Best way to parse a log file in C#

Question

I have the following log file:

START:SOME_STRING
BL:2
LK:3
LH:5
end
START:SOME_STRING
BL:5
LK:6
LH:6
end

Which has multiple START: -> end structures inside. Is there a better 'non-sloppy' way of parsing this file rather than reading line by line and using SPLIT?

score 2 · Accepted Answer · edited May 23 '17 at 09:58

You can try to formalize your ini-file's grammar, and you some of parser generators. See this question for more detail.

Be aware howeveer that for such a simple grammar as yours it might be easier to parse manually :-P

class IniEntry
{
    public int BL;
    public int LK;
    public int LH;
    IniEntry Clone() { return new IniEntry { BL = BL, LK = LK, LH = LH }; }
}

IEnumerable<IniEntry> Parse()
{
    IniEntry ie = new IniEntry();
    while (ParseEntry(out ie))
        yield return ie.Clone();
}

bool ParseEntry(out IniEntry ie)
{
    ie = new IniEntry();
    return ParseStart(ie) &&
               ParseBL(ie) &&
               ParseLK(ie) &&
               ParseLH(ie) &&
               ParseEnd(ie);
}

bool ParseStart(IniEntry ie)
{
    string dummy;
    return ParseLine("START", out dummy);
}

bool ParseBL(IniEntry ie)
{
    string BL;
    return ParseLine("BL", out BL) && int.TryParse(BL, out ie.BL);
}

bool ParseLK(IniEntry ie)
{
    string LK;
    return ParseLine("LK", out LK) && int.TryParse(LK, out ie.LK);
}

bool ParseLH(IniEntry ie)
{
    string LH;
    return ParseLine("LH", out LH) && string.TryParse(LH, out ie.LH);
}

bool ParseLine(string key, out string value)
{
    string line = GetNextLine();
    var parts = line.Split(":");
    if (parts.Count != 2) return false;
    if (parts[0] != key) return false;
    value = parts[1];
}

etc.

Thats the log file I was given to parse, unfortunately not subject to change. — rvk, Nov 03 '10 at 09:05

score 0 · Answer 2 · answered Nov 03 '10 at 09:01

0

This is a good candidate for a while loop and a state machine. With this approach you would use even use less memory and have greater performance than using string.split()

answered Nov 03 '10 at 09:01

Karim Agha

3,606
4
32
48

1

Some further details on the state machine will get an up vote from me. – Zephyr was a Friend of Mine Nov 03 '10 at 09:04
State machine? I'll have a look at it. – rvk Nov 03 '10 at 09:05
Avoid the pitfall of doing things way too 'non-sloppy'. Unless you can reuse somehwre else the state mechine implementation you better thing twice if it's wort it. You coudl do the SPLIT option in far less time. – Liviu Mandras Nov 03 '10 at 09:18

score 0 · Answer 3 · answered Nov 03 '10 at 09:14

0

If it is certain that the START/END are always matched, (apologies, my C# is embarrassing, so plain English):

Read the whole file with System.IO.ReadToEnd
Parse the whole thing in one go with a regular expression
Iterate over regex results

The regex would be something like "(START:([^$]+)$BL:([^$]+)$LK:([^$]+)$LH:([^$]+)$end$)+", off the top of my head, you'll need to validate/adjust according to how your parameters BL/LK etc. occur

answered Nov 03 '10 at 09:14

smirkingman

6,167
4
34
47

1

This doesn't sounds efficient when some files I need read are over 6000 lines. Unless I am mistaken? – rvk Nov 03 '10 at 09:18
That's really a bad solution. Just imagine the file has a size of several hundred megs. Better read it line by line (or chunk by chunk) and use some kind of state machine. – Oliver Nov 03 '10 at 09:27
Is memory really a problem? Even if it momentarily takes 1Gb of memory, so what? Of course, a state machine is more elegant, but this solves the problem quickly – smirkingman Nov 03 '10 at 09:39

Best way to parse a log file in C#

3 Answers3