X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fretrieve.c;h=1afe24f6aec635d8ba6400ce264c868a2d878c7f;hb=161aa7805930f6b91ccea027e4afa4ccce41c379;hp=010c3c77c48618010d40b199542df483708de241;hpb=7a41489c9fb0495b4baaad3a00190c5b5a876ead;p=idzebra-moved-to-github.git diff --git a/index/retrieve.c b/index/retrieve.c index 010c3c7..1afe24f 100644 --- a/index/retrieve.c +++ b/index/retrieve.c @@ -1,4 +1,4 @@ -/* $Id: retrieve.c,v 1.49 2006-11-14 14:32:13 marc Exp $ +/* $Id: retrieve.c,v 1.61 2006-12-18 23:40:07 adam Exp $ Copyright (C) 1995-2006 Index Data ApS @@ -37,7 +37,7 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include - +#define ZEBRA_XML_HEADER_STR "reg->zei, *rec); - if ((*rec)->size[recInfo_storeData] > 0){ + if ((*rec)->size[recInfo_storeData] > 0) zebra_create_stream_mem(stream, (*rec)->info[recInfo_storeData], (*rec)->size[recInfo_storeData]); - } - else { + else + { char full_rep[1024]; int fd; @@ -74,31 +74,40 @@ static int zebra_create_record_stream(ZebraHandle zh, -static void parse_zebra_elem(const char *elem, - const char **index, size_t *index_len, - const char **type, size_t *type_len) +static int parse_zebra_elem(const char *elem, + const char **index, size_t *index_len, + const char **type, size_t *type_len) { - *type = 0; - *type_len = 0; - *index = 0; *index_len = 0; + *type = 0; + *type_len = 0; + if (elem && *elem) { - const char *cp = strchr(elem, ':'); + char *cp; + /* verify that '::' is in the beginning of *elem + and something more follows */ + if (':' != *elem + || !(elem +1) || ':' != *(elem +1) + || !(elem +2) || '\0' == *(elem +2)) + return 0; + + /* pick out info from string after '::' */ + elem = elem + 2; + cp = strchr(elem, ':'); - if (!cp) /* no colon */ + if (!cp) /* index, no colon, no type */ { *index = elem; *index_len = strlen(elem); } - else if (cp[1] == '\0') /* 'index:' */ + else if (cp[1] == '\0') /* colon, but no following type */ { - *index = elem; - *index_len = cp - elem; + return 0; } - else + else /* index, colon and type */ { *index = elem; *index_len = cp - elem; @@ -106,65 +115,184 @@ static void parse_zebra_elem(const char *elem, *type_len = strlen(cp+1); } } + return 1; } -int zebra_special_index_fetch(ZebraHandle zh, SYSNO sysno, ODR odr, - Record rec, - const char *elemsetname, - oid_value input_format, - oid_value *output_format, - char **rec_bufp, int *rec_lenp) +int zebra_special_sort_fetch(ZebraHandle zh, zint sysno, ODR odr, + const char *elemsetname, + oid_value input_format, + oid_value *output_format, + char **rec_bufp, int *rec_lenp) { const char *retrieval_index; size_t retrieval_index_len; const char *retrieval_type; size_t retrieval_type_len; - - /* int return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; */ + char retrieval_index_cstr[256]; + int ord; + + /* only accept XML and SUTRS requests */ + if (input_format != VAL_TEXT_XML && input_format != VAL_SUTRS) + { + yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s", + elemsetname); + *output_format = VAL_NONE; + return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST; + } + + if (!parse_zebra_elem(elemsetname, + &retrieval_index, &retrieval_index_len, + &retrieval_type, &retrieval_type_len)) + { + return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; + } + + if (retrieval_type_len != 0 && retrieval_type_len != 1) + { + return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; + } + if (!retrieval_index_len || + retrieval_index_len >= sizeof(retrieval_index_cstr)-1) + { + return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; + } + + memcpy(retrieval_index_cstr, retrieval_index, retrieval_index_len); + retrieval_index_cstr[retrieval_index_len] = '\0'; + + ord = zebraExplain_lookup_attr_str(zh->reg->zei, + zinfo_index_category_sort, + (retrieval_type_len == 0 ? -1 : + retrieval_type[0]), + retrieval_index_cstr); + if (ord == -1) + return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; + else + { + char dst_buf[IT_MAX_WORD]; + char str[IT_MAX_WORD]; + int index_type; + const char *db = 0; + const char *string_index = 0; + WRBUF wrbuf = wrbuf_alloc(); + + zebra_sort_sysno(zh->reg->sort_index, sysno); + zebra_sort_type(zh->reg->sort_index, ord); + zebra_sort_read(zh->reg->sort_index, str); - WRBUF wrbuf = wrbuf_alloc(); + zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, &string_index); + + zebra_term_untrans(zh, index_type, dst_buf, str); + + + if (input_format == VAL_TEXT_XML) + { + *output_format = VAL_TEXT_XML; + wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR + " sysno=\"" ZINT_FORMAT "\"" + " set=\"zebra::sort%s/\">\n", + sysno, elemsetname); + + wrbuf_printf(wrbuf, " ", index_type); + wrbuf_xmlputs(wrbuf, dst_buf); + wrbuf_printf(wrbuf, "\n"); + wrbuf_printf(wrbuf, "\n"); + } + else if (input_format == VAL_SUTRS) + { + *output_format = VAL_SUTRS; + + wrbuf_printf(wrbuf, "%s %c %s\n", string_index, index_type, + dst_buf); + } + *rec_lenp = wrbuf_len(wrbuf); + *rec_bufp = odr_malloc(odr, *rec_lenp); + memcpy(*rec_bufp, wrbuf_buf(wrbuf), *rec_lenp); + wrbuf_free(wrbuf, 1); + return 0; + } +} + +int zebra_special_index_fetch(ZebraHandle zh, zint sysno, ODR odr, + Record rec, + const char *elemsetname, + oid_value input_format, + oid_value *output_format, + char **rec_bufp, int *rec_lenp) +{ + const char *retrieval_index; + size_t retrieval_index_len; + const char *retrieval_type; + size_t retrieval_type_len; + WRBUF wrbuf = 0; zebra_rec_keys_t keys; + /* set output variables before processing possible error states */ + /* *rec_lenp = 0; */ + /* only accept XML and SUTRS requests */ - if (input_format != VAL_TEXT_XML - && input_format != VAL_SUTRS){ + if (input_format != VAL_TEXT_XML && input_format != VAL_SUTRS) + { yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s", elemsetname); *output_format = VAL_NONE; - return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; + return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST; + } + + if (!parse_zebra_elem(elemsetname, + &retrieval_index, &retrieval_index_len, + &retrieval_type, &retrieval_type_len)) + return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; + + if (retrieval_type_len != 0 && retrieval_type_len != 1) + { + return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; + } + + if (retrieval_index_len) + { + char retrieval_index_cstr[256]; + + if (retrieval_index_len < sizeof(retrieval_index_cstr) -1) + { + memcpy(retrieval_index_cstr, retrieval_index, retrieval_index_len); + retrieval_index_cstr[retrieval_index_len] = '\0'; + + if (zebraExplain_lookup_attr_str(zh->reg->zei, + zinfo_index_category_index, + (retrieval_type_len == 0 ? -1 : + retrieval_type[0]), + retrieval_index_cstr) == -1) + return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; + } } keys = zebra_rec_keys_open(); zebra_rec_keys_set_buf(keys, rec->info[recInfo_delKeys], rec->size[recInfo_delKeys], 0); - parse_zebra_elem(elemsetname, - &retrieval_index, &retrieval_index_len, - &retrieval_type, &retrieval_type_len); - + wrbuf = wrbuf_alloc(); if (zebra_rec_keys_rewind(keys)){ size_t slen; const char *str; struct it_key key_in; - if (input_format == VAL_TEXT_XML){ + if (input_format == VAL_TEXT_XML) + { *output_format = VAL_TEXT_XML; - /*wrbuf_printf(wrbuf, - "\n");*/ - - wrbuf_printf(wrbuf, - "\n", + " set=\"zebra::index%s/\">\n", sysno, elemsetname); } else if (input_format == VAL_SUTRS) *output_format = VAL_SUTRS; - - while(zebra_rec_keys_read(keys, &str, &slen, &key_in)){ + while (zebra_rec_keys_read(keys, &str, &slen, &key_in)) + { int i; int ord = CAST_ZINT_TO_INT(key_in.mem[0]); int index_type; @@ -176,49 +304,56 @@ int zebra_special_index_fetch(ZebraHandle zh, SYSNO sysno, ODR odr, zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, &string_index); string_index_len = strlen(string_index); + + /* process only if index is not defined, + or if defined and matching */ if (retrieval_index == 0 || (string_index_len == retrieval_index_len && !memcmp(string_index, retrieval_index, - string_index_len))){ - + string_index_len))) + { + /* process only if type is not defined, or is matching */ if (retrieval_type == 0 || (retrieval_type_len == 1 - && retrieval_type[0] == index_type)){ - - if (input_format == VAL_TEXT_XML){ - wrbuf_printf(wrbuf, " ", + key_in.mem[key_in.len -1]); - wrbuf_printf(wrbuf, " seq=\"" ZINT_FORMAT "\">", - key_in.mem[key_in.len -1]); + wrbuf_xmlputs(wrbuf, dst_buf); + wrbuf_printf(wrbuf, "\n"); + } + else if (input_format == VAL_SUTRS){ + wrbuf_printf(wrbuf, "%s ", string_index); + + wrbuf_printf(wrbuf, "%c", index_type); + + for (i = 1; i < key_in.len; i++) + wrbuf_printf(wrbuf, " " ZINT_FORMAT, + key_in.mem[i]); + + /* zebra_term_untrans(zh, index_type, dst_buf, str); */ + wrbuf_printf(wrbuf, " %s", dst_buf); - zebra_term_untrans(zh, index_type, dst_buf, str); - wrbuf_xmlputs(wrbuf, dst_buf); - wrbuf_printf(wrbuf, "\n"); + wrbuf_printf(wrbuf, "\n"); + } } - else if (input_format == VAL_SUTRS){ - wrbuf_printf(wrbuf, "%s ", string_index); - - wrbuf_printf(wrbuf, "%c", index_type); - for (i = 1; i < key_in.len; i++) - wrbuf_printf(wrbuf, " " ZINT_FORMAT, - key_in.mem[i]); - - zebra_term_untrans(zh, index_type, dst_buf, str); - wrbuf_printf(wrbuf, " %s", dst_buf); - - wrbuf_printf(wrbuf, "\n"); - } } } } if (input_format == VAL_TEXT_XML) wrbuf_printf(wrbuf, "\n"); - } - + } *rec_lenp = wrbuf_len(wrbuf); *rec_bufp = odr_malloc(odr, *rec_lenp); memcpy(*rec_bufp, wrbuf_buf(wrbuf), *rec_lenp); @@ -228,72 +363,100 @@ int zebra_special_index_fetch(ZebraHandle zh, SYSNO sysno, ODR odr, } -int zebra_special_fetch(ZebraHandle zh, SYSNO sysno, ODR odr, +static void retrieve_puts_attr(WRBUF wrbuf, const char *name, + const char *value) +{ + if (value) + { + wrbuf_printf(wrbuf, " %s=\"", name); + wrbuf_xmlputs(wrbuf, value); + wrbuf_printf(wrbuf, "\""); + } +} + +static void retrieve_puts_attr_int(WRBUF wrbuf, const char *name, + const int value) +{ + wrbuf_printf(wrbuf, " %s=\"%i\"", name, value); +} + +static void retrieve_puts_str(WRBUF wrbuf, const char *name, + const char *value) +{ + if (value) + wrbuf_printf(wrbuf, "%s %s\n", name, value); +} + +static void retrieve_puts_int(WRBUF wrbuf, const char *name, + const int value) +{ + wrbuf_printf(wrbuf, "%s %i\n", name, value); +} + +int zebra_special_fetch(ZebraHandle zh, zint sysno, int score, ODR odr, const char *elemsetname, oid_value input_format, oid_value *output_format, char **rec_bufp, int *rec_lenp) { Record rec; + + /* set output variables before processing possible error states */ + /* *rec_lenp = 0; */ + - /* only accept XML and SUTRS requests */ - if (input_format != VAL_TEXT_XML - && input_format != VAL_SUTRS){ - yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s", - elemsetname); - return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; - } /* processing zebra::meta::sysno elemset without fetching binary data */ - if (elemsetname && 0 == strcmp(elemsetname, "meta::sysno")) + if (elemsetname && 0 == strcmp(elemsetname, "meta::sysno")) { - char rec_str[128]; - if (input_format == VAL_SUTRS){ - sprintf(rec_str, ZINT_FORMAT, sysno); + int ret = 0; + WRBUF wrbuf = wrbuf_alloc(); + if (input_format == VAL_SUTRS) + { + wrbuf_printf(wrbuf, ZINT_FORMAT, sysno); *output_format = VAL_SUTRS; } - else if (input_format == VAL_TEXT_XML){ - sprintf(rec_str, "\n", - sysno, elemsetname); + else if (input_format == VAL_TEXT_XML) + { + wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR + " sysno=\"" ZINT_FORMAT "\"/>\n", + sysno); *output_format = VAL_TEXT_XML; } - *rec_lenp = strlen(rec_str); - if (*rec_lenp){ - *rec_bufp = odr_strdup(odr, rec_str); - return 0; - } else { - return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; - } + *rec_lenp = wrbuf_len(wrbuf); + if (*rec_lenp) + *rec_bufp = odr_strdup(odr, wrbuf_buf(wrbuf)); + else + ret = YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST; + wrbuf_free(wrbuf, 1); + return ret; } + /* processing special elementsetnames zebra::sort:: */ + if (elemsetname && 0 == strncmp(elemsetname, "sort", 4)) + { + return zebra_special_sort_fetch(zh, sysno, odr, + elemsetname + 4, + input_format, output_format, + rec_bufp, rec_lenp); + } + + /* fetching binary record up for all other display elementsets */ rec = rec_get(zh->reg->records, sysno); - if (!rec){ - yaz_log(YLOG_DEBUG, "rec_get fail on sysno=" ZINT_FORMAT, sysno); + if (!rec) + { + yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno); return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; } - /* processing special elementsetnames zebra::index:: */ - if (elemsetname && 0 == strncmp(elemsetname, "index::", 7)){ - - int ret = zebra_special_index_fetch(zh, sysno, odr, rec, - elemsetname + 7, - input_format, output_format, - rec_bufp, rec_lenp); - - rec_free(&rec); - return ret; - } - /* processing special elementsetnames zebra::data */ - if (elemsetname && 0 == strcmp(elemsetname, "data")){ + if (elemsetname && 0 == strcmp(elemsetname, "data")) + { struct ZebraRecStream stream; RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec); zebra_create_record_stream(zh, &rec, &stream); - *output_format = VAL_SUTRS; + *output_format = input_format; *rec_lenp = recordAttr->recordSize; *rec_bufp = (char *) odr_malloc(odr, *rec_lenp); stream.readf(&stream, *rec_bufp, *rec_lenp); @@ -302,13 +465,90 @@ int zebra_special_fetch(ZebraHandle zh, SYSNO sysno, ODR odr, return 0; } + /* only accept XML and SUTRS requests from now */ + if (input_format != VAL_TEXT_XML && input_format != VAL_SUTRS) + { + yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s", + elemsetname); + return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST; + } + + + /* processing special elementsetnames zebra::meta:: */ + if (elemsetname && 0 == strcmp(elemsetname, "meta")) + { + int ret = 0; + WRBUF wrbuf = wrbuf_alloc(); + RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec); + + if (input_format == VAL_TEXT_XML) + { + *output_format = VAL_TEXT_XML; + + wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR + " sysno=\"" ZINT_FORMAT "\"", sysno); + retrieve_puts_attr(wrbuf, "base", rec->info[recInfo_databaseName]); + retrieve_puts_attr(wrbuf, "file", rec->info[recInfo_filename]); + retrieve_puts_attr(wrbuf, "type", rec->info[recInfo_fileType]); + if (score >= 0) + retrieve_puts_attr_int(wrbuf, "score", score); + + wrbuf_printf(wrbuf, + " rank=\"" ZINT_FORMAT "\"" + " size=\"%i\"" + " set=\"zebra::%s\"/>\n", + recordAttr->staticrank, + recordAttr->recordSize, + elemsetname); + } + else if (input_format == VAL_SUTRS) + { + *output_format = VAL_SUTRS; + wrbuf_printf(wrbuf, "sysno " ZINT_FORMAT "\n", sysno); + retrieve_puts_str(wrbuf, "base", rec->info[recInfo_databaseName]); + retrieve_puts_str(wrbuf, "file", rec->info[recInfo_filename]); + retrieve_puts_str(wrbuf, "type", rec->info[recInfo_fileType]); + if (score >= 0) + retrieve_puts_int(wrbuf, "score", score); + + wrbuf_printf(wrbuf, + "rank " ZINT_FORMAT "\n" + "size %i\n" + "set zebra::%s\n", + recordAttr->staticrank, + recordAttr->recordSize, + elemsetname); + } + *rec_lenp = wrbuf_len(wrbuf); + if (*rec_lenp) + *rec_bufp = odr_strdup(odr, wrbuf_buf(wrbuf)); + else + ret = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; + + wrbuf_free(wrbuf, 1); + rec_free(&rec); + return ret; + } + + /* processing special elementsetnames zebra::index:: */ + if (elemsetname && 0 == strncmp(elemsetname, "index", 5)) + { + int ret = zebra_special_index_fetch(zh, sysno, odr, rec, + elemsetname + 5, + input_format, output_format, + rec_bufp, rec_lenp); + + rec_free(&rec); + return ret; + } + if (rec) rec_free(&rec); - return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; + return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; } -int zebra_record_fetch(ZebraHandle zh, SYSNO sysno, int score, +int zebra_record_fetch(ZebraHandle zh, zint sysno, int score, zebra_snippets *hit_snippet, ODR odr, oid_value input_format, Z_RecordComposition *comp, oid_value *output_format, @@ -327,9 +567,9 @@ int zebra_record_fetch(ZebraHandle zh, SYSNO sysno, int score, *addinfo = 0; elemsetname = yaz_get_esn(comp); - /* processing zebra special elementset names of form 'zebra:: */ - if (elemsetname && 0 == strncmp(elemsetname, "zebra::", 7)) - return zebra_special_fetch(zh, sysno, odr, + /* processing zebra special elementset names of form 'zebra:: */ + if (elemsetname && 0 == strncmp(elemsetname, "zebra::", 7)) + return zebra_special_fetch(zh, sysno, score, odr, elemsetname + 7, input_format, output_format, rec_bufp, rec_lenp); @@ -339,7 +579,7 @@ int zebra_record_fetch(ZebraHandle zh, SYSNO sysno, int score, rec = rec_get(zh->reg->records, sysno); if (!rec) { - yaz_log(YLOG_DEBUG, "rec_get fail on sysno=" ZINT_FORMAT, sysno); + yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno); *basenamep = 0; return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; } @@ -353,9 +593,8 @@ int zebra_record_fetch(ZebraHandle zh, SYSNO sysno, int score, *basenamep = (char *) odr_malloc (odr, strlen(basename)+1); strcpy (*basenamep, basename); - - yaz_log (YLOG_DEBUG, "retrieve localno=" ZINT_FORMAT " score=%d", - sysno, score); + yaz_log(YLOG_DEBUG, "retrieve localno=" ZINT_FORMAT " score=%d", + sysno, score); zebra_create_record_stream(zh, &rec, &stream); @@ -408,8 +647,8 @@ int zebra_record_fetch(ZebraHandle zh, SYSNO sysno, int score, zebra_snippets_log(snippet, YLOG_LOG); #endif - if (!(rt = recType_byName (zh->reg->recTypes, zh->res, - file_type, &clientData))) + if (!(rt = recType_byName(zh->reg->recTypes, zh->res, + file_type, &clientData))) { return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; } @@ -423,6 +662,7 @@ int zebra_record_fetch(ZebraHandle zh, SYSNO sysno, int score, *rec_lenp = retrieveCtrl.rec_len; *addinfo = retrieveCtrl.addinfo; } + zebra_snippets_destroy(snippet); zebra_snippets_destroy(retrieveCtrl.doc_snippet); }