X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fmod_dom.c;h=f16fba7a576c07a0c0025811b9abeec4c0d0524a;hb=7598c76f1a4989a91003bd4fbd90f30a7c7255ef;hp=0dcf999820d9bc686a8445311a5fe3e9778101e8;hpb=50d5984f510e0434dc6811b46c822fc2301c4450;p=idzebra-moved-to-github.git diff --git a/index/mod_dom.c b/index/mod_dom.c index 0dcf999..f16fba7 100644 --- a/index/mod_dom.c +++ b/index/mod_dom.c @@ -338,7 +338,7 @@ static ZEBRA_RES parse_convert(struct filter_info *tinfo, xmlNodePtr ptr, *l = p; l = &p->next; } - else if (!XML_STRCMP(ptr->name, "meta")) + else if (!XML_STRCMP(ptr->name, "process-meta")) { struct _xmlAttr *attr; struct convert_s *p = nmem_malloc(tinfo->nmem_config, sizeof(*p)); @@ -376,13 +376,12 @@ static int process_meta(struct filter_info *tinfo, xmlDocPtr doc, xmlNodePtr nod struct _xmlAttr *attr; for (attr = node->properties; attr; attr = attr->next) { - if (attr_content(attr, "element_set_name", &element_set_name)) + if (attr_content(attr, "name", &element_set_name)) ; else { dom_log(YLOG_WARN, tinfo, node, - "bad attribute @%s, expected @element_set_name", - attr->name); + "bad attribute @%s, expected @name", attr->name); } } if (element_set_name) @@ -400,11 +399,11 @@ static int process_meta(struct filter_info *tinfo, xmlDocPtr doc, xmlNodePtr nod if (ret == 0) { xmlDocPtr sub_doc = - xmlParseMemory( wrbuf_buf(result), wrbuf_len(result)); + xmlParseMemory(wrbuf_buf(result), wrbuf_len(result)); if (sub_doc) { xmlNodePtr t = xmlDocGetRootElement(sub_doc); - xmlAddChild(node, xmlCopyNode(t, 1)); + xmlReplaceNode(node, xmlCopyNode(t, 1)); xmlFreeDoc(sub_doc); } } @@ -814,10 +813,18 @@ static void index_value_of(struct filter_info *tinfo, { xmlChar *text = xmlNodeGetContent(node); size_t text_len = strlen((const char *)text); - + /* if there is no text, we do not need to proceed */ if (text_len) { + /* keep seqno base so that all text will have + identical seqno's for multiple fields , e.g + .. */ + + zint seqno_base = recword->seqno; + zint seqno_max = recword->seqno; + + const char *look = index_p; const char *bval; const char *eval; @@ -865,6 +872,7 @@ static void index_value_of(struct filter_info *tinfo, /* actually indexing the text given */ + recword->seqno = seqno_base; recword->index_name = (const char *)index; if (*type) recword->index_type = (const char *) type; @@ -879,12 +887,16 @@ static void index_value_of(struct filter_info *tinfo, (extctr->tokenAdd)(recword); + if (seqno_max < recword->seqno) + seqno_max = recword->seqno; + /* eat whitespaces */ if (*look && ' ' == *look) { look++; } } + recword->seqno = seqno_max; } xmlFree(text); } @@ -1151,7 +1163,6 @@ static int convert_extract_doc(struct filter_info *tinfo, int len_out; const char *params[10]; xsltStylesheetPtr last_xsp = 0; - xmlDocPtr store_doc = 0; /* per default do not ingest record */ tinfo->record_info_invoked = 0; @@ -1168,41 +1179,44 @@ static int convert_extract_doc(struct filter_info *tinfo, { xmlChar *buf_out; int len_out; + xmlDocDumpMemory(doc, &buf_out, &len_out); #if 0 FILE *outf = fopen("extract.xml", "w"); - xmlDocDumpMemory(doc, &buf_out, &len_out); fwrite(buf_out, 1, len_out, outf); -#endif - yaz_log(YLOG_LOG, "Extract Doc: %.*s", len_out, buf_out); -#if 0 fclose(outf); #endif + yaz_log(YLOG_LOG, "Extract Doc: %.*s", len_out, buf_out); } - /* input conversion */ - perform_convert(tinfo, p, 0, input->convert, params, &doc, 0); - - if (tinfo->store) + if (p->setStoreData) { - /* store conversion */ - store_doc = xmlCopyDoc(doc, 1); - perform_convert(tinfo, p, 0, tinfo->store->convert, - params, &store_doc, &last_xsp); - } - - /* saving either store doc or original doc in case no store doc exists */ - if (last_xsp) - xsltSaveResultToString(&buf_out, &len_out, - store_doc ? store_doc : doc, last_xsp); - else - xmlDocDumpMemory(store_doc ? store_doc : doc, &buf_out, &len_out); + xmlDocPtr store_doc = 0; - if (p->setStoreData) - (*p->setStoreData)(p, buf_out, len_out); - xmlFree(buf_out); + /* input conversion */ + perform_convert(tinfo, p, 0, input->convert, params, &doc, 0); + + if (tinfo->store) + { + /* store conversion */ + store_doc = xmlCopyDoc(doc, 1); + perform_convert(tinfo, p, 0, tinfo->store->convert, + params, &store_doc, &last_xsp); + } + + /* saving either store doc or original doc in case no store doc exists */ + if (last_xsp) + xsltSaveResultToString(&buf_out, &len_out, + store_doc ? store_doc : doc, last_xsp); + else + xmlDocDumpMemory(store_doc ? store_doc : doc, &buf_out, &len_out); + + if (p->setStoreData) + (*p->setStoreData)(p, buf_out, len_out); + xmlFree(buf_out); + if (store_doc) + xmlFreeDoc(store_doc); + } - if (store_doc) - xmlFreeDoc(store_doc); /* extract conversion */ perform_convert(tinfo, p, 0, tinfo->extract->convert, params, &doc, 0); @@ -1258,7 +1272,7 @@ static int extract_xml_split(struct filter_info *tinfo, ptr = xmlTextReaderExpand(input->u.xmlreader.reader); if (ptr) - { + { /* we have a new document */ xmlNodePtr ptr2 = xmlCopyNode(ptr, 1);