/// <summary>
/// Retrieves J
/// </summary>
/// <param name="article"></param>
/// <returns></returns>
public string GetJsonArticle(Article article)
{
var unpackedDictionary = new Dictionary <String, String>();
var jsonFileName = Zipper.Unzip(article.ZippedJsonFileName.ToByteArray());
var bucketFileName = Zipper.Unzip(article.ZippedBucketFileName.ToByteArray());
using (var fs = new FileStream(Path.Combine(_inputDirectory, bucketFileName), FileMode.Open))
{
using (var zip = new ZipArchive(fs))
{
var entry = zip.Entries.First();
using (StreamReader sr = new StreamReader(entry.Open()))
{
BinaryFormatter serializer = new BinaryFormatter();
unpackedDictionary = serializer.Deserialize(sr.BaseStream) as Dictionary <String, String>;
}
}
}
var json = String.Empty;
if (unpackedDictionary.ContainsKey(jsonFileName))
{
json = unpackedDictionary[jsonFileName];
}
return(json);
}
/// <summary>
/// Searches for a specific list of articles containing key words in abstract.
/// </summary>
/// <param name="keyWords"></param>
/// <returns></returns>
public Articles SearchArticles(string phraseToSearch)
{
var result = new Articles();
var smallKeyWords = phraseToSearch
.ToLower()
.Split(new[] { ' ' }, StringSplitOptions.RemoveEmptyEntries)
.Distinct()
.ToList();
var commonArticles = new Articles();
bool haveCommonArticlesBeenSet = false;
// First check if a key word is found or not.
foreach (var keyWord in smallKeyWords)
{
if (keyWord.Length < 3)
{
// no intereset in short words
continue;
}
var keyWordIndex = keyWord.ToLower().Substring(0, 3);
if (!ArticlesHubDictionary.ContainsKey(keyWordIndex))
{
keyWordIndex = REST;
}
// now look at contents of index file
var articlesHub = new ArticlesHub();
var indexPath = Path.Combine(_outputDirectory, "index", $"index_{keyWordIndex}.bin.zip");
if (!File.Exists(indexPath))
{
continue;
}
articlesHub.Deserialize(indexPath);
// now search for keywords in this hub
if (articlesHub.KeyWordsDictionary.Dictionary.Keys.Contains(keyWord))
{
var articlesFound = articlesHub.KeyWordsDictionary.Dictionary[keyWord];
var commonArticlesCopy = new Articles();
if (haveCommonArticlesBeenSet)
{
var commonArticlesHashSet = commonArticles
.List.Select(x => Zipper.Unzip(x.ZippedJsonFileName.ToByteArray()))
.ToHashSet();
foreach (var articleFound in articlesFound.List)
{
var articleFoundFileName = Zipper.Unzip(articleFound.ZippedJsonFileName.ToByteArray());
var hasCommonArticle = commonArticlesHashSet.Contains(articleFoundFileName);
if (hasCommonArticle)
{
var isTitleEmpty =
String.IsNullOrWhiteSpace(Zipper.Unzip(articleFound.ZippedTitle.ToByteArray()));
commonArticlesCopy.List.Add(articleFound);
}
}
commonArticles = commonArticlesCopy;
}
else
{
var nonEmptyTitlesList = articlesFound.List
.Where(x => !String.IsNullOrWhiteSpace(Zipper.Unzip(x.ZippedTitle.ToByteArray())));
commonArticles.List.AddRange(nonEmptyTitlesList);
haveCommonArticlesBeenSet = true;
}
}
}
return(commonArticles);
}