public async Task IndexDocument(string fileName, string filePath, string contents)
{
var id = HashUtility.Hash(filePath);
var doc = new Doc()
{
Id = id,
FileName = fileName,
Path = filePath,
Text = contents
};
await CreateElasticClient().IndexDocumentAsync(doc);
}
static async Task BuildIndex(string startPath)
{
Console.WriteLine("Indexes Directory: " + startPath);
// Use TikaOnDotNet to extract the contents of the document.
var textExtractor = new TextExtractor();
// Use ElasticSearch to index the document
var client = CreateElasticClient();
// Loop through all files in the passed in directory
var directory = new DirectoryInfo(startPath);
if (directory.Exists)
{
foreach (var file in directory.GetFiles())
{
var contents = textExtractor.Extract(file.FullName);
if (!string.IsNullOrWhiteSpace(contents.Text))
{
Console.WriteLine("Indexing File " + file.FullName);
var id = HashUtility.Hash(file.FullName);
var doc = new Doc()
{
Id = id,
FileName = file.Name,
Path = file.FullName,
Text = contents.Text
};
await client.IndexDocumentAsync(doc);
}
}
foreach (var dir in directory.GetDirectories())
{
await BuildIndex(dir.FullName);
}
}
}
