从文本文件中提取数组

Question

I have a huge (~500K) text file, which looks like this:

{   // H-20e180a.wav 
    {-7,5,-4,-1,-9,2,-5,-1,2,-12,12,-33,34,-48,41,-40,16,20,730,4751,3861},
    {-7,5,-4,-1,-9,2,-5,-1,2,-12,12,-33,34,-48,41,-40,16,20,}
}

(NOTE: in the actual file there is some extra clutter, and the array-pairs are much longer each ~140 elts).

I am looking to create a C# / .NET routine that allows me to extract a pair of arrays:

int [] [] elev_neg20__azi_180 = ArraysForLocation( -20, 180 );

What would be my basic strategy?

From my days of coding BASIC, I would read in one line at a time, looking for '// H', then extract the 2 numbers, and if they match I would process the next two lines. But things have probably moved on since then!

I'm guessing that there is no shortcut to reading through the entire file...

Answer 1

From my days of coding BASIC, I would read in one line at a time, looking for '// H', then extract the 2 numbers, and if they match I would process the next two lines.

Approach it the same way. Using System.IO.StreamReader you can repeatedly ReadLine until you find the desired section, read the next two lines of data, and Close . Then String.Split the comma separated values and Convert.ToInt32 .

And actually you probably wouldn't explicitly call Close . The StreamReader class implements IDisposable so a best practice is to wrap it in a using statement (which automatically calls Dispose which will close the stream).

using (var reader = new StreamReader("somefile.txt"))
{
   string line = reader.ReadLine();
}

Parsing a string containing a line of your data could be done like this:

string line = "{-7,5,-4,-1,-9,2,-5,-1,2,-12,12,-33,34,-48,41,-40,16,20,730,4751,3861},";

var regex = new Regex("[{},]");
int[] ints = regex.Replace(line, " ").Trim().
                   Split(new char[] { ' ' }).Select(int.Parse).ToArray();

An option for returning the arrays from a method is to use out values. That way your normal return value could be used to indicate success. A method signature like this:

public bool ArraysForLocation(int x, int y, out int[] array1, out int[] array2)

could be called like this:

int[] a1;
int[] a2;
bool ok = ArraysForLocation(-20, 180, out a1, out a2);

I'm guessing that there is no shortcut to reading through the entire file...

You won't read the entire file unless what you are seeking is at the tail. You are streaming the data so only a line is read at a time. Unless the file content is sorted so that you could do a binary search with FileStream.Seek , then you'll need to read through the file until you find the data you are looking for.

Answer 2

You can use the string.Split(Char[]) method: http://msdn.microsoft.com/en-us/library/b873y76a.aspx

This method returns an array of strings.

The char parameter is the delimiter you want to split on. So you would call it once to split your long string into the two arrays you want, and then on each one split on the comma to get the respective arrays of individual values. After that you could convert the strings to int if needed.

Answer 3

If you are doing much lookup and speed is more important than memory you might want to process the file once and put the information into a dictionary. That way lookup is very fast and you only have to read the file once.

Here's some code that will parse data like the example you gave:

class Program
{
    static void Main(string[] args)
    {
        string filename = "example.txt";

        Dictionary<string, int[][]> myDictionary = new Dictionary<string, int[][]>();

        BuildMyDataDictionary(filename, myDictionary);

        //lookup via key
        int x = 20;
        int y = 180;
        string key = string.Format("{0}.{1}", x, y);
        int[][] values = myDictionary[key];

        //print the values to check
        foreach (int[] array in values)
            foreach (int i in array)
                Console.Write(i + ", ");
        Console.WriteLine();

        Console.ReadKey();
    }

    private static void BuildMyDataDictionary(string filename, Dictionary<string, int[][]> myDictionary)
    {
        using (StreamReader r = new StreamReader(filename))
        {
            string line = r.ReadLine();
            // read through the file line by line and build the dictionary
            while (line != null)
            {
                Regex regx = new Regex(@"//\s*H\-(\d*)\w(\d*)");
                Match m = regx.Match(line);
                if (m.Success)
                {
                    // make a key of the two parts int 1 and int2 separated by a "."
                    string key = string.Format("{0}.{1}", m.Groups[1], m.Groups[2]);

                    // continue reading the block
                    List<int[]> intList = new List<int[]>();
                    line = r.ReadLine();
                    while (!Regex.IsMatch(line, @"^\s*\}"))
                    {
                        Regex regex = new Regex("[{},]");
                        intList.Add(regex.Replace(line, " ").Trim().Split(new char[] { ' ' }).Select(int.Parse).ToArray());
                        line = r.ReadLine();
                    }
                    myDictionary.Add(key, intList.ToArray());
                }
                line = r.ReadLine();
            }
        }
    }
}

The example file I tested with was:

{   // H-20e180a.wav 
    {-7,5,-4,-1,-9,2,-5,-1,2,-12,12,-33,34,-48,41,-40,16,20,730,4751,3861},
    {-7,5,-4,-1,-9,2,-5,-1,2,-12,12,-33,34,-48,41,-40,16,20,}
}
{   // H-21e181a.wav 
    {-7,5,-4,-1,-9,2,-5,-1,2,-12,12,-33,34,-48,41,-40,16,20,730,4751,3861},
    {-7,5,-4,-1,-9,2,-5,-1,2,-12,12,-33,34,-48,41,-40,16,20,}
    {-7,5,-4,-1,-9,2,-5,-1,2,-12,12,-33,34,-48,41,-40,16,20,730,4751,3861},
}

I borrowed the int[] parsing and creation from jltrem above.