-/*
- * Copyright (C) 1994-1995, Index Data I/S
- * All rights reserved.
- * Sebastian Hammer, Adam Dickmeiss
- *
- * $Log: extract.c,v $
- * Revision 1.27 1995-11-20 16:59:45 adam
- * New update method: the 'old' keys are saved for each records.
- *
- * Revision 1.26 1995/11/20 11:56:24 adam
- * Work on new traversal.
- *
- * Revision 1.25 1995/11/16 15:34:54 adam
- * Uses new record management system in both indexer and server.
- *
- * Revision 1.24 1995/11/15 19:13:08 adam
- * Work on record management.
- *
- * Revision 1.23 1995/10/27 14:00:10 adam
- * Implemented detection of database availability.
- *
- * Revision 1.22 1995/10/17 18:02:07 adam
- * New feature: databases. Implemented as prefix to words in dictionary.
- *
- * Revision 1.21 1995/10/10 12:24:38 adam
- * Temporary sort files are compressed.
- *
- * Revision 1.20 1995/10/06 13:52:05 adam
- * Bug fixes. Handler may abort further scanning.
- *
- * Revision 1.19 1995/10/04 12:55:16 adam
- * Bug fix in ranked search. Use=Any keys inserted.
- *
- * Revision 1.18 1995/10/04 09:37:08 quinn
- * Fixed bug.
- *
- * Revision 1.17 1995/10/03 14:28:57 adam
- * Buffered read in extract works.
- *
- * Revision 1.16 1995/10/03 14:28:45 adam
- * Work on more effecient read handler in extract.
- *
- * Revision 1.15 1995/10/02 15:42:53 adam
- * Extract uses file descriptors instead of FILE pointers.
- *
- * Revision 1.14 1995/10/02 15:29:13 adam
- * More logging in file_extract.
- *
- * Revision 1.13 1995/09/29 14:01:39 adam
- * Bug fixes.
- *
- * Revision 1.12 1995/09/28 14:22:56 adam
- * Sort uses smaller temporary files.
- *
- * Revision 1.11 1995/09/28 12:10:31 adam
- * Bug fixes. Field prefix used in queries.
- *
- * Revision 1.10 1995/09/28 09:19:41 adam
- * xfree/xmalloc used everywhere.
- * Extract/retrieve method seems to work for text records.
- *
- * Revision 1.9 1995/09/27 12:22:28 adam
- * More work on extract in record control.
- * Field name is not in isam keys but in prefix in dictionary words.
- *
- * Revision 1.8 1995/09/14 07:48:22 adam
- * Record control management.
- *
- * Revision 1.7 1995/09/11 13:09:32 adam
- * More work on relevance feedback.
- *
- * Revision 1.6 1995/09/08 14:52:27 adam
- * Minor changes. Dictionary is lower case now.
- *
- * Revision 1.5 1995/09/06 16:11:16 adam
- * Option: only one word key per file.
- *
- * Revision 1.4 1995/09/05 15:28:39 adam
- * More work on search engine.
- *
- * Revision 1.3 1995/09/04 12:33:41 adam
- * Various cleanup. YAZ util used instead.
- *
- * Revision 1.2 1995/09/04 09:10:34 adam
- * More work on index add/del/update.
- * Merge sort implemented.
- * Initial work on z39 server.
- *
- * Revision 1.1 1995/09/01 14:06:35 adam
- * Split of work into more files.
- *
- */
+/* $Id: extract.c,v 1.229 2006-09-08 14:40:52 adam Exp $
+ Copyright (C) 1995-2006
+ Index Data ApS
+
+This file is part of the Zebra server.
+
+Zebra is free software; you can redistribute it and/or modify it under
+the terms of the GNU General Public License as published by the Free
+Software Foundation; either version 2, or (at your option) any later
+version.
+
+Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
+WARRANTY; without even the implied warranty of MERCHANTABILITY or
+FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
+for more details.
+
+You should have received a copy of the GNU General Public License
+along with this program; if not, write to the Free Software
+Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
+
+*/
+
#include <stdio.h>
#include <assert.h>
+#include <ctype.h>
+#ifdef WIN32
+#include <io.h>
+#endif
+#if HAVE_UNISTD_H
#include <unistd.h>
+#endif
#include <fcntl.h>
-#include <ctype.h>
-#include <alexutil.h>
-#include <recctrl.h>
#include "index.h"
+#include "orddict.h"
+#include <direntz.h>
+#include <charmap.h>
-#if RECORD_BASE
-#include "recindex.h"
-#endif
+#define ENCODE_BUFLEN 768
+struct encode_info {
+ void *encode_handle;
+ void *decode_handle;
+ char buf[ENCODE_BUFLEN];
+};
-static Dict file_idx;
+static int log_level = 0;
+static int log_level_initialized = 0;
+static void zebra_init_log_level()
+{
+ if (!log_level_initialized)
+ {
+ log_level = yaz_log_module_level("extract");
+ log_level_initialized = 1;
+ }
+}
-#if RECORD_BASE
-static Records records = NULL;
-#else
-static int sys_idx_fd = -1;
-static SYSNO sysno_next;
-#endif
+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 int key_cmd;
-static int key_sysno;
-static const char *key_databaseName;
-static char **key_buf;
-static size_t ptr_top;
-static size_t ptr_i;
-static size_t key_buf_used;
-static int key_file_no;
+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 int key_del_max;
-static int key_del_used;
-static char *key_del_buf;
+#define USE_SHELLSORT 0
-void key_open (int mem)
+#if USE_SHELLSORT
+static void shellsort(void *ar, int r, size_t s,
+ int (*cmp)(const void *a, const void *b))
{
-#if !RECORD_BASE
- void *file_key;
+ 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);
+ }
+}
#endif
- if (mem < 50000)
- mem = 50000;
- key_buf = xmalloc (mem);
- ptr_top = mem/sizeof(char*);
- ptr_i = 0;
- key_buf_used = 0;
- key_file_no = 0;
+static void logRecord (ZebraHandle zh)
+{
+ ++zh->records_processed;
+ if (!(zh->records_processed % 1000))
+ {
+ yaz_log(YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
+ ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT,
+ zh->records_processed, zh->records_inserted,
+ zh->records_updated, zh->records_deleted);
+ }
+}
+
+static void extract_add_index_string (RecWord *p,
+ zinfo_index_category_t cat,
+ const char *str, int length);
+
+static void extract_set_store_data_prepare(struct recExtractCtrl *p);
+
+static void extract_init(struct recExtractCtrl *p, RecWord *w)
+{
+ w->seqno = 1;
+ w->index_name = "any";
+ w->index_type = 'w';
+ w->extractCtrl = p;
+ w->record_id = 0;
+ w->section_id = 0;
+ w->segment = 0;
+}
+
+static void searchRecordKey(ZebraHandle zh,
+ zebra_rec_keys_t reckeys,
+ const char *index_name,
+ const char **ws, int ws_length)
+{
+ int i;
+ int ch = -1;
+ zinfo_index_category_t cat = zinfo_index_category_index;
+
+ for (i = 0; i<ws_length; i++)
+ ws[i] = NULL;
- key_del_buf = NULL;
- key_del_max = 0;
+ if (ch < 0)
+ 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);
+ if (ch < 0)
+ ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'w', index_name);
- if (!(file_idx = dict_open (FNAME_FILE_DICT, 40, 1)))
+ if (ch < 0)
+ return ;
+
+ if (zebra_rec_keys_rewind(reckeys))
{
- logf (LOG_FATAL, "dict_open fail of %s", "fileidx");
- exit (1);
+ zint startSeq = -1;
+ const char *str;
+ size_t slen;
+ struct it_key key;
+ zint seqno;
+ while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
+ {
+ assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
+
+ seqno = key.mem[key.len-1];
+
+ if (key.mem[0] == ch)
+ {
+ zint woff;
+
+ if (startSeq == -1)
+ startSeq = seqno;
+ woff = seqno - startSeq;
+ if (woff >= 0 && woff < ws_length)
+ ws[woff] = str;
+ }
+ }
}
-#if RECORD_BASE
- assert (!records);
- records = rec_open (1);
-#else
- file_key = dict_lookup (file_idx, ".");
- if (file_key)
- memcpy (&sysno_next, (char*)file_key+1, sizeof(sysno_next));
- else
- sysno_next = 1;
- if ((sys_idx_fd = open (FNAME_SYS_IDX, O_RDWR|O_CREAT, 0666)) == -1)
+}
+
+#define FILE_MATCH_BLANK "\t "
+
+static char *fileMatchStr (ZebraHandle zh,
+ zebra_rec_keys_t reckeys,
+ const char *fname, const char *spec)
+{
+ static char dstBuf[2048]; /* static here ??? */
+ char *dst = dstBuf;
+ const char *s = spec;
+
+ while (1)
{
- logf (LOG_FATAL|LOG_ERRNO, "open %s", FNAME_SYS_IDX);
- exit (1);
+ for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
+ ;
+ if (!*s)
+ break;
+ if (*s == '(')
+ {
+ const char *ws[32];
+ char attset_str[64], attname_str[64];
+ int i;
+ int first = 1;
+
+ for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
+ ;
+ for (i = 0; *s && *s != ',' && *s != ')' &&
+ !strchr(FILE_MATCH_BLANK, *s); s++)
+ if (i+1 < sizeof(attset_str))
+ attset_str[i++] = *s;
+ attset_str[i] = '\0';
+
+ for (; strchr(FILE_MATCH_BLANK, *s); s++)
+ ;
+ if (*s != ',')
+ strcpy(attname_str, attset_str);
+ else
+ {
+ for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
+ ;
+ for (i = 0; *s && *s != ')' &&
+ !strchr(FILE_MATCH_BLANK, *s); s++)
+ if (i+1 < sizeof(attname_str))
+ attname_str[i++] = *s;
+ attname_str[i] = '\0';
+ }
+
+ searchRecordKey (zh, reckeys, attname_str, ws, 32);
+
+ if (*s != ')')
+ {
+ yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
+ spec, zh->m_group ? zh->m_group : "none");
+ return NULL;
+ }
+ s++;
+
+ for (i = 0; i<32; i++)
+ if (ws[i])
+ {
+ if (first)
+ {
+ *dst++ = ' ';
+ first = 0;
+ }
+ strcpy (dst, ws[i]);
+ dst += strlen(ws[i]);
+ }
+ if (first)
+ {
+ yaz_log (YLOG_WARN, "Record didn't contain match"
+ " fields in (%s,%s)", attset_str, attname_str);
+ return NULL;
+ }
+ }
+ else if (*s == '$')
+ {
+ int spec_len;
+ char special[64];
+ const char *spec_src = NULL;
+ const char *s1 = ++s;
+ while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
+ s1++;
+
+ spec_len = s1 - s;
+ if (spec_len > sizeof(special)-1)
+ spec_len = sizeof(special)-1;
+ memcpy (special, s, spec_len);
+ special[spec_len] = '\0';
+ s = s1;
+
+ if (!strcmp (special, "group"))
+ spec_src = zh->m_group;
+ else if (!strcmp (special, "database"))
+ spec_src = zh->basenames[0];
+ else if (!strcmp (special, "filename")) {
+ spec_src = fname;
+ }
+ 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);
+ }
+ }
+ else if (*s == '\"' || *s == '\'')
+ {
+ int stopMarker = *s++;
+ char tmpString[64];
+ int i = 0;
+
+ while (*s && *s != stopMarker)
+ {
+ if (i+1 < sizeof(tmpString))
+ tmpString[i++] = *s++;
+ }
+ if (*s)
+ s++;
+ tmpString[i] = '\0';
+ strcpy (dst, tmpString);
+ dst += strlen (tmpString);
+ }
+ else
+ {
+ yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
+ spec, zh->m_group ? zh->m_group : "none");
+ return NULL;
+ }
+ *dst++ = 1;
}
-#endif
+ if (dst == dstBuf)
+ {
+ yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
+ fname, zh->m_group ? zh->m_group : "none");
+ return NULL;
+ }
+ *dst = '\0';
+ return dstBuf;
}
-struct encode_info {
- int sysno;
- int seqno;
- char buf[512];
+struct recordLogInfo {
+ const char *fname;
+ int recordOffset;
+ struct recordGroup *rGroup;
};
-void encode_key_init (struct encode_info *i)
+static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
{
- i->sysno = 0;
- i->seqno = 0;
+ 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)
+{
+ RecWord word;
+ extract_init(ctrl, &word);
+ word.index_name = "_ALLRECORDS";
+ word.index_type = 'w';
+ word.seqno = 1;
+ extract_add_index_string (&word, zinfo_index_category_alwaysmatches,
+ "", 0);
}
-char *encode_key_int (int d, char *bp)
+ZEBRA_RES zebra_extract_file(ZebraHandle zh, SYSNO *sysno, const char *fname,
+ int deleteFlag)
{
- if (d <= 63)
- *bp++ = d;
- else if (d <= 16383)
+ ZEBRA_RES r = ZEBRA_OK;
+ int i, fd;
+ char gprefix[128];
+ char ext[128];
+ char ext_res[128];
+ struct file_read_info *fi = 0;
+ const char *original_record_type = 0;
+ RecType recType;
+ void *recTypeClientData;
+ struct ZebraRecStream stream, *streamp;
+
+ zebra_init_log_level();
+
+ if (!zh->m_group || !*zh->m_group)
+ *gprefix = '\0';
+ else
+ sprintf (gprefix, "%s.", zh->m_group);
+
+ yaz_log(log_level, "zebra_extract_file %s", fname);
+
+ /* determine file extension */
+ *ext = '\0';
+ for (i = strlen(fname); --i >= 0; )
+ if (fname[i] == '/')
+ break;
+ else if (fname[i] == '.')
+ {
+ 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);
+ }
+ if (!zh->m_record_type)
{
- *bp++ = 64 + (d>>8);
- *bp++ = d & 255;
+ if (zh->records_processed < zh->m_file_verbose_limit)
+ yaz_log (YLOG_LOG, "? %s", fname);
+ 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);
+ }
+
+ if (!(recType =
+ 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);
+ return ZEBRA_FAIL;
+ }
+
+ switch(recType->version)
+ {
+ case 0:
+ break;
+ default:
+ yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
}
- else if (d <= 4194303)
+ if (sysno && deleteFlag)
{
- *bp++ = 128 + (d>>16);
- *bp++ = (d>>8) & 255;
- *bp++ = d & 255;
+ streamp = 0;
+ fi = 0;
}
else
{
- *bp++ = 192 + (d>>24);
- *bp++ = (d>>16) & 255;
- *bp++ = (d>>8) & 255;
- *bp++ = d & 255;
+ char full_rep[1024];
+
+ if (zh->path_reg && !yaz_is_abspath (fname))
+ {
+ strcpy (full_rep, zh->path_reg);
+ strcat (full_rep, "/");
+ strcat (full_rep, fname);
+ }
+ else
+ strcpy (full_rep, fname);
+
+ if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
+ {
+ 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;
+ }
}
- return bp;
+ if (streamp)
+ stream.destroy(streamp);
+ zh->m_record_type = original_record_type;
+ return r;
}
-void encode_key_write (char *k, struct encode_info *i, FILE *outf)
+/*
+ If sysno is provided, then it's used to identify the reocord.
+ If not, and match_criteria is provided, then sysno is guessed
+ If not, and a record is provided, then sysno is got from there
+
+ */
+
+ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh,
+ const char *buf, size_t buf_size,
+ int delete_flag,
+ int test_mode,
+ const char *recordType,
+ SYSNO *sysno,
+ const char *match_criteria,
+ const char *fname,
+ int force_update,
+ int allow_update)
{
- struct it_key key;
- char *bp = i->buf;
+ struct ZebraRecStream stream;
+ ZEBRA_RES res;
+ void *clientData;
+ RecType recType = 0;
- while ((*bp++ = *k++))
- ;
- memcpy (&key, k+1, sizeof(struct it_key));
- bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
- if (i->sysno != key.sysno)
+ if (recordType && *recordType)
{
- i->sysno = key.sysno;
- i->seqno = 0;
+ yaz_log(log_level, "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");
+ return ZEBRA_FAIL;
+ }
+ yaz_log(log_level, "Get record type from rgroup: %s",
+ zh->m_record_type);
+ recType = recType_byName (zh->reg->recTypes, zh->res,
+ zh->m_record_type, &clientData);
+ recordType = zh->m_record_type;
}
- bp = encode_key_int (key.seqno - i->seqno, bp);
- i->seqno = key.seqno;
- if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
+
+ if (!recType)
{
- logf (LOG_FATAL|LOG_ERRNO, "fwrite");
- exit (1);
+ 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);
+ stream.destroy(&stream);
+ return res;
}
-void key_flush (void)
+
+ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
+ struct ZebraRecStream *stream,
+ int delete_flag,
+ int test_mode,
+ const char *recordType,
+ SYSNO *sysno,
+ const char *match_criteria,
+ const char *fname,
+ int force_update,
+ int allow_update,
+ RecType recType,
+ void *recTypeClientData)
+
{
- FILE *outf;
- char out_fname[200];
- char *prevcp, *cp;
- struct encode_info encode_info;
-
- if (ptr_i <= 0)
- return;
+ SYSNO sysno0 = 0;
+ RecordAttr *recordAttr;
+ struct recExtractCtrl extractCtrl;
+ int r;
+ const char *matchStr = 0;
+ Record rec;
+ off_t start_offset = 0;
+ const char *pr_fname = fname; /* filename to print .. */
+ int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
- key_file_no++;
- logf (LOG_LOG, "sorting section %d", key_file_no);
- qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
- sprintf (out_fname, TEMP_FNAME, key_file_no);
+ zebra_init_log_level();
- if (!(outf = fopen (out_fname, "w")))
+ if (!pr_fname)
+ pr_fname = "<no file>"; /* make it printable if file is omitted */
+
+ zebra_rec_keys_reset(zh->reg->keys);
+ zebra_rec_keys_reset(zh->reg->sortKeys);
+
+ if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
{
- logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname);
- exit (1);
+ if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
+ zh->m_explain_database))
+ return ZEBRA_FAIL;
}
- logf (LOG_LOG, "writing section %d", key_file_no);
- prevcp = cp = key_buf[ptr_top-ptr_i];
-
- encode_key_init (&encode_info);
- encode_key_write (cp, &encode_info, outf);
- while (--ptr_i > 0)
+
+ if (stream)
{
- cp = key_buf[ptr_top-ptr_i];
- if (strcmp (cp, prevcp))
+ off_t null_offset = 0;
+ extractCtrl.stream = stream;
+
+ start_offset = stream->tellf(stream);
+
+ extractCtrl.first_record = start_offset ? 0 : 1;
+
+ stream->endf(stream, &null_offset);;
+
+ extractCtrl.init = extract_init;
+ extractCtrl.tokenAdd = extract_token_add;
+ extractCtrl.schemaAdd = extract_schema_add;
+ extractCtrl.dh = zh->reg->dh;
+ extractCtrl.handle = zh;
+ extractCtrl.match_criteria[0] = '\0';
+ extractCtrl.staticrank = 0;
+
+
+ init_extractCtrl(zh, &extractCtrl);
+
+ extract_set_store_data_prepare(&extractCtrl);
+
+ r = (*recType->extract)(recTypeClientData, &extractCtrl);
+
+ if (r == RECCTRL_EXTRACT_EOF)
+ return ZEBRA_FAIL;
+ else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
{
- encode_key_init (&encode_info);
- encode_key_write (cp, &encode_info, outf);
- prevcp = cp;
+ /* error occured during extraction ... */
+ yaz_log (YLOG_WARN, "extract error: generic");
+ return ZEBRA_FAIL;
}
- else
- encode_key_write (cp + strlen(cp), &encode_info, outf);
+ else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
+ {
+ /* error occured during extraction ... */
+ yaz_log (YLOG_WARN, "extract error: no such filter");
+ return ZEBRA_FAIL;
+ }
+
+ all_matches_add(&extractCtrl);
+
+ if (extractCtrl.match_criteria[0])
+ match_criteria = extractCtrl.match_criteria;
}
- if (fclose (outf))
+ 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);
+ if (!matchStr)
+ {
+ yaz_log (YLOG_WARN, "Bad match criteria (recordID)");
+ return ZEBRA_FAIL;
+ }
+ }
+ }
+ if (matchStr)
+ {
+ int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
+ char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
+ matchStr);
+ if (rinfo)
+ {
+ assert(*rinfo == sizeof(*sysno));
+ memcpy (sysno, rinfo+1, sizeof(*sysno));
+ }
+ }
+ }
+ if (zebra_rec_keys_empty(zh->reg->keys))
{
- logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
- exit (1);
+ /* the extraction process returned no information - the record
+ is probably empty - unless flagShowRecords is in use */
+ if (test_mode)
+ return ZEBRA_OK;
}
- logf (LOG_LOG, "finished section %d", key_file_no);
- ptr_i = 0;
- key_buf_used = 0;
-}
-int key_close (void)
-{
- key_flush ();
- xfree (key_buf);
-#if RECORD_BASE
- rec_close (&records);
-#else
- close (sys_idx_fd);
- dict_insert (file_idx, ".", sizeof(sysno_next), &sysno_next);
-#endif
- dict_close (file_idx);
+ if (! *sysno)
+ {
+ /* new record */
+ if (delete_flag)
+ {
+ 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;
+ }
+ if (show_progress)
+ yaz_log (YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
+ (zint) start_offset);
+ rec = rec_new (zh->reg->records);
- xfree (key_del_buf);
- key_del_buf = NULL;
- key_del_max = 0;
- return key_file_no;
-}
+ *sysno = rec->sysno;
-static void wordInit (RecWord *p)
-{
- p->attrSet = 1;
- p->attrUse = 1016;
- p->which = Word_String;
-}
+ recordAttr = rec_init_attr (zh->reg->zei, rec);
+ recordAttr->staticrank = extractCtrl.staticrank;
-static void wordAdd (const RecWord *p)
-{
- struct it_key key;
- size_t i;
-
- if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
- key_flush ();
- ++ptr_i;
- key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
- key_buf_used += index_word_prefix ((char*)key_buf + key_buf_used,
- p->attrSet, p->attrUse,
- key_databaseName);
- switch (p->which)
- {
- case Word_String:
- for (i = 0; p->u.string[i]; i++)
- ((char*)key_buf) [key_buf_used++] =
- index_char_cvt (p->u.string[i]);
- ((char*)key_buf) [key_buf_used++] = '\0';
- break;
- default:
- return ;
- }
- ((char*) key_buf)[key_buf_used++] = ((key_cmd == 'a') ? 1 : 0);
- key.sysno = key_sysno;
- key.seqno = p->seqno;
- memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
- key_buf_used += sizeof(key);
+ if (matchStr)
+ {
+ int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
+ dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
+ sizeof(*sysno), sysno);
+ }
- if (key_cmd == 'a' && key_del_used >= 0)
+
+ extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
+ extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
+ recordAttr->staticrank);
+ zh->records_inserted++;
+ }
+ else
{
- char attrSet;
- short attrUse;
- if (key_del_used + 1024 > key_del_max)
+ /* record already exists */
+ zebra_rec_keys_t delkeys = zebra_rec_keys_open();
+ zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
+ if (!allow_update)
+ {
+ 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);
+
+ recordAttr = rec_init_attr (zh->reg->zei, rec);
+
+ zebra_rec_keys_set_buf(delkeys,
+ rec->info[recInfo_delKeys],
+ rec->size[recInfo_delKeys],
+ 0);
+ zebra_rec_keys_set_buf(sortKeys,
+ rec->info[recInfo_sortKeys],
+ rec->size[recInfo_sortKeys],
+ 0);
+
+ extract_flushSortKeys (zh, *sysno, 0, sortKeys);
+ extract_flushRecordKeys (zh, *sysno, 0, delkeys,
+ recordAttr->staticrank);
+ if (delete_flag)
{
- char *kbn;
-
- if (!(kbn = malloc (key_del_max += 64000)))
+ /* record going to be deleted */
+ if (zebra_rec_keys_empty(delkeys))
+ {
+ yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
+ pr_fname, (zint) start_offset);
+ yaz_log(YLOG_WARN, "cannot delete file above, "
+ "storeKeys false (3)");
+ }
+ else
{
- logf (LOG_FATAL, "malloc");
- exit (1);
+ if (show_progress)
+ yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
+ pr_fname, (zint) start_offset);
+ zh->records_deleted++;
+ if (matchStr)
+ {
+ 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);
}
- if (key_del_buf)
- memcpy (kbn, key_del_buf, key_del_used);
- free (key_del_buf);
- key_del_buf = kbn;
+ rec_rm (&rec);
+ logRecord(zh);
+ return ZEBRA_OK;
}
- switch (p->which)
+ else
{
- case Word_String:
- for (i = 0; p->u.string[i]; i++)
- ((char*)key_del_buf) [key_del_used++] = p->u.string[i];
- ((char*)key_del_buf) [key_del_used++] = '\0';
- break;
- default:
- return ;
+ if (show_progress)
+ yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
+ pr_fname, (zint) ZINT_FORMAT);
+ 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++;
}
- attrSet = p->attrSet;
- memcpy (key_del_buf + key_del_used, &attrSet, sizeof(attrSet));
- key_del_used += sizeof(attrSet);
+ zebra_rec_keys_close(delkeys);
+ zebra_rec_keys_close(sortKeys);
+ }
+ /* update file type */
+ xfree (rec->info[recInfo_fileType]);
+ rec->info[recInfo_fileType] =
+ rec_strdup (recordType, &rec->size[recInfo_fileType]);
- attrUse = p->attrUse;
- memcpy (key_del_buf + key_del_used, &attrUse, sizeof(attrUse));
- key_del_used += sizeof(attrUse);
+ /* update filename */
+ xfree (rec->info[recInfo_filename]);
+ rec->info[recInfo_filename] =
+ rec_strdup (fname, &rec->size[recInfo_filename]);
- memcpy (key_del_buf + key_del_used, &p->seqno, sizeof(p->seqno));
- key_del_used += sizeof(p->seqno);
+ /* update delete keys */
+ 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->info[recInfo_delKeys],
+ &rec->size[recInfo_delKeys]);
}
-}
+ else
+ {
+ rec->info[recInfo_delKeys] = NULL;
+ rec->size[recInfo_delKeys] = 0;
+ }
+ /* update sort keys */
+ xfree (rec->info[recInfo_sortKeys]);
-static void wordAddAny (const RecWord *p)
-{
- if (p->attrSet != 1 || p->attrUse != 1016)
+ 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)
{
- RecWord w;
+ off_t end_offset = stream->endf(stream, 0);
- memcpy (&w, p, sizeof(w));
- w.attrSet = 1;
- w.attrUse = 1016;
- wordAdd (&w);
+ 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);
}
- wordAdd (p);
-}
-static char *file_buf;
-static int file_offset;
-static int file_bufsize;
+ /* set run-number for this record */
+ recordAttr->runNumber =
+ zebraExplain_runNumberIncrement (zh->reg->zei, 0);
-static void file_read_start (int fd)
-{
- file_offset = 0;
- file_buf = xmalloc (4096);
- file_bufsize = read (fd, file_buf, 4096);
+ /* update store data */
+ 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;
+ }
+ else if (zh->m_store_data)
+ {
+ off_t cur_offset = stream->tellf(stream);
+
+ rec->size[recInfo_storeData] = recordAttr->recordSize;
+ rec->info[recInfo_storeData] = (char *)
+ xmalloc (recordAttr->recordSize);
+ stream->seekf(stream, start_offset);
+ stream->readf(stream, rec->info[recInfo_storeData],
+ recordAttr->recordSize);
+ stream->seekf(stream, cur_offset);
+ }
+ else
+ {
+ rec->info[recInfo_storeData] = NULL;
+ rec->size[recInfo_storeData] = 0;
+ }
+ /* update database name */
+ xfree (rec->info[recInfo_databaseName]);
+ rec->info[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);
+ logRecord(zh);
+ return ZEBRA_OK;
}
-static void file_read_stop (int fd)
+ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
{
- xfree (file_buf);
+ ZebraHandle zh = (ZebraHandle) handle;
+ struct recExtractCtrl extractCtrl;
+
+ if (zebraExplain_curDatabase (zh->reg->zei,
+ rec->info[recInfo_databaseName]))
+ {
+ abort();
+ if (zebraExplain_newDatabase (zh->reg->zei,
+ rec->info[recInfo_databaseName], 0))
+ abort ();
+ }
+
+ zebra_rec_keys_reset(zh->reg->keys);
+ zebra_rec_keys_reset(zh->reg->sortKeys);
+
+ extractCtrl.init = extract_init;
+ extractCtrl.tokenAdd = extract_token_add;
+ extractCtrl.schemaAdd = extract_schema_add;
+ extractCtrl.dh = zh->reg->dh;
+
+ init_extractCtrl(zh, &extractCtrl);
+
+ extractCtrl.flagShowRecords = 0;
+ extractCtrl.match_criteria[0] = '\0';
+ extractCtrl.staticrank = 0;
+ extractCtrl.handle = handle;
+ extractCtrl.first_record = 1;
+
+ extract_set_store_data_prepare(&extractCtrl);
+
+ if (n)
+ grs_extract_tree(&extractCtrl, n);
+
+ if (rec->size[recInfo_delKeys])
+ {
+ zebra_rec_keys_t delkeys = zebra_rec_keys_open();
+
+ zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
+
+ zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
+ rec->size[recInfo_delKeys],
+ 0);
+ extract_flushRecordKeys (zh, rec->sysno, 0, delkeys, 0);
+ 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);
+ 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]);
+ zebra_rec_keys_get_buf(zh->reg->keys,
+ &rec->info[recInfo_delKeys],
+ &rec->size[recInfo_delKeys]);
+
+ xfree (rec->info[recInfo_sortKeys]);
+ zebra_rec_keys_get_buf(zh->reg->sortKeys,
+ &rec->info[recInfo_sortKeys],
+ &rec->size[recInfo_sortKeys]);
+ return ZEBRA_OK;
}
-static int file_read (int fd, char *buf, size_t count)
+void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
+ zebra_rec_keys_t reckeys)
{
- int l = file_bufsize - file_offset;
+ ZebraExplainInfo zei = zh->reg->zei;
+ struct ord_stat {
+ int no;
+ int ord;
+ struct ord_stat *next;
+ };
- if (count > l)
+ if (zebra_rec_keys_rewind(reckeys))
{
- int r;
- if (l > 0)
- memcpy (buf, file_buf + file_offset, l);
- count = count-l;
- if (count > file_bufsize)
+ struct ord_stat *ord_list = 0;
+ struct ord_stat *p;
+ size_t slen;
+ const char *str;
+ struct it_key key_in;
+ while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
{
- if ((r = read (fd, buf + l, count)) == -1)
+ int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
+
+ for (p = ord_list; p ; p = p->next)
+ if (p->ord == ord)
+ {
+ p->no++;
+ break;
+ }
+ if (!p)
{
- logf (LOG_FATAL|LOG_ERRNO, "read");
- exit (1);
+ p = xmalloc(sizeof(*p));
+ p->no = 1;
+ p->ord = ord;
+ p->next = ord_list;
+ ord_list = p;
}
- file_bufsize = 0;
- file_offset = 0;
- return r;
- }
- file_bufsize = r = read (fd, file_buf, 4096);
- if (r == -1)
- {
- logf (LOG_FATAL|LOG_ERRNO, "read");
- exit (1);
}
- else if (r <= count)
- {
- file_offset = r;
- memcpy (buf + l, file_buf, r);
- return l + r;
- }
- else
+
+ p = ord_list;
+ while (p)
{
- file_offset = count;
- memcpy (buf + l, file_buf, count - l);
- return count;
+ struct ord_stat *p1 = p;
+
+ if (is_insert)
+ zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
+ else
+ zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
+ p = p->next;
+ xfree(p1);
}
}
- memcpy (buf, file_buf + file_offset, count);
- file_offset += count;
- return count;
}
-SYSNO file_extract (int cmd, const char *fname, const char *kname,
- char *databaseName)
+void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
+ int cmd,
+ zebra_rec_keys_t reckeys,
+ zint staticrank)
{
- int i, r;
- char ext[128];
- SYSNO sysno;
- char ext_res[128];
- const char *file_type;
- void *file_info;
- struct recExtractCtrl extractCtrl;
- RecType rt;
+ ZebraExplainInfo zei = zh->reg->zei;
- key_del_used = -1;
- key_databaseName = databaseName;
- for (i = strlen(fname); --i >= 0; )
- if (fname[i] == '/')
+ extract_rec_keys_adjust(zh, cmd, reckeys);
+
+ if (!zh->reg->key_buf)
+ {
+ 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;
+ }
+ zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
+
+ if (zebra_rec_keys_rewind(reckeys))
+ {
+ 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);
+ }
+ }
+}
+
+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)
+ {
+ 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))
{
- strcpy (ext, "");
- break;
+ encode_key_flush ( &encode_info, outf);
+ encode_key_init (&encode_info);
+ encode_key_write (cp, &encode_info, outf);
+ prevcp = cp;
}
- else if (fname[i] == '.')
+ else
+ encode_key_write (cp + strlen(cp), &encode_info, outf);
+ }
+ 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]))
{
- strcpy (ext, fname+i+1);
- break;
+ 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];
}
- sprintf (ext_res, "fileExtension.%s", ext);
- if (!(file_type = res_get (common_resource, ext_res)))
- return 0;
- if (!(rt = recType_byName (file_type)))
- return 0;
- logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
- file_info = dict_lookup (file_idx, kname);
- if (!file_info)
- {
-#if RECORD_BASE
- Record rec = rec_new (records);
-
- sysno = rec->sysno;
- dict_insert (file_idx, kname, sizeof(sysno), &sysno);
- rec->info[0] = rec_strdup (file_type, &rec->size[0]);
- rec->info[1] = rec_strdup (kname, &rec->size[1]);
- rec_put (records, &rec);
-#else
- sysno = sysno_next++;
- dict_insert (file_idx, kname, sizeof(sysno), &sysno);
- lseek (sys_idx_fd, sysno * SYS_IDX_ENTRY_LEN, SEEK_SET);
- write (sys_idx_fd, file_type, strlen (file_type)+1);
- write (sys_idx_fd, kname, strlen(kname)+1);
#endif
+ if (fclose (outf))
+ {
+ yaz_log (YLOG_FATAL|YLOG_ERRNO, "fclose %s", out_fname);
+ exit (1);
}
- else
- memcpy (&sysno, (char*) file_info+1, sizeof(sysno));
+ yaz_log (YLOG_LOG, "finished section %d", zh->reg->key_file_no);
+ zh->reg->ptr_i = 0;
+ zh->reg->key_buf_used = 0;
+}
- if ((extractCtrl.fd = open (fname, O_RDONLY)) == -1)
+ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh,
+ zebra_rec_keys_t reckeys,
+ zebra_snippets *snippets)
+{
+ NMEM nmem = nmem_create();
+ if (zebra_rec_keys_rewind(reckeys))
{
- logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
- return 0;
+ const char *str;
+ size_t slen;
+ struct it_key key;
+ while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
+ {
+ char dst_buf[IT_MAX_WORD];
+ char *dst_term = dst_buf;
+ int ord;
+ zint seqno;
+ int index_type;
+
+ assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
+ seqno = key.mem[key.len-1];
+ ord = CAST_ZINT_TO_INT(key.mem[0]);
+
+ zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
+ 0/* db */, 0 /* string_index */);
+ assert(index_type);
+ zebra_term_untrans_iconv(zh, nmem, index_type,
+ &dst_term, str);
+ zebra_snippets_append(snippets, seqno, ord, dst_term);
+ nmem_reset(nmem);
+ }
+ }
+ nmem_destroy(nmem);
+ return ZEBRA_OK;
+}
+
+void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
+{
+ yaz_log(YLOG_LOG, "print_rec_keys");
+ if (zebra_rec_keys_rewind(reckeys))
+ {
+ const char *str;
+ size_t slen;
+ struct it_key key;
+ while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
+ {
+ char dst_buf[IT_MAX_WORD];
+ zint seqno;
+ int 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);
+
+ zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
+
+ seqno = key.mem[key.len-1];
+
+ zebra_term_untrans(zh, index_type, dst_buf, str);
+
+ yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT
+ " term=%s", ord, seqno, dst_buf);
+ }
}
- extractCtrl.subType = "";
- extractCtrl.init = wordInit;
- extractCtrl.add = wordAddAny;
+}
+
+static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
+ const char *str, int length)
+{
+ struct it_key key;
+ ZebraHandle zh = p->extractCtrl->handle;
+ ZebraExplainInfo zei = zh->reg->zei;
+ int ch, i;
- file_read_start (extractCtrl.fd);
+ ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
+ if (ch < 0)
+ ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
- extractCtrl.readf = file_read;
- key_sysno = sysno;
- key_cmd = cmd;
- r = (*rt->extract)(&extractCtrl);
+ i = 0;
+ key.mem[i++] = ch;
+ key.mem[i++] = p->record_id;
+ key.mem[i++] = p->section_id;
- file_read_stop (extractCtrl.fd);
+ if (zh->m_segment_indexing)
+ key.mem[i++] = p->segment;
+ key.mem[i++] = p->seqno;
+ key.len = i;
- close (extractCtrl.fd);
- if (r)
- logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);
- return sysno;
+ zebra_rec_keys_write(zh->reg->keys, str, length, &key);
}
-int fileExtract (SYSNO *sysno, const char *fname, const char *databaseName,
- int deleteFlag)
+static void extract_add_sort_string(RecWord *p, const char *str, int length)
{
- int i, r;
- char ext[128];
- char ext_res[128];
- const char *file_type;
- struct recExtractCtrl extractCtrl;
- RecType rt;
- Record rec;
+ struct it_key key;
+ ZebraHandle zh = p->extractCtrl->handle;
+ ZebraExplainInfo zei = zh->reg->zei;
+ int ch;
+ zinfo_index_category_t cat = zinfo_index_category_sort;
- logf (LOG_DEBUG, "fileExtractAdd %s", fname);
+ ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
+ if (ch < 0)
+ ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
+ key.len = 2;
+ key.mem[0] = ch;
+ key.mem[1] = p->record_id;
- key_del_used = 0;
- for (i = strlen(fname); --i >= 0; )
- if (fname[i] == '/')
+ zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
+}
+
+static void extract_add_string(RecWord *p, const char *string, int length)
+{
+ ZebraHandle zh = p->extractCtrl->handle;
+ 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
+ {
+ extract_add_index_string(p, zinfo_index_category_index,
+ string, length);
+ if (zebra_maps_is_alwaysmatches(zh->reg->zebra_maps, p->index_type))
{
- strcpy (ext, "");
- break;
+ RecWord word;
+ memcpy(&word, p, sizeof(word));
+
+ word.seqno = 1;
+ extract_add_index_string(
+ &word, zinfo_index_category_alwaysmatches, "", 0);
}
- else if (fname[i] == '.')
+ }
+}
+
+static void extract_add_incomplete_field(RecWord *p)
+{
+ ZebraHandle zh = p->extractCtrl->handle;
+ const char *b = p->term_buf;
+ int remain = p->term_len;
+ const char **map = 0;
+
+ if (remain > 0)
+ map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
+
+ if (map)
+ {
+ if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
{
- strcpy (ext, fname+i+1);
- break;
+ /* first in field marker */
+ extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
+ p->seqno++;
}
- sprintf (ext_res, "fileExtension.%s", ext);
- if (!(file_type = res_get (common_resource, ext_res)))
- return 0;
- if (!(rt = recType_byName (file_type)))
- return 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);
+ if (remain > 0)
+ map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
+ remain, 0);
+ else
+ map = 0;
+ }
+ if (!map)
+ break;
+ 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);
+ if (remain > 0)
+ map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
+ else
+ map = 0;
+ }
+ if (!i)
+ return;
+ extract_add_string (p, buf, i);
+ p->seqno++;
+ }
+}
- if ((extractCtrl.fd = open (fname, O_RDONLY)) == -1)
+static void extract_add_complete_field (RecWord *p)
+{
+ 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);
+
+ while (remain > 0 && i < IT_MAX_WORD)
{
- logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
- return 0;
+ while (map && *map && **map == *CHR_SPACE)
+ {
+ remain = p->term_len - (b - p->term_buf);
+
+ if (remain > 0)
+ {
+ int first = i ? 0 : 1; /* first position */
+ map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &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++);
+ }
+ 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);
+ }
+ else
+ map = 0;
+ }
}
+ if (!i)
+ return;
+ extract_add_string (p, buf, i);
+}
- extractCtrl.subType = "";
- extractCtrl.init = wordInit;
- extractCtrl.add = wordAddAny;
+static void extract_token_add(RecWord *p)
+{
+ ZebraHandle zh = p->extractCtrl->handle;
+ WRBUF wrbuf;
- if (! *sysno)
+ if (log_level)
{
- logf (LOG_LOG, "add record %s", fname);
- rec = rec_new (records);
- *sysno = rec->sysno;
- rec->info[0] = rec_strdup (file_type, &rec->size[0]);
- rec->info[1] = rec_strdup (fname, &rec->size[1]);
- rec->info[3] = rec_strdup (databaseName, &rec->size[3]);
+ yaz_log(log_level, "extract_token_add "
+ "type=%c 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)))
+ {
+ p->term_buf = wrbuf_buf(wrbuf);
+ p->term_len = wrbuf_len(wrbuf);
}
+ if (zebra_maps_is_complete (zh->reg->zebra_maps, p->index_type))
+ extract_add_complete_field (p);
else
+ extract_add_incomplete_field(p);
+}
+
+static void extract_set_store_data_cb(struct recExtractCtrl *p,
+ void *buf, size_t sz)
+{
+ ZebraHandle zh = (ZebraHandle) p->handle;
+
+ xfree(zh->store_data_buf);
+ zh->store_data_buf = 0;
+ zh->store_data_size = 0;
+ if (buf && sz)
{
- size_t off;
- char *kb;
+ zh->store_data_buf = xmalloc(sz);
+ zh->store_data_size = sz;
+ memcpy(zh->store_data_buf, buf, sz);
+ }
+}
- if (deleteFlag)
- logf (LOG_LOG, "delete record %s", fname);
- else
- logf (LOG_LOG, "update record %s", fname);
- rec = rec_get (records, *sysno);
-
- key_cmd = 'd';
- key_sysno = *sysno;
- key_databaseName = rec->info[3];
- kb = rec->info[2];
- for (off = 0; off < rec->size[2]; )
+static void extract_set_store_data_prepare(struct recExtractCtrl *p)
+{
+ ZebraHandle zh = (ZebraHandle) p->handle;
+ xfree(zh->store_data_buf);
+ zh->store_data_buf = 0;
+ zh->store_data_size = 0;
+ p->setStoreData = extract_set_store_data_cb;
+}
+
+static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
+{
+ ZebraHandle zh = (ZebraHandle) p->handle;
+ zebraExplain_addSchema (zh->reg->zei, oid);
+}
+
+void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
+ int cmd, zebra_rec_keys_t reckeys)
+{
+ if (zebra_rec_keys_rewind(reckeys))
+ {
+ SortIdx sortIdx = zh->reg->sortIdx;
+ size_t slen;
+ const char *str;
+ struct it_key key_in;
+
+ sortIdx_sysno (sortIdx, sysno);
+
+ while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
{
- RecWord rw;
- char attrSet;
- short attrUse;
-
- rw.which = Word_String;
- rw.u.string = kb + off;
- while (kb[off])
- off++;
- off++;
- memcpy (&attrSet, kb + off, sizeof(attrSet));
- off += sizeof(attrSet);
- memcpy (&attrUse, kb + off, sizeof(attrUse));
- off += sizeof(attrUse);
- memcpy (&rw.seqno, kb + off, sizeof(rw.seqno));
- off += sizeof(rw.seqno);
- rw.attrSet = attrSet;
- rw.attrUse = attrUse;
-
- (*extractCtrl.add) (&rw);
+ int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
+
+ sortIdx_type(sortIdx, ord);
+ if (cmd == 1)
+ sortIdx_add(sortIdx, str, slen);
+ else
+ sortIdx_add(sortIdx, "", 1);
}
- assert (off == rec->size[2]);
- free (rec->info[2]);
- rec->info[2] = NULL;
- rec->size[2] = 0;
-
- free (rec->info[3]);
- rec->info[3] = rec_strdup (databaseName, &rec->size[3]);
}
+}
+
+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 (deleteFlag)
- {
#if 0
- rec_del (records, *sysno);
- rec_rm (&rec);
+ /* debugging */
+ key_logdump_txt(YLOG_LOG, &key, *k ? "i" : "d");
#endif
- return 1;
- }
-
- key_databaseName = databaseName;
- key_sysno = *sysno;
- key_cmd = 'a';
+ assert(key.mem[0] >= 0);
- file_read_start (extractCtrl.fd);
- extractCtrl.readf = file_read;
- r = (*rt->extract)(&extractCtrl);
- file_read_stop (extractCtrl.fd);
+ bp0 = bp++;
+ iscz1_encode(i->encode_handle, &bp, &src);
- close (extractCtrl.fd);
- if (r)
+ *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */
+ if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
{
- rec_rm (&rec);
- logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);
- return 0;
+ yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
+ exit (1);
}
- if (key_del_used > 0)
+
+#if 0
+ /* debugging */
+ if (1)
{
- rec->size[2] = key_del_used;
- rec->info[2] = malloc (rec->size[2]);
- memcpy (rec->info[2], key_del_buf, rec->size[2]);
+ 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]);
}
- rec_put (records, &rec);
- return 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
+ * indent-tabs-mode: nil
+ * End:
+ * vim: shiftwidth=4 tabstop=8 expandtab
+ */
+
projects / idzebra-moved-to-github.git / blobdiff