X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fsortidx.c;h=5490139e492c91ff0f0592988ee8f0b56376e541;hb=78b13a3ac6a79768fb609c14db2a8e0c94a9c4da;hp=65bff6bbc1d75f558c8fe64e769902b78d1044e4;hpb=0ef68e6e54cfd7a91a28ee3b1f305924ac34e787;p=idzebra-moved-to-github.git diff --git a/index/sortidx.c b/index/sortidx.c index 65bff6b..5490139 100644 --- a/index/sortidx.c +++ b/index/sortidx.c @@ -47,7 +47,7 @@ static void sort_term_log_item(int level, const void *b, const char *txt) memcpy(&a1, b, sizeof(a1)); yaz_log(level, "%s " ZINT_FORMAT " %.*s", txt, a1.sysno, - (int) a1.length, a1.term); + (int) a1.length-1, a1.term); } static int sort_term_compare(const void *a, const void *b) @@ -102,8 +102,8 @@ static void sort_term_decode1(void *p, char **dst, const char **src) zebra_zint_decode(src, &a1.sysno); strcpy(a1.term, *src); - slen = strlen(a1.term); - *src += slen + 1; + slen = 1 + strlen(a1.term); + *src += slen; a1.length = slen; memcpy(*dst, &a1, sizeof(a1)); @@ -315,25 +315,26 @@ int zebra_sort_type(zebra_sort_index_t si, int id) return 0; } +static void zebra_sortf_rewind(struct sortFile *sf) +{ + if (sf->isam_pp) + isamb_pp_close(sf->isam_pp); + sf->isam_pp = 0; + sf->no_inserted = 0; + sf->no_deleted = 0; +} + void zebra_sort_sysno(zebra_sort_index_t si, zint sysno) { - struct sortFile *sf = si->current_file; zint new_sysno = rec_sysno_to_int(sysno); + struct sortFile *sf; for (sf = si->files; sf; sf = sf->next) { if (sf->no_inserted || sf->no_deleted) - { - isamb_pp_close(sf->isam_pp); - sf->isam_pp = 0; - } - else if (sf->isam_pp && new_sysno < si->sysno && sf->isam_pp) - { - isamb_pp_close(sf->isam_pp); - sf->isam_pp = 0; - } - sf->no_inserted = 0; - sf->no_deleted = 0; + zebra_sortf_rewind(sf); + else if (sf->isam_pp && new_sysno <= si->sysno) + zebra_sortf_rewind(sf); } si->sysno = new_sysno; } @@ -348,7 +349,8 @@ void zebra_sort_delete(zebra_sort_index_t si) switch(si->type) { case ZEBRA_SORT_TYPE_FLAT: - zebra_sort_add(si, "", 0); + memset(si->entry_buf, 0, SORT_IDX_ENTRYSIZE); + bf_write(sf->u.bf, si->sysno+1, 0, 0, si->entry_buf); break; case ZEBRA_SORT_TYPE_ISAMB: case ZEBRA_SORT_TYPE_MULTI: @@ -374,40 +376,44 @@ void zebra_sort_delete(zebra_sort_index_t si) } } -void zebra_sort_add(zebra_sort_index_t si, const char *buf, int len) +void zebra_sort_add(zebra_sort_index_t si, WRBUF wrbuf) { struct sortFile *sf = si->current_file; + int len; if (!sf || !sf->u.bf) return; switch(si->type) { case ZEBRA_SORT_TYPE_FLAT: + /* take first entry from wrbuf - itself is 0-terminated */ + len = strlen(wrbuf_buf(wrbuf)); if (len > SORT_IDX_ENTRYSIZE) - { len = SORT_IDX_ENTRYSIZE; - memcpy(si->entry_buf, buf, len); - } - else - { - memcpy(si->entry_buf, buf, len); + + memcpy(si->entry_buf, wrbuf_buf(wrbuf), len); + if (len < SORT_IDX_ENTRYSIZE-len) memset(si->entry_buf+len, 0, SORT_IDX_ENTRYSIZE-len); - } bf_write(sf->u.bf, si->sysno+1, 0, 0, si->entry_buf); break; case ZEBRA_SORT_TYPE_ISAMB: assert(sf->u.isamb); + if (sf->no_inserted == 0) { struct sort_term_stream s; ISAMC_I isamc_i; + /* take first entry from wrbuf - itself is 0-terminated */ - s.st.sysno = si->sysno; - if (len >= SORT_MAX_TERM) - len = SORT_MAX_TERM-1; - memcpy(s.st.term, buf, len); - s.st.term[len] = '\0'; + len = wrbuf_len(wrbuf); + if (len > SORT_MAX_TERM) + { + len = SORT_MAX_TERM; + wrbuf_buf(wrbuf)[len-1] = '\0'; + } + memcpy(s.st.term, wrbuf_buf(wrbuf), len); s.st.length = len; + s.st.sysno = si->sysno; s.no = 1; s.insert_flag = 1; isamc_i.clientData = &s; @@ -423,12 +429,15 @@ void zebra_sort_add(zebra_sort_index_t si, const char *buf, int len) { struct sort_term_stream s; ISAMC_I isamc_i; - - s.st.sysno = si->sysno; - if (len >= SORT_MAX_MULTI) - len = SORT_MAX_MULTI-1; - memcpy(s.st.term, buf, len); + len = wrbuf_len(wrbuf); + if (len > SORT_MAX_MULTI) + { + len = SORT_MAX_MULTI; + wrbuf_buf(wrbuf)[len-1] = '\0'; + } + memcpy(s.st.term, wrbuf_buf(wrbuf), len); s.st.length = len; + s.st.sysno = si->sysno; s.no = 1; s.insert_flag = 1; isamc_i.clientData = &s; @@ -441,6 +450,7 @@ void zebra_sort_add(zebra_sort_index_t si, const char *buf, int len) } } + int zebra_sort_read(zebra_sort_index_t si, WRBUF w) { int r; @@ -455,43 +465,37 @@ int zebra_sort_read(zebra_sort_index_t si, WRBUF w) case ZEBRA_SORT_TYPE_FLAT: r = bf_read(sf->u.bf, si->sysno+1, 0, 0, tbuf); if (r && *tbuf) + { wrbuf_puts(w, tbuf); - else - return 0; + wrbuf_putc(w, '\0'); + return 1; + } break; case ZEBRA_SORT_TYPE_ISAMB: case ZEBRA_SORT_TYPE_MULTI: - if (!sf->isam_p) - return 0; - else + if (sf->isam_p) { - struct sort_term st, st_untilbuf; if (!sf->isam_pp) sf->isam_pp = isamb_pp_open(sf->u.isamb, sf->isam_p, 1); - if (!sf->isam_pp) - return 0; - - st_untilbuf.sysno = si->sysno; - st_untilbuf.length = 0; - st_untilbuf.term[0] = '\0'; - r = isamb_pp_forward(sf->isam_pp, &st, &st_untilbuf); - if (!r) - return 0; - if (r) + if (sf->isam_pp) { - if (st.sysno != si->sysno) + struct sort_term st, st_untilbuf; + + st_untilbuf.sysno = si->sysno; + st_untilbuf.length = 0; + st_untilbuf.term[0] = '\0'; + r = isamb_pp_forward(sf->isam_pp, &st, &st_untilbuf); + if (r && st.sysno == si->sysno) { - yaz_log(YLOG_LOG, "Received sysno=" ZINT_FORMAT " looking for " - ZINT_FORMAT, st.sysno, si->sysno); - return 0; + wrbuf_write(w, st.term, st.length); + return 1; } - wrbuf_write(w, st.term, st.length); } } break; } - return 1; + return 0; } /* * Local variables: