X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=a085d728215cdc2a36589fc4b5146c72277c063d;hb=4097bac1dce126e79ce1550fad8b2034a311ff80;hp=bd1b4f0f72d99f7796380fb54b4423ad08747ae3;hpb=ea01436a957572aaaa878b59469a4cedce7e5e21;p=idzebra-moved-to-github.git diff --git a/index/extract.c b/index/extract.c index bd1b4f0..be000c2 100644 --- a/index/extract.c +++ b/index/extract.c @@ -1,652 +1,1886 @@ -/* - * Copyright (C) 1994-1995, Index Data I/S - * All rights reserved. - * Sebastian Hammer, Adam Dickmeiss - * - * $Log: extract.c,v $ - * 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. - * - */ +/* $Id: extract.c,v 1.275 2007-12-20 11:15:42 adam Exp $ + Copyright (C) 1995-2007 + Index Data ApS + +This file is part of the Zebra server. + +Zebra is free software; you can redistribute it and/or modify it under +the terms of the GNU General Public License as published by the Free +Software Foundation; either version 2, or (at your option) any later +version. + +Zebra is distributed in the hope that it will be useful, but WITHOUT ANY +WARRANTY; without even the implied warranty of MERCHANTABILITY or +FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +for more details. + +You should have received a copy of the GNU General Public License +along with this program; if not, write to the Free Software +Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA + +*/ + +/** \file + \brief indexes records and extract tokens for indexing and sorting +*/ + #include #include -#include -#include #include - -#include -#include -#include "index.h" - -#if RECORD_BASE -#include "recindex.h" +#ifdef WIN32 +#include +#endif +#if HAVE_UNISTD_H +#include #endif +#include -static Dict file_idx; +#include "index.h" +#include "orddict.h" +#include +#include +#include -#if RECORD_BASE -static Records records = NULL; -#else -static int sys_idx_fd = -1; -static SYSNO sysno_next; -#endif +static int log_level_extract = 0; +static int log_level_details = 0; +static int log_level_initialized = 0; -static int key_cmd; -static int key_sysno; -static const char *key_databaseName; -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; +/* 1 if we use eliminitate identical delete/insert keys */ +/* eventually this the 0-case code will be removed */ +#define FLUSH2 1 -static int key_del_max; -static int key_del_used; -static char *key_del_buf; +void extract_flush_record_keys2(ZebraHandle zh, zint sysno, + zebra_rec_keys_t ins_keys, + zint ins_rank, + zebra_rec_keys_t del_keys, + zint del_rank); -void key_open (int mem) +static void zebra_init_log_level(void) { -#if !RECORD_BASE - void *file_key; -#endif - if (mem < 50000) - mem = 50000; - key_buf = xmalloc (mem); - ptr_top = mem/sizeof(char*); - ptr_i = 0; + if (!log_level_initialized) + { + log_level_initialized = 1; - key_buf_used = 0; - key_file_no = 0; + log_level_extract = yaz_log_module_level("extract"); + log_level_details = yaz_log_module_level("indexdetails"); + } +} - key_del_buf = NULL; - key_del_max = 0; +static void extract_flush_sort_keys(ZebraHandle zh, zint 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 (!(file_idx = dict_open (FNAME_FILE_DICT, 40, 1))) +static void check_log_limit(ZebraHandle zh) +{ + if (zh->records_processed + zh->records_skipped == zh->m_file_verbose_limit) { - logf (LOG_FATAL, "dict_open fail of %s", "fileidx"); - exit (1); + yaz_log(YLOG_LOG, "More than %d file log entries. Omitting rest", + zh->m_file_verbose_limit); } -#if RECORD_BASE - assert (!records); - records = rec_open (1); -#else - file_key = dict_lookup (file_idx, "."); - if (file_key) - memcpy (&sysno_next, (char*)file_key+1, sizeof(sysno_next)); - else - sysno_next = 1; - if ((sys_idx_fd = open (FNAME_SYS_IDX, O_RDWR|O_CREAT, 0666)) == -1) +} + +static void logRecord(ZebraHandle zh) +{ + check_log_limit(zh); + ++zh->records_processed; + if (!(zh->records_processed % 1000)) { - logf (LOG_FATAL|LOG_ERRNO, "open %s", FNAME_SYS_IDX); - exit (1); + 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); } -#endif } -struct encode_info { - int sysno; - int seqno; - char buf[512]; -}; +static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl) +{ + ctrl->flagShowRecords = !zh->m_flag_rw; +} -void encode_key_init (struct encode_info *i) + +static void extract_add_index_string(RecWord *p, + zinfo_index_category_t cat, + const char *str, int length); + +static void extract_set_store_data_prepare(struct recExtractCtrl *p); + +static void extract_init(struct recExtractCtrl *p, RecWord *w) { - i->sysno = 0; - i->seqno = 0; + 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; } -char *encode_key_int (int d, char *bp) +struct snip_rec_info { + ZebraHandle zh; + zebra_snippets *snippets; +}; + + +static void snippet_add_complete_field(RecWord *p, int ord, + zebra_map_t zm) { - if (d <= 63) - *bp++ = d; - else if (d <= 16383) - { - *bp++ = 64 + (d>>8); - *bp++ = d & 255; - } - else if (d <= 4194303) + struct snip_rec_info *h = 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; + const char *start = b; + const char *last = 0; + + if (remain > 0) + map = zebra_maps_input(zm, &b, remain, 1); + + while (remain > 0 && i < IT_MAX_WORD) { - *bp++ = 128 + (d>>16); - *bp++ = (d>>8) & 255; - *bp++ = d & 255; + while (map && *map && **map == *CHR_SPACE) + { + remain = p->term_len - (b - p->term_buf); + + if (i == 0) + start = b; /* set to first non-ws area */ + if (remain > 0) + { + int first = i ? 0 : 1; /* first position */ + + map = zebra_maps_input(zm, &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++); + } + last = b; + remain = p->term_len - (b - p->term_buf); + if (remain > 0) + { + map = zebra_maps_input(zm, &b, remain, 0); + } + else + map = 0; + } } - else + if (!i) + return; + if (last && start != last) + zebra_snippets_appendn(h->snippets, p->seqno, 0, ord, + start, last - start); +} + +static void snippet_add_incomplete_field(RecWord *p, int ord, zebra_map_t zm) +{ + struct snip_rec_info *h = p->extractCtrl->handle; + const char *b = p->term_buf; + int remain = p->term_len; + int first = 1; + const char **map = 0; + const char *start = b; + const char *last = b; + + if (remain > 0) + map = zebra_maps_input(zm, &b, remain, 0); + + while (map) { - *bp++ = 192 + (d>>24); - *bp++ = (d>>16) & 255; - *bp++ = (d>>8) & 255; - *bp++ = d & 255; + 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); + last = b; + if (remain > 0) + map = zebra_maps_input(zm, &b, remain, 0); + else + map = 0; + } + if (!map) + break; + if (start != last) + { + zebra_snippets_appendn(h->snippets, p->seqno, 1, ord, + start, last - start); + + } + start = last; + + 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); + last = b; + if (remain > 0) + map = zebra_maps_input(zm, &b, remain, 0); + else + map = 0; + } + if (!i) + return; + + if (first) + { + first = 0; + if (zebra_maps_is_first_in_field(zm)) + { + /* first in field marker */ + p->seqno++; + } + } + if (start != last) + zebra_snippets_appendn(h->snippets, p->seqno, 0, ord, + start, last - start); + start = last; + p->seqno++; } - return bp; + } -void encode_key_write (char *k, struct encode_info *i, FILE *outf) +static void snippet_add_icu(RecWord *p, int ord, zebra_map_t zm) { - struct it_key key; - char *bp = i->buf; + struct snip_rec_info *h = p->extractCtrl->handle; + + const char *res_buf = 0; + size_t res_len = 0; - 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) + const char *display_buf = 0; + size_t display_len = 0; + + zebra_map_tokenize_start(zm, p->term_buf, p->term_len); + while (zebra_map_tokenize_next(zm, &res_buf, &res_len, + &display_buf, &display_len)) { - i->sysno = key.sysno; - i->seqno = 0; + zebra_snippets_appendn(h->snippets, p->seqno, 0, ord, + display_buf, display_len); + p->seqno++; } - bp = encode_key_int (key.seqno - i->seqno, bp); - i->seqno = key.seqno; - if (fwrite (i->buf, bp - i->buf, 1, outf) != 1) +} + +static void snippet_token_add(RecWord *p) +{ + struct snip_rec_info *h = p->extractCtrl->handle; + ZebraHandle zh = h->zh; + zebra_map_t zm = zebra_map_get(zh->reg->zebra_maps, p->index_type); + + if (zm && zebra_maps_is_index(zm)) { - logf (LOG_FATAL|LOG_ERRNO, "fwrite"); - exit (1); + ZebraExplainInfo zei = zh->reg->zei; + int ch = zebraExplain_lookup_attr_str( + zei, zinfo_index_category_index, p->index_type, p->index_name); + + if (zebra_maps_is_icu(zm)) + snippet_add_icu(p, ch, zm); + else + { + if (zebra_maps_is_complete(zm)) + snippet_add_complete_field(p, ch, zm); + else + snippet_add_incomplete_field(p, ch, zm); + } } } -void key_flush (void) +static void snippet_schema_add( + struct recExtractCtrl *p, Odr_oid *oid) +{ + +} + +void extract_snippet(ZebraHandle zh, zebra_snippets *sn, + struct ZebraRecStream *stream, + RecType rt, void *recTypeClientData) { - FILE *outf; - char out_fname[200]; - char *prevcp, *cp; - struct encode_info encode_info; + struct recExtractCtrl extractCtrl; + struct snip_rec_info info; + int r; + + extractCtrl.stream = stream; + extractCtrl.first_record = 1; + extractCtrl.init = extract_init; + extractCtrl.tokenAdd = snippet_token_add; + extractCtrl.schemaAdd = snippet_schema_add; + assert(zh->reg); + assert(zh->reg->dh); + + extractCtrl.dh = zh->reg->dh; - if (ptr_i <= 0) - return; + info.zh = zh; + info.snippets = sn; + extractCtrl.handle = &info; + extractCtrl.match_criteria[0] = '\0'; + extractCtrl.staticrank = 0; + extractCtrl.action = action_insert; + + init_extractCtrl(zh, &extractCtrl); + + extractCtrl.setStoreData = 0; + + r = (*rt->extract)(recTypeClientData, &extractCtrl); + +} - 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 searchRecordKey(ZebraHandle zh, + zebra_rec_keys_t reckeys, + const char *index_name, + const char **ws, int ws_length) +{ + int i; + int ch = -1; + zinfo_index_category_t cat = zinfo_index_category_index; + + for (i = 0; ireg->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); + + if (ch < 0) + return ; - if (!(outf = fopen (out_fname, "w"))) + if (zebra_rec_keys_rewind(reckeys)) { - logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname); - exit (1); + 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; + } + } } - 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) +} + +#define FILE_MATCH_BLANK "\t " + +static char *get_match_from_spec(ZebraHandle zh, + zebra_rec_keys_t reckeys, + const char *fname, const char *spec) +{ + static char dstBuf[2048]; /* static here ??? */ + char *dst = dstBuf; + const char *s = spec; + + while (1) { - cp = key_buf[ptr_top-ptr_i]; - if (strcmp (cp, prevcp)) + for (; *s && strchr(FILE_MATCH_BLANK, *s); s++) + ; + if (!*s) + break; + if (*s == '(') + { + const char *ws[32]; + char attset_str[64], attname_str[64]; + int i; + int first = 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 != ')') + { + 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 (ws[i]) + { + if (first) + { + *dst++ = ' '; + first = 0; + } + strcpy(dst, ws[i]); + dst += strlen(ws[i]); + } + if (first) + { + yaz_log(YLOG_WARN, "Record didn't contain match" + " fields in (%s,%s)", attset_str, attname_str); + return NULL; + } + } + else if (*s == '$') { - encode_key_init (&encode_info); - encode_key_write (cp, &encode_info, outf); - prevcp = cp; + int spec_len; + char special[64]; + const char *spec_src = NULL; + const char *s1 = ++s; + while (*s1 && !strchr(FILE_MATCH_BLANK, *s1)) + s1++; + + spec_len = s1 - s; + 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 = zh->m_group; + else if (!strcmp(special, "database")) + spec_src = zh->basenames[0]; + else if (!strcmp(special, "filename")) { + spec_src = fname; + } + else if (!strcmp(special, "type")) + spec_src = zh->m_record_type; + else + spec_src = NULL; + if (spec_src) + { + strcpy(dst, spec_src); + dst += strlen(spec_src); + } + } + else if (*s == '\"' || *s == '\'') + { + int stopMarker = *s++; + char tmpString[64]; + int i = 0; + + while (*s && *s != stopMarker) + { + if (i+1 < sizeof(tmpString)) + tmpString[i++] = *s++; + } + if (*s) + s++; + tmpString[i] = '\0'; + strcpy(dst, tmpString); + dst += strlen(tmpString); } else - encode_key_write (cp + strlen(cp), &encode_info, outf); + { + 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 (fclose (outf)) + if (dst == dstBuf) { - logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname); - exit (1); + yaz_log(YLOG_WARN, "No match criteria for record %s in group %s", + fname, zh->m_group ? zh->m_group : "none"); + return NULL; } - logf (LOG_LOG, "finished section %d", key_file_no); - ptr_i = 0; - key_buf_used = 0; + *dst = '\0'; + return dstBuf; } -int key_close (void) +struct recordLogInfo { + const char *fname; + int recordOffset; + struct recordGroup *rGroup; +}; + +/** \brief add the always-matches index entry and map to real record ID + \param ctrl record control + \param record_id custom record ID + \param sysno system record ID + + This function serves two purposes.. It adds the always matches + entry and makes a pointer from the custom record ID (if defined) + back to the system record ID (sysno) + See zebra_recid_to_sysno . + */ +static void all_matches_add(struct recExtractCtrl *ctrl, zint record_id, + zint sysno) { - key_flush (); - xfree (key_buf); -#if RECORD_BASE - rec_close (&records); -#else - close (sys_idx_fd); - dict_insert (file_idx, ".", sizeof(sysno_next), &sysno_next); -#endif - dict_close (file_idx); + RecWord word; + extract_init(ctrl, &word); + word.record_id = record_id; + /* we use the seqno as placeholder for a way to get back to + record database from _ALLRECORDS.. This is used if a custom + RECORD was defined */ + word.seqno = sysno; + word.index_name = "_ALLRECORDS"; + word.index_type = "w"; - xfree (key_del_buf); - key_del_buf = NULL; - key_del_max = 0; - return key_file_no; + extract_add_index_string(&word, zinfo_index_category_alwaysmatches, + "", 0); } -static void wordInit (RecWord *p) -{ - p->attrSet = 1; - p->attrUse = 1016; - p->which = Word_String; -} +ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh, + struct ZebraRecStream *stream, + enum zebra_recctrl_action_t action, + int test_mode, + const char *recordType, + zint *sysno, + const char *match_criteria, + const char *fname, + RecType recType, + void *recTypeClientData); + -static void wordAdd (const RecWord *p) +ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname, + enum zebra_recctrl_action_t action) { - struct it_key key; - size_t i; + 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; - 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, - p->attrSet, p->attrUse, - key_databaseName); - switch (p->which) - { - case Word_String: - for (i = 0; p->u.string[i]; i++) - ((char*)key_buf) [key_buf_used++] = - index_char_cvt (p->u.string[i]); - ((char*)key_buf) [key_buf_used++] = '\0'; - break; - default: - return ; + zebra_init_log_level(); + + if (!zh->m_group || !*zh->m_group) + *gprefix = '\0'; + else + sprintf(gprefix, "%s.", zh->m_group); + + yaz_log(log_level_extract, "zebra_extract_file %s", fname); + + /* determine file extension */ + *ext = '\0'; + for (i = strlen(fname); --i >= 0; ) + if (fname[i] == '/') + break; + else if (fname[i] == '.') + { + 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); + } + if (!zh->m_record_type) + { + check_log_limit(zh); + if (zh->records_processed + zh->records_skipped + < zh->m_file_verbose_limit) + yaz_log(YLOG_LOG, "? %s", fname); + zh->records_skipped++; + return 0; + } + /* determine match criteria */ + if (!zh->m_record_id) + { + sprintf(ext_res, "%srecordId.%s", gprefix, ext); + zh->m_record_id = res_get(zh->res, ext_res); + } + + 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; } - ((char*) key_buf)[key_buf_used++] = ((key_cmd == 'a') ? 1 : 0); - key.sysno = key_sysno; - key.seqno = p->seqno; - memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key)); - key_buf_used += sizeof(key); - if (key_cmd == 'a' && key_del_used >= 0) + switch(recType->version) + { + case 0: + break; + default: + yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type); + } + if (sysno && (action == action_delete || action == action_a_delete)) { - char attrSet; - short attrUse; - if (key_del_used + 1024 > key_del_max) + streamp = 0; + fi = 0; + } + else + { + char full_rep[1024]; + + if (zh->path_reg && !yaz_is_abspath(fname)) { - char *kbn; - - if (!(kbn = malloc (key_del_max += 64000))) - { - logf (LOG_FATAL, "malloc"); - exit (1); - } - if (key_del_buf) - memcpy (kbn, key_del_buf, key_del_used); - free (key_del_buf); - key_del_buf = kbn; + strcpy(full_rep, zh->path_reg); + strcat(full_rep, "/"); + strcat(full_rep, fname); } - switch (p->which) + else + strcpy(full_rep, fname); + + if ((fd = open(full_rep, O_BINARY|O_RDONLY)) == -1) { - case Word_String: - for (i = 0; p->u.string[i]; i++) - ((char*)key_del_buf) [key_del_used++] = p->u.string[i]; - ((char*)key_del_buf) [key_del_used++] = '\0'; - break; - default: - return ; + yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", full_rep); + zh->m_record_type = original_record_type; + return ZEBRA_FAIL; } - attrSet = p->attrSet; - memcpy (key_del_buf + key_del_used, &attrSet, sizeof(attrSet)); - key_del_used += sizeof(attrSet); + streamp = &stream; + zebra_create_stream_fd(streamp, fd, 0); + } + r = zebra_extract_records_stream(zh, streamp, + action, + 0, /* tst_mode */ + zh->m_record_type, + sysno, + 0, /*match_criteria */ + fname, + recType, recTypeClientData); + if (streamp) + stream.destroy(streamp); + zh->m_record_type = original_record_type; + return r; +} - attrUse = p->attrUse; - memcpy (key_del_buf + key_del_used, &attrUse, sizeof(attrUse)); - key_del_used += sizeof(attrUse); +/* + 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 + + */ - memcpy (key_del_buf + key_del_used, &p->seqno, sizeof(p->seqno)); - key_del_used += sizeof(p->seqno); +ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh, + const char *buf, size_t buf_size, + enum zebra_recctrl_action_t action, + int test_mode, + const char *recordType, + zint *sysno, + const char *match_criteria, + const char *fname) +{ + struct ZebraRecStream stream; + ZEBRA_RES res; + void *clientData; + RecType recType = 0; + + if (recordType && *recordType) + { + yaz_log(log_level_extract, + "Record type explicitly specified: %s", recordType); + recType = recType_byName(zh->reg->recTypes, zh->res, recordType, + &clientData); + } + else + { + if (!(zh->m_record_type)) + { + yaz_log(YLOG_WARN, "No such record type defined"); + return ZEBRA_FAIL; + } + yaz_log(log_level_extract, "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; + } + + zebra_create_stream_mem(&stream, buf, buf_size); + + res = zebra_extract_records_stream(zh, &stream, + action, + test_mode, + recordType, + sysno, + match_criteria, + fname, + recType, clientData); + stream.destroy(&stream); + return res; } -static void wordAddAny (const RecWord *p) +ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh, + struct ZebraRecStream *stream, + enum zebra_recctrl_action_t action, + int test_mode, + const char *recordType, + zint *sysno, + const char *match_criteria, + const char *fname, + RecType recType, + void *recTypeClientData) { - if (p->attrSet != 1 || p->attrUse != 1016) + ZEBRA_RES res = ZEBRA_OK; + while (1) { - RecWord w; - - memcpy (&w, p, sizeof(w)); - w.attrSet = 1; - w.attrUse = 1016; - wordAdd (&w); + int more = 0; + res = zebra_extract_record_stream(zh, stream, + action, + test_mode, + recordType, + sysno, + match_criteria, + fname, + recType, recTypeClientData, &more); + if (!more) + { + res = ZEBRA_OK; + break; + } + if (res != ZEBRA_OK) + break; + if (sysno) + break; } - wordAdd (p); + return res; } -static char *file_buf; -static int file_offset; -static int file_bufsize; -static void file_read_start (int fd) +static WRBUF wrbuf_hex_str(const char *cstr) { - file_offset = 0; - file_buf = xmalloc (4096); - file_bufsize = read (fd, file_buf, 4096); + size_t i; + WRBUF w = wrbuf_alloc(); + for (i = 0; cstr[i]; i++) + { + if (cstr[i] < ' ' || cstr[i] > 126) + wrbuf_printf(w, "\\%02X", cstr[i] & 0xff); + else + wrbuf_putc(w, cstr[i]); + } + return w; } -static void file_read_stop (int fd) -{ - xfree (file_buf); -} +ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, + struct ZebraRecStream *stream, + enum zebra_recctrl_action_t action, + int test_mode, + const char *recordType, + zint *sysno, + const char *match_criteria, + const char *fname, + RecType recType, + void *recTypeClientData, + int *more) -static int file_read (int fd, char *buf, size_t count) { - int l = file_bufsize - file_offset; + zint sysno0 = 0; + RecordAttr *recordAttr; + struct recExtractCtrl extractCtrl; + int r; + const char *matchStr = 0; + Record rec; + off_t start_offset = 0, end_offset = 0; + const char *pr_fname = fname; /* filename to print .. */ + int show_progress = zh->records_processed + zh->records_skipped + < 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])) + { + if (zebraExplain_newDatabase(zh->reg->zei, zh->basenames[0], + zh->m_explain_database)) + return ZEBRA_FAIL; + } - if (count > l) + if (stream) { - int r; - if (l > 0) - memcpy (buf, file_buf + file_offset, l); - count = count-l; - if (count > file_bufsize) + 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; + extractCtrl.action = action; + + init_extractCtrl(zh, &extractCtrl); + + extract_set_store_data_prepare(&extractCtrl); + + r = (*recType->extract)(recTypeClientData, &extractCtrl); + + if (action == action_update) { - if ((r = read (fd, buf + l, count)) == -1) + action = extractCtrl.action; + } + + switch (r) + { + case RECCTRL_EXTRACT_EOF: + return ZEBRA_FAIL; + case RECCTRL_EXTRACT_ERROR_GENERIC: + /* error occured during extraction ... */ + yaz_log(YLOG_WARN, "extract error: generic"); + return ZEBRA_FAIL; + case RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER: + /* error occured during extraction ... */ + yaz_log(YLOG_WARN, "extract error: no such filter"); + return ZEBRA_FAIL; + case RECCTRL_EXTRACT_SKIP: + if (show_progress) + yaz_log(YLOG_LOG, "skip %s %s " ZINT_FORMAT, + recordType, pr_fname, (zint) start_offset); + *more = 1; + + end_offset = stream->endf(stream, 0); + if (end_offset) + stream->seekf(stream, end_offset); + + return ZEBRA_OK; + case RECCTRL_EXTRACT_OK: + break; + default: + yaz_log(YLOG_WARN, "extract error: unknown error: %d", r); + return ZEBRA_FAIL; + } + end_offset = stream->endf(stream, 0); + if (end_offset) + stream->seekf(stream, end_offset); + else + end_offset = stream->tellf(stream); + + if (extractCtrl.match_criteria[0]) + match_criteria = extractCtrl.match_criteria; + } + + *more = 1; + if (!sysno) + { + sysno = &sysno0; + + if (match_criteria && *match_criteria) { + matchStr = match_criteria; + } else { + if (zh->m_record_id && *zh->m_record_id) { + matchStr = get_match_from_spec(zh, zh->reg->keys, pr_fname, + zh->m_record_id); + if (!matchStr) + { + yaz_log(YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType, + pr_fname, (zint) start_offset); + return ZEBRA_FAIL; + } + } + } + if (matchStr) + { + int db_ord = zebraExplain_get_database_ord(zh->reg->zei); + char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord, + matchStr); + + + if (log_level_extract) { - logf (LOG_FATAL|LOG_ERRNO, "read"); - exit (1); + WRBUF w = wrbuf_hex_str(matchStr); + yaz_log(log_level_extract, "matchStr: %s", wrbuf_cstr(w)); + wrbuf_destroy(w); } - file_bufsize = 0; - file_offset = 0; - return r; + 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 AKA does not exist already */ + if (action == action_delete) + { + 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; } - file_bufsize = r = read (fd, file_buf, 4096); - if (r == -1) + else if (action == action_a_delete) { - logf (LOG_FATAL|LOG_ERRNO, "read"); - exit (1); + if (show_progress) + yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType, + pr_fname, (zint) start_offset); + return ZEBRA_OK; } - else if (r <= count) + else if (action == action_replace) + { + yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType, + pr_fname, (zint) start_offset); + yaz_log(YLOG_WARN, "cannot update 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; + + + if (stream) { - file_offset = r; - memcpy (buf + l, file_buf, r); - return l + r; + all_matches_add(&extractCtrl, + zebra_rec_keys_get_custom_record_id(zh->reg->keys), + *sysno); } - else + + + recordAttr = rec_init_attr(zh->reg->zei, rec); + if (extractCtrl.staticrank < 0) { - file_offset = count; - memcpy (buf + l, file_buf, count - l); - return count; + yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0"); + extractCtrl.staticrank = 0; } - } - memcpy (buf, file_buf + file_offset, count); - file_offset += count; - return count; -} -SYSNO file_extract (int cmd, const char *fname, const char *kname, - char *databaseName) -{ - int i, r; - char ext[128]; - SYSNO sysno; - char ext_res[128]; - const char *file_type; - void *file_info; - struct recExtractCtrl extractCtrl; - RecType rt; + if (matchStr) + { + int db_ord = zebraExplain_get_database_ord(zh->reg->zei); + dict_insert_ord(zh->reg->matchDict, db_ord, matchStr, + sizeof(*sysno), sysno); + } - key_del_used = -1; - key_databaseName = databaseName; - for (i = strlen(fname); --i >= 0; ) - if (fname[i] == '/') + extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys); +#if FLUSH2 + extract_flush_record_keys2(zh, *sysno, + zh->reg->keys, extractCtrl.staticrank, + 0, recordAttr->staticrank); +#else + extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys, + extractCtrl.staticrank); +#endif + recordAttr->staticrank = extractCtrl.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 (action == action_insert) + { + 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); + + if (stream) { - strcpy (ext, ""); - break; + all_matches_add(&extractCtrl, + zebra_rec_keys_get_custom_record_id(zh->reg->keys), + *sysno); } - else if (fname[i] == '.') + + recordAttr = rec_init_attr(zh->reg->zei, rec); + + /* decrease total size */ + zebraExplain_recordBytesIncrement(zh->reg->zei, + - recordAttr->recordSize); + + 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_flush_sort_keys(zh, *sysno, 0, sortKeys); +#if !FLUSH2 + extract_flush_record_keys(zh, *sysno, 0, delkeys, + recordAttr->staticrank); +#endif + if (action == action_delete || action == action_a_delete) { - strcpy (ext, fname+i+1); - break; + /* record going to be deleted */ +#if FLUSH2 + extract_flush_record_keys2(zh, *sysno, 0, recordAttr->staticrank, + delkeys, recordAttr->staticrank); +#endif + if (zebra_rec_keys_empty(delkeys)) + { + 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); + } + zebra_rec_keys_close(delkeys); + zebra_rec_keys_close(sortKeys); + rec_free(&rec); + logRecord(zh); + return ZEBRA_OK; } - sprintf (ext_res, "fileExtension.%s", ext); - if (!(file_type = res_get (common_resource, ext_res))) - return 0; - if (!(rt = recType_byName (file_type))) - return 0; - logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname); - file_info = dict_lookup (file_idx, kname); - if (!file_info) - { -#if RECORD_BASE - Record rec = rec_new (records); - - sysno = rec->sysno; - dict_insert (file_idx, kname, sizeof(sysno), &sysno); - rec->info[0] = rec_strdup (file_type, &rec->size[0]); - rec->info[1] = rec_strdup (kname, &rec->size[1]); - rec_put (records, &rec); + else + { /* update or special_update */ + if (show_progress) + yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType, + pr_fname, (zint) start_offset); + extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys); + +#if FLUSH2 + extract_flush_record_keys2(zh, *sysno, + zh->reg->keys, extractCtrl.staticrank, + delkeys, recordAttr->staticrank); #else - sysno = sysno_next++; - dict_insert (file_idx, kname, sizeof(sysno), &sysno); - lseek (sys_idx_fd, sysno * SYS_IDX_ENTRY_LEN, SEEK_SET); - write (sys_idx_fd, file_type, strlen (file_type)+1); - write (sys_idx_fd, kname, strlen(kname)+1); + extract_flush_record_keys(zh, *sysno, 1, + zh->reg->keys, extractCtrl.staticrank); #endif + recordAttr->staticrank = extractCtrl.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 - memcpy (&sysno, (char*) file_info+1, sizeof(sysno)); + { + 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]); - if ((extractCtrl.fd = open (fname, O_RDONLY)) == -1) + if (stream) { - logf (LOG_WARN|LOG_ERRNO, "open %s", fname); - return 0; + recordAttr->recordSize = end_offset - start_offset; + zebraExplain_recordBytesIncrement(zh->reg->zei, + recordAttr->recordSize); } - extractCtrl.subType = ""; - extractCtrl.init = wordInit; - extractCtrl.add = wordAddAny; - file_read_start (extractCtrl.fd); + /* set run-number for this record */ + recordAttr->runNumber = + zebraExplain_runNumberIncrement(zh->reg->zei, 0); + + /* update store data */ + xfree(rec->info[recInfo_storeData]); - extractCtrl.readf = file_read; - key_sysno = sysno; - key_cmd = cmd; - r = (*rt->extract)(&extractCtrl); + /* 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; + recordAttr->recordSize = zh->store_data_size; + } + else if (zh->m_store_data) + { + off_t cur_offset = stream->tellf(stream); - file_read_stop (extractCtrl.fd); + 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(zh->basenames[0], &rec->size[recInfo_databaseName]); - close (extractCtrl.fd); - if (r) - logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r); - return sysno; + /* 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 char *databaseName, - int deleteFlag) +ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n) { - int i, r; - char ext[128]; - char ext_res[128]; - const char *file_type; + ZebraHandle zh = (ZebraHandle) handle; struct recExtractCtrl extractCtrl; - RecType rt; - Record rec; - logf (LOG_DEBUG, "fileExtractAdd %s", fname); + if (zebraExplain_curDatabase(zh->reg->zei, + rec->info[recInfo_databaseName])) + { + abort(); + if (zebraExplain_newDatabase(zh->reg->zei, + rec->info[recInfo_databaseName], 0)) + abort(); + } - key_del_used = 0; - for (i = strlen(fname); --i >= 0; ) - if (fname[i] == '/') + zebra_rec_keys_reset(zh->reg->keys); + zebra_rec_keys_reset(zh->reg->sortKeys); + + 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.action = action_update; + + 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]) + { + 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); +#if FLUSH2 + extract_flush_record_keys2(zh, rec->sysno, + zh->reg->keys, 0, delkeys, 0); +#else + extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0); + extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0); +#endif + zebra_rec_keys_close(delkeys); + + zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys], + rec->size[recInfo_sortKeys], + 0); + + extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys); + zebra_rec_keys_close(sortkeys); + } + else + { +#if FLUSH2 + extract_flush_record_keys2(zh, rec->sysno, zh->reg->keys, 0, 0, 0); +#else + extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0); +#endif + } + extract_flush_sort_keys(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_log(ZebraHandle zh, int is_insert, + zebra_rec_keys_t reckeys, + int level) +{ + if (zebra_rec_keys_rewind(reckeys)) + { + size_t slen; + const char *str; + struct it_key key; + NMEM nmem = nmem_create(); + + while(zebra_rec_keys_read(reckeys, &str, &slen, &key)) { - strcpy (ext, ""); - break; + char keystr[200]; /* room for zints to print */ + char *dst_term = 0; + int ord = CAST_ZINT_TO_INT(key.mem[0]); + const char *index_type; + int i; + const char *string_index; + + zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, + 0/* db */, &string_index); + assert(index_type); + zebra_term_untrans_iconv(zh, nmem, index_type, + &dst_term, str); + *keystr = '\0'; + for (i = 0; ireg->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)) { - strcpy (ext, fname+i+1); - break; + 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) + { + p = xmalloc(sizeof(*p)); + p->no = 1; + p->ord = ord; + p->next = ord_list; + ord_list = p; + } } - sprintf (ext_res, "fileExtension.%s", ext); - if (!(file_type = res_get (common_resource, ext_res))) - return 0; - if (!(rt = recType_byName (file_type))) - return 0; - if ((extractCtrl.fd = open (fname, O_RDONLY)) == -1) + 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); + } + } +} + +void extract_flush_record_keys2(ZebraHandle zh, zint sysno, + zebra_rec_keys_t ins_keys, zint ins_rank, + zebra_rec_keys_t del_keys, zint del_rank) +{ + ZebraExplainInfo zei = zh->reg->zei; + int normal = 0; + int optimized = 0; + + if (!zh->reg->key_block) { - logf (LOG_WARN|LOG_ERRNO, "open %s", fname); - return 0; + int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8")); + const char *key_tmp_dir = res_get_def(zh->res, "keyTmpDir", "."); + int use_threads = atoi(res_get_def(zh->res, "threads", "1")); + zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads); } - extractCtrl.subType = ""; - extractCtrl.init = wordInit; - extractCtrl.add = wordAddAny; + if (ins_keys) + { + extract_rec_keys_adjust(zh, 1, ins_keys); + if (!del_keys) + zebraExplain_recordCountIncrement(zei, 1); + zebra_rec_keys_rewind(ins_keys); + } + if (del_keys) + { + extract_rec_keys_adjust(zh, 0, del_keys); + if (!ins_keys) + zebraExplain_recordCountIncrement(zei, -1); + zebra_rec_keys_rewind(del_keys); + } - if (! *sysno) + while (1) { - logf (LOG_LOG, "add record %s", fname); - rec = rec_new (records); - *sysno = rec->sysno; - rec->info[0] = rec_strdup (file_type, &rec->size[0]); - rec->info[1] = rec_strdup (fname, &rec->size[1]); - rec->info[3] = rec_strdup (databaseName, &rec->size[3]); + size_t del_slen; + const char *del_str; + struct it_key del_key_in; + int del = 0; + + size_t ins_slen; + const char *ins_str; + struct it_key ins_key_in; + int ins = 0; + + if (del_keys) + del = zebra_rec_keys_read(del_keys, &del_str, &del_slen, + &del_key_in); + if (ins_keys) + ins = zebra_rec_keys_read(ins_keys, &ins_str, &ins_slen, + &ins_key_in); + + if (del && ins && ins_rank == del_rank + && !key_compare(&del_key_in, &ins_key_in) + && ins_slen == del_slen && !memcmp(del_str, ins_str, del_slen)) + { + optimized++; + continue; + } + if (!del && !ins) + break; + + normal++; + if (del) + key_block_write(zh->reg->key_block, sysno, + &del_key_in, 0, del_str, del_slen, + del_rank, zh->m_staticrank); + if (ins) + key_block_write(zh->reg->key_block, sysno, + &ins_key_in, 1, ins_str, ins_slen, + ins_rank, zh->m_staticrank); } - else + yaz_log(log_level_extract, "normal=%d optimized=%d", normal, optimized); +} + + +ZEBRA_RES zebra_rec_keys_to_snippets1(ZebraHandle zh, + zebra_rec_keys_t reckeys, + zebra_snippets *snippets) +{ + NMEM nmem = nmem_create(); + if (zebra_rec_keys_rewind(reckeys)) { - size_t off; - char *kb; + const char *str; + size_t slen; + struct it_key key; + while (zebra_rec_keys_read(reckeys, &str, &slen, &key)) + { + char *dst_term = 0; + int ord; + zint seqno; + const char *index_type; - if (deleteFlag) - logf (LOG_LOG, "delete record %s", fname); - else - logf (LOG_LOG, "update record %s", fname); - rec = rec_get (records, *sysno); - - key_cmd = 'd'; - key_sysno = *sysno; - key_databaseName = rec->info[3]; - kb = rec->info[2]; - for (off = 0; off < rec->size[2]; ) + 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, 0, 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; + const char *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); + + 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_staticrank_string(RecWord *p, + const char *str, int length) +{ + char valz[40]; + struct recExtractCtrl *ctrl = p->extractCtrl; + + if (length > sizeof(valz)-1) + length = sizeof(valz)-1; + + memcpy(valz, str, length); + valz[length] = '\0'; + ctrl->staticrank = atozint(valz); +} + +static void extract_add_string(RecWord *p, zebra_map_t zm, + const char *string, int length) +{ + assert(length > 0); + + if (!p->index_name) + return; + + if (zebra_maps_is_index(zm)) + { + extract_add_index_string(p, zinfo_index_category_index, + string, length); + if (zebra_maps_is_alwaysmatches(zm)) { - RecWord rw; - char attrSet; - short attrUse; - - rw.which = Word_String; - rw.u.string = kb + off; - while (kb[off]) - off++; - off++; - memcpy (&attrSet, kb + off, sizeof(attrSet)); - off += sizeof(attrSet); - memcpy (&attrUse, kb + off, sizeof(attrUse)); - off += sizeof(attrUse); - memcpy (&rw.seqno, kb + off, sizeof(rw.seqno)); - off += sizeof(rw.seqno); - rw.attrSet = attrSet; - rw.attrUse = attrUse; - - (*extractCtrl.add) (&rw); + RecWord word; + memcpy(&word, p, sizeof(word)); + + word.seqno = 1; + extract_add_index_string( + &word, zinfo_index_category_alwaysmatches, "", 0); } - assert (off == rec->size[2]); - free (rec->info[2]); - rec->info[2] = NULL; - rec->size[2] = 0; + } + else if (zebra_maps_is_sort(zm)) + { + extract_add_sort_string(p, string, length); + } + else if (zebra_maps_is_staticrank(zm)) + { + extract_add_staticrank_string(p, string, length); + } +} + +static void extract_add_incomplete_field(RecWord *p, zebra_map_t zm) +{ + const char *b = p->term_buf; + int remain = p->term_len; + int first = 1; + const char **map = 0; + + if (remain > 0) + map = zebra_maps_input(zm, &b, remain, 0); + + 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(zm, &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(zm, &b, remain, 0); + else + map = 0; + } + if (!i) + return; - free (rec->info[3]); - rec->info[3] = rec_strdup (databaseName, &rec->size[3]); + if (first) + { + first = 0; + if (zebra_maps_is_first_in_field(zm)) + { + /* first in field marker */ + extract_add_string(p, zm, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN); + p->seqno++; + } + } + extract_add_string(p, zm, buf, i); + p->seqno++; } +} - if (deleteFlag) +static void extract_add_complete_field(RecWord *p, zebra_map_t zm) +{ + 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(zm, &b, remain, 1); + + while (remain > 0 && i < IT_MAX_WORD) { -#if 0 - rec_del (records, *sysno); - rec_rm (&rec); -#endif - return 1; + 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(zm, &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(zm, &b, remain, 0); + } + else + map = 0; + } + } + if (!i) + return; + extract_add_string(p, zm, buf, i); +} + +static void extract_add_icu(RecWord *p, zebra_map_t zm) +{ + const char *res_buf = 0; + size_t res_len = 0; + + zebra_map_tokenize_start(zm, p->term_buf, p->term_len); + while (zebra_map_tokenize_next(zm, &res_buf, &res_len, 0, 0)) + { + extract_add_string(p, zm, res_buf, res_len); + p->seqno++; } +} + + +/** \brief top-level indexing handler for recctrl system + \param p token data to be indexed + + Call sequence: + extract_token_add + extract_add_{in}_complete / extract_add_icu + extract_add_string + + extract_add_index_string + or + extract_add_sort_string + or + extract_add_staticrank_string - key_databaseName = databaseName; - key_sysno = *sysno; - key_cmd = 'a'; +*/ +static void extract_token_add(RecWord *p) +{ + ZebraHandle zh = p->extractCtrl->handle; + zebra_map_t zm = zebra_map_get_or_add(zh->reg->zebra_maps, p->index_type); + WRBUF wrbuf; - file_read_start (extractCtrl.fd); - extractCtrl.readf = file_read; - r = (*rt->extract)(&extractCtrl); - file_read_stop (extractCtrl.fd); + if (log_level_details) + { + yaz_log(log_level_details, "extract_token_add " + "type=%s 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(zm, 0, p->term_buf, p->term_len))) + { + p->term_buf = wrbuf_buf(wrbuf); + p->term_len = wrbuf_len(wrbuf); + } + if (zebra_maps_is_icu(zm)) + { + extract_add_icu(p, zm); + } + else + { + if (zebra_maps_is_complete(zm)) + extract_add_complete_field(p, zm); + else + extract_add_incomplete_field(p, zm); + } +} + +static void extract_set_store_data_cb(struct recExtractCtrl *p, + void *buf, size_t sz) +{ + ZebraHandle zh = (ZebraHandle) p->handle; - close (extractCtrl.fd); - if (r) + xfree(zh->store_data_buf); + zh->store_data_buf = 0; + zh->store_data_size = 0; + if (buf && sz) { - rec_rm (&rec); - logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r); - return 0; + zh->store_data_buf = xmalloc(sz); + zh->store_data_size = sz; + memcpy(zh->store_data_buf, buf, sz); } - if (key_del_used > 0) +} + +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_flush_sort_keys(ZebraHandle zh, zint sysno, + int cmd, zebra_rec_keys_t reckeys) +{ +#if 0 + yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT, + cmd, sysno); + extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG); +#endif + + if (zebra_rec_keys_rewind(reckeys)) { - rec->size[2] = key_del_used; - rec->info[2] = malloc (rec->size[2]); - memcpy (rec->info[2], key_del_buf, rec->size[2]); + zebra_sort_index_t si = zh->reg->sort_index; + size_t slen; + const char *str; + struct it_key key_in; + + zebra_sort_sysno(si, sysno); + + while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in)) + { + int ord = CAST_ZINT_TO_INT(key_in.mem[0]); + + zebra_sort_type(si, ord); + if (cmd == 1) + zebra_sort_add(si, str, slen); + else + zebra_sort_delete(si); + } } - rec_put (records, &rec); - return 1; } + +/* + * Local variables: + * c-basic-offset: 4 + * indent-tabs-mode: nil + * End: + * vim: shiftwidth=4 tabstop=8 expandtab + */ +