/// <summary>
/// Reads file with synonyms from MOBY format
/// </summary>
/// <param name="filename">The filename.</param>
/// <param name="datastructure">The in memory storage structure.</param>
/// <exception cref="System.IO.FileNotFoundException">Passed filename with keywords not found.</exception>
/// <exception cref="System.IO.IOException">Failed to read file.</exception>
public void ReadFile(string filename, ref IinMemoryStorageStructure datastructure)
{
if (datastructure == null)
{
throw new ArgumentNullException("data structure must me initialized and could not be null");
}
if (!File.Exists(filename))
{
throw new FileNotFoundException("Passed filename with keywords not found.", filename);
}
int stringCounter = 0;
using (StreamReader reader = new System.IO.StreamReader(filename))
{
while (!reader.EndOfStream)
{
stringCounter++;
var input = reader.ReadLine();
if (input.IndexOf(',') > 0)
{
var keyword = input.Substring(0, input.IndexOf(","));
if (this.validator != null && !this.validator.Validate(keyword))
{
throw new FormatException(string.Format("Format on the line {0} is not right. Keyword '{1}' contains illegal characters", stringCounter, keyword));
}
var collection = input.Substring(input.IndexOf(",") + 1).Trim().Split(',').Select(s => s.Trim()).ToArray();
List<string> checkedCollection = new List<string>();
foreach (var item in collection)
{
if (this.validator != null && !this.validator.Validate(item))
{
throw new FormatException(string.Format("Format on the line {0} is not right. Illegal characters value: '{1}'", stringCounter, item));
}
if (item.Trim().Equals(string.Empty))
{
// skip silently
}
else
{
checkedCollection.Add(item.Trim());
}
}
datastructure.AddSynonyms(keyword.Trim(), checkedCollection.ToArray());
}
}
}
}
/// <summary>
/// Initializes a new instance of the <see cref="InMemorySuggester"/> class.
/// </summary>
/// <param name="storageStructure">Storage DI</param>
/// <param name="reader">reader that reads file to memory from specified format</param>
/// <param name="filename">Filename</param>
/// <param name="sourceName">Optional name of the source</param>
public InMemorySuggester(IinMemoryStorageStructure storageStructure, IToMemoryReader reader, string filename, string sourceName = "")
{
if (storageStructure == null)
{
throw new ArgumentNullException("storageSturcture could not be null");
}
if (reader == null)
{
throw new ArgumentNullException("reader could not be null");
}
this.reader = reader;
this.storage = storageStructure;
this.reader.ReadFile(filename, ref this.storage);
this.id = Guid.NewGuid();
this.name = sourceName.Equals(string.Empty) ? this.id.ToString() : sourceName;
}
/// <summary>
/// Reads file with synonyms from Ripicts format (word, syn1;raiing1,syn2;raiting2 etc.)
/// </summary>
/// <param name="filename">The filename.</param>
/// <param name="datastructure">The in memory storage structure.</param>
/// <exception cref="System.ArgumentNullException">data structure must me initialized and could not be null</exception>
/// <exception cref="System.IO.FileNotFoundException">Passed filename with keywords not found.</exception>
/// <exception cref="System.IO.IOException">Failed to read file.</exception>
public void ReadFile(string filename, ref IinMemoryStorageStructure datastructure)
{
if (datastructure == null)
{
throw new ArgumentNullException("datastructure must me initialized and could not be null");
}
if (!File.Exists(filename))
{
throw new FileNotFoundException("Passed filename with keywords not found.", filename);
}
int stringCounter = 0;
using (StreamReader reader = new System.IO.StreamReader(filename))
{
KeywordValidator validator = new KeywordValidator();
while (!reader.EndOfStream)
{
stringCounter++;
var input = reader.ReadLine();
// empty string ignore them
if (!input.Trim().Equals(string.Empty))
{
string keyword;
var synRelevanceCollection = new List<Tuple<string, int>>();
if (input.IndexOf(',') > 0)
{
keyword = input.Substring(0, input.IndexOf(","));
var collection = input.Substring(input.IndexOf(",") + 1).Trim().Split(',').Select(s => s.Trim()).ToArray();
for (int i = 0; i < collection.Length; i++)
{
var element = collection[i];
var pair = element.Trim().Split(';');
if (pair.Length > 0 && validator.Validate(pair[0]))
{
var synvalue = pair[0].Trim();
int relValue = 0;
if (pair.Length > 1)
{
int.TryParse(pair[1], out relValue);
}
if (!synvalue.Equals(string.Empty))
{
synRelevanceCollection.Add(new Tuple<string, int>(synvalue, relValue));
}
}
else
{
throw new FormatException(string.Format("Format on the line {0} is not right. Missing ';' symbol or value '{1}' illegal.", stringCounter, element));
}
}
}
else
{
// means that this is just a word without synonyms;
keyword = input.Trim();
}
try
{
datastructure.AddSynonyms(keyword, synRelevanceCollection.ToArray());
}
catch (Exception ex)
{
throw new Exception(string.Format("Format on the line {0}.", stringCounter), ex);
}
}
}
}
}