X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Frecords.c;h=1414ddd6999da6c9a52330610d3f73988c5ee069;hb=38b8c213f9be6890ddba9d5564dc978c2ae2bc3d;hp=4a24e3541cdb6b80f2ffbfcb92fd28dc05cc48b1;hpb=8fa59ec675af0ea4f15f157401c61121b5cb5782;p=idzebra-moved-to-github.git diff --git a/index/records.c b/index/records.c index 4a24e35..1414ddd 100644 --- a/index/records.c +++ b/index/records.c @@ -1,8 +1,5 @@ -/* $Id: records.c,v 1.1 2007-11-23 13:52:52 adam Exp $ - Copyright (C) 1995-2007 - Index Data ApS - -This file is part of the Zebra server. +/* This file is part of the Zebra server. + Copyright (C) 1994-2009 Index Data 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 @@ -41,12 +38,70 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include #include -#include "recindxp.h" +#include +#include "recindex.h" #if HAVE_BZLIB_H #include #endif +#define REC_BLOCK_TYPES 2 +#define REC_HEAD_MAGIC "recindex" +#define REC_VERSION 5 + +struct records_info { + int rw; + int compression_method; + + recindex_t recindex; + + char *data_fname[REC_BLOCK_TYPES]; + BFile data_BFile[REC_BLOCK_TYPES]; + + char *tmp_buf; + int tmp_size; + + struct record_cache_entry *record_cache; + int cache_size; + int cache_cur; + int cache_max; + + int compression_chunk_size; + + Zebra_mutex mutex; + + struct records_head { + char magic[8]; + char version[4]; + zint block_size[REC_BLOCK_TYPES]; + zint block_free[REC_BLOCK_TYPES]; + zint block_last[REC_BLOCK_TYPES]; + zint block_used[REC_BLOCK_TYPES]; + zint block_move[REC_BLOCK_TYPES]; + + zint total_bytes; + zint index_last; + zint index_free; + zint no_records; + + } head; +}; + +enum recordCacheFlag { recordFlagNop, recordFlagWrite, recordFlagNew, + recordFlagDelete }; + +struct record_cache_entry { + Record rec; + enum recordCacheFlag flag; +}; + +struct record_index_entry { + zint next; /* first block of record info / next free entry */ + int size; /* size of record or 0 if free entry */ +}; + +Record rec_cp(Record rec); + /* Modify argument to if below: 1=normal, 0=sysno testing */ #if 1 /* If this is used sysno are not converted (no testing) */ @@ -116,6 +171,7 @@ static ZEBRA_RES rec_release_blocks(Records p, zint sysno) memcpy(block_and_ref + sizeof(freeblock), &ref, sizeof(ref)); if (ref) { + /* there is still a reference to this block.. */ if (bf_write(p->data_BFile[dst_type], freeblock, 0, sizeof(block_and_ref), block_and_ref)) { @@ -124,7 +180,8 @@ static ZEBRA_RES rec_release_blocks(Records p, zint sysno) } return ZEBRA_OK; } - first = 0; + /* the list of blocks can all be removed (ref == 0) */ + first = 0; } if (bf_write(p->data_BFile[dst_type], freeblock, 0, sizeof(freeblock), @@ -232,8 +289,9 @@ Records rec_open(BFiles bfs, int rw, int compression_method) p->compression_method = compression_method; p->rw = rw; p->tmp_size = 1024; - p->recindex = recindex_open(bfs, rw); p->tmp_buf = (char *) xmalloc(p->tmp_size); + p->compression_chunk_size = 0; + p->recindex = recindex_open(bfs, rw, 0 /* 1=isamb for recindex */); r = recindex_read_head(p->recindex, p->tmp_buf); switch (r) { @@ -259,7 +317,8 @@ Records rec_open(BFiles bfs, int rw, int compression_method) } if (rw) { - if (recindex_write_head(p->recindex, &p->head, sizeof(p->head)) != ZEBRA_OK) + if (recindex_write_head(p->recindex, + &p->head, sizeof(p->head)) != ZEBRA_OK) ret = ZEBRA_FAIL; } break; @@ -279,6 +338,7 @@ Records rec_open(BFiles bfs, int rw, int compression_method) recindex_get_fname(p->recindex), version, REC_VERSION); ret = ZEBRA_FAIL; } + p->compression_chunk_size = 90000; /* good for BZIP2 */ break; } for (i = 0; iflag = recordFlagNop; break; - default: + case recordFlagNop: break; + default: + break; } } @@ -581,7 +643,7 @@ static ZEBRA_RES rec_cache_insert(Records p, Record rec, enum recordCacheFlag fl for (j = 0; jsize[j]; } - if (used > 90000) + if (used > p->compression_chunk_size) ret = rec_cache_flush(p, 1); } assert(p->cache_cur < p->cache_max); @@ -924,9 +986,67 @@ char *rec_strdup(const char *s, size_t *len) return p; } +void rec_prstat(Records records, int verbose) +{ + int i; + zint total_bytes = 0; + + yaz_log (YLOG_LOG, + "Total records %8" ZINT_FORMAT0, + records->head.no_records); + + for (i = 0; i< REC_BLOCK_TYPES; i++) + { + yaz_log (YLOG_LOG, "Record blocks of size "ZINT_FORMAT, + records->head.block_size[i]); + yaz_log (YLOG_LOG, + " Used/Total/Bytes used " + ZINT_FORMAT "/" ZINT_FORMAT "/" ZINT_FORMAT, + records->head.block_used[i], records->head.block_last[i]-1, + records->head.block_used[i] * records->head.block_size[i]); + total_bytes += + records->head.block_used[i] * records->head.block_size[i]; + + yaz_log(YLOG_LOG, " Block Last " ZINT_FORMAT, records->head.block_last[i]); + if (verbose) + { /* analyse free lists */ + zint no_free = 0; + zint block_free = records->head.block_free[i]; + WRBUF w = wrbuf_alloc(); + while (block_free) + { + zint nblock; + no_free++; + wrbuf_printf(w, " " ZINT_FORMAT, block_free); + if (bf_read(records->data_BFile[i], + block_free, 0, sizeof(nblock), &nblock) != 1) + { + yaz_log(YLOG_FATAL|YLOG_ERRNO, "read in %s at free block " + ZINT_FORMAT, + records->data_fname[i], block_free); + break; + } + block_free = nblock; + } + yaz_log (YLOG_LOG, + " Number in free list %8" ZINT_FORMAT0, no_free); + if (no_free) + yaz_log(YLOG_LOG, "%s", wrbuf_cstr(w)); + wrbuf_destroy(w); + } + } + yaz_log (YLOG_LOG, + "Total size of record index in bytes %8" ZINT_FORMAT0, + records->head.total_bytes); + yaz_log (YLOG_LOG, + "Total size with overhead %8" ZINT_FORMAT0, + total_bytes); +} + /* * Local variables: * c-basic-offset: 4 + * c-file-style: "Stroustrup" * indent-tabs-mode: nil * End: * vim: shiftwidth=4 tabstop=8 expandtab