//--- Constructors ---
public SearchResultItem(uint typeId, SearchResultType type, string title, double rank, DateTime modified) {
TypeId = typeId;
Type = type;
Title = title;
Modified = modified;
Detail = null;
_rank = rank;
}
public void Can_roundtrip_SearchResultDetail_with_null_string()
{
var data = new SearchResultDetail();
data["null"] = null;
using (var ms = new MemoryStream()) {
_serializer.Serialize(ms, data);
ms.Position = 0;
var data2 = _serializer.Deserialize <SearchResultDetail>(ms);
Assert.IsNull(data2["null"]);
}
}
public void Can_roundtrip_SearchResultDetail_with_empty_string2()
{
var data = new SearchResultDetail();
data["empty"] = "";
using (var ms = new MemoryStream()) {
_serializer.Serialize(ms, data);
ms.Position = 0;
var data2 = _serializer.Deserialize <SearchResultDetail>(ms);
Assert.AreEqual("", data2["empty"]);
}
}
public void Can_roundtrip_SearchResultDetail() {
var data = new SearchResultDetail();
data["foo"] = "bar";
data["bzz"] = "bonk";
using(var ms = new MemoryStream()) {
_serializer.Serialize(ms, data);
ms.Position = 0;
var data2 = _serializer.Deserialize<SearchResultDetail>(ms);
Assert.AreEqual(data["foo"], data2["foo"]);
Assert.AreEqual(data["bzz"], data2["bzz"]);
}
}
public void Can_roundtrip_SearchResultDetail()
{
var data = new SearchResultDetail();
data["foo"] = "bar";
data["bzz"] = "bonk";
using (var ms = new MemoryStream()) {
_serializer.Serialize(ms, data);
ms.Position = 0;
var data2 = _serializer.Deserialize <SearchResultDetail>(ms);
Assert.AreEqual(data["foo"], data2["foo"]);
Assert.AreEqual(data["bzz"], data2["bzz"]);
}
}
public void FormatResultSet_generates_special_output_for_tracked_searches() {
// Arrange
var item = new SearchResultItem(1, SearchResultType.Page, "page", 1, DateTime.Parse("2010/10/10").ToSafeUniversalTime());
var detail = new SearchResultDetail(Pairs(
"id.page", 1,
"path","path",
"uri","http://uri",
"title", "foo",
"author", "bob",
"preview", "preview"
));
var set = new SearchResult("parsed", new[] { item });
var discriminator = new SetDiscriminator() { Ascending = true, Limit = 100, Offset = 0, SortField = "rank" };
var explain = false;
TrackingInfo trackingInfo = new TrackingInfo() { QueryId = 123, PreviousQueryId = 456 };
_cache.Setup(x => x.TryGet(item.DetailKey, out detail)).Returns(true).AtMostOnce().Verifiable();
// Act
var xml = Search.FormatResultSet(set, discriminator, explain, trackingInfo, new Result<XDoc>()).Wait();
// Assert
_cache.VerifyAll();
var expectedXml = XDocFactory.From(@"
<search querycount=""1"" ranking=""adaptive"" queryid=""123"" count=""1"">
<parsedQuery>parsed</parsedQuery>
<result>
<id>1</id>
<uri>http://uri</uri>
<uri.track>mock://api/site/query/123?pageid=1&rank=1&type=page&position=1</uri.track>
<rank>1</rank>
<title>foo</title>
<page>
<title>foo</title>
<path>path</path>
</page>
<author>bob</author>
<date.modified>2010-10-10T00:00:00Z</date.modified>
<content>preview</content>
<type>page</type>
</result>
</search>", MimeType.TEXT_XML);
Assert.AreEqual(expectedXml.ToCompactString(), xml.ToCompactString());
}
public void Can_roundtrip_SearchResultDetail_with_null_string() {
var data = new SearchResultDetail();
data["null"] = null;
using(var ms = new MemoryStream()) {
_serializer.Serialize(ms, data);
ms.Position = 0;
var data2 = _serializer.Deserialize<SearchResultDetail>(ms);
Assert.IsNull(data2["null"]);
}
}
public void Can_roundtrip_SearchResultDetail_with_empty_string2() {
var data = new SearchResultDetail();
data["empty"] = "";
using(var ms = new MemoryStream()) {
_serializer.Serialize(ms, data);
ms.Position = 0;
var data2 = _serializer.Deserialize<SearchResultDetail>(ms);
Assert.AreEqual("", data2["empty"]);
}
}
/// <summary>
///
/// </summary>
/// <param name="userSessionId"></param>
/// <param name="searchText"></param>
/// <returns></returns>
/// <remarks>
/// I should update parameter for GetSessionDetailByUserSessionId to be a string and decouple dependency from MongoDb ObjectId Model
/// I should also create my own "OcrResult" to decouple dependency from Microsoft Azure Models
/// </remarks>
public async Task <SearchResultsDto> SearchUserSession(string userSessionId, string searchText)
{
var result = new SearchResultsDto();
var sessionDetailsFound = await _sessionDetailRepo.GetSessionDetailByUserSessionId(ObjectId.Parse(userSessionId)).ConfigureAwait(false);
//var rawText = String.Empty;
try
{
var detail = sessionDetailsFound.PrintedTextResult;
var regions = detail.Regions;
foreach (OcrRegion region in regions)
{
var lines = region.Lines;
foreach (OcrLine line in lines)
{
var words = line.Words;
foreach (OcrWord word in words)
{
if (word.Text.ToLower().Contains(searchText.ToLower())) //Searh for match
{
//find how many regions to exclude
var regionFoundIndex = regions.IndexOf(region);
var skipLastCount = regions.Count - regionFoundIndex;
if (skipLastCount < 0)
{
skipLastCount = 0;
}
//find haw many lines of text there are to find the word
var reg = regions.SkipLast(skipLastCount);
var linesInPriorRegions = 0;
if (reg.Any())
{
linesInPriorRegions = reg.Sum(x => x.Lines.Count());
}
var lineNumber = linesInPriorRegions + region.Lines.IndexOf(line) + 1;
//find how many words to exclude
var wordsFoundIndex = words.IndexOf(word);
var skipLastWordCount = words.Count - wordsFoundIndex;
if (skipLastWordCount < 0)
{
skipLastWordCount = 0;
}
//find how many characters to find the word in the line
var filteredWords = words.SkipLast(skipLastWordCount);
var charInPriorWords = 0;
if (filteredWords.Any())
{
charInPriorWords = filteredWords.Sum(w => w.Text.Length);
}
var intPosition = charInPriorWords + filteredWords.Count(); //count number of words and spaces. will not include special characters.
//Map from Ocr classes to custom classes (to reduce dependency on Ocr classes
var resultText = new SearchResultText()
{
BoundingBox = word.BoundingBox,
Text = word.Text
};
var resultLine = new SearchResultLine()
{
BoundingBox = line.BoundingBox,
Text = words.Select(w => new SearchResultText()
{
BoundingBox = w.BoundingBox, Text = w.Text
}).ToList()
};
var resultRegion = new SearchResultRegion()
{
BoundingBox = region.BoundingBox,
Lines = new List <SearchResultLine>()
{
resultLine
}
};
var resultDetail = new SearchResultDetail()
{
Language = detail.Language,
TextAngle = detail.TextAngle,
Orientation = detail.Orientation,
Regions = new List <SearchResultRegion>()
{
resultRegion
}
};
//build result details
var resultDetails = new SearchResultDetails(resultText, resultLine, resultRegion, resultDetail, lineNumber, intPosition);
//add to results
result.ResultDetails.Add(resultDetails);
}
}
}
}
}
finally
{
if (sessionDetailsFound != null)
{
sessionDetailsFound = null;
}
}
return(result);
}
/// <summary>
/// Retrieve the search results from the database
/// </summary>
/// <param name="searchterm"></param>
/// <param name="searchclass"></param>
/// <param name="searchlimit"></param>
/// <returns></returns>
private List<SearchResult> getSearchResults(string searchterm, string searchclass, int searchlimit)
{
// Prepare a list of results to return
List<SearchResult> results = new List<SearchResult>();
// Put catcher in for a null search term
if (searchterm == null) return results;
// Get the connection info
string connectionInfo = ConfigurationManager.ConnectionStrings["ViewerSettings"].ConnectionString;
// Create the connection
SqlConnection conn = new SqlConnection(connectionInfo);
// Generate the command object
string sqlstring = "pLocationSearch_NZTM";
SqlCommand cmd = new SqlCommand(sqlstring, conn);
cmd.CommandType = CommandType.StoredProcedure;
cmd.Parameters.Add("@searchlimit", SqlDbType.Int).Value = searchlimit;
cmd.Parameters.Add("@searchterm", SqlDbType.VarChar).Value = searchterm;
cmd.Parameters.Add("@searchclass", SqlDbType.VarChar).Value = searchclass;
// Get the data
try
{
// Open the connection to the database
conn.Open();
// Execute the reader
SqlDataReader myReader = cmd.ExecuteReader();
// Prepare the results
while (myReader.Read())
{
// Create a details object
SearchResultDetail detail = new SearchResultDetail()
{
searchkey = (string)myReader.GetString(0),
searchclass = (string)myReader.GetString(1),
keydescription = (string)myReader.GetString(5),
x = (double)myReader.GetDouble(6),
y = (double)myReader.GetDouble(7),
outSR = (int)myReader.GetInt32(8)
};
// Create a result
SearchResult myresult = new SearchResult()
{
label = (string)myReader.GetString(4),
value = detail
};
// Add result to the results list
results.Add(myresult);
}
// Sort the results
results = results.OrderBy(o => o.label).ToList();
// Check for no results - add a black result
if (results.Count == 0)
{
// Create a blank result
// Create a details object
SearchResultDetail detail = new SearchResultDetail()
{
searchkey = "No Matching Results",
searchclass = "",
keydescription = "No Matching Results",
x = 0,
y = 0,
outSR = 0
};
SearchResult myresult = new SearchResult()
{
label = "",
value = detail
};
// Add result to the results list
results.Add(myresult);
}
// Cleanup
myReader.Close();
conn.Close();
}
catch (Exception e)
{
throw e;
}
return results;
}
private Yield PopulateDetail(List <SearchResultItem> items, Result result)
{
var pages = new List <SearchResultItem>();
var files = new List <SearchResultItem>();
var comments = new List <SearchResultItem>();
var users = new List <SearchResultItem>();
var hits = 0;
foreach (var item in items)
{
SearchResultDetail detail;
if (_cache.TryGet(item.DetailKey, out detail))
{
_log.TraceFormat("got {0} from cache", item.DetailKey);
}
if (detail == null)
{
// cache miss
switch (item.Type)
{
case SearchResultType.User:
users.Add(item);
break;
case SearchResultType.File:
files.Add(item);
break;
case SearchResultType.Comment:
comments.Add(item);
break;
default:
pages.Add(item);
break;
}
}
else
{
hits++;
item.Detail = detail;
}
}
_log.DebugFormat("got {0}/{1} items from cache", hits, items.Count);
if (pages.Any() || files.Any() || comments.Any())
{
// query lucene for meta data
var itemLookup = new Dictionary <string, SearchResultItem>();
var queries = new List <String>();
if (pages.Any())
{
BuildLookupQueries(pages, itemLookup, queries, "(id.page:(", ") AND type:wiki) ");
}
if (files.Any())
{
BuildLookupQueries(files, itemLookup, queries, "(id.file:(", ")) ");
}
if (comments.Any())
{
BuildLookupQueries(comments, itemLookup, queries, "(id.comment:(", ")) ");
}
if (users.Any())
{
BuildLookupQueries(users, itemLookup, queries, "(id.user:(", ")) ");
}
var count = 0;
foreach (var query in queries)
{
count++;
_log.DebugFormat("querying lucene for details ({0}/{1})", count, queries.Count);
DreamMessage luceneResult = null;
yield return(_searchPlug.With("wikiid", _wikiid).With("q", query).With("max", "all").GetAsync().Set(x => luceneResult = x));
if (!luceneResult.IsSuccessful)
{
// TODO (arnec): need error handling story here
throw new Exception("unable to query lucene for details");
}
var details = luceneResult.ToDocument()["document"];
_log.DebugFormat("retrieved {0} detail records", details.ListLength);
foreach (var detailDoc in details)
{
var detail = SearchResultDetail.FromXDoc(detailDoc);
SearchResultItem item;
if (detail["id.file"] != null)
{
itemLookup.TryGetValue(SearchResultType.File + ":" + detail["id.file"], out item);
}
else if (detail["id.comment"] != null)
{
itemLookup.TryGetValue(SearchResultType.Comment + ":" + detail["id.comment"], out item);
}
else if (detail["id.user"] != null)
{
itemLookup.TryGetValue(SearchResultType.User + ":" + detail["id.user"], out item);
}
else
{
itemLookup.TryGetValue(SearchResultType.Page + ":" + detail["id.page"], out item);
}
if (item != null)
{
_log.TraceFormat("got {0} from lucene", item.DetailKey);
_cache.Set(item.DetailKey, detail, TimeSpan.FromSeconds(_settings.GetValue("search/date-cache-time", 60d)));
item.Detail = detail;
}
}
}
_log.Debug("finished populating and caching detail records");
}
result.Return();
}
