wrbuf_destroy(wrbuf);
}
zebra_rec_keys_close(keys);
return ret_code;
wrbuf_printf(wrbuf, "%s %i\n", name, value);
}
-int zebra_special_fetch(ZebraHandle zh, zint sysno, int score, ODR odr,
- const char *elemsetname,
- oid_value input_format,
- oid_value *output_format,
- char **rec_bufp, int *rec_lenp)
+
+static void snippet_xml_record(ZebraHandle zh, WRBUF wrbuf, zebra_snippets *doc)
+{
+ const zebra_snippet_word *doc_w;
+ int mark_state = 0;
+
+ wrbuf_printf(wrbuf, "%s>\n", ZEBRA_XML_HEADER_STR);
+ for (doc_w = zebra_snippets_constlist(doc); doc_w; doc_w = doc_w->next)
+ {
+ if (doc_w->mark)
+ {
+ int index_type;
+ const char *db = 0;
+ const char *string_index = 0;
+
+ zebraExplain_lookup_ord(zh->reg->zei, doc_w->ord,
+ &index_type, &db, &string_index);
+
+ if (mark_state == 0)
+ {
+ wrbuf_printf(wrbuf, " <snippet name=\"%s\"", string_index);
+ wrbuf_printf(wrbuf, " type=\"%c\">", index_type);
+ }
+ if (doc_w->match)
+ wrbuf_puts(wrbuf, "<s>");
+ /* not printing leading ws */
+ if (mark_state || !doc_w->ws || doc_w->match)
+ wrbuf_xmlputs(wrbuf, doc_w->term);
+ if (doc_w->match)
+ wrbuf_puts(wrbuf, "</s>");
+ }
+ else if (mark_state == 1)
+ {
+ wrbuf_puts(wrbuf, "</snippet>\n");
+ }
+ mark_state = doc_w->mark;
+ }
+ if (mark_state == 1)
+ {
+ wrbuf_puts(wrbuf, "</snippet>\n");
+ }
+ wrbuf_printf(wrbuf, "</record>");
+}
+
+int zebra_get_rec_snippets(ZebraHandle zh, zint sysno,
+ zebra_snippets *snippets)
+{
+ int return_code = 0;
+ Record rec = rec_get(zh->reg->records, sysno);
+ if (!rec)
+ {
+ yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno);
+ return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
+ }
+ else
+ {
+ const char *file_type = rec->info[recInfo_fileType];
+ void *recTypeClientData;
+ RecType rt = recType_byName(zh->reg->recTypes, zh->res,
+ file_type, &recTypeClientData);
+
+ if (!rt)
+ return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
+ else
+ {
+ struct ZebraRecStream stream;
+ return_code = zebra_create_record_stream(zh, &rec, &stream);
+ if (return_code == 0)
+ {
+ extract_snippet(zh, snippets, &stream,
+ rt, recTypeClientData);
+
+ stream.destroy(&stream);
+ }
+ }
+ rec_free(&rec);
+ }
+ return return_code;
+}
+
+int zebra_special_snippet_fetch(ZebraHandle zh, const char *setname,
+ zint sysno, ODR odr,
+ const char *elemsetname,
+ const Odr_oid *input_format,
+ const Odr_oid **output_format,
+ char **rec_bufp, int *rec_lenp)
+{
+ zebra_snippets *rec_snippets = zebra_snippets_create();
+ int return_code = zebra_get_rec_snippets(zh, sysno, rec_snippets);
+
+ if (!return_code)
+ {
+ WRBUF wrbuf = wrbuf_alloc();
+ zebra_snippets *hit_snippet = zebra_snippets_create();
+
+ zebra_snippets_hit_vector(zh, setname, sysno, hit_snippet);
+
+#if 0
+ /* for debugging purposes */
+ yaz_log(YLOG_LOG, "---------------------------");
+ yaz_log(YLOG_LOG, "REC SNIPPET:");
+ zebra_snippets_log(rec_snippet, YLOG_LOG, 1);
+ yaz_log(YLOG_LOG, "---------------------------");
+ yaz_log(YLOG_LOG, "HIT SNIPPET:");
+ zebra_snippets_log(hit_snippet, YLOG_LOG, 1);
+#endif
+
+ zebra_snippets_ring(rec_snippets, hit_snippet, 5, 5);
+
+#if 0
+ yaz_log(YLOG_LOG, "---------------------------");
+ yaz_log(YLOG_LOG, "RING SNIPPET:");
+ zebra_snippets_log(rec_snippets, YLOG_LOG, 1);
+#endif
+ snippet_xml_record(zh, wrbuf, rec_snippets);
+
+ *output_format = yaz_oid_recsyn_xml;
+
+ if (return_code == 0)
+ {
+ *rec_lenp = wrbuf_len(wrbuf);
+ *rec_bufp = odr_strdup(odr, wrbuf_cstr(wrbuf));
+ }
+ wrbuf_destroy(wrbuf);
+ zebra_snippets_destroy(hit_snippet);
+ }
+ zebra_snippets_destroy(rec_snippets);
+ return return_code;
+}
+
+int zebra_special_fetch(ZebraHandle zh, const char *setname,
+ zint sysno, int score, ODR odr,
+ const char *elemsetname,
+ const Odr_oid *input_format,
+ const Odr_oid **output_format,
+ char **rec_bufp, int *rec_lenp)
{
Record rec;
/* *rec_lenp = 0; */
+ if (elemsetname && 0 == strcmp(elemsetname, "snippet"))
+ {
+ return zebra_special_snippet_fetch(zh, setname, sysno, odr,
+ elemsetname + 7,
+ input_format, output_format,
+ rec_bufp, rec_lenp);
+ }
/* processing zebra::meta::sysno elemset without fetching binary data */
if (elemsetname && 0 == strcmp(elemsetname, "meta::sysno"))
{
int ret = 0;
WRBUF wrbuf = wrbuf_alloc();
- if (input_format == VAL_SUTRS)
+ if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
{
wrbuf_printf(wrbuf, ZINT_FORMAT, sysno);
- *output_format = VAL_SUTRS;
+ *output_format = input_format;
}
- else if (input_format == VAL_TEXT_XML)
+ else if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
{
wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
" sysno=\"" ZINT_FORMAT "\"/>\n",
sysno);
- *output_format = VAL_TEXT_XML;
+ *output_format = input_format;
}
*rec_lenp = wrbuf_len(wrbuf);
if (*rec_lenp)
- *rec_bufp = odr_strdup(odr, wrbuf_buf(wrbuf));
+ *rec_bufp = odr_strdup(odr, wrbuf_cstr(wrbuf));
else
ret = YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
- wrbuf_free(wrbuf, 1);
+ wrbuf_destroy(wrbuf);
return ret;
}
}
/* only accept XML and SUTRS requests from now */
- if (input_format != VAL_TEXT_XML && input_format != VAL_SUTRS)
+ if (oid_oidcmp(input_format, yaz_oid_recsyn_xml)
+ && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
{
yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s",
elemsetname);
WRBUF wrbuf = wrbuf_alloc();
RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec);
- if (input_format == VAL_TEXT_XML)
+ if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
{
- *output_format = VAL_TEXT_XML;
+ *output_format = input_format;
wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
" sysno=\"" ZINT_FORMAT "\"", sysno);
recordAttr->recordSize,
elemsetname);
}
- else if (input_format == VAL_SUTRS)
+ else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs))
{
- *output_format = VAL_SUTRS;
+ *output_format = input_format;
wrbuf_printf(wrbuf, "sysno " ZINT_FORMAT "\n", sysno);
retrieve_puts_str(wrbuf, "base", rec->info[recInfo_databaseName]);
retrieve_puts_str(wrbuf, "file", rec->info[recInfo_filename]);
}
*rec_lenp = wrbuf_len(wrbuf);
if (*rec_lenp)
- *rec_bufp = odr_strdup(odr, wrbuf_buf(wrbuf));
+ *rec_bufp = odr_strdup(odr, wrbuf_cstr(wrbuf));
else
ret = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
- wrbuf_free(wrbuf, 1);
+ wrbuf_destroy(wrbuf);
rec_free(&rec);
return ret;
}
}
-int zebra_record_fetch(ZebraHandle zh, zint sysno, int score,
+int zebra_record_fetch(ZebraHandle zh, const char *setname,
+ zint sysno, int score,
zebra_snippets *hit_snippet, ODR odr,
- oid_value input_format, Z_RecordComposition *comp,
- oid_value *output_format,
+ const Odr_oid *input_format, Z_RecordComposition *comp,
+ const Odr_oid **output_format,
char **rec_bufp, int *rec_lenp, char **basenamep,
char **addinfo)
{
/* processing zebra special elementset names of form 'zebra:: */
if (elemsetname && 0 == strncmp(elemsetname, "zebra::", 7))
- return zebra_special_fetch(zh, sysno, score, odr,
+ return zebra_special_fetch(zh, setname, sysno, score, odr,
elemsetname + 7,
input_format, output_format,
rec_bufp, rec_lenp);
if (rec)
{
- zebra_snippets *snippet;
zebra_rec_keys_t reckeys = zebra_rec_keys_open();
RecType rt;
struct recRetrieveCtrl retrieveCtrl;
zebra_rec_keys_to_snippets(zh, reckeys, retrieveCtrl.doc_snippet);
zebra_rec_keys_close(reckeys);
-#if 0
- /* for debugging purposes */
- yaz_log(YLOG_LOG, "DOC SNIPPET:");
- zebra_snippets_log(retrieveCtrl.doc_snippet, YLOG_LOG);
- yaz_log(YLOG_LOG, "HIT SNIPPET:");
- zebra_snippets_log(retrieveCtrl.hit_snippet, YLOG_LOG);
-#endif
- snippet = zebra_snippets_window(retrieveCtrl.doc_snippet,
- retrieveCtrl.hit_snippet,
- 10);
-#if 0
- /* for debugging purposes */
- yaz_log(YLOG_LOG, "WINDOW SNIPPET:");
- zebra_snippets_log(snippet, YLOG_LOG);
-#endif
-
if (!(rt = recType_byName(zh->reg->recTypes, zh->res,
file_type, &clientData)))
{
*addinfo = retrieveCtrl.addinfo;
}
- zebra_snippets_destroy(snippet);
zebra_snippets_destroy(retrieveCtrl.doc_snippet);
stream.destroy(&stream);