X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Fextract.c;h=b1269a0ad93a9e82cd436ecee5e8081522023706;hp=5055e72245e3d091adbdd1dcf39afe876c6868a1;hb=c33ea56e3771c3b80ba66ef8fda3a09cad171ebb;hpb=3a417319ed74c6e699e497b2355a2eda2cc18820 diff --git a/index/extract.c b/index/extract.c index 5055e72..8189504 100644 --- a/index/extract.c +++ b/index/extract.c @@ -1,717 +1,248 @@ -/* - * Copyright (C) 1994-1995, Index Data I/S - * All rights reserved. - * Sebastian Hammer, Adam Dickmeiss - * - * $Log: extract.c,v $ - * Revision 1.52 1996-04-25 13:27:57 adam - * Function recordExtract modified so that files with no keys (possibly empty) - * are ignored. - * - * Revision 1.51 1996/03/19 11:08:42 adam - * Bug fix: Log preamble wasn't always turned off after recordExtract. - * - * Revision 1.50 1996/02/12 18:45:36 adam - * New fileVerboseFlag in record group control. - * - * Revision 1.49 1996/02/05 12:29:57 adam - * Logging reduced a bit. - * The remaining running time is estimated during register merge. - * - * Revision 1.48 1996/02/01 20:53:26 adam - * The temporary per-record keys are compacted a little, and duplication - * of the per-records keys are avoided when they are saved in the record - * information buffer. - * - * Revision 1.47 1996/01/17 14:57:48 adam - * Prototype changed for reader functions in extract/retrieve. File - * is identified by 'void *' instead of 'int. - * - * Revision 1.46 1995/12/15 14:57:16 adam - * Bug fix. - * - * Revision 1.45 1995/12/15 12:37:41 adam - * In addRecordKeyAny: Writes key only when attrSet != -1. - * - * Revision 1.44 1995/12/12 16:00:54 adam - * System call sync(2) used after update/commit. - * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK - * and F_RDLCK. - * - * Revision 1.43 1995/12/11 09:12:46 adam - * The rec_get function returns NULL if record doesn't exist - will - * happen in the server if the result set records have been deleted since - * the creation of the set (i.e. the search). - * The server saves a result temporarily if it is 'volatile', i.e. the - * set is register dependent. - * - * Revision 1.42 1995/12/07 17:38:46 adam - * Work locking mechanisms for concurrent updates/commit. - * - * Revision 1.41 1995/12/06 16:06:42 adam - * Better diagnostics. Work on 'real' dictionary deletion. - * - * Revision 1.40 1995/12/05 16:57:40 adam - * More work on regular patterns. - * - * Revision 1.39 1995/12/05 13:20:18 adam - * Bug fix: file_read sometimes returned early EOF. - * - * Revision 1.38 1995/12/04 17:59:21 adam - * More work on regular expression conversion. - * - * Revision 1.37 1995/12/04 14:22:27 adam - * Extra arg to recType_byName. - * Started work on new regular expression parsed input to - * structured records. - * - * Revision 1.36 1995/11/30 08:34:29 adam - * Started work on commit facility. - * Changed a few malloc/free to xmalloc/xfree. - * - * Revision 1.35 1995/11/28 14:26:21 adam - * Bug fix: recordId with constant wasn't right. - * Bug fix: recordId dictionary entry wasn't deleted when needed. - * - * Revision 1.34 1995/11/28 09:09:38 adam - * Zebra config renamed. - * Use setting 'recordId' to identify record now. - * Bug fix in recindex.c: rec_release_blocks was invokeded even - * though the blocks were already released. - * File traversal properly deletes records when needed. - * - * Revision 1.33 1995/11/27 09:56:20 adam - * Record info elements better enumerated. Internal store of records. - * - * Revision 1.32 1995/11/25 10:24:05 adam - * More record fields - they are enumerated now. - * New options: flagStoreData flagStoreKey. - * - * Revision 1.31 1995/11/24 11:31:35 adam - * Commands add & del read filenames from stdin if source directory is - * empty. - * Match criteria supports 'constant' strings. - * - * Revision 1.30 1995/11/22 17:19:16 adam - * Record management uses the bfile system. - * - * Revision 1.29 1995/11/21 15:01:14 adam - * New general match criteria implemented. - * New feature: document groups. - * - * Revision 1.28 1995/11/21 09:20:30 adam - * Yet more work on record match. - * - * Revision 1.27 1995/11/20 16:59:45 adam - * New update method: the 'old' keys are saved for each records. - * - * Revision 1.26 1995/11/20 11:56:24 adam - * Work on new traversal. - * - * Revision 1.25 1995/11/16 15:34:54 adam - * Uses new record management system in both indexer and server. - * - * Revision 1.24 1995/11/15 19:13:08 adam - * Work on record management. - * - * Revision 1.23 1995/10/27 14:00:10 adam - * Implemented detection of database availability. - * - * Revision 1.22 1995/10/17 18:02:07 adam - * New feature: databases. Implemented as prefix to words in dictionary. - * - * Revision 1.21 1995/10/10 12:24:38 adam - * Temporary sort files are compressed. - * - * Revision 1.20 1995/10/06 13:52:05 adam - * Bug fixes. Handler may abort further scanning. - * - * Revision 1.19 1995/10/04 12:55:16 adam - * Bug fix in ranked search. Use=Any keys inserted. - * - * Revision 1.18 1995/10/04 09:37:08 quinn - * Fixed bug. - * - * Revision 1.17 1995/10/03 14:28:57 adam - * Buffered read in extract works. - * - * Revision 1.16 1995/10/03 14:28:45 adam - * Work on more effecient read handler in extract. - * - * Revision 1.15 1995/10/02 15:42:53 adam - * Extract uses file descriptors instead of FILE pointers. - * - * Revision 1.14 1995/10/02 15:29:13 adam - * More logging in file_extract. - * - * Revision 1.13 1995/09/29 14:01:39 adam - * Bug fixes. - * - * Revision 1.12 1995/09/28 14:22:56 adam - * Sort uses smaller temporary files. - * - * Revision 1.11 1995/09/28 12:10:31 adam - * Bug fixes. Field prefix used in queries. - * - * Revision 1.10 1995/09/28 09:19:41 adam - * xfree/xmalloc used everywhere. - * Extract/retrieve method seems to work for text records. - * - * Revision 1.9 1995/09/27 12:22:28 adam - * More work on extract in record control. - * Field name is not in isam keys but in prefix in dictionary words. - * - * Revision 1.8 1995/09/14 07:48:22 adam - * Record control management. - * - * Revision 1.7 1995/09/11 13:09:32 adam - * More work on relevance feedback. - * - * Revision 1.6 1995/09/08 14:52:27 adam - * Minor changes. Dictionary is lower case now. - * - * Revision 1.5 1995/09/06 16:11:16 adam - * Option: only one word key per file. - * - * Revision 1.4 1995/09/05 15:28:39 adam - * More work on search engine. - * - * Revision 1.3 1995/09/04 12:33:41 adam - * Various cleanup. YAZ util used instead. - * - * Revision 1.2 1995/09/04 09:10:34 adam - * More work on index add/del/update. - * Merge sort implemented. - * Initial work on z39 server. - * - * Revision 1.1 1995/09/01 14:06:35 adam - * Split of work into more files. - * - */ -#include -#include -#include -#include -#include - -#include -#include -#include "index.h" - -#include "recindex.h" +/* $Id: extract.c,v 1.229 2006-09-08 14:40:52 adam Exp $ + Copyright (C) 1995-2006 + Index Data ApS -static Dict matchDict; +This file is part of the Zebra server. -static Records records = NULL; +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. -static char **key_buf; -static size_t ptr_top; -static size_t ptr_i; -static size_t key_buf_used; -static int key_file_no; +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. -static int records_inserted = 0; -static int records_updated = 0; -static int records_deleted = 0; +You should have received a copy of the GNU General Public License +along with this program; if not, write to the Free Software +Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA -void key_open (int mem) -{ - if (mem < 50000) - mem = 50000; - key_buf = xmalloc (mem); - ptr_top = mem/sizeof(char*); - ptr_i = 0; +*/ - key_buf_used = 0; - key_file_no = 0; +#include +#include +#include +#ifdef WIN32 +#include +#endif +#if HAVE_UNISTD_H +#include +#endif +#include - if (!(matchDict = dict_open (GMATCH_DICT, 50, 1))) - { - logf (LOG_FATAL, "dict_open fail of %s", GMATCH_DICT); - exit (1); - } - assert (!records); - records = rec_open (1); -} +#include "index.h" +#include "orddict.h" +#include +#include +#define ENCODE_BUFLEN 768 struct encode_info { - int sysno; - int seqno; - char buf[512]; + void *encode_handle; + void *decode_handle; + char buf[ENCODE_BUFLEN]; }; -void encode_key_init (struct encode_info *i) -{ - i->sysno = 0; - i->seqno = 0; -} - -char *encode_key_int (int d, char *bp) -{ - if (d <= 63) - *bp++ = d; - else if (d <= 16383) - { - *bp++ = 64 + (d>>8); - *bp++ = d & 255; - } - else if (d <= 4194303) - { - *bp++ = 128 + (d>>16); - *bp++ = (d>>8) & 255; - *bp++ = d & 255; - } - else - { - *bp++ = 192 + (d>>24); - *bp++ = (d>>16) & 255; - *bp++ = (d>>8) & 255; - *bp++ = d & 255; - } - return bp; -} +static int log_level = 0; +static int log_level_initialized = 0; -void encode_key_write (char *k, struct encode_info *i, FILE *outf) +static void zebra_init_log_level() { - struct it_key key; - char *bp = i->buf; - - while ((*bp++ = *k++)) - ; - memcpy (&key, k+1, sizeof(struct it_key)); - bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp); - if (i->sysno != key.sysno) - { - i->sysno = key.sysno; - i->seqno = 0; - } - bp = encode_key_int (key.seqno - i->seqno, bp); - i->seqno = key.seqno; - if (fwrite (i->buf, bp - i->buf, 1, outf) != 1) + if (!log_level_initialized) { - logf (LOG_FATAL|LOG_ERRNO, "fwrite"); - exit (1); + log_level = yaz_log_module_level("extract"); + log_level_initialized = 1; } } -void key_flush (void) -{ - FILE *outf; - char out_fname[200]; - char *prevcp, *cp; - struct encode_info encode_info; - - if (ptr_i <= 0) - return; - - key_file_no++; - logf (LOG_LOG, "sorting section %d", key_file_no); - qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare); - sprintf (out_fname, TEMP_FNAME, key_file_no); +static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno, + int cmd, zebra_rec_keys_t reckeys, + zint staticrank); +static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno, + int cmd, zebra_rec_keys_t skp); +static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid); +static void extract_token_add (RecWord *p); - if (!(outf = fopen (out_fname, "w"))) - { - logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname); - exit (1); - } - logf (LOG_LOG, "writing section %d", key_file_no); - prevcp = cp = key_buf[ptr_top-ptr_i]; - - encode_key_init (&encode_info); - encode_key_write (cp, &encode_info, outf); - while (--ptr_i > 0) - { - cp = key_buf[ptr_top-ptr_i]; - if (strcmp (cp, prevcp)) - { - encode_key_init (&encode_info); - encode_key_write (cp, &encode_info, outf); - prevcp = cp; - } - else - encode_key_write (cp + strlen(cp), &encode_info, outf); - } - if (fclose (outf)) - { - logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname); - exit (1); - } - logf (LOG_LOG, "finished section %d", key_file_no); - ptr_i = 0; - key_buf_used = 0; -} +static void encode_key_init (struct encode_info *i); +static void encode_key_write (char *k, struct encode_info *i, FILE *outf); +static void encode_key_flush (struct encode_info *i, FILE *outf); -int key_close (void) -{ - key_flush (); - xfree (key_buf); - rec_close (&records); - dict_close (matchDict); - - logf (LOG_LOG, "Records inserted %6d", records_inserted); - logf (LOG_LOG, "Records updated %6d", records_updated); - logf (LOG_LOG, "Records deleted %6d", records_deleted); - return key_file_no; -} +#define USE_SHELLSORT 0 -static void wordInit (RecWord *p) +#if USE_SHELLSORT +static void shellsort(void *ar, int r, size_t s, + int (*cmp)(const void *a, const void *b)) { - p->attrSet = 1; - p->attrUse = 1016; - p->which = Word_String; + char *a = ar; + char v[100]; + int h, i, j, k; + static const int incs[16] = { 1391376, 463792, 198768, 86961, 33936, + 13776, 4592, 1968, 861, 336, + 112, 48, 21, 7, 3, 1 }; + for ( k = 0; k < 16; k++) + for (h = incs[k], i = h; i < r; i++) + { + memcpy (v, a+s*i, s); + j = i; + while (j > h && (*cmp)(a + s*(j-h), v) > 0) + { + memcpy (a + s*j, a + s*(j-h), s); + j -= h; + } + memcpy (a+s*j, v, s); + } } +#endif -struct recKeys { - int buf_used; - int buf_max; - char *buf; - char prevAttrSet; - short prevAttrUse; -} reckeys; - -static void addRecordKey (const RecWord *p) +static void logRecord (ZebraHandle zh) { - char *dst; - char attrSet; - short attrUse; - size_t i; - int lead = 0; - - if (reckeys.buf_used+1024 > reckeys.buf_max) + ++zh->records_processed; + if (!(zh->records_processed % 1000)) { - char *b; - - b = xmalloc (reckeys.buf_max += 128000); - if (reckeys.buf_used > 0) - memcpy (b, reckeys.buf, reckeys.buf_used); - xfree (reckeys.buf); - reckeys.buf = b; + yaz_log(YLOG_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); } - dst = reckeys.buf + reckeys.buf_used; - - attrSet = p->attrSet; - if (reckeys.buf_used > 0 && reckeys.prevAttrSet == attrSet) - lead |= 1; - else - reckeys.prevAttrSet = attrSet; - attrUse = p->attrUse; - if (reckeys.buf_used > 0 && reckeys.prevAttrUse == attrUse) - lead |= 2; - else - reckeys.prevAttrUse = attrUse; - - switch (p->which) - { - case Word_String: - *dst++ = lead; - - if (!(lead & 1)) - { - memcpy (dst, &attrSet, sizeof(attrSet)); - dst += sizeof(attrSet); - } - if (!(lead & 2)) - { - memcpy (dst, &attrUse, sizeof(attrUse)); - dst += sizeof(attrUse); - } - for (i = 0; p->u.string[i]; i++) - *dst++ = p->u.string[i]; - *dst++ = '\0'; +} - memcpy (dst, &p->seqno, sizeof(p->seqno)); - dst += sizeof(p->seqno); +static void extract_add_index_string (RecWord *p, + zinfo_index_category_t cat, + const char *str, int length); - break; - default: - return; - } - reckeys.buf_used = dst - reckeys.buf; -} +static void extract_set_store_data_prepare(struct recExtractCtrl *p); -static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys, - const char *databaseName) +static void extract_init(struct recExtractCtrl *p, RecWord *w) { - char attrSet = -1; - short attrUse = -1; - int off = 0; - while (off < reckeys->buf_used) - { - const char *src = reckeys->buf + off; - struct it_key key; - int lead; - - lead = *src++; - - if (!(lead & 1)) - { - memcpy (&attrSet, src, sizeof(attrSet)); - src += sizeof(attrSet); - } - if (!(lead & 2)) - { - memcpy (&attrUse, src, sizeof(attrUse)); - src += sizeof(attrUse); - } - if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*)) - key_flush (); - ++ptr_i; - key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used; - key_buf_used += index_word_prefix ((char*)key_buf + key_buf_used, - attrSet, attrUse, databaseName); - while (*src) - ((char*)key_buf) [key_buf_used++] = index_char_cvt (*src++); - src++; - ((char*)key_buf) [key_buf_used++] = '\0'; - - ((char*) key_buf)[key_buf_used++] = cmd; - - memcpy (&key.seqno, src, sizeof(key.seqno)); - src += sizeof(key.seqno); - key.sysno = sysno; - memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key)); - key_buf_used += sizeof(key); - off = src - reckeys->buf; - } - assert (off == reckeys->buf_used); + w->seqno = 1; + w->index_name = "any"; + w->index_type = 'w'; + w->extractCtrl = p; + w->record_id = 0; + w->section_id = 0; + w->segment = 0; } -static const char **searchRecordKey (struct recKeys *reckeys, - int attrSetS, int attrUseS) +static void searchRecordKey(ZebraHandle zh, + zebra_rec_keys_t reckeys, + const char *index_name, + const char **ws, int ws_length) { - static const char *ws[32]; - int off = 0; - int startSeq = -1; int i; + int ch = -1; + zinfo_index_category_t cat = zinfo_index_category_index; - for (i = 0; i<32; i++) + for (i = 0; ibuf_used) - { - const char *src = reckeys->buf + off; - char attrSet; - short attrUse; - int seqno; - const char *wstart; - - memcpy (&attrSet, src, sizeof(attrSet)); - src += sizeof(attrSet); - - memcpy (&attrUse, src, sizeof(attrUse)); - src += sizeof(attrUse); - - wstart = src; - while (*src++) - ; - - memcpy (&seqno, src, sizeof(seqno)); - src += sizeof(seqno); - -#if 0 - logf (LOG_LOG, "(%d,%d) %d %s", attrSet, attrUse, seqno, wstart); -#endif - if (attrUseS == attrUse && attrSetS == attrSet) - { - int woff; - - - if (startSeq == -1) - startSeq = seqno; - woff = seqno - startSeq; - if (woff >= 0 && woff < 31) - ws[woff] = wstart; - } - - off = src - reckeys->buf; - } - assert (off == reckeys->buf_used); - return ws; -} - -static void addRecordKeyAny (const RecWord *p) -{ - if (p->attrSet != 1 || p->attrUse != 1016) - { - RecWord w; - - memcpy (&w, p, sizeof(w)); - w.attrSet = 1; - w.attrUse = 1016; - addRecordKey (&w); - } - if (p->attrSet != -1) - addRecordKey (p); -} - -#define FILE_READ_BUFSIZE 4096 -struct file_read_info { - int file_noread; - int fd; -#if FILE_READ_BUFSIZE - char *file_buf; - int file_offset; - int file_bufsize; -#endif -}; - -static struct file_read_info *file_read_start (int fd) -{ - struct file_read_info *fi = xmalloc (sizeof(*fi)); - - fi->fd = fd; - fi->file_noread = 0; -#if FILE_READ_BUFSIZE - fi->file_offset = 0; - fi->file_buf = xmalloc (FILE_READ_BUFSIZE); - fi->file_bufsize = read (fd, fi->file_buf, FILE_READ_BUFSIZE); -#endif - return fi; -} -static void file_read_stop (struct file_read_info *fi) -{ - assert (fi); -#if FILE_READ_BUFSIZE - xfree (fi->file_buf); - fi->file_buf = NULL; -#endif - xfree (fi); -} + if (ch < 0) + ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, '0', index_name); + if (ch < 0) + ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'p', index_name); + if (ch < 0) + ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'w', index_name); -static int file_read (void *handle, char *buf, size_t count) -{ - struct file_read_info *p = handle; - int fd = p->fd; -#if FILE_READ_BUFSIZE - int l = p->file_bufsize - p->file_offset; + if (ch < 0) + return ; - if (count > l) + if (zebra_rec_keys_rewind(reckeys)) { - int r; - if (l > 0) - memcpy (buf, p->file_buf + p->file_offset, l); - count = count-l; - if (count > FILE_READ_BUFSIZE) - { - if ((r = read (fd, buf + l, count)) == -1) - { - logf (LOG_FATAL|LOG_ERRNO, "read"); - exit (1); - } - p->file_bufsize = 0; - p->file_offset = 0; - p->file_noread += l+r; - return l+r; - } - p->file_bufsize = r = read (fd, p->file_buf, FILE_READ_BUFSIZE); - if (r == -1) - { - logf (LOG_FATAL|LOG_ERRNO, "read"); - exit (1); - } - else if (r <= count) - { - p->file_offset = r; - memcpy (buf + l, p->file_buf, r); - p->file_noread += l+r; - return l+r; - } - else - { - p->file_offset = count; - memcpy (buf + l, p->file_buf, count - l); - p->file_noread += count; - return count; - } + zint startSeq = -1; + const char *str; + size_t slen; + struct it_key key; + zint seqno; + while (zebra_rec_keys_read(reckeys, &str, &slen, &key)) + { + assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2); + + seqno = key.mem[key.len-1]; + + if (key.mem[0] == ch) + { + zint woff; + + if (startSeq == -1) + startSeq = seqno; + woff = seqno - startSeq; + if (woff >= 0 && woff < ws_length) + ws[woff] = str; + } + } } - memcpy (buf, p->file_buf + p->file_offset, count); - p->file_offset += count; - p->file_noread += count; - return count; -#else - int r; - r = read (fd, buf, count); - if (r > 0) - p->file_noread += r; - return r; -#endif } -static int atois (const char **s) -{ - int val = 0, c; - while ( (c=**s) >= '0' && c <= '9') - { - val = val*10 + c - '0'; - ++(*s); - } - return val; -} +#define FILE_MATCH_BLANK "\t " -static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup, - const char *fname, - const char *spec) +static char *fileMatchStr (ZebraHandle zh, + zebra_rec_keys_t reckeys, + const char *fname, const char *spec) { - static char dstBuf[2048]; + static char dstBuf[2048]; /* static here ??? */ char *dst = dstBuf; const char *s = spec; - static const char **w; - int i; while (1) { - while (*s == ' ' || *s == '\t') - s++; + for (; *s && strchr(FILE_MATCH_BLANK, *s); s++) + ; if (!*s) break; if (*s == '(') { - char matchFlag[32]; - int attrSet, attrUse; + const char *ws[32]; + char attset_str[64], attname_str[64]; + int i; int first = 1; - - s++; - attrSet = atois (&s); - if (*s != ',') - { - logf (LOG_WARN, "Missing , in match criteria %s in group %s", - spec, rGroup->groupName ? rGroup->groupName : "none"); - return NULL; - } - s++; - attrUse = atois (&s); - w = searchRecordKey (reckeys, attrSet, attrUse); - assert (w); - - if (*s == ')') - { - for (i = 0; i<32; i++) - matchFlag[i] = 1; - } + + for (s++; strchr(FILE_MATCH_BLANK, *s); s++) + ; + for (i = 0; *s && *s != ',' && *s != ')' && + !strchr(FILE_MATCH_BLANK, *s); s++) + if (i+1 < sizeof(attset_str)) + attset_str[i++] = *s; + attset_str[i] = '\0'; + + for (; strchr(FILE_MATCH_BLANK, *s); s++) + ; + if (*s != ',') + strcpy(attname_str, attset_str); else + { + for (s++; strchr(FILE_MATCH_BLANK, *s); s++) + ; + for (i = 0; *s && *s != ')' && + !strchr(FILE_MATCH_BLANK, *s); s++) + if (i+1 < sizeof(attname_str)) + attname_str[i++] = *s; + attname_str[i] = '\0'; + } + + searchRecordKey (zh, reckeys, attname_str, ws, 32); + + if (*s != ')') { - logf (LOG_WARN, "Missing ) in match criteria %s in group %s", - spec, rGroup->groupName ? rGroup->groupName : "none"); + yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s", + spec, zh->m_group ? zh->m_group : "none"); return NULL; } s++; for (i = 0; i<32; i++) - if (matchFlag[i] && w[i]) + if (ws[i]) { if (first) { *dst++ = ' '; first = 0; } - strcpy (dst, w[i]); - dst += strlen(w[i]); + strcpy (dst, ws[i]); + dst += strlen(ws[i]); } if (first) { - logf (LOG_WARN, "Record didn't contain match" - " fields in (%d,%d)", attrSet, attrUse); + yaz_log (YLOG_WARN, "Record didn't contain match" + " fields in (%s,%s)", attset_str, attname_str); return NULL; } } @@ -721,24 +252,25 @@ static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup, char special[64]; const char *spec_src = NULL; const char *s1 = ++s; - while (*s1 && *s1 != ' ' && *s1 != '\t') + while (*s1 && !strchr(FILE_MATCH_BLANK, *s1)) s1++; spec_len = s1 - s; - if (spec_len > 63) - spec_len = 63; + if (spec_len > sizeof(special)-1) + spec_len = sizeof(special)-1; memcpy (special, s, spec_len); special[spec_len] = '\0'; s = s1; if (!strcmp (special, "group")) - spec_src = rGroup->groupName; + spec_src = zh->m_group; else if (!strcmp (special, "database")) - spec_src = rGroup->databaseName; - else if (!strcmp (special, "filename")) + spec_src = zh->basenames[0]; + else if (!strcmp (special, "filename")) { spec_src = fname; + } else if (!strcmp (special, "type")) - spec_src = rGroup->recordType; + spec_src = zh->m_record_type; else spec_src = NULL; if (spec_src) @@ -755,7 +287,7 @@ static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup, while (*s && *s != stopMarker) { - if (i < 63) + if (i+1 < sizeof(tmpString)) tmpString[i++] = *s++; } if (*s) @@ -766,371 +298,1294 @@ static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup, } else { - logf (LOG_WARN, "Syntax error in match criteria %s in group %s", - spec, rGroup->groupName ? rGroup->groupName : "none"); + yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s", + spec, zh->m_group ? zh->m_group : "none"); return NULL; } *dst++ = 1; } if (dst == dstBuf) { - logf (LOG_WARN, "No match criteria for record %s in group %s", - fname, rGroup->groupName ? rGroup->groupName : "none"); + yaz_log (YLOG_WARN, "No match criteria for record %s in group %s", + fname, zh->m_group ? zh->m_group : "none"); return NULL; } + *dst = '\0'; return dstBuf; } struct recordLogInfo { const char *fname; - char *op; + int recordOffset; struct recordGroup *rGroup; }; - -static void recordLogPreamble (int level, const char *msg, void *info) + +static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl) { - struct recordLogInfo *p = info; - FILE *outf = log_file (); - - if (level & LOG_LOG) - return ; - if (p->op) - fprintf (outf, "%s of ", p->op); - fprintf (outf, "%s type %s\n", p->rGroup->recordType, p->fname); - log_event_start (NULL, NULL); + int i; + for (i = 0; i<256; i++) + { + if (zebra_maps_is_positioned(zh->reg->zebra_maps, i)) + ctrl->seqno[i] = 1; + else + ctrl->seqno[i] = 0; + } + ctrl->flagShowRecords = !zh->m_flag_rw; } -static int recordExtract (SYSNO *sysno, const char *fname, - struct recordGroup *rGroup, int deleteFlag, - struct file_read_info *fi, RecType recType, - char *subType) +static void all_matches_add(struct recExtractCtrl *ctrl) { - struct recExtractCtrl extractCtrl; - int r; - char *matchStr; - SYSNO sysnotmp; - Record rec; - struct recordLogInfo logInfo; + RecWord word; + extract_init(ctrl, &word); + word.index_name = "_ALLRECORDS"; + word.index_type = 'w'; + word.seqno = 1; + extract_add_index_string (&word, zinfo_index_category_alwaysmatches, + "", 0); +} - logInfo.fname = fname; - logInfo.op = NULL; - logInfo.rGroup = rGroup; - log_event_start (recordLogPreamble, &logInfo); +ZEBRA_RES zebra_extract_file(ZebraHandle zh, SYSNO *sysno, const char *fname, + int deleteFlag) +{ + ZEBRA_RES r = ZEBRA_OK; + int i, fd; + char gprefix[128]; + char ext[128]; + char ext_res[128]; + struct file_read_info *fi = 0; + const char *original_record_type = 0; + RecType recType; + void *recTypeClientData; + struct ZebraRecStream stream, *streamp; + + zebra_init_log_level(); + + if (!zh->m_group || !*zh->m_group) + *gprefix = '\0'; + else + sprintf (gprefix, "%s.", zh->m_group); - if (fi->fd != -1) - { - extractCtrl.fh = fi; - /* extract keys */ - extractCtrl.subType = subType; - extractCtrl.init = wordInit; - extractCtrl.add = addRecordKeyAny; - - reckeys.buf_used = 0; - reckeys.prevAttrUse = -1; - reckeys.prevAttrSet = -1; - extractCtrl.readf = file_read; - r = (*recType->extract)(&extractCtrl); - - if (r) - { - logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r); - return 0; - } - if (reckeys.buf_used == 0) + yaz_log(log_level, "zebra_extract_file %s", fname); + + /* determine file extension */ + *ext = '\0'; + for (i = strlen(fname); --i >= 0; ) + if (fname[i] == '/') + break; + else if (fname[i] == '.') { - logf (LOG_WARN, "Empty file %s", fname); - return 0; + strcpy (ext, fname+i+1); + 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); + zh->m_record_type = res_get (zh->res, ext_res); } - - /* perform match if sysno not known and if match criteria is specified */ - - matchStr = NULL; - if (!sysno) + if (!zh->m_record_type) { - sysnotmp = 0; - sysno = &sysnotmp; - if (rGroup->recordId && *rGroup->recordId) - { - char *rinfo; - - matchStr = fileMatchStr (&reckeys, rGroup, fname, - rGroup->recordId); - if (matchStr) - { - rinfo = dict_lookup (matchDict, matchStr); - if (rinfo) - memcpy (sysno, rinfo+1, sizeof(*sysno)); - } - else - { - logf (LOG_WARN, "Bad match criteria"); - return 0; - } - } + if (zh->records_processed < zh->m_file_verbose_limit) + yaz_log (YLOG_LOG, "? %s", fname); + return 0; } - - /* new record ? */ - if (! *sysno) + /* determine match criteria */ + if (!zh->m_record_id) { - if (deleteFlag) - { - logf (LOG_LOG, "Cannot delete new record"); - return 1; - } - logInfo.op = "add"; - if (rGroup->fileVerboseFlag) - logf (LOG_LOG, "add %s %s", rGroup->recordType, fname); - rec = rec_new (records); - *sysno = rec->sysno; + sprintf (ext_res, "%srecordId.%s", gprefix, ext); + zh->m_record_id = res_get (zh->res, ext_res); + } - if (matchStr) - { - dict_insert (matchDict, matchStr, sizeof(*sysno), sysno); - } - flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName); + if (!(recType = + recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type, + &recTypeClientData))) + { + yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type); + return ZEBRA_FAIL; + } - records_inserted++; + switch(recType->version) + { + case 0: + break; + default: + yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type); + } + if (sysno && deleteFlag) + { + streamp = 0; + fi = 0; } else { - struct recKeys delkeys; + char full_rep[1024]; - rec = rec_get (records, *sysno); - assert (rec); - delkeys.buf_used = rec->size[recInfo_delKeys]; - delkeys.buf = rec->info[recInfo_delKeys]; - flushRecordKeys (*sysno, 0, &delkeys, rec->info[recInfo_databaseName]); - if (deleteFlag) + if (zh->path_reg && !yaz_is_abspath (fname)) { - logInfo.op = "delete"; - if (!delkeys.buf_used) - { - logf (LOG_WARN, "cannot delete; storeKeys false"); - } - else - { - if (rGroup->fileVerboseFlag) - logf (LOG_LOG, "delete %s %s", rGroup->recordType, fname); - records_deleted++; - if (matchStr) - dict_delete (matchDict, matchStr); - rec_del (records, &rec); - } - return 1; + strcpy (full_rep, zh->path_reg); + strcat (full_rep, "/"); + strcat (full_rep, fname); } else + strcpy (full_rep, fname); + + if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1) { - logInfo.op = "update"; - if (!delkeys.buf_used) - { - logf (LOG_WARN, "cannot update; storeKeys false"); - } - else - { - if (rGroup->fileVerboseFlag) - logf (LOG_LOG, "update %s %s", rGroup->recordType, fname); - flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName); - records_updated++; - } + yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep); + zh->m_record_type = original_record_type; + return ZEBRA_FAIL; } + streamp = &stream; + zebra_create_stream_fd(streamp, fd, 0); } - xfree (rec->info[recInfo_fileType]); - rec->info[recInfo_fileType] = - rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]); + while(1) + { + r = zebra_extract_record_stream(zh, streamp, + deleteFlag, + 0, /* tst_mode */ + zh->m_record_type, + sysno, + 0, /*match_criteria */ + fname, + 1, /* force_update */ + 1, /* allow_update */ + recType, recTypeClientData); + if (r != ZEBRA_OK) + { + break; + } + if (sysno) + { + break; + } + } + if (streamp) + stream.destroy(streamp); + zh->m_record_type = original_record_type; + return r; +} - xfree (rec->info[recInfo_filename]); - rec->info[recInfo_filename] = - rec_strdup (fname, &rec->size[recInfo_filename]); +/* + If sysno is provided, then it's used to identify the reocord. + If not, and match_criteria is provided, then sysno is guessed + If not, and a record is provided, then sysno is got from there + + */ - xfree (rec->info[recInfo_delKeys]); - if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1) +ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh, + const char *buf, size_t buf_size, + int delete_flag, + int test_mode, + const char *recordType, + SYSNO *sysno, + const char *match_criteria, + const char *fname, + int force_update, + int allow_update) +{ + struct ZebraRecStream stream; + ZEBRA_RES res; + void *clientData; + RecType recType = 0; + + if (recordType && *recordType) { -#if 1 - rec->size[recInfo_delKeys] = reckeys.buf_used; - rec->info[recInfo_delKeys] = reckeys.buf; - reckeys.buf = NULL; - reckeys.buf_max = 0; -#else - rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used); - rec->size[recInfo_delKeys] = reckeys.buf_used; - memcpy (rec->info[recInfo_delKeys], reckeys.buf, - rec->size[recInfo_delKeys]); -#endif - } + yaz_log(log_level, "Record type explicitly specified: %s", recordType); + recType = recType_byName (zh->reg->recTypes, zh->res, recordType, + &clientData); + } else { - rec->info[recInfo_delKeys] = NULL; - rec->size[recInfo_delKeys] = 0; + if (!(zh->m_record_type)) + { + yaz_log (YLOG_WARN, "No such record type defined"); + return ZEBRA_FAIL; + } + yaz_log(log_level, "Get record type from rgroup: %s", + zh->m_record_type); + recType = recType_byName (zh->reg->recTypes, zh->res, + zh->m_record_type, &clientData); + recordType = zh->m_record_type; + } + + if (!recType) + { + yaz_log (YLOG_WARN, "No such record type: %s", recordType); + return ZEBRA_FAIL; } - xfree (rec->info[recInfo_storeData]); - if (rGroup->flagStoreData == 1) + + + zebra_create_stream_mem(&stream, buf, buf_size); + + res = zebra_extract_record_stream(zh, &stream, + delete_flag, + test_mode, + recordType, + sysno, + match_criteria, + fname, + force_update, + allow_update, + recType, clientData); + stream.destroy(&stream); + return res; +} + + +ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, + struct ZebraRecStream *stream, + int delete_flag, + int test_mode, + const char *recordType, + SYSNO *sysno, + const char *match_criteria, + const char *fname, + int force_update, + int allow_update, + RecType recType, + void *recTypeClientData) + +{ + SYSNO sysno0 = 0; + RecordAttr *recordAttr; + struct recExtractCtrl extractCtrl; + int r; + const char *matchStr = 0; + Record rec; + off_t start_offset = 0; + const char *pr_fname = fname; /* filename to print .. */ + int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0; + + zebra_init_log_level(); + + if (!pr_fname) + pr_fname = ""; /* make it printable if file is omitted */ + + zebra_rec_keys_reset(zh->reg->keys); + zebra_rec_keys_reset(zh->reg->sortKeys); + + if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0])) { - rec->size[recInfo_storeData] = fi->file_noread; - rec->info[recInfo_storeData] = xmalloc (fi->file_noread); -#if FILE_READ_BUFSIZE - if (fi->file_noread < FILE_READ_BUFSIZE) - memcpy (rec->info[recInfo_storeData], fi->file_buf, - fi->file_noread); - else -#endif + if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0], + zh->m_explain_database)) + return ZEBRA_FAIL; + } + + if (stream) + { + off_t null_offset = 0; + extractCtrl.stream = stream; + + start_offset = stream->tellf(stream); + + extractCtrl.first_record = start_offset ? 0 : 1; + + stream->endf(stream, &null_offset);; + + extractCtrl.init = extract_init; + extractCtrl.tokenAdd = extract_token_add; + extractCtrl.schemaAdd = extract_schema_add; + extractCtrl.dh = zh->reg->dh; + extractCtrl.handle = zh; + extractCtrl.match_criteria[0] = '\0'; + extractCtrl.staticrank = 0; + + + init_extractCtrl(zh, &extractCtrl); + + extract_set_store_data_prepare(&extractCtrl); + + r = (*recType->extract)(recTypeClientData, &extractCtrl); + + if (r == RECCTRL_EXTRACT_EOF) + return ZEBRA_FAIL; + else if (r == RECCTRL_EXTRACT_ERROR_GENERIC) { - if (lseek (fi->fd, 0L, SEEK_SET) < 0) - { - logf (LOG_ERRNO|LOG_FATAL, "seek to 0 in %s", fname); - exit (1); + /* error occured during extraction ... */ + yaz_log (YLOG_WARN, "extract error: generic"); + return ZEBRA_FAIL; + } + else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER) + { + /* error occured during extraction ... */ + yaz_log (YLOG_WARN, "extract error: no such filter"); + return ZEBRA_FAIL; + } + + all_matches_add(&extractCtrl); + + if (extractCtrl.match_criteria[0]) + match_criteria = extractCtrl.match_criteria; + } + if (!sysno) { + + sysno = &sysno0; + + if (match_criteria && *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) + { + yaz_log (YLOG_WARN, "Bad match criteria (recordID)"); + return ZEBRA_FAIL; + } } - if (read (fi->fd, rec->info[recInfo_storeData], fi->file_noread) - < fi->file_noread) + } + if (matchStr) + { + int db_ord = zebraExplain_get_database_ord(zh->reg->zei); + char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord, + matchStr); + if (rinfo) + { + assert(*rinfo == sizeof(*sysno)); + memcpy (sysno, rinfo+1, sizeof(*sysno)); + } + } + } + if (zebra_rec_keys_empty(zh->reg->keys)) + { + /* the extraction process returned no information - the record + is probably empty - unless flagShowRecords is in use */ + if (test_mode) + return ZEBRA_OK; + } + + if (! *sysno) + { + /* new record */ + if (delete_flag) + { + yaz_log (YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType, + pr_fname, (zint) start_offset); + yaz_log (YLOG_WARN, "cannot delete record above (seems new)"); + return ZEBRA_FAIL; + } + if (show_progress) + yaz_log (YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname, + (zint) start_offset); + rec = rec_new (zh->reg->records); + + *sysno = rec->sysno; + + recordAttr = rec_init_attr (zh->reg->zei, rec); + recordAttr->staticrank = extractCtrl.staticrank; + + if (matchStr) + { + int db_ord = zebraExplain_get_database_ord(zh->reg->zei); + dict_insert_ord(zh->reg->matchDict, db_ord, matchStr, + sizeof(*sysno), sysno); + } + + + extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys); + extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys, + recordAttr->staticrank); + zh->records_inserted++; + } + else + { + /* record already exists */ + zebra_rec_keys_t delkeys = zebra_rec_keys_open(); + zebra_rec_keys_t sortKeys = zebra_rec_keys_open(); + if (!allow_update) + { + yaz_log (YLOG_LOG, "skipped %s %s " ZINT_FORMAT, + recordType, pr_fname, (zint) start_offset); + logRecord(zh); + return ZEBRA_FAIL; + } + + rec = rec_get (zh->reg->records, *sysno); + assert (rec); + + recordAttr = rec_init_attr (zh->reg->zei, rec); + + zebra_rec_keys_set_buf(delkeys, + rec->info[recInfo_delKeys], + rec->size[recInfo_delKeys], + 0); + zebra_rec_keys_set_buf(sortKeys, + rec->info[recInfo_sortKeys], + rec->size[recInfo_sortKeys], + 0); + + extract_flushSortKeys (zh, *sysno, 0, sortKeys); + extract_flushRecordKeys (zh, *sysno, 0, delkeys, + recordAttr->staticrank); + if (delete_flag) + { + /* record going to be deleted */ + if (zebra_rec_keys_empty(delkeys)) { - logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s", - fi->file_noread, fname); - exit (1); + yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType, + pr_fname, (zint) start_offset); + yaz_log(YLOG_WARN, "cannot delete file above, " + "storeKeys false (3)"); + } + else + { + if (show_progress) + yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType, + pr_fname, (zint) start_offset); + zh->records_deleted++; + if (matchStr) + { + int db_ord = zebraExplain_get_database_ord(zh->reg->zei); + dict_delete_ord(zh->reg->matchDict, db_ord, matchStr); + } + rec_del (zh->reg->records, &rec); } + rec_rm (&rec); + logRecord(zh); + return ZEBRA_OK; + } + else + { + if (show_progress) + yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType, + pr_fname, (zint) ZINT_FORMAT); + recordAttr->staticrank = extractCtrl.staticrank; + extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys); + extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys, + recordAttr->staticrank); + zh->records_updated++; } + zebra_rec_keys_close(delkeys); + zebra_rec_keys_close(sortKeys); + } + /* update file type */ + xfree (rec->info[recInfo_fileType]); + rec->info[recInfo_fileType] = + rec_strdup (recordType, &rec->size[recInfo_fileType]); + + /* update filename */ + xfree (rec->info[recInfo_filename]); + rec->info[recInfo_filename] = + rec_strdup (fname, &rec->size[recInfo_filename]); + + /* update delete keys */ + xfree (rec->info[recInfo_delKeys]); + if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1) + { + zebra_rec_keys_get_buf(zh->reg->keys, + &rec->info[recInfo_delKeys], + &rec->size[recInfo_delKeys]); + } + else + { + rec->info[recInfo_delKeys] = NULL; + rec->size[recInfo_delKeys] = 0; + } + /* update sort keys */ + xfree (rec->info[recInfo_sortKeys]); + + zebra_rec_keys_get_buf(zh->reg->sortKeys, + &rec->info[recInfo_sortKeys], + &rec->size[recInfo_sortKeys]); + + /* save file size of original record */ + zebraExplain_recordBytesIncrement (zh->reg->zei, + - recordAttr->recordSize); + if (stream) + { + off_t end_offset = stream->endf(stream, 0); + + if (!end_offset) + end_offset = stream->tellf(stream); + else + stream->seekf(stream, end_offset); + + recordAttr->recordSize = end_offset - start_offset; + zebraExplain_recordBytesIncrement(zh->reg->zei, + recordAttr->recordSize); + } + + /* set run-number for this record */ + recordAttr->runNumber = + zebraExplain_runNumberIncrement (zh->reg->zei, 0); + + /* update store data */ + xfree (rec->info[recInfo_storeData]); + + /* update store data */ + if (zh->store_data_buf) + { + rec->size[recInfo_storeData] = zh->store_data_size; + rec->info[recInfo_storeData] = zh->store_data_buf; + zh->store_data_buf = 0; + } + else if (zh->m_store_data) + { + off_t cur_offset = stream->tellf(stream); + + rec->size[recInfo_storeData] = recordAttr->recordSize; + rec->info[recInfo_storeData] = (char *) + xmalloc (recordAttr->recordSize); + stream->seekf(stream, start_offset); + stream->readf(stream, rec->info[recInfo_storeData], + recordAttr->recordSize); + stream->seekf(stream, cur_offset); } else { rec->info[recInfo_storeData] = NULL; rec->size[recInfo_storeData] = 0; } + /* update database name */ xfree (rec->info[recInfo_databaseName]); rec->info[recInfo_databaseName] = - rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]); + rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]); - rec_put (records, &rec); - return 1; + /* update offset */ + recordAttr->recordOffset = start_offset; + + /* commit this record */ + rec_put (zh->reg->records, &rec); + logRecord(zh); + return ZEBRA_OK; } -int fileExtract (SYSNO *sysno, const char *fname, - const struct recordGroup *rGroupP, int deleteFlag) +ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n) { - int r, i, fd; - char gprefix[128]; - char ext[128]; - char ext_res[128]; - char subType[128]; - RecType recType; - struct recordGroup rGroupM; - struct recordGroup *rGroup = &rGroupM; - struct file_read_info *fi; + ZebraHandle zh = (ZebraHandle) handle; + struct recExtractCtrl extractCtrl; - memcpy (rGroup, rGroupP, sizeof(*rGroupP)); - - if (!rGroup->groupName || !*rGroup->groupName) - *gprefix = '\0'; - else - sprintf (gprefix, "%s.", rGroup->groupName); + if (zebraExplain_curDatabase (zh->reg->zei, + rec->info[recInfo_databaseName])) + { + abort(); + if (zebraExplain_newDatabase (zh->reg->zei, + rec->info[recInfo_databaseName], 0)) + abort (); + } - logf (LOG_DEBUG, "fileExtract %s", fname); + zebra_rec_keys_reset(zh->reg->keys); + zebra_rec_keys_reset(zh->reg->sortKeys); - /* determine file extension */ - for (i = strlen(fname); --i >= 0; ) - if (fname[i] == '/') - { - strcpy (ext, ""); - break; - } - else if (fname[i] == '.') - { - strcpy (ext, fname+i+1); - break; - } - /* determine file type - depending on extension */ - if (!rGroup->recordType) + extractCtrl.init = extract_init; + extractCtrl.tokenAdd = extract_token_add; + extractCtrl.schemaAdd = extract_schema_add; + extractCtrl.dh = zh->reg->dh; + + init_extractCtrl(zh, &extractCtrl); + + extractCtrl.flagShowRecords = 0; + extractCtrl.match_criteria[0] = '\0'; + extractCtrl.staticrank = 0; + extractCtrl.handle = handle; + extractCtrl.first_record = 1; + + extract_set_store_data_prepare(&extractCtrl); + + if (n) + grs_extract_tree(&extractCtrl, n); + + if (rec->size[recInfo_delKeys]) { - sprintf (ext_res, "%srecordType.%s", gprefix, ext); - if (!(rGroup->recordType = res_get (common_resource, ext_res))) + zebra_rec_keys_t delkeys = zebra_rec_keys_open(); + + zebra_rec_keys_t sortkeys = zebra_rec_keys_open(); + + zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys], + rec->size[recInfo_delKeys], + 0); + extract_flushRecordKeys (zh, rec->sysno, 0, delkeys, 0); + zebra_rec_keys_close(delkeys); + + zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys], + rec->size[recInfo_sortKeys], + 0); + + extract_flushSortKeys (zh, rec->sysno, 0, sortkeys); + zebra_rec_keys_close(sortkeys); + } + extract_flushRecordKeys (zh, rec->sysno, 1, zh->reg->keys, 0); + extract_flushSortKeys (zh, rec->sysno, 1, zh->reg->sortKeys); + + xfree (rec->info[recInfo_delKeys]); + zebra_rec_keys_get_buf(zh->reg->keys, + &rec->info[recInfo_delKeys], + &rec->size[recInfo_delKeys]); + + xfree (rec->info[recInfo_sortKeys]); + zebra_rec_keys_get_buf(zh->reg->sortKeys, + &rec->info[recInfo_sortKeys], + &rec->size[recInfo_sortKeys]); + return ZEBRA_OK; +} + +void extract_rec_keys_adjust(ZebraHandle zh, int is_insert, + zebra_rec_keys_t reckeys) +{ + ZebraExplainInfo zei = zh->reg->zei; + struct ord_stat { + int no; + int ord; + struct ord_stat *next; + }; + + if (zebra_rec_keys_rewind(reckeys)) + { + struct ord_stat *ord_list = 0; + struct ord_stat *p; + size_t slen; + const char *str; + struct it_key key_in; + while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in)) { - sprintf (ext_res, "%srecordType", gprefix); - if (!(rGroup->recordType = res_get (common_resource, ext_res))) + int ord = CAST_ZINT_TO_INT(key_in.mem[0]); + + for (p = ord_list; p ; p = p->next) + if (p->ord == ord) + { + p->no++; + break; + } + if (!p) { - if (rGroup->fileVerboseFlag) - logf (LOG_LOG, "? %s", fname); - return 0; + p = xmalloc(sizeof(*p)); + p->no = 1; + p->ord = ord; + p->next = ord_list; + ord_list = p; } } + + p = ord_list; + while (p) + { + struct ord_stat *p1 = p; + + if (is_insert) + zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1); + else + zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1); + p = p->next; + xfree(p1); + } } - if (!rGroup->recordType) +} + +void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno, + int cmd, + zebra_rec_keys_t reckeys, + zint staticrank) +{ + ZebraExplainInfo zei = zh->reg->zei; + + extract_rec_keys_adjust(zh, cmd, reckeys); + + if (!zh->reg->key_buf) { - if (rGroup->fileVerboseFlag) - logf (LOG_LOG, "? record %s", fname); - return 0; + int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8")); + if (mem <= 0) + { + yaz_log(YLOG_WARN, "Invalid memory setting, using default 8 MB"); + mem= 1024*1024*8; + } + /* FIXME: That "8" should be in a default settings include */ + /* not hard-coded here! -H */ + zh->reg->key_buf = (char**) xmalloc (mem); + zh->reg->ptr_top = mem/sizeof(char*); + zh->reg->ptr_i = 0; + zh->reg->key_buf_used = 0; + zh->reg->key_file_no = 0; } - if (!(recType = recType_byName (rGroup->recordType, subType))) + zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1); + + if (zebra_rec_keys_rewind(reckeys)) { - logf (LOG_WARN, "No such record type: %s", rGroup->recordType); - return 0; + size_t slen; + const char *str; + struct it_key key_in; + while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in)) + { + int ch = 0; + int i, j = 0; + struct it_key key_out; + + assert(key_in.len >= 2); + assert(key_in.len <= IT_KEY_LEVEL_MAX); + + /* check for buffer overflow */ + 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; + + /* key_in.mem[0] ord/ch */ + /* key_in.mem[1] filter specified record ID */ + + /* encode the ordinal value (field/use/attribute) .. */ + ch = CAST_ZINT_TO_INT(key_in.mem[0]); + zh->reg->key_buf_used += + key_SU_encode(ch, (char*)zh->reg->key_buf + + zh->reg->key_buf_used); + + /* copy the 0-terminated stuff from str to output */ + memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used, str, slen); + zh->reg->key_buf_used += slen; + ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = '\0'; + + /* the delete/insert indicator */ + ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = cmd; + + if (zh->m_staticrank) /* rank config enabled ? */ + { + if (staticrank < 0) + { + yaz_log(YLOG_WARN, "staticrank = %ld. Setting to 0", + (long) staticrank); + staticrank = 0; + } + key_out.mem[j++] = staticrank; + } + + if (key_in.mem[1]) /* filter specified record ID */ + key_out.mem[j++] = key_in.mem[1]; + else + key_out.mem[j++] = sysno; + for (i = 2; i < key_in.len; i++) + key_out.mem[j++] = key_in.mem[i]; + key_out.len = j; + + memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used, + &key_out, sizeof(key_out)); + (zh->reg->key_buf_used) += sizeof(key_out); + } } +} - /* determine match criteria */ - if (!rGroup->recordId) +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) { - sprintf (ext_res, "%srecordId.%s", gprefix, ext); - rGroup->recordId = res_get (common_resource, ext_res); + yaz_log(log_level, " nothing to flush section=%d buf=%p i=%d", + zh->reg->key_file_no, zh->reg->key_buf, ptr_i); + return; } - /* determine database name */ - if (!rGroup->databaseName) + (zh->reg->key_file_no)++; + yaz_log (YLOG_LOG, "sorting section %d", (zh->reg->key_file_no)); + yaz_log(log_level, " 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') { + yaz_log (YLOG_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"))) + { + yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname); + exit (1); + } + yaz_log (YLOG_LOG, "writing section %d", zh->reg->key_file_no); + prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i]; + + encode_key_init (&encode_info); + encode_key_write (cp, &encode_info, outf); + + while (--ptr_i > 0) { - sprintf (ext_res, "%sdatabase.%s", gprefix, ext); - if (!(rGroup->databaseName = res_get (common_resource, ext_res))) + cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i]; + if (strcmp (cp, prevcp)) { - sprintf (ext_res, "%sdatabase", gprefix); - rGroup->databaseName = res_get (common_resource, ext_res); + encode_key_flush ( &encode_info, outf); + encode_key_init (&encode_info); + encode_key_write (cp, &encode_info, outf); + prevcp = cp; } + else + encode_key_write (cp + strlen(cp), &encode_info, outf); } - if (!rGroup->databaseName) - rGroup->databaseName = "Default"; + encode_key_flush ( &encode_info, outf); +#else + qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare); + extract_get_fname_tmp (out_fname, key_file_no); - if (rGroup->flagStoreData == -1) + if (!(outf = fopen (out_fname, "wb"))) { - const char *sval; - sprintf (ext_res, "%sstoreData.%s", gprefix, ext); - if (!(sval = res_get (common_resource, ext_res))) + yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname); + exit (1); + } + yaz_log (YLOG_LOG, "writing section %d", key_file_no); + i = ptr_i; + prevcp = key_buf[ptr_top-i]; + while (1) + if (!--i || strcmp (prevcp, key_buf[ptr_top-i])) { - sprintf (ext_res, "%sstoreData", gprefix); - sval = res_get (common_resource, ext_res); + key_y_len = strlen(prevcp)+1; +#if 0 + yaz_log (YLOG_LOG, "key_y_len: %2d %02x %02x %s", + key_y_len, prevcp[0], prevcp[1], 2+prevcp); +#endif + qsort (key_buf + ptr_top-ptr_i, ptr_i - i, + sizeof(char*), key_y_compare); + cp = key_buf[ptr_top-ptr_i]; + --key_y_len; + encode_key_init (&encode_info); + encode_key_write (cp, &encode_info, outf); + while (--ptr_i > i) + { + cp = key_buf[ptr_top-ptr_i]; + encode_key_write (cp+key_y_len, &encode_info, outf); + } + encode_key_flush ( &encode_info, outf); + if (!i) + break; + prevcp = key_buf[ptr_top-ptr_i]; } - if (sval) - rGroup->flagStoreData = atoi (sval); +#endif + if (fclose (outf)) + { + yaz_log (YLOG_FATAL|YLOG_ERRNO, "fclose %s", out_fname); + exit (1); } - if (rGroup->flagStoreData == -1) - rGroup->flagStoreData = 0; + yaz_log (YLOG_LOG, "finished section %d", zh->reg->key_file_no); + zh->reg->ptr_i = 0; + zh->reg->key_buf_used = 0; +} - if (rGroup->flagStoreKeys == -1) +ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh, + zebra_rec_keys_t reckeys, + zebra_snippets *snippets) +{ + NMEM nmem = nmem_create(); + if (zebra_rec_keys_rewind(reckeys)) { - const char *sval; + const char *str; + size_t slen; + struct it_key key; + while (zebra_rec_keys_read(reckeys, &str, &slen, &key)) + { + char dst_buf[IT_MAX_WORD]; + char *dst_term = dst_buf; + int ord; + zint seqno; + int index_type; + + assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2); + seqno = key.mem[key.len-1]; + ord = CAST_ZINT_TO_INT(key.mem[0]); + + zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, + 0/* db */, 0 /* string_index */); + assert(index_type); + zebra_term_untrans_iconv(zh, nmem, index_type, + &dst_term, str); + zebra_snippets_append(snippets, seqno, ord, dst_term); + nmem_reset(nmem); + } + } + nmem_destroy(nmem); + return ZEBRA_OK; +} + +void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys) +{ + yaz_log(YLOG_LOG, "print_rec_keys"); + if (zebra_rec_keys_rewind(reckeys)) + { + const char *str; + size_t slen; + struct it_key key; + while (zebra_rec_keys_read(reckeys, &str, &slen, &key)) + { + char dst_buf[IT_MAX_WORD]; + zint seqno; + int index_type; + int ord = CAST_ZINT_TO_INT(key.mem[0]); + const char *db = 0; + assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2); + + zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0); + + seqno = key.mem[key.len-1]; + + zebra_term_untrans(zh, index_type, dst_buf, str); + + yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT + " term=%s", ord, seqno, dst_buf); + } + } +} + +static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat, + const char *str, int length) +{ + struct it_key key; + ZebraHandle zh = p->extractCtrl->handle; + ZebraExplainInfo zei = zh->reg->zei; + int ch, i; + + ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name); + if (ch < 0) + ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name); - sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext); - if (!(sval = res_get (common_resource, ext_res))) + i = 0; + key.mem[i++] = ch; + key.mem[i++] = p->record_id; + key.mem[i++] = p->section_id; + + if (zh->m_segment_indexing) + key.mem[i++] = p->segment; + key.mem[i++] = p->seqno; + key.len = i; + + zebra_rec_keys_write(zh->reg->keys, str, length, &key); +} + +static void extract_add_sort_string(RecWord *p, const char *str, int length) +{ + struct it_key key; + ZebraHandle zh = p->extractCtrl->handle; + ZebraExplainInfo zei = zh->reg->zei; + int ch; + zinfo_index_category_t cat = zinfo_index_category_sort; + + ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name); + if (ch < 0) + ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name); + key.len = 2; + key.mem[0] = ch; + key.mem[1] = p->record_id; + + zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key); +} + +static void extract_add_string(RecWord *p, const char *string, int length) +{ + ZebraHandle zh = p->extractCtrl->handle; + assert (length > 0); + + if (!p->index_name) + return; + + if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type)) + extract_add_sort_string(p, string, length); + else + { + extract_add_index_string(p, zinfo_index_category_index, + string, length); + if (zebra_maps_is_alwaysmatches(zh->reg->zebra_maps, p->index_type)) { - sprintf (ext_res, "%sstoreKeys", gprefix); - sval = res_get (common_resource, ext_res); + RecWord word; + memcpy(&word, p, sizeof(word)); + + word.seqno = 1; + extract_add_index_string( + &word, zinfo_index_category_alwaysmatches, "", 0); } - if (sval) - rGroup->flagStoreKeys = atoi (sval); } - if (rGroup->flagStoreKeys == -1) - rGroup->flagStoreKeys = 0; +} - if (sysno && deleteFlag) - fd = -1; +static void extract_add_incomplete_field(RecWord *p) +{ + ZebraHandle zh = p->extractCtrl->handle; + const char *b = p->term_buf; + int remain = p->term_len; + const char **map = 0; + + if (remain > 0) + map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0); + + if (map) + { + if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type)) + { + /* first in field marker */ + extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN); + p->seqno++; + } + } + while (map) + { + char buf[IT_MAX_WORD+1]; + int i, remain; + + /* Skip spaces */ + while (map && *map && **map == *CHR_SPACE) + { + remain = p->term_len - (b - p->term_buf); + if (remain > 0) + map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, + remain, 0); + else + map = 0; + } + if (!map) + break; + i = 0; + while (map && *map && **map != *CHR_SPACE) + { + const char *cp = *map; + + while (i < IT_MAX_WORD && *cp) + buf[i++] = *(cp++); + remain = p->term_len - (b - p->term_buf); + if (remain > 0) + map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0); + else + map = 0; + } + if (!i) + return; + extract_add_string (p, buf, i); + p->seqno++; + } +} + +static void extract_add_complete_field (RecWord *p) +{ + ZebraHandle zh = p->extractCtrl->handle; + const char *b = p->term_buf; + char buf[IT_MAX_WORD+1]; + const char **map = 0; + int i = 0, remain = p->term_len; + + if (remain > 0) + map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1); + + while (remain > 0 && i < IT_MAX_WORD) + { + while (map && *map && **map == *CHR_SPACE) + { + remain = p->term_len - (b - p->term_buf); + + if (remain > 0) + { + int first = i ? 0 : 1; /* first position */ + map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, first); + } + else + map = 0; + } + if (!map) + break; + + if (i && i < IT_MAX_WORD) + buf[i++] = *CHR_SPACE; + while (map && *map && **map != *CHR_SPACE) + { + const char *cp = *map; + + if (**map == *CHR_CUT) + { + i = 0; + } + else + { + if (i >= IT_MAX_WORD) + break; + while (i < IT_MAX_WORD && *cp) + buf[i++] = *(cp++); + } + remain = p->term_len - (b - p->term_buf); + if (remain > 0) + { + map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, + remain, 0); + } + else + map = 0; + } + } + if (!i) + return; + extract_add_string (p, buf, i); +} + +static void extract_token_add(RecWord *p) +{ + ZebraHandle zh = p->extractCtrl->handle; + WRBUF wrbuf; + + if (log_level) + { + yaz_log(log_level, "extract_token_add " + "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s", + p->index_type, p->index_name, + p->seqno, p->term_len, p->term_buf); + } + if ((wrbuf = zebra_replace(zh->reg->zebra_maps, p->index_type, 0, + p->term_buf, p->term_len))) + { + p->term_buf = wrbuf_buf(wrbuf); + p->term_len = wrbuf_len(wrbuf); + } + if (zebra_maps_is_complete (zh->reg->zebra_maps, p->index_type)) + extract_add_complete_field (p); else + extract_add_incomplete_field(p); +} + +static void extract_set_store_data_cb(struct recExtractCtrl *p, + void *buf, size_t sz) +{ + ZebraHandle zh = (ZebraHandle) p->handle; + + xfree(zh->store_data_buf); + zh->store_data_buf = 0; + zh->store_data_size = 0; + if (buf && sz) + { + zh->store_data_buf = xmalloc(sz); + zh->store_data_size = sz; + memcpy(zh->store_data_buf, buf, sz); + } +} + +static void extract_set_store_data_prepare(struct recExtractCtrl *p) +{ + ZebraHandle zh = (ZebraHandle) p->handle; + xfree(zh->store_data_buf); + zh->store_data_buf = 0; + zh->store_data_size = 0; + p->setStoreData = extract_set_store_data_cb; +} + +static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid) +{ + ZebraHandle zh = (ZebraHandle) p->handle; + zebraExplain_addSchema (zh->reg->zei, oid); +} + +void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno, + int cmd, zebra_rec_keys_t reckeys) +{ + if (zebra_rec_keys_rewind(reckeys)) { - if ((fd = open (fname, O_RDONLY)) == -1) + SortIdx sortIdx = zh->reg->sortIdx; + size_t slen; + const char *str; + struct it_key key_in; + + sortIdx_sysno (sortIdx, sysno); + + while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in)) { - logf (LOG_WARN|LOG_ERRNO, "open %s", fname); - return 0; + int ord = CAST_ZINT_TO_INT(key_in.mem[0]); + + sortIdx_type(sortIdx, ord); + if (cmd == 1) + sortIdx_add(sortIdx, str, slen); + else + sortIdx_add(sortIdx, "", 1); } } - fi = file_read_start (fd); - r = recordExtract (sysno, fname, rGroup, deleteFlag, fi, recType, subType); - log_event_start (NULL, NULL); - file_read_stop (fi); - if (fd != -1) - close (fd); - return r; } +static void encode_key_init(struct encode_info *i) +{ + i->encode_handle = iscz1_start(); + i->decode_handle = iscz1_start(); +} + +static void encode_key_write (char *k, struct encode_info *i, FILE *outf) +{ + struct it_key key; + char *bp = i->buf, *bp0; + const char *src = (char *) &key; + + /* copy term to output buf */ + while ((*bp++ = *k++)) + ; + /* and copy & align key so we can mangle */ + memcpy (&key, k+1, sizeof(struct it_key)); /* *k is insert/delete */ + +#if 0 + /* debugging */ + key_logdump_txt(YLOG_LOG, &key, *k ? "i" : "d"); +#endif + assert(key.mem[0] >= 0); + + bp0 = bp++; + iscz1_encode(i->encode_handle, &bp, &src); + + *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */ + if (fwrite (i->buf, bp - i->buf, 1, outf) != 1) + { + yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite"); + exit (1); + } + +#if 0 + /* debugging */ + if (1) + { + struct it_key key2; + const char *src = bp0+1; + char *dst = (char*) &key2; + iscz1_decode(i->decode_handle, &dst, &src); + + key_logdump_txt(YLOG_LOG, &key2, *k ? "i" : "d"); + + assert(key2.mem[1]); + } +#endif +} + +static void encode_key_flush (struct encode_info *i, FILE *outf) +{ + iscz1_stop(i->encode_handle); + iscz1_stop(i->decode_handle); +} + +/* + * Local variables: + * c-basic-offset: 4 + * indent-tabs-mode: nil + * End: + * vim: shiftwidth=4 tabstop=8 expandtab + */ +