Added single column anonymization feature

segovoni · Oct 23, 2022 · 1922030 · 1922030
1 parent 97eb5a5
commit 1922030
Show file tree

Hide file tree

Showing 7 changed files with 174 additions and 8 deletions.
diff --git a/Packages/d103/sqlcmdcli.dproj b/Packages/d103/sqlcmdcli.dproj
@@ -112,8 +112,8 @@
         <VerInfo_Locale>1033</VerInfo_Locale>
         <VerInfo_IncludeVerInfo>true</VerInfo_IncludeVerInfo>
         <VerInfo_AutoIncVersion>true</VerInfo_AutoIncVersion>
-        <VerInfo_Keys>CompanyName=;FileDescription=Command-line utility for ad hoc, interactive execution of commands on SQL Server;FileVersion=1.0.0.67;InternalName=sqlcmdcli;LegalCopyright=;LegalTrademarks=;OriginalFilename=sqlcmdcli.exe;ProgramID=com.embarcadero.$(MSBuildProjectName);ProductName=$(MSBuildProjectName);ProductVersion=1.0;Comments=https://github.com/segovoni/sqlcmdcli</VerInfo_Keys>
-        <VerInfo_Build>67</VerInfo_Build>
+        <VerInfo_Keys>CompanyName=;FileDescription=Command-line utility for ad hoc, interactive execution of commands on SQL Server;FileVersion=1.0.0.77;InternalName=sqlcmdcli;LegalCopyright=;LegalTrademarks=;OriginalFilename=sqlcmdcli.exe;ProgramID=com.embarcadero.$(MSBuildProjectName);ProductName=$(MSBuildProjectName);ProductVersion=1.0;Comments=https://github.com/segovoni/sqlcmdcli</VerInfo_Keys>
+        <VerInfo_Build>77</VerInfo_Build>
     </PropertyGroup>
     <ItemGroup>
         <DelphiCompile Include="$(MainSource)">

diff --git a/Packages/d103/sqlcmdcli.dres b/Packages/d103/sqlcmdcli.dres
diff --git a/Sources/sqlcmdcli.AnonymizeDB.pas b/Sources/sqlcmdcli.AnonymizeDB.pas
@@ -22,8 +22,8 @@   TAnonymizeDB = class(TObject)
     class procedure DisableTriggers(const AConnection: TADOConnection;
       const ATriggers: TDictionary<string, string>);
   public
-    class procedure Run(const AServerName, ADatabaseName, AUserName, APassword: string;
-      const AVerbose: Boolean);
+    class procedure Run(const AServerName, ADatabaseName, AUserName, APassword,
+      ASchemaName, ATableName, AColumnName: string; const AVerbose: Boolean);
   end;
 
 implementation
@@ -174,7 +174,7 @@ class procedure TAnonymizeDB.EnableTriggers(const AConnection: TADOConnection;
 end;
 
 class procedure TAnonymizeDB.Run(const AServerName, ADatabaseName, AUserName,
-  APassword: string; const AVerbose: Boolean);
+  APassword, ASchemaName, ATableName, AColumnName: string; const AVerbose: Boolean);
 var
   LConnection: TADOConnection;
   LDBSchema: TDBSchema;
@@ -224,7 +224,24 @@ class procedure TAnonymizeDB.Run(const AServerName, ADatabaseName, AUserName,
         if (AVerbose) then
           TConsole.Log(Format('Extract schema for %s ...', [ADatabaseName]), Success, False);
 
-        LDBSchemaExtractor.ExtractSchema(stText);
+        if (Trim(ASchemaName) = '') and
+           (Trim(ATableName) = '') and
+           (Trim(AColumnName) = '') then
+        begin
+          LDBSchemaExtractor.ExtractSchema(stText);
+        end
+        else if (Trim(ASchemaName) <> '') and
+                (Trim(ATableName) <> '') and
+                (Trim(AColumnName) <> '') then
+        begin
+          LDBSchemaExtractor.ExtractSchema(ASchemaName, ATableName, AColumnName);
+        end
+        else begin
+          if (AVerbose) then
+            TConsole.Log('Invalid parametes', Error, True);
+          raise Exception.Create('Invalid parametes');
+        end;
+
         LDBSchema := LDBSchemaExtractor.DBSchema;
         //LDBSchemaIndex := LDBSchemaExtractor.DBSchemaIndex;
       finally

diff --git a/Sources/sqlcmdcli.Boot.pas b/Sources/sqlcmdcli.Boot.pas
@@ -157,11 +157,40 @@ class procedure TBootCLI.Boot;
   // Command: anonymizedb
   LCommand := TOptionsRegistry.RegisterCommand('anonymizedb', 'anondb',
     RS_CMD_ANONYMIZEDB_DESCRIPTION, RS_CMD_ANONYMIZEDB_INFO,
-    'anonymizedb -servername:<name> -databasename:<dbname> -username:<name> -password:<password>');
+    'anonymizedb -servername:<name> -databasename:<dbname> -username:<name> ' +
+    '-password:<password> -schemaname:<tableschema> -tablename:<tablename> ' +
+    '-columnname:<columnname>');
   LCommand.Examples.Add('anonymizedb -servername:MARCONI -databasename:AdventureWorks -username:sgovoni -password:royalbreeze489');
   LCommand.Examples.Add('anonymizedb -s:MARCONI -d:AdventureWorks -u:sgovoni -p:royalbreeze489');
   LCommand.Examples.Add('anondb -s:MARCONI -d:AdventureWorks -u:sgovoni -p:royalbreeze489');
 
+  // Option: "schemaname"
+  LOption := LCommand.RegisterOption<string>('schemaname', 'schema',
+    RS_CMD_ANONYMIZEDB_SCHEMANAMEINFO,
+    procedure(const AValue: string)
+    begin
+      TAnonymizeDBOptions.SchemaName := AValue
+    end);
+  LOption.Required := False;
+
+  // Option: "tablename"
+  LOption := LCommand.RegisterOption<string>('tablename', 'table',
+    RS_CMD_ANONYMIZEDB_TABLENAMEINFO,
+    procedure(const AValue: string)
+    begin
+      TAnonymizeDBOptions.TableName := AValue
+    end);
+  LOption.Required := False;
+
+  // Option: "columnname"
+  LOption := LCommand.RegisterOption<string>('columnname', 'column',
+    RS_CMD_ANONYMIZEDB_COLUMNNAMEINFO,
+    procedure(const AValue: string)
+    begin
+      TAnonymizeDBOptions.ColumnName := AValue
+    end);
+  LOption.Required := False;
+
   TCommandHandler.RegisterCommand('anonymizedb',
     procedure()
     begin
@@ -170,6 +199,9 @@ class procedure TBootCLI.Boot;
         TGlobalOptions.DatabaseName,
         TGlobalOptions.UserName,
         TGlobalOptions.Password,
+        TAnonymizeDBOptions.SchemaName,
+        TAnonymizeDBOptions.TableName,
+        TAnonymizeDBOptions.ColumnName,
         TGlobalOptions.Verbose);
     end);
 

diff --git a/Sources/sqlcmdcli.CommandOptions.pas b/Sources/sqlcmdcli.CommandOptions.pas
@@ -27,6 +27,13 @@   TAlterColumnOptions = class
     class var DataType: string;
   end;
 
+  TAnonymizeDBOptions = class
+  public
+    class var SchemaName: string;
+    class var TableName: string;
+    class var ColumnName: string;
+  end;
+
   TGlobalOptions = class
   public
     class var ServerName: string;

diff --git a/Sources/sqlcmdcli.ResourceStrings.pas b/Sources/sqlcmdcli.ResourceStrings.pas
@@ -40,6 +40,9 @@ interface
     'removing personally identifiable information from data sets, so that the ' +
     'people whom the data describe remain anonymous (source ' +
     'https://en.wikipedia.org/wiki/Data_anonymization)';
+  RS_CMD_ANONYMIZEDB_SCHEMANAMEINFO = 'The name of the table schema';
+  RS_CMD_ANONYMIZEDB_TABLENAMEINFO = 'The name of the table';
+  RS_CMD_ANONYMIZEDB_COLUMNNAMEINFO = 'The name of the column you want to modify';
 
   RS_CMD_ANONYMIZEDB_DISABLE_FK_START = 'Disable foreign key constraints on %s ...';
   RS_CMD_ANONYMIZEDB_DISABLE_FK_END = 'Foreign key constraints disabled successfully!';

diff --git a/Sources/sqlcmdcli.SchemaExtractor.pas b/Sources/sqlcmdcli.SchemaExtractor.pas
@@ -41,14 +41,16 @@   TSQLDBSchemaExtractor = class(TObject)
     function GetDBSchema: TDBSchema;
     //function GetDBSchemaIndex: TDBSchemaIndex;
     function GetSQLbySchemaType(ASchemaType: TSchemaType): string;
+    function GetSQLbySchema(const ASchemaName, ATableName, AColumnName: string): string;
   protected
     FConnection: TADOConnection;
     FDBSchema: TDBSchema;
     //FDBSchemaIndex: TDBSchemaIndex;
   public
     constructor Create(AConnection: TADOConnection);
     destructor Destroy; override;
-    procedure ExtractSchema(ASchemaType: TSchemaType);
+    procedure ExtractSchema(ASchemaType: TSchemaType); overload;
+    procedure ExtractSchema(const ASchemaName, ATableName, AColumnName: string); overload;
     property DBSchema: TDBSchema read GetDBSchema;
     //property DBSchemaIndex: TDBSchemaIndex read GetDBSchemaIndex;
   end;
@@ -78,6 +80,75 @@ destructor TSQLDBSchemaExtractor.Destroy;
   inherited;
 end;
 
+procedure TSQLDBSchemaExtractor.ExtractSchema(const ASchemaName, ATableName,
+  AColumnName: string);
+var
+  LQry: TADOQuery;
+  LCurrentTable: string;
+  LList: TObjectList<TSQLDBTableInfo>;
+begin
+  // Database schema extractor logic
+
+  LQry := TADOQuery.Create(nil);
+  LList := TObjectList<TSQLDBTableInfo>.Create();
+  try
+    LQry.Connection := FConnection;
+    LQry.SQL.Text := GetSQLbySchema(ASchemaName, ATableName, AColumnName);
+    LQry.Open;
+
+    //
+    if not (LQry.Eof) then
+      LCurrentTable :=
+        TSQLObjectNameFormatter.Format(LQry.FieldByName('TABLE_SCHEMA').AsString) +
+        '.' +
+        TSQLObjectNameFormatter.Format(LQry.FieldByName('TABLE_NAME').AsString);
+
+    while (not LQry.Eof) do
+    begin
+      if CompareText(LCurrentTable,
+                     TSQLObjectNameFormatter.Format(LQry.FieldByName('TABLE_SCHEMA').AsString) + '.' +
+                     TSQLObjectNameFormatter.Format(LQry.FieldByName('TABLE_NAME').AsString)) = 0 then
+      begin
+        LList.Add(TSQLDBTableInfo.Create(
+          TSQLObjectNameFormatter.Format(LQry.FieldByName('TABLE_SCHEMA').AsString),
+          TSQLObjectNameFormatter.Format(LQry.FieldByName('TABLE_NAME').AsString),
+          TSQLObjectNameFormatter.Format(LQry.FieldByName('COLUMN_NAME').AsString),
+          TSQLObjectNameFormatter.Format(LQry.FieldByName('DATA_TYPE').AsString),
+          LQry.FieldByName('MAX_LENGHT').AsInteger,
+          LQry.FieldByName('COLUMN_IDENTITY').AsBoolean
+                                         )
+                 )
+      end
+      else begin
+        FDBSchema.Add(LCurrentTable, LList);
+        LList := TObjectList<TSQLDBTableInfo>.Create();
+        LList.Add(TSQLDBTableInfo.Create(
+          TSQLObjectNameFormatter.Format(LQry.FieldByName('TABLE_SCHEMA').AsString),
+          TSQLObjectNameFormatter.Format(LQry.FieldByName('TABLE_NAME').AsString),
+          TSQLObjectNameFormatter.Format(LQry.FieldByName('COLUMN_NAME').AsString),
+          TSQLObjectNameFormatter.Format(LQry.FieldByName('DATA_TYPE').AsString),
+          LQry.FieldByName('MAX_LENGHT').AsInteger,
+          LQry.FieldByName('COLUMN_IDENTITY').AsBoolean
+                                         )
+                 );
+        LCurrentTable :=
+          TSQLObjectNameFormatter.Format(LQry.FieldByName('TABLE_SCHEMA').AsString) +
+          '.' +
+          TSQLObjectNameFormatter.Format(LQry.FieldByName('TABLE_NAME').AsString);
+      end;
+
+      LQry.Next;
+    end;
+
+    FDBSchema.Add(TSQLObjectNameFormatter.Format(LQry.FieldByName('TABLE_SCHEMA').AsString) +
+                                                 '.' +
+                                                 TSQLObjectNameFormatter.Format(LQry.FieldByName('TABLE_NAME').AsString), LList);
+
+  finally
+    FreeAndNil(LQry);
+  end;
+end;
+
 procedure TSQLDBSchemaExtractor.ExtractSchema(ASchemaType: TSchemaType);
 var
   LQry: TADOQuery;
@@ -169,6 +240,42 @@ function TSQLDBSchemaExtractor.GetDBSchema: TDictionary<string, TOBjectList<TSQL
   Result := FDBSchema;
 end;
 
+function TSQLDBSchemaExtractor.GetSQLbySchema(const ASchemaName, ATableName,
+  AColumnName: string): string;
+begin
+  Result :=
+    'SELECT ' +
+      'TABLE_SCHEMA = SCHEMA_NAME(T.schema_id) ' +
+      ',TABLE_NAME = T.name ' +
+      ',ORDINAL_POSITION = C.column_id ' +
+      ',COLUMN_NAME = C.name ' +
+      ',MAX_LENGHT = C.max_length ' +
+      ',DATA_TYPE = TYPE_NAME(C.system_type_id) ' +
+      ',COLUMN_IDENTITY = C.is_identity ' +
+      //',ROWS = P.rows ' +
+    'FROM ' +
+      'sys.tables AS T ' +
+    'JOIN ' +
+      'sys.columns AS C ON T.object_id=C.object_id ' +
+    //'JOIN ' +
+    //  'sys.partitions AS P on P.object_id=T.object_id ' +
+    'WHERE ' +
+      '(T.type = ''U'') ' +
+      //'AND (T.is_memory_optimized = 0) ' +
+      'AND (T.is_ms_shipped = 0) ' +
+      'AND (T.temporal_type = 0) ' +
+      'AND (T.is_replicated = 0) ' +
+      'AND (C.is_computed = 0) ' +
+      'AND (SCHEMA_NAME(T.schema_id) = ''' + ASchemaName + ''') ' +
+      'AND (T.name = ''' + ATableName + ''') ' +
+      'AND (C.name = ''' + AColumnName + ''') ' +
+    'ORDER BY ' +
+      'TABLE_SCHEMA ' +
+      ',TABLE_NAME ' +
+      ',COLUMN_NAME ' +
+      ',ORDINAL_POSITION';
+end;
+
 function TSQLDBSchemaExtractor.GetSQLbySchemaType(ASchemaType: TSchemaType): string;
 begin
   if (ASchemaType = stFull) then