public MarkovBrain()
{
_markovChainTrainer = new Trainer(_markovChainString);
}
public static MarkovChainString Train(Config parameters, Action <string> output)
{
Func <int, string> formatInteger = (n) => n.ToString("N0");
Func <float, string> formatFloat = (n) => n.ToString("0.");
Func <string, string> getMessageText = (m) => m.Split('>').Last().Trim();
Func <string, string> getMessageNick = (m) => m.Split(' ').Skip(2).First();
Func <string, string> getMessageTime = (m) => m.Split(']').First().TrimStart('[');
Func <IEnumerable <string>, IEnumerable <string> > removeIrcEvents = (m) => m.Skip(1).Where(message => message.Length > 10 && message.Contains("<"));
output?.Invoke($"Initializing trainer '{parameters.Load}'");
var markovChainString = new MarkovChainString();
var markovChainTrainer = new Trainer(markovChainString);
output?.Invoke($"Loading '{parameters.Load}'");
var lines = File.ReadAllLines(parameters.Load);
var numLinesFormatted = formatInteger(lines.Length);
output?.Invoke($"Number of lines: {numLinesFormatted}");
output?.Invoke($"Removing IRC events from log");
var messages = removeIrcEvents(lines);
if (parameters.Sections > 1)
{
output?.Invoke($"Splitting log into {parameters.Sections} parts. Generating text from one part");
messages = messages.Reverse().Take(messages.Count() / (int)parameters.Sections);
}
var numMessagesFormatted = formatInteger(messages.Count());
output?.Invoke($"Number of messages: {numMessagesFormatted}");
if (parameters.Filter != "")
{
output?.Invoke($"Applying filter '{parameters.Filter}'");
messages = messages.Where(message => {
var text = getMessageText(message);
var nick = getMessageNick(message);
return(Regex.IsMatch(text, parameters.Filter, RegexOptions.IgnoreCase) || Regex.IsMatch(nick, parameters.Filter, RegexOptions.IgnoreCase));
});
numMessagesFormatted = formatInteger(messages.Count());
output?.Invoke($"Number of messages: {numMessagesFormatted}");
}
var i = 0;
var numReports = parameters.Reports;
var messageCount = messages.Count();
var step = messageCount / numReports;
output?.Invoke($"Processing messages");
messages.ForEach(message =>
{
if (i++ % step == 0)
{
var percentage = (float)100 / messageCount * i;
var percentageFormatted = formatFloat(percentage);
var wordCountFormatted = formatInteger(markovChainTrainer.WordCount);
var sentencesFormatted = formatInteger(markovChainTrainer.SentenceCount);
var messageTimeFormatted = getMessageTime(message);
output?.Invoke($"Processed: {percentageFormatted} %, {wordCountFormatted} words, {sentencesFormatted} sentences, logtime {messageTimeFormatted}");
}
var messageText = getMessageText(message);
markovChainTrainer.Train(messageText);
});
var uniqueWordsFormatted = formatInteger(markovChainString.Nodes.Count);
output?.Invoke($"Finished training");
output?.Invoke($"Unique words in brain: {uniqueWordsFormatted}");
return(markovChainString);
}
