public override void Execute(Database database, IProviderCollection providers)
{
ITableImportance importance = providers.GetProvider <TableDataImportance>();
if (importance == null)
{
importance = providers.GetProvider <PageRank>();
}
this._importanceBackend = importance;
}
private void Write(String dirName, DatabaseLint dblint)
{
this.scoring = new IScoringImpl();
this.scoring.CalculateScores(dblint);
if (DBLint.Settings.IsNormalContext)
{
//Save run for incremental viewing
//File name is a timestamp
DateTime now = DateTime.Now;
String fileName = String.Format("{0}{1}{2}{3}{4}{5}.xml", now.Year, now.Month, now.Day, now.Hour,
now.Minute, now.Second);
//Folder, i.e.: runs/dbname/
String folder = Settings.INCREMENTAL_FOLDER + "testtest"; // dblint.DatabaseModel.DatabaseName;
String filePath = folder + "/" + fileName;
if (!Directory.Exists(folder))
{
Directory.CreateDirectory(folder);
}
//Create run
DBLint.IncrementalRuns.Run run = new IncrementalRuns.Run(dblint.DatabaseModel, dblint.IssueCollector, scoring.GetScores());
//Write run
using (FileStream writer = new FileStream(filePath, FileMode.Create))
{
DataContractSerializer ser = new DataContractSerializer(typeof(DBLint.IncrementalRuns.Run));
ser.WriteObject(writer, run);
writer.Flush();
}
}
DirectoryInfo dir = new DirectoryInfo(dirName);
int tableNameCounter = 1;
foreach (Table table in dblint.DatabaseModel.Tables)
{
String tName = "table" + tableNameCounter.ToString();
this.tableNames.Add(table, tName);
this.tableFiles.Add(table, "tables/" + tName + ".html");
tableNameCounter++;
}
this.dblint = dblint;
this.formatter = new HTMLDescriptionFormatter(this.tableFiles);
IssueCollector issues = dblint.IssueCollector;
//Create result directory if it does not exist
if (!dir.Exists)
{
dir.Create();
}
VelocityContext context = new VelocityContext();
context.Put("db", dblint.DatabaseModel);
context.Put("totalScore", this.scoring.GetScore());
context.Put("issuesTotal", issues.Count());
context.Put("rulesExecuted", this.getRulesExecuted());
context.Put("ruleTypes", this.getRuleTypes());
context.Put("formatter", this.formatter);
context.Put("HTMLBuilder", this);
context.Put("summaries", this.dblint.ExecutionSummary);
context.Put("executionTime", this.formatTimeSpan(this.dblint.ExecutionSummary.ExecutionTime));
//Pagerank
IProviderCollection providers = dblint.RuleController.ProviderCollection;
var rank = providers.GetProvider <DBLint.Rules.SchemaProviders.ImportanceProvider>();
//List all tables
var tables = (from t in dblint.DatabaseModel.Tables
select new
{
Table = t,
Name = t.TableName,
IssueCount = issues.GetIssues(t).Count(),
Score = this.scoring.GetScore(t),
Importance = Math.Round(rank[t], 1)
}).ToList();
context.Put("tables", tables);
//Bottom tables
var bottom = tables.OrderBy(t => t.Score).Take(5).ToList();
context.Put("bottomTables", bottom);
int groupId = 0; //Used in the template to identify a group of issues
//Group issues by name
var issueGroups = (from i in issues
group i by i.Name into g
orderby g.First().Severity
select new
{
Name = g.Key,
Count = g.Count(),
Issues = g,
GroupID = ++groupId,
Severity = g.First().Severity
}).ToList();
context.Put("issueGroups", issueGroups);
//Put issueGroups into severity groups
var severityGroups = (from issueGroup in issueGroups
group issueGroup by issueGroup.Severity into g
orderby g.First().Severity
select new
{
Severity = g.First().Severity,
IssueGroups = g
}
);
context.Put("severityGroups", severityGroups);
//Incremental runs list
var diffs = new List <DBLint.IncrementalRuns.Diff>();
if (DBLint.Settings.IsNormalContext)
{
//Incremental runs
try
{
var runs = DBLint.IncrementalRuns.Run.GetRuns(dblint.DatabaseModel.DatabaseName, 5).ToList();
for (int i = 1; i < runs.Count; i++)
{
var diff = new DBLint.IncrementalRuns.Diff();
diff.Compare(runs[i], runs[i - 1]);
diffs.Add(diff);
}
}
catch { }
context.Put("diffs", diffs);
}
//Create template for the main html page
Template template = Velocity.GetTemplate("mainpage.vm");
//Create outputstream for the main page
TextWriter htmlOut = new StreamWriter(Path.Combine(dir.FullName, "mainpage.html"));
//Write template
template.Merge(context, htmlOut);
htmlOut.Close();
//Write issue groups
String issuePath = Path.Combine(dir.FullName, "issues");
if (!Directory.Exists(issuePath))
{
Directory.CreateDirectory(issuePath);
}
Template issueGroupTemplate = Velocity.GetTemplate("issuegroup.vm");
formatter.PathPrefix = "../";
foreach (var g in issueGroups)
{
context.Put("groupIssues", g.Issues);
TextWriter issueOut = new StreamWriter(Path.Combine(issuePath, g.GroupID.ToString() + ".html"));
issueGroupTemplate.Merge(context, issueOut);
issueOut.Close();
}
if (DBLint.Settings.IsNormalContext)
{
//Write diffs/increments to files:
String incPath = Path.Combine(dir.FullName, "increments");
if (!Directory.Exists(incPath))
{
Directory.CreateDirectory(incPath);
}
Template incrementTemplate = Velocity.GetTemplate("increment.vm");
int diffId = 0;
foreach (var diff in diffs)
{
diffId++;
context.Put("diff", diff);
TextWriter incOut = new StreamWriter(Path.Combine(incPath, diffId.ToString() + ".html"));
incrementTemplate.Merge(context, incOut);
incOut.Close();
}
}
formatter.PathPrefix = "";
writeTableViews(dirName);
}
public override void Execute(DataTable table, IIssueCollector issueCollector, IProviderCollection providers)
{
var informationContent = providers.GetProvider <InformationContent>();
if (table.PrimaryKey == null || table.Cardinality < MinRows.Value)
{
return;
}
var pkcolumns = table.PrimaryKey.Columns;
var escaper = table.Database.Escaper;
// Heuristic: Only check columns storing a lot of entropy
var entropyOrderedCOlumns = pkcolumns.OrderByDescending(col => informationContent[col]).ToArray();
double maxPossibleCardinality = 1;
var currentColumns = new List <Column>(entropyOrderedCOlumns.Length);
var first = true;
var currentColumnsString = new StringBuilder();
foreach (var col in entropyOrderedCOlumns)
{
currentColumns.Add(col);
if (currentColumns.Count == entropyOrderedCOlumns.Length)
{
break; // Last column added. It is known to be a key.
}
maxPossibleCardinality *= informationContent[col];
if (table.Cardinality - maxPossibleCardinality > 0.2f)
{
continue; // If not enough entropy to generate a higher card. no need to query the data.
}
var escapedCol = escaper.Escape(col);
if (first)
{
currentColumnsString.AppendFormat("{0}", escapedCol);
first = false;
}
else
{
currentColumnsString.AppendFormat(", {0}", escapedCol);
}
var query = string.Format(@"SELECT COUNT(*)
FROM ( SELECT COUNT(*) AS rowcnt
FROM {0}
GROUP BY {1}
) AS exp1
WHERE rowcnt > 1 ", escaper.Escape(table), currentColumnsString.ToString());
var res = table.QueryTable(query);
if (res is DBNull)
{
break;
}
var num = Convert.ToInt32(res);
if (num == 0)
{
issueCollector.ReportIssue(new Issue(this, this.Severity)
{
Name = "Defined Primary Key is not a Minimal Key",
Context = new TableContext(table),
Description = new Description("Primary key for table {0}, is a superkey.", table),
ExtendedDescription = new Description("Columns {0} are enough to uniquely identify a tuple. Currently used are {1}", currentColumns, table.PrimaryKey.Columns),
});
break;
}
}
}
public override void Finalize(Model.Database database, IProviderCollection providers)
{
var informationContent = providers.GetProvider <InformationContent>();
var fks = database.Tables.SelectMany(t => t.ForeignKeys);
DatabaseDictionary <TableID, List <JoinEdge> > dbJoinEdges = DictionaryFactory.CreateTableID <List <JoinEdge> >();
DatabaseDictionary <TableID, double> tableTotalEntropyTransfer = DictionaryFactory.CreateTableID <double>();
foreach (var tbl in database.Tables)
{
dbJoinEdges[tbl] = new List <JoinEdge>(4);
tableTotalEntropyTransfer[tbl] = 0;
}
foreach (var foreignKey in fks)
{
var pkColumns = (from cp in foreignKey.ColumnPairs
select cp.PKColumn).ToArray();
var fkColumns = (from cp in foreignKey.ColumnPairs
select cp.FKColumn).ToArray();
double fkEdgeEntropy;
if (foreignKey.IsSingleColumn)
{
fkEdgeEntropy = informationContent[foreignKey.FKColumn];
}
else
{
fkEdgeEntropy = informationContent.GetMultiColumnEntropy((DataTable)foreignKey.FKTable, fkColumns);
}
var pkEdgeEntropy = Math.Log(Math.Max(foreignKey.PKTable.Cardinality, 1), 2); // Primary key guarantees uniqueness across pkcolumns, hence entropy equals log of cardinality.
dbJoinEdges[foreignKey.PKTable].Add(new JoinEdge {
Table = foreignKey.FKTable, Columns = fkColumns, EdgeEntropy = fkEdgeEntropy
});
dbJoinEdges[foreignKey.FKTable].Add(new JoinEdge {
Table = foreignKey.PKTable, Columns = pkColumns, EdgeEntropy = pkEdgeEntropy
});
tableTotalEntropyTransfer[foreignKey.PKTable] += pkEdgeEntropy;
tableTotalEntropyTransfer[foreignKey.FKTable] += fkEdgeEntropy;
}
DatabaseDictionary <TableID, DatabaseDictionary <TableID, double> > pmatrix = DictionaryFactory.CreateTableID <DatabaseDictionary <TableID, double> >();
foreach (var tbl in database.Tables)
{
pmatrix[tbl] = DictionaryFactory.CreateTableID <double>();
}
foreach (var toTable in database.Tables)
{
var joinEdges = dbJoinEdges[toTable];
foreach (var joinEdge in joinEdges)
{
var fromTable = joinEdge.Table;
var columnsEntropy = joinEdge.EdgeEntropy;
var tableInformationContent = informationContent[fromTable];
var tableTotalTransfer = tableTotalEntropyTransfer[fromTable];
var todic = pmatrix[toTable];
if (!todic.ContainsKey(fromTable))
{
todic[fromTable] = 0;
}
if (tableInformationContent + tableTotalTransfer > 0)
{
todic[fromTable] += columnsEntropy / (tableInformationContent + tableTotalTransfer);
}
}
var toTableRow = pmatrix[toTable];
}
foreach (var keyRow in pmatrix.Keys)
{
double selfLoopValue = 1d;
foreach (var keyColumn in pmatrix.Keys)
{
var row = pmatrix[keyColumn];
if (row.ContainsKey(keyRow))
{
selfLoopValue -= row[keyRow];
}
}
if (selfLoopValue < 0)
{
}
pmatrix[keyRow][keyRow] = selfLoopValue;
}
DatabaseDictionary <TableID, double> importanceVector = DictionaryFactory.CreateTableID <double>();
DatabaseDictionary <TableID, double> calculateVector = DictionaryFactory.CreateTableID <double>();
foreach (var table in database.Tables)
{
importanceVector[table] = Math.Max(informationContent[table], 0);
}
for (int i = 0; i < 100; i++)
{
foreach (var table in importanceVector.Keys)
{
double newRank = 0;
var fromTables = pmatrix[table];
foreach (var fromTable in fromTables)
{
var fromRank = importanceVector[fromTable.Key];
newRank += fromRank * fromTable.Value;
}
calculateVector[table] = newRank;
}
{
var tmp = calculateVector;
calculateVector = importanceVector;
importanceVector = tmp;
}
}
var totalEntropy = importanceVector.Values.Sum();
foreach (var k in importanceVector.Keys)
{
if (importanceVector[k] < 0)
{
}
if (totalEntropy == 0)
{
importanceVector[k] = 100f / database.Tables.Count;
}
else
{
importanceVector[k] *= 100f / totalEntropy;
}
}
this._importanceVector = importanceVector;
}
public override void Execute(DataTable table, IIssueCollector issueCollector, IProviderCollection providers)
{
var dataTypes = new[] { DataType.CHAR, DataType.NCHAR, DataType.NVARCHAR, DataType.VARCHAR };
var informationContent = providers.GetProvider <InformationContent>();
var columnsToCheck = (from c in table.Columns
where dataTypes.Contains(c.DataType) && informationContent[c] < 3 // Avoid checking columns with more than 8 unique values
select c).ToArray();
if (columnsToCheck.Length == 0)
{
return;
}
var columnNotBooleanCount = DictionaryFactory.CreateColumnID <int>();
foreach (var col in columnsToCheck)
{
columnNotBooleanCount[col] = 0;
}
int rowCount = 0;
using (var rowEnumerable = table.GetTableRowEnumerable())
foreach (var row in rowEnumerable)
{
rowCount++;
foreach (var column in columnsToCheck)
{
var value = row[column.ColumnName];
if (value is DBNull || !Classifier.IsBool(value.ToString()))
{
columnNotBooleanCount[column] += 1;
}
}
// Foreach 128th row, check that all columns are likely to be boolean
if ((rowCount & 127) == 0)
{
var allowedDirtiness = rowCount * DirtinessFactor.Value / 100f;
columnsToCheck = columnsToCheck.Where(c => columnNotBooleanCount[c] < allowedDirtiness).ToArray();
if (columnsToCheck.Length == 0)
{
return;
}
}
}
foreach (var column in columnsToCheck)
{
var allowedDirtiness = rowCount * DirtinessFactor.Value / 100f;
if (columnNotBooleanCount[column] < allowedDirtiness)
{
issueCollector.ReportIssue(new Issue(this, this.Severity)
{
Name = "Text Column Used for Boolean Values",
Context = new ColumnContext(column),
Description = new Description("The column '{0}' contains boolean values. Consider using another data type", column),
Severity = this.Severity
});
}
}
}
public override void Execute(Database database, IIssueCollector issueCollector, IProviderCollection providers)
{
if (database.Tables.Count == 0)
{
return;
}
float invalidThreshold = this.InvalidThreshold.Value / 100f;
int columnsTotal = database.Columns.Count;
INameConventionDetector naming;
if (this.NamingConventionRepresentation.Value == DataTypes.NamingConventionRepresentation.Markov)
{
naming = new MarkovConventionDetector(MarkovTolerance.Value / 100f);
}
else
{
naming = new TrieNameDetector(30);
}
var columnNames = new List <String>();
var importance = providers.GetProvider <SchemaProviders.ImportanceProvider>();
foreach (var column in database.Columns)
{
double tableRank = importance[column.Table];
int weight;
if (tableRank < 1)
{
weight = 1;
}
else
{
weight = (int)(tableRank);
}
//for (int i = 0; i < weight; i++)
// columnNames.Add(column.ColumnName);
columnNames.Add(column.ColumnName);
}
Regex numPat = new Regex("[0-9]");
columnNames.RemoveAll(name => numPat.Match(name).Success);
//Detect convention
bool detected = naming.DetectConvention(columnNames);
var invalidColumns = (from col in database.Columns
where naming.IsValid(col.ColumnName) == false
select col).ToList();
float percentInvalid = (float)invalidColumns.Count / columnsTotal;
if (detected == false || percentInvalid > invalidThreshold)
{
var issue = new Issue(this, NoNamingConventionSeverity.Value);
issue.Name = this.nonExisting;
issue.Context = new DatabaseContext(database);
issue.Description = new Description("Unable to find a naming convention for columns");
issueCollector.ReportIssue(issue);
}
else
{
//Raise an issue for all columns that don't use the convention
foreach (Column col in invalidColumns)
{
var issue = new Issue(this, this.DefaultSeverity.Value);
issue.Name = this.inconsistent;
issue.Context = new ColumnContext(col);
issue.Description = new Description("Column '{0}' in table {1} does not follow the naming convention", col, col.Table);
issueCollector.ReportIssue(issue);
}
}
//Tables
if (this.NamingConventionRepresentation.Value == DataTypes.NamingConventionRepresentation.Markov)
{
naming = new MarkovConventionDetector(MarkovTolerance.Value / 100f);
}
else
{
naming = new TrieNameDetector(30);
}
var tableNames = database.Tables.Select(t => t.TableName);
detected = naming.DetectConvention(tableNames);
var invalidTables = (from table in database.Tables
where naming.IsValid(table.TableName) == false
select table).ToList();
float percentInvalidTables = (float)invalidTables.Count / database.Tables.Count;
if (detected == false || percentInvalidTables > invalidThreshold)
{
var issue = new Issue(this, NoNamingConventionSeverity.Value);
issue.Name = this.nonExisting;
issue.Context = new DatabaseContext(database);
issue.Description = new Description("Unable to find a naming convention for tables");
issueCollector.ReportIssue(issue);
}
else
{
foreach (Table table in invalidTables)
{
var issue = new Issue(this, this.DefaultSeverity.Value);
issue.Name = this.inconsistent;
issue.Context = new TableContext(table);
issue.Description = new Description("Table name '{0}' does not follow the naming convention", table);
issueCollector.ReportIssue(issue);
}
}
}
