private static int DoIndex(IndexOptions opts)
{
using (var builder = IndexFactory.CreateBuilder(new PersistentIndexName(".", opts.DictionaryType, opts.FieldsType, opts.PostingType, opts.TextEncoding)))
{
builder.Start();
var timer = Stopwatch.StartNew();
var documents = 0;
foreach (var file in Directory.EnumerateFiles(opts.InputPath, opts.Filter, SearchOption.AllDirectories).Select(f => new FileInfo(f)))
{
PrintConsole(ConsoleColor.Gray, $"{file.FullName}");
if (opts.InputType == "text")
{
builder.AddFile(
file.FullName,
"{filename:\"" + file.FullName + "\", size:\"" + file.Length + "\", created:\"" + file.CreationTime.ToString("o") + "\"}");
}
else if (opts.InputType == "name")
{
builder.AddText(
file.FullName,
"{filename:\"" + file.FullName + "\", size:\"" + file.Length + "\", created:\"" + file.CreationTime.ToString("o") + "\"}");
}
else
{
throw new Exception("Unsupported input type");
}
++documents;
}
var stat = builder.StopAndWait();
PrintConsole(ConsoleColor.White, $"Indexed documents: {documents}, terms: {stat.Terms}, occurrences: {stat.Occurrences}, time: {timer.Elapsed}");
}
return(0);
}
public static IFullTextIndex AddToIndex(IIndexName indexName, string text)
{
using (var builder = IndexFactory.CreateBuilder(indexName))
{
builder.Start();
builder.AddText(text, null);
builder.StopAndWait();
}
return(IndexFactory.OpenIndex(indexName));
}
public static IFullTextIndex PrepareIndexForSearch(IIndexName indexName)
{
using (var builder = IndexFactory.CreateBuilder(indexName))
{
builder.Start();
builder.AddText("Hello World!", null);
builder.AddText("Petro Petrolium Petrol", null);
builder.AddText("This is test document for search unit tests", null);
builder.AddText("This test document is used for search operators", null);
builder.AddText("This full-text search only supports boolean operators: and, or", null);
builder.AddText("Programming is very exciting. Programs can help. This is fantastic!!!", null);
builder.StopAndWait();
}
return(IndexFactory.OpenIndex(indexName));
}
static void Main(string[] args)
{
PrintConsole(ConsoleColor.Green, "PMS Full-Text Search (c) Petro Protsyk 2017-2018");
if (args.Length < 1)
{
PrintHelp();
return;
}
if (args[0] == "index")
{
var fieldsType = (args.Length > 3 && args[2] == "--fieldsType") ? args[3] : "List";
using (var builder = IndexFactory.CreateBuilder(new PersistentIndexName(".", fieldsType)))
{
builder.Start();
var timer = Stopwatch.StartNew();
var documents = 0;
foreach (var file in Directory.EnumerateFiles(args[1], "*.txt", SearchOption.AllDirectories).Select(f => new FileInfo(f)))
{
PrintConsole(ConsoleColor.Gray, $"{file.FullName}");
builder.AddFile(
file.FullName,
"{filename:\"" + file.FullName + "\", size:\"" + file.Length + "\", created:\"" + file.CreationTime.ToString("o") + "\"}");
++documents;
}
builder.StopAndWait();
PrintConsole(ConsoleColor.White, $"Indexed documents: {documents}, time: {timer.Elapsed}");
}
}
else if (args[0] == "search")
{
var timer = Stopwatch.StartNew();
var documentsCount = 0;
var matchesCount = 0;
using (var index = IndexFactory.OpenIndex(new PersistentIndexName(".")))
{
using (var compiler = new FullTextQueryCompiler(index))
{
var searchQuery = compiler.Compile(args[1]);
var prevDoc = Occurrence.NoId;
foreach (var match in searchQuery.AsEnumerable())
{
if (match.DocumentId != prevDoc)
{
if (prevDoc != Occurrence.NoId)
{
PrintConsole(ConsoleColor.Gray, String.Empty);
}
PrintConsole(ConsoleColor.Gray, index.Fields.GetMetadata(match.DocumentId));
prevDoc = match.DocumentId;
documentsCount++;
}
++matchesCount;
PrintConsole(ConsoleColor.Gray, $"{match} ");
}
if (prevDoc != Occurrence.NoId)
{
PrintConsole(ConsoleColor.Gray, String.Empty);
}
}
}
PrintConsole(ConsoleColor.White, $"Documents found: {documentsCount}, matches: {matchesCount}, time: {timer.Elapsed}");
}
else if (args[0] == "print")
{
var timer = Stopwatch.StartNew();
var terms = 0;
using (var index = IndexFactory.OpenIndex(new PersistentIndexName(".")))
{
index.Visit(new PrintVisitor(index));
++terms;
}
PrintConsole(ConsoleColor.White, $"Terms: {terms}, time: {timer.Elapsed}");
}
else if (args[0] == "lookup")
{
var timer = Stopwatch.StartNew();
var termsFound = 0;
using (var index = IndexFactory.OpenIndex(new PersistentIndexName(".")))
{
int tilda = args[1].IndexOf("~");
IEnumerable <DictionaryTerm> terms = null;
if (tilda == -1)
{
terms = index.GetTerms(args[1]);
}
else
{
terms = index.GetTerms(args[1].Substring(0, tilda), int.Parse(args[1].Substring(tilda + 1)));
}
foreach (var term in terms)
{
++termsFound;
PrintConsole(ConsoleColor.Gray, term.Key);
}
}
PrintConsole(ConsoleColor.White, $"Terms found: {termsFound}, time: {timer.Elapsed}");
}
}
