source: branches/2911_TSQLSymbolicExpressionFormatter/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Formatters/TSQLExpressionFormatter.cs @ 15901

Last change on this file since 15901 was 15901, checked in by fbaching, 4 years ago

#2911: change to lowercase string identifier

File size: 12.9 KB
Line 
1#region License Information
2/* HeuristicLab
3 * Copyright (C) 2002-2018 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
4 *
5 * This file is part of HeuristicLab.
6 *
7 * HeuristicLab is free software: you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation, either version 3 of the License, or
10 * (at your option) any later version.
11 *
12 * HeuristicLab is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
15 * GNU General Public License for more details.
16 *
17 * You should have received a copy of the GNU General Public License
18 * along with HeuristicLab. If not, see <http://www.gnu.org/licenses/>.
19 */
20#endregion
21
22using System;
23using System.Collections.Generic;
24using System.Globalization;
25using System.Linq;
26using System.Text;
27using HeuristicLab.Common;
28using HeuristicLab.Core;
29using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
30using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
31
32namespace HeuristicLab.Problems.DataAnalysis.Symbolic {
33  internal static class StringBuilderExtensions {
34    internal static void AppendIndented(this StringBuilder strBuilder, int level, string text) {
35      strBuilder.Append(new string(' ', level * 2));
36      strBuilder.Append(text);
37    }
38    internal static void AppendLineIndented(this StringBuilder strBuilder, int level, string text) {
39      strBuilder.Append(new string(' ', level * 2));
40      strBuilder.AppendLine(text);
41    }
42  }
43
44  [Item("TSQL String Formatter", "String formatter for string representations of symbolic data analysis expressions in TSQL syntax.")]
45  [StorableClass]
46  public sealed class TSQLExpressionFormatter : NamedItem, ISymbolicExpressionTreeStringFormatter {
47    [StorableConstructor]
48    private TSQLExpressionFormatter(bool deserializing) : base(deserializing) { }
49    private TSQLExpressionFormatter(TSQLExpressionFormatter original, Cloner cloner) : base(original, cloner) { }
50    public TSQLExpressionFormatter()
51      : base() {
52      Name = ItemName;
53      Description = ItemDescription;
54    }
55    public override IDeepCloneable Clone(Cloner cloner) {
56      return new TSQLExpressionFormatter(this, cloner);
57    }
58
59    public string Format(ISymbolicExpressionTree symbolicExpressionTree) {
60      // skip root and start symbols
61      StringBuilder strBuilder = new StringBuilder();
62      GenerateHeader(strBuilder, symbolicExpressionTree);
63
64      //generate function body
65      FormatRecursively(1, symbolicExpressionTree.Root.GetSubtree(0).GetSubtree(0), strBuilder);
66
67      GenerateFooter(strBuilder);
68      return strBuilder.ToString();
69    }
70
71    private void GenerateHeader(StringBuilder strBuilder, ISymbolicExpressionTree symbolicExpressionTree) {
72      HashSet<string> floatVarNames = new HashSet<string>();
73      foreach (var node in symbolicExpressionTree.IterateNodesPostfix().Where(x => x is VariableTreeNode || x is VariableConditionTreeNode)) {
74        floatVarNames.Add(((IVariableTreeNode)node).VariableName);
75      }
76      var sortedFloatIdentifiers = floatVarNames.OrderBy(n => n, new NaturalStringComparer()).Select(n => VariableName2Identifier(n)).ToList();
77
78      HashSet<string> varcharVarNames = new HashSet<string>();
79      foreach (var node in symbolicExpressionTree.IterateNodesPostfix().Where(x => x is BinaryFactorVariableTreeNode || x is FactorVariableTreeNode)) {
80        varcharVarNames.Add(((IVariableTreeNode)node).VariableName);
81      }
82      var sortedVarcharIdentifiers = varcharVarNames.OrderBy(n => n, new NaturalStringComparer()).Select(n => VariableName2Identifier(n)).ToList();
83
84      //Generate comment and instructions
85      strBuilder.Append("-- generated. created function can be used like 'SELECT dbo.REGRESSIONMODEL(");
86      strBuilder.Append(string.Join(", ", sortedVarcharIdentifiers));
87      if (varcharVarNames.Any() && floatVarNames.Any())
88        strBuilder.Append(",");
89      strBuilder.Append(string.Join(", ", sortedFloatIdentifiers));
90      strBuilder.AppendLine(")'");
91      strBuilder.AppendLine("-- use the expression after the RETURN statement if you want to incorporate the model in a query without creating a function.");
92
93      //Generate function header
94      strBuilder.Append("CREATE FUNCTION dbo.REGRESSIONMODEL(");
95      strBuilder.Append(string.Join(", ", sortedVarcharIdentifiers.Select(n => string.Format("{0} NVARCHAR(max)",n))));
96      if (varcharVarNames.Any() && floatVarNames.Any())
97        strBuilder.Append(",");
98      strBuilder.Append(string.Join(", ", sortedFloatIdentifiers.Select(n => string.Format("{0} FLOAT",n))));
99      strBuilder.AppendLine(")");
100
101      //start function body
102      strBuilder.AppendLine("RETURNS FLOAT");
103      strBuilder.AppendLine("BEGIN");
104
105      //add variable declaration for convenience
106      strBuilder.AppendLineIndented(1, "-- added variable declaration for convenience");
107      foreach (var name in sortedVarcharIdentifiers)
108        strBuilder.AppendLineIndented(1, string.Format("-- DECLARE {0} NVARCHAR(max) = ''", name));
109      foreach (var name in sortedFloatIdentifiers)
110        strBuilder.AppendLineIndented(1, string.Format("-- DECLARE {0} FLOAT = 0.0", name));
111      strBuilder.AppendLineIndented(1, "-- SELECT");
112      strBuilder.AppendLine("RETURN ");
113    }
114
115    private void FormatRecursively(int level, ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
116      if (node.Subtrees.Any()) {
117        if (node.Symbol is Addition) {
118          FormatOperator(level, node, "+", strBuilder);
119        } else if (node.Symbol is And) {
120          FormatOperator(level, node, "AND", strBuilder);
121        } else if (node.Symbol is Average) {
122          FormatAverage(level, node, strBuilder);
123        } else if (node.Symbol is Cosine) {
124          FormatFunction(level, node, "COS", strBuilder);
125        } else if (node.Symbol is Division) {
126          FormatDivision(level, node, strBuilder);
127        } else if (node.Symbol is Exponential) {
128          FormatFunction(level, node, "EXP", strBuilder);
129        } else if (node.Symbol is GreaterThan) {
130          FormatOperator(level, node, ">", strBuilder);
131        } else if (node.Symbol is IfThenElse) {
132          FormatIfThenElse(level, node, strBuilder);
133        } else if (node.Symbol is LessThan) {
134          FormatOperator(level, node, "<", strBuilder);
135        } else if (node.Symbol is Logarithm) {
136          FormatFunction(level, node, "LOG", strBuilder);
137        } else if (node.Symbol is Multiplication) {
138          FormatOperator(level, node, "*", strBuilder);
139        } else if (node.Symbol is Not) {
140          FormatOperator(level, node, "NOT LIKE", strBuilder);
141        } else if (node.Symbol is Or) {
142          FormatOperator(level, node, "OR", strBuilder);
143        } else if (node.Symbol is Xor) {
144          throw new NotSupportedException(string.Format("Symbol {0} not yet supported.", node.Symbol.GetType().Name));
145        } else if (node.Symbol is Sine) {
146          FormatFunction(level, node, "SIN", strBuilder);
147        } else if (node.Symbol is Subtraction) {
148          FormatSubtraction(level, node, strBuilder);
149        } else if (node.Symbol is Tangent) {
150          FormatFunction(level, node, "TAN", strBuilder);
151        } else if (node.Symbol is Square) {
152          FormatFunction(level, node, "SQUARE", strBuilder);
153        } else if (node.Symbol is SquareRoot) {
154          FormatFunction(level, node, "SQRT", strBuilder);
155        } else if (node.Symbol is Power) {
156          FormatFunction(level, node, "POWER", strBuilder);
157        } else if (node.Symbol is Root) {
158          FormatRoot(level, node, strBuilder);
159        } else {
160          throw new NotSupportedException("Formatting of symbol: " + node.Symbol + " not supported for TSQL symbolic expression tree formatter.");
161        }
162      } else {
163        if (node is VariableTreeNode) {
164          var varNode = node as VariableTreeNode;
165          strBuilder.AppendFormat("{0} * {1}", VariableName2Identifier(varNode.VariableName), varNode.Weight.ToString("g17", CultureInfo.InvariantCulture));
166        } else if (node is ConstantTreeNode) {
167          var constNode = node as ConstantTreeNode;
168          strBuilder.Append(constNode.Value.ToString("g17", CultureInfo.InvariantCulture));
169        } else if (node.Symbol is FactorVariable) {
170          var factorNode = node as FactorVariableTreeNode;
171          FormatFactor(level, factorNode, strBuilder);
172        } else if (node.Symbol is BinaryFactorVariable) {
173          var binFactorNode = node as BinaryFactorVariableTreeNode;
174          throw new NotSupportedException(string.Format("Symbol {0} not yet supported.", node.Symbol.GetType().Name));
175        } else {
176          throw new NotSupportedException("Formatting of symbol: " + node.Symbol + " not supported for TSQL symbolic expression tree formatter.");
177        }
178      }
179    }
180
181    private void FormatIfThenElse(int level, ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
182      strBuilder.Append("CASE ISNULL((SELECT 1 WHERE");
183      FormatRecursively(level,node.GetSubtree(0), strBuilder);
184      strBuilder.AppendLine("),0)");
185      strBuilder.AppendIndented(level,"WHEN 1 THEN ");
186      FormatRecursively(level, node.GetSubtree(1), strBuilder);
187      strBuilder.AppendLine();
188      strBuilder.AppendIndented(level, "WHEN 0 THEN ");
189      FormatRecursively(level, node.GetSubtree(2), strBuilder);
190      strBuilder.AppendLine();
191      strBuilder.AppendIndented(level, "END");
192    }
193
194    private void FormatAverage(int level, ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
195      strBuilder.Append("(");
196      foreach (var child in node.Subtrees) {
197        FormatRecursively(level, child, strBuilder);
198        if (child != node.Subtrees.Last())
199          strBuilder.Append(" + ");
200      }
201      strBuilder.AppendFormat(") / {0}", node.Subtrees.Count());
202    }
203
204    private string VariableName2Identifier(string variableName) {
205      return "@"+variableName.Replace(' ', '_');
206    }
207
208    private void GenerateFooter(StringBuilder strBuilder) {
209      strBuilder.Append(Environment.NewLine);
210      strBuilder.AppendLine("END");
211    }
212
213
214    private void FormatOperator(int level, ISymbolicExpressionTreeNode node, string symbol, StringBuilder strBuilder) {
215      strBuilder.Append("(");
216      foreach (var child in node.Subtrees) {
217        FormatRecursively(level, child, strBuilder);
218        if (child != node.Subtrees.Last())
219          strBuilder.Append(" " + symbol + " ");
220      }
221      strBuilder.Append(")");
222    }
223
224    private void FormatFunction(int level, ISymbolicExpressionTreeNode node, string function, StringBuilder strBuilder) {
225      strBuilder.Append(function + "(");
226      foreach (var child in node.Subtrees) {
227        FormatRecursively(level++, child, strBuilder);
228        if (child != node.Subtrees.Last())
229          strBuilder.Append(", ");
230      }
231      strBuilder.Append(")");
232    }
233
234    private void FormatDivision(int level, ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
235      if (node.SubtreeCount == 1) {
236        strBuilder.Append("1.0 / ");
237        FormatRecursively(level, node.GetSubtree(0), strBuilder);
238      } else {
239        FormatRecursively(level, node.GetSubtree(0), strBuilder);
240        strBuilder.Append("/ (");
241        for (int i = 1; i < node.SubtreeCount; i++) {
242          if (i > 1) strBuilder.Append(" * ");
243          FormatRecursively(level, node.GetSubtree(i), strBuilder);
244        }
245        strBuilder.Append(")");
246      }
247    }
248
249    private void FormatSubtraction(int level, ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
250      if (node.SubtreeCount == 1) {
251        strBuilder.Append("-");
252        FormatRecursively(level, node.GetSubtree(0), strBuilder);
253        return;
254      }
255      //Default case: more than 1 child
256      FormatOperator(level, node, "-", strBuilder);
257    }
258
259    private void FormatRoot(int level, ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
260      strBuilder.AppendLine("POWER(");
261      FormatRecursively(level, node.GetSubtree(0), strBuilder);
262      strBuilder.AppendLineIndented(level, " , 1.0 / (");
263      FormatRecursively(level, node.GetSubtree(1), strBuilder);
264      strBuilder.AppendLineIndented(level, "))");
265    }
266
267    private void FormatFactor(int level, FactorVariableTreeNode node, StringBuilder strBuilder) {
268      strBuilder.AppendLine("( ");
269      strBuilder.AppendLineIndented(level + 1, string.Format("CASE {0}", VariableName2Identifier(node.VariableName)));
270      foreach (var name in node.Symbol.GetVariableValues(node.VariableName)) {
271        strBuilder.AppendLineIndented(level + 2, string.Format("WHEN '{0}' THEN {1}", name, node.GetValue(name).ToString(CultureInfo.InvariantCulture)));
272      }
273      strBuilder.AppendLineIndented(level + 1, "ELSE NULL END");
274      strBuilder.AppendIndented(level, ")");
275    }
276
277  }
278}
Note: See TracBrowser for help on using the repository browser.