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);
}
