X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fkinput.c;h=23d9366b6548da1201f2a81ab00ad4c842ddc68e;hb=44082770642275dd993aa621d9dc5c5b0dc454d3;hp=72eef0915bdcc79dd4c8ea1a958c46329fe369d1;hpb=3d38b3e2b49a3f87882d7b564fa845b9b36ac189;p=idzebra-moved-to-github.git diff --git a/index/kinput.c b/index/kinput.c index 72eef09..23d9366 100644 --- a/index/kinput.c +++ b/index/kinput.c @@ -1,6 +1,6 @@ -/* $Id: kinput.c,v 1.59 2004-06-15 10:56:31 adam Exp $ - Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002 - Index Data Aps +/* $Id: kinput.c,v 1.66 2005-06-14 20:28:54 adam Exp $ + Copyright (C) 1995-2005 + Index Data ApS This file is part of the Zebra server. @@ -19,13 +19,12 @@ 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 +#endif +#if HAVE_UNISTD_H #include #endif #include @@ -49,8 +48,7 @@ struct key_file { size_t chunk; /* number of bytes allocated */ size_t buf_ptr; /* current position in buffer */ char *prev_name; /* last word read */ - int sysno; /* last sysno */ - int seqno; /* last seqno */ + void *decode_handle; off_t length; /* length of file */ /* handler invoked in each read */ void (*readHandler)(struct key_file *keyp, void *rinfo); @@ -85,21 +83,21 @@ void key_file_chunk_read (struct key_file *f) f->buf_size = 0; if (fd == -1) { - logf (LOG_WARN|LOG_ERRNO, "cannot open %s", fname); + yaz_log (YLOG_WARN|YLOG_ERRNO, "cannot open %s", fname); return ; } if (!f->length) { if ((f->length = lseek (fd, 0L, SEEK_END)) == (off_t) -1) { - logf (LOG_WARN|LOG_ERRNO, "cannot seek %s", fname); + yaz_log (YLOG_WARN|YLOG_ERRNO, "cannot seek %s", fname); close (fd); return ; } } if (lseek (fd, f->offset, SEEK_SET) == -1) { - logf (LOG_WARN|LOG_ERRNO, "cannot seek %s", fname); + yaz_log (YLOG_WARN|YLOG_ERRNO, "cannot seek %s", fname); close(fd); return ; } @@ -112,7 +110,7 @@ void key_file_chunk_read (struct key_file *f) } if (r == -1) { - logf (LOG_WARN|LOG_ERRNO, "read of %s", fname); + yaz_log (YLOG_WARN|YLOG_ERRNO, "read of %s", fname); close (fd); return; } @@ -124,6 +122,7 @@ void key_file_chunk_read (struct key_file *f) void key_file_destroy (struct key_file *f) { + iscz1_stop(f->decode_handle); xfree (f->buf); xfree (f->prev_name); xfree (f); @@ -135,8 +134,7 @@ struct key_file *key_file_init (int no, int chunk, Res res) f = (struct key_file *) xmalloc (sizeof(*f)); f->res = res; - f->sysno = 0; - f->seqno = 0; + f->decode_handle = iscz1_start(); f->no = no; f->chunk = chunk; f->offset = 0; @@ -191,8 +189,11 @@ int key_file_decode (struct key_file *f) int key_file_read (struct key_file *f, char *key) { - int i, d, c; - struct it_key itkey; + int i, c; + char srcbuf[128]; + const char *src = srcbuf; + char *dst; + int j; c = key_file_getc (f); if (c == 0) @@ -209,22 +210,16 @@ int key_file_read (struct key_file *f, char *key) while ((key[i++] = key_file_getc (f))) ; strcpy (f->prev_name, key); - f->sysno = 0; + iscz1_reset(f->decode_handle); } - d = key_file_decode (f); - key[i++] = d & 1; - d = d >> 1; - itkey.sysno = d + f->sysno; - if (d) - { - f->sysno = itkey.sysno; - f->seqno = 0; - } - d = key_file_decode (f); - itkey.seqno = d + f->seqno; - f->seqno = itkey.seqno; - memcpy (key + i, &itkey, sizeof(struct it_key)); - return i + sizeof (struct it_key); + 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 { @@ -294,8 +289,8 @@ struct heap_info *key_heap_init_buff ( ZebraHandle zh, void key_heap_destroy (struct heap_info *hi, int nkeys) { int i; - yaz_log (LOG_DEBUG, "key_heap_destroy"); - yaz_log (LOG_DEBUG, "key_heap_destroy nk=%d",nkeys); + 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]); @@ -369,7 +364,7 @@ static int heap_read_one_raw (struct heap_info *hi, char *name, char *key) return 0; --(zh->reg->ptr_i); cp=(zh->reg->key_buf)[zh->reg->ptr_top - ptr_i]; - logf (LOG_DEBUG, " raw: i=%ld top=%ld cp=%p", (long) 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); @@ -552,7 +547,7 @@ int heap_inpc (struct heap_info *hi) while (hci.more) { char this_name[INP_NAME_MAX]; - ISAMC_P isamc_p, isamc_p2; + ISAM_P isamc_p, isamc_p2; char *dict_info; strcpy (this_name, hci.cur_name); @@ -560,8 +555,9 @@ int heap_inpc (struct heap_info *hi) 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); + memcpy (&isamc_p, dict_info+1, sizeof(ISAM_P)); + isamc_p2 = isamc_p; + isamc_merge (hi->reg->isamc, &isamc_p2, isamc_i); if (!isamc_p2) { hi->no_deletions++; @@ -573,14 +569,15 @@ int heap_inpc (struct heap_info *hi) hi->no_updates++; if (isamc_p2 != isamc_p) dict_insert (hi->reg->dict, this_name, - sizeof(ISAMC_P), &isamc_p2); + sizeof(ISAM_P), &isamc_p2); } } else { - isamc_p = isc_merge (hi->reg->isamc, 0, isamc_i); + isamc_p = 0; + isamc_merge (hi->reg->isamc, &isamc_p, isamc_i); hi->no_insertions++; - dict_insert (hi->reg->dict, this_name, sizeof(ISAMC_P), &isamc_p); + dict_insert (hi->reg->dict, this_name, sizeof(ISAM_P), &isamc_p); } } xfree (isamc_i); @@ -609,7 +606,7 @@ static void print_dict_item (ZebraMaps zm, const char *s) *to++ = *res++; } *to = '\0'; - yaz_log (LOG_LOG, "%s", keybuf); + yaz_log (YLOG_LOG, "%s", keybuf); } #endif @@ -632,7 +629,7 @@ int heap_inpb (struct heap_info *hi) while (hci.more) { char this_name[INP_NAME_MAX]; - ISAMC_P isamc_p, isamc_p2; + ISAM_P isamc_p, isamc_p2; char *dict_info; strcpy (this_name, hci.cur_name); @@ -644,8 +641,9 @@ int heap_inpb (struct heap_info *hi) #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); + memcpy (&isamc_p, dict_info+1, sizeof(ISAM_P)); + isamc_p2 = isamc_p; + isamb_merge (hi->reg->isamb, &isamc_p2, isamc_i); if (!isamc_p2) { hi->no_deletions++; @@ -657,14 +655,15 @@ int heap_inpb (struct heap_info *hi) hi->no_updates++; if (isamc_p2 != isamc_p) dict_insert (hi->reg->dict, this_name, - sizeof(ISAMC_P), &isamc_p2); + sizeof(ISAM_P), &isamc_p2); } } else { - isamc_p = isamb_merge (hi->reg->isamb, 0, isamc_i); + isamc_p = 0; + isamb_merge (hi->reg->isamb, &isamc_p, isamc_i); hi->no_insertions++; - dict_insert (hi->reg->dict, this_name, sizeof(ISAMC_P), &isamc_p); + dict_insert (hi->reg->dict, this_name, sizeof(ISAM_P), &isamc_p); } } xfree (isamc_i); @@ -674,158 +673,6 @@ int heap_inpb (struct heap_info *hi) return 0; } -int heap_inpd (struct heap_info *hi) -{ - struct heap_cread_info hci; - ISAMD_I isamd_i = (ISAMD_I) xmalloc (sizeof(*isamd_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); - - isamd_i->clientData = &hci; - isamd_i->read_item = heap_cread_item; - - while (hci.more) - { - char this_name[INP_NAME_MAX]; - char *dict_info; - char dictentry[ISAMD_MAX_DICT_LEN+1]; - char dictlen; - - strcpy (this_name, hci.cur_name); - - /* print_dict_item (hi->reg->zebra_maps, hci.cur_name); */ - /*!*/ /* FIXME: depend on isamd-debug */ - - assert (hci.cur_name[1]); - hi->no_diffs++; - if ((dict_info = dict_lookup (hi->reg->dict, hci.cur_name))) - { - dictlen=dict_info[0]; - memcpy (dictentry, dict_info+1, dictlen ); -#ifdef SKIPTHIS - logf(LOG_LOG,"dictentry before. len=%d: %d %d %d %d %d %d %d %d %d", - dictlen,dictentry[0], dictentry[1], dictentry[2], - dictentry[3], dictentry[4], dictentry[5], - dictentry[6], dictentry[7], dictentry[8]); /*!*/ -#endif - dictlen= isamd_append(hi->reg->isamd, dictentry, dictlen, isamd_i); - /* logf dictentry after */ - if (dictlen) - { - hi->no_updates++; - if ( (dictlen!=dict_info[0]) || - (0!=memcmp(dictentry, dict_info+1, dictlen)) ) - { - dict_insert(hi->reg->dict, this_name, - dictlen,dictentry); - } - } - else - { - hi->no_deletions++; - if (!dict_delete (hi->reg->dict, this_name)) - { - logf (LOG_FATAL, "dict_delete failed"); - abort(); - } - } - } - else - { - dictlen=0; - memset (dictentry, '\0', ISAMD_MAX_DICT_LEN); - dictlen= isamd_append(hi->reg->isamd, dictentry, dictlen, isamd_i); - /* logf dictentry first */ - hi->no_insertions++; - if (dictlen) - dict_insert(hi->reg->dict, this_name, - dictlen,dictentry); - } - } - xfree (isamd_i); - xfree (hci.key); - xfree (hci.key_1); - xfree (hci.key_2); - return 0; -} - -int heap_inp (struct heap_info *hi) -{ - char *info; - char next_name[INP_NAME_MAX]; - char cur_name[INP_NAME_MAX]; - int key_buf_size = INP_BUF_START; - int key_buf_ptr; - char *next_key; - char *key_buf; - int more; - - next_key = (char *) xmalloc (KEY_SIZE); - key_buf = (char *) xmalloc (key_buf_size); - more = heap_read_one (hi, cur_name, key_buf); - while (more) /* EOF ? */ - { - int nmemb; - key_buf_ptr = KEY_SIZE; - while (1) - { - if (!(more = heap_read_one (hi, next_name, next_key))) - break; - if (*next_name && strcmp (next_name, cur_name)) - break; - memcpy (key_buf + key_buf_ptr, next_key, KEY_SIZE); - key_buf_ptr += KEY_SIZE; - if (key_buf_ptr+(int) KEY_SIZE >= key_buf_size) - { - char *new_key_buf; - new_key_buf = (char *) 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_diffs++; - nmemb = key_buf_ptr / KEY_SIZE; - assert (nmemb * (int) KEY_SIZE == key_buf_ptr); - if ((info = dict_lookup (hi->reg->dict, cur_name))) - { - ISAM_P isam_p, isam_p2; - memcpy (&isam_p, info+1, sizeof(ISAM_P)); - isam_p2 = is_merge (hi->reg->isam, isam_p, nmemb, key_buf); - if (!isam_p2) - { - hi->no_deletions++; - if (!dict_delete (hi->reg->dict, cur_name)) - abort (); - } - else - { - hi->no_updates++; - if (isam_p2 != isam_p) - dict_insert (hi->reg->dict, cur_name, - sizeof(ISAM_P), &isam_p2); - } - } - else - { - ISAM_P isam_p; - hi->no_insertions++; - isam_p = is_merge (hi->reg->isam, 0, nmemb, key_buf); - dict_insert (hi->reg->dict, cur_name, sizeof(ISAM_P), &isam_p); - } - memcpy (key_buf, next_key, KEY_SIZE); - strcpy (cur_name, next_name); - } - return 0; -} - int heap_inps (struct heap_info *hi) { struct heap_cread_info hci; @@ -845,7 +692,7 @@ int heap_inps (struct heap_info *hi) while (hci.more) { char this_name[INP_NAME_MAX]; - ISAMS_P isams_p; + ISAM_P isams_p; char *dict_info; strcpy (this_name, hci.cur_name); @@ -855,11 +702,11 @@ int heap_inps (struct heap_info *hi) { isams_p = isams_merge (hi->reg->isams, isams_i); hi->no_insertions++; - dict_insert (hi->reg->dict, this_name, sizeof(ISAMS_P), &isams_p); + dict_insert (hi->reg->dict, this_name, sizeof(ISAM_P), &isams_p); } else { - logf (LOG_FATAL, "isams doesn't support this kind of update"); + yaz_log (YLOG_FATAL, "isams doesn't support this kind of update"); break; } } @@ -889,10 +736,10 @@ void progressFunc (struct key_file *keyp, void *info) remaining = (time_t) ((now - p->startTime)* ((double) p->totalBytes/p->totalOffset - 1.0)); if (remaining <= 130) - logf (LOG_LOG, "Merge %2.1f%% completed; %ld seconds remaining", + yaz_log (YLOG_LOG, "Merge %2.1f%% completed; %ld seconds remaining", (100.0*p->totalOffset) / p->totalBytes, (long) remaining); else - logf (LOG_LOG, "Merge %2.1f%% completed; %ld minutes remaining", + 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; @@ -912,7 +759,7 @@ void zebra_index_merge (ZebraHandle zh) int nkeys = zh->reg->key_file_no; int usefile; - logf (LOG_DEBUG, " index_merge called with nk=%d b=%p", + 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 */ @@ -964,10 +811,6 @@ void zebra_index_merge (ZebraHandle zh) heap_inps (hi); if (zh->reg->isamc) heap_inpc (hi); - if (zh->reg->isam) - heap_inp (hi); - if (zh->reg->isamd) - heap_inpd (hi); if (zh->reg->isamb) heap_inpb (hi); @@ -984,15 +827,13 @@ void zebra_index_merge (ZebraHandle zh) } if (hi->no_iterations) { /* do not log if nothing happened */ - logf (LOG_LOG, "Iterations . . .%7d", hi->no_iterations); - logf (LOG_LOG, "Distinct words .%7d", hi->no_diffs); - logf (LOG_LOG, "Updates. . . . .%7d", hi->no_updates); - logf (LOG_LOG, "Deletions. . . .%7d", hi->no_deletions); - logf (LOG_LOG, "Insertions . . .%7d", hi->no_insertions); + 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); } - -