X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Fextract.c;h=18cac3d597c58577d9346d69cc32ed0805d2f728;hp=09902fdbfdb3a386a209a8ff5e9e780c22b08d49;hb=7c3a0352f0492609a3b6b26b63a72b0b2d207aab;hpb=2a6f64379961f30bfe6376964ce279ab5242da1f diff --git a/index/extract.c b/index/extract.c index 09902fd..18cac3d 100644 --- a/index/extract.c +++ b/index/extract.c @@ -1,4 +1,4 @@ -/* $Id: extract.c,v 1.147 2004-01-22 11:27:21 adam Exp $ +/* $Id: extract.c,v 1.161 2004-09-14 14:38:07 quinn Exp $ Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004 Index Data Aps @@ -20,9 +20,9 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ - #include #include +#include #ifdef WIN32 #include #else @@ -72,7 +72,8 @@ static void logRecord (ZebraHandle zh) ++zh->records_processed; if (!(zh->records_processed % 1000)) { - logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d", + logf (LOG_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); } @@ -92,6 +93,51 @@ static const char **searchRecordKey (ZebraHandle zh, struct recKeys *reckeys, int attrSetS, int attrUseS) { +#if IT_KEY_NEW +/* #error searchRecordKey does not work yet in this mode.. */ + static const char *ws[32]; + void *decode_handle = iscz1_start(); + int off = 0; + int startSeq = -1; + int seqno = 0; + int i; + + for (i = 0; i<32; i++) + ws[i] = NULL; + + while (off < reckeys->buf_used) + { + const char *src = reckeys->buf + off; + struct it_key key; + char *dst = (char*) &key; + int attrSet, attrUse; + + iscz1_decode(decode_handle, &dst, &src); + assert(key.len < 4 && key.len > 2); + + attrSet = (int) key.mem[0]; + attrUse = (int) key.mem[1]; + seqno = (int) key.mem[2]; + + if (attrUseS == attrUse && attrSetS == attrSet) + { + int woff; + + if (startSeq == -1) + startSeq = seqno; + woff = seqno - startSeq; + if (woff >= 0 && woff < 31) + ws[woff] = src; + } + + while (*src++) + ; + off = src - reckeys->buf; + } + iscz1_stop(decode_handle); + assert (off == reckeys->buf_used); + return ws; +#else static const char *ws[32]; int off = 0; int startSeq = -1; @@ -170,6 +216,7 @@ static const char **searchRecordKey (ZebraHandle zh, } assert (off == reckeys->buf_used); return ws; +#endif } struct file_read_info { @@ -414,6 +461,18 @@ struct recordLogInfo { int recordOffset; struct recordGroup *rGroup; }; + +void create_rec_keys_codec(struct recKeys *keys) +{ + keys->buf_used = 0; +#if IT_KEY_NEW + iscz1_reset(keys->codec_handle); +#else + keys->prevAttrUse = -1; + keys->prevAttrSet = -1; + keys->prevSeqNo = 0; +#endif +} static int file_extract_record(ZebraHandle zh, SYSNO *sysno, const char *fname, @@ -423,7 +482,7 @@ static int file_extract_record(ZebraHandle zh, { RecordAttr *recordAttr; int r; - char *matchStr; + const char *matchStr; SYSNO sysnotmp; Record rec; off_t recordOffset = 0; @@ -454,10 +513,8 @@ static int file_extract_record(ZebraHandle zh, /* we are going to read from a file, so prepare the extraction */ int i; - zh->reg->keys.buf_used = 0; - zh->reg->keys.prevAttrUse = -1; - zh->reg->keys.prevAttrSet = -1; - zh->reg->keys.prevSeqNo = 0; + create_rec_keys_codec(&zh->reg->keys); + zh->reg->sortKeys.buf_used = 0; recordOffset = fi->file_moffset; @@ -550,7 +607,10 @@ static int file_extract_record(ZebraHandle zh, { rinfo = dict_lookup (zh->reg->matchDict, matchStr); if (rinfo) + { + assert(*rinfo == sizeof(*sysno)); memcpy (sysno, rinfo+1, sizeof(*sysno)); + } } else { @@ -755,6 +815,7 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname, char ext[128]; char ext_res[128]; struct file_read_info *fi; + const char *original_record_type = 0; if (!zh->m_group || !*zh->m_group) *gprefix = '\0'; @@ -774,6 +835,7 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname, 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); @@ -811,6 +873,7 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname, if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1) { logf (LOG_WARN|LOG_ERRNO, "open %s", full_rep); + zh->m_record_type = original_record_type; return 0; } } @@ -823,6 +886,7 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname, file_read_stop (fi); if (fd != -1) close (fd); + zh->m_record_type = original_record_type; return r; } @@ -837,7 +901,7 @@ int buffer_extract_record (ZebraHandle zh, int delete_flag, int test_mode, const char *recordType, - int *sysno, + SYSNO *sysno, const char *match_criteria, const char *fname, int force_update, @@ -846,7 +910,7 @@ int buffer_extract_record (ZebraHandle zh, RecordAttr *recordAttr; struct recExtractCtrl extractCtrl; int i, r; - char *matchStr = 0; + const char *matchStr = 0; RecType recType = NULL; char subType[1024]; void *clientData; @@ -872,10 +936,8 @@ int buffer_extract_record (ZebraHandle zh, extractCtrl.endf = zebra_record_int_end; extractCtrl.fh = &fc; - zh->reg->keys.buf_used = 0; - zh->reg->keys.prevAttrUse = -1; - zh->reg->keys.prevAttrSet = -1; - zh->reg->keys.prevSeqNo = 0; + create_rec_keys_codec(&zh->reg->keys); + zh->reg->sortKeys.buf_used = 0; if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0])) @@ -953,20 +1015,25 @@ int buffer_extract_record (ZebraHandle zh, if (! *sysno) { char *rinfo; if (match_criteria && *match_criteria) { - matchStr = (char *)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) + { + logf (LOG_WARN, "Bad match criteria (recordID)"); + return 1; + } } } if (matchStr) { rinfo = dict_lookup (zh->reg->matchDict, matchStr); if (rinfo) + { + assert(*rinfo == sizeof(*sysno)); memcpy (sysno, rinfo+1, sizeof(*sysno)); - } else { - logf (LOG_WARN, "Bad match criteria (recordID)"); - return 0; + } } } @@ -1025,7 +1092,7 @@ int buffer_extract_record (ZebraHandle zh, extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys); rec_rm (&rec); logRecord(zh); - return 0; + return -1; } } @@ -1171,10 +1238,8 @@ int explain_extract (void *handle, Record rec, data1_node *n) abort (); } - zh->reg->keys.buf_used = 0; - zh->reg->keys.prevAttrUse = -1; - zh->reg->keys.prevAttrSet = -1; - zh->reg->keys.prevSeqNo = 0; + create_rec_keys_codec(&zh->reg->keys); + zh->reg->sortKeys.buf_used = 0; extractCtrl.init = extract_init; @@ -1225,12 +1290,16 @@ int explain_extract (void *handle, Record rec, data1_node *n) void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno, int cmd, struct recKeys *reckeys) { +#if IT_KEY_NEW + void *decode_handle = iscz1_start(); +#else + int seqno = 0; #if SU_SCHEME #else unsigned char attrSet = (unsigned char) -1; unsigned short attrUse = (unsigned short) -1; #endif - int seqno = 0; +#endif int off = 0; int ch = 0; ZebraExplainInfo zei = zh->reg->zei; @@ -1252,6 +1321,52 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno, zh->reg->key_file_no = 0; } zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1); +#if IT_KEY_NEW + while (off < reckeys->buf_used) + { + const char *src = reckeys->buf + off; + struct it_key key; + char *dst = (char*) &key; + int attrSet, attrUse; + + iscz1_decode(decode_handle, &dst, &src); + assert(key.len < 4 && key.len > 2); + + attrSet = (int) key.mem[0]; + attrUse = (int) key.mem[1]; /* sequence in mem[2] */ + + 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; + + ch = zebraExplain_lookupSU (zei, attrSet, attrUse); + if (ch < 0) + ch = zebraExplain_addSU (zei, attrSet, attrUse); + + assert (ch > 0); + zh->reg->key_buf_used += + key_SU_encode (ch,((char*)zh->reg->key_buf) + + zh->reg->key_buf_used); + while (*src) + ((char*)zh->reg->key_buf) [(zh->reg->key_buf_used)++] = *src++; + src++; + ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = '\0'; + ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = cmd; + + key.len = 2; + key.mem[0] = sysno; + key.mem[1] = key.mem[2]; /* sequence .. */ + + memcpy ((char*)zh->reg->key_buf + zh->reg->key_buf_used, + &key, sizeof(key)); + (zh->reg->key_buf_used) += sizeof(key); + off = src - reckeys->buf; + } +#else while (off < reckeys->buf_used) { const char *src = reckeys->buf + off; @@ -1280,8 +1395,10 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno, #endif if (zh->reg->key_buf_used + 1024 > (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*)) - extract_flushWriteKeys (zh); + extract_flushWriteKeys (zh,0); + assert(zh->reg->ptr_i >= 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; #if SU_SCHEME @@ -1314,27 +1431,73 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno, (zh->reg->key_buf_used) += sizeof(key); off = src - reckeys->buf; } +#endif assert (off == reckeys->buf_used); +#if IT_KEY_NEW + iscz1_stop(decode_handle); +#endif } -void extract_flushWriteKeys (ZebraHandle zh) +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) + { + logf (LOG_DEBUG, " nothing to flush section=%d buf=%p i=%d", + zh->reg->key_file_no, zh->reg->key_buf, ptr_i); + logf (LOG_DEBUG, " buf=%p ", + zh->reg->key_buf); + logf (LOG_DEBUG, " ptr=%d ",zh->reg->ptr_i); + logf (LOG_DEBUG, " reg=%p ",zh->reg); + return; + } (zh->reg->key_file_no)++; logf (LOG_LOG, "sorting section %d", (zh->reg->key_file_no)); + logf (LOG_DEBUG, " 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') { + logf (LOG_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"))) @@ -1409,18 +1572,22 @@ void extract_flushWriteKeys (ZebraHandle zh) zh->reg->key_buf_used = 0; } -void extract_add_index_string (RecWord *p, const char *string, - int length) +void extract_add_index_string (RecWord *p, const char *str, int length) { char *dst; + ZebraHandle zh = p->extractCtrl->handle; + struct recKeys *keys = &zh->reg->keys; +#if IT_KEY_NEW + struct it_key key; + const char *src = (char*) &key; +#else unsigned char attrSet; unsigned short attrUse; int lead = 0; int diff = 0; int *pseqno = &p->seqno; - ZebraHandle zh = p->extractCtrl->handle; ZebraExplainInfo zei = zh->reg->zei; - struct recKeys *keys = &zh->reg->keys; +#endif if (keys->buf_used+1024 > keys->buf_max) { @@ -1434,6 +1601,29 @@ void extract_add_index_string (RecWord *p, const char *string, } dst = keys->buf + keys->buf_used; +#if IT_KEY_NEW + key.len = 3; + key.mem[0] = p->attrSet; + key.mem[1] = p->attrUse; + key.mem[2] = p->seqno; + +#if 0 + /* just for debugging .. */ + yaz_log(LOG_LOG, "set=%d use=%d seqno=%d", p->attrSet, p->attrUse, + p->seqno); +#endif + + iscz1_encode(keys->codec_handle, &dst, &src); + + *dst++ = p->reg_type; + memcpy (dst, str, length); + dst += length; + *dst++ = '\0'; +#else + /* leader byte is encoded as follows: + bit 0 : 1 if attrset is unchanged; 0 if attrset is changed + bit 1 : 1 if attruse is unchanged; 0 if attruse is changed + */ attrSet = p->attrSet; if (keys->buf_used > 0 && keys->prevAttrSet == attrSet) lead |= 1; @@ -1444,13 +1634,13 @@ void extract_add_index_string (RecWord *p, const char *string, lead |= 2; else keys->prevAttrUse = attrUse; -#if 1 + diff = 1 + *pseqno - keys->prevSeqNo; if (diff >= 1 && diff <= 15) lead |= (diff << 2); else diff = 0; -#endif + keys->prevSeqNo = *pseqno; *dst++ = lead; @@ -1482,7 +1672,7 @@ void extract_add_index_string (RecWord *p, const char *string, } #endif *dst++ = p->reg_type; - memcpy (dst, string, length); + memcpy (dst, str, length); dst += length; *dst++ = '\0'; @@ -1491,10 +1681,11 @@ void extract_add_index_string (RecWord *p, const char *string, memcpy (dst, pseqno, sizeof(*pseqno)); dst += sizeof(*pseqno); } +#endif keys->buf_used = dst - keys->buf; } -static void extract_add_sort_string (RecWord *p, const char *string, +static void extract_add_sort_string (RecWord *p, const char *str, int length) { ZebraHandle zh = p->extractCtrl->handle; @@ -1527,7 +1718,7 @@ static void extract_add_sort_string (RecWord *p, const char *string, off += key_SU_encode(p->attrSet, sk->buf + off); off += key_SU_encode(p->attrUse, sk->buf + off); off += key_SU_encode(length, sk->buf + off); - memcpy (sk->buf + off, string, length); + memcpy (sk->buf + off, str, length); sk->buf_used = off + length; } @@ -1547,7 +1738,7 @@ static void extract_add_incomplete_field (RecWord *p) const char **map = 0; if (remain > 0) - map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain); + map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0); while (map) { @@ -1559,7 +1750,7 @@ static void extract_add_incomplete_field (RecWord *p) { remain = p->length - (b - p->string); if (remain > 0) - map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain); + map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0); else map = 0; } @@ -1574,7 +1765,7 @@ static void extract_add_incomplete_field (RecWord *p) buf[i++] = *(cp++); remain = p->length - (b - p->string); if (remain > 0) - map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain); + map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0); else map = 0; } @@ -1591,17 +1782,24 @@ static void extract_add_complete_field (RecWord *p) char buf[IT_MAX_WORD+1]; const char **map = 0; int i = 0, remain = p->length; + int first; /* first position */ + +yaz_log(LOG_DEBUG, "Complete field, w='%s'", p->string); if (remain > 0) - map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain); + map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain, 1); while (remain > 0 && i < IT_MAX_WORD) { while (map && *map && **map == *CHR_SPACE) { remain = p->length - (b - p->string); + if (remain > 0) - map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain); + { + first = i ? 0 : 1; + map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, first); + } else map = 0; } @@ -1614,14 +1812,24 @@ static void extract_add_complete_field (RecWord *p) { const char *cp = *map; - if (i >= IT_MAX_WORD) - break; - while (i < IT_MAX_WORD && *cp) - buf[i++] = *(cp++); + if (**map == *CHR_CUT) + { + i = 0; + } + else + { + if (i >= IT_MAX_WORD) + break; +yaz_log(LOG_DEBUG, "Adding string to index '%d'", *map); + while (i < IT_MAX_WORD && *cp) + buf[i++] = *(cp++); + } remain = p->length - (b - p->string); if (remain > 0) + { map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, - remain); + remain, 0); + } else map = 0; } @@ -1692,8 +1900,13 @@ void encode_key_init (struct encode_info *i) i->prevseq=0; i->prevcmd=-1; i->keylen=0; +#if IT_KEY_NEW + i->encode_handle = iscz1_start(); +#endif } +#if IT_KEY_NEW +#else char *encode_key_int (int d, char *bp) { if (d <= 63) @@ -1718,6 +1931,8 @@ char *encode_key_int (int d, char *bp) } return bp; } +#endif + #define OLDENCODE 1 #ifdef OLDENCODE @@ -1728,11 +1943,19 @@ char *encode_key_int (int d, char *bp) void encode_key_write (char *k, struct encode_info *i, FILE *outf) { struct it_key key; - char *bp = i->buf; + char *bp = i->buf, *bp0; + const char *src = (char *) &key; + /* copy term to output buf */ while ((*bp++ = *k++)) ; - memcpy (&key, k+1, sizeof(struct it_key)); + /* and copy & align key so we can mangle */ + memcpy (&key, k+1, sizeof(struct it_key)); /* *k is insert/delete */ +#if IT_KEY_NEW + bp0 = bp++; + iscz1_encode(i->encode_handle, &bp, &src); + *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */ +#else bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp); if (i->sysno != key.sysno) { @@ -1744,6 +1967,7 @@ void encode_key_write (char *k, struct encode_info *i, FILE *outf) bp = encode_key_int (key.seqno - i->seqno, bp); i->seqno = key.seqno; i->cmd = *k; +#endif if (fwrite (i->buf, bp - i->buf, 1, outf) != 1) { logf (LOG_FATAL|LOG_ERRNO, "fwrite"); @@ -1753,6 +1977,9 @@ void encode_key_write (char *k, struct encode_info *i, FILE *outf) void encode_key_flush (struct encode_info *i, FILE *outf) { /* dummy routine */ +#if IT_KEY_NEW + iscz1_stop(i->encode_handle); +#endif } #else