static void WriteCsv(CsvZipPackage csvz, string csv)
{
Console.WriteLine($"Processing {Path.GetFileName(csv)}: ");
var data = CsvDataReader.Create(csv);
var analyzer = new SchemaAnalyzer();
Console.Write($" Analyzing. ");
var sw = Stopwatch.StartNew();
var result = analyzer.Analyze(data);
Console.WriteLine(sw.Elapsed.ToString());
var schema = result.GetSchema();
var csvSchema = new CsvSchema(schema);
var csvOpts = new CsvDataReaderOptions {
Schema = csvSchema
};
Console.Write($" Writing. ");
data = CsvDataReader.Create(csv, csvOpts);
sw = Stopwatch.StartNew();
var entry = csvz.CreateEntry(Path.GetFileName(csv));
entry.WriteData(data);
Console.WriteLine(sw.Elapsed.ToString());
}
public override int Execute(CommandContext context, AnalyzeSettings settings)
{
var filename = settings.File;
var output = settings.Schema ?? filename + ".schema";
Stream oStream =
output == "."
? Console.OpenStandardOutput()
: File.Create(output);
var tr = new StreamReader(filename);
for (int i = 0; i < settings.Skip; i++)
{
tr.ReadLine();
}
var csv = CsvDataReader.Create(tr);
var a = new SchemaAnalyzer(new SchemaAnalyzerOptions {
AnalyzeRowCount = settings.Lines
});
var re = a.Analyze(csv);
var sb = re.GetSchemaBuilder();
foreach (var col in sb)
{
// TODO: think more about how to handle columns size
col.ColumnSize = null;
}
var schema = sb.Build();
using var tw = new StreamWriter(oStream);
tw.Write(schema.ToString().Replace(",", "," + Environment.NewLine));
return(0);
}