X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Fextract.c;h=32b92eced8576225a525edeadd93ba958c98a1ba;hp=5be148815d0c6c1c593b9f652ef5a71f48211f21;hb=2cf8341fde67a92369c46d9b12e6d056e5bc5a0e;hpb=3731bdaf94aeba2550fc553aebe34831c203dc36 diff --git a/index/extract.c b/index/extract.c index 5be1488..32b92ec 100644 --- a/index/extract.c +++ b/index/extract.c @@ -1,8 +1,5 @@ -/* $Id: extract.c,v 1.270 2007-11-30 12:19:08 adam Exp $ - Copyright (C) 1995-2007 - Index Data ApS - -This file is part of the Zebra server. +/* This file is part of the Zebra server. + Copyright (C) 1995-2008 Index Data 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 @@ -187,7 +184,7 @@ static void snippet_add_complete_field(RecWord *p, int ord, } if (!i) return; - if (last && start != last) + if (last && start != last && zebra_maps_is_index(zm)) zebra_snippets_appendn(h->snippets, p->seqno, 0, ord, start, last - start); } @@ -222,7 +219,7 @@ static void snippet_add_incomplete_field(RecWord *p, int ord, zebra_map_t zm) } if (!map) break; - if (start != last) + if (start != last && zebra_maps_is_index(zm)) { zebra_snippets_appendn(h->snippets, p->seqno, 1, ord, start, last - start); @@ -256,7 +253,7 @@ static void snippet_add_incomplete_field(RecWord *p, int ord, zebra_map_t zm) p->seqno++; } } - if (start != last) + if (start != last && zebra_maps_is_index(zm)) zebra_snippets_appendn(h->snippets, p->seqno, 0, ord, start, last - start); start = last; @@ -265,22 +262,48 @@ static void snippet_add_incomplete_field(RecWord *p, int ord, zebra_map_t zm) } +static void snippet_add_icu(RecWord *p, int ord, zebra_map_t zm) +{ + struct snip_rec_info *h = p->extractCtrl->handle; + + const char *res_buf = 0; + size_t res_len = 0; + + 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)) + { + if (zebra_maps_is_index(zm)) + zebra_snippets_appendn(h->snippets, p->seqno, 0, ord, + display_buf, display_len); + p->seqno++; + } +} + 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)) + if (zm) { 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_complete(zm)) - snippet_add_complete_field(p, ch, zm); + if (zebra_maps_is_icu(zm)) + snippet_add_icu(p, ch, zm); else - snippet_add_incomplete_field(p, ch, zm); + { + if (zebra_maps_is_complete(zm)) + snippet_add_complete_field(p, ch, zm); + else + snippet_add_incomplete_field(p, ch, zm); + } } } @@ -560,7 +583,7 @@ ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh, ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname, - int deleteFlag) + enum zebra_recctrl_action_t action) { ZEBRA_RES r = ZEBRA_OK; int i, fd; @@ -630,7 +653,7 @@ ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname, default: yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type); } - if (sysno && deleteFlag) + if (sysno && (action == action_delete || action == action_a_delete)) { streamp = 0; fi = 0; @@ -658,8 +681,7 @@ ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname, zebra_create_stream_fd(streamp, fd, 0); } r = zebra_extract_records_stream(zh, streamp, - deleteFlag ? - action_delete : action_update, + action, 0, /* tst_mode */ zh->m_record_type, sysno, @@ -944,14 +966,21 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, if (! *sysno) { - /* new record */ + /* 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_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; } + else if (action == action_a_delete) + { + if (show_progress) + yaz_log(YLOG_LOG, "adelete %s %s " ZINT_FORMAT, recordType, + pr_fname, (zint) start_offset); + return ZEBRA_OK; + } else if (action == action_replace) { yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType, @@ -1044,7 +1073,7 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, extract_flush_record_keys(zh, *sysno, 0, delkeys, recordAttr->staticrank); #endif - if (action == action_delete) + if (action == action_delete || action == action_a_delete) { /* record going to be deleted */ #if FLUSH2 @@ -1456,7 +1485,7 @@ void extract_flush_record_keys2(ZebraHandle zh, zint sysno, } -ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh, +ZEBRA_RES zebra_rec_keys_to_snippets1(ZebraHandle zh, zebra_rec_keys_t reckeys, zebra_snippets *snippets) { @@ -1583,7 +1612,15 @@ static void extract_add_string(RecWord *p, zebra_map_t zm, if (!p->index_name) return; + if (log_level_details) + { + WRBUF w = wrbuf_alloc(); + + wrbuf_write_escaped(w, string, length); + yaz_log(log_level_details, "extract_add_string: %s", wrbuf_cstr(w)); + wrbuf_destroy(w); + } if (zebra_maps_is_index(zm)) { extract_add_index_string(p, zinfo_index_category_index, @@ -1725,32 +1762,14 @@ static void extract_add_complete_field(RecWord *p, zebra_map_t zm) static void extract_add_icu(RecWord *p, zebra_map_t zm) { - struct it_key key; const char *res_buf = 0; size_t res_len = 0; - ZebraHandle zh = p->extractCtrl->handle; - int r = zebra_map_tokenize(zm, p->term_buf, p->term_len, - &res_buf, &res_len); - int cat = zinfo_index_category_index; - int ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, p->index_type, p->index_name); - if (ch < 0) - ch = zebraExplain_add_attr_str(zh->reg->zei, cat, p->index_type, p->index_name); - while (r) - { - int 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, res_buf, res_len, &key); - + 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++; - r = zebra_map_tokenize(zm, 0, 0, &res_buf, &res_len); } } @@ -1759,8 +1778,8 @@ static void extract_add_icu(RecWord *p, zebra_map_t zm) \param p token data to be indexed Call sequence: - extract_token - zebra_add_{in}_complete + extract_token_add + extract_add_{in}_complete / extract_add_icu extract_add_string extract_add_index_string @@ -1848,18 +1867,57 @@ void extract_flush_sort_keys(ZebraHandle zh, zint sysno, const char *str; struct it_key key_in; - zebra_sort_sysno(si, sysno); + NMEM nmem = nmem_create(); + struct sort_add_ent { + int ord; + int cmd; + struct sort_add_ent *next; + WRBUF wrbuf; + zint sysno; + }; + struct sort_add_ent *sort_ent_list = 0; while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in)) { int ord = CAST_ZINT_TO_INT(key_in.mem[0]); + zint filter_sysno = key_in.mem[1]; + + struct sort_add_ent **e = &sort_ent_list; + while (*e && (*e)->ord != ord) + e = &(*e)->next; + if (!*e) + { + *e = nmem_malloc(nmem, sizeof(**e)); + (*e)->next = 0; + (*e)->wrbuf = wrbuf_alloc(); + (*e)->ord = ord; + (*e)->cmd = cmd; + (*e)->sysno = filter_sysno ? filter_sysno : sysno; + } - zebra_sort_type(si, ord); - if (cmd == 1) - zebra_sort_add(si, str, slen); - else - zebra_sort_delete(si); + wrbuf_write((*e)->wrbuf, str, slen); + wrbuf_putc((*e)->wrbuf, '\0'); + } + if (sort_ent_list) + { + zint last_sysno = 0; + struct sort_add_ent *e = sort_ent_list; + for (; e; e = e->next) + { + if (last_sysno != e->sysno) + { + zebra_sort_sysno(si, e->sysno); + last_sysno = e->sysno; + } + zebra_sort_type(si, e->ord); + if (e->cmd == 1) + zebra_sort_add(si, e->wrbuf); + else + zebra_sort_delete(si); + wrbuf_destroy(e->wrbuf); + } } + nmem_destroy(nmem); } }