+#else
+static void extract_flush_record_keys(
+ ZebraHandle zh, zint sysno, int cmd,
+ zebra_rec_keys_t reckeys,
+ zint staticrank)
+{
+ ZebraExplainInfo zei = zh->reg->zei;
+
+ extract_rec_keys_adjust(zh, cmd, reckeys);
+
+ if (log_level_details)
+ {
+ yaz_log(log_level_details, "Keys for record " ZINT_FORMAT " %s",
+ sysno, cmd ? "insert" : "delete");
+ extract_rec_keys_log(zh, cmd, reckeys, log_level_details);
+ }
+
+ if (!zh->reg->key_block)
+ {
+ 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);
+ }
+ zebraExplain_recordCountIncrement(zei, cmd ? 1 : -1);
+
+#if 0
+ yaz_log(YLOG_LOG, "sysno=" ZINT_FORMAT " cmd=%d", sysno, cmd);
+ print_rec_keys(zh, reckeys);
+#endif
+ if (zebra_rec_keys_rewind(reckeys))
+ {
+ size_t slen;
+ const char *str;
+ struct it_key key_in;
+ while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
+ {
+ key_block_write(zh->reg->key_block, sysno,
+ &key_in, cmd, str, slen,
+ staticrank, zh->m_staticrank);
+ }
+ }
+}
+#endif
+
+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))
+ {
+ 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;
+
+ 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 = 3;
+ key.mem[0] = ch;
+ key.mem[1] = p->record_id;
+ key.mem[2] = p->section_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 (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,
+ string, length);
+ if (zebra_maps_is_alwaysmatches(zm))
+ {
+ RecWord word;
+ memcpy(&word, p, sizeof(word));
+
+ word.seqno = 1;
+ extract_add_index_string(
+ &word, zinfo_index_category_alwaysmatches, "", 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;
+
+ 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++;
+ }
+}
+
+static void extract_add_complete_field(RecWord *p, zebra_map_t zm)
+{
+ char buf[IT_MAX_WORD+1];
+ int i = parse_complete_field(p, zm, buf);
+ if (!i)
+ return;
+ extract_add_string(p, zm, buf, i);
+ p->seqno++;
+}
+
+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))
+ {
+ if (res_len > IT_MAX_WORD)
+ {
+ yaz_log(YLOG_LOG, "Truncating long term %ld", (long) res_len);
+ res_len = IT_MAX_WORD;
+ }
+ 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
+
+*/
+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);
+
+ 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 (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;
+
+ 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_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))
+ {
+ zebra_sort_index_t si = zh->reg->sort_index;
+ size_t slen;
+ const char *str;
+ struct it_key key_in;
+
+ NMEM nmem = nmem_create();
+ struct sort_add_ent {
+ int ord;
+ int cmd;
+ struct sort_add_ent *next;
+ WRBUF wrbuf;
+ zint sysno;
+ zint section_id;
+ };
+ 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];
+ zint section_id = key_in.mem[2];
+
+ struct sort_add_ent **e = &sort_ent_list;
+ for (; *e; e = &(*e)->next)
+ if ((*e)->ord == ord && section_id == (*e)->section_id)
+ break;
+ 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;
+ (*e)->section_id = section_id;
+ }
+
+ 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->section_id, e->wrbuf);
+ else
+ zebra_sort_delete(si, e->section_id);
+ wrbuf_destroy(e->wrbuf);
+ }
+ }
+ nmem_destroy(nmem);
+ }
+}
+
+/*
+ * Local variables:
+ * c-basic-offset: 4
+ * c-file-style: "Stroustrup"
+ * indent-tabs-mode: nil
+ * End:
+ * vim: shiftwidth=4 tabstop=8 expandtab
+ */
+