source: trunk/sources/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Formatters/SymbolicDataAnalysisExpressionCSharpFormatter.cs @ 14826

Last change on this file since 14826 was 14826, checked in by gkronber, 6 months ago

#2650: merged the factors branch into trunk

File size: 14.0 KB
Line 
1#region License Information
2/* HeuristicLab
3 * Copyright (C) 2002-2016 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 System.Text.RegularExpressions;
28using HeuristicLab.Common;
29using HeuristicLab.Core;
30using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
31using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
32
33namespace HeuristicLab.Problems.DataAnalysis.Symbolic {
34  [Item("C# Symbolic Expression Tree Formatter", "A string formatter that converts symbolic expression trees to C# code.")]
35  [StorableClass]
36  public sealed class CSharpSymbolicExpressionTreeStringFormatter : NamedItem, ISymbolicExpressionTreeStringFormatter {
37    [StorableConstructor]
38    private CSharpSymbolicExpressionTreeStringFormatter(bool deserializing) : base(deserializing) { }
39    private CSharpSymbolicExpressionTreeStringFormatter(CSharpSymbolicExpressionTreeStringFormatter original, Cloner cloner) : base(original, cloner) { }
40    public CSharpSymbolicExpressionTreeStringFormatter()
41      : base() {
42      Name = ItemName;
43      Description = ItemDescription;
44    }
45    public override IDeepCloneable Clone(Cloner cloner) {
46      return new CSharpSymbolicExpressionTreeStringFormatter(this, cloner);
47    }
48
49    public string Format(ISymbolicExpressionTree symbolicExpressionTree) {
50      // skip root and start symbols
51      StringBuilder strBuilder = new StringBuilder();
52      GenerateHeader(strBuilder, symbolicExpressionTree);
53      FormatRecursively(symbolicExpressionTree.Root.GetSubtree(0).GetSubtree(0), strBuilder);
54      GenerateFooter(strBuilder);
55      return strBuilder.ToString();
56    }
57
58    private string VariableName2Identifier(string name) {     
59      /*
60       * identifier-start-character:
61       *    letter-character
62       *    _ (the underscore character U+005F)
63       *  identifier-part-characters:
64       *    identifier-part-character
65       *    identifier-part-characters   identifier-part-character
66       *  identifier-part-character:
67       *    letter-character
68       *    decimal-digit-character
69       *    connecting-character
70       *    combining-character
71       *    formatting-character
72       *  letter-character:
73       *    A Unicode character of classes Lu, Ll, Lt, Lm, Lo, or Nl
74       *    A unicode-escape-sequence representing a character of classes Lu, Ll, Lt, Lm, Lo, or Nl
75       *  combining-character:
76       *    A Unicode character of classes Mn or Mc
77       *    A unicode-escape-sequence representing a character of classes Mn or Mc
78       *  decimal-digit-character:
79       *    A Unicode character of the class Nd
80       *    A unicode-escape-sequence representing a character of the class Nd
81       *  connecting-character:
82       *    A Unicode character of the class Pc
83       *    A unicode-escape-sequence representing a character of the class Pc
84       *  formatting-character:
85       *    A Unicode character of the class Cf
86       *    A unicode-escape-sequence representing a character of the class Cf
87       */
88
89      var invalidIdentifierStarts = new Regex(@"[^_\p{Lu}\p{Ll}\p{Lt}\p{Lm}\p{Lo}\p{Nl}]");
90      var invalidIdentifierParts = new Regex(@"[^\p{Lu}\p{Ll}\p{Lt}\p{Lm}\p{Lo}\p{Nl}\p{Mn}\p{Mc}\p{Nd}\p{Pc}\p{Cf}]");
91      return "@" +
92        (invalidIdentifierStarts.IsMatch(name.Substring(0, 1)) ? "_" : "") + // prepend '_' if necessary
93        invalidIdentifierParts.Replace(name, "_");
94    }
95
96    private void FormatRecursively(ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
97      // TODO: adapt to interpreter semantics. The HL interpreter also allows Boolean operations on reals
98      if (node.Subtrees.Any()) {
99        if (node.Symbol is Addition) {
100          FormatOperator(node, "+", strBuilder);
101        } else if (node.Symbol is And) {
102          FormatOperator(node, "&&", strBuilder);
103        } else if (node.Symbol is Average) {
104          FormatFunction(node, "Average", strBuilder);
105        } else if (node.Symbol is Cosine) {
106          FormatFunction(node, "Math.Cos", strBuilder);
107        } else if (node.Symbol is Division) {
108          FormatDivision(node, strBuilder);
109        } else if (node.Symbol is Exponential) {
110          FormatFunction(node, "Math.Exp", strBuilder);
111        } else if (node.Symbol is GreaterThan) {
112          FormatOperator(node, ">", strBuilder);
113        } else if (node.Symbol is IfThenElse) {
114          FormatFunction(node, "EvaluateIf", strBuilder);
115        } else if (node.Symbol is LessThan) {
116          FormatOperator(node, "<", strBuilder);
117        } else if (node.Symbol is Logarithm) {
118          FormatFunction(node, "Math.Log", strBuilder);
119        } else if (node.Symbol is Multiplication) {
120          FormatOperator(node, "*", strBuilder);
121        } else if (node.Symbol is Not) {
122          FormatOperator(node, "!", strBuilder);
123        } else if (node.Symbol is Or) {
124          FormatOperator(node, "||", strBuilder);
125        } else if (node.Symbol is Xor) {
126          FormatOperator(node, "^", strBuilder);
127        } else if (node.Symbol is Sine) {
128          FormatFunction(node, "Math.Sin", strBuilder);
129        } else if (node.Symbol is Subtraction) {
130          FormatSubtraction(node, strBuilder);
131        } else if (node.Symbol is Tangent) {
132          FormatFunction(node, "Math.Tan", strBuilder);
133        } else if (node.Symbol is Square) {
134          FormatSquare(node, strBuilder);
135        } else if (node.Symbol is SquareRoot) {
136          FormatFunction(node, "Math.Sqrt", strBuilder);
137        } else if (node.Symbol is Power) {
138          FormatFunction(node, "Math.Pow", strBuilder);
139        } else if (node.Symbol is Root) {
140          FormatRoot(node, strBuilder);
141        } else {
142          throw new NotSupportedException("Formatting of symbol: " + node.Symbol + " not supported for C# symbolic expression tree formatter.");
143        }
144      } else {
145        if (node is VariableTreeNode) {
146          var varNode = node as VariableTreeNode;
147          strBuilder.AppendFormat("{0} * {1}", VariableName2Identifier(varNode.VariableName), varNode.Weight.ToString("g17", CultureInfo.InvariantCulture));
148        } else if (node is ConstantTreeNode) {
149          var constNode = node as ConstantTreeNode;
150          strBuilder.Append(constNode.Value.ToString("g17", CultureInfo.InvariantCulture));
151        } else if (node.Symbol is FactorVariable) {
152          var factorNode = node as FactorVariableTreeNode;
153          FormatFactor(factorNode, strBuilder);
154        } else if (node.Symbol is BinaryFactorVariable) {
155          var binFactorNode = node as BinaryFactorVariableTreeNode;
156          FormatBinaryFactor(binFactorNode, strBuilder);
157        } else {
158          throw new NotSupportedException("Formatting of symbol: " + node.Symbol + " not supported for C# symbolic expression tree formatter.");
159        }
160      }
161    }
162
163    private void FormatFactor(FactorVariableTreeNode node, StringBuilder strBuilder) {
164      strBuilder.AppendFormat("EvaluateFactor({0}, new [] {{ {1} }}, new [] {{ {2} }})", VariableName2Identifier(node.VariableName),
165        string.Join(",", node.Symbol.GetVariableValues(node.VariableName).Select(name => "\"" + name + "\"")), string.Join(",", node.Weights.Select(v => v.ToString(CultureInfo.InvariantCulture))));
166    }
167
168    private void FormatBinaryFactor(BinaryFactorVariableTreeNode node, StringBuilder strBuilder) {
169      strBuilder.AppendFormat(CultureInfo.InvariantCulture, "EvaluateBinaryFactor({0}, \"{1}\", {2})", VariableName2Identifier(node.VariableName), node.VariableValue, node.Weight);
170    }
171
172    private void FormatSquare(ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
173      strBuilder.Append("Math.Pow(");
174      FormatRecursively(node.GetSubtree(0), strBuilder);
175      strBuilder.Append(", 2)");
176    }
177
178    private void FormatRoot(ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
179      strBuilder.Append("Math.Pow(");
180      FormatRecursively(node.GetSubtree(0), strBuilder);
181      strBuilder.Append(", 1.0 / (");
182      FormatRecursively(node.GetSubtree(1), strBuilder);
183      strBuilder.Append("))");
184    }
185
186    private void FormatDivision(ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
187      if (node.SubtreeCount == 1) {
188        strBuilder.Append("1.0 / ");
189        FormatRecursively(node.GetSubtree(0), strBuilder);
190      } else {
191        FormatRecursively(node.GetSubtree(0), strBuilder);
192        strBuilder.Append("/ (");
193        for (int i = 1; i < node.SubtreeCount; i++) {
194          if (i > 1) strBuilder.Append(" * ");
195          FormatRecursively(node.GetSubtree(i), strBuilder);
196        }
197        strBuilder.Append(")");
198      }
199    }
200
201    private void FormatSubtraction(ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
202      if (node.SubtreeCount == 1) {
203        strBuilder.Append("-");
204        FormatRecursively(node.GetSubtree(0), strBuilder);
205        return;
206      }
207      //Default case: more than 1 child
208      FormatOperator(node, "-", strBuilder);
209    }
210
211    private void FormatOperator(ISymbolicExpressionTreeNode node, string symbol, StringBuilder strBuilder) {
212      strBuilder.Append("(");
213      foreach (var child in node.Subtrees) {
214        FormatRecursively(child, strBuilder);
215        if (child != node.Subtrees.Last())
216          strBuilder.Append(" " + symbol + " ");
217      }
218      strBuilder.Append(")");
219    }
220
221    private void FormatFunction(ISymbolicExpressionTreeNode node, string function, StringBuilder strBuilder) {
222      strBuilder.Append(function + "(");
223      foreach (var child in node.Subtrees) {
224        FormatRecursively(child, strBuilder);
225        if (child != node.Subtrees.Last())
226          strBuilder.Append(", ");
227      }
228      strBuilder.Append(")");
229    }
230
231    private void GenerateHeader(StringBuilder strBuilder, ISymbolicExpressionTree symbolicExpressionTree) {
232      strBuilder.AppendLine("using System;");
233      strBuilder.AppendLine("using System.Linq;" + Environment.NewLine);
234      strBuilder.AppendLine("namespace HeuristicLab.Models {");
235      strBuilder.AppendLine("public static class Model {");
236      GenerateAverageSource(strBuilder);
237      GenerateIfThenElseSource(strBuilder);
238      GenerateFactorSource(strBuilder);
239      GenerateBinaryFactorSource(strBuilder);
240      strBuilder.Append(Environment.NewLine + "public static double Evaluate (");
241
242      // here we don't have access to problemData to determine the type for each variable (double/string) therefore we must distinguish based on the symbol type
243      HashSet<string> doubleVarNames = new HashSet<string>();
244      foreach (var node in symbolicExpressionTree.IterateNodesPostfix().Where(x => x is VariableTreeNode || x is VariableConditionTreeNode)) {
245        doubleVarNames.Add(((IVariableTreeNode)node).VariableName);
246      }
247
248      HashSet<string> stringVarNames = new HashSet<string>();
249      foreach (var node in symbolicExpressionTree.IterateNodesPostfix().Where(x => x is BinaryFactorVariableTreeNode || x is FactorVariableTreeNode)) {
250        stringVarNames.Add(((IVariableTreeNode)node).VariableName);
251      }
252
253      var orderedNames = stringVarNames.OrderBy(n => n, new NaturalStringComparer()).Select(n => "string " + VariableName2Identifier(n) + " /* " + n + " */");
254      strBuilder.Append(string.Join(", ", orderedNames));
255
256      if (stringVarNames.Any() && doubleVarNames.Any())
257        strBuilder.AppendLine(",");
258      orderedNames = doubleVarNames.OrderBy(n => n, new NaturalStringComparer()).Select(n => "double " + VariableName2Identifier(n) + " /* " + n + " */");
259      strBuilder.Append(string.Join(", ", orderedNames));
260
261
262      strBuilder.AppendLine(") {");
263      strBuilder.Append("double result = ");
264    }
265
266    private void GenerateFooter(StringBuilder strBuilder) {
267      strBuilder.AppendLine(";");
268
269      strBuilder.AppendLine("return result;");
270      strBuilder.AppendLine("}");
271      strBuilder.AppendLine("}");
272      strBuilder.AppendLine("}");
273    }
274
275    private void GenerateAverageSource(StringBuilder strBuilder) {
276      strBuilder.AppendLine("private static double Average(params double[] values) {");
277      strBuilder.AppendLine("  return values.Average();");
278      strBuilder.AppendLine("}");
279    }
280
281    private void GenerateIfThenElseSource(StringBuilder strBuilder) {
282      strBuilder.AppendLine("private static double EvaluateIf(bool condition, double then, double @else) {");
283      strBuilder.AppendLine("   return condition ? then : @else;");
284      strBuilder.AppendLine("}");
285    }
286
287    private void GenerateFactorSource(StringBuilder strBuilder) {
288      strBuilder.AppendLine("private static double EvaluateFactor(string factorValue, string[] factorValues, double[] constants) {");
289      strBuilder.AppendLine("   for(int i=0;i<factorValues.Length;i++) " +
290                            "      if(factorValues[i] == factorValue) return constants[i];" +
291                            "   throw new ArgumentException();");
292      strBuilder.AppendLine("}");
293    }
294
295    private void GenerateBinaryFactorSource(StringBuilder strBuilder) {
296      strBuilder.AppendLine("private static double EvaluateBinaryFactor(string factorValue, string targetValue, double weight) {");
297      strBuilder.AppendLine("  return factorValue == targetValue ? weight : 0.0;");
298      strBuilder.AppendLine("}");
299    }
300
301  }
302}
Note: See TracBrowser for help on using the repository browser.