-/* $Id: recindex.c,v 1.54 2006-11-21 22:17:49 adam Exp $
- Copyright (C) 1995-2006
- Index Data ApS
-
-This file is part of the Zebra server.
+/* This file is part of the Zebra server.
+ Copyright (C) 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
*/
-#define RIDX_CHUNK 128
-
-/*
- * Format of first block
- * next (8 bytes)
- * ref_count (2 bytes)
- * block (500 bytes)
- *
- * Format of subsequent blocks
- * next (8 bytes)
- * block (502 bytes)
- *
- * Format of each record
- * sysno
- * (length, data) - pairs
- * length = 0 if same as previous
- */
+#if HAVE_CONFIG_H
+#include <config.h>
+#endif
#include <stdio.h>
#include <stdlib.h>
#include <assert.h>
#include <string.h>
+#include <idzebra/isamb.h>
#include <yaz/yaz-util.h>
-#include "recindxp.h"
+#include "recindex.h"
-#if HAVE_BZLIB_H
-#include <bzlib.h>
-#endif
+#define RIDX_CHUNK 128
-/* Modify argument to if below: 1=normal, 0=sysno testing */
-#if 1
-/* If this is used sysno are not converted (no testing) */
-#define FAKE_OFFSET 0
-#define USUAL_RANGE 6000000000LL
-#else
-/* Use a fake > 2^32 offset so we can test for proper 64-bit handling */
-#define FAKE_OFFSET 6000000000LL
-#define USUAL_RANGE 2000000000LL
-#endif
+struct recindex {
+ char *index_fname;
+ BFile index_BFile;
+ ISAMB isamb;
+ ISAM_P isam_p;
+};
-static zint rec_sysno_to_ext(zint sysno)
-{
- assert(sysno >= 0 && sysno <= USUAL_RANGE);
- return sysno + FAKE_OFFSET;
-}
+struct record_index_entry {
+ zint next; /* first block of record info / next free entry */
+ int size; /* size of record or 0 if free entry */
+} ent;
-zint rec_sysno_to_int(zint sysno)
-{
- assert(sysno >= FAKE_OFFSET && sysno <= FAKE_OFFSET + USUAL_RANGE);
- return sysno - FAKE_OFFSET;
-}
-static ZEBRA_RES rec_write_head(Records p)
+static void rect_log_item(int level, const void *b, const char *txt)
{
- int r;
+ zint sys;
+ int len;
- assert(p);
- assert(p->index_BFile);
- r = bf_write(p->index_BFile, 0, 0, sizeof(p->head), &p->head);
- if (r)
- {
- yaz_log(YLOG_FATAL|YLOG_ERRNO, "write head of %s", p->index_fname);
- return ZEBRA_FAIL;
- }
- return ZEBRA_OK;
-}
+ memcpy(&sys, b, sizeof(sys));
+ len = ((const char *) b)[sizeof(sys)];
-static void rec_tmp_expand(Records p, int size)
-{
- if (p->tmp_size < size + 2048 ||
- p->tmp_size < p->head.block_size[REC_BLOCK_TYPES-1]*2)
+ if (len == sizeof(struct record_index_entry))
{
- xfree(p->tmp_buf);
- p->tmp_size = size + (int)
- (p->head.block_size[REC_BLOCK_TYPES-1])*2 + 2048;
- p->tmp_buf = (char *) xmalloc(p->tmp_size);
+ memcpy(&ent, (const char *)b + sizeof(sys) + 1, len);
+ yaz_log(YLOG_LOG, "%s " ZINT_FORMAT " next=" ZINT_FORMAT " sz=%d", txt, sys,
+ ent.next, ent.size);
+
}
+ else
+ yaz_log(YLOG_LOG, "%s " ZINT_FORMAT, txt, sys);
}
-static int read_indx(Records p, zint sysno, void *buf, int itemsize,
- int ignoreError)
+int rect_compare(const void *a, const void *b)
{
- int r;
- zint pos = (sysno-1)*itemsize;
- int off = CAST_ZINT_TO_INT(pos%RIDX_CHUNK);
- int sz1 = RIDX_CHUNK - off; /* sz1 is size of buffer to read.. */
+ zint s_a, s_b;
- if (sz1 > itemsize)
- sz1 = itemsize; /* no more than itemsize bytes */
+ memcpy(&s_a, a, sizeof(s_a));
+ memcpy(&s_b, b, sizeof(s_b));
- r = bf_read(p->index_BFile, 1+pos/RIDX_CHUNK, off, sz1, buf);
- if (r == 1 && sz1 < itemsize) /* boundary? - must read second part */
- r = bf_read(p->index_BFile, 2+pos/RIDX_CHUNK, 0, itemsize - sz1,
- (char*) buf + sz1);
- if (r != 1 && !ignoreError)
- {
- yaz_log(YLOG_FATAL|YLOG_ERRNO, "read in %s at pos %ld",
- p->index_fname, (long) pos);
- }
- return r;
+ if (s_a > s_b)
+ return 1;
+ else if (s_a < s_b)
+ return -1;
+ return 0;
}
-static void write_indx(Records p, zint sysno, void *buf, int itemsize)
+void *rect_code_start(void)
{
- zint pos = (sysno-1)*itemsize;
- int off = CAST_ZINT_TO_INT(pos%RIDX_CHUNK);
- int sz1 = RIDX_CHUNK - off; /* sz1 is size of buffer to read.. */
-
- if (sz1 > itemsize)
- sz1 = itemsize; /* no more than itemsize bytes */
-
- bf_write(p->index_BFile, 1+pos/RIDX_CHUNK, off, sz1, buf);
- if (sz1 < itemsize) /* boundary? must write second part */
- bf_write(p->index_BFile, 2+pos/RIDX_CHUNK, 0, itemsize - sz1,
- (char*) buf + sz1);
+ return 0;
}
-static ZEBRA_RES rec_release_blocks(Records p, zint sysno)
+void rect_encode(void *p, char **dst, const char **src)
{
- struct record_index_entry entry;
- zint freeblock;
- char block_and_ref[sizeof(zint) + sizeof(short)];
- int dst_type;
- int first = 1;
+ zint sys;
+ int len;
- if (read_indx(p, sysno, &entry, sizeof(entry), 1) != 1)
- return ZEBRA_FAIL;
+ memcpy(&sys, *src, sizeof(sys));
+ zebra_zint_encode(dst, sys);
+ (*src) += sizeof(sys);
- freeblock = entry.next;
- assert(freeblock > 0);
- dst_type = CAST_ZINT_TO_INT(freeblock & 7);
- assert(dst_type < REC_BLOCK_TYPES);
- freeblock = freeblock / 8;
- while (freeblock)
- {
- if (bf_read(p->data_BFile[dst_type], freeblock, 0,
- first ? sizeof(block_and_ref) : sizeof(zint),
- block_and_ref) != 1)
- {
- yaz_log(YLOG_FATAL|YLOG_ERRNO, "read in rec_del_single");
- return ZEBRA_FAIL;
- }
- if (first)
- {
- short ref;
- memcpy(&ref, block_and_ref + sizeof(freeblock), sizeof(ref));
- --ref;
- memcpy(block_and_ref + sizeof(freeblock), &ref, sizeof(ref));
- if (ref)
- {
- if (bf_write(p->data_BFile[dst_type], freeblock, 0,
- sizeof(block_and_ref), block_and_ref))
- {
- yaz_log(YLOG_FATAL|YLOG_ERRNO, "write in rec_del_single");
- return ZEBRA_FAIL;
- }
- return ZEBRA_OK;
- }
- first = 0;
- }
-
- if (bf_write(p->data_BFile[dst_type], freeblock, 0, sizeof(freeblock),
- &p->head.block_free[dst_type]))
- {
- yaz_log(YLOG_FATAL|YLOG_ERRNO, "write in rec_del_single");
- return ZEBRA_FAIL;
- }
- p->head.block_free[dst_type] = freeblock;
- memcpy(&freeblock, block_and_ref, sizeof(freeblock));
+ len = **src;
+ **dst = len;
+ (*src)++;
+ (*dst)++;
- p->head.block_used[dst_type]--;
- }
- p->head.total_bytes -= entry.size;
- return ZEBRA_OK;
+ memcpy(*dst, *src, len);
+ *dst += len;
+ *src += len;
}
-static ZEBRA_RES rec_delete_single(Records p, Record rec)
+void rect_decode(void *p, char **dst, const char **src)
{
- struct record_index_entry entry;
+ zint sys;
+ int len;
- if (rec_release_blocks(p, rec_sysno_to_int(rec->sysno)) != ZEBRA_OK)
- return ZEBRA_FAIL;
+ zebra_zint_decode(src, &sys);
+ memcpy(*dst, &sys, sizeof(sys));
+ *dst += sizeof(sys);
- entry.next = p->head.index_free;
- entry.size = 0;
- p->head.index_free = rec_sysno_to_int(rec->sysno);
- write_indx(p, rec_sysno_to_int(rec->sysno), &entry, sizeof(entry));
- return ZEBRA_OK;
-}
+ len = **src;
+ **dst = len;
+ (*src)++;
+ (*dst)++;
-static ZEBRA_RES rec_write_tmp_buf(Records p, int size, zint *sysnos)
-{
- struct record_index_entry entry;
- int no_written = 0;
- char *cptr = p->tmp_buf;
- zint block_prev = -1, block_free;
- int dst_type = 0;
- int i;
-
- for (i = 1; i<REC_BLOCK_TYPES; i++)
- if (size >= p->head.block_move[i])
- dst_type = i;
- while (no_written < size)
- {
- block_free = p->head.block_free[dst_type];
- if (block_free)
- {
- if (bf_read(p->data_BFile[dst_type],
- block_free, 0, sizeof(*p->head.block_free),
- &p->head.block_free[dst_type]) != 1)
- {
- yaz_log(YLOG_FATAL|YLOG_ERRNO, "read in %s at free block "
- ZINT_FORMAT,
- p->data_fname[dst_type], block_free);
- return ZEBRA_FAIL;
- }
- }
- else
- block_free = p->head.block_last[dst_type]++;
- if (block_prev == -1)
- {
- entry.next = block_free*8 + dst_type;
- entry.size = size;
- p->head.total_bytes += size;
- while (*sysnos > 0)
- {
- write_indx(p, *sysnos, &entry, sizeof(entry));
- sysnos++;
- }
- }
- else
- {
- memcpy(cptr, &block_free, sizeof(block_free));
- bf_write(p->data_BFile[dst_type], block_prev, 0, 0, cptr);
- cptr = p->tmp_buf + no_written;
- }
- block_prev = block_free;
- no_written += CAST_ZINT_TO_INT(p->head.block_size[dst_type])
- - sizeof(zint);
- p->head.block_used[dst_type]++;
- }
- assert(block_prev != -1);
- block_free = 0;
- memcpy(cptr, &block_free, sizeof(block_free));
- bf_write(p->data_BFile[dst_type], block_prev, 0,
- sizeof(block_free) + (p->tmp_buf+size) - cptr, cptr);
- return ZEBRA_OK;
+ memcpy(*dst, *src, len);
+ *dst += len;
+ *src += len;
}
-Records rec_open(BFiles bfs, int rw, int compression_method)
+void rect_code_reset(void *p)
{
- Records p;
- int i, r;
- int version;
- ZEBRA_RES ret = ZEBRA_OK;
-
- p = (Records) xmalloc(sizeof(*p));
- p->compression_method = compression_method;
- p->rw = rw;
- p->tmp_size = 1024;
- p->index_fname = "reci";
- p->index_BFile = bf_open(bfs, p->index_fname, RIDX_CHUNK, rw);
- if (p->index_BFile == NULL)
- {
- yaz_log(YLOG_FATAL|YLOG_ERRNO, "open %s", p->index_fname);
- xfree(p);
- return 0;
- }
- p->tmp_buf = (char *) xmalloc(p->tmp_size);
- r = bf_read(p->index_BFile, 0, 0, 0, p->tmp_buf);
- switch (r)
- {
- case 0:
- memcpy(p->head.magic, REC_HEAD_MAGIC, sizeof(p->head.magic));
- sprintf(p->head.version, "%3d", REC_VERSION);
- p->head.index_free = 0;
- p->head.index_last = 1;
- p->head.no_records = 0;
- p->head.total_bytes = 0;
- for (i = 0; i<REC_BLOCK_TYPES; i++)
- {
- p->head.block_free[i] = 0;
- p->head.block_last[i] = 1;
- p->head.block_used[i] = 0;
- }
- p->head.block_size[0] = 128;
- p->head.block_move[0] = 0;
- for (i = 1; i<REC_BLOCK_TYPES; i++)
- {
- p->head.block_size[i] = p->head.block_size[i-1] * 4;
- p->head.block_move[i] = p->head.block_size[i] * 24;
- }
- if (rw)
- {
- if (rec_write_head(p) != ZEBRA_OK)
- ret = ZEBRA_FAIL;
- }
- break;
- case 1:
- memcpy(&p->head, p->tmp_buf, sizeof(p->head));
- if (memcmp(p->head.magic, REC_HEAD_MAGIC, sizeof(p->head.magic)))
- {
- yaz_log(YLOG_FATAL, "file %s has bad format", p->index_fname);
- ret = ZEBRA_FAIL;
- }
- version = atoi(p->head.version);
- if (version != REC_VERSION)
- {
- yaz_log(YLOG_FATAL, "file %s is version %d, but version"
- " %d is required", p->index_fname, version, REC_VERSION);
- ret = ZEBRA_FAIL;
- }
- break;
- }
- for (i = 0; i<REC_BLOCK_TYPES; i++)
- {
- char str[80];
- sprintf(str, "recd%c", i + 'A');
- p->data_fname[i] = (char *) xmalloc(strlen(str)+1);
- strcpy(p->data_fname[i], str);
- p->data_BFile[i] = NULL;
- }
- for (i = 0; i<REC_BLOCK_TYPES; i++)
- {
- if (!(p->data_BFile[i] =
- bf_open(bfs, p->data_fname[i],
- CAST_ZINT_TO_INT(p->head.block_size[i]), rw)))
- {
- yaz_log(YLOG_FATAL|YLOG_ERRNO, "bf_open %s", p->data_fname[i]);
- ret = ZEBRA_FAIL;
- break;
- }
- }
- p->cache_max = 400;
- p->cache_cur = 0;
- p->record_cache = (struct record_cache_entry *)
- xmalloc(sizeof(*p->record_cache)*p->cache_max);
- zebra_mutex_init(&p->mutex);
- if (ret == ZEBRA_FAIL)
- rec_close(&p);
- return p;
}
-static void rec_encode_unsigned(unsigned n, unsigned char *buf, int *len)
+void rect_code_stop(void *p)
{
- (*len) = 0;
- while (n > 127)
- {
- buf[*len] = 128 + (n & 127);
- n = n >> 7;
- (*len)++;
- }
- buf[*len] = n;
- (*len)++;
}
-static void rec_decode_unsigned(unsigned *np, unsigned char *buf, int *len)
+
+recindex_t recindex_open(BFiles bfs, int rw, int use_isamb)
{
- unsigned n = 0;
- unsigned w = 1;
- (*len) = 0;
+ recindex_t p = xmalloc(sizeof(*p));
+ p->index_BFile = 0;
+ p->isamb = 0;
- while (buf[*len] > 127)
+ p->index_fname = "reci";
+ p->index_BFile = bf_open(bfs, p->index_fname, RIDX_CHUNK, rw);
+ if (p->index_BFile == NULL)
{
- n += w*(buf[*len] & 127);
- w = w << 7;
- (*len)++;
+ yaz_log(YLOG_FATAL|YLOG_ERRNO, "open %s", p->index_fname);
+ xfree(p);
+ return 0;
}
- n += w * buf[*len];
- (*len)++;
- *np = n;
-}
-static void rec_encode_zint(zint n, unsigned char *buf, int *len)
-{
- (*len) = 0;
- while (n > 127)
+ if (use_isamb)
{
- buf[*len] = (unsigned) (128 + (n & 127));
- n = n >> 7;
- (*len)++;
- }
- buf[*len] = (unsigned) n;
- (*len)++;
-}
+ int isam_block_size = 4096;
+ ISAMC_M method;
-static void rec_decode_zint(zint *np, unsigned char *buf, int *len)
-{
- zint n = 0;
- zint w = 1;
- (*len) = 0;
+ method.compare_item = rect_compare;
+ method.log_item = rect_log_item;
+ method.codec.start = rect_code_start;
+ method.codec.encode = rect_encode;
+ method.codec.decode = rect_decode;
+ method.codec.reset = rect_code_reset;
+ method.codec.stop = rect_code_stop;
- while (buf[*len] > 127)
- {
- n += w*(buf[*len] & 127);
- w = w << 7;
- (*len)++;
- }
- n += w * buf[*len];
- (*len)++;
- *np = n;
-}
+ p->index_fname = "rect";
+ p->isamb = isamb_open2(bfs, p->index_fname, rw, &method,
+ /* cache */ 0,
+ /* no_cat */ 1, &isam_block_size,
+ /* use_root_ptr */ 1);
-static void rec_cache_flush_block1(Records p, Record rec, Record last_rec,
- char **out_buf, int *out_size,
- int *out_offset)
-{
- int i;
- int len;
+ p->isam_p = 0;
+ if (p->isamb)
+ p->isam_p = isamb_get_root_ptr(p->isamb);
- for (i = 0; i<REC_NO_INFO; i++)
- {
- if (*out_offset + CAST_ZINT_TO_INT(rec->size[i]) + 20 > *out_size)
- {
- int new_size = *out_offset + rec->size[i] + 65536;
- char *np = (char *) xmalloc(new_size);
- if (*out_offset)
- memcpy(np, *out_buf, *out_offset);
- xfree(*out_buf);
- *out_size = new_size;
- *out_buf = np;
- }
- if (i == 0)
- {
- rec_encode_zint(rec_sysno_to_int(rec->sysno),
- (unsigned char *) *out_buf + *out_offset, &len);
- (*out_offset) += len;
- }
- if (rec->size[i] == 0)
- {
- rec_encode_unsigned(1, (unsigned char *) *out_buf + *out_offset,
- &len);
- (*out_offset) += len;
- }
- else if (last_rec && rec->size[i] == last_rec->size[i] &&
- !memcmp(rec->info[i], last_rec->info[i], rec->size[i]))
- {
- rec_encode_unsigned(0, (unsigned char *) *out_buf + *out_offset,
- &len);
- (*out_offset) += len;
- }
- else
- {
- rec_encode_unsigned(rec->size[i]+1,
- (unsigned char *) *out_buf + *out_offset,
- &len);
- (*out_offset) += len;
- memcpy(*out_buf + *out_offset, rec->info[i], rec->size[i]);
- (*out_offset) += rec->size[i];
- }
}
+ return p;
}
-static ZEBRA_RES rec_write_multiple(Records p, int saveCount)
+static void log_pr(const char *txt)
{
- int i;
- short ref_count = 0;
- char compression_method;
- Record last_rec = 0;
- int out_size = 1000;
- int out_offset = 0;
- char *out_buf = (char *) xmalloc(out_size);
- zint *sysnos = (zint *) xmalloc(sizeof(*sysnos) * (p->cache_cur + 1));
- zint *sysnop = sysnos;
- ZEBRA_RES ret = ZEBRA_OK;
-
- for (i = 0; i<p->cache_cur - saveCount; i++)
- {
- struct record_cache_entry *e = p->record_cache + i;
- switch (e->flag)
- {
- case recordFlagNew:
- rec_cache_flush_block1(p, e->rec, last_rec, &out_buf,
- &out_size, &out_offset);
- *sysnop++ = rec_sysno_to_int(e->rec->sysno);
- ref_count++;
- e->flag = recordFlagNop;
- last_rec = e->rec;
- break;
- case recordFlagWrite:
- if (rec_release_blocks(p, rec_sysno_to_int(e->rec->sysno))
- != ZEBRA_OK)
- ret = ZEBRA_FAIL;
-
- rec_cache_flush_block1(p, e->rec, last_rec, &out_buf,
- &out_size, &out_offset);
- *sysnop++ = rec_sysno_to_int(e->rec->sysno);
- ref_count++;
- e->flag = recordFlagNop;
- last_rec = e->rec;
- break;
- case recordFlagDelete:
- if (rec_delete_single(p, e->rec) != ZEBRA_OK)
- ret = ZEBRA_FAIL;
-
- e->flag = recordFlagNop;
- break;
- default:
- break;
- }
- }
-
- *sysnop = -1;
- if (ref_count)
- {
- unsigned int csize = 0; /* indicate compression "not performed yet" */
- compression_method = p->compression_method;
- switch (compression_method)
- {
- case REC_COMPRESS_BZIP2:
-#if HAVE_BZLIB_H
- csize = out_offset + (out_offset >> 6) + 620;
- rec_tmp_expand(p, csize);
-#ifdef BZ_CONFIG_ERROR
- i = BZ2_bzBuffToBuffCompress
-#else
- i = bzBuffToBuffCompress
-#endif
- (p->tmp_buf+sizeof(zint)+sizeof(short)+
- sizeof(char),
- &csize, out_buf, out_offset, 1, 0, 30);
- if (i != BZ_OK)
- {
- yaz_log(YLOG_WARN, "bzBuffToBuffCompress error code=%d", i);
- csize = 0;
- }
- yaz_log(YLOG_LOG, "compress %4d %5d %5d", ref_count, out_offset,
- csize);
-#endif
- break;
- case REC_COMPRESS_NONE:
- break;
- }
- if (!csize)
- {
- /* either no compression or compression not supported ... */
- csize = out_offset;
- rec_tmp_expand(p, csize);
- memcpy(p->tmp_buf + sizeof(zint) + sizeof(short) + sizeof(char),
- out_buf, out_offset);
- csize = out_offset;
- compression_method = REC_COMPRESS_NONE;
- }
- memcpy(p->tmp_buf + sizeof(zint), &ref_count, sizeof(ref_count));
- memcpy(p->tmp_buf + sizeof(zint)+sizeof(short),
- &compression_method, sizeof(compression_method));
-
- /* -------- compression */
- if (rec_write_tmp_buf(p, csize + sizeof(short) + sizeof(char), sysnos)
- != ZEBRA_OK)
- ret = ZEBRA_FAIL;
- }
- xfree(out_buf);
- xfree(sysnos);
- return ret;
+ yaz_log(YLOG_LOG, "%s", txt);
}
-static ZEBRA_RES rec_cache_flush(Records p, int saveCount)
-{
- int i, j;
- ZEBRA_RES ret;
- if (saveCount >= p->cache_cur)
- saveCount = 0;
-
- ret = rec_write_multiple(p, saveCount);
-
- for (i = 0; i<p->cache_cur - saveCount; i++)
- {
- struct record_cache_entry *e = p->record_cache + i;
- rec_free(&e->rec);
- }
- /* i still being used ... */
- for (j = 0; j<saveCount; j++, i++)
- memcpy(p->record_cache+j, p->record_cache+i,
- sizeof(*p->record_cache));
- p->cache_cur = saveCount;
- return ret;
-}
-
-static Record *rec_cache_lookup(Records p, zint sysno,
- enum recordCacheFlag flag)
+void recindex_close(recindex_t p)
{
- int i;
- for (i = 0; i<p->cache_cur; i++)
+ if (p)
{
- struct record_cache_entry *e = p->record_cache + i;
- if (e->rec->sysno == sysno)
+ if (p->index_BFile)
+ bf_close(p->index_BFile);
+ if (p->isamb)
{
- if (flag != recordFlagNop && e->flag == recordFlagNop)
- e->flag = flag;
- return &e->rec;
+ isamb_set_root_ptr(p->isamb, p->isam_p);
+ isamb_dump(p->isamb, p->isam_p, log_pr);
+ isamb_close(p->isamb);
}
+ xfree(p);
}
- return NULL;
}
-static ZEBRA_RES rec_cache_insert(Records p, Record rec, enum recordCacheFlag flag)
+int recindex_read_head(recindex_t p, void *buf)
{
- struct record_cache_entry *e;
- ZEBRA_RES ret = ZEBRA_OK;
-
- if (p->cache_cur == p->cache_max)
- ret = rec_cache_flush(p, 1);
- else if (p->cache_cur > 0)
- {
- int i, j;
- int used = 0;
- for (i = 0; i<p->cache_cur; i++)
- {
- Record r = (p->record_cache + i)->rec;
- for (j = 0; j<REC_NO_INFO; j++)
- used += r->size[j];
- }
- if (used > 90000)
- ret = rec_cache_flush(p, 1);
- }
- assert(p->cache_cur < p->cache_max);
-
- e = p->record_cache + (p->cache_cur)++;
- e->flag = flag;
- e->rec = rec_cp(rec);
- return ret;
+ return bf_read(p->index_BFile, 0, 0, 0, buf);
}
-ZEBRA_RES rec_close(Records *pp)
+const char *recindex_get_fname(recindex_t p)
{
- Records p = *pp;
- int i;
- ZEBRA_RES ret = ZEBRA_OK;
-
- if (!p)
- return ret;
-
- zebra_mutex_destroy(&p->mutex);
- if (rec_cache_flush(p, 0) != ZEBRA_OK)
- ret = ZEBRA_FAIL;
+ return p->index_fname;
+}
- xfree(p->record_cache);
+ZEBRA_RES recindex_write_head(recindex_t p, const void *buf, size_t len)
+{
+ int r;
- if (p->rw)
- {
- if (rec_write_head(p) != ZEBRA_OK)
- ret = ZEBRA_FAIL;
- }
+ assert(p);
- if (p->index_BFile)
- bf_close(p->index_BFile);
+ assert(p->index_BFile);
- for (i = 0; i<REC_BLOCK_TYPES; i++)
+ r = bf_write(p->index_BFile, 0, 0, len, buf);
+ if (r)
{
- if (p->data_BFile[i])
- bf_close(p->data_BFile[i]);
- xfree(p->data_fname[i]);
+ yaz_log(YLOG_FATAL|YLOG_ERRNO, "write head of %s", p->index_fname);
+ return ZEBRA_FAIL;
}
- xfree(p->tmp_buf);
- xfree(p);
- *pp = NULL;
- return ret;
+ return ZEBRA_OK;
}
-static Record rec_get_int(Records p, zint sysno)
+int recindex_read_indx(recindex_t p, zint sysno, void *buf, int itemsize,
+ int ignoreError)
{
- int i, in_size, r;
- Record rec, *recp;
- struct record_index_entry entry;
- zint freeblock;
- int dst_type;
- char *nptr, *cptr;
- char *in_buf = 0;
- char *bz_buf = 0;
-#if HAVE_BZLIB_H
- unsigned int bz_size;
-#endif
- char compression_method;
-
- assert(sysno > 0);
- assert(p);
-
- if ((recp = rec_cache_lookup(p, sysno, recordFlagNop)))
- return rec_cp(*recp);
-
- if (read_indx(p, rec_sysno_to_int(sysno), &entry, sizeof(entry), 1) < 1)
- return NULL; /* record is not there! */
-
- if (!entry.size)
- return NULL; /* record is deleted */
+ int r = 0;
+ if (p->isamb)
+ {
+ if (p->isam_p)
+ {
+ char item[256];
+ char *st = item;
+ char untilbuf[sizeof(zint) + 1];
- dst_type = (int) (entry.next & 7);
- assert(dst_type < REC_BLOCK_TYPES);
- freeblock = entry.next / 8;
+ ISAMB_PP isam_pp = isamb_pp_open(p->isamb, p->isam_p, 1);
- assert(freeblock > 0);
-
- rec_tmp_expand(p, entry.size);
+ memcpy(untilbuf, &sysno, sizeof(sysno));
+ untilbuf[sizeof(sysno)] = 0;
+ r = isamb_pp_forward(isam_pp, st, untilbuf);
- cptr = p->tmp_buf;
- r = bf_read(p->data_BFile[dst_type], freeblock, 0, 0, cptr);
- if (r < 0)
- return 0;
- memcpy(&freeblock, cptr, sizeof(freeblock));
+ isamb_pp_close(isam_pp);
+ if (!r)
+ return 0;
- while (freeblock)
- {
- zint tmp;
-
- cptr += p->head.block_size[dst_type] - sizeof(freeblock);
-
- memcpy(&tmp, cptr, sizeof(tmp));
- r = bf_read(p->data_BFile[dst_type], freeblock, 0, 0, cptr);
- if (r < 0)
- return 0;
- memcpy(&freeblock, cptr, sizeof(freeblock));
- memcpy(cptr, &tmp, sizeof(tmp));
+ if (item[sizeof(sysno)] != itemsize)
+ {
+ yaz_log(YLOG_WARN, "unexpected entry size %d != %d",
+ item[sizeof(sysno)], itemsize);
+ return 0;
+ }
+ memcpy(buf, item + sizeof(sysno) + 1, itemsize);
+ }
}
-
- rec = (Record) xmalloc(sizeof(*rec));
- rec->sysno = sysno;
- memcpy(&compression_method, p->tmp_buf + sizeof(zint) + sizeof(short),
- sizeof(compression_method));
- in_buf = p->tmp_buf + sizeof(zint) + sizeof(short) + sizeof(char);
- in_size = entry.size - sizeof(short) - sizeof(char);
- switch (compression_method)
+ else
{
- case REC_COMPRESS_BZIP2:
-#if HAVE_BZLIB_H
- bz_size = entry.size * 20 + 100;
- while (1)
- {
- bz_buf = (char *) xmalloc(bz_size);
-#ifdef BZ_CONFIG_ERROR
- i = BZ2_bzBuffToBuffDecompress
-#else
- i = bzBuffToBuffDecompress
-#endif
- (bz_buf, &bz_size, in_buf, in_size, 0, 0);
- yaz_log(YLOG_LOG, "decompress %5d %5d", in_size, bz_size);
- if (i == BZ_OK)
- break;
- yaz_log(YLOG_LOG, "failed");
- xfree(bz_buf);
- bz_size *= 2;
- }
- in_buf = bz_buf;
- in_size = bz_size;
-#else
- yaz_log(YLOG_FATAL, "cannot decompress record(s) in BZIP2 format");
- return 0;
-#endif
- break;
- case REC_COMPRESS_NONE:
- break;
- }
- for (i = 0; i<REC_NO_INFO; i++)
- rec->info[i] = 0;
+ zint pos = (sysno-1)*itemsize;
+ int off = CAST_ZINT_TO_INT(pos%RIDX_CHUNK);
+ int sz1 = RIDX_CHUNK - off; /* sz1 is size of buffer to read.. */
- nptr = in_buf; /* skip ref count */
- while (nptr < in_buf + in_size)
- {
- zint this_sysno;
- int len;
- rec_decode_zint(&this_sysno, (unsigned char *) nptr, &len);
- nptr += len;
-
- for (i = 0; i < REC_NO_INFO; i++)
- {
- unsigned int this_size;
- rec_decode_unsigned(&this_size, (unsigned char *) nptr, &len);
- nptr += len;
-
- if (this_size == 0)
- continue;
- rec->size[i] = this_size-1;
-
- if (rec->size[i])
- {
- rec->info[i] = nptr;
- nptr += rec->size[i];
- }
- else
- rec->info[i] = NULL;
- }
- if (this_sysno == rec_sysno_to_int(sysno))
- break;
+ if (sz1 > itemsize)
+ sz1 = itemsize; /* no more than itemsize bytes */
+
+ r = bf_read(p->index_BFile, 1+pos/RIDX_CHUNK, off, sz1, buf);
+ if (r == 1 && sz1 < itemsize) /* boundary? - must read second part */
+ r = bf_read(p->index_BFile, 2+pos/RIDX_CHUNK, 0, itemsize - sz1,
+ (char*) buf + sz1);
+ if (r != 1 && !ignoreError)
+ {
+ yaz_log(YLOG_FATAL|YLOG_ERRNO, "read in %s at pos %ld",
+ p->index_fname, (long) pos);
+ }
}
- for (i = 0; i<REC_NO_INFO; i++)
+#if 0
{
- if (rec->info[i] && rec->size[i])
- {
- char *np = xmalloc(rec->size[i]+1);
- memcpy(np, rec->info[i], rec->size[i]);
- np[rec->size[i]] = '\0';
- rec->info[i] = np;
- }
- else
- {
- assert(rec->info[i] == 0);
- assert(rec->size[i] == 0);
- }
+ struct record_index_entry *ep = buf;
+ yaz_log(YLOG_LOG, "read r=%d sysno=" ZINT_FORMAT " next=" ZINT_FORMAT
+ " sz=%d", r, sysno, ep->next, ep->size);
}
- xfree(bz_buf);
- if (rec_cache_insert(p, rec, recordFlagNop) != ZEBRA_OK)
- return 0;
- return rec;
+#endif
+ return r;
}
-Record rec_get(Records p, zint sysno)
-{
- Record rec;
- zebra_mutex_lock(&p->mutex);
-
- rec = rec_get_int(p, sysno);
- zebra_mutex_unlock(&p->mutex);
- return rec;
-}
+struct code_read_data {
+ int no;
+ zint sysno;
+ void *buf;
+ int itemsize;
+ int insert_flag;
+};
-Record rec_get_root(Records p)
+int bt_code_read(void *vp, char **dst, int *insertMode)
{
- return rec_get(p, rec_sysno_to_ext(1));
-}
+ struct code_read_data *s = (struct code_read_data *) vp;
-static Record rec_new_int(Records p)
-{
- int i;
- zint sysno;
- Record rec;
+ if (s->no == 0)
+ return 0;
- assert(p);
- rec = (Record) xmalloc(sizeof(*rec));
- if (1 || p->head.index_free == 0)
- sysno = (p->head.index_last)++;
- else
- {
- struct record_index_entry entry;
-
- if (read_indx(p, p->head.index_free, &entry, sizeof(entry), 0) < 1)
- {
- xfree(rec);
- return 0;
- }
- sysno = p->head.index_free;
- p->head.index_free = entry.next;
- }
- (p->head.no_records)++;
- rec->sysno = rec_sysno_to_ext(sysno);
- for (i = 0; i < REC_NO_INFO; i++)
- {
- rec->info[i] = NULL;
- rec->size[i] = 0;
- }
- rec_cache_insert(p, rec, recordFlagNew);
- return rec;
-}
+ (s->no)--;
-Record rec_new(Records p)
-{
- Record rec;
- zebra_mutex_lock(&p->mutex);
-
- rec = rec_new_int(p);
- zebra_mutex_unlock(&p->mutex);
- return rec;
+ memcpy(*dst, &s->sysno, sizeof(zint));
+ *dst += sizeof(zint);
+ **dst = s->itemsize;
+ (*dst)++;
+ memcpy(*dst, s->buf, s->itemsize);
+ *dst += s->itemsize;
+ *insertMode = s->insert_flag;
+ return 1;
}
-ZEBRA_RES rec_del(Records p, Record *recpp)
+void recindex_write_indx(recindex_t p, zint sysno, void *buf, int itemsize)
{
- Record *recp;
- ZEBRA_RES ret = ZEBRA_OK;
-
- zebra_mutex_lock(&p->mutex);
- (p->head.no_records)--;
- if ((recp = rec_cache_lookup(p, (*recpp)->sysno, recordFlagDelete)))
- {
- rec_free(recp);
- *recp = *recpp;
- }
- else
+#if 0
+ yaz_log(YLOG_LOG, "write_indx sysno=" ZINT_FORMAT, sysno);
+#endif
+ if (p->isamb)
{
- ret = rec_cache_insert(p, *recpp, recordFlagDelete);
- rec_free(recpp);
- }
- zebra_mutex_unlock(&p->mutex);
- *recpp = NULL;
- return ret;
-}
+ struct code_read_data input;
+ ISAMC_I isamc_i;
-ZEBRA_RES rec_put(Records p, Record *recpp)
-{
- Record *recp;
- ZEBRA_RES ret = ZEBRA_OK;
+ input.sysno = sysno;
+ input.buf = buf;
+ input.itemsize = itemsize;
- zebra_mutex_lock(&p->mutex);
- if ((recp = rec_cache_lookup(p, (*recpp)->sysno, recordFlagWrite)))
- {
- rec_free(recp);
- *recp = *recpp;
+ isamc_i.clientData = &input;
+ isamc_i.read_item = bt_code_read;
+
+ input.no = 1;
+ input.insert_flag = 2;
+ isamb_merge(p->isamb, &p->isam_p, &isamc_i);
}
else
{
- ret = rec_cache_insert(p, *recpp, recordFlagWrite);
- rec_free(recpp);
- }
- zebra_mutex_unlock(&p->mutex);
- *recpp = NULL;
- return ret;
-}
-
-void rec_free(Record *recpp)
-{
- int i;
-
- if (!*recpp)
- return ;
- for (i = 0; i < REC_NO_INFO; i++)
- xfree((*recpp)->info[i]);
- xfree(*recpp);
- *recpp = NULL;
-}
-
-Record rec_cp(Record rec)
-{
- Record n;
- int i;
-
- n = (Record) xmalloc(sizeof(*n));
- n->sysno = rec->sysno;
- for (i = 0; i < REC_NO_INFO; i++)
- if (!rec->info[i])
- {
- n->info[i] = NULL;
- n->size[i] = 0;
- }
- else
- {
- n->size[i] = rec->size[i];
- n->info[i] = (char *) xmalloc(rec->size[i]+1);
- memcpy(n->info[i], rec->info[i], rec->size[i]);
- n->info[i][rec->size[i]] = '\0';
- }
- return n;
-}
-
+ zint pos = (sysno-1)*itemsize;
+ int off = CAST_ZINT_TO_INT(pos%RIDX_CHUNK);
+ int sz1 = RIDX_CHUNK - off; /* sz1 is size of buffer to read.. */
-char *rec_strdup(const char *s, size_t *len)
-{
- char *p;
+ if (sz1 > itemsize)
+ sz1 = itemsize; /* no more than itemsize bytes */
- if (!s)
- {
- *len = 0;
- return NULL;
+ bf_write(p->index_BFile, 1+pos/RIDX_CHUNK, off, sz1, buf);
+ if (sz1 < itemsize) /* boundary? must write second part */
+ bf_write(p->index_BFile, 2+pos/RIDX_CHUNK, 0, itemsize - sz1,
+ (char*) buf + sz1);
}
- *len = strlen(s)+1;
- p = (char *) xmalloc(*len);
- strcpy(p, s);
- return p;
}
+
/*
* Local variables:
* c-basic-offset: 4
+ * c-file-style: "Stroustrup"
* indent-tabs-mode: nil
* End:
* vim: shiftwidth=4 tabstop=8 expandtab
projects / idzebra-moved-to-github.git / blobdiff