X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=51d2b94feec0ec7eefa21053a1c9db0270344c57;hb=cf66499bac7c49c5bdd363a2c927295fa92f547a;hp=7dcbbee1df931f002d30ff2b95f4d167781a59c1;hpb=12a519a0bc231a70b7c3211cc21e9fb1f98510f0;p=idzebra-moved-to-github.git diff --git a/index/extract.c b/index/extract.c index 7dcbbee..51d2b94 100644 --- a/index/extract.c +++ b/index/extract.c @@ -1,5 +1,5 @@ -/* $Id: extract.c,v 1.243 2006-11-28 08:44:10 adam Exp $ - Copyright (C) 1995-2006 +/* $Id: extract.c,v 1.250 2007-03-01 10:35:46 adam Exp $ + Copyright (C) 1995-2007 Index Data ApS This file is part of the Zebra server. @@ -59,8 +59,18 @@ static void extract_flush_sort_keys(ZebraHandle zh, zint sysno, static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid); static void extract_token_add (RecWord *p); +static void check_log_limit(ZebraHandle zh) +{ + if (zh->records_processed + zh->records_skipped == zh->m_file_verbose_limit) + { + yaz_log(YLOG_LOG, "More than %d file log entries. Omitting rest", + zh->m_file_verbose_limit); + } +} + static void logRecord (ZebraHandle zh) { + check_log_limit(zh); ++zh->records_processed; if (!(zh->records_processed % 1000)) { @@ -351,8 +361,11 @@ ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname, } if (!zh->m_record_type) { - if (zh->records_processed < zh->m_file_verbose_limit) + 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 */ @@ -521,7 +534,8 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, 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->m_file_verbose_limit ? 1:0; + int show_progress = zh->records_processed + zh->records_skipped + < zh->m_file_verbose_limit ? 1:0; zebra_init_log_level(); @@ -562,38 +576,48 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, 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) + + 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; - } - else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER) - { + 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); + all_matches_add(&extractCtrl); if (extractCtrl.match_criteria[0]) match_criteria = extractCtrl.match_criteria; - - - end_offset = stream->endf(stream, 0); - - if (!end_offset) - end_offset = stream->tellf(stream); - else - stream->seekf(stream, end_offset); - } - *more = 1; if (!sysno) { @@ -1064,8 +1088,7 @@ ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh, struct it_key key; while (zebra_rec_keys_read(reckeys, &str, &slen, &key)) { - char dst_buf[IT_MAX_WORD]; - char *dst_term = dst_buf; + char *dst_term = 0; int ord; zint seqno; int index_type; @@ -1159,6 +1182,20 @@ static void extract_add_sort_string(RecWord *p, const char *str, int length) 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, const char *string, int length) { ZebraHandle zh = p->extractCtrl->handle; @@ -1167,9 +1204,7 @@ static void extract_add_string(RecWord *p, const char *string, int length) 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 + if (zebra_maps_is_index(zh->reg->zebra_maps, p->index_type)) { extract_add_index_string(p, zinfo_index_category_index, string, length); @@ -1183,6 +1218,14 @@ static void extract_add_string(RecWord *p, const char *string, int length) &word, zinfo_index_category_alwaysmatches, "", 0); } } + else if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type)) + { + extract_add_sort_string(p, string, length); + } + else if (zebra_maps_is_staticrank(zh->reg->zebra_maps, p->index_type)) + { + extract_add_staticrank_string(p, string, length); + } } static void extract_add_incomplete_field(RecWord *p) @@ -1190,20 +1233,12 @@ static void extract_add_incomplete_field(RecWord *p) ZebraHandle zh = p->extractCtrl->handle; 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(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]; @@ -1236,6 +1271,17 @@ static void extract_add_incomplete_field(RecWord *p) } if (!i) return; + + if (first) + { + first = 0; + 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++; + } + } extract_add_string (p, buf, i); p->seqno++; } @@ -1359,24 +1405,30 @@ static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *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)) { - SortIdx sortIdx = zh->reg->sortIdx; + zebra_sort_index_t si = zh->reg->sort_index; size_t slen; const char *str; struct it_key key_in; - sortIdx_sysno (sortIdx, sysno); + 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]); - sortIdx_type(sortIdx, ord); + zebra_sort_type(si, ord); if (cmd == 1) - sortIdx_add(sortIdx, str, slen); + zebra_sort_add(si, str, slen); else - sortIdx_add(sortIdx, "", 1); + zebra_sort_delete(si); } } }