CSV文件解析到DataTable

  1 public class CsvParsingHelper
  2     {
  3         /// <summary>
  4         /// 将csv文件的数据转成datatable
  5         /// </summary>
  6         /// <param name="csvfilePath">csv文件路径</param>
  7         /// <param name="firstIsRowHead">是否将第一行作为字段名</param>
  8         /// <returns></returns>
  9         public static DataTable CsvToDataTable(string csvfilePath, bool firstIsRowHead)
 10         {
 11             DataTable dtResult = null;
 12             if (File.Exists(csvfilePath))
 13             {
 14                 string csvstr = File.ReadAllText(csvfilePath, Encoding.Default);
 15                 if (!string.IsNullOrEmpty(csvstr))
 16                 {
 17                     dtResult = ToDataTable(csvstr, firstIsRowHead);
 18                 }
 19             }
 20             return dtResult;
 21         }
 22 
 23         /// <summary>
 24         /// 将CSV数据转换为DataTable
 25         /// </summary>
 26         /// <param name="csv">包含以","分隔的CSV数据的字符串</param>
 27         /// <param name="isRowHead">是否将第一行作为字段名</param>
 28         /// <returns></returns>
 29         private static DataTable ToDataTable(string csv, bool isRowHead)
 30         {
 31             DataTable dt = null;
 32             if (!string.IsNullOrEmpty(csv))
 33             {
 34                 dt = new DataTable();
 35                 string[] csvRows = csv.Split(new string[] { Environment.NewLine }, StringSplitOptions.RemoveEmptyEntries);
 36                 string[] csvColumns = null;
 37                 if (csvRows != null)
 38                 {
 39                     if (csvRows.Length > 0)
 40                     {
 41                         //第一行作为字段名,添加第一行记录并删除csvRows中的第一行数据
 42                         if (isRowHead)
 43                         {
 44                             csvColumns = FromCsvLine(csvRows[0]);
 45                             csvRows[0] = null;
 46                             for (int i = 0; i < csvColumns.Length; i++)
 47                             {
 48                                 dt.Columns.Add(csvColumns[i]);
 49                             }
 50                         }
 51 
 52                         for (int i = 0; i < csvRows.Length; i++)
 53                         {
 54                             if (csvRows[i] != null)
 55                             {
 56                                 csvColumns = FromCsvLine(csvRows[i]);
 57                                 //检查列数是否足够,不足则补充
 58                                 if (dt.Columns.Count < csvColumns.Length)
 59                                 {
 60                                     int columnCount = csvColumns.Length - dt.Columns.Count;
 61                                     for (int c = 0; c < columnCount; c++)
 62                                     {
 63                                         dt.Columns.Add();
 64                                     }
 65                                 }
 66                                 dt.Rows.Add(csvColumns);
 67                             }
 68                         }
 69                     }
 70                 }
 71             }
 72 
 73             return dt;
 74         }
 75         /// <summary>
 76         /// 解析一行CSV数据
 77         /// </summary>
 78         /// <param name="csv">csv数据行</param>
 79         /// <returns></returns>
 80         public static string[] FromCsvLine(string csv)
 81         {
 82             List<string> csvLiAsc = new List<string>();
 83             List<string> csvLiDesc = new List<string>();
 84 
 85             if (!string.IsNullOrEmpty(csv))
 86             {
 87                 //顺序查找
 88                 int lastIndex = 0;
 89                 int quotCount = 0;
 90                 //剩余的字符串
 91                 string lstr = string.Empty;
 92                 for (int i = 0; i < csv.Length; i++)
 93                 {
 94                     if (csv[i] == ‘"‘)
 95                     {
 96                         quotCount++;
 97                     }
 98                     else if (csv[i] == ‘,‘ && quotCount % 2 == 0)
 99                     {
100                         csvLiAsc.Add(ReplaceQuote(csv.Substring(lastIndex, i - lastIndex)));
101                         lastIndex = i + 1;
102                     }
103                     if (i == csv.Length - 1 && lastIndex < csv.Length)
104                     {
105                         lstr = csv.Substring(lastIndex, i - lastIndex + 1);
106                     }
107                 }
108                 if (!string.IsNullOrEmpty(lstr))
109                 {
110                     //倒序查找
111                     lastIndex = 0;
112                     quotCount = 0;
113                     string revStr = Reverse(lstr);
114                     for (int i = 0; i < revStr.Length; i++)
115                     {
116                         if (revStr[i] == ‘"‘)
117                         {
118                             quotCount++;
119                         }
120                         else if (revStr[i] == ‘,‘ && quotCount % 2 == 0)
121                         {
122                             csvLiDesc.Add(ReplaceQuote(Reverse(revStr.Substring(lastIndex, i - lastIndex))));
123                             lastIndex = i + 1;
124                         }
125                         if (i == revStr.Length - 1 && lastIndex < revStr.Length)
126                         {
127                             csvLiDesc.Add(ReplaceQuote(Reverse(revStr.Substring(lastIndex, i - lastIndex + 1))));
128                             lastIndex = i + 1;
129                         }
130 
131                     }
132                     string[] tmpStrs = csvLiDesc.ToArray();
133                     Array.Reverse(tmpStrs);
134                     csvLiAsc.AddRange(tmpStrs);
135                 }
136             }
137 
138             return csvLiAsc.ToArray();
139         }
140         /// <summary>
141         /// 反转字符串
142         /// </summary>
143         /// <param name="str"></param>
144         /// <returns></returns>
145         private static string Reverse(string str)
146         {
147             string revStr = string.Empty;
148             foreach (char chr in str)
149             {
150                 revStr = chr.ToString() + revStr;
151             }
152             return revStr;
153         }
154         /// <summary>
155         /// 替换CSV中的双引号转义符为正常双引号,并去掉左右双引号
156         /// </summary>
157         /// <param name="csvValue">csv格式的数据</param>
158         /// <returns></returns>
159         private static string ReplaceQuote(string csvValue)
160         {
161             string rtnStr = csvValue;
162             if (!string.IsNullOrEmpty(csvValue))
163             {
164                 //首尾都是"
165                 Match m = Regex.Match(csvValue, "^\"(.*?)\"$");
166                 if (m.Success)
167                 {
168                     rtnStr = m.Result("${1}").Replace("\"\"", "\"");
169                 }
170                 else
171                 {
172                     rtnStr = rtnStr.Replace("\"\"", "\"");
173                 }
174             }
175             return rtnStr;
176 
177         }
178     }