Free cookie consent management tool by TermsFeed Policy Generator

source: trunk/sources/HeuristicLab.Problems.Instances.DataAnalysis/3.3/TableFileParser.cs @ 14791

Last change on this file since 14791 was 14408, checked in by mkommend, 8 years ago

#2661: Fixed a bug in the TableFileparser that omits the correct detection of DateTime columns.

File size: 26.9 KB
RevLine 
[7849]1#region License Information
2/* HeuristicLab
[14185]3 * Copyright (C) 2002-2016 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
[7849]4 *
5 * This file is part of HeuristicLab.
6 *
7 * HeuristicLab is free software: you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation, either version 3 of the License, or
10 * (at your option) any later version.
11 *
12 * HeuristicLab is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
15 * GNU General Public License for more details.
16 *
17 * You should have received a copy of the GNU General Public License
18 * along with HeuristicLab. If not, see <http://www.gnu.org/licenses/>.
19 */
20#endregion
21
22
23using System;
24using System.Collections;
25using System.Collections.Generic;
[13526]26using System.Diagnostics.Contracts;
[7849]27using System.Globalization;
28using System.IO;
29using System.Linq;
[13440]30using System.Text;
[7849]31
32namespace HeuristicLab.Problems.Instances.DataAnalysis {
[13414]33  public class TableFileParser : Progress<long> { // reports the number of bytes read
[8564]34    private const int BUFFER_SIZE = 65536;
[9652]35    // char used to symbolize whitespaces (no missing values can be handled with whitespaces)
36    private const char WHITESPACECHAR = (char)0;
37    private static readonly char[] POSSIBLE_SEPARATORS = new char[] { ',', ';', '\t', WHITESPACECHAR };
[7849]38    private Tokenizer tokenizer;
[13440]39    private int estimatedNumberOfLines = 200; // initial capacity for columns, will be set automatically when data is read from a file
[7849]40
[13584]41
42    private Encoding encoding = Encoding.Default;
43
44    public Encoding Encoding {
45      get { return encoding; }
46      set {
47        if (value == null) throw new ArgumentNullException("Encoding");
48        encoding = value;
49      }
50    }
51
52
[7849]53    private int rows;
54    public int Rows {
55      get { return rows; }
56      set { rows = value; }
57    }
58
59    private int columns;
60    public int Columns {
61      get { return columns; }
62      set { columns = value; }
63    }
64
65    private List<IList> values;
66    public List<IList> Values {
67      get {
68        return values;
69      }
70    }
71
72    private List<string> variableNames;
73    public IEnumerable<string> VariableNames {
74      get {
75        if (variableNames.Count > 0) return variableNames;
76        else {
77          string[] names = new string[columns];
78          for (int i = 0; i < names.Length; i++) {
79            names[i] = "X" + i.ToString("000");
80          }
81          return names;
82        }
83      }
84    }
85
86    public TableFileParser() {
87      variableNames = new List<string>();
88    }
89
[9608]90    public bool AreColumnNamesInFirstLine(string fileName) {
91      NumberFormatInfo numberFormat;
92      DateTimeFormatInfo dateTimeFormatInfo;
93      char separator;
94      DetermineFileFormat(fileName, out numberFormat, out dateTimeFormatInfo, out separator);
95      using (var stream = new FileStream(fileName, FileMode.Open, FileAccess.Read, FileShare.ReadWrite)) {
96        return AreColumnNamesInFirstLine(stream, numberFormat, dateTimeFormatInfo, separator);
97      }
98    }
99
100    public bool AreColumnNamesInFirstLine(Stream stream) {
101      NumberFormatInfo numberFormat = NumberFormatInfo.InvariantInfo;
102      DateTimeFormatInfo dateTimeFormatInfo = DateTimeFormatInfo.InvariantInfo;
103      char separator = ',';
104      return AreColumnNamesInFirstLine(stream, numberFormat, dateTimeFormatInfo, separator);
105    }
106
107    public bool AreColumnNamesInFirstLine(string fileName, NumberFormatInfo numberFormat,
108                                         DateTimeFormatInfo dateTimeFormatInfo, char separator) {
109      using (var stream = new FileStream(fileName, FileMode.Open, FileAccess.Read, FileShare.ReadWrite)) {
110        return AreColumnNamesInFirstLine(stream, numberFormat, dateTimeFormatInfo, separator);
111      }
112    }
113
114    public bool AreColumnNamesInFirstLine(Stream stream, NumberFormatInfo numberFormat,
115                                          DateTimeFormatInfo dateTimeFormatInfo, char separator) {
[13584]116      using (StreamReader reader = new StreamReader(stream, Encoding)) {
[9608]117        tokenizer = new Tokenizer(reader, numberFormat, dateTimeFormatInfo, separator);
[13440]118        return (tokenizer.PeekType() != TokenTypeEnum.Double);
[9608]119      }
120    }
121
[7851]122    /// <summary>
123    /// Parses a file and determines the format first
124    /// </summary>
125    /// <param name="fileName">file which is parsed</param>
[9608]126    /// <param name="columnNamesInFirstLine"></param>
[13413]127    public void Parse(string fileName, bool columnNamesInFirstLine, int lineLimit = -1) {
[7851]128      NumberFormatInfo numberFormat;
129      DateTimeFormatInfo dateTimeFormatInfo;
130      char separator;
[9608]131      DetermineFileFormat(fileName, out numberFormat, out dateTimeFormatInfo, out separator);
[13440]132      EstimateNumberOfLines(fileName);
[13413]133      Parse(new FileStream(fileName, FileMode.Open, FileAccess.Read, FileShare.ReadWrite), numberFormat, dateTimeFormatInfo, separator, columnNamesInFirstLine, lineLimit);
[7851]134    }
135
136    /// <summary>
137    /// Parses a file with the given formats
138    /// </summary>
139    /// <param name="fileName">file which is parsed</param>
140    /// <param name="numberFormat">Format of numbers</param>
141    /// <param name="dateTimeFormatInfo">Format of datetime</param>
142    /// <param name="separator">defines the separator</param>
[9608]143    /// <param name="columnNamesInFirstLine"></param>
[13413]144    public void Parse(string fileName, NumberFormatInfo numberFormat, DateTimeFormatInfo dateTimeFormatInfo, char separator, bool columnNamesInFirstLine, int lineLimit = -1) {
[13440]145      EstimateNumberOfLines(fileName);
[9608]146      using (var stream = new FileStream(fileName, FileMode.Open, FileAccess.Read, FileShare.ReadWrite)) {
[13413]147        Parse(stream, numberFormat, dateTimeFormatInfo, separator, columnNamesInFirstLine, lineLimit);
[9608]148      }
[7849]149    }
150
[13440]151    // determines the number of newline characters in the first 64KB to guess the number of rows for a file
152    private void EstimateNumberOfLines(string fileName) {
153      var len = new System.IO.FileInfo(fileName).Length;
[13526]154      var buf = new char[1024 * 1024];
[13584]155      using (var reader = new StreamReader(fileName, Encoding)) {
[13445]156        reader.ReadBlock(buf, 0, buf.Length);
157      }
[13440]158      int numNewLine = 0;
[13442]159      int charsInCurrentLine = 0, charsInFirstLine = 0; // the first line (names) and the last line (incomplete) are not representative
160      foreach (var ch in buf) {
161        charsInCurrentLine++;
162        if (ch == '\n') {
163          if (numNewLine == 0) charsInFirstLine = charsInCurrentLine; // store the number of chars in the first line
164          charsInCurrentLine = 0;
165          numNewLine++;
166        }
167      }
168      if (numNewLine <= 1) {
[13440]169        // fail -> keep the default setting
170        return;
171      } else {
[13442]172        double charsPerLineFactor = (buf.Length - charsInFirstLine - charsInCurrentLine) / ((double)numNewLine - 1);
[13440]173        double estimatedLines = len / charsPerLineFactor;
174        estimatedNumberOfLines = (int)Math.Round(estimatedLines * 1.1); // pessimistic allocation of 110% to make sure that the list is very likely large enough
175      }
176    }
177
[7851]178    /// <summary>
179    /// Takes a Stream and parses it with default format. NumberFormatInfo.InvariantInfo, DateTimeFormatInfo.InvariantInfo and separator = ','
180    /// </summary>
181    /// <param name="stream">stream which is parsed</param>
[9608]182    /// <param name="columnNamesInFirstLine"></param>
[13413]183    public void Parse(Stream stream, bool columnNamesInFirstLine, int lineLimit = -1) {
[7851]184      NumberFormatInfo numberFormat = NumberFormatInfo.InvariantInfo;
185      DateTimeFormatInfo dateTimeFormatInfo = DateTimeFormatInfo.InvariantInfo;
186      char separator = ',';
[13413]187      Parse(stream, numberFormat, dateTimeFormatInfo, separator, columnNamesInFirstLine, lineLimit);
[7851]188    }
189
190    /// <summary>
191    /// Parses a stream with the given formats.
192    /// </summary>
193    /// <param name="stream">Stream which is parsed</param>   
194    /// <param name="numberFormat">Format of numbers</param>
195    /// <param name="dateTimeFormatInfo">Format of datetime</param>
196    /// <param name="separator">defines the separator</param>
[9608]197    /// <param name="columnNamesInFirstLine"></param>
[13413]198    public void Parse(Stream stream, NumberFormatInfo numberFormat, DateTimeFormatInfo dateTimeFormatInfo, char separator, bool columnNamesInFirstLine, int lineLimit = -1) {
[14296]199      if (lineLimit > 0) estimatedNumberOfLines = lineLimit;
200
201      using (var reader = new StreamReader(stream)) {
[7849]202        tokenizer = new Tokenizer(reader, numberFormat, dateTimeFormatInfo, separator);
[14296]203        var strValues = new List<List<string>>();
[13440]204        values = new List<IList>();
[14296]205        Prepare(columnNamesInFirstLine, strValues);
[13447]206
207        int nLinesParsed = 0;
208        int colIdx = 0;
209        while (tokenizer.HasNext() && (lineLimit < 0 || nLinesParsed < lineLimit)) {
210          if (tokenizer.PeekType() == TokenTypeEnum.NewLine) {
211            tokenizer.Skip();
212
213            // all rows have to have the same number of values
[14296]214            // the first row defines how many elements are needed
215            if (colIdx > 0 && values.Count != colIdx) {
216              // read at least one value in the row (support for skipping empty lines)
217              Error("The first row of the dataset has " + values.Count + " columns." + Environment.NewLine +
[13447]218                    "Line " + tokenizer.CurrentLineNumber + " has " + colIdx + " columns.", "",
[14296]219                tokenizer.CurrentLineNumber);
[13440]220            }
[13447]221            OnReport(tokenizer.BytesRead);
222
223            nLinesParsed++;
224            colIdx = 0;
[13440]225          } else {
[13447]226            // read one value
[14296]227            TokenTypeEnum type;
228            string strVal;
229            double dblVal;
230            DateTime dateTimeVal;
[13447]231            tokenizer.Next(out type, out strVal, out dblVal, out dateTimeVal);
232
[14296]233            if (colIdx == values.Count) {
234              Error("The first row of the dataset has " + values.Count + " columns." + Environment.NewLine +
[13447]235                    "Line " + tokenizer.CurrentLineNumber + " has more columns.", "",
236                tokenizer.CurrentLineNumber);
[13440]237            }
[13526]238            if (!IsColumnTypeCompatible(values[colIdx], type)) {
[14296]239              values[colIdx] = strValues[colIdx];
[13526]240            }
[14296]241
[13447]242            // add the value to the column
[14296]243            AddValue(type, values[colIdx], strVal, dblVal, dateTimeVal);
244            if (!(values[colIdx] is List<string>)) { // optimization: don't store the string values in another list if the column is list<string>
245              strValues[colIdx].Add(strVal);
246            }
247            colIdx++;
[13440]248          }
[7849]249        }
250      }
251
[14296]252      if (!values.Any() || values.First().Count == 0)
253        Error("Couldn't parse data values. Probably because of incorrect number format " +
254              "(the parser expects english number format with a '.' as decimal separator).", "", tokenizer.CurrentLineNumber);
255
[13441]256      this.rows = values.First().Count;
[13447]257      this.columns = values.Count;
[13441]258
[13440]259      // after everything has been parsed make sure the lists are as compact as possible
260      foreach (var l in values) {
261        var dblList = l as List<double>;
262        var byteList = l as List<byte>;
263        var dateList = l as List<DateTime>;
264        var stringList = l as List<string>;
265        var objList = l as List<object>;
266        if (dblList != null) dblList.TrimExcess();
267        if (byteList != null) byteList.TrimExcess();
268        if (dateList != null) dateList.TrimExcess();
269        if (stringList != null) stringList.TrimExcess();
270        if (objList != null) objList.TrimExcess();
[7849]271      }
[13442]272
273      // for large files we created a lot of memory pressure, cannot hurt to run GC.Collect here (TableFileParser is called seldomly on user interaction)
274      GC.Collect(2, GCCollectionMode.Forced);
[7849]275    }
276
[14296]277    private void Prepare(bool columnNamesInFirstLine, List<List<string>> strValues) {
278      if (columnNamesInFirstLine) {
279        ParseVariableNames();
280        if (!tokenizer.HasNext())
281          Error(
282            "Couldn't parse data values. Probably because of incorrect number format (the parser expects english number format with a '.' as decimal separator).",
283            "", tokenizer.CurrentLineNumber);
284      }
285      // read first line to determine types and allocate specific lists
286      // read values... start in first row
287      int colIdx = 0;
288      while (tokenizer.PeekType() != TokenTypeEnum.NewLine) {
289        // read one value
290        TokenTypeEnum type; string strVal; double dblVal; DateTime dateTimeVal;
291        tokenizer.Next(out type, out strVal, out dblVal, out dateTimeVal);
292
293        // initialize column
294        values.Add(CreateList(type, estimatedNumberOfLines));
295        if (type == TokenTypeEnum.String)
296          strValues.Add(new List<string>(0)); // optimization: don't store the string values in another list if the column is list<string>
297        else
298          strValues.Add(new List<string>(estimatedNumberOfLines));
299
300        AddValue(type, values[colIdx], strVal, dblVal, dateTimeVal);
301        if (type != TokenTypeEnum.String)
302          strValues[colIdx].Add(strVal);
303        colIdx++;
304      }
305      tokenizer.Skip(); // skip newline
306    }
307
[13447]308    #region type-dependent dispatch
[13526]309    private bool IsColumnTypeCompatible(IList list, TokenTypeEnum tokenType) {
310      return (list is List<string>) || // all tokens can be added to a string list
311             (tokenType == TokenTypeEnum.Missing) || // empty entries are allowed in all columns
312             (tokenType == TokenTypeEnum.Double && list is List<double>) ||
313             (tokenType == TokenTypeEnum.DateTime && list is List<DateTime>);
314    }
315
316    // all columns are converted to string columns when we find an non-empty value that has incorrect type
317    private IList ConvertToStringColumn(IList list) {
318      var dblL = list as List<double>;
319      if (dblL != null) {
320        var l = new List<string>(dblL.Capacity);
321        l.AddRange(dblL.Select(dbl => dbl.ToString()));
322        return l;
[13447]323      }
[13526]324
325      var dtL = list as List<DateTime>;
326      if (dtL != null) {
327        var l = new List<string>(dtL.Capacity);
328        l.AddRange(dtL.Select(dbl => dbl.ToString()));
329        return l;
330      }
331
332      if (list is List<string>) return list;
333
334      throw new InvalidProgramException(string.Format("Cannot convert column of type {0} to string column", list.GetType()));
[13447]335    }
336
[13526]337    private void AddValue(TokenTypeEnum type, IList list, string strVal, double dblVal, DateTime dateTimeVal) {
[13447]338      var dblList = list as List<double>;
[13526]339      if (dblList != null) {
340        AddValue(type, dblList, dblVal);
341        return;
[13447]342      }
343
344      var strList = list as List<string>;
[13526]345      if (strList != null) {
346        AddValue(type, strList, strVal);
347        return;
[13447]348      }
[13526]349      var dtList = list as List<DateTime>;
350      if (dtList != null) {
351        AddValue(type, dtList, dateTimeVal);
352        return;
353      }
354
355      list.Add(strVal); // assumes List<object>
[13447]356    }
357
[13526]358    private void AddValue(TokenTypeEnum type, List<double> list, double dblVal) {
359      Contract.Assert(type == TokenTypeEnum.Missing || type == TokenTypeEnum.Double);
360      list.Add(type == TokenTypeEnum.Missing ? double.NaN : dblVal);
[13447]361    }
362
[13526]363    private void AddValue(TokenTypeEnum type, List<string> list, string strVal) {
364      // assumes that strVal is always set to the original token read from the input file
365      list.Add(type == TokenTypeEnum.Missing ? string.Empty : strVal);
366    }
367
368    private void AddValue(TokenTypeEnum type, List<DateTime> list, DateTime dtVal) {
369      Contract.Assert(type == TokenTypeEnum.Missing || type == TokenTypeEnum.DateTime);
370      list.Add(type == TokenTypeEnum.Missing ? DateTime.MinValue : dtVal);
371    }
372
[13447]373    private IList CreateList(TokenTypeEnum type, int estimatedNumberOfLines) {
374      switch (type) {
375        case TokenTypeEnum.String:
376          return new List<string>(estimatedNumberOfLines);
377        case TokenTypeEnum.Double:
[13526]378        case TokenTypeEnum.Missing: // assume double columns
[13447]379          return new List<double>(estimatedNumberOfLines);
380        case TokenTypeEnum.DateTime:
381          return new List<DateTime>(estimatedNumberOfLines);
382        default:
383          throw new InvalidOperationException();
384      }
385    }
386    #endregion
387
[7849]388    public static void DetermineFileFormat(string path, out NumberFormatInfo numberFormat, out DateTimeFormatInfo dateTimeFormatInfo, out char separator) {
[8885]389      DetermineFileFormat(new FileStream(path, FileMode.Open, FileAccess.Read, FileShare.ReadWrite), out numberFormat, out dateTimeFormatInfo, out separator);
[7849]390    }
391
392    public static void DetermineFileFormat(Stream stream, out NumberFormatInfo numberFormat, out DateTimeFormatInfo dateTimeFormatInfo, out char separator) {
393      using (StreamReader reader = new StreamReader(stream)) {
394        // skip first line
395        reader.ReadLine();
396        // read a block
397        char[] buffer = new char[BUFFER_SIZE];
398        int charsRead = reader.ReadBlock(buffer, 0, BUFFER_SIZE);
399        // count frequency of special characters
400        Dictionary<char, int> charCounts = buffer.Take(charsRead)
401          .GroupBy(c => c)
402          .ToDictionary(g => g.Key, g => g.Count());
403
404        // depending on the characters occuring in the block
405        // we distinghish a number of different cases based on the the following rules:
406        // many points => it must be English number format, the other frequently occuring char is the separator
407        // no points but many commas => this is the problematic case. Either German format (real numbers) or English format (only integer numbers) with ',' as separator
408        //   => check the line in more detail:
409        //            English: 0, 0, 0, 0
410        //            German:  0,0 0,0 0,0 ...
411        //            => if commas are followed by space => English format
412        // no points no commas => English format (only integer numbers) use the other frequently occuring char as separator
413        // in all cases only treat ' ' as separator if no other separator is possible (spaces can also occur additionally to separators)
414        if (OccurrencesOf(charCounts, '.') > 10) {
415          numberFormat = NumberFormatInfo.InvariantInfo;
416          dateTimeFormatInfo = DateTimeFormatInfo.InvariantInfo;
417          separator = POSSIBLE_SEPARATORS
418            .Where(c => OccurrencesOf(charCounts, c) > 10)
419            .OrderBy(c => -OccurrencesOf(charCounts, c))
420            .DefaultIfEmpty(' ')
421            .First();
422        } else if (OccurrencesOf(charCounts, ',') > 10) {
423          // no points and many commas
424          // count the number of tokens (chains of only digits and commas) that contain multiple comma characters
425          int tokensWithMultipleCommas = 0;
426          for (int i = 0; i < charsRead; i++) {
427            int nCommas = 0;
428            while (i < charsRead && (buffer[i] == ',' || Char.IsDigit(buffer[i]))) {
429              if (buffer[i] == ',') nCommas++;
430              i++;
431            }
432            if (nCommas > 2) tokensWithMultipleCommas++;
433          }
434          if (tokensWithMultipleCommas > 1) {
435            // English format (only integer values) with ',' as separator
436            numberFormat = NumberFormatInfo.InvariantInfo;
437            dateTimeFormatInfo = DateTimeFormatInfo.InvariantInfo;
438            separator = ',';
439          } else {
[13526]440            char[] disallowedSeparators = new char[] { ',' }; // n. def. contains a space so ' ' should be disallowed to, however existing unit tests would fail
[7849]441            // German format (real values)
442            numberFormat = NumberFormatInfo.GetInstance(new CultureInfo("de-DE"));
443            dateTimeFormatInfo = DateTimeFormatInfo.GetInstance(new CultureInfo("de-DE"));
444            separator = POSSIBLE_SEPARATORS
445              .Except(disallowedSeparators)
446              .Where(c => OccurrencesOf(charCounts, c) > 10)
447              .OrderBy(c => -OccurrencesOf(charCounts, c))
[13584]448              .DefaultIfEmpty(' ')
[7849]449              .First();
450          }
451        } else {
452          // no points and no commas => English format
453          numberFormat = NumberFormatInfo.InvariantInfo;
454          dateTimeFormatInfo = DateTimeFormatInfo.InvariantInfo;
455          separator = POSSIBLE_SEPARATORS
456            .Where(c => OccurrencesOf(charCounts, c) > 10)
457            .OrderBy(c => -OccurrencesOf(charCounts, c))
458            .DefaultIfEmpty(' ')
459            .First();
460        }
461      }
462    }
463
464    private static int OccurrencesOf(Dictionary<char, int> charCounts, char c) {
465      return charCounts.ContainsKey(c) ? charCounts[c] : 0;
466    }
467
468    #region tokenizer
[13447]469    // the tokenizer reads full lines and returns separated tokens in the line as well as a terminating end-of-line character
[7849]470    internal enum TokenTypeEnum {
[13526]471      NewLine, String, Double, DateTime, Missing
[7849]472    }
473
474    internal class Tokenizer {
475      private StreamReader reader;
[13411]476      // we assume that a buffer of 1024 tokens for a line is sufficient most of the time (the buffer is increased below if necessary)
477      private TokenTypeEnum[] tokenTypes = new TokenTypeEnum[1024];
478      private string[] stringVals = new string[1024];
479      private double[] doubleVals = new double[1024];
480      private DateTime[] dateTimeVals = new DateTime[1024];
481      private int tokenPos;
482      private int numTokens;
[7849]483      private NumberFormatInfo numberFormatInfo;
484      private DateTimeFormatInfo dateTimeFormatInfo;
485      private char separator;
486
[13447]487      // arrays for string.Split()
488      private readonly char[] whiteSpaceSeparators = new char[0]; // string split uses separators as default
489      private readonly char[] separators;
490
[7849]491      private int currentLineNumber = 0;
492      public int CurrentLineNumber {
493        get { return currentLineNumber; }
494        private set { currentLineNumber = value; }
495      }
496      private string currentLine;
497      public string CurrentLine {
498        get { return currentLine; }
499        private set { currentLine = value; }
500      }
[13414]501      public long BytesRead {
502        get;
503        private set;
504      }
[7849]505
506      public Tokenizer(StreamReader reader, NumberFormatInfo numberFormatInfo, DateTimeFormatInfo dateTimeFormatInfo, char separator) {
507        this.reader = reader;
508        this.numberFormatInfo = numberFormatInfo;
509        this.dateTimeFormatInfo = dateTimeFormatInfo;
510        this.separator = separator;
[13447]511        this.separators = new char[] { separator };
[7849]512        ReadNextTokens();
513      }
514
[13447]515      public bool HasNext() {
516        return numTokens > tokenPos || !reader.EndOfStream;
517      }
518
519      public TokenTypeEnum PeekType() {
520        return tokenTypes[tokenPos];
521      }
522
523      public void Skip() {
524        // simply skips one token without returning the result values
525        tokenPos++;
526        if (numTokens == tokenPos) {
527          ReadNextTokens();
528        }
529      }
530
531      public void Next(out TokenTypeEnum type, out string strVal, out double dblVal, out DateTime dateTimeVal) {
532        type = tokenTypes[tokenPos];
533        strVal = stringVals[tokenPos];
534        dblVal = doubleVals[tokenPos];
535        dateTimeVal = dateTimeVals[tokenPos];
536        Skip();
537      }
538
[7849]539      private void ReadNextTokens() {
540        if (!reader.EndOfStream) {
541          CurrentLine = reader.ReadLine();
[13447]542          CurrentLineNumber++;
[13925]543          if (reader.BaseStream.CanSeek) {
[13414]544            BytesRead = reader.BaseStream.Position;
[13925]545          } else {
[13414]546            BytesRead += CurrentLine.Length + 2; // guess
[13584]547          }
[13411]548          int i = 0;
[13447]549          if (!string.IsNullOrWhiteSpace(CurrentLine)) {
550            foreach (var tok in Split(CurrentLine)) {
551              TokenTypeEnum type;
[13411]552              double doubleVal;
553              DateTime dateTimeValue;
[13447]554              type = TokenTypeEnum.String; // default
555              stringVals[i] = tok.Trim();
556              if (double.TryParse(tok, NumberStyles.Float, numberFormatInfo, out doubleVal)) {
[13411]557                type = TokenTypeEnum.Double;
558                doubleVals[i] = doubleVal;
[14296]559              } else if (DateTime.TryParse(tok, dateTimeFormatInfo, DateTimeStyles.NoCurrentDateDefault, out dateTimeValue)
[14408]560                && (dateTimeValue.Year > 1 || dateTimeValue.Month > 1 || dateTimeValue.Day > 1)// if no date is given it is returned as 1.1.0001 -> don't allow this
[14296]561                ) {
[13411]562                type = TokenTypeEnum.DateTime;
563                dateTimeVals[i] = dateTimeValue;
[13526]564              } else if (string.IsNullOrWhiteSpace(tok)) {
565                type = TokenTypeEnum.Missing;
[13411]566              }
[7849]567
[13447]568              // couldn't parse the token as an int or float number or datetime value so return a string token
[13411]569
570              tokenTypes[i] = type;
571              i++;
572
573              if (i >= tokenTypes.Length) {
574                // increase buffer size if necessary
575                IncreaseCapacity(ref tokenTypes);
576                IncreaseCapacity(ref doubleVals);
577                IncreaseCapacity(ref stringVals);
578                IncreaseCapacity(ref dateTimeVals);
579              }
580            }
581          }
582          tokenTypes[i] = TokenTypeEnum.NewLine;
583          numTokens = i + 1;
584          tokenPos = 0;
[7849]585        }
586      }
587
[13447]588      private IEnumerable<string> Split(string line) {
589        return separator == WHITESPACECHAR ?
590          line.Split(whiteSpaceSeparators, StringSplitOptions.RemoveEmptyEntries) :
591          line.Split(separators);
592      }
593
[13411]594      private static void IncreaseCapacity<T>(ref T[] arr) {
595        int n = (int)Math.Floor(arr.Length * 1.7); // guess
596        T[] arr2 = new T[n];
597        Array.Copy(arr, arr2, arr.Length);
598        arr = arr2;
599      }
[7849]600    }
601    #endregion
602
603    #region parsing
604
605    private void ParseVariableNames() {
606      // the first line must contain variable names
[13411]607      List<string> varNames = new List<string>();
608
609      TokenTypeEnum type;
610      string strVal;
611      double dblVal;
612      DateTime dateTimeVal;
613
614      tokenizer.Next(out type, out strVal, out dblVal, out dateTimeVal);
615
616      // the first token must be a variable name
617      if (type != TokenTypeEnum.String)
618        throw new ArgumentException("Error: Expected " + TokenTypeEnum.String + " got " + type);
619      varNames.Add(strVal);
620
[13447]621      while (tokenizer.HasNext() && tokenizer.PeekType() != TokenTypeEnum.NewLine) {
[13411]622        tokenizer.Next(out type, out strVal, out dblVal, out dateTimeVal);
623        varNames.Add(strVal);
[7849]624      }
[13411]625      ExpectType(TokenTypeEnum.NewLine);
626
627      variableNames = varNames;
[7849]628    }
629
[13411]630    private void ExpectType(TokenTypeEnum expectedToken) {
631      if (tokenizer.PeekType() != expectedToken)
632        throw new ArgumentException("Error: Expected " + expectedToken + " got " + tokenizer.PeekType());
633      tokenizer.Skip();
[7849]634    }
635
636    private void Error(string message, string token, int lineNumber) {
[14285]637      throw new IOException(string.Format("Error while parsing. {0} (token: {1} lineNumber: {2}).", message, token, lineNumber));
[7849]638    }
639    #endregion
640  }
641}
Note: See TracBrowser for help on using the repository browser.