public void A05_AsyncRegExCrawlerTwoExpressionMultipleResultTest()
{
// Read the content
string content = File.ReadAllText("./Testdata/testdata.log");
// build search term one
var regExSearchTerm1 = new RegExSearchTerm()
{
Expression = @"(\d{4}-\d{2}-\d{2}).+(CatalogCacheUpdateJob perform)(.+)",
ExpressionFriendlyName = "SearchTerm 1"
};
// build search term two
var regExSearchTerm2 = new RegExSearchTerm()
{
Expression = @"2015-06-27 12:03:04,721 INFO \[EJB default - 5\] RuleChangeCommand executeChange - Setting entry 00.2285 to MODIFY",
ExpressionFriendlyName = "Ex1"
};
// fill the search term list
var regExSearchList = new List<RegExSearchTerm>() { regExSearchTerm1, regExSearchTerm2 };
// build the expected findigs
// build finding one
var expectedFinding1 = new Finding()
{
Expression = @"(\d{4}-\d{2}-\d{2}).+(CatalogCacheUpdateJob perform)(.+)",
ExpressionFriendlyName = "SearchTerm 1",
FileFolder = @"D:\WORK\RegExTractor\src\RegExTractor\RegExTractorTests\bin\Debug\Testdata",
FileName = "testdata.log",
Match = new List<RegExTractorMatchCollection>()
{
// first match
new RegExTractorMatchCollection()
{
Id = 1,
MatchCollection = new List<RegExTractorMatch>()
{
new RegExTractorMatch()
{
Id = 1,
Match = "2015-06-27 13:00:00,005 INFO [EJB default - 2] CatalogCacheUpdateJob perform - Performing scheduled reload of catalog cache...\r"
},
new RegExTractorMatch()
{
Id = 2,
Match = "2015-06-27"
},
new RegExTractorMatch()
{
Id = 3,
Match = "CatalogCacheUpdateJob perform"
},
new RegExTractorMatch()
{
Id = 4,
Match = " - Performing scheduled reload of catalog cache...\r"
}
}
},
// second match
new RegExTractorMatchCollection()
{
Id = 2,
MatchCollection = new List<RegExTractorMatch>()
{
new RegExTractorMatch()
{
Id = 1,
Match = "2015-06-27 13:01:30,110 INFO [EJB default - 2] CatalogCacheUpdateJob perform - Scheduled catalog cache reload successfully completed.\r"
},
new RegExTractorMatch()
{
Id = 2,
Match = "2015-06-27"
},
new RegExTractorMatch()
{
Id = 3,
Match = "CatalogCacheUpdateJob perform"
},
new RegExTractorMatch()
{
Id = 4,
Match = " - Scheduled catalog cache reload successfully completed.\r"
}
}
},
// third match
new RegExTractorMatchCollection()
{
Id = 3,
MatchCollection = new List<RegExTractorMatch>()
{
new RegExTractorMatch()
{
Id = 1,
Match = "2015-06-28 01:00:00,006 INFO [EJB default - 6] CatalogCacheUpdateJob perform - Performing scheduled reload of catalog cache...\r"
},
new RegExTractorMatch()
{
Id = 2,
Match = "2015-06-28"
},
new RegExTractorMatch()
{
Id = 3,
Match = "CatalogCacheUpdateJob perform"
},
new RegExTractorMatch()
{
Id = 4,
Match = " - Performing scheduled reload of catalog cache...\r"
}
}
},
// fourth match
new RegExTractorMatchCollection()
{
Id = 4,
MatchCollection = new List<RegExTractorMatch>()
{
new RegExTractorMatch()
{
Id = 1,
Match = "2015-06-28 01:01:29,226 INFO [EJB default - 6] CatalogCacheUpdateJob perform - Scheduled catalog cache reload successfully completed.\r"
},
new RegExTractorMatch()
{
Id = 2,
Match = "2015-06-28"
},
new RegExTractorMatch()
{
Id = 3,
Match = "CatalogCacheUpdateJob perform"
},
new RegExTractorMatch()
{
Id = 4,
Match = " - Scheduled catalog cache reload successfully completed.\r"
}
}
}
}
};
// build finding two
var expectedFinding2 = new Finding()
{
Expression = @"2015-06-27 12:03:04,721 INFO \[EJB default - 5\] RuleChangeCommand executeChange - Setting entry 00.2285 to MODIFY",
ExpressionFriendlyName = "Ex1",
FileFolder = @"D:\WORK\RegExTractor\src\RegExTractor\RegExTractorTests\bin\Debug\Testdata",
FileName = "testdata.log",
Match = new List<RegExTractorMatchCollection>()
{
new RegExTractorMatchCollection()
{
Id = 1,
MatchCollection = new List<RegExTractorMatch>()
{
new RegExTractorMatch()
{
Id = 1,
Match = "2015-06-27 12:03:04,721 INFO [EJB default - 5] RuleChangeCommand executeChange - Setting entry 00.2285 to MODIFY"
}
}
}
}
};
// build expected findings list
var expected = new List<Finding>() { expectedFinding1, expectedFinding2 };
// do the magic and crawl!
IRegExCrawler crawler = new AsyncRegExCrawler();
var actual = crawler.Crawl(regExSearchList, content, expectedFinding1.FileName, expectedFinding1.FileFolder);
var comparer = new CompareLogic();
comparer.Config.IgnoreCollectionOrder = true;
var compareResult = comparer.Compare(expected, actual);
Assert.IsTrue(compareResult.AreEqual, compareResult.DifferencesString);
Assert.AreEqual(expected.Count, actual.Count, "Another result set expected.");
}
public void A01_SimpleRexExCrawlerTest()
{
string content = File.ReadAllText("./Testdata/testdata.log");
var regExSearchTerm1 = new RegExSearchTerm()
{
Expression = @"(\d{4}-\d{2}-\d{2}).+(CatalogCacheUpdateJob perform)(.+)",
ExpressionFriendlyName = "SearchTerm 1"
};
var regExSearchList = new List<RegExSearchTerm>() { regExSearchTerm1 };
var expectedFinding1 = new Finding()
{
Expression = @"(\d{4}-\d{2}-\d{2}).+(CatalogCacheUpdateJob perform)(.+)",
ExpressionFriendlyName = "SearchTerm 1",
FileFolder = @"D:\WORK\RegExTractor\src\RegExTractor\RegExTractorTests\bin\Debug\Testdata",
FileName = "testdata.log",
Match = new List<RegExTractorMatchCollection>()
{
// first match
new RegExTractorMatchCollection()
{
Id = 1,
MatchCollection = new List<RegExTractorMatch>()
{
new RegExTractorMatch()
{
Id = 1,
Match = "2015-06-27 13:00:00,005 INFO [EJB default - 2] CatalogCacheUpdateJob perform - Performing scheduled reload of catalog cache...\r"
},
new RegExTractorMatch()
{
Id = 2,
Match = "2015-06-27"
},
new RegExTractorMatch()
{
Id = 3,
Match = "CatalogCacheUpdateJob perform"
},
new RegExTractorMatch()
{
Id = 4,
Match = " - Performing scheduled reload of catalog cache...\r"
}
}
},
// second match
new RegExTractorMatchCollection()
{
Id = 2,
MatchCollection = new List<RegExTractorMatch>()
{
new RegExTractorMatch()
{
Id = 1,
Match = "2015-06-27 13:01:30,110 INFO [EJB default - 2] CatalogCacheUpdateJob perform - Scheduled catalog cache reload successfully completed.\r"
},
new RegExTractorMatch()
{
Id = 2,
Match = "2015-06-27"
},
new RegExTractorMatch()
{
Id = 3,
Match = "CatalogCacheUpdateJob perform"
},
new RegExTractorMatch()
{
Id = 4,
Match = " - Scheduled catalog cache reload successfully completed.\r"
}
}
},
// third match
new RegExTractorMatchCollection()
{
Id = 3,
MatchCollection = new List<RegExTractorMatch>()
{
new RegExTractorMatch()
{
Id = 1,
Match = "2015-06-28 01:00:00,006 INFO [EJB default - 6] CatalogCacheUpdateJob perform - Performing scheduled reload of catalog cache...\r"
},
new RegExTractorMatch()
{
Id = 2,
Match = "2015-06-28"
},
new RegExTractorMatch()
{
Id = 3,
Match = "CatalogCacheUpdateJob perform"
},
new RegExTractorMatch()
{
Id = 4,
Match = " - Performing scheduled reload of catalog cache...\r"
}
}
},
// fourth match
new RegExTractorMatchCollection()
{
Id = 4,
MatchCollection = new List<RegExTractorMatch>()
{
new RegExTractorMatch()
{
Id = 1,
Match = "2015-06-28 01:01:29,226 INFO [EJB default - 6] CatalogCacheUpdateJob perform - Scheduled catalog cache reload successfully completed.\r"
},
new RegExTractorMatch()
{
Id = 2,
Match = "2015-06-28"
},
new RegExTractorMatch()
{
Id = 3,
Match = "CatalogCacheUpdateJob perform"
},
new RegExTractorMatch()
{
Id = 4,
Match = " - Scheduled catalog cache reload successfully completed.\r"
}
}
}
}
};
var expected = new List<Finding>() { expectedFinding1 };
IRegExCrawler crawler = new SimpleRegExCrawler();
var actual = crawler.Crawl(regExSearchList, content, expected[0].FileName, expected[0].FileFolder);
var comparer = new CompareLogic();
var compareResult = comparer.Compare(expected, actual);
Assert.IsTrue(compareResult.AreEqual);
}
public void A02_SimpleRegexCrawlerOneExpressionOneResultTest()
{
string content = File.ReadAllText("./Testdata/testdata.log");
var regExSearchTerm = new RegExSearchTerm()
{
Expression = @"2015-06-27 12:03:04,721 INFO \[EJB default - 5\] RuleChangeCommand executeChange - Setting entry 00.2285 to MODIFY",
ExpressionFriendlyName = "Ex1"
};
var expectedFinding = new Finding()
{
Expression = @"2015-06-27 12:03:04,721 INFO \[EJB default - 5\] RuleChangeCommand executeChange - Setting entry 00.2285 to MODIFY",
ExpressionFriendlyName = "Ex1",
Match = new List<RegExTractorMatchCollection>()
{
new RegExTractorMatchCollection()
{
Id = 1,
MatchCollection = new List<RegExTractorMatch>()
{
new RegExTractorMatch()
{
Id = 1,
Match = "2015-06-27 12:03:04,721 INFO [EJB default - 5] RuleChangeCommand executeChange - Setting entry 00.2285 to MODIFY"
}
}
}
}
};
IRegExCrawler crawler = new SimpleRegExCrawler();
var actual = crawler.Crawl(new List<RegExSearchTerm>(){ regExSearchTerm}, content, expectedFinding.FileName, expectedFinding.FileFolder);
var expected = new List<Finding>() { expectedFinding };
Assert.AreEqual(expected[0].Expression, actual[0].Expression);
Assert.AreEqual(expected[0].ExpressionFriendlyName, actual[0].ExpressionFriendlyName);
Assert.AreEqual(expected[0].FileFolder, actual[0].FileFolder);
Assert.AreEqual(expected[0].FileName, actual[0].FileName);
var matchCount = actual[0].Match.Count();
Assert.AreEqual(1, matchCount);
for (int i = 0; i <= matchCount; i++)
{
var expectedMatch = expected[0].Match[0];
var actualMatch = actual[0].Match[0];
Assert.AreEqual(expectedMatch.Id, actualMatch.Id);
Assert.AreEqual(expectedMatch.MatchCollection[0].Id, actualMatch.MatchCollection[0].Id);
}
}
public void A03_SimpleRegExCrawlerTwoExpressionMultipleResultTest()
{
// Read the content
string content = File.ReadAllText("./Testdata/testdata.log");
// build search term one
var regExSearchTerm1 = new RegExSearchTerm()
{
Expression = @"(\d{4}-\d{2}-\d{2}).+(CatalogCacheUpdateJob perform)(.+)",
ExpressionFriendlyName = "SearchTerm 1"
};
// build search term two
var regExSearchTerm2 = new RegExSearchTerm()
{
Expression = @"2015-06-27 12:03:04,721 INFO \[EJB default - 5\] RuleChangeCommand executeChange - Setting entry 00.2285 to MODIFY",
ExpressionFriendlyName = "Ex1"
};
// fill the search term list
var regExSearchList = new List<RegExSearchTerm>() { regExSearchTerm1, regExSearchTerm2 };
// build the expected findigs
// build finding one
var expectedFinding1 = new Finding()
{
Expression = @"(\d{4}-\d{2}-\d{2}).+(CatalogCacheUpdateJob perform)(.+)",
ExpressionFriendlyName = "SearchTerm 1",
FileFolder = @"D:\WORK\RegExTractor\src\RegExTractor\RegExTractorTests\bin\Debug\Testdata",
FileName = "testdata.log",
Match = new List<RegExTractorMatchCollection>()
{
// first match
new RegExTractorMatchCollection()
{
Id = 1,
MatchCollection = new List<RegExTractorMatch>()
{
new RegExTractorMatch()
{
Id = 1,
Match = "2015-06-27 13:00:00,005 INFO [EJB default - 2] CatalogCacheUpdateJob perform - Performing scheduled reload of catalog cache...\r"
},
new RegExTractorMatch()
{
Id = 2,
Match = "2015-06-27"
},
new RegExTractorMatch()
{
Id = 3,
Match = "CatalogCacheUpdateJob perform"
},
new RegExTractorMatch()
{
Id = 4,
Match = " - Performing scheduled reload of catalog cache...\r"
}
}
},
// second match
new RegExTractorMatchCollection()
{
Id = 2,
MatchCollection = new List<RegExTractorMatch>()
{
new RegExTractorMatch()
{
Id = 1,
Match = "2015-06-27 13:01:30,110 INFO [EJB default - 2] CatalogCacheUpdateJob perform - Scheduled catalog cache reload successfully completed.\r"
},
new RegExTractorMatch()
{
Id = 2,
Match = "2015-06-27"
},
new RegExTractorMatch()
{
Id = 3,
Match = "CatalogCacheUpdateJob perform"
},
new RegExTractorMatch()
{
Id = 4,
Match = " - Scheduled catalog cache reload successfully completed.\r"
}
}
},
// third match
new RegExTractorMatchCollection()
{
Id = 3,
MatchCollection = new List<RegExTractorMatch>()
{
new RegExTractorMatch()
{
Id = 1,
Match = "2015-06-28 01:00:00,006 INFO [EJB default - 6] CatalogCacheUpdateJob perform - Performing scheduled reload of catalog cache...\r"
},
new RegExTractorMatch()
{
Id = 2,
Match = "2015-06-28"
},
new RegExTractorMatch()
{
Id = 3,
Match = "CatalogCacheUpdateJob perform"
},
new RegExTractorMatch()
{
Id = 4,
Match = " - Performing scheduled reload of catalog cache...\r"
}
}
},
// fourth match
new RegExTractorMatchCollection()
{
Id = 4,
MatchCollection = new List<RegExTractorMatch>()
{
new RegExTractorMatch()
{
Id = 1,
Match = "2015-06-28 01:01:29,226 INFO [EJB default - 6] CatalogCacheUpdateJob perform - Scheduled catalog cache reload successfully completed.\r"
},
new RegExTractorMatch()
{
Id = 2,
Match = "2015-06-28"
},
new RegExTractorMatch()
{
Id = 3,
Match = "CatalogCacheUpdateJob perform"
},
new RegExTractorMatch()
{
Id = 4,
Match = " - Scheduled catalog cache reload successfully completed.\r"
}
}
}
}
};
// build finding two
var expectedFinding2 = new Finding()
{
Expression = @"2015-06-27 12:03:04,721 INFO \[EJB default - 5\] RuleChangeCommand executeChange - Setting entry 00.2285 to MODIFY",
ExpressionFriendlyName = "Ex1",
FileFolder = @"D:\WORK\RegExTractor\src\RegExTractor\RegExTractorTests\bin\Debug\Testdata",
FileName = "testdata.log",
Match = new List<RegExTractorMatchCollection>()
{
new RegExTractorMatchCollection()
{
Id = 1,
MatchCollection = new List<RegExTractorMatch>()
{
new RegExTractorMatch()
{
Id = 1,
Match = "2015-06-27 12:03:04,721 INFO [EJB default - 5] RuleChangeCommand executeChange - Setting entry 00.2285 to MODIFY"
}
}
}
}
};
// build expected findings list
var expected = new List<Finding>() { expectedFinding1, expectedFinding2 };
// do the magic and crawl!
IRegExCrawler crawler = new SimpleRegExCrawler();
var actual = crawler.Crawl(regExSearchList, content, expectedFinding1.FileName, expectedFinding1.FileFolder);
KellermanSoftware.CompareNetObjects.CompareLogic l = new KellermanSoftware.CompareNetObjects.CompareLogic();
var result = l.Compare(expected, actual);
Assert.IsTrue(result.AreEqual);
// check the results
// result for finding one
Assert.AreEqual(expected[0].Expression, actual[0].Expression);
Assert.AreEqual(expected[0].ExpressionFriendlyName, actual[0].ExpressionFriendlyName);
//Assert.AreEqual(expected.FirstOrDefault().FileFolder, actual.FirstOrDefault().FileFolder);
//Assert.AreEqual(expected.FirstOrDefault().FileName, actual.FirstOrDefault().FileName);
var matchCount = actual[0].Match.Count();
Assert.AreEqual(4, matchCount);
for (int i = 0; i <= matchCount; i++)
{
var expectedMatch = expected[0].Match[0];
var actualMatch = actual[0].Match[0];
Assert.AreEqual(expectedMatch.Id, actualMatch.Id);
Assert.AreEqual(expectedMatch.MatchCollection[0].Id, actualMatch.MatchCollection[0].Id);
Assert.AreEqual(expectedMatch.MatchCollection[1].Id, actualMatch.MatchCollection[1].Id);
Assert.AreEqual(expectedMatch.MatchCollection[2].Id, actualMatch.MatchCollection[2].Id);
}
// result for finding two
Assert.AreEqual(expected[1].Expression, actual[1].Expression);
Assert.AreEqual(expected[1].ExpressionFriendlyName, actual[1].ExpressionFriendlyName);
//Assert.AreEqual(expected.FirstOrDefault().FileFolder, actual.FirstOrDefault().FileFolder);
//Assert.AreEqual(expected.FirstOrDefault().FileName, actual.FirstOrDefault().FileName);
matchCount = actual[1].Match.Count();
Assert.AreEqual(1, matchCount);
for (int i = 0; i <= matchCount; i++)
{
var expectedMatch = expected[1].Match[0];
var actualMatch = actual[1].Match[0];
Assert.AreEqual(expectedMatch.Id, actualMatch.Id);
Assert.AreEqual(expectedMatch.MatchCollection[0].Id, actualMatch.MatchCollection[0].Id);
}
}
/// <summary>
/// Parse the given content with the given search terms.
/// </summary>
/// <param name="SearchTerms"></param>
/// <param name="Content"></param>
/// <param name="FileName"></param>
/// <param name="FileFolder"></param>
/// <returns></returns>
public List<Finding> Crawl(List<RegExSearchTerm> SearchTerms, string Content, string FileName, string FileFolder)
{
var findingResultList = new List<Finding>();
// iterate through search terms
// each search term may result in a finding
// this finding will be added to finding result list
foreach (var searchTerm in SearchTerms)
{
var finding = new Finding()
{
Expression = searchTerm.Expression,
ExpressionFriendlyName = searchTerm.ExpressionFriendlyName,
FileName = FileName,
FileFolder = FileFolder,
Match = new List<RegExTractorMatchCollection>()
};
var regEx = new Regex(searchTerm.Expression);
var regExMatchCollection = regEx.Matches(Content);
// the System.Text.RegularExpression.Regex.Matches has not index
// so we have to count the iterations by ourself
int matchCount = 0;
// iterate through match collection
foreach (Match regExMatch in regExMatchCollection)
{
// increase matchCount
matchCount++;
// create a new RegExtractorMatchCollection
var resultMatchCollection = new RegExTractorMatchCollection()
{
// set the id of the collection
Id = matchCount, MatchCollection = new List<RegExTractorMatch>()
};
// count match groups
var groupsCount = regExMatch.Groups.Count;
// loop throug groups
for (int g = 0; g < groupsCount; g++)
{
// create a single RegExTractorMatch
var resultMatch = new RegExTractorMatch()
{
Id = g + 1,
Match = regExMatch.Groups[g].Value
};
// add this match to result collection
resultMatchCollection.MatchCollection.Add(resultMatch);
}
// add the result match collection to finding result
finding.Match.Add(resultMatchCollection);
}
// add finding to findig list
findingResultList.Add(finding);
// report progress
var eventArgs = new ReportProgressEventArgs();
eventArgs.Message = String.Format(@"Finished search for expression {0} in file {1}\{2}", searchTerm.ExpressionFriendlyName, FileFolder, FileName);
OnSingleFileCrawlFinished(eventArgs);
}
if (findingResultList == null) throw new ArgumentNullException();
return findingResultList;
}
