InvertedIndex.cs

usingSystem.Collections.Generic;
usingSystem.Linq;

namespaceDataStructures;

/// <summary>
/// Inverted index is the simplest form of document indexing,
/// allowing performing boolean queries on text data.
///
/// This realization is just simplified for better understanding the process of indexing
/// and working on straightforward string inputs.
/// </summary>
publicclassInvertedIndex
{
privatereadonlyDictionary<string,List<string>>invertedIndex=new();

/// <summary>
/// Build inverted index with source name and source content.
/// </summary>
/// <param name="sourceName">Name of the source.</param>
/// <param name="sourceContent">Content of the source.</param>
publicvoidAddToIndex(stringsourceName,stringsourceContent)
{
varcontext=sourceContent.Split(' ').Distinct();
foreach(varwordincontext)
{
if(!invertedIndex.ContainsKey(word))
{
invertedIndex.Add(word,newList<string>{sourceName});
}
else
{
invertedIndex[word].Add(sourceName);
}
}
}

/// <summary>
/// Returns the source names contains ALL terms inside at same time.
/// </summary>
/// <param name="terms">List of terms.</param>
/// <returns>Source names.</returns>
publicIEnumerable<string>And(IEnumerable<string>terms)
{
varentries=terms
.Select(term =>invertedIndex
.Where(x =>x.Key.Equals(term))
.SelectMany(x =>x.Value))
.ToList();

varintersection=entries
.Skip(1)
.Aggregate(newHashSet<string>(entries.First()),(hashSet,enumerable)=>
{
hashSet.IntersectWith(enumerable);
returnhashSet;
});

returnintersection;
}

/// <summary>
/// Returns the source names contains AT LEAST ONE from terms inside.
/// </summary>
/// <param name="terms">List of terms.</param>
/// <returns>Source names.</returns>
publicIEnumerable<string>Or(IEnumerable<string>terms)
{
varsources=newList<string>();
foreach(varterminterms)
{
varsource=invertedIndex
.Where(x =>x.Key.Equals(term))
.SelectMany(x =>x.Value);

sources.AddRange(source);
}

returnsources.Distinct();
}
}