public SearchResult ToEySearchResult()
{
return(new SearchResult
{
DocId = docId,
FileName = docFileName,
FileUrl = docUrl + AzureStorageHelper.GenerateSasUrl(docFileName),
Categories = Categories,
LastUpdate = InsertDate,
Score = Score,
Highlights = Highlights.content.ToArray()
});
}
public static void DeleteDocById(string id)
{
// Get the doc first
var indexClient = GetSearchIndexClient();
SearchDocument document = indexClient.Documents.Get <SearchDocument>(id);
if (document != null)
{
// Remove file from azure
AzureStorageHelper.DeleteBlob(document.DocFileName);
// Now remove from index
var batch = IndexBatch.Delete(new[] { document });
PerformIndexOperation(batch);
}
}
public static SearchResult[] DoSearch(SearchRequest request)
{
var indexClient = GetSearchIndexClient();
// Set up search parameters
var sp = new SearchParameters();
// Set up highlights
var hlFields = new List <string> {
"content"
};
sp.HighlightFields = hlFields;
sp.HighlightPreTag = "<b>";
sp.HighlightPostTag = "</b>";
// Set mode to 'all' (must match all terms/clauses). Default is 'any'
sp.SearchMode = SearchMode.All;
// Use the lucerne query engine
sp.QueryType = QueryType.Full;
// Filter by docid if requested
if (!string.IsNullOrEmpty(request.DocId))
{
sp.Filter = "docId eq '" + request.DocId + "'";
}
// Filter by clientid if requested
if (!string.IsNullOrEmpty(request.ClientId))
{
// add the and clause if needed
if (!string.IsNullOrEmpty(sp.Filter))
{
sp.Filter += " and ";
}
sp.Filter = "clientId eq '" + request.ClientId + "'";
}
// Filter by category(s) if requested
if (request.Categories != null)
{
for (int i = 0; i < request.Categories.Length; i++)
{
// start the filter expression
if (i == 0)
{
// add the and clause if needed
if (!string.IsNullOrEmpty(sp.Filter))
{
sp.Filter += " and ";
}
sp.Filter += "(";
}
// Add category filter
sp.Filter += "category eq '" + request.Categories[i];
// end expression or add 'and'
if (i == request.Categories.Length - 1)
{
// end expression
sp.Filter += ")";
}
else
{
sp.Filter += " and ";
}
}
}
// Do not return the content
sp.Select = new List <string> {
"docId", "docFileName", "docUrl", "categories", "insertDate"
};
// Perform the search
DocumentSearchResult <Document> response = null;
try
{
response = indexClient.Documents.Search <Document>(request.SearchTerm, sp);
}
catch (Exception ex)
{
Debug.WriteLine("Error doing search. " + ex.Message);
}
var ret = new List <SearchResult>();
if (response != null)
{
foreach (SearchResult <Document> result in response.Results)
{
var sResult = new SearchResult
{
Score = result.Score,
DocId = (string)result.Document["docId"],
FileName = (string)result.Document["docFileName"],
FileUrl = (string)result.Document["docUrl"],
Categories = (IEnumerable <string>)result.Document["categories"],
LastUpdate = ((DateTimeOffset)result.Document["insertDate"]).DateTime
};
if (result.Highlights != null && result.Highlights.ContainsKey("content"))
{
sResult.Highlights = result.Highlights["content"].ToArray();
}
sResult.FileUrl += AzureStorageHelper.GenerateSasUrl(sResult.FileName);
ret.Add(sResult);
}
}
return(ret.ToArray());
}
public static SearchResult[] CategorizeDocument(string docId)
{
string catScoreThreshold = ConfigurationManager.AppSettings["CategorizationScoreThreshold"];
double catTheshold;
if (!double.TryParse(catScoreThreshold, out catTheshold))
{
throw new Exception("Bad or missing Categorization Score Threshold value.");
}
var categoryList = AzureStorageHelper.GetTableRows();
if (categoryList != null && categoryList.Length > 0)
{
// Get the doc
var indexClient = GetSearchIndexClient();
SearchDocument document = indexClient.Documents.Get <SearchDocument>(docId);
if (document != null)
{
// Create list of terms per category
Dictionary <string, List <string> > catTerms = new Dictionary <string, List <string> >();
foreach (var categoryPoco in categoryList)
{
if (catTerms.ContainsKey(categoryPoco.CategoryName))
{
catTerms[categoryPoco.CategoryName].Add(categoryPoco.SearchTerm);
}
else
{
catTerms.Add(categoryPoco.CategoryName, new List <string> {
categoryPoco.SearchTerm
});
}
}
List <SearchResult> resultList = new List <SearchResult>();
List <string> documentCategoryList = new List <string>();
// Do the search for each category
foreach (var catTerm in catTerms)
{
// create each phrase and separate by an OR clause
SearchRequest request = new SearchRequest
{
DocId = docId,
SearchTerm = "(" + string.Join(")|(", catTerm.Value) + ")"
};
// Do the search
var results = DoSearch(request);
if (results != null && results.Length > 0)
{
// Found a match, but we should only get one result b/c we filtered on the unique docId
if (results.Length > 1)
{
throw new Exception("More than one result found for document id: " + docId);
}
var firstResult = results[0];
// add the category to the item if the score is high enough
bool addedToCategory = false;
if (firstResult.Score >= catTheshold)
{
documentCategoryList.Add(catTerm.Key);
addedToCategory = true;
}
// Put the category and it's search term in the categories field (a bit of a hack)
firstResult.Categories = new[] { catTerm.Key, request.SearchTerm, addedToCategory.ToString().ToLower() };
resultList.Add(firstResult);
}
}
// Set category for item
document.Categories = documentCategoryList.ToArray();
// Resave item
UpdateDocument(document);
// return results
return(resultList.OrderByDescending(r => r.Score).ToArray());
}
}
return(null);
}
