X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Fkinput.c;h=aca8fe9a854a04e736b57df9b782d75541f9e319;hp=9f834a9da6699cf87747c075371c5bf054f8eb15;hb=6c9fcd3b5d3108702fa1ffc92dab4ab6060f9a19;hpb=caa7fe057dd8617129577a6725d2c4e3da066857 diff --git a/index/kinput.c b/index/kinput.c index 9f834a9..aca8fe9 100644 --- a/index/kinput.c +++ b/index/kinput.c @@ -1,149 +1,834 @@ -/* - * Copyright (C) 1994-1995, Index Data I/S - * All rights reserved. - * Sebastian Hammer, Adam Dickmeiss - * - * $Log: kinput.c,v $ - * Revision 1.3 1995-09-06 16:11:17 adam - * Option: only one word key per file. - * - * Revision 1.2 1995/09/04 12:33:42 adam - * Various cleanup. YAZ util used instead. - * - * Revision 1.1 1995/09/04 09:10:37 adam - * More work on index add/del/update. - * Merge sort implemented. - * Initial work on z39 server. - * - */ +/* $Id: kinput.c,v 1.64 2005-01-15 19:38:26 adam Exp $ + Copyright (C) 1995-2005 + 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 Zebra; see the file LICENSE.zebra. If not, write to the +Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA +02111-1307, USA. +*/ + +#include +#ifdef WIN32 +#include +#else +#include +#endif #include #include #include -#include #include #include "index.h" #define KEY_SIZE (1+sizeof(struct it_key)) -#define INP_NAME_MAX 8192 +#define INP_NAME_MAX 768 #define INP_BUF_START 60000 #define INP_BUF_ADD 400000 -static int no_diffs = 0; -static int no_updates = 0; -static int no_insertions = 0; -static int no_iterations = 0; -static int read_one (FILE *inf, char *name, char *key) +struct key_file { + int no; /* file no */ + off_t offset; /* file offset */ + unsigned char *buf; /* buffer block */ + size_t buf_size; /* number of read bytes in block */ + size_t chunk; /* number of bytes allocated */ + size_t buf_ptr; /* current position in buffer */ + char *prev_name; /* last word read */ + void *decode_handle; + off_t length; /* length of file */ + /* handler invoked in each read */ + void (*readHandler)(struct key_file *keyp, void *rinfo); + void *readInfo; + Res res; +}; + +void getFnameTmp (Res res, char *fname, int no) +{ + const char *pre; + + pre = res_get_def (res, "keyTmpDir", "."); + sprintf (fname, "%s/key%d.tmp", pre, no); +} + +void extract_get_fname_tmp (ZebraHandle zh, char *fname, int no) +{ + const char *pre; + + pre = res_get_def (zh->res, "keyTmpDir", "."); + sprintf (fname, "%s/key%d.tmp", pre, no); +} + +void key_file_chunk_read (struct key_file *f) { - int c; - int i = 0; - name[0] = 0; - do + int nr = 0, r = 0, fd; + char fname[1024]; + getFnameTmp (f->res, fname, f->no); + fd = open (fname, O_BINARY|O_RDONLY); + + f->buf_ptr = 0; + f->buf_size = 0; + if (fd == -1) { - if ((c=getc(inf)) == EOF) - return 0; - name[i++] = c; - } while (c); - for (i = 0; ilength) + { + if ((f->length = lseek (fd, 0L, SEEK_END)) == (off_t) -1) + { + yaz_log (YLOG_WARN|YLOG_ERRNO, "cannot seek %s", fname); + close (fd); + return ; + } + } + if (lseek (fd, f->offset, SEEK_SET) == -1) + { + yaz_log (YLOG_WARN|YLOG_ERRNO, "cannot seek %s", fname); + close(fd); + return ; + } + while (f->chunk - nr > 0) + { + r = read (fd, f->buf + nr, f->chunk - nr); + if (r <= 0) + break; + nr += r; + } + if (r == -1) + { + yaz_log (YLOG_WARN|YLOG_ERRNO, "read of %s", fname); + close (fd); + return; + } + f->buf_size = nr; + if (f->readHandler) + (*f->readHandler)(f, f->readInfo); + close (fd); +} + +void key_file_destroy (struct key_file *f) +{ + iscz1_stop(f->decode_handle); + xfree (f->buf); + xfree (f->prev_name); + xfree (f); +} + +struct key_file *key_file_init (int no, int chunk, Res res) +{ + struct key_file *f; + + f = (struct key_file *) xmalloc (sizeof(*f)); + f->res = res; + f->decode_handle = iscz1_start(); + f->no = no; + f->chunk = chunk; + f->offset = 0; + f->length = 0; + f->readHandler = NULL; + f->buf = (unsigned char *) xmalloc (f->chunk); + f->prev_name = (char *) xmalloc (INP_NAME_MAX); + *f->prev_name = '\0'; + key_file_chunk_read (f); + return f; +} + +int key_file_getc (struct key_file *f) +{ + if (f->buf_ptr < f->buf_size) + return f->buf[(f->buf_ptr)++]; + if (f->buf_size < f->chunk) + return EOF; + f->offset += f->buf_size; + key_file_chunk_read (f); + if (f->buf_ptr < f->buf_size) + return f->buf[(f->buf_ptr)++]; + else + return EOF; } -static int inp (Dict dict, ISAM isam, const char *name) +int key_file_decode (struct key_file *f) { - FILE *inf; - char *info; - char next_name[INP_NAME_MAX+1]; - char cur_name[INP_NAME_MAX+1]; - int key_buf_size = INP_BUF_START; - int key_buf_ptr; - char *next_key; - char *key_buf; + int c, d; + + c = key_file_getc (f); + switch (c & 192) + { + case 0: + d = c; + break; + case 64: + d = ((c&63) << 8) + (key_file_getc (f) & 0xff); + break; + case 128: + d = ((c&63) << 8) + (key_file_getc (f) & 0xff); + d = (d << 8) + (key_file_getc (f) & 0xff); + break; + default: /* 192 */ + d = ((c&63) << 8) + (key_file_getc (f) & 0xff); + d = (d << 8) + (key_file_getc (f) & 0xff); + d = (d << 8) + (key_file_getc (f) & 0xff); + break; + } + return d; +} + +int key_file_read (struct key_file *f, char *key) +{ + int i, c; + char srcbuf[128]; + const char *src = srcbuf; + char *dst; + int j; + + c = key_file_getc (f); + if (c == 0) + { + strcpy (key, f->prev_name); + i = 1+strlen (key); + } + else if (c == EOF) + return 0; + else + { + i = 0; + key[i++] = c; + while ((key[i++] = key_file_getc (f))) + ; + strcpy (f->prev_name, key); + iscz1_reset(f->decode_handle); + } + c = key_file_getc(f); /* length + insert/delete combined */ + key[i++] = c & 128; + c = c & 127; + for (j = 0; j < c; j++) + srcbuf[j] = key_file_getc(f); + dst = key + i; + iscz1_decode(f->decode_handle, &dst, &src); + return i + sizeof(struct it_key); +} + +struct heap_info { + struct { + struct key_file **file; + char **buf; + } info; + int heapnum; + int *ptr; + int (*cmp)(const void *p1, const void *p2); + struct zebra_register *reg; + ZebraHandle zh; /* only used for raw reading that bypasses the heaps */ + int no_diffs; + int no_updates; + int no_deletions; + int no_insertions; + int no_iterations; +}; + +static struct heap_info *key_heap_malloc() +{ /* malloc and clear it */ + struct heap_info *hi; + hi = (struct heap_info *) xmalloc (sizeof(*hi)); + hi->info.file = 0; + hi->info.buf = 0; + hi->heapnum = 0; + hi->ptr = 0; + hi->zh=0; + hi->no_diffs = 0; + hi->no_diffs = 0; + hi->no_updates = 0; + hi->no_deletions = 0; + hi->no_insertions = 0; + hi->no_iterations = 0; + return hi; +} + +struct heap_info *key_heap_init (int nkeys, + int (*cmp)(const void *p1, const void *p2)) +{ + struct heap_info *hi; + int i; + + hi = key_heap_malloc(); + hi->info.file = (struct key_file **) + xmalloc (sizeof(*hi->info.file) * (1+nkeys)); + hi->info.buf = (char **) xmalloc (sizeof(*hi->info.buf) * (1+nkeys)); + hi->ptr = (int *) xmalloc (sizeof(*hi->ptr) * (1+nkeys)); + hi->cmp = cmp; + for (i = 0; i<= nkeys; i++) + { + hi->ptr[i] = i; + hi->info.buf[i] = (char *) xmalloc (INP_NAME_MAX); + } + return hi; +} + +struct heap_info *key_heap_init_buff ( ZebraHandle zh, + int (*cmp)(const void *p1, const void *p2)) +{ + struct heap_info *hi=key_heap_malloc(); + hi->cmp=cmp; + hi->zh=zh; + return hi; +} + +void key_heap_destroy (struct heap_info *hi, int nkeys) +{ + int i; + yaz_log (YLOG_DEBUG, "key_heap_destroy"); + yaz_log (YLOG_DEBUG, "key_heap_destroy nk=%d",nkeys); + if (!hi->zh) + for (i = 0; i<=nkeys; i++) + xfree (hi->info.buf[i]); - next_key = xmalloc (KEY_SIZE); - key_buf = xmalloc (key_buf_size * (KEY_SIZE)); - if (!(inf = fopen (name, "r"))) + xfree (hi->info.buf); + xfree (hi->ptr); + xfree (hi->info.file); + xfree (hi); +} + +static void key_heap_swap (struct heap_info *hi, int i1, int i2) +{ + int swap; + + swap = hi->ptr[i1]; + hi->ptr[i1] = hi->ptr[i2]; + hi->ptr[i2] = swap; +} + + +static void key_heap_delete (struct heap_info *hi) +{ + int cur = 1, child = 2; + + assert (hi->heapnum > 0); + + key_heap_swap (hi, 1, hi->heapnum); + hi->heapnum--; + while (child <= hi->heapnum) { + if (child < hi->heapnum && + (*hi->cmp)(&hi->info.buf[hi->ptr[child]], + &hi->info.buf[hi->ptr[child+1]]) > 0) + child++; + if ((*hi->cmp)(&hi->info.buf[hi->ptr[cur]], + &hi->info.buf[hi->ptr[child]]) > 0) + { + key_heap_swap (hi, cur, child); + cur = child; + child = 2*cur; + } + else + break; + } +} + +static void key_heap_insert (struct heap_info *hi, const char *buf, int nbytes, + struct key_file *kf) +{ + int cur, parent; + + cur = ++(hi->heapnum); + memcpy (hi->info.buf[hi->ptr[cur]], buf, nbytes); + hi->info.file[hi->ptr[cur]] = kf; + + parent = cur/2; + while (parent && (*hi->cmp)(&hi->info.buf[hi->ptr[parent]], + &hi->info.buf[hi->ptr[cur]]) > 0) { - logf (LOG_FATAL|LOG_ERRNO, "cannot open `%s'", name); - exit (1); + key_heap_swap (hi, cur, parent); + cur = parent; + parent = cur/2; } - read_one (inf, cur_name, key_buf); - while (cur_name[0]) /* EOF ? */ +} + +static int heap_read_one_raw (struct heap_info *hi, char *name, char *key) +{ + ZebraHandle zh=hi->zh; + size_t ptr_i = zh->reg->ptr_i; + char *cp; + if (!ptr_i) + return 0; + --(zh->reg->ptr_i); + cp=(zh->reg->key_buf)[zh->reg->ptr_top - ptr_i]; + yaz_log (YLOG_DEBUG, " raw: i=%ld top=%ld cp=%p", (long) ptr_i, + (long) zh->reg->ptr_top, cp); + strcpy(name, cp); + memcpy(key, cp+strlen(name)+1, KEY_SIZE); + hi->no_iterations++; + return 1; +} + +static int heap_read_one (struct heap_info *hi, char *name, char *key) +{ + int n, r; + char rbuf[INP_NAME_MAX]; + struct key_file *kf; + + if (hi->zh) /* bypass the heap stuff, we have a readymade buffer */ + return heap_read_one_raw(hi, name, key); + + if (!hi->heapnum) + return 0; + n = hi->ptr[1]; + strcpy (name, hi->info.buf[n]); + kf = hi->info.file[n]; + r = strlen(name); + memcpy (key, hi->info.buf[n] + r+1, KEY_SIZE); + key_heap_delete (hi); + if ((r = key_file_read (kf, rbuf))) + key_heap_insert (hi, rbuf, r, kf); + hi->no_iterations++; + return 1; +} + +#define PR_KEY 0 + +#if PR_KEY +static void pkey(const char *b, int mode) +{ + struct it_key *key = (struct it_key *) b; + printf ("%c %d:%d\n", mode + 48, key->sysno, key->seqno); +} +#endif + +struct heap_cread_info { + char prev_name[INP_NAME_MAX]; + char cur_name[INP_NAME_MAX]; + char *key; + char *key_1, *key_2; + int mode_1, mode_2; + int sz_1, sz_2; + struct heap_info *hi; + int first_in_list; + int more; + int ret; +}; + +static int heap_cread_item (void *vp, char **dst, int *insertMode); + +int heap_cread_item2 (void *vp, char **dst, int *insertMode) +{ + struct heap_cread_info *p = (struct heap_cread_info *) vp; + int level = 0; + + if (p->ret == 0) /* lookahead was 0?. Return that in read next round */ + { + p->ret = -1; + return 0; + } + else if (p->ret == -1) /* Must read new item ? */ + { + char *dst_1 = p->key_1; + p->ret = heap_cread_item(vp, &dst_1, &p->mode_1); + p->sz_1 = dst_1 - p->key_1; + } + else + { /* lookahead in 2 . Now in 1. */ + p->sz_1 = p->sz_2; + p->mode_1 = p->mode_2; + memcpy (p->key_1, p->key_2, p->sz_2); + } + if (p->mode_1) + level = 1; /* insert */ + else + level = -1; /* delete */ + while(1) { - int nmemb; - key_buf_ptr = KEY_SIZE; - while (1) + char *dst_2 = p->key_2; + p->ret = heap_cread_item(vp, &dst_2, &p->mode_2); + if (!p->ret) { - if (!read_one (inf, next_name, next_key)) + if (level) break; - if (strcmp (next_name, cur_name)) + p->ret = -1; + return 0; + } + p->sz_2 = dst_2 - p->key_2; + if (p->sz_1 == p->sz_2 && memcmp(p->key_1, p->key_2, p->sz_1) == 0) + { + if (p->mode_2) /* adjust level according to deletes/inserts */ + level++; + else + level--; + } + else + { + if (level) break; - memcpy (key_buf + key_buf_ptr, next_key, KEY_SIZE); - key_buf_ptr += KEY_SIZE; - if (key_buf_ptr+KEY_SIZE >= key_buf_size) + /* all the same. new round .. */ + p->sz_1 = p->sz_2; + p->mode_1 = p->mode_2; + memcpy (p->key_1, p->key_2, p->sz_1); + if (p->mode_1) + level = 1; /* insert */ + else + level = -1; /* delete */ + } + } + /* outcome is insert (1) or delete (0) depending on final level */ + if (level > 0) + *insertMode = 1; + else + *insertMode = 0; + memcpy (*dst, p->key_1, p->sz_1); +#if PR_KEY + printf ("top: "); + pkey(*dst, *insertMode); fflush(stdout); +#endif + (*dst) += p->sz_1; + return 1; +} + +int heap_cread_item (void *vp, char **dst, int *insertMode) +{ + struct heap_cread_info *p = (struct heap_cread_info *) vp; + struct heap_info *hi = p->hi; + + if (p->first_in_list) + { + *insertMode = p->key[0]; + memcpy (*dst, p->key+1, sizeof(struct it_key)); +#if PR_KEY + printf ("sub1: "); + pkey(*dst, *insertMode); +#endif + (*dst) += sizeof(struct it_key); + p->first_in_list = 0; + return 1; + } + strcpy (p->prev_name, p->cur_name); + if (!(p->more = heap_read_one (hi, p->cur_name, p->key))) + return 0; + if (*p->cur_name && strcmp (p->cur_name, p->prev_name)) + { + p->first_in_list = 1; + return 0; + } + *insertMode = p->key[0]; + memcpy (*dst, p->key+1, sizeof(struct it_key)); +#if PR_KEY + printf ("sub2: "); + pkey(*dst, *insertMode); +#endif + (*dst) += sizeof(struct it_key); + return 1; +} + +int heap_inpc (struct heap_info *hi) +{ + struct heap_cread_info hci; + ISAMC_I *isamc_i = (ISAMC_I *) xmalloc (sizeof(*isamc_i)); + + hci.key = (char *) xmalloc (KEY_SIZE); + hci.key_1 = (char *) xmalloc (KEY_SIZE); + hci.key_2 = (char *) xmalloc (KEY_SIZE); + hci.ret = -1; + hci.first_in_list = 1; + hci.hi = hi; + hci.more = heap_read_one (hi, hci.cur_name, hci.key); + + isamc_i->clientData = &hci; + isamc_i->read_item = heap_cread_item2; + + while (hci.more) + { + char this_name[INP_NAME_MAX]; + ISAMC_P isamc_p, isamc_p2; + char *dict_info; + + strcpy (this_name, hci.cur_name); + assert (hci.cur_name[1]); + hi->no_diffs++; + if ((dict_info = dict_lookup (hi->reg->dict, hci.cur_name))) + { + memcpy (&isamc_p, dict_info+1, sizeof(ISAMC_P)); + isamc_p2 = isc_merge (hi->reg->isamc, isamc_p, isamc_i); + if (!isamc_p2) { - char *new_key_buf; - new_key_buf = xmalloc (key_buf_size + INP_BUF_ADD); - memcpy (new_key_buf, key_buf, key_buf_size); - key_buf_size += INP_BUF_ADD; - xfree (key_buf); - key_buf = new_key_buf; + hi->no_deletions++; + if (!dict_delete (hi->reg->dict, this_name)) + abort(); } - } - no_diffs++; - nmemb = key_buf_ptr / KEY_SIZE; - assert (nmemb*KEY_SIZE == key_buf_ptr); - if ((info = dict_lookup (dict, cur_name))) + else + { + hi->no_updates++; + if (isamc_p2 != isamc_p) + dict_insert (hi->reg->dict, this_name, + sizeof(ISAMC_P), &isamc_p2); + } + } + else { - ISAM_P isam_p, isam_p2; - logf (LOG_DEBUG, "updating %s", cur_name); - no_updates++; - memcpy (&isam_p, info+1, sizeof(ISAM_P)); - isam_p2 = is_merge (isam, isam_p, nmemb, key_buf); - if (isam_p2 != isam_p) - dict_insert (dict, cur_name, sizeof(ISAM_P), &isam_p2); + isamc_p = isc_merge (hi->reg->isamc, 0, isamc_i); + hi->no_insertions++; + dict_insert (hi->reg->dict, this_name, sizeof(ISAMC_P), &isamc_p); } + } + xfree (isamc_i); + xfree (hci.key); + xfree (hci.key_1); + xfree (hci.key_2); + return 0; +} + +#if 0 +/* for debugging only */ +static void print_dict_item (ZebraMaps zm, const char *s) +{ + int reg_type = s[1]; + char keybuf[IT_MAX_WORD+1]; + char *to = keybuf; + const char *from = s + 2; + + while (*from) + { + const char *res = zebra_maps_output (zm, reg_type, &from); + if (!res) + *to++ = *from++; + else + while (*res) + *to++ = *res++; + } + *to = '\0'; + yaz_log (YLOG_LOG, "%s", keybuf); +} +#endif + +int heap_inpb (struct heap_info *hi) +{ + struct heap_cread_info hci; + ISAMC_I *isamc_i = (ISAMC_I *) xmalloc (sizeof(*isamc_i)); + + hci.key = (char *) xmalloc (KEY_SIZE); + hci.key_1 = (char *) xmalloc (KEY_SIZE); + hci.key_2 = (char *) xmalloc (KEY_SIZE); + hci.ret = -1; + hci.first_in_list = 1; + hci.hi = hi; + hci.more = heap_read_one (hi, hci.cur_name, hci.key); + + isamc_i->clientData = &hci; + isamc_i->read_item = heap_cread_item2; + + while (hci.more) + { + char this_name[INP_NAME_MAX]; + ISAMC_P isamc_p, isamc_p2; + char *dict_info; + + strcpy (this_name, hci.cur_name); + assert (hci.cur_name[1]); + hi->no_diffs++; + +#if 0 + print_dict_item (hi->reg->zebra_maps, hci.cur_name); +#endif + if ((dict_info = dict_lookup (hi->reg->dict, hci.cur_name))) + { + memcpy (&isamc_p, dict_info+1, sizeof(ISAMC_P)); + isamc_p2 = isamb_merge (hi->reg->isamb, isamc_p, isamc_i); + if (!isamc_p2) + { + hi->no_deletions++; + if (!dict_delete (hi->reg->dict, this_name)) + abort(); + } + else + { + hi->no_updates++; + if (isamc_p2 != isamc_p) + dict_insert (hi->reg->dict, this_name, + sizeof(ISAMC_P), &isamc_p2); + } + } else { - ISAM_P isam_p; - logf (LOG_DEBUG, "inserting %s", cur_name); - no_insertions++; - isam_p = is_merge (isam, 0, nmemb, key_buf); - dict_insert (dict, cur_name, sizeof(ISAM_P), &isam_p); + isamc_p = isamb_merge (hi->reg->isamb, 0, isamc_i); + hi->no_insertions++; + dict_insert (hi->reg->dict, this_name, sizeof(ISAMC_P), &isamc_p); } - memcpy (key_buf, next_key, KEY_SIZE); - strcpy (cur_name, next_name); } - fclose (inf); + xfree (isamc_i); + xfree (hci.key); + xfree (hci.key_1); + xfree (hci.key_2); return 0; +} + +int heap_inps (struct heap_info *hi) +{ + struct heap_cread_info hci; + ISAMS_I isams_i = (ISAMS_I) xmalloc (sizeof(*isams_i)); + + hci.key = (char *) xmalloc (KEY_SIZE); + hci.key_1 = (char *) xmalloc (KEY_SIZE); + hci.key_2 = (char *) xmalloc (KEY_SIZE); + hci.first_in_list = 1; + hci.ret = -1; + hci.hi = hi; + hci.more = heap_read_one (hi, hci.cur_name, hci.key); + + isams_i->clientData = &hci; + isams_i->read_item = heap_cread_item; + + while (hci.more) + { + char this_name[INP_NAME_MAX]; + ISAMS_P isams_p; + char *dict_info; + + strcpy (this_name, hci.cur_name); + assert (hci.cur_name[1]); + hi->no_diffs++; + if (!(dict_info = dict_lookup (hi->reg->dict, hci.cur_name))) + { + isams_p = isams_merge (hi->reg->isams, isams_i); + hi->no_insertions++; + dict_insert (hi->reg->dict, this_name, sizeof(ISAMS_P), &isams_p); + } + else + { + yaz_log (YLOG_FATAL, "isams doesn't support this kind of update"); + break; + } + } + xfree (isams_i); + return 0; +} + +struct progressInfo { + time_t startTime; + time_t lastTime; + off_t totalBytes; + off_t totalOffset; +}; + +void progressFunc (struct key_file *keyp, void *info) +{ + struct progressInfo *p = (struct progressInfo *) info; + time_t now, remaining; + + if (keyp->buf_size <= 0 || p->totalBytes <= 0) + return ; + time (&now); + + if (now >= p->lastTime+10) + { + p->lastTime = now; + remaining = (time_t) ((now - p->startTime)* + ((double) p->totalBytes/p->totalOffset - 1.0)); + if (remaining <= 130) + yaz_log (YLOG_LOG, "Merge %2.1f%% completed; %ld seconds remaining", + (100.0*p->totalOffset) / p->totalBytes, (long) remaining); + else + yaz_log (YLOG_LOG, "Merge %2.1f%% completed; %ld minutes remaining", + (100.0*p->totalOffset) / p->totalBytes, (long) remaining/60); + } + p->totalOffset += keyp->buf_size; } -void key_input (const char *dict_fname, const char *isam_fname, - const char *key_fname, int cache) +#ifndef R_OK +#define R_OK 4 +#endif + +void zebra_index_merge (ZebraHandle zh) { - Dict dict; - ISAM isam; + struct key_file **kf = 0; + char rbuf[1024]; + int i, r; + struct heap_info *hi; + struct progressInfo progressInfo; + int nkeys = zh->reg->key_file_no; + int usefile; + + yaz_log (YLOG_DEBUG, " index_merge called with nk=%d b=%p", + nkeys, zh->reg->key_buf); + if ( (nkeys==0) && (zh->reg->key_buf==0) ) + return; /* nothing to merge - probably flush after end-trans */ + + usefile = (nkeys!=0); - dict = dict_open (dict_fname, cache, 1); - if (!dict) + if (usefile) { - logf (LOG_FATAL, "dict_open fail of `%s'", dict_fname); - exit (1); + if (nkeys < 0) + { + char fname[1024]; + nkeys = 0; + while (1) + { + extract_get_fname_tmp (zh, fname, nkeys+1); + if (access (fname, R_OK) == -1) + break; + nkeys++; + } + if (!nkeys) + return ; + } + kf = (struct key_file **) xmalloc ((1+nkeys) * sizeof(*kf)); + progressInfo.totalBytes = 0; + progressInfo.totalOffset = 0; + time (&progressInfo.startTime); + time (&progressInfo.lastTime); + for (i = 1; i<=nkeys; i++) + { + kf[i] = key_file_init (i, 8192, zh->res); + kf[i]->readHandler = progressFunc; + kf[i]->readInfo = &progressInfo; + progressInfo.totalBytes += kf[i]->length; + progressInfo.totalOffset += kf[i]->buf_size; + } + hi = key_heap_init (nkeys, key_qsort_compare); + hi->reg = zh->reg; + + for (i = 1; i<=nkeys; i++) + if ((r = key_file_read (kf[i], rbuf))) + key_heap_insert (hi, rbuf, r, kf[i]); + } /* use file */ + else + { /* do not use file, read straight from buffer */ + hi = key_heap_init_buff (zh,key_qsort_compare); + hi->reg = zh->reg; } - isam = is_open (isam_fname, key_compare, 1, sizeof(struct it_key)); - if (!isam) + if (zh->reg->isams) + heap_inps (hi); + if (zh->reg->isamc) + heap_inpc (hi); + if (zh->reg->isamb) + heap_inpb (hi); + + if (usefile) { - logf (LOG_FATAL, "is_open fail of `%s'", isam_fname); - exit (1); + for (i = 1; i<=nkeys; i++) + { + extract_get_fname_tmp (zh, rbuf, i); + unlink (rbuf); + } + for (i = 1; i<=nkeys; i++) + key_file_destroy (kf[i]); + xfree (kf); } - inp (dict, isam, key_fname); - dict_close (dict); - is_close (isam); - logf (LOG_LOG, "Iterations . . .%7d", no_iterations); - logf (LOG_LOG, "Distinct words .%7d", no_diffs); - logf (LOG_LOG, "Updates. . . . .%7d", no_updates); - logf (LOG_LOG, "Insertions . . .%7d", no_insertions); + if (hi->no_iterations) + { /* do not log if nothing happened */ + yaz_log (YLOG_LOG, "Iterations . . .%7d", hi->no_iterations); + yaz_log (YLOG_LOG, "Distinct words .%7d", hi->no_diffs); + yaz_log (YLOG_LOG, "Updates. . . . .%7d", hi->no_updates); + yaz_log (YLOG_LOG, "Deletions. . . .%7d", hi->no_deletions); + yaz_log (YLOG_LOG, "Insertions . . .%7d", hi->no_insertions); + } + zh->reg->key_file_no = 0; + + key_heap_destroy (hi, nkeys); }