public static string readOCR (Uri ocr_url) {
AzureSearchServiceController checkUrl = new AzureSearchServiceController();
string ocrPrevText = "";
if (checkUrl.RemoteFileExists(ocr_url.ToString()))
{
System.Net.WebClient wc = new System.Net.WebClient();
byte[] raw = wc.DownloadData(ocr_url);
string webData = System.Text.Encoding.UTF8.GetString(raw);
string[] ocrSplit = webData.Split(' ');
for (int i = 0; i < 300; i++)
{
ocrPrevText += ocrSplit[i];
ocrPrevText += " ";
}
return ocrPrevText;
}
else
{
ocrPrevText ="Unfortunately OCR is not available for this document";
} return ocrPrevText;
}
public ActionResult Index(string queryID, int? page)
{
ViewBag.queryID = queryID;
if (!serviceUrl.EndsWith("/"))
{
serviceUrl += "/";
}
if (!serviceUrl.EndsWith("indexes/"))
{
serviceUrl += "indexes/";
}
try
{
//Delete the Index if it exists
DeleteIndex(serviceUrl + indexName);
// Create the index
if (CreateIndex() == false)
{
ViewBag.Error = "Error From Create Index";
}
// Add documents to index - loading in 100 document batches
if (PostDocuments("data2.json") == false)
{
ViewBag.ErrorPostDoc = "Error from post documents0";
}
if (PostDocuments("data3.json") == false)
{
ViewBag.ErrorPostDoc = "Error from post documents0";
}
//Wait 5 seconds
// Console.WriteLine("Waiting 5 seconds for data to be indexed...");
Thread.Sleep(TimeSpan.FromSeconds(5));
setUserInput(queryID);
//Construct of an Object
AzureSearchServiceController azureSearchService = new AzureSearchServiceController();
List<int> booleanSelection;
string[] keywordOfFields; //declare keywordOfFields as string array of any size
String toPrintQuerySelection = "";
List<List<BookModel>> listOfQueries;
azureSearchService.initialiseListsFromAzureSearch(queryID, out keywordOfFields, out booleanSelection, out listOfQueries, ref toPrintQuerySelection);
ViewBag.searchQuery = toPrintQuerySelection;
var title = listOfQueries[0].Select(item => item.title);
var finalResults = listOfQueries[7].Where(item => title.Contains(item.title));
int position;
List<BookModel> finalList = new List<BookModel>();
azureSearchService.booleanFunctionality(queryID,booleanSelection, keywordOfFields, out position, listOfQueries, ref finalResults, ref finalList);
insertElements(finalResults, ref finalList);
finalList = finalList.GroupBy(x => x.title).Select(x => x.First()).ToList();
List<BookModel> bookListIntersection = new List<BookModel>();
bookListIntersection = finalList.ToList();
/*
ViewBag.list = listOfQueries;
ViewBag.keywords = keywordOfFields;
ViewBag.boolean = booleanSelection;
ViewBag.count = position;
ViewBag.final = finalList;
ViewBag.finalCount = finalList.Count;
*/
//ViewBag.countIntersection = bookListIntersection.Count();
ViewBag.ListResponse = bookListIntersection;
resultingList = bookListIntersection.ToList();
int bookListCount;
if ((bookListIntersection.Count() != 0))
{
bookListCount = bookListIntersection.Count();
}
else
{
bookListCount = 0;
}
bookListTotal = bookListCount;
ViewBag.resultCount = bookListCount;
}
catch (Exception e)
{
// Console.WriteLine("Unhandled exception caught:");
while (e != null)
{
//Console.WriteLine("\t{0}", e.Message);
e = e.InnerException;
}
//Console.WriteLine("\nDid you remember to paste your service URL and API key into App.config?\n");
}
// Console.Write("Complete. Press <enter> to continue: ");
//var name = Console.ReadLine();
var pagedResultList = resultingList.ToList();
int pageResultTotal = pagedResultList.Count();
ViewBag.resultTotal = pageResultTotal;
var pageNumber = page ?? 1;
var onePageOfProducts = pagedResultList.ToPagedList(pageNumber, 15);
ViewBag.OnePageOfProducts = onePageOfProducts;
if (page > 1)
{
ViewBag.PageNumber = page;
}
else
{
ViewBag.PageNumber = 1;
}
/********************* CODE FOR "DISPLAYING XX - YY ... RESULTS" *************************/
//amount of pages
int pageTotal = 0;
if (bookListTotal != 0)
{
pageTotal = (bookListTotal / 15) + 1;
}
else
{
pageTotal = 1;
}
// from and to
int fromResult = 0;
int toResult = 0;
int lastPageResults = (bookListTotal % 15); //amount of results to display on last page
//calculating resutlts to and from indexes
//last page - checked first
if (pageNumber == pageTotal) //last page
{
if (bookListTotal == 0) //if no results
{
fromResult = 0;
toResult = 0;
}
else
{
fromResult = (bookListTotal - lastPageResults) + 1;
toResult = bookListTotal;
}
}
else
{
if (pageNumber != 1) //pages 2 - penulti
{
fromResult = (15 * (pageNumber - 1)) + 1;
toResult = (15 * pageNumber);
}
else //first page
{
fromResult = 1;
if (bookListTotal < 15) //when less than 15 results to display
{
toResult = bookListTotal;
}
else
{
toResult = 15;
}
}
}
ViewBag.fromResult = fromResult;
ViewBag.toResult = toResult;
return View();
}
//
// GET: /StatisticalAnalysis/
public ActionResult Index(Uri website, string titleOfBook)
{
AzureSearchServiceController checkUrl = new AzureSearchServiceController();
if (checkUrl.RemoteFileExists(website.ToString()))
{
List<string> MostFrequentWords = new List<string>();
List<int> FrequencyOfMostFrequentWords = new List<int>();
frequentWords(website, out MostFrequentWords, out FrequencyOfMostFrequentWords);
ViewBag.FrequencyOfMostFrequentWords = FrequencyOfMostFrequentWords;
ViewBag.MostFrequentWords = MostFrequentWords;
ViewBag.titleOfBook = titleOfBook;
ViewBag.website = website;
//stores the user keyword to identify the frequency of a word
string keyword = Request["txtSearch"];
ViewBag.keyword = Request["txtSearch"];
/////////////////////////////////////////
//stores the user keyword to identify the frequency of a biagram
string keywordBiagram = Request["txtSearchBiagram"];
ViewBag.keywordBiagram = Request["txtSearchBiagram"];
////////////////////////////////////////////////////
//stores the user keyword to identify the frequency of a triagram
string keywordTriagram = Request["txtSearchTriagram"];
ViewBag.keywordBiagram = Request["txtSearchTriagram"];
char[] delimiters = new char[] { ' ' };
int frequencyWordInOCR = 0;
if (String.IsNullOrEmpty(keyword))
{
ViewBag.initialState = 0;
}
else
{
string[] userKeyword = keyword.Split(delimiters, StringSplitOptions.RemoveEmptyEntries);
//display a message to the user (the user has entered more than one word)
if (userKeyword.Count() != 1) { ViewBag.MessageOneWord = "Please enter one word"; }
else
{
frequencyWordInOCR = frequentWordInOCR(website, keyword.ToLower().Trim());
if (frequencyWordInOCR == 0) { ViewBag.WordNotFound = "Word: " + keyword + " " + "was not found"; }
else { ViewBag.wordFound = "The frequency of the word " + keyword + " " + "is " + frequencyWordInOCR; }
}
}
//////////////////////////////Biagram/////////////////////////////
int frequencyOfBiagram = 0;
if (String.IsNullOrEmpty(keywordBiagram))
{
ViewBag.initialStateBiagram = 0;
}
else
{
string[] wordsBiagram = keywordBiagram.Split(delimiters, StringSplitOptions.RemoveEmptyEntries);
//check whether the user has entered two words
if (wordsBiagram.Count() != 2) { ViewBag.MessageBiagram = "Please enter a valid biagram(biagram consists of two words)"; }
else
{
frequencyOfBiagram = frequentOfBiagrams(website, keywordBiagram.ToLower().Trim());
if (frequencyOfBiagram == 0) { ViewBag.WordNotFoundBiagram = "Word: " + keywordBiagram + " " + "was not found"; }
else { ViewBag.wordFoundBiagram = "The frequency of a biagram " + keywordBiagram + " " + "is " + frequencyOfBiagram; }
}
}
///////////////////////////////////////////////////////////////////////////////////////
//this section calculate the frequency word of a triagram and checks whether the user has entered three words in the document
int frequencyOfTriagram = 0;
if (String.IsNullOrEmpty(keywordTriagram))
{
ViewBag.initialStateTriagram = 0;
}
else
{
string[] wordsTriagram = keywordTriagram.Split(delimiters, StringSplitOptions.RemoveEmptyEntries);
//check whether the user has entered three words
if (wordsTriagram.Count() != 3) { ViewBag.MessageTriagram = "Please enter a valid triagram(triagram consists of two words)"; }
else
{
frequencyOfTriagram = frequentOfBiagrams(website, keywordTriagram.ToLower().Trim());
if (frequencyOfTriagram == 0) { ViewBag.WordNotFoundTriagram = "Word: " + keywordTriagram + " " + "was not found"; }
else { ViewBag.wordFoundTriagram = "The frequency of a triagram " + keywordTriagram + " " + "is " + frequencyOfTriagram; }
}
}
return View();
}
else
{
return RedirectToAction("ErrorPage", "ErrorPage");
}
}
public void AzureSearch()
{
AzureSearchServiceController SearchByPublisher = new AzureSearchServiceController();
var results = SearchByPublisher.insertANDElemensToFinalList()
results.RouteValues["action"].Equals("Index");
results.RouteValues["controller"].Equals("Home");
Assert.AreEqual("Index", results.RouteValues["action"]);
Assert.AreEqual("AzureSearchService", results.RouteValues["controller"]);
}
public void Search_By_Publisher()
{
//Create Mock Data
List<string> tagsList = new List<string>();
tagsList.Add("lewis");
tagsList.Add("Shetlander");
tagsList.Add("shetlander");
tagsList.Add("Lewis");
List<string> creators_and_contributorsList = new List<string>();
creators_and_contributorsList.Add("GUNN, John - M.A., D.Sc");
//Create a mock model to pass as a parameter
BookModel model = new BookModel
{
azure_url = "http://blmc.blob.core.windows.net/1894/001544024_0_000191_1_1894_plates.jpg",
creators_and_contributors = creators_and_contributorsList,
date = 1894,
electronicsysnum = 014812388,
flickr_original_jpeg = "http://farm4.staticflickr.com/3811/11230026636_0021861570_o.jpg",
flickr_url = "https://flickr.com/photos/britishlibrary/11230026636",
fromshelfmark = "British Library HMNTS 012630.e.11.",
height = 1896,
idx = 1,
// This OCR Text is INVALID
ocrtext = "http://blmc.blob.core.windows.net/ocrplaintext/001544024_0.txt",
place = "London",
printsysnum = 001218010,
publisher = "T. Nelson & Sons",
scannumber = 000191,
sizebracket = "plates",
tags = tagsList,
title = "Sons of the Vikings. An Orkney story",
vol = 0,
width = 1256
};
AzureSearchServiceController SearchByPublisher = new AzureSearchServiceController();
var results = (RedirectToRouteResult)SearchByPublisher.SearchByPublisher(model.publisher);
results.RouteValues["action"].Equals("Index");
results.RouteValues["controller"].Equals("Home");
Assert.AreEqual("Index", results.RouteValues["action"]);
Assert.AreEqual("AzureSearchService", results.RouteValues["controller"]);
}
public void Download_Test_VALID_URL()
{
//Create Mock Data
List<string> tagsList = new List<string>();
tagsList.Add("lewis");
tagsList.Add("Shetlander");
tagsList.Add("shetlander");
tagsList.Add("Lewis");
List<string> creators_and_contributorsList = new List<string>();
creators_and_contributorsList.Add("GUNN, John - M.A., D.Sc");
//Create a mock model to pass as a parameter
BookModel model = new BookModel
{
azure_url = "http://blmc.blob.core.windows.net/1894/001544024_0_000191_1_1894_plates.jpg",
creators_and_contributors = creators_and_contributorsList,
date = 1894,
electronicsysnum = 014812388,
flickr_original_jpeg = "http://farm4.staticflickr.com/3811/11230026636_0021861570_o.jpg",
flickr_url = "https://flickr.com/photos/britishlibrary/11230026636",
fromshelfmark = "British Library HMNTS 012630.e.11.",
height = 1896,
idx = 1,
// This OCR Text is INVALID
ocrtext = "http://blmc.blob.core.windows.net/ocrplaintext/001544024_0.txt",
place = "London",
printsysnum = 001218010,
publisher = "T. Nelson & Sons",
scannumber = 000191,
sizebracket = "plates",
tags = tagsList,
title = "Sons of the Vikings. An Orkney story",
vol = 0,
width = 1256
};
AzureSearchServiceController DownloadTest = new AzureSearchServiceController();
Uri testUri = new Uri(model.ocrtext);
var results = DownloadTest.Download(testUri, model.azure_url, model.creators_and_contributors, model.date.ToString(), model.electronicsysnum.ToString(), model.flickr_original_jpeg
, model.flickr_url, model.fromshelfmark, model.height.ToString(), model.idx.ToString(), model.place, model.printsysnum.ToString(), model.publisher, model.scannumber.ToString(), model.sizebracket, model.tags.ToString()
, model.title, model.vol.ToString(), model.width.ToString());
Assert.AreEqual(typeof(FileStreamResult), results.GetType());
}
