CsvStreamReader.cs 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548
  1. using System;
  2. using System.Collections;
  3. using System.Collections.Generic;
  4. using System.Data;
  5. using System.IO;
  6. using System.Linq;
  7. using System.Text;
  8. using System.Threading.Tasks;
  9. namespace ToolsClassLibrary
  10. {
  11. /// <summary>
  12. /// CSV工具类
  13. /// </summary>
  14. public class CsvStreamReader
  15. {
  16. /// <summary>
  17. /// 行链表,CSV文件的每一行就是一个链
  18. /// </summary>
  19. private ArrayList rowAL; //行链表,CSV文件的每一行就是一个链
  20. /// <summary>
  21. /// 文件名
  22. /// </summary>
  23. private string fileName; //文件名
  24. /// <summary>
  25. /// 编码
  26. /// </summary>
  27. private Encoding encoding; //编码
  28. public CsvStreamReader()
  29. {
  30. this.rowAL = new ArrayList();
  31. this.fileName = "";
  32. this.encoding = Encoding.Default;
  33. }
  34. /// <summary>
  35. ///
  36. /// </summary>
  37. /// <param name="fileName">文件名,包括文件路径</param>
  38. public CsvStreamReader(string fileName)
  39. {
  40. this.rowAL = new ArrayList();
  41. this.fileName = fileName;
  42. this.encoding = Encoding.Default;
  43. LoadCsvFile();
  44. }
  45. /// <summary>
  46. ///
  47. /// </summary>
  48. /// <param name="fileName">文件名,包括文件路径</param>
  49. /// <param name="encoding">文件编码</param>
  50. public CsvStreamReader(string fileName, Encoding encoding)
  51. {
  52. this.rowAL = new ArrayList();
  53. this.fileName = fileName;
  54. this.encoding = encoding;
  55. LoadCsvFile();
  56. }
  57. /// <summary>
  58. /// 文件名,包括文件路径
  59. /// </summary>
  60. public string FileName
  61. {
  62. set
  63. {
  64. this.fileName = value;
  65. LoadCsvFile();
  66. }
  67. }
  68. /// <summary>
  69. /// 文件编码
  70. /// </summary>
  71. public Encoding FileEncoding
  72. {
  73. set
  74. {
  75. this.encoding = value;
  76. }
  77. }
  78. /// <summary>
  79. /// 获取行数
  80. /// </summary>
  81. public int RowCount
  82. {
  83. get
  84. {
  85. return this.rowAL.Count;
  86. }
  87. }
  88. /// <summary>
  89. /// 获取列数
  90. /// </summary>
  91. public int ColCount
  92. {
  93. get
  94. {
  95. int maxCol;
  96. maxCol = 0;
  97. for (int i = 0; i < this.rowAL.Count; i++)
  98. {
  99. ArrayList colAL = (ArrayList)this.rowAL[i];
  100. maxCol = (maxCol > colAL.Count) ? maxCol : colAL.Count;
  101. }
  102. return maxCol;
  103. }
  104. }
  105. /// <summary>
  106. /// 获取某行某列的数据
  107. /// row:行,row = 1代表第一行
  108. /// col:列,col = 1代表第一列
  109. /// </summary>
  110. public string this[int row, int col]
  111. {
  112. get
  113. {
  114. //数据有效性验证
  115. CheckRowValid(row);
  116. CheckColValid(col);
  117. ArrayList colAL = (ArrayList)this.rowAL[row - 1];
  118. //如果请求列数据大于当前行的列时,返回空值
  119. if (colAL.Count < col)
  120. {
  121. return "";
  122. }
  123. return colAL[col - 1].ToString();
  124. }
  125. }
  126. /// <summary>
  127. /// 根据最小行,最大行,最小列,最大列,来生成一个DataTable类型的数据
  128. /// 行等于1代表第一行
  129. /// 列等于1代表第一列
  130. /// maxrow: -1代表最大行
  131. /// maxcol: -1代表最大列
  132. /// </summary>
  133. public DataTable this[int minRow, int maxRow, int minCol, int maxCol]
  134. {
  135. get
  136. {
  137. //数据有效性验证
  138. CheckRowValid(minRow);
  139. CheckMaxRowValid(maxRow);
  140. CheckColValid(minCol);
  141. CheckMaxColValid(maxCol);
  142. if (maxRow == -1)
  143. {
  144. maxRow = RowCount;
  145. }
  146. if (maxCol == -1)
  147. {
  148. maxCol = ColCount;
  149. }
  150. if (maxRow < minRow)
  151. {
  152. throw new Exception("最大行数不能小于最小行数");
  153. }
  154. if (maxCol < minCol)
  155. {
  156. throw new Exception("最大列数不能小于最小列数");
  157. }
  158. DataTable csvDT = new DataTable();
  159. int i;
  160. int col;
  161. int row;
  162. //增加列
  163. for (i = minCol; i <= maxCol; i++)
  164. {
  165. csvDT.Columns.Add(i.ToString());
  166. }
  167. for (row = minRow; row <= maxRow; row++)
  168. {
  169. DataRow csvDR = csvDT.NewRow();
  170. i = 0;
  171. for (col = minCol; col <= maxCol; col++)
  172. {
  173. csvDR[i] = this[row, col];
  174. i++;
  175. }
  176. csvDT.Rows.Add(csvDR);
  177. }
  178. return csvDT;
  179. }
  180. }
  181. /// <summary>
  182. /// 检查行数是否是有效的
  183. /// </summary>
  184. /// <param name="col"></param>
  185. private void CheckRowValid(int row)
  186. {
  187. if (row <= 0)
  188. {
  189. throw new Exception("行数不能小于0");
  190. }
  191. if (row > RowCount)
  192. {
  193. throw new Exception("没有当前行的数据");
  194. }
  195. }
  196. /// <summary>
  197. /// 检查最大行数是否是有效的
  198. /// </summary>
  199. /// <param name="col"></param>
  200. private void CheckMaxRowValid(int maxRow)
  201. {
  202. if (maxRow <= 0 && maxRow != -1)
  203. {
  204. throw new Exception("行数不能等于0或小于-1");
  205. }
  206. if (maxRow > RowCount)
  207. {
  208. throw new Exception("没有当前行的数据");
  209. }
  210. }
  211. /// <summary>
  212. /// 检查列数是否是有效的
  213. /// </summary>
  214. /// <param name="col"></param>
  215. private void CheckColValid(int col)
  216. {
  217. if (col <= 0)
  218. {
  219. throw new Exception("列数不能小于0");
  220. }
  221. if (col > ColCount)
  222. {
  223. throw new Exception("没有当前列的数据");
  224. }
  225. }
  226. /// <summary>
  227. /// 检查检查最大列数是否是有效的
  228. /// </summary>
  229. /// <param name="col"></param>
  230. private void CheckMaxColValid(int maxCol)
  231. {
  232. if (maxCol <= 0 && maxCol != -1)
  233. {
  234. throw new Exception("列数不能等于0或小于-1");
  235. }
  236. if (maxCol > ColCount)
  237. {
  238. throw new Exception("没有当前列的数据");
  239. }
  240. }
  241. /// <summary>
  242. /// 载入CSV文件
  243. /// </summary>
  244. private void LoadCsvFile()
  245. {
  246. //对数据的有效性进行验证
  247. if (this.fileName == null)
  248. {
  249. throw new Exception("请指定要载入的CSV文件名");
  250. }
  251. else if (!File.Exists(this.fileName))
  252. {
  253. throw new Exception("指定的CSV文件不存在");
  254. }
  255. else
  256. {
  257. }
  258. if (this.encoding == null)
  259. {
  260. this.encoding = Encoding.Default;
  261. }
  262. StreamReader sr = new StreamReader(this.fileName, this.encoding);
  263. string csvDataLine;
  264. csvDataLine = "";
  265. while (true)
  266. {
  267. string fileDataLine;
  268. fileDataLine = sr.ReadLine();
  269. if (fileDataLine == null)
  270. {
  271. break;
  272. }
  273. if (csvDataLine == "")
  274. {
  275. csvDataLine = fileDataLine;//GetDeleteQuotaDataLine(fileDataLine);
  276. }
  277. else
  278. {
  279. csvDataLine += "\r\n" + fileDataLine;//GetDeleteQuotaDataLine(fileDataLine);
  280. }
  281. //如果包含偶数个引号,说明该行数据中出现回车符或包含逗号
  282. if (!IfOddQuota(csvDataLine))
  283. {
  284. AddNewDataLine(csvDataLine);
  285. csvDataLine = "";
  286. }
  287. }
  288. sr.Close();
  289. //数据行出现奇数个引号
  290. if (csvDataLine.Length > 0)
  291. {
  292. throw new Exception("CSV文件的格式有错误");
  293. }
  294. }
  295. /// <summary>
  296. /// 获取两个连续引号变成单个引号的数据行
  297. /// </summary>
  298. /// <param name="fileDataLine">文件数据行</param>
  299. /// <returns></returns>
  300. private string GetDeleteQuotaDataLine(string fileDataLine)
  301. {
  302. return fileDataLine.Replace("\"\"", "\"");
  303. }
  304. /// <summary>
  305. /// 判断字符串是否包含奇数个引号
  306. /// </summary>
  307. /// <param name="dataLine">数据行</param>
  308. /// <returns>为奇数时,返回为真;否则返回为假</returns>
  309. private bool IfOddQuota(string dataLine)
  310. {
  311. int quotaCount;
  312. bool oddQuota;
  313. quotaCount = 0;
  314. for (int i = 0; i < dataLine.Length; i++)
  315. {
  316. if (dataLine[i] == '\"')
  317. {
  318. quotaCount++;
  319. }
  320. }
  321. oddQuota = false;
  322. if (quotaCount % 2 == 1)
  323. {
  324. oddQuota = true;
  325. }
  326. return oddQuota;
  327. }
  328. /// <summary>
  329. /// 判断是否以奇数个引号开始
  330. /// </summary>
  331. /// <param name="dataCell"></param>
  332. /// <returns></returns>
  333. private bool IfOddStartQuota(string dataCell)
  334. {
  335. int quotaCount;
  336. bool oddQuota;
  337. quotaCount = 0;
  338. for (int i = 0; i < dataCell.Length; i++)
  339. {
  340. if (dataCell[i] == '\"')
  341. {
  342. quotaCount++;
  343. }
  344. else
  345. {
  346. break;
  347. }
  348. }
  349. oddQuota = false;
  350. if (quotaCount % 2 == 1)
  351. {
  352. oddQuota = true;
  353. }
  354. return oddQuota;
  355. }
  356. /// <summary>
  357. /// 判断是否以奇数个引号结尾
  358. /// </summary>
  359. /// <param name="dataCell"></param>
  360. /// <returns></returns>
  361. private bool IfOddEndQuota(string dataCell)
  362. {
  363. int quotaCount;
  364. bool oddQuota;
  365. quotaCount = 0;
  366. for (int i = dataCell.Length - 1; i >= 0; i--)
  367. {
  368. if (dataCell[i] == '\"')
  369. {
  370. quotaCount++;
  371. }
  372. else
  373. {
  374. break;
  375. }
  376. }
  377. oddQuota = false;
  378. if (quotaCount % 2 == 1)
  379. {
  380. oddQuota = true;
  381. }
  382. return oddQuota;
  383. }
  384. /// <summary>
  385. /// 加入新的数据行
  386. /// </summary>
  387. /// <param name="newDataLine">新的数据行</param>
  388. private void AddNewDataLine(string newDataLine)
  389. {
  390. //System.Diagnostics.Debug.WriteLine("NewLine:" + newDataLine);
  391. ////return;
  392. ArrayList colAL = new ArrayList();
  393. string[] dataArray = newDataLine.Split(',');
  394. bool oddStartQuota; //是否以奇数个引号开始
  395. string cellData;
  396. oddStartQuota = false;
  397. cellData = "";
  398. for (int i = 0; i < dataArray.Length; i++)
  399. {
  400. if (oddStartQuota)
  401. {
  402. //因为前面用逗号分割,所以要加上逗号
  403. cellData += "," + dataArray[i];
  404. //是否以奇数个引号结尾
  405. if (IfOddEndQuota(dataArray[i]))
  406. {
  407. colAL.Add(GetHandleData(cellData));
  408. oddStartQuota = false;
  409. continue;
  410. }
  411. }
  412. else
  413. {
  414. //是否以奇数个引号开始
  415. if (IfOddStartQuota(dataArray[i]))
  416. {
  417. //是否以奇数个引号结尾,不能是一个双引号,并且不是奇数个引号
  418. if (IfOddEndQuota(dataArray[i]) && dataArray[i].Length > 2 && !IfOddQuota(dataArray[i]))
  419. {
  420. colAL.Add(GetHandleData(dataArray[i]));
  421. oddStartQuota = false;
  422. continue;
  423. }
  424. else
  425. {
  426. oddStartQuota = true;
  427. cellData = dataArray[i];
  428. continue;
  429. }
  430. }
  431. else
  432. {
  433. colAL.Add(GetHandleData(dataArray[i]));
  434. }
  435. }
  436. }
  437. if (oddStartQuota)
  438. {
  439. throw new Exception("数据格式有问题");
  440. }
  441. this.rowAL.Add(colAL);
  442. }
  443. /// <summary>
  444. /// 去掉格子的首尾引号,把双引号变成单引号
  445. /// </summary>
  446. /// <param name="fileCellData"></param>
  447. /// <returns></returns>
  448. private string GetHandleData(string fileCellData)
  449. {
  450. if (fileCellData == "")
  451. {
  452. return "";
  453. }
  454. if (IfOddStartQuota(fileCellData))
  455. {
  456. if (IfOddEndQuota(fileCellData))
  457. {
  458. return fileCellData.Substring(1, fileCellData.Length - 2).Replace("\"\"", "\""); //去掉首尾引号,然后把双引号变成单引号
  459. }
  460. else
  461. {
  462. throw new Exception("数据引号无法匹配" + fileCellData);
  463. }
  464. }
  465. else
  466. {
  467. //考虑形如"" """" """"""
  468. if (fileCellData.Length > 2 && fileCellData[0] == '\"')
  469. {
  470. fileCellData = fileCellData.Substring(1, fileCellData.Length - 2).Replace("\"\"", "\""); //去掉首尾引号,然后把双引号变成单引号
  471. }
  472. }
  473. return fileCellData;
  474. }
  475. }
  476. }