public Processor(Configuration config, IDatabase database)
{
_config = config;
_database = database;
_listeners = new List<IMatchListener>();
_passthrough = new PassThroughFilter();
_choosebest = new ChooseBestFilter();
// precomputing for later optimizations
_proporder = new List<Property>();
foreach (Property p in _config.GetProperties())
{
if (!p.IsIdProperty)
_proporder.Add(p);
}
_proporder.Sort(new PropertyComparator());
// still precomputing
double prob = 0.5;
_accprob = new double[_proporder.Count];
for (int ix = _proporder.Count - 1; ix >= 0; ix--)
{
prob = StandardUtils.ComputeBayes(prob, _proporder[ix].HighProbability);
_accprob[ix] = prob;
}
}
private int _sizeix; // position in prevsizes
#endregion Fields
#region Constructors
public QueryResultTracker(Configuration config, Analyzer analyzer, IndexSearcher searcher, int maxSearchHits, float minRelevance)
{
_limit = 100;
_prevsizes = new int[10];
_config = config;
_analyzer = analyzer;
_maxSearchHits = maxSearchHits;
_searcher = searcher;
_minRelevance = minRelevance;
}
public LuceneDatabase(Configuration config, bool overwrite, DatabaseProperties dbprops)
{
_config = config;
_analyzer = new StandardAnalyzer(Version.LUCENE_29);
_maintracker = new QueryResultTracker(config, _analyzer, _searcher, dbprops.MaxSearchHits,
dbprops.MinRelevance);
_maxSearchHits = dbprops.MaxSearchHits;
_minRelevance = dbprops.MinRelevance;
try
{
OpenIndexes(overwrite);
OpenSearchers();
}
catch (Exception ex)
{
logger.Error("Error initializing object: {0}", ex.Message);
}
}
public Processor(Configuration config, bool overwrite)
: this(config, config.CreateDatabase(overwrite))
{
}
public Processor(Configuration config)
: this(config, true)
{
}
//Note that if file starts with 'classpath:' the resource is looked
// up on the classpath instead.
public static Configuration Load(string file)
{
var cfg = new Configuration();
var properties = new List<Property>();
// Get the appropriate nodes using Linq to XML
XElement xml = XElement.Load(file);
// Get the threshold
double threshold =
xml.Elements("schema").Descendants("threshold").Select(x => double.Parse(x.Value)).FirstOrDefault();
cfg.Threshold = threshold;
// Get all of the properties
IEnumerable<XElement> xmlProperties = from s in xml.Elements("schema")
from p in s.Descendants("property")
select p;
foreach (XElement xElement in xmlProperties)
{
string propName = xElement.Descendants("name").First().Value;
var property = new Property(propName);
// Check to see if this is an id property
XAttribute xAttribute = xElement.Attribute("type");
if (xAttribute != null)
{
string id = xAttribute.Value;
if (id != null && id == "id")
{
property.IsIdProperty = true;
}
}
else
{
string comparatorName = xElement.Descendants("comparator").FirstOrDefault().Value;
property.Comparator = GetComparatorFromString(comparatorName);
property.LowProbability =
xElement.Descendants("low").Select(x => double.Parse(x.Value)).FirstOrDefault();
property.HighProbability =
xElement.Descendants("high").Select(x => double.Parse(x.Value)).FirstOrDefault();
properties.Add(property);
}
}
cfg.SetProperties(properties);
//// Get the datasources
//XPathNodeIterator dsi = xpn.Select("/duke/*[not(self::schema)]");
//while (dsi.MoveNext())
//{
// if (dsi.Current != null && xpi.Current.Name == "csv")
// {
// var datasource = GetCsvDataSourceFromXml(dsi, xpn);
// }
//}
IEnumerable<XElement> dataSources = from d in xml.Elements()
where d.Name != "schema"
select d;
foreach (XElement dataSource in dataSources)
{
if (dataSource.Name == "csv")
{
var csvDs = new CsvDataSource();
Hashtable csvParams = GetParametersTable(dataSource);
csvDs.File = csvParams["input-file"].ToString();
if (csvParams.Contains("header-line"))
csvDs.HasHeader = (csvParams["header-line"].ToString().ToLower() == "true");
if (csvParams.Contains("skip-lines"))
{
int skipLines = 0;
csvDs.SkipLines = Int32.TryParse(csvParams["skip-lines"].ToString(), out skipLines)
? skipLines
: 0;
}
csvDs.FileEncoding = csvParams.Contains("encoding")
? GetTextEncodingFromString(csvParams["encoding"].ToString())
: Encoding.Default;
List<Column> cols = GetDataSourceColumns(dataSource);
foreach (Column column in cols)
{
csvDs.AddColumn(column);
}
cfg.AddDataSource(0, csvDs);
}
}
return cfg;
}
