using System.Collections.Generic;\r
using System.Text;\r
using System.Text.RegularExpressions;\r
-\r
+using System.Data;\r
\r
namespace Karinto\r
{\r
- public class CsvReader<TDataSet> where TDataSet : IDataSet, new()\r
+ public class CsvReader : CsvReader<DataTable>\r
{\r
- private string separator = ",";\r
+ private CsvReader()\r
+ { \r
+ }\r
\r
- public string Separator\r
+ public CsvReader(DataTable prototype)\r
+ : base(prototype)\r
{\r
- get { return separator; }\r
}\r
+ }\r
\r
- /// <summary>\r
- /// 区切り文字を指定する\r
- /// カンマ以外だったらCSVじゃないような...?\r
- /// </summary>\r
- /// <param name="separator">区切り文字(正規表現可)</param>\r
- public void SetSeparator(string separator)\r
+ public class CsvReader<TDataTable> where TDataTable : DataTable, new()\r
+ {\r
+ #region private fields\r
+\r
+ private string separator;\r
+ private Regex sepPattern;\r
+ private TDataTable prototype;\r
+ private DataColumnCollection cols;\r
+\r
+ public delegate bool Parser(ref string input, out object output);\r
+ static private Dictionary<Type, Parser> parsers;\r
+ private Parser[] lineParser;\r
+\r
+ #endregion\r
+\r
+ #region constructors\r
+ static CsvReader()\r
{\r
- this.separator = separator;\r
+ parsers = new Dictionary<Type, Parser>();\r
+ parsers[typeof(Boolean)] = ParseBoolean;\r
+ parsers[typeof(Double)] = ParseDouble;\r
+ parsers[typeof(Single)] = ParseSingle;\r
+ parsers[typeof(Decimal)] = ParseDecimal;\r
+ parsers[typeof(Int32)] = ParseInt32;\r
+ parsers[typeof(Int64)] = ParseInt64;\r
}\r
\r
- /// <summary>\r
- /// CSVファイルを読み込む\r
- /// </summary>\r
- /// <param name="path">CSVファイルのパス</param>\r
- /// <param name="proto">行の雛型</param>\r
- /// <returns>指定した型のリスト</returns>\r
- public List<TDataSet> Read(string path)\r
+ public CsvReader()\r
+ : this(new TDataTable())\r
{\r
- return Read(path, new TDataSet());\r
}\r
\r
+ public CsvReader(TDataTable prototype)\r
+ {\r
+ Separator = @",";\r
+ this.prototype = (TDataTable)prototype.Clone();\r
+ cols = prototype.Columns;\r
+ lineParser = new Parser[cols.Count];\r
+ for (int i = 0; i < cols.Count; ++i)\r
+ {\r
+ Type t = cols[i].DataType;\r
+ if (parsers[t] != null)\r
+ {\r
+ lineParser[i] = parsers[t];\r
+ }\r
+ else if (t == typeof(DateTime))\r
+ {\r
+ lineParser[i] = ParseDateTime;\r
+ }\r
+ else\r
+ {\r
+ lineParser[i] = ParseString;\r
+ }\r
+ }\r
+ }\r
+ #endregion\r
+\r
+ #region properties\r
+\r
/// <summary>\r
- /// 雛型を指定してCSVファイルを読み込む\r
+ /// 区切り文字\r
/// </summary>\r
- /// <param name="path">CSVファイルのパス</param>\r
- /// <param name="proto">行の雛型</param>\r
- /// <returns>指定した型のリスト</returns>\r
- public List<TDataSet> Read(string path, TDataSet proto)\r
+ public string Separator\r
{\r
- List<TDataSet> records = new List<TDataSet>();\r
+ get\r
+ {\r
+ return separator;\r
+ }\r
+ set\r
+ {\r
+ separator = value;\r
+ sepPattern = new Regex(separator, RegexOptions.Compiled);\r
+ }\r
+ }\r
+\r
+ #endregion\r
\r
+ #region public methods\r
+\r
+ public TDataTable Read(FilePath path)\r
+ {\r
+ TDataTable table = (TDataTable)prototype.Clone();\r
try\r
{\r
- using (FileStream fs = new FileStream(path, FileMode.Open, FileAccess.Read, FileShare.ReadWrite))\r
+ using (FileStream fs = new FileStream(\r
+ path, FileMode.Open, FileAccess.Read, FileShare.ReadWrite))\r
{\r
using (StreamReader reader = new StreamReader(fs))\r
{\r
- string line = "";\r
- // 数値にマッチする正規表現\r
- string rfloat = @"\s*([-+]?(?:[0-9]+(\.[0-9]*)?|(\.[0-9]+))([eE][-+]?[0-9]+)?)\s*";\r
- Regex rCsv = new Regex(rfloat + separator, RegexOptions.Compiled);\r
-\r
while (reader.Peek() >= 0)\r
{\r
- line = reader.ReadLine() + separator;\r
- //Console.WriteLine("> " + line);\r
- MatchCollection mc = rCsv.Matches(line);\r
-\r
- // 現状ではNaNが含まれる行は受理しない\r
- if (!proto.Width.Includes(mc.Count)) continue;\r
-\r
- if (proto.KeyRange != null)\r
+ DataRow row = table.NewRow();\r
+ string line = reader.ReadLine();\r
+ SetRow(row, line);\r
+ if (row.HasErrors)\r
{\r
- double key = Double.Parse(mc[0].Groups[1].Value);\r
- if (!proto.KeyRange.Includes(key)) continue;\r
+ if (table.Rows.Count < 1)\r
+ {\r
+ SetHeader(table.Columns, line);\r
+ }\r
+ continue;\r
}\r
-\r
- TDataSet c = new TDataSet();\r
- foreach (Match m in mc)\r
- {\r
- c.Push(Double.Parse(m.Groups[1].Value));\r
- }\r
- records.Add(c);\r
+ table.Rows.Add(row);\r
}\r
}\r
}\r
}\r
- catch (Exception e)\r
+ catch (Exception ex)\r
+ {\r
+ throw ex;\r
+ }\r
+ return table;\r
+ }\r
+\r
+ public TDataTable ReadAll(FilePath path)\r
+ {\r
+ throw new NotImplementedException();\r
+ return null;\r
+ }\r
+\r
+ public Parser GetParser(int column)\r
+ {\r
+ return lineParser[column];\r
+ }\r
+\r
+ public Parser GetParser(string column)\r
+ {\r
+ return lineParser[prototype.Columns.IndexOf(column)];\r
+ }\r
+\r
+ public Parser GetParser(DataColumn column)\r
+ {\r
+ return lineParser[prototype.Columns.IndexOf(column.ColumnName)];\r
+ }\r
+\r
+ static public Parser GetParser(Type type)\r
+ {\r
+ return parsers[type];\r
+ }\r
+\r
+ public void SetParser(int column, Parser parser)\r
+ {\r
+ lineParser[column] = parser;\r
+ }\r
+\r
+ public void SetParser(string column, Parser parser)\r
+ {\r
+ lineParser[prototype.Columns.IndexOf(column)] = parser;\r
+ }\r
+\r
+ public void SetParser(DataColumn column, Parser parser)\r
+ { \r
+ lineParser[prototype.Columns.IndexOf(column.ColumnName)] = parser;\r
+ }\r
+\r
+ static public void SetParser(Type type, Parser parser)\r
+ {\r
+ parsers[type] = parser;\r
+ }\r
+\r
+\r
+\r
+ #endregion\r
+\r
+ #region private methods\r
+\r
+ private void SetRow(DataRow row, string line)\r
+ {\r
+ for (int i = 0; i < lineParser.Length; ++i)\r
+ {\r
+ object value;\r
+ if (lineParser[i](ref line, out value))\r
+ {\r
+ row[i] = value;\r
+ }\r
+ else\r
+ {\r
+ row.RowError = line;\r
+ }\r
+ Match m = sepPattern.Match(line);\r
+ line = line.Substring(m.Length);\r
+ }\r
+ return;\r
+ }\r
+\r
+ private void SetHeader(DataColumnCollection columns, string line)\r
+ {\r
+ for (int i = 0; i < columns.Count; ++i)\r
+ {\r
+ object value;\r
+ if (ParseString(ref line, out value))\r
+ {\r
+ string name = value as string;\r
+ Match comment = Regex.Match(name, @"^#\s*");\r
+ columns[i].Caption = name.Substring(comment.Length);\r
+ }\r
+ else\r
+ {\r
+ return;\r
+ }\r
+ Match m = sepPattern.Match(line);\r
+ line = line.Substring(m.Length);\r
+ }\r
+ }\r
+\r
+ #region parsers\r
+\r
+ static private bool ParseBoolean(ref string input, out object output)\r
+ {\r
+ Match m = Regex.Match(input, \r
+ "^" + Boolean.TrueString, RegexOptions.IgnoreCase);\r
+ if (!m.Success)\r
+ {\r
+ m = Regex.Match(input, \r
+ "^" + Boolean.FalseString, RegexOptions.IgnoreCase);\r
+ if (!m.Success)\r
+ {\r
+ output = false;\r
+ return false;\r
+ }\r
+ }\r
+ output = true;\r
+ input = input.Substring(m.Length);\r
+ return true;\r
+ }\r
+\r
+ static private bool ParseDouble(ref string input, out object output)\r
+ {\r
+ Match m = RegexSet.DecimalFloat.Match(input);\r
+ if (!m.Success)\r
+ {\r
+ output = 0;\r
+ return false;\r
+ }\r
+ output = Double.Parse(m.Groups[1].Value);\r
+ input = input.Substring(m.Length);\r
+ return true;\r
+ }\r
+\r
+ static private bool ParseSingle(ref string input, out object output)\r
+ {\r
+ Match m = RegexSet.DecimalFloat.Match(input);\r
+ if (!m.Success)\r
+ {\r
+ output = 0;\r
+ return false;\r
+ }\r
+ output = Single.Parse(m.Groups[1].Value);\r
+ input = input.Substring(m.Length);\r
+ return true;\r
+ }\r
+\r
+ static private bool ParseDecimal(ref string input, out object output)\r
+ {\r
+ Match m = RegexSet.DecimalFloat.Match(input);\r
+ if (!m.Success)\r
+ {\r
+ output = 0;\r
+ return false;\r
+ }\r
+ output = Decimal.Parse(m.Groups[1].Value);\r
+ input = input.Substring(m.Length);\r
+ return true;\r
+ }\r
+\r
+ static private bool ParseInt32(ref string input, out object output)\r
+ {\r
+ Match m = RegexSet.DecimalFloat.Match(input);\r
+ if (!m.Success)\r
+ {\r
+ output = 0;\r
+ return false;\r
+ }\r
+ output = (Int32)Double.Parse(m.Groups[1].Value);\r
+ input = input.Substring(m.Length);\r
+ return true;\r
+ }\r
+\r
+ static private bool ParseInt64(ref string input, out object output)\r
+ {\r
+ Match m = RegexSet.DecimalFloat.Match(input);\r
+ if (!m.Success)\r
+ {\r
+ output = 0;\r
+ return false;\r
+ }\r
+ output = (Int64)Double.Parse(m.Groups[1].Value);\r
+ input = input.Substring(m.Length);\r
+ return true;\r
+ }\r
+\r
+ private bool ParseDateTime(ref string input, out object output)\r
+ {\r
+ object value;\r
+ ParseString(ref input, out value);\r
+ DateTime dt;\r
+ if (DateTime.TryParse(value as string, out dt))\r
{\r
- Console.WriteLine("The file could not be read:");\r
- Console.WriteLine("> " + path);\r
- Console.WriteLine(e.Message);\r
- throw e;\r
+ output = dt;\r
+ return true;\r
}\r
+ output = DBNull.Value;\r
+ return false;\r
+ }\r
\r
- return records;\r
+ private bool ParseString(ref string input, out object output)\r
+ {\r
+ Match quotedMatch = RegexSet.QuotedString.Match(input);\r
+ if (quotedMatch.Success)\r
+ {\r
+ string content = quotedMatch.Groups[1].Value;\r
+ output = content.Replace("\"\"", "\"");\r
+ input = input.Substring(quotedMatch.Length);\r
+ return true;\r
+ }\r
+\r
+ Match m = sepPattern.Match(input);\r
+ if (m.Success)\r
+ {\r
+ output = input.Substring(0, m.Index);\r
+ input = input.Substring(m.Index);\r
+ }\r
+ else\r
+ {\r
+ output = input;\r
+ input = "";\r
+ }\r
+ return true;\r
}\r
+ #endregion\r
+\r
+ #endregion\r
}\r
}\r
OSDN Git Service
