-/* $Id: extract.c,v 1.233 2006-10-30 11:18:26 adam Exp $
- Copyright (C) 1995-2006
+/* $Id: extract.c,v 1.275 2007-12-20 11:15:42 adam Exp $
+ Copyright (C) 1995-2007
Index Data ApS
This file is part of the Zebra server.
*/
+/** \file
+ \brief indexes records and extract tokens for indexing and sorting
+*/
+
#include <stdio.h>
#include <assert.h>
#include <ctype.h>
#endif
#include <fcntl.h>
+
#include "index.h"
#include "orddict.h"
#include <direntz.h>
#include <charmap.h>
+#include <yaz/snprintf.h>
-#define ENCODE_BUFLEN 768
-struct encode_info {
- void *encode_handle;
- void *decode_handle;
- char buf[ENCODE_BUFLEN];
-};
-
-static int log_level = 0;
+static int log_level_extract = 0;
+static int log_level_details = 0;
static int log_level_initialized = 0;
+/* 1 if we use eliminitate identical delete/insert keys */
+/* eventually this the 0-case code will be removed */
+#define FLUSH2 1
+
+void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
+ zebra_rec_keys_t ins_keys,
+ zint ins_rank,
+ zebra_rec_keys_t del_keys,
+ zint del_rank);
+
static void zebra_init_log_level(void)
{
if (!log_level_initialized)
{
- log_level = yaz_log_module_level("extract");
log_level_initialized = 1;
+
+ log_level_extract = yaz_log_module_level("extract");
+ log_level_details = yaz_log_module_level("indexdetails");
}
}
-static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
- int cmd, zebra_rec_keys_t reckeys,
- zint staticrank);
-static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
- int cmd, zebra_rec_keys_t skp);
-static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid);
-static void extract_token_add (RecWord *p);
-
-static void encode_key_init (struct encode_info *i);
-static void encode_key_write (char *k, struct encode_info *i, FILE *outf);
-static void encode_key_flush (struct encode_info *i, FILE *outf);
+static void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
+ int cmd, zebra_rec_keys_t skp);
+static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid);
+static void extract_token_add(RecWord *p);
-#define USE_SHELLSORT 0
-
-#if USE_SHELLSORT
-static void shellsort(void *ar, int r, size_t s,
- int (*cmp)(const void *a, const void *b))
+static void check_log_limit(ZebraHandle zh)
{
- char *a = ar;
- char v[100];
- int h, i, j, k;
- static const int incs[16] = { 1391376, 463792, 198768, 86961, 33936,
- 13776, 4592, 1968, 861, 336,
- 112, 48, 21, 7, 3, 1 };
- for ( k = 0; k < 16; k++)
- for (h = incs[k], i = h; i < r; i++)
- {
- memcpy (v, a+s*i, s);
- j = i;
- while (j > h && (*cmp)(a + s*(j-h), v) > 0)
- {
- memcpy (a + s*j, a + s*(j-h), s);
- j -= h;
- }
- memcpy (a+s*j, v, s);
- }
+ if (zh->records_processed + zh->records_skipped == zh->m_file_verbose_limit)
+ {
+ yaz_log(YLOG_LOG, "More than %d file log entries. Omitting rest",
+ zh->m_file_verbose_limit);
+ }
}
-#endif
-static void logRecord (ZebraHandle zh)
+static void logRecord(ZebraHandle zh)
{
+ check_log_limit(zh);
++zh->records_processed;
if (!(zh->records_processed % 1000))
{
}
}
-static void extract_add_index_string (RecWord *p,
+static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
+{
+ ctrl->flagShowRecords = !zh->m_flag_rw;
+}
+
+
+static void extract_add_index_string(RecWord *p,
zinfo_index_category_t cat,
const char *str, int length);
{
w->seqno = 1;
w->index_name = "any";
- w->index_type = 'w';
+ w->index_type = "w";
w->extractCtrl = p;
w->record_id = 0;
w->section_id = 0;
w->segment = 0;
}
+struct snip_rec_info {
+ ZebraHandle zh;
+ zebra_snippets *snippets;
+};
+
+
+static void snippet_add_complete_field(RecWord *p, int ord,
+ zebra_map_t zm)
+{
+ struct snip_rec_info *h = p->extractCtrl->handle;
+
+ const char *b = p->term_buf;
+ char buf[IT_MAX_WORD+1];
+ const char **map = 0;
+ int i = 0, remain = p->term_len;
+ const char *start = b;
+ const char *last = 0;
+
+ if (remain > 0)
+ map = zebra_maps_input(zm, &b, remain, 1);
+
+ while (remain > 0 && i < IT_MAX_WORD)
+ {
+ while (map && *map && **map == *CHR_SPACE)
+ {
+ remain = p->term_len - (b - p->term_buf);
+
+ if (i == 0)
+ start = b; /* set to first non-ws area */
+ if (remain > 0)
+ {
+ int first = i ? 0 : 1; /* first position */
+
+ map = zebra_maps_input(zm, &b, remain, first);
+ }
+ else
+ map = 0;
+ }
+ if (!map)
+ break;
+
+ if (i && i < IT_MAX_WORD)
+ buf[i++] = *CHR_SPACE;
+ while (map && *map && **map != *CHR_SPACE)
+ {
+ const char *cp = *map;
+
+ if (**map == *CHR_CUT)
+ {
+ i = 0;
+ }
+ else
+ {
+ if (i >= IT_MAX_WORD)
+ break;
+ while (i < IT_MAX_WORD && *cp)
+ buf[i++] = *(cp++);
+ }
+ last = b;
+ remain = p->term_len - (b - p->term_buf);
+ if (remain > 0)
+ {
+ map = zebra_maps_input(zm, &b, remain, 0);
+ }
+ else
+ map = 0;
+ }
+ }
+ if (!i)
+ return;
+ if (last && start != last)
+ zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
+ start, last - start);
+}
+
+static void snippet_add_incomplete_field(RecWord *p, int ord, zebra_map_t zm)
+{
+ struct snip_rec_info *h = p->extractCtrl->handle;
+ const char *b = p->term_buf;
+ int remain = p->term_len;
+ int first = 1;
+ const char **map = 0;
+ const char *start = b;
+ const char *last = b;
+
+ if (remain > 0)
+ map = zebra_maps_input(zm, &b, remain, 0);
+
+ while (map)
+ {
+ char buf[IT_MAX_WORD+1];
+ int i, remain;
+
+ /* Skip spaces */
+ while (map && *map && **map == *CHR_SPACE)
+ {
+ remain = p->term_len - (b - p->term_buf);
+ last = b;
+ if (remain > 0)
+ map = zebra_maps_input(zm, &b, remain, 0);
+ else
+ map = 0;
+ }
+ if (!map)
+ break;
+ if (start != last)
+ {
+ zebra_snippets_appendn(h->snippets, p->seqno, 1, ord,
+ start, last - start);
+
+ }
+ start = last;
+
+ i = 0;
+ while (map && *map && **map != *CHR_SPACE)
+ {
+ const char *cp = *map;
+
+ while (i < IT_MAX_WORD && *cp)
+ buf[i++] = *(cp++);
+ remain = p->term_len - (b - p->term_buf);
+ last = b;
+ if (remain > 0)
+ map = zebra_maps_input(zm, &b, remain, 0);
+ else
+ map = 0;
+ }
+ if (!i)
+ return;
+
+ if (first)
+ {
+ first = 0;
+ if (zebra_maps_is_first_in_field(zm))
+ {
+ /* first in field marker */
+ p->seqno++;
+ }
+ }
+ if (start != last)
+ zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
+ start, last - start);
+ start = last;
+ p->seqno++;
+ }
+
+}
+
+static void snippet_add_icu(RecWord *p, int ord, zebra_map_t zm)
+{
+ struct snip_rec_info *h = p->extractCtrl->handle;
+
+ const char *res_buf = 0;
+ size_t res_len = 0;
+
+ const char *display_buf = 0;
+ size_t display_len = 0;
+
+ zebra_map_tokenize_start(zm, p->term_buf, p->term_len);
+ while (zebra_map_tokenize_next(zm, &res_buf, &res_len,
+ &display_buf, &display_len))
+ {
+ zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
+ display_buf, display_len);
+ p->seqno++;
+ }
+}
+
+static void snippet_token_add(RecWord *p)
+{
+ struct snip_rec_info *h = p->extractCtrl->handle;
+ ZebraHandle zh = h->zh;
+ zebra_map_t zm = zebra_map_get(zh->reg->zebra_maps, p->index_type);
+
+ if (zm && zebra_maps_is_index(zm))
+ {
+ ZebraExplainInfo zei = zh->reg->zei;
+ int ch = zebraExplain_lookup_attr_str(
+ zei, zinfo_index_category_index, p->index_type, p->index_name);
+
+ if (zebra_maps_is_icu(zm))
+ snippet_add_icu(p, ch, zm);
+ else
+ {
+ if (zebra_maps_is_complete(zm))
+ snippet_add_complete_field(p, ch, zm);
+ else
+ snippet_add_incomplete_field(p, ch, zm);
+ }
+ }
+}
+
+static void snippet_schema_add(
+ struct recExtractCtrl *p, Odr_oid *oid)
+{
+
+}
+
+void extract_snippet(ZebraHandle zh, zebra_snippets *sn,
+ struct ZebraRecStream *stream,
+ RecType rt, void *recTypeClientData)
+{
+ struct recExtractCtrl extractCtrl;
+ struct snip_rec_info info;
+ int r;
+
+ extractCtrl.stream = stream;
+ extractCtrl.first_record = 1;
+ extractCtrl.init = extract_init;
+ extractCtrl.tokenAdd = snippet_token_add;
+ extractCtrl.schemaAdd = snippet_schema_add;
+ assert(zh->reg);
+ assert(zh->reg->dh);
+
+ extractCtrl.dh = zh->reg->dh;
+
+ info.zh = zh;
+ info.snippets = sn;
+ extractCtrl.handle = &info;
+ extractCtrl.match_criteria[0] = '\0';
+ extractCtrl.staticrank = 0;
+ extractCtrl.action = action_insert;
+
+ init_extractCtrl(zh, &extractCtrl);
+
+ extractCtrl.setStoreData = 0;
+
+ r = (*rt->extract)(recTypeClientData, &extractCtrl);
+
+}
+
static void searchRecordKey(ZebraHandle zh,
zebra_rec_keys_t reckeys,
const char *index_name,
ws[i] = NULL;
if (ch < 0)
- ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, '0', index_name);
+ ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "0", index_name);
if (ch < 0)
- ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'p', index_name);
+ ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "p", index_name);
if (ch < 0)
- ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'w', index_name);
+ ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "w", index_name);
if (ch < 0)
return ;
#define FILE_MATCH_BLANK "\t "
-static char *fileMatchStr (ZebraHandle zh,
- zebra_rec_keys_t reckeys,
- const char *fname, const char *spec)
+static char *get_match_from_spec(ZebraHandle zh,
+ zebra_rec_keys_t reckeys,
+ const char *fname, const char *spec)
{
static char dstBuf[2048]; /* static here ??? */
char *dst = dstBuf;
attname_str[i] = '\0';
}
- searchRecordKey (zh, reckeys, attname_str, ws, 32);
+ searchRecordKey(zh, reckeys, attname_str, ws, 32);
if (*s != ')')
{
- yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
+ yaz_log(YLOG_WARN, "Missing ) in match criteria %s in group %s",
spec, zh->m_group ? zh->m_group : "none");
return NULL;
}
*dst++ = ' ';
first = 0;
}
- strcpy (dst, ws[i]);
+ strcpy(dst, ws[i]);
dst += strlen(ws[i]);
}
if (first)
{
- yaz_log (YLOG_WARN, "Record didn't contain match"
+ yaz_log(YLOG_WARN, "Record didn't contain match"
" fields in (%s,%s)", attset_str, attname_str);
return NULL;
}
spec_len = s1 - s;
if (spec_len > sizeof(special)-1)
spec_len = sizeof(special)-1;
- memcpy (special, s, spec_len);
+ memcpy(special, s, spec_len);
special[spec_len] = '\0';
s = s1;
- if (!strcmp (special, "group"))
+ if (!strcmp(special, "group"))
spec_src = zh->m_group;
- else if (!strcmp (special, "database"))
+ else if (!strcmp(special, "database"))
spec_src = zh->basenames[0];
- else if (!strcmp (special, "filename")) {
+ else if (!strcmp(special, "filename")) {
spec_src = fname;
}
- else if (!strcmp (special, "type"))
+ else if (!strcmp(special, "type"))
spec_src = zh->m_record_type;
else
spec_src = NULL;
if (spec_src)
{
- strcpy (dst, spec_src);
- dst += strlen (spec_src);
+ strcpy(dst, spec_src);
+ dst += strlen(spec_src);
}
}
else if (*s == '\"' || *s == '\'')
if (*s)
s++;
tmpString[i] = '\0';
- strcpy (dst, tmpString);
- dst += strlen (tmpString);
+ strcpy(dst, tmpString);
+ dst += strlen(tmpString);
}
else
{
- yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
+ yaz_log(YLOG_WARN, "Syntax error in match criteria %s in group %s",
spec, zh->m_group ? zh->m_group : "none");
return NULL;
}
}
if (dst == dstBuf)
{
- yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
+ yaz_log(YLOG_WARN, "No match criteria for record %s in group %s",
fname, zh->m_group ? zh->m_group : "none");
return NULL;
}
struct recordGroup *rGroup;
};
-static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
-{
- int i;
- for (i = 0; i<256; i++)
- {
- if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
- ctrl->seqno[i] = 1;
- else
- ctrl->seqno[i] = 0;
- }
- ctrl->flagShowRecords = !zh->m_flag_rw;
-}
-
-static void all_matches_add(struct recExtractCtrl *ctrl)
+/** \brief add the always-matches index entry and map to real record ID
+ \param ctrl record control
+ \param record_id custom record ID
+ \param sysno system record ID
+
+ This function serves two purposes.. It adds the always matches
+ entry and makes a pointer from the custom record ID (if defined)
+ back to the system record ID (sysno)
+ See zebra_recid_to_sysno .
+ */
+static void all_matches_add(struct recExtractCtrl *ctrl, zint record_id,
+ zint sysno)
{
RecWord word;
extract_init(ctrl, &word);
+ word.record_id = record_id;
+ /* we use the seqno as placeholder for a way to get back to
+ record database from _ALLRECORDS.. This is used if a custom
+ RECORD was defined */
+ word.seqno = sysno;
word.index_name = "_ALLRECORDS";
- word.index_type = 'w';
- word.seqno = 1;
- extract_add_index_string (&word, zinfo_index_category_alwaysmatches,
+ word.index_type = "w";
+
+ extract_add_index_string(&word, zinfo_index_category_alwaysmatches,
"", 0);
}
-ZEBRA_RES zebra_extract_file(ZebraHandle zh, SYSNO *sysno, const char *fname,
- int deleteFlag)
+ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
+ struct ZebraRecStream *stream,
+ enum zebra_recctrl_action_t action,
+ int test_mode,
+ const char *recordType,
+ zint *sysno,
+ const char *match_criteria,
+ const char *fname,
+ RecType recType,
+ void *recTypeClientData);
+
+
+ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
+ enum zebra_recctrl_action_t action)
{
ZEBRA_RES r = ZEBRA_OK;
int i, fd;
if (!zh->m_group || !*zh->m_group)
*gprefix = '\0';
else
- sprintf (gprefix, "%s.", zh->m_group);
+ sprintf(gprefix, "%s.", zh->m_group);
- yaz_log(log_level, "zebra_extract_file %s", fname);
+ yaz_log(log_level_extract, "zebra_extract_file %s", fname);
/* determine file extension */
*ext = '\0';
break;
else if (fname[i] == '.')
{
- strcpy (ext, fname+i+1);
+ strcpy(ext, fname+i+1);
break;
}
/* determine file type - depending on extension */
original_record_type = zh->m_record_type;
if (!zh->m_record_type)
{
- sprintf (ext_res, "%srecordType.%s", gprefix, ext);
- zh->m_record_type = res_get (zh->res, ext_res);
+ sprintf(ext_res, "%srecordType.%s", gprefix, ext);
+ zh->m_record_type = res_get(zh->res, ext_res);
}
if (!zh->m_record_type)
{
- if (zh->records_processed < zh->m_file_verbose_limit)
- yaz_log (YLOG_LOG, "? %s", fname);
+ check_log_limit(zh);
+ if (zh->records_processed + zh->records_skipped
+ < zh->m_file_verbose_limit)
+ yaz_log(YLOG_LOG, "? %s", fname);
+ zh->records_skipped++;
return 0;
}
/* determine match criteria */
if (!zh->m_record_id)
{
- sprintf (ext_res, "%srecordId.%s", gprefix, ext);
- zh->m_record_id = res_get (zh->res, ext_res);
+ sprintf(ext_res, "%srecordId.%s", gprefix, ext);
+ zh->m_record_id = res_get(zh->res, ext_res);
}
if (!(recType =
- recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
+ recType_byName(zh->reg->recTypes, zh->res, zh->m_record_type,
&recTypeClientData)))
{
yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
default:
yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
}
- if (sysno && deleteFlag)
+ if (sysno && (action == action_delete || action == action_a_delete))
{
streamp = 0;
fi = 0;
{
char full_rep[1024];
- if (zh->path_reg && !yaz_is_abspath (fname))
+ if (zh->path_reg && !yaz_is_abspath(fname))
{
- strcpy (full_rep, zh->path_reg);
- strcat (full_rep, "/");
- strcat (full_rep, fname);
+ strcpy(full_rep, zh->path_reg);
+ strcat(full_rep, "/");
+ strcat(full_rep, fname);
}
else
- strcpy (full_rep, fname);
+ strcpy(full_rep, fname);
- if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
+ if ((fd = open(full_rep, O_BINARY|O_RDONLY)) == -1)
{
- yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
+ yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
zh->m_record_type = original_record_type;
return ZEBRA_FAIL;
}
streamp = &stream;
zebra_create_stream_fd(streamp, fd, 0);
}
- while(1)
- {
- r = zebra_extract_record_stream(zh, streamp,
- deleteFlag,
- 0, /* tst_mode */
- zh->m_record_type,
- sysno,
- 0, /*match_criteria */
- fname,
- 1, /* force_update */
- 1, /* allow_update */
- recType, recTypeClientData);
- if (r != ZEBRA_OK)
- {
- break;
- }
- if (sysno)
- {
- break;
- }
- }
+ r = zebra_extract_records_stream(zh, streamp,
+ action,
+ 0, /* tst_mode */
+ zh->m_record_type,
+ sysno,
+ 0, /*match_criteria */
+ fname,
+ recType, recTypeClientData);
if (streamp)
stream.destroy(streamp);
zh->m_record_type = original_record_type;
ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh,
const char *buf, size_t buf_size,
- int delete_flag,
+ enum zebra_recctrl_action_t action,
int test_mode,
const char *recordType,
- SYSNO *sysno,
+ zint *sysno,
const char *match_criteria,
- const char *fname,
- int force_update,
- int allow_update)
+ const char *fname)
{
struct ZebraRecStream stream;
ZEBRA_RES res;
if (recordType && *recordType)
{
- yaz_log(log_level, "Record type explicitly specified: %s", recordType);
- recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
+ yaz_log(log_level_extract,
+ "Record type explicitly specified: %s", recordType);
+ recType = recType_byName(zh->reg->recTypes, zh->res, recordType,
&clientData);
}
else
{
if (!(zh->m_record_type))
{
- yaz_log (YLOG_WARN, "No such record type defined");
+ yaz_log(YLOG_WARN, "No such record type defined");
return ZEBRA_FAIL;
}
- yaz_log(log_level, "Get record type from rgroup: %s",
+ yaz_log(log_level_extract, "Get record type from rgroup: %s",
zh->m_record_type);
- recType = recType_byName (zh->reg->recTypes, zh->res,
+ recType = recType_byName(zh->reg->recTypes, zh->res,
zh->m_record_type, &clientData);
recordType = zh->m_record_type;
}
if (!recType)
{
- yaz_log (YLOG_WARN, "No such record type: %s", recordType);
+ yaz_log(YLOG_WARN, "No such record type: %s", recordType);
return ZEBRA_FAIL;
}
-
-
zebra_create_stream_mem(&stream, buf, buf_size);
- res = zebra_extract_record_stream(zh, &stream,
- delete_flag,
- test_mode,
- recordType,
- sysno,
- match_criteria,
- fname,
- force_update,
- allow_update,
- recType, clientData);
+ res = zebra_extract_records_stream(zh, &stream,
+ action,
+ test_mode,
+ recordType,
+ sysno,
+ match_criteria,
+ fname,
+ recType, clientData);
stream.destroy(&stream);
return res;
}
+ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
+ struct ZebraRecStream *stream,
+ enum zebra_recctrl_action_t action,
+ int test_mode,
+ const char *recordType,
+ zint *sysno,
+ const char *match_criteria,
+ const char *fname,
+ RecType recType,
+ void *recTypeClientData)
+{
+ ZEBRA_RES res = ZEBRA_OK;
+ while (1)
+ {
+ int more = 0;
+ res = zebra_extract_record_stream(zh, stream,
+ action,
+ test_mode,
+ recordType,
+ sysno,
+ match_criteria,
+ fname,
+ recType, recTypeClientData, &more);
+ if (!more)
+ {
+ res = ZEBRA_OK;
+ break;
+ }
+ if (res != ZEBRA_OK)
+ break;
+ if (sysno)
+ break;
+ }
+ return res;
+}
+
+
+static WRBUF wrbuf_hex_str(const char *cstr)
+{
+ size_t i;
+ WRBUF w = wrbuf_alloc();
+ for (i = 0; cstr[i]; i++)
+ {
+ if (cstr[i] < ' ' || cstr[i] > 126)
+ wrbuf_printf(w, "\\%02X", cstr[i] & 0xff);
+ else
+ wrbuf_putc(w, cstr[i]);
+ }
+ return w;
+}
ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
struct ZebraRecStream *stream,
- int delete_flag,
+ enum zebra_recctrl_action_t action,
int test_mode,
const char *recordType,
- SYSNO *sysno,
+ zint *sysno,
const char *match_criteria,
const char *fname,
- int force_update,
- int allow_update,
RecType recType,
- void *recTypeClientData)
+ void *recTypeClientData,
+ int *more)
{
- SYSNO sysno0 = 0;
+ zint sysno0 = 0;
RecordAttr *recordAttr;
struct recExtractCtrl extractCtrl;
int r;
const char *matchStr = 0;
Record rec;
- off_t start_offset = 0;
+ off_t start_offset = 0, end_offset = 0;
const char *pr_fname = fname; /* filename to print .. */
- int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
+ int show_progress = zh->records_processed + zh->records_skipped
+ < zh->m_file_verbose_limit ? 1:0;
zebra_init_log_level();
zebra_rec_keys_reset(zh->reg->keys);
zebra_rec_keys_reset(zh->reg->sortKeys);
- if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
+ if (zebraExplain_curDatabase(zh->reg->zei, zh->basenames[0]))
{
- if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
+ if (zebraExplain_newDatabase(zh->reg->zei, zh->basenames[0],
zh->m_explain_database))
return ZEBRA_FAIL;
}
extractCtrl.handle = zh;
extractCtrl.match_criteria[0] = '\0';
extractCtrl.staticrank = 0;
+ extractCtrl.action = action;
-
init_extractCtrl(zh, &extractCtrl);
-
+
extract_set_store_data_prepare(&extractCtrl);
r = (*recType->extract)(recTypeClientData, &extractCtrl);
+
+ if (action == action_update)
+ {
+ action = extractCtrl.action;
+ }
- if (r == RECCTRL_EXTRACT_EOF)
- return ZEBRA_FAIL;
- else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
+ switch (r)
{
+ case RECCTRL_EXTRACT_EOF:
+ return ZEBRA_FAIL;
+ case RECCTRL_EXTRACT_ERROR_GENERIC:
/* error occured during extraction ... */
- yaz_log (YLOG_WARN, "extract error: generic");
+ yaz_log(YLOG_WARN, "extract error: generic");
return ZEBRA_FAIL;
- }
- else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
- {
+ case RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER:
/* error occured during extraction ... */
- yaz_log (YLOG_WARN, "extract error: no such filter");
+ yaz_log(YLOG_WARN, "extract error: no such filter");
+ return ZEBRA_FAIL;
+ case RECCTRL_EXTRACT_SKIP:
+ if (show_progress)
+ yaz_log(YLOG_LOG, "skip %s %s " ZINT_FORMAT,
+ recordType, pr_fname, (zint) start_offset);
+ *more = 1;
+
+ end_offset = stream->endf(stream, 0);
+ if (end_offset)
+ stream->seekf(stream, end_offset);
+
+ return ZEBRA_OK;
+ case RECCTRL_EXTRACT_OK:
+ break;
+ default:
+ yaz_log(YLOG_WARN, "extract error: unknown error: %d", r);
return ZEBRA_FAIL;
}
-
- all_matches_add(&extractCtrl);
-
+ end_offset = stream->endf(stream, 0);
+ if (end_offset)
+ stream->seekf(stream, end_offset);
+ else
+ end_offset = stream->tellf(stream);
+
if (extractCtrl.match_criteria[0])
match_criteria = extractCtrl.match_criteria;
}
- if (!sysno) {
+ *more = 1;
+ if (!sysno)
+ {
sysno = &sysno0;
if (match_criteria && *match_criteria) {
matchStr = match_criteria;
} else {
if (zh->m_record_id && *zh->m_record_id) {
- matchStr = fileMatchStr (zh, zh->reg->keys, pr_fname,
- zh->m_record_id);
+ matchStr = get_match_from_spec(zh, zh->reg->keys, pr_fname,
+ zh->m_record_id);
if (!matchStr)
{
- yaz_log (YLOG_WARN, "Bad match criteria (recordID)");
+ yaz_log(YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType,
+ pr_fname, (zint) start_offset);
return ZEBRA_FAIL;
}
}
int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
matchStr);
+
+
+ if (log_level_extract)
+ {
+ WRBUF w = wrbuf_hex_str(matchStr);
+ yaz_log(log_level_extract, "matchStr: %s", wrbuf_cstr(w));
+ wrbuf_destroy(w);
+ }
if (rinfo)
{
assert(*rinfo == sizeof(*sysno));
- memcpy (sysno, rinfo+1, sizeof(*sysno));
+ memcpy(sysno, rinfo+1, sizeof(*sysno));
}
- }
+ }
}
+
if (zebra_rec_keys_empty(zh->reg->keys))
{
/* the extraction process returned no information - the record
if (! *sysno)
{
- /* new record */
- if (delete_flag)
+ /* new record AKA does not exist already */
+ if (action == action_delete)
{
- yaz_log (YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
- pr_fname, (zint) start_offset);
- yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
+ yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
+ pr_fname, (zint) start_offset);
+ yaz_log(YLOG_WARN, "cannot delete record above (seems new)");
return ZEBRA_FAIL;
}
+ else if (action == action_a_delete)
+ {
+ if (show_progress)
+ yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
+ pr_fname, (zint) start_offset);
+ return ZEBRA_OK;
+ }
+ else if (action == action_replace)
+ {
+ yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
+ pr_fname, (zint) start_offset);
+ yaz_log(YLOG_WARN, "cannot update record above (seems new)");
+ return ZEBRA_FAIL;
+ }
if (show_progress)
- yaz_log (YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
+ yaz_log(YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
(zint) start_offset);
- rec = rec_new (zh->reg->records);
+ rec = rec_new(zh->reg->records);
*sysno = rec->sysno;
- recordAttr = rec_init_attr (zh->reg->zei, rec);
- recordAttr->staticrank = extractCtrl.staticrank;
+
+ if (stream)
+ {
+ all_matches_add(&extractCtrl,
+ zebra_rec_keys_get_custom_record_id(zh->reg->keys),
+ *sysno);
+ }
+
+
+ recordAttr = rec_init_attr(zh->reg->zei, rec);
+ if (extractCtrl.staticrank < 0)
+ {
+ yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0");
+ extractCtrl.staticrank = 0;
+ }
if (matchStr)
{
sizeof(*sysno), sysno);
}
-
- extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
- extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
- recordAttr->staticrank);
+ extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
+#if FLUSH2
+ extract_flush_record_keys2(zh, *sysno,
+ zh->reg->keys, extractCtrl.staticrank,
+ 0, recordAttr->staticrank);
+#else
+ extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
+ extractCtrl.staticrank);
+#endif
+ recordAttr->staticrank = extractCtrl.staticrank;
zh->records_inserted++;
}
else
/* record already exists */
zebra_rec_keys_t delkeys = zebra_rec_keys_open();
zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
- if (!allow_update)
+ if (action == action_insert)
{
- yaz_log (YLOG_LOG, "skipped %s %s " ZINT_FORMAT,
+ yaz_log(YLOG_LOG, "skipped %s %s " ZINT_FORMAT,
recordType, pr_fname, (zint) start_offset);
logRecord(zh);
return ZEBRA_FAIL;
}
- rec = rec_get (zh->reg->records, *sysno);
- assert (rec);
+ rec = rec_get(zh->reg->records, *sysno);
+ assert(rec);
+
+ if (stream)
+ {
+ all_matches_add(&extractCtrl,
+ zebra_rec_keys_get_custom_record_id(zh->reg->keys),
+ *sysno);
+ }
- recordAttr = rec_init_attr (zh->reg->zei, rec);
+ recordAttr = rec_init_attr(zh->reg->zei, rec);
+
+ /* decrease total size */
+ zebraExplain_recordBytesIncrement(zh->reg->zei,
+ - recordAttr->recordSize);
zebra_rec_keys_set_buf(delkeys,
rec->info[recInfo_delKeys],
rec->size[recInfo_sortKeys],
0);
- extract_flushSortKeys (zh, *sysno, 0, sortKeys);
- extract_flushRecordKeys (zh, *sysno, 0, delkeys,
- recordAttr->staticrank);
- if (delete_flag)
+ extract_flush_sort_keys(zh, *sysno, 0, sortKeys);
+#if !FLUSH2
+ extract_flush_record_keys(zh, *sysno, 0, delkeys,
+ recordAttr->staticrank);
+#endif
+ if (action == action_delete || action == action_a_delete)
{
/* record going to be deleted */
+#if FLUSH2
+ extract_flush_record_keys2(zh, *sysno, 0, recordAttr->staticrank,
+ delkeys, recordAttr->staticrank);
+#endif
if (zebra_rec_keys_empty(delkeys))
{
yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
}
- rec_del (zh->reg->records, &rec);
+ rec_del(zh->reg->records, &rec);
}
+ zebra_rec_keys_close(delkeys);
+ zebra_rec_keys_close(sortKeys);
rec_free(&rec);
logRecord(zh);
return ZEBRA_OK;
}
else
- {
+ { /* update or special_update */
if (show_progress)
- yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
- pr_fname, (zint) ZINT_FORMAT);
+ yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
+ pr_fname, (zint) start_offset);
+ extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
+
+#if FLUSH2
+ extract_flush_record_keys2(zh, *sysno,
+ zh->reg->keys, extractCtrl.staticrank,
+ delkeys, recordAttr->staticrank);
+#else
+ extract_flush_record_keys(zh, *sysno, 1,
+ zh->reg->keys, extractCtrl.staticrank);
+#endif
recordAttr->staticrank = extractCtrl.staticrank;
- extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
- extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
- recordAttr->staticrank);
zh->records_updated++;
}
zebra_rec_keys_close(delkeys);
zebra_rec_keys_close(sortKeys);
}
/* update file type */
- xfree (rec->info[recInfo_fileType]);
+ xfree(rec->info[recInfo_fileType]);
rec->info[recInfo_fileType] =
- rec_strdup (recordType, &rec->size[recInfo_fileType]);
+ rec_strdup(recordType, &rec->size[recInfo_fileType]);
/* update filename */
- xfree (rec->info[recInfo_filename]);
+ xfree(rec->info[recInfo_filename]);
rec->info[recInfo_filename] =
- rec_strdup (fname, &rec->size[recInfo_filename]);
+ rec_strdup(fname, &rec->size[recInfo_filename]);
/* update delete keys */
- xfree (rec->info[recInfo_delKeys]);
+ xfree(rec->info[recInfo_delKeys]);
if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
{
zebra_rec_keys_get_buf(zh->reg->keys,
rec->size[recInfo_delKeys] = 0;
}
/* update sort keys */
- xfree (rec->info[recInfo_sortKeys]);
+ xfree(rec->info[recInfo_sortKeys]);
zebra_rec_keys_get_buf(zh->reg->sortKeys,
&rec->info[recInfo_sortKeys],
&rec->size[recInfo_sortKeys]);
- /* save file size of original record */
- zebraExplain_recordBytesIncrement (zh->reg->zei,
- - recordAttr->recordSize);
if (stream)
{
- off_t end_offset = stream->endf(stream, 0);
-
- if (!end_offset)
- end_offset = stream->tellf(stream);
- else
- stream->seekf(stream, end_offset);
-
recordAttr->recordSize = end_offset - start_offset;
zebraExplain_recordBytesIncrement(zh->reg->zei,
recordAttr->recordSize);
/* set run-number for this record */
recordAttr->runNumber =
- zebraExplain_runNumberIncrement (zh->reg->zei, 0);
+ zebraExplain_runNumberIncrement(zh->reg->zei, 0);
/* update store data */
- xfree (rec->info[recInfo_storeData]);
+ xfree(rec->info[recInfo_storeData]);
/* update store data */
if (zh->store_data_buf)
rec->size[recInfo_storeData] = zh->store_data_size;
rec->info[recInfo_storeData] = zh->store_data_buf;
zh->store_data_buf = 0;
+ recordAttr->recordSize = zh->store_data_size;
}
else if (zh->m_store_data)
{
rec->size[recInfo_storeData] = recordAttr->recordSize;
rec->info[recInfo_storeData] = (char *)
- xmalloc (recordAttr->recordSize);
+ xmalloc(recordAttr->recordSize);
stream->seekf(stream, start_offset);
stream->readf(stream, rec->info[recInfo_storeData],
recordAttr->recordSize);
rec->size[recInfo_storeData] = 0;
}
/* update database name */
- xfree (rec->info[recInfo_databaseName]);
+ xfree(rec->info[recInfo_databaseName]);
rec->info[recInfo_databaseName] =
- rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
+ rec_strdup(zh->basenames[0], &rec->size[recInfo_databaseName]);
/* update offset */
recordAttr->recordOffset = start_offset;
/* commit this record */
- rec_put (zh->reg->records, &rec);
+ rec_put(zh->reg->records, &rec);
logRecord(zh);
return ZEBRA_OK;
}
ZebraHandle zh = (ZebraHandle) handle;
struct recExtractCtrl extractCtrl;
- if (zebraExplain_curDatabase (zh->reg->zei,
+ if (zebraExplain_curDatabase(zh->reg->zei,
rec->info[recInfo_databaseName]))
{
abort();
- if (zebraExplain_newDatabase (zh->reg->zei,
+ if (zebraExplain_newDatabase(zh->reg->zei,
rec->info[recInfo_databaseName], 0))
- abort ();
+ abort();
}
zebra_rec_keys_reset(zh->reg->keys);
extractCtrl.flagShowRecords = 0;
extractCtrl.match_criteria[0] = '\0';
extractCtrl.staticrank = 0;
+ extractCtrl.action = action_update;
+
extractCtrl.handle = handle;
extractCtrl.first_record = 1;
zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
rec->size[recInfo_delKeys],
0);
- extract_flushRecordKeys (zh, rec->sysno, 0, delkeys, 0);
+#if FLUSH2
+ extract_flush_record_keys2(zh, rec->sysno,
+ zh->reg->keys, 0, delkeys, 0);
+#else
+ extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0);
+ extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
+#endif
zebra_rec_keys_close(delkeys);
zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
rec->size[recInfo_sortKeys],
0);
- extract_flushSortKeys (zh, rec->sysno, 0, sortkeys);
+ extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys);
zebra_rec_keys_close(sortkeys);
}
- extract_flushRecordKeys (zh, rec->sysno, 1, zh->reg->keys, 0);
- extract_flushSortKeys (zh, rec->sysno, 1, zh->reg->sortKeys);
-
- xfree (rec->info[recInfo_delKeys]);
+ else
+ {
+#if FLUSH2
+ extract_flush_record_keys2(zh, rec->sysno, zh->reg->keys, 0, 0, 0);
+#else
+ extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
+#endif
+ }
+ extract_flush_sort_keys(zh, rec->sysno, 1, zh->reg->sortKeys);
+
+ xfree(rec->info[recInfo_delKeys]);
zebra_rec_keys_get_buf(zh->reg->keys,
&rec->info[recInfo_delKeys],
&rec->size[recInfo_delKeys]);
- xfree (rec->info[recInfo_sortKeys]);
+ xfree(rec->info[recInfo_sortKeys]);
zebra_rec_keys_get_buf(zh->reg->sortKeys,
&rec->info[recInfo_sortKeys],
&rec->size[recInfo_sortKeys]);
return ZEBRA_OK;
}
+void extract_rec_keys_log(ZebraHandle zh, int is_insert,
+ zebra_rec_keys_t reckeys,
+ int level)
+{
+ if (zebra_rec_keys_rewind(reckeys))
+ {
+ size_t slen;
+ const char *str;
+ struct it_key key;
+ NMEM nmem = nmem_create();
+
+ while(zebra_rec_keys_read(reckeys, &str, &slen, &key))
+ {
+ char keystr[200]; /* room for zints to print */
+ char *dst_term = 0;
+ int ord = CAST_ZINT_TO_INT(key.mem[0]);
+ const char *index_type;
+ int i;
+ const char *string_index;
+
+ zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
+ 0/* db */, &string_index);
+ assert(index_type);
+ zebra_term_untrans_iconv(zh, nmem, index_type,
+ &dst_term, str);
+ *keystr = '\0';
+ for (i = 0; i<key.len; i++)
+ {
+ sprintf(keystr + strlen(keystr), ZINT_FORMAT " ", key.mem[i]);
+ }
+
+ if (*str < CHR_BASE_CHAR)
+ {
+ int i;
+ char dst_buf[200]; /* room for special chars */
+
+ strcpy(dst_buf , "?");
+
+ if (!strcmp(str, ""))
+ strcpy(dst_buf, "alwaysmatches");
+ if (!strcmp(str, FIRST_IN_FIELD_STR))
+ strcpy(dst_buf, "firstinfield");
+ else if (!strcmp(str, CHR_UNKNOWN))
+ strcpy(dst_buf, "unknown");
+ else if (!strcmp(str, CHR_SPACE))
+ strcpy(dst_buf, "space");
+
+ for (i = 0; i<slen; i++)
+ {
+ sprintf(dst_buf + strlen(dst_buf), " %d", str[i] & 0xff);
+ }
+ yaz_log(level, "%s%s %s %s", keystr, index_type,
+ string_index, dst_buf);
+
+ }
+ else
+ yaz_log(level, "%s%s %s \"%s\"", keystr, index_type,
+ string_index, dst_term);
+
+ nmem_reset(nmem);
+ }
+ nmem_destroy(nmem);
+ }
+}
+
void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
zebra_rec_keys_t reckeys)
{
}
}
-void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
- int cmd,
- zebra_rec_keys_t reckeys,
- zint staticrank)
+void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
+ zebra_rec_keys_t ins_keys, zint ins_rank,
+ zebra_rec_keys_t del_keys, zint del_rank)
{
ZebraExplainInfo zei = zh->reg->zei;
+ int normal = 0;
+ int optimized = 0;
- extract_rec_keys_adjust(zh, cmd, reckeys);
-
- if (!zh->reg->key_buf)
+ if (!zh->reg->key_block)
{
- int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
- if (mem <= 0)
- {
- yaz_log(YLOG_WARN, "Invalid memory setting, using default 8 MB");
- mem= 1024*1024*8;
- }
- /* FIXME: That "8" should be in a default settings include */
- /* not hard-coded here! -H */
- zh->reg->key_buf = (char**) xmalloc (mem);
- zh->reg->ptr_top = mem/sizeof(char*);
- zh->reg->ptr_i = 0;
- zh->reg->key_buf_used = 0;
- zh->reg->key_file_no = 0;
+ int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
+ const char *key_tmp_dir = res_get_def(zh->res, "keyTmpDir", ".");
+ int use_threads = atoi(res_get_def(zh->res, "threads", "1"));
+ zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
}
- zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
- if (zebra_rec_keys_rewind(reckeys))
+ if (ins_keys)
{
- size_t slen;
- const char *str;
- struct it_key key_in;
- while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
- {
- int ch = 0;
- int i, j = 0;
- struct it_key key_out;
-
- assert(key_in.len >= 2);
- assert(key_in.len <= IT_KEY_LEVEL_MAX);
-
- /* check for buffer overflow */
- if (zh->reg->key_buf_used + 1024 >
- (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
- extract_flushWriteKeys (zh, 0);
-
- ++(zh->reg->ptr_i);
- assert(zh->reg->ptr_i > 0);
- (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
- (char*)zh->reg->key_buf + zh->reg->key_buf_used;
-
- /* key_in.mem[0] ord/ch */
- /* key_in.mem[1] filter specified record ID */
-
- /* encode the ordinal value (field/use/attribute) .. */
- ch = CAST_ZINT_TO_INT(key_in.mem[0]);
- zh->reg->key_buf_used +=
- key_SU_encode(ch, (char*)zh->reg->key_buf +
- zh->reg->key_buf_used);
-
- /* copy the 0-terminated stuff from str to output */
- memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used, str, slen);
- zh->reg->key_buf_used += slen;
- ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = '\0';
-
- /* the delete/insert indicator */
- ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = cmd;
-
- if (zh->m_staticrank) /* rank config enabled ? */
- {
- if (staticrank < 0)
- {
- yaz_log(YLOG_WARN, "staticrank = %ld. Setting to 0",
- (long) staticrank);
- staticrank = 0;
- }
- key_out.mem[j++] = staticrank;
- }
-
- if (key_in.mem[1]) /* filter specified record ID */
- key_out.mem[j++] = key_in.mem[1];
- else
- key_out.mem[j++] = sysno;
- for (i = 2; i < key_in.len; i++)
- key_out.mem[j++] = key_in.mem[i];
- key_out.len = j;
-
- memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used,
- &key_out, sizeof(key_out));
- (zh->reg->key_buf_used) += sizeof(key_out);
- }
+ extract_rec_keys_adjust(zh, 1, ins_keys);
+ if (!del_keys)
+ zebraExplain_recordCountIncrement(zei, 1);
+ zebra_rec_keys_rewind(ins_keys);
}
-}
-
-void extract_flushWriteKeys (ZebraHandle zh, int final)
- /* optimizing: if final=1, and no files written yet */
- /* push the keys directly to merge, sidestepping the */
- /* temp file altogether. Speeds small updates */
-{
- FILE *outf;
- char out_fname[200];
- char *prevcp, *cp;
- struct encode_info encode_info;
- int ptr_i = zh->reg->ptr_i;
- int temp_policy;
-#if SORT_EXTRA
- int i;
-#endif
- if (!zh->reg->key_buf || ptr_i <= 0)
+ if (del_keys)
{
- yaz_log(log_level, " nothing to flush section=%d buf=%p i=%d",
- zh->reg->key_file_no, zh->reg->key_buf, ptr_i);
- return;
- }
-
- (zh->reg->key_file_no)++;
- yaz_log (YLOG_LOG, "sorting section %d", (zh->reg->key_file_no));
- yaz_log(log_level, " sort_buff at %p n=%d",
- zh->reg->key_buf + zh->reg->ptr_top - ptr_i,ptr_i);
-#if !SORT_EXTRA
- qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
- sizeof(char*), key_qsort_compare);
-
- /* zebra.cfg: tempfiles:
- Y: always use temp files (old way)
- A: use temp files, if more than one (auto)
- = if this is both the last and the first
- N: never bother with temp files (new) */
-
- temp_policy=toupper(res_get_def(zh->res,"tempfiles","auto")[0]);
- if (temp_policy != 'Y' && temp_policy != 'N' && temp_policy != 'A') {
- yaz_log (YLOG_WARN, "Illegal tempfiles setting '%c'. using 'Auto' ",
- temp_policy);
- temp_policy='A';
- }
-
- if ( ( temp_policy =='N' ) || /* always from memory */
- ( ( temp_policy =='A' ) && /* automatic */
- (zh->reg->key_file_no == 1) && /* this is first time */
- (final) ) ) /* and last (=only) time */
- { /* go directly from memory */
- zh->reg->key_file_no =0; /* signal not to read files */
- zebra_index_merge(zh);
- zh->reg->ptr_i = 0;
- zh->reg->key_buf_used = 0;
- return;
- }
-
- /* Not doing directly from memory, write into a temp file */
- extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no);
-
- if (!(outf = fopen (out_fname, "wb")))
- {
- yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
- exit (1);
- }
- yaz_log (YLOG_LOG, "writing section %d", zh->reg->key_file_no);
- prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
-
- encode_key_init (&encode_info);
- encode_key_write (cp, &encode_info, outf);
-
- while (--ptr_i > 0)
- {
- cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
- if (strcmp (cp, prevcp))
- {
- encode_key_flush ( &encode_info, outf);
- encode_key_init (&encode_info);
- encode_key_write (cp, &encode_info, outf);
- prevcp = cp;
- }
- else
- encode_key_write (cp + strlen(cp), &encode_info, outf);
+ extract_rec_keys_adjust(zh, 0, del_keys);
+ if (!ins_keys)
+ zebraExplain_recordCountIncrement(zei, -1);
+ zebra_rec_keys_rewind(del_keys);
}
- encode_key_flush ( &encode_info, outf);
-#else
- qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
- extract_get_fname_tmp (out_fname, key_file_no);
- if (!(outf = fopen (out_fname, "wb")))
- {
- yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
- exit (1);
- }
- yaz_log (YLOG_LOG, "writing section %d", key_file_no);
- i = ptr_i;
- prevcp = key_buf[ptr_top-i];
while (1)
- if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
+ {
+ size_t del_slen;
+ const char *del_str;
+ struct it_key del_key_in;
+ int del = 0;
+
+ size_t ins_slen;
+ const char *ins_str;
+ struct it_key ins_key_in;
+ int ins = 0;
+
+ if (del_keys)
+ del = zebra_rec_keys_read(del_keys, &del_str, &del_slen,
+ &del_key_in);
+ if (ins_keys)
+ ins = zebra_rec_keys_read(ins_keys, &ins_str, &ins_slen,
+ &ins_key_in);
+
+ if (del && ins && ins_rank == del_rank
+ && !key_compare(&del_key_in, &ins_key_in)
+ && ins_slen == del_slen && !memcmp(del_str, ins_str, del_slen))
{
- key_y_len = strlen(prevcp)+1;
-#if 0
- yaz_log (YLOG_LOG, "key_y_len: %2d %02x %02x %s",
- key_y_len, prevcp[0], prevcp[1], 2+prevcp);
-#endif
- qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
- sizeof(char*), key_y_compare);
- cp = key_buf[ptr_top-ptr_i];
- --key_y_len;
- encode_key_init (&encode_info);
- encode_key_write (cp, &encode_info, outf);
- while (--ptr_i > i)
- {
- cp = key_buf[ptr_top-ptr_i];
- encode_key_write (cp+key_y_len, &encode_info, outf);
- }
- encode_key_flush ( &encode_info, outf);
- if (!i)
- break;
- prevcp = key_buf[ptr_top-ptr_i];
+ optimized++;
+ continue;
}
-#endif
- if (fclose (outf))
- {
- yaz_log (YLOG_FATAL|YLOG_ERRNO, "fclose %s", out_fname);
- exit (1);
+ if (!del && !ins)
+ break;
+
+ normal++;
+ if (del)
+ key_block_write(zh->reg->key_block, sysno,
+ &del_key_in, 0, del_str, del_slen,
+ del_rank, zh->m_staticrank);
+ if (ins)
+ key_block_write(zh->reg->key_block, sysno,
+ &ins_key_in, 1, ins_str, ins_slen,
+ ins_rank, zh->m_staticrank);
}
- yaz_log (YLOG_LOG, "finished section %d", zh->reg->key_file_no);
- zh->reg->ptr_i = 0;
- zh->reg->key_buf_used = 0;
+ yaz_log(log_level_extract, "normal=%d optimized=%d", normal, optimized);
}
-ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh,
- zebra_rec_keys_t reckeys,
- zebra_snippets *snippets)
+
+ZEBRA_RES zebra_rec_keys_to_snippets1(ZebraHandle zh,
+ zebra_rec_keys_t reckeys,
+ zebra_snippets *snippets)
{
NMEM nmem = nmem_create();
if (zebra_rec_keys_rewind(reckeys))
struct it_key key;
while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
{
- char dst_buf[IT_MAX_WORD];
- char *dst_term = dst_buf;
+ char *dst_term = 0;
int ord;
zint seqno;
- int index_type;
+ const char *index_type;
assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
seqno = key.mem[key.len-1];
assert(index_type);
zebra_term_untrans_iconv(zh, nmem, index_type,
&dst_term, str);
- zebra_snippets_append(snippets, seqno, ord, dst_term);
+ zebra_snippets_append(snippets, seqno, 0, ord, dst_term);
nmem_reset(nmem);
}
}
{
char dst_buf[IT_MAX_WORD];
zint seqno;
- int index_type;
+ const char *index_type;
int ord = CAST_ZINT_TO_INT(key.mem[0]);
const char *db = 0;
assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
}
-static void extract_add_string(RecWord *p, const char *string, int length)
+static void extract_add_staticrank_string(RecWord *p,
+ const char *str, int length)
{
- ZebraHandle zh = p->extractCtrl->handle;
- assert (length > 0);
+ char valz[40];
+ struct recExtractCtrl *ctrl = p->extractCtrl;
+
+ if (length > sizeof(valz)-1)
+ length = sizeof(valz)-1;
+
+ memcpy(valz, str, length);
+ valz[length] = '\0';
+ ctrl->staticrank = atozint(valz);
+}
+
+static void extract_add_string(RecWord *p, zebra_map_t zm,
+ const char *string, int length)
+{
+ assert(length > 0);
if (!p->index_name)
return;
- if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
- extract_add_sort_string(p, string, length);
- else
+ if (zebra_maps_is_index(zm))
{
extract_add_index_string(p, zinfo_index_category_index,
string, length);
- if (zebra_maps_is_alwaysmatches(zh->reg->zebra_maps, p->index_type))
+ if (zebra_maps_is_alwaysmatches(zm))
{
RecWord word;
memcpy(&word, p, sizeof(word));
&word, zinfo_index_category_alwaysmatches, "", 0);
}
}
+ else if (zebra_maps_is_sort(zm))
+ {
+ extract_add_sort_string(p, string, length);
+ }
+ else if (zebra_maps_is_staticrank(zm))
+ {
+ extract_add_staticrank_string(p, string, length);
+ }
}
-static void extract_add_incomplete_field(RecWord *p)
+static void extract_add_incomplete_field(RecWord *p, zebra_map_t zm)
{
- ZebraHandle zh = p->extractCtrl->handle;
const char *b = p->term_buf;
int remain = p->term_len;
+ int first = 1;
const char **map = 0;
if (remain > 0)
- map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
+ map = zebra_maps_input(zm, &b, remain, 0);
- if (map)
- {
- if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
- {
- /* first in field marker */
- extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
- p->seqno++;
- }
- }
while (map)
{
char buf[IT_MAX_WORD+1];
{
remain = p->term_len - (b - p->term_buf);
if (remain > 0)
- map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
- remain, 0);
+ map = zebra_maps_input(zm, &b, remain, 0);
else
map = 0;
}
buf[i++] = *(cp++);
remain = p->term_len - (b - p->term_buf);
if (remain > 0)
- map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
+ map = zebra_maps_input(zm, &b, remain, 0);
else
map = 0;
}
if (!i)
return;
- extract_add_string (p, buf, i);
+
+ if (first)
+ {
+ first = 0;
+ if (zebra_maps_is_first_in_field(zm))
+ {
+ /* first in field marker */
+ extract_add_string(p, zm, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
+ p->seqno++;
+ }
+ }
+ extract_add_string(p, zm, buf, i);
p->seqno++;
}
}
-static void extract_add_complete_field (RecWord *p)
+static void extract_add_complete_field(RecWord *p, zebra_map_t zm)
{
- ZebraHandle zh = p->extractCtrl->handle;
const char *b = p->term_buf;
char buf[IT_MAX_WORD+1];
const char **map = 0;
int i = 0, remain = p->term_len;
if (remain > 0)
- map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1);
+ map = zebra_maps_input(zm, &b, remain, 1);
while (remain > 0 && i < IT_MAX_WORD)
{
if (remain > 0)
{
int first = i ? 0 : 1; /* first position */
- map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, first);
+ map = zebra_maps_input(zm, &b, remain, first);
}
else
map = 0;
remain = p->term_len - (b - p->term_buf);
if (remain > 0)
{
- map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b,
- remain, 0);
+ map = zebra_maps_input(zm, &b, remain, 0);
}
else
map = 0;
}
if (!i)
return;
- extract_add_string (p, buf, i);
+ extract_add_string(p, zm, buf, i);
}
+static void extract_add_icu(RecWord *p, zebra_map_t zm)
+{
+ const char *res_buf = 0;
+ size_t res_len = 0;
+
+ zebra_map_tokenize_start(zm, p->term_buf, p->term_len);
+ while (zebra_map_tokenize_next(zm, &res_buf, &res_len, 0, 0))
+ {
+ extract_add_string(p, zm, res_buf, res_len);
+ p->seqno++;
+ }
+}
+
+
+/** \brief top-level indexing handler for recctrl system
+ \param p token data to be indexed
+
+ Call sequence:
+ extract_token_add
+ extract_add_{in}_complete / extract_add_icu
+ extract_add_string
+
+ extract_add_index_string
+ or
+ extract_add_sort_string
+ or
+ extract_add_staticrank_string
+
+*/
static void extract_token_add(RecWord *p)
{
ZebraHandle zh = p->extractCtrl->handle;
+ zebra_map_t zm = zebra_map_get_or_add(zh->reg->zebra_maps, p->index_type);
WRBUF wrbuf;
- if (log_level)
+ if (log_level_details)
{
- yaz_log(log_level, "extract_token_add "
- "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s",
+ yaz_log(log_level_details, "extract_token_add "
+ "type=%s index=%s seqno=" ZINT_FORMAT " s=%.*s",
p->index_type, p->index_name,
p->seqno, p->term_len, p->term_buf);
}
- if ((wrbuf = zebra_replace(zh->reg->zebra_maps, p->index_type, 0,
- p->term_buf, p->term_len)))
+ if ((wrbuf = zebra_replace(zm, 0, p->term_buf, p->term_len)))
+ {
+ p->term_buf = wrbuf_buf(wrbuf);
+ p->term_len = wrbuf_len(wrbuf);
+ }
+ if (zebra_maps_is_icu(zm))
{
- p->term_buf = wrbuf_buf(wrbuf);
- p->term_len = wrbuf_len(wrbuf);
+ extract_add_icu(p, zm);
}
- if (zebra_maps_is_complete (zh->reg->zebra_maps, p->index_type))
- extract_add_complete_field (p);
else
- extract_add_incomplete_field(p);
+ {
+ if (zebra_maps_is_complete(zm))
+ extract_add_complete_field(p, zm);
+ else
+ extract_add_incomplete_field(p, zm);
+ }
}
static void extract_set_store_data_cb(struct recExtractCtrl *p,
p->setStoreData = extract_set_store_data_cb;
}
-static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
+static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid)
{
ZebraHandle zh = (ZebraHandle) p->handle;
- zebraExplain_addSchema (zh->reg->zei, oid);
+ zebraExplain_addSchema(zh->reg->zei, oid);
}
-void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
- int cmd, zebra_rec_keys_t reckeys)
+void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
+ int cmd, zebra_rec_keys_t reckeys)
{
+#if 0
+ yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT,
+ cmd, sysno);
+ extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG);
+#endif
+
if (zebra_rec_keys_rewind(reckeys))
{
- SortIdx sortIdx = zh->reg->sortIdx;
+ zebra_sort_index_t si = zh->reg->sort_index;
size_t slen;
const char *str;
struct it_key key_in;
- sortIdx_sysno (sortIdx, sysno);
+ zebra_sort_sysno(si, sysno);
while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
{
int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
- sortIdx_type(sortIdx, ord);
+ zebra_sort_type(si, ord);
if (cmd == 1)
- sortIdx_add(sortIdx, str, slen);
+ zebra_sort_add(si, str, slen);
else
- sortIdx_add(sortIdx, "", 1);
+ zebra_sort_delete(si);
}
}
}
-static void encode_key_init(struct encode_info *i)
-{
- i->encode_handle = iscz1_start();
- i->decode_handle = iscz1_start();
-}
-
-static void encode_key_write (char *k, struct encode_info *i, FILE *outf)
-{
- struct it_key key;
- char *bp = i->buf, *bp0;
- const char *src = (char *) &key;
-
- /* copy term to output buf */
- while ((*bp++ = *k++))
- ;
- /* and copy & align key so we can mangle */
- memcpy (&key, k+1, sizeof(struct it_key)); /* *k is insert/delete */
-
-#if 0
- /* debugging */
- key_logdump_txt(YLOG_LOG, &key, *k ? "i" : "d");
-#endif
- assert(key.mem[0] >= 0);
-
- bp0 = bp++;
- iscz1_encode(i->encode_handle, &bp, &src);
-
- *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */
- if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
- {
- yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
- exit (1);
- }
-
-#if 0
- /* debugging */
- if (1)
- {
- struct it_key key2;
- const char *src = bp0+1;
- char *dst = (char*) &key2;
- iscz1_decode(i->decode_handle, &dst, &src);
-
- key_logdump_txt(YLOG_LOG, &key2, *k ? "i" : "d");
-
- assert(key2.mem[1]);
- }
-#endif
-}
-
-static void encode_key_flush (struct encode_info *i, FILE *outf)
-{
- iscz1_stop(i->encode_handle);
- iscz1_stop(i->decode_handle);
-}
-
/*
* Local variables:
* c-basic-offset: 4
projects / idzebra-moved-to-github.git / blobdiff