-/* $Id: retrieve.c,v 1.59 2006-11-24 12:21:31 marc Exp $
- Copyright (C) 1995-2006
+/* $Id: retrieve.c,v 1.71 2007-08-21 11:06:47 adam Exp $
+ Copyright (C) 1995-2007
Index Data ApS
This file is part of the Zebra server.
#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);
if ((*rec)->size[recInfo_storeData] > 0)
yaz_log (YLOG_WARN|YLOG_ERRNO, "Retrieve fail; missing file: %s",
full_rep);
rec_free(rec);
- return 14;
+ return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
}
zebra_create_stream_fd(stream, fd, recordAttr->recordOffset);
}
static int parse_zebra_elem(const char *elem,
- const char **index, size_t *index_len,
- const char **type, size_t *type_len)
+ const char **index, size_t *index_len,
+ const char **type, size_t *type_len)
{
*index = 0;
*index_len = 0;
}
+int zebra_special_sort_fetch(ZebraHandle zh, zint sysno, ODR odr,
+ const char *elemsetname,
+ const Odr_oid *input_format,
+ const Odr_oid **output_format,
+ char **rec_bufp, int *rec_lenp)
+{
+ const char *retrieval_index;
+ size_t retrieval_index_len;
+ const char *retrieval_type;
+ size_t retrieval_type_len;
+ char retrieval_index_cstr[256];
+ int ord;
+
+ /* only accept XML and SUTRS requests */
+ 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 = 0;
+ return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
+ }
+
+ if (!parse_zebra_elem(elemsetname,
+ &retrieval_index, &retrieval_index_len,
+ &retrieval_type, &retrieval_type_len))
+ {
+ return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
+ }
+
+ if (retrieval_type_len == 0)
+ return -1; /* must have a register type specified */
+ if (!retrieval_index_len ||
+ retrieval_index_len >= sizeof(retrieval_index_cstr)-1)
+ {
+ return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
+ }
+
+ memcpy(retrieval_index_cstr, retrieval_index, retrieval_index_len);
+ retrieval_index_cstr[retrieval_index_len] = '\0';
+
+ ord = zebraExplain_lookup_attr_str(zh->reg->zei,
+ zinfo_index_category_sort,
+ retrieval_type[0],
+ retrieval_index_cstr);
+ if (ord == -1)
+ return -1; /* is not a sort index */
+ else
+ {
+ char dst_buf[IT_MAX_WORD];
+ char str[IT_MAX_WORD];
+ int index_type;
+ const char *db = 0;
+ const char *string_index = 0;
+ WRBUF wrbuf = wrbuf_alloc();
+
+ zebra_sort_sysno(zh->reg->sort_index, sysno);
+ zebra_sort_type(zh->reg->sort_index, ord);
+ zebra_sort_read(zh->reg->sort_index, str);
+
+ zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, &string_index);
+
+ zebra_term_untrans(zh, index_type, dst_buf, str);
+
+
+ if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml))
+ {
+ *output_format = yaz_oid_recsyn_xml;
+ wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR
+ " sysno=\"" ZINT_FORMAT "\""
+ " set=\"zebra::index%s/\">\n",
+ sysno, elemsetname);
+
+ wrbuf_printf(wrbuf, " <index name=\"%s\"",
+ string_index);
+ wrbuf_printf(wrbuf, " type=\"%c\">", index_type);
+ wrbuf_xmlputs(wrbuf, dst_buf);
+ wrbuf_printf(wrbuf, "</index>\n");
+ wrbuf_printf(wrbuf, "</record>\n");
+ }
+ else if (!oid_oidcmp(input_format, yaz_oid_recsyn_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_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;
size_t retrieval_index_len;
const char *retrieval_type;
size_t retrieval_type_len;
- WRBUF wrbuf = 0;
zebra_rec_keys_t keys;
+ int ret_code = 0;
/* set output variables before processing possible error states */
/* *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;
}
{
char retrieval_index_cstr[256];
- if (retrieval_index_len < sizeof(retrieval_index_cstr) -1)
+ if (retrieval_index_len < sizeof(retrieval_index_cstr) -1)
{
memcpy(retrieval_index_cstr, retrieval_index, retrieval_index_len);
retrieval_index_cstr[retrieval_index_len] = '\0';
zebra_rec_keys_set_buf(keys, rec->info[recInfo_delKeys],
rec->size[recInfo_delKeys], 0);
- wrbuf = wrbuf_alloc();
- if (zebra_rec_keys_rewind(keys)){
+ if (!zebra_rec_keys_rewind(keys))
+ {
+ ret_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
+ }
+ else
+ {
size_t slen;
const char *str;
struct it_key key_in;
-
- if (input_format == VAL_TEXT_XML)
+ WRBUF wrbuf = wrbuf_alloc();
+
+ 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)){
+ while (zebra_rec_keys_read(keys, &str, &slen, &key_in))
+ {
int i;
int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
int index_type;
if (retrieval_index == 0
|| (string_index_len == retrieval_index_len
&& !memcmp(string_index, retrieval_index,
- string_index_len))){
-
+ string_index_len)))
+ {
/* process only if type is not defined, or is matching */
if (retrieval_type == 0
|| (retrieval_type_len == 1
- && retrieval_type[0] == index_type)){
-
-
+ && retrieval_type[0] == index_type))
+ {
zebra_term_untrans(zh, index_type, dst_buf, str);
- if (strlen(dst_buf)){
-
- if (input_format == VAL_TEXT_XML){
+ if (strlen(dst_buf))
+ {
+ 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);
+ *rec_lenp = wrbuf_len(wrbuf);
+ *rec_bufp = odr_malloc(odr, *rec_lenp);
+ memcpy(*rec_bufp, wrbuf_buf(wrbuf), *rec_lenp);
+ wrbuf_destroy(wrbuf);
+ }
zebra_rec_keys_close(keys);
- return 0;
+ 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_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)
+{
+ int return_code = 0;
+ Record rec;
+
+ 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);
+ zebra_snippets *hit_snippet = zebra_snippets_create();
+ WRBUF wrbuf = wrbuf_alloc();
+
+ zebra_snippets_hit_vector(zh, setname, sysno, hit_snippet);
+
+ 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)
+ {
+ zebra_snippets *rec_snippet = zebra_snippets_create();
+ extract_snippet(zh, rec_snippet, &stream,
+ rt, recTypeClientData);
+
+#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_snippet, hit_snippet, 5, 5);
+
+#if 0
+ yaz_log(YLOG_LOG, "---------------------------");
+ yaz_log(YLOG_LOG, "RING SNIPPET:");
+ zebra_snippets_log(rec_snippet, YLOG_LOG, 1);
+#endif
+
+ snippet_xml_record(zh, wrbuf, rec_snippet);
+
+ *output_format = yaz_oid_recsyn_xml;
+
+
+ zebra_snippets_destroy(rec_snippet);
+ }
+ stream.destroy(&stream);
+ }
+ if (return_code == 0)
+ {
+ *rec_lenp = wrbuf_len(wrbuf);
+ *rec_bufp = odr_strdup(odr, wrbuf_cstr(wrbuf));
+ }
+ wrbuf_destroy(wrbuf);
+ rec_free(&rec);
+ zebra_snippets_destroy(hit_snippet);
+ }
+ 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 "\""
- " set=\"zebra::%s\"/>\n",
- sysno, elemsetname);
- *output_format = VAL_TEXT_XML;
+ " sysno=\"" ZINT_FORMAT "\"/>\n",
+ sysno);
+ *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;
}
+ /* processing special elementsetname zebra::index:: for sort elements */
+ if (elemsetname && 0 == strncmp(elemsetname, "index", 5))
+ {
+ int ret = zebra_special_sort_fetch(zh, sysno, odr,
+ elemsetname + 5,
+ input_format, output_format,
+ rec_bufp, rec_lenp);
+ if (ret != -1)
+ return ret;
+ /* not a sort index so we continue to get the full record */
+ }
+
+
/* fetching binary record up for all other display elementsets */
rec = rec_get(zh->reg->records, sysno);
if (!rec)
}
/* 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;
}
/* processing special elementsetnames zebra::index:: */
- if (elemsetname && 0 == strncmp(elemsetname, "index", 5)){
-
+ if (elemsetname && 0 == strncmp(elemsetname, "index", 5))
+ {
int ret = zebra_special_index_fetch(zh, sysno, odr, rec,
elemsetname + 5,
input_format, output_format,
}
-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);
yaz_log(YLOG_DEBUG, "retrieve localno=" ZINT_FORMAT " score=%d",
sysno, score);
- zebra_create_record_stream(zh, &rec, &stream);
-
+ return_code = zebra_create_record_stream(zh, &rec, &stream);
+
+ if (rec)
{
- /* snippets code */
- 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)))
{
+ char addinfo_str[100];
+
+ sprintf(addinfo_str, "Could not handle record type %.40s",
+ file_type);
+
+ *addinfo = odr_strdup(odr, addinfo_str);
return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
}
else
*addinfo = retrieveCtrl.addinfo;
}
- zebra_snippets_destroy(snippet);
zebra_snippets_destroy(retrieveCtrl.doc_snippet);
- }
- stream.destroy(&stream);
- rec_free(&rec);
+ stream.destroy(&stream);
+ rec_free(&rec);
+ }
return return_code;
}
projects / idzebra-moved-to-github.git / blobdiff