137 lines
4.0 KiB
C#
137 lines
4.0 KiB
C#
using System.Text;
|
|
using ConsoleApp2.Entities;
|
|
using ConsoleApp2.Helpers;
|
|
using Microsoft.Extensions.Logging;
|
|
|
|
namespace ConsoleApp2.Services;
|
|
|
|
public class NewCsvSource
|
|
{
|
|
private readonly string _filePath;
|
|
private readonly StreamReader _reader;
|
|
private readonly ILogger? _logger;
|
|
private readonly string _tableName;
|
|
|
|
public DataRecord Current { get; protected set; }
|
|
public string[]? Headers { get; }
|
|
public string? CurrentRaw { get; private set; }
|
|
public string Delimiter { get; private set; }
|
|
public char QuoteChar { get; private set; }
|
|
|
|
public NewCsvSource(string filePath, string[]? headers = null, string delimiter = ",", char quoteChar = '"',
|
|
ILogger? logger = null)
|
|
{
|
|
_filePath = filePath;
|
|
Headers = headers;
|
|
_logger = logger;
|
|
Delimiter = delimiter;
|
|
QuoteChar = quoteChar;
|
|
|
|
var fs = File.OpenRead(filePath);
|
|
_reader = new StreamReader(fs);
|
|
_tableName = DumpDataHelper.GetTableName(filePath);
|
|
}
|
|
|
|
public async ValueTask<bool> ReadAsync()
|
|
{
|
|
var str = await _reader.ReadLineAsync();
|
|
if (string.IsNullOrWhiteSpace(str))
|
|
return false;
|
|
|
|
CurrentRaw = str;
|
|
|
|
var fields = ParseRow2(str, QuoteChar, Delimiter);
|
|
Current = new DataRecord(fields, _tableName, Headers);
|
|
return true;
|
|
}
|
|
|
|
public string[] ParseRow(string row, char quoteChar, string delimiter)
|
|
{
|
|
var span = row.AsSpan();
|
|
var result = new List<string>();
|
|
|
|
if (span.Length == 0)
|
|
throw new ArgumentException("The row is empty", nameof(row));
|
|
|
|
var isInQuote = span[0] == quoteChar;
|
|
var start = 0;
|
|
for (var i = 1; i < span.Length; i++)
|
|
{
|
|
if (span[i] == quoteChar)
|
|
{
|
|
isInQuote = !isInQuote;
|
|
}
|
|
// delimiter需要足够复杂
|
|
else if (/*!isInQuote && */span.Length > i + delimiter.Length && span[i..(i + delimiter.Length)].Equals(delimiter, StringComparison.CurrentCulture)) // field matched
|
|
{
|
|
string field;
|
|
if (span[start] == quoteChar && span[i - 1] == quoteChar) // enclosed by quoteChar
|
|
field = span[(start + 1)..(i - 1)].ToString(); // escape quoteChar
|
|
else
|
|
field = span[start..i].ToString();
|
|
|
|
start = i + delimiter.Length;
|
|
|
|
if (field == "\\N")
|
|
field = "NULL";
|
|
|
|
result.Add(field);
|
|
|
|
continue;
|
|
}
|
|
}
|
|
|
|
result.Add(span[start..].ToString());
|
|
|
|
|
|
for (var i = 0; i < result.Count; i++)
|
|
{
|
|
var field = result[i];
|
|
if (DumpDataHelper.CheckHexField(field) && StringExtensions.CheckJsonHex(field))
|
|
{
|
|
result[i] = StringExtensions.FromHex(field);
|
|
}
|
|
}
|
|
|
|
return result.ToArray();
|
|
}
|
|
|
|
public string[] ParseRow2(ReadOnlySpan<char> source, char quoteChar, string delimiter)
|
|
{
|
|
var result = new List<string>();
|
|
var index = -1;
|
|
StringBuilder current = new StringBuilder();
|
|
bool hasQuote = false;
|
|
bool hasSlash = false;
|
|
while (index < source.Length-1)
|
|
{
|
|
index++;
|
|
if (hasSlash == false && source[index] == '\\')
|
|
{
|
|
hasSlash = true;
|
|
current.Append('\\');
|
|
continue;
|
|
}
|
|
if (hasSlash ==false && source[index] == quoteChar)
|
|
{
|
|
hasQuote = !hasQuote;
|
|
current.Append(source[index]);
|
|
continue;
|
|
}
|
|
if (hasQuote==false && source[index] == delimiter[0])
|
|
{
|
|
result.Add(current.ToString());
|
|
current.Clear();
|
|
}
|
|
else
|
|
{
|
|
current.Append(source[index]);
|
|
}
|
|
|
|
hasSlash = false;
|
|
}
|
|
|
|
result.Add(current.ToString());
|
|
return result.ToArray();
|
|
}
|
|
} |