static void Main(string[] args)
{
string data = File.ReadAllText("Data.txt");
//setup rules that will be applied
var splitSentences = new SingleRegExpRule(@"(?<=[.!?])\s+(?=[A-Z])", RegExpRuleType.RegExpSplit, "Split sentences");
var filterQuestions = new SingleRegExpRule(@"[?]$", RegExpRuleType.RegExpFilter, "Filter quesions");
var filterSentencesWithDate = new MultipleRegExpRules(new List <string>
{
@"\d{1,2}[/-]\d{1,2}[/-]\d{2,4}",
@"(?:\d{1,2} )?(?:Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)[a-z]* (?:\d{1,2}, )?\d{4}"
}, RegExpRuleType.RegExpFilter, RulesConnectionType.Union, "Filter dates");
//setup rules results logging
var consoleVisualizer = new RegExpResultConsoleVisualizer();
EventHandler <RegExpParserResultEventArgs> textParsed = consoleVisualizer.Visualize;
var textParsedPublisher = new RegExpParserResultPublisher(textParsed);
//apply rules
var initData = new RegExpParserResult(data, null);
var result = initData.SplitByRegExp(splitSentences, textParsedPublisher)
.FilterByRegExp(filterQuestions, textParsedPublisher)
.FilterByRegExp(filterSentencesWithDate, textParsedPublisher);
//Console.WriteLine("INITIAL TEXT:\n{0}", data);
//var freq = SimpleRexExpForTest.CalcFrequencyOfWordI(data);
//var sentences = SimpleRexExpForTest.SplitOnSentences(data, true);
//var questions = SimpleRexExpForTest.FilterQuestions(sentences, true);
//SimpleRexExpForTest.FilterWithDates(questions, true);
}
public static RegExpParserResult SplitByRegExp(this RegExpParserResult prevParserResult,
SingleRegExpRule rule, IRegExpParserResultPublisher <RegExpRuleType, RegExpParserResultEventArgs> resultPublisher)
{
//check if regular expression type matches operation type
if (rule.RuleType != RegExpRuleType.RegExpSplit)
{
throw new NotSupportedException(
String.Format("Method accepts regular expressions that are aimed to split text, but this has type {0}",
rule.RuleType.ToString()));
}
//split blank text
var entryArray = Regex.Split(prevParserResult.SourceText, rule.RegularExpression);
//identify entries indexes
var entryIndexDictionary = new List <KeyValuePair <int, string> >();
foreach (var entry in entryArray)
{
int entryIndex = prevParserResult.SourceText.IndexOf(entry);
entryIndexDictionary.Add(new KeyValuePair <int, string>(entryIndex, entry));
}
resultPublisher.RegExpParserResult = new RegExpParserResult(
(prevParserResult.Entries != null && prevParserResult.Entries.Count() != 0) ?
String.Join("\n", prevParserResult.Entries.Select(e => e.Value)) : prevParserResult.SourceText,
entryIndexDictionary, rule.Title, prevParserResult);
resultPublisher.Publish();
return((RegExpParserResult)resultPublisher.RegExpParserResult);
}
public void SplitByRegExp_WithSingleRegExpRule_Split9Sentences()
{
//Arrange
var splitSentences = new SingleRegExpRule(@"(?<=[.!?])\s+(?=[A-Z])", RegExpRuleType.RegExpSplit, "Split sentences");
var initData = new RegExpParserResult(_data, null);
//Act
var sentences = initData.SplitByRegExp(splitSentences);
//Assert
Assert.IsTrue(sentences.Entries.ToList().Count == 9);
}
public void ChainOfSplitByRegExpAndFilterByRegExp_WithSingleRegExpRule_Split9SentencesAdnFiltersQuestions()
{
//Arrange
var splitSentences = new SingleRegExpRule(@"(?<=[.!?])\s+(?=[A-Z])", RegExpRuleType.RegExpSplit, "Split sentences");
var filterQuestions = new SingleRegExpRule(@"[?]$", RegExpRuleType.RegExpFilter, "Filter quesions");
var initData = new RegExpParserResult(_data, null);
//Act
var questions = initData.SplitByRegExp(splitSentences).FilterByRegExp(filterQuestions);
//Assert
Assert.IsTrue(questions.Entries.ToList().Count == 3);
}
public void ChainOfSplitByRegExpAndFilterByRegExp_WithMultipleRegExpRule_Split9SentencesAdnFiltersSentencesWithDates()
{
//Arrange
var splitSentences = new SingleRegExpRule(@"(?<=[.!?])\s+(?=[A-Z])", RegExpRuleType.RegExpSplit, "Split sentences");
var filterSentencesWithDate = new MultipleRegExpRules(new List <string>
{
@"\d{1,2}[/-]\d{1,2}[/-]\d{2,4}",
@"(?:\d{1,2} )?(?:Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)[a-z]* (?:\d{1,2}, )?\d{4}"
}, RegExpRuleType.RegExpSplit, RulesConnectionType.Union, "Filter dates");
var initData = new RegExpParserResult(_data, null);
//Act
var sentencesWithBothFormsOfDates = initData.SplitByRegExp(splitSentences).FilterByRegExp(filterSentencesWithDate);
//Assert
Assert.IsTrue(sentencesWithBothFormsOfDates.Entries.ToList().Count == 6);
}
public static RegExpParserResult FindByRegExp(this RegExpParserResult prevParserResult,
RegExpRule rule, IRegExpParserResultPublisher <RegExpRuleType, RegExpParserResultEventArgs> resultPublisher)
{
//check if regular expression type matches operation type
if (rule.RuleType != RegExpRuleType.RegExpFind)
{
throw new NotSupportedException(
String.Format("Method accepts regular expressions that are aimed to filter text, but this has type {0}",
rule.RuleType.ToString()));
}
//check if entires are not null as next calculations are based on it
if (prevParserResult.Entries == null)
{
throw new ArgumentNullException("Something went wrong. Entries of previous action are null.");
}
//apply different filtration logic depending on type of rule
IEnumerable <KeyValuePair <int, string> > entryIndexDictionary = new List <KeyValuePair <int, string> >();
if (rule is SingleRegExpRule)
{
entryIndexDictionary = FindEntriesWithSingleRexExp(prevParserResult.Entries, (rule as SingleRegExpRule));
}
else if (rule is MultipleRegExpRules)
{
entryIndexDictionary = FindEntriesWithMultipleRegExp(prevParserResult.Entries, (rule as MultipleRegExpRules));
}
else
{
throw new NotSupportedException("Passed rule type is not supported in this method.");
}
resultPublisher.RegExpParserResult = new RegExpParserResult(
(prevParserResult.Entries != null && prevParserResult.Entries.Count() != 0) ?
String.Join("\n", prevParserResult.Entries.Select(e => e.Value)) : prevParserResult.SourceText,
entryIndexDictionary, rule.Title, prevParserResult);
resultPublisher.Publish();
return((RegExpParserResult)resultPublisher.RegExpParserResult);
}