/// <summary> Optimize the solr/lucene core used for searching within a single document </summary>
/// <param name="SolrPageUrl"> URL for the solr/lucene core used for searching within a single document for matching pages </param>
public static void Optimize_Page_Index(string SolrPageUrl)
{
// Create the solr worker
var solrPageWorker = Solr_Operations_Cache <SolrPage> .GetSolrOperations(SolrPageUrl);
try
{
solrPageWorker.Optimize();
}
catch (Exception)
{
// Do not do anything here. It may throw an exception when it runs very longs
}
}
/// <summary> Deletes an existing resource from both solr/lucene core indexes </summary>
/// <param name="SolrDocumentUrl"> URL for the solr/lucene core used for searching for a single document within the library </param>
/// <param name="SolrPageUrl"> URL for the solr/lucene core used for searching within a single document for matching pages </param>
/// <param name="BibID"> Bibliographic identifier for the item to remove from the solr/lucene indexes </param>
/// <param name="VID"> Volume identifier for the item to remove from the solr/lucene indexes </param>
/// <returns> TRUE if successful, otherwise FALSE </returns>
public static bool Delete_Resource_From_Index(string SolrDocumentUrl, string SolrPageUrl, string BibID, string VID)
{
try
{
// Create the solr workers
var solrDocumentWorker = Solr_Operations_Cache <SolrDocument> .GetSolrOperations(SolrDocumentUrl);
var solrPageWorker = Solr_Operations_Cache <SolrPage> .GetSolrOperations(SolrPageUrl);
// For the object, we can use the unique identifier
solrDocumentWorker.Delete(BibID + ":" + VID);
// For the pages, we need to search by id
solrPageWorker.Delete(new SolrQuery("did:\"" + BibID + ":" + VID + "\""));
// Comit the changes to the solr/lucene index
try
{
solrDocumentWorker.Commit();
}
catch
{
Thread.Sleep(10 * 60 * 1000);
}
try
{
solrPageWorker.Commit();
}
catch
{
Thread.Sleep(10 * 60 * 1000);
}
return(true);
}
catch (Exception)
{
return(false);
}
}
/// <summary> Perform an search for documents with matching parameters </summary>
/// <param name="AggregationCode"> Aggregation code within which to search </param>
/// <param name="QueryString"> Quert string for the actual search to perform aggainst the Solr/Lucene engine </param>
/// <param name="ResultsPerPage"> Number of results to display per a "page" of results </param>
/// <param name="Page_Number"> Which page of results to return ( one-based, so the first page is page number of one )</param>
/// <param name="Sort"> Sort to apply before returning the results of the search </param>
/// <param name="Tracer"> Trace object keeps a list of each method executed and important milestones in rendering </param>
/// <param name="Complete_Result_Set_Info"> [OUT] Information about the entire set of results </param>
/// <param name="Paged_Results"> [OUT] List of search results for the requested page of results </param>
/// <returns> Page search result object with all relevant result information </returns>
public static bool Search(string AggregationCode, string QueryString, int ResultsPerPage, int Page_Number, ushort Sort, Custom_Tracer Tracer, out Search_Results_Statistics Complete_Result_Set_Info, out List <iSearch_Title_Result> Paged_Results)
{
if (Tracer != null)
{
Tracer.Add_Trace("Solr_Documents_Searcher.Search", String.Empty);
}
// Set output initially to null
Paged_Results = new List <iSearch_Title_Result>();
Complete_Result_Set_Info = null;
try
{
// Ensure page is not erroneously set to zero or negative
if (Page_Number <= 0)
{
Page_Number = 1;
}
// Create the solr worker to query the document index
var solrWorker = Solr_Operations_Cache <Solr_Document_Result> .GetSolrOperations(Engine_ApplicationCache_Gateway.Settings.Servers.Document_Solr_Index_URL);
// Create the query options
QueryOptions options = new QueryOptions
{
Rows = ResultsPerPage,
Start = (Page_Number - 1) * ResultsPerPage,
Fields = new[] { "did", "score", "url", "aleph", "donor", "edition", "format", "holdinglocation", "sourceinstitution", "maintitle", "materialtype", "oclc", "pubdate_display", "author_display", "publisher_display", "mainthumbnail" },
Highlight = new HighlightingParameters {
Fields = new[] { "fulltext" },
},
ExtraParams = new Dictionary <string, string> {
{ "hl.useFastVectorHighlighter", "true" }
}
};
// Set the sort value
if (Sort != 0)
{
options.OrderBy.Clear();
switch (Sort)
{
case 1:
options.OrderBy.Add(new SortOrder("maintitle_sort"));
break;
case 2:
options.OrderBy.Add(new SortOrder("bibid", Order.ASC));
break;
case 3:
options.OrderBy.Add(new SortOrder("bibid", Order.DESC));
break;
case 10:
options.OrderBy.Add(new SortOrder("pubdate", Order.ASC));
break;
case 11:
options.OrderBy.Add(new SortOrder("pubdate", Order.DESC));
break;
}
}
// If there was an aggregation code included, put that at the beginning of the search
if ((AggregationCode.Length > 0) && (AggregationCode.ToUpper() != "ALL"))
{
QueryString = "(aggregation_code:" + AggregationCode.ToUpper() + ")AND(" + QueryString + ")";
}
// Perform this search
SolrQueryResults <Solr_Document_Result> results = solrWorker.Query(QueryString, options);
// Create the search statistcs
List <string> metadataLabels = new List <string> {
"Author", "Publisher", "Format", "Edition", "Institution", "Donor"
};
Complete_Result_Set_Info = new Search_Results_Statistics(metadataLabels)
{
Total_Titles = results.NumFound,
Total_Items = results.NumFound,
QueryTime = results.Header.QTime
};
// Pass all the results into the List and add the highlighted text to each result as well
foreach (Solr_Document_Result thisResult in results)
{
// Add the highlight snipper
if ((results.Highlights.ContainsKey(thisResult.DID)) && (results.Highlights[thisResult.DID].Count > 0) && (results.Highlights[thisResult.DID].ElementAt(0).Value.Count > 0))
{
thisResult.Snippet = results.Highlights[thisResult.DID].ElementAt(0).Value.ElementAt(0);
}
// Add this results
Paged_Results.Add(thisResult);
}
return(true);
}
catch
{
return(false);
}
}
/// <summary> Indexes a single digital resource within a SobekCM library </summary>
/// <param name="SolrDocumentUrl"> URL for the solr/lucene core used for searching for a single document within the library </param>
/// <param name="SolrPageUrl"> URL for the solr/lucene core used for searching within a single document for matching pages </param>
/// <param name="Resource"> Digital resource to index</param>
/// <param name="Include_Text"> Flag indicates whether to look for and include full text </param>
public static void Update_Index(string SolrDocumentUrl, string SolrPageUrl, SobekCM_Item Resource, bool Include_Text)
{
// Create the solr workers
var solrDocumentWorker = Solr_Operations_Cache <SolrDocument> .GetSolrOperations(SolrDocumentUrl);
var solrPageWorker = Solr_Operations_Cache <SolrPage> .GetSolrOperations(SolrPageUrl);
// Get the list of all items in this collection
List <SolrDocument> index_files = new List <SolrDocument>();
List <SolrPage> index_pages = new List <SolrPage>();
// Add this document to the list of documents to index
index_files.Add(new SolrDocument(Resource, Resource.Source_Directory));
bool document_success = false;
int document_attempts = 0;
while (!document_success)
{
try
{
solrDocumentWorker.Add(index_files);
document_success = true;
}
catch (Exception)
{
if (document_attempts > 5)
{
throw;
}
document_attempts++;
Console.WriteLine(@"ERROR {0}", document_attempts);
Thread.Sleep(document_attempts * 1000);
}
}
// Add each page to be indexed
foreach (SolrDocument document in index_files)
{
index_pages.AddRange(document.Solr_Pages);
}
bool page_success = false;
int page_attempts = 0;
while (!page_success)
{
try
{
solrPageWorker.Add(index_pages);
page_success = true;
}
catch (Exception)
{
if (page_attempts > 5)
{
throw;
}
page_attempts++;
Thread.Sleep(page_attempts * 1000);
}
}
// Comit the changes to the solr/lucene index
try
{
solrDocumentWorker.Commit();
}
catch
{
Thread.Sleep(10 * 60 * 1000);
}
try
{
solrPageWorker.Commit();
}
catch
{
Thread.Sleep(10 * 60 * 1000);
}
}
/// <summary> Perform an in-document search for pages with matching full-text </summary>
/// <param name="BibID"> Bibliographic identifier (BibID) for the item to search </param>
/// <param name="VID"> Volume identifier for the item to search </param>
/// <param name="Search_Terms"> Terms to search for within the page text </param>
/// <param name="ResultsPerPage"> Number of results to display per a "page" of results </param>
/// <param name="ResultsPage"> Which page of results to return ( one-based, so the first page is page number of one )</param>
/// <param name="Sort_By_Score"> Flag indicates whether to sort the results by relevancy score, rather than the default page order </param>
/// <returns> Page search result object with all relevant result information </returns>
public static Solr_Page_Results Search(string BibID, string VID, List <string> Search_Terms, int ResultsPerPage, int ResultsPage, bool Sort_By_Score)
{
// Ensure page is not erroneously set to zero or negative
if (ResultsPage <= 0)
{
ResultsPage = 1;
}
// Create the solr worker to query the page index
var solrWorker = Solr_Operations_Cache <Solr_Page_Result> .GetSolrOperations(Engine_ApplicationCache_Gateway.Settings.Servers.Page_Solr_Index_URL);
// Create the query options
QueryOptions options = new QueryOptions
{
Rows = ResultsPerPage,
Start = (ResultsPage - 1) * ResultsPerPage,
Fields = new [] { "pageid", "pagename", "pageorder", "score", "thumbnail" },
Highlight = new HighlightingParameters {
Fields = new[] { "pagetext" },
},
ExtraParams = new Dictionary <string, string> {
{ "hl.useFastVectorHighlighter", "true" }
}
};
// If this is not the default Solr sort (by score) request sort by the page order
if (!Sort_By_Score)
{
options.OrderBy = new[] { new SortOrder("pageorder", Order.ASC) }
}
;
// Build the query string
StringBuilder queryStringBuilder = new StringBuilder("(bibid:" + BibID + ")AND(vid:" + VID + ")AND(");
bool first_value = true;
foreach (string searchTerm in Search_Terms)
{
if (searchTerm.Length > 1)
{
// Skip any AND NOT for now
if (searchTerm[0] != '-')
{
// Find the joiner
if (first_value)
{
if (searchTerm.IndexOf(" ") > 0)
{
if ((searchTerm[0] == '+') || (searchTerm[0] == '=') || (searchTerm[0] == '-'))
{
queryStringBuilder.Append("(pagetext:\"" + searchTerm.Substring(1).Replace(":", "") + "\")");
}
else
{
queryStringBuilder.Append("(pagetext:\"" + searchTerm.Replace(":", "") + "\")");
}
}
else
{
if ((searchTerm[0] == '+') || (searchTerm[0] == '=') || (searchTerm[0] == '-'))
{
queryStringBuilder.Append("(pagetext:" + searchTerm.Substring(1).Replace(":", "") + ")");
}
else
{
queryStringBuilder.Append("(pagetext:" + searchTerm.Replace(":", "") + ")");
}
}
first_value = false;
}
else
{
if ((searchTerm[0] == '+') || (searchTerm[0] == '=') || (searchTerm[0] == '-'))
{
queryStringBuilder.Append(searchTerm[0] == '=' ? " OR " : " AND ");
if (searchTerm.IndexOf(" ") > 0)
{
queryStringBuilder.Append("(pagetext:\"" + searchTerm.Substring(1).Replace(":", "") + "\")");
}
else
{
queryStringBuilder.Append("(pagetext:" + searchTerm.Substring(1).Replace(":", "") + ")");
}
}
else
{
if (searchTerm.IndexOf(" ") > 0)
{
queryStringBuilder.Append(" AND (pagetext:\"" + searchTerm.Replace(":", "") + "\")");
}
else
{
queryStringBuilder.Append(" AND (pagetext:" + searchTerm.Replace(":", "") + ")");
}
}
}
}
}
}
queryStringBuilder.Append(")");
// Perform this search
SolrQueryResults <Solr_Page_Result> results = solrWorker.Query(queryStringBuilder.ToString(), options);
// Create the results object to pass back out
var searchResults = new Solr_Page_Results
{
QueryTime = results.Header.QTime,
TotalResults = results.NumFound,
Query = queryStringBuilder.ToString(),
Sort_By_Score = Sort_By_Score,
Page_Number = ResultsPage
};
// Pass all the results into the List and add the highlighted text to each result as well
foreach (Solr_Page_Result thisResult in results)
{
// Add the highlight snipper
if ((results.Highlights.ContainsKey(thisResult.PageID)) && (results.Highlights[thisResult.PageID].Count > 0) && (results.Highlights[thisResult.PageID].ElementAt(0).Value.Count > 0))
{
thisResult.Snippet = results.Highlights[thisResult.PageID].ElementAt(0).Value.ElementAt(0);
}
// Add this results
searchResults.Add_Result(thisResult);
}
return(searchResults);
}
}
