def getSuggestedNames(self): # search common forms lookupNames = [] surname = self.__metadata.getList("surname").get(0) firstName = self.__metadata.getList("firstName").get(0) firstInitial = firstName[0].upper() secondName = self.__metadata.getList("secondName") if not secondName.isEmpty(): secondName = secondName.get(0) if secondName and secondName != "": secondInitial = secondName[0].upper() lookupNames.append("%s, %s. %s." % (surname, firstInitial, secondInitial)) lookupNames.append("%s, %s %s." % (surname, firstName, secondInitial)) lookupNames.append("%s, %s %s" % (surname, firstName, secondName)) lookupNames.append("%s %s %s" % (firstName, secondName, surname)) lookupNames.append("%s, %s." % (surname, firstInitial)) lookupNames.append("%s, %s" % (surname, firstName)) lookupNames.append("%s %s" % (firstName, surname)) query = '" OR dc_title:"'.join(lookupNames) # general word search from each part of the name parts = [p for p in self.getPackageTitle().split(" ") if len(p) > 0] query2 = " OR dc_title:".join(parts) req = SearchRequest('(dc_title:"%s")^2.5 OR (dc_title:%s)^0.5' % (query, query2)) self.log.info("suggestedNames query={}", req.query) req.setParam("fq", 'recordtype:"author"') req.addParam("fq", 'item_type:"object"') req.setParam("rows", "9999") req.setParam("fl", "score") req.setParam("sort", "score desc") # Make sure 'fq' has already been set in the session ##security_roles = self.authentication.get_roles_list(); ##security_query = 'security_filter:("' + '" OR "'.join(security_roles) + '")' ##req.addParam("fq", security_query) out = ByteArrayOutputStream() indexer = self.services.getIndexer() indexer.search(req, out) result = JsonConfigHelper(ByteArrayInputStream(out.toByteArray())) #self.log.info("result={}", result.toString()) docs = result.getJsonList("response/docs") map = LinkedHashMap() for doc in docs: authorName = doc.getList("dc_title").get(0) if map.containsKey(authorName): authorDocs = map.get(authorName) else: authorDocs = ArrayList() map.put(authorName, authorDocs) authorDocs.add(doc) self.__maxScore = max(1.0, float(result.get("response/maxScore"))) return map
class OrganiseData: def __init__(self): self.__portal = Services.portalManager.get(portalId) self.__result = JsonConfigHelper() self.__pageNum = sessionState.get("pageNum", 1) self.__selected = [] self.__storage = Services.storage uri = URLDecoder.decode(request.getAttribute("RequestURI")) basePath = portalId + "/" + pageName self.__oid = uri[len(basePath)+1:] slash = self.__oid.rfind("/") self.__pid = self.__oid[slash+1:] payload = self.__storage.getPayload(self.__oid, self.__pid) if payload is not None: self.__mimeType = payload.contentType else: self.__mimeType = "application/octet-stream" self.__metadata = JsonConfigHelper() print " * combined.py: uri='%s' oid='%s' pid='%s' mimeType='%s'" % (uri, self.__oid, self.__pid, self.__mimeType) self.__search() def getManifestItem(self): hashId = md5.new(self.__oid).hexdigest() return self.getPortal().getMap("manifest//node-%s" % hashId) def getMimeType(self): return self.__mimeType def __search(self): req = SearchRequest('id:"%s"' % self.__oid) out = ByteArrayOutputStream() Services.indexer.search(req, out) self.__json = JsonConfigHelper(ByteArrayInputStream(out.toByteArray())) self.__metadata = SolrDoc(self.__json) def getManifest(self): return self.getPortal().getJsonMap("manifest") def getContent(self): content = "" return content def getPortal(self): return Services.portalManager.get(portalId) def getPortalName(self): return Services.portalManager.get(portalId).description def __search(self): recordsPerPage = self.__portal.recordsPerPage query = None if query is None or query == "": query = formData.get("query") if query is None or query == "": query = "*:*" req = SearchRequest(query) req.setParam("facet", "true") req.setParam("rows", "1000") req.setParam("facet.field", self.__portal.facetFieldList) req.setParam("facet.sort", "true") req.setParam("facet.limit", str(self.__portal.facetCount)) req.setParam("sort", "f_dc_title asc") # setup facets action = formData.get("verb") value = formData.get("value") fq = sessionState.get("fq") if fq is not None: self.__pageNum = 1 req.setParam("fq", fq) if action == "add_fq": self.__pageNum = 1 name = formData.get("name") print " * add_fq: %s" % value req.addParam("fq", URLDecoder.decode(value, "UTF-8")) elif action == "remove_fq": self.__pageNum = 1 req.removeParam("fq", URLDecoder.decode(value, "UTF-8")) elif action == "clear_fq": self.__pageNum = 1 req.removeParam("fq") elif action == "select-page": self.__pageNum = int(value) req.addParam("fq", 'item_type:"object"') portalQuery = self.__portal.query print " * portalQuery=%s" % portalQuery if portalQuery: req.addParam("fq", portalQuery) self.__selected = req.getParams("fq") sessionState.set("fq", self.__selected) sessionState.set("pageNum", self.__pageNum) req.setParam("start", str((self.__pageNum - 1) * recordsPerPage)) print " * combined.py:", req.toString(), self.__pageNum out = ByteArrayOutputStream() Services.indexer.search(req, out) self.__result = JsonConfigHelper(ByteArrayInputStream(out.toByteArray())) if self.__result is not None: self.__paging = Pagination(self.__pageNum, int(self.__result.get("response/numFound")), self.__portal.recordsPerPage) print " * combined.py: updating manifest..." portal = self.getPortal() manifest = portal.getJsonMap("manifest") #add new items from search for doc in self.__result.getList("response/docs"): hashId = md5.new(doc.get("id")).hexdigest() node = portal.get("manifest//node-%s" % hashId) print " ********node=", node if node is None: portal.set("manifest/node-%s/title" % hashId, doc.get("dc_title").get(0)) portal.set("manifest/node-%s/id" % hashId, doc.get("id")) #remove manifest items missing from search result print manifest for key in manifest.keySet(): item = manifest.get(key) id = item.get("id") doc = self.__result.getList('response/docs[@id="%s"]' % id) if len(doc) == 0: portal.removePath("manifest//%s" % key) Services.getPortalManager().save(portal) def getQueryTime(self): return int(self.__result.get("responseHeader/QTime")) / 1000.0; def getPaging(self): return self.__paging def getResult(self): return self.__result def getFacetField(self, key): return self.__portal.facetFields.get(key) def getFacetName(self, key): return self.__portal.facetFields.get(key).get("label") def getFacetCounts(self, key): values = LinkedHashMap() valueList = self.__result.getList("facet_counts/facet_fields/%s" % key) for i in range(0,len(valueList),2): name = valueList[i] count = valueList[i+1] if count > 0: values.put(name, count) return values def hasSelectedFacets(self): return (self.__selected is not None and len(self.__selected) > 1) and \ not (self.__portal.query in self.__selected and len(self.__selected) == 2) def getSelectedFacets(self): return self.__selected def isPortalQueryFacet(self, fq): return fq == self.__portal.query def isSelected(self, fq): return fq in self.__selected def getSelectedFacetIds(self): return [md5.new(fq).hexdigest() for fq in self.__selected] def getFileName(self, path): return os.path.split(path)[1] def getFacetQuery(self, name, value): return '%s:"%s"' % (name, value) def isImage(self, format): return format.startswith("image/") def getThumbnail(self, oid): ext = os.path.splitext(oid)[1] url = oid[oid.rfind("/")+1:-len(ext)] + ".thumb.jpg" if Services.getStorage().getPayload(oid, url): return url return None def getContent(self): contentStr = "<div>" portal = self.getPortal() manifest = portal.getJsonMap("manifest") for key in manifest.keySet(): item = manifest.get(key) if item.get("hidden", "false") == "false": oid = item.get("id") slash = oid.rfind("/") pid = oid[slash+1:] contentStr += "<div class='combined-item' rel='%s'><a name='content-%s'><!-- --></a><h2>%s</h2>" % (oid, key, item.get("title")) contentStr += self.__getPayloadContent(oid, pid) contentStr += "</div>" contentStr += "<div class='clear'></div>" return contentStr + "</div>" def __getPayloadContent(self, oid, pid): print " * combined.py: oid='%s' pid='%s'" % (oid, pid) payload = self.__storage.getPayload(oid, pid) if payload is None: return "<div>Error: No content for '%s'</div>" % oid mimeType = payload.contentType contentStr = "" if mimeType.startswith("text/"): if mimeType == "text/html": contentStr = '<iframe class="iframe-preview" src="%s/download/%s"></iframe>' % \ (portalPath, oid) else: sw = StringWriter() sw.write("<pre>") IOUtils.copy(payload.getInputStream(), sw) sw.write("</pre>") sw.flush() contentStr = sw.toString() elif mimeType == "application/pdf" or mimeType.find("vnd.ms")>-1 or mimeType.find("vnd.oasis.opendocument.")>-1: # get the html version if exist... pid = os.path.splitext(pid)[0] + ".htm" print " * combined.py: pid=%s" % pid payload = self.__storage.getPayload(oid, pid) saxReader = SAXReader(False) try: document = saxReader.read(payload.getInputStream()) slideNode = document.selectSingleNode("//*[local-name()='body']") slideNode.setName("div") out = ByteArrayOutputStream() format = OutputFormat.createPrettyPrint() format.setSuppressDeclaration(True) format.setExpandEmptyElements(True) writer = XMLWriter(out, format) writer.write(slideNode) writer.close() contentStr = out.toString("UTF-8") except: traceback.print_exc() contentStr = "<p class=\"error\">No preview available</p>" elif mimeType.startswith("image/"): src = "%s/%s" % (oid, pid) contentStr = '<a class="image" href="%(src)s" style="max-width:98%%">' \ '<img src="%(src)s" style="max-width:100%%" /></a>' % { "src": pid } return contentStr def getOid(self): return self.__oid
def getSuggestedNames(self): # search common forms lookupNames = [] surname = self.__metadata.getList("surname").get(0) firstName = self.__metadata.getList("firstName").get(0) firstInitial = firstName[0].upper() secondName = self.__metadata.getList("secondName") if not secondName.isEmpty(): secondName = secondName.get(0) if secondName and secondName != "": secondInitial = secondName[0].upper() lookupNames.append("%s, %s. %s." % (surname, firstInitial, secondInitial)) lookupNames.append("%s, %s %s." % (surname, firstName, secondInitial)) lookupNames.append("%s, %s %s" % (surname, firstName, secondName)) lookupNames.append("%s %s %s" % (firstName, secondName, surname)) lookupNames.append("%s, %s." % (surname, firstInitial)) lookupNames.append("%s, %s" % (surname, firstName)) lookupNames.append("%s %s" % (firstName, surname)) query = '" OR dc_title:"'.join(lookupNames) # general word search from each part of the name parts = [p for p in self.getPackageTitle().split(" ") if len(p) > 0] query2 = " OR dc_title:".join(parts) req = SearchRequest('(dc_title:"%s")^2.5 OR (dc_title:%s)^0.5' % (query, query2)) self.log.info("suggestedNames query={}", req.query) req.setParam("fq", 'recordtype:"author"') req.addParam("fq", 'item_type:"object"') req.setParam("rows", "9999") req.setParam("fl", "score") req.setParam("sort", "score desc") # Make sure 'fq' has already been set in the session ##security_roles = self.authentication.get_roles_list(); ##security_query = 'security_filter:("' + '" OR "'.join(security_roles) + '")' ##req.addParam("fq", security_query) out = ByteArrayOutputStream() indexer = self.services.getIndexer() indexer.search(req, out) result = JsonConfigHelper(ByteArrayInputStream(out.toByteArray())) #self.log.info("result={}", result.toString()) docs = result.getJsonList("response/docs") map = LinkedHashMap() for doc in docs: authorName = doc.getList("dc_title").get(0) if map.containsKey(authorName): authorDocs = map.get(authorName) else: authorDocs = ArrayList() map.put(authorName, authorDocs) authorDocs.add(doc) self.__maxScore = max(1.0, float(result.get("response/maxScore"))) return map