Free cookie consent management tool by TermsFeed Policy Generator

source: trunk/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Formatters/SymbolicDataAnalysisExpressionCSharpFormatter.cs @ 16702

Last change on this file since 16702 was 16702, checked in by bburlacu, 5 years ago

#2866: Add tanh to all formatters except the TSQLExpressionFormatter, since TSQL does not have a TANH function (could be approximated but ugly).

File size: 14.1 KB
Line 
1#region License Information
2/* HeuristicLab
3 * Copyright (C) 2002-2019 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 HEAL.Attic;
29using HeuristicLab.Common;
30using HeuristicLab.Core;
31using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
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  [StorableType("88298836-6087-405A-9354-D4E6864887EB")]
36  public sealed class CSharpSymbolicExpressionTreeStringFormatter : NamedItem, ISymbolicExpressionTreeStringFormatter {
37    [StorableConstructor]
38    private CSharpSymbolicExpressionTreeStringFormatter(StorableConstructorFlag _) : base(_) { }
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 HyperbolicTangent) {
134          FormatFunction(node, "Math.Tanh", strBuilder);
135        } else if (node.Symbol is Square) {
136          FormatSquare(node, strBuilder);
137        } else if (node.Symbol is SquareRoot) {
138          FormatFunction(node, "Math.Sqrt", strBuilder);
139        } else if (node.Symbol is Power) {
140          FormatFunction(node, "Math.Pow", strBuilder);
141        } else if (node.Symbol is Root) {
142          FormatRoot(node, strBuilder);
143        } else {
144          throw new NotSupportedException("Formatting of symbol: " + node.Symbol + " not supported for C# symbolic expression tree formatter.");
145        }
146      } else {
147        if (node is VariableTreeNode) {
148          var varNode = node as VariableTreeNode;
149          strBuilder.AppendFormat("{0} * {1}", VariableName2Identifier(varNode.VariableName), varNode.Weight.ToString("g17", CultureInfo.InvariantCulture));
150        } else if (node is ConstantTreeNode) {
151          var constNode = node as ConstantTreeNode;
152          strBuilder.Append(constNode.Value.ToString("g17", CultureInfo.InvariantCulture));
153        } else if (node.Symbol is FactorVariable) {
154          var factorNode = node as FactorVariableTreeNode;
155          FormatFactor(factorNode, strBuilder);
156        } else if (node.Symbol is BinaryFactorVariable) {
157          var binFactorNode = node as BinaryFactorVariableTreeNode;
158          FormatBinaryFactor(binFactorNode, strBuilder);
159        } else {
160          throw new NotSupportedException("Formatting of symbol: " + node.Symbol + " not supported for C# symbolic expression tree formatter.");
161        }
162      }
163    }
164
165    private void FormatFactor(FactorVariableTreeNode node, StringBuilder strBuilder) {
166      strBuilder.AppendFormat("EvaluateFactor({0}, new [] {{ {1} }}, new [] {{ {2} }})", VariableName2Identifier(node.VariableName),
167        string.Join(",", node.Symbol.GetVariableValues(node.VariableName).Select(name => "\"" + name + "\"")), string.Join(",", node.Weights.Select(v => v.ToString(CultureInfo.InvariantCulture))));
168    }
169
170    private void FormatBinaryFactor(BinaryFactorVariableTreeNode node, StringBuilder strBuilder) {
171      strBuilder.AppendFormat(CultureInfo.InvariantCulture, "EvaluateBinaryFactor({0}, \"{1}\", {2})", VariableName2Identifier(node.VariableName), node.VariableValue, node.Weight);
172    }
173
174    private void FormatSquare(ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
175      strBuilder.Append("Math.Pow(");
176      FormatRecursively(node.GetSubtree(0), strBuilder);
177      strBuilder.Append(", 2)");
178    }
179
180    private void FormatRoot(ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
181      strBuilder.Append("Math.Pow(");
182      FormatRecursively(node.GetSubtree(0), strBuilder);
183      strBuilder.Append(", 1.0 / (");
184      FormatRecursively(node.GetSubtree(1), strBuilder);
185      strBuilder.Append("))");
186    }
187
188    private void FormatDivision(ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
189      if (node.SubtreeCount == 1) {
190        strBuilder.Append("1.0 / ");
191        FormatRecursively(node.GetSubtree(0), strBuilder);
192      } else {
193        FormatRecursively(node.GetSubtree(0), strBuilder);
194        strBuilder.Append("/ (");
195        for (int i = 1; i < node.SubtreeCount; i++) {
196          if (i > 1) strBuilder.Append(" * ");
197          FormatRecursively(node.GetSubtree(i), strBuilder);
198        }
199        strBuilder.Append(")");
200      }
201    }
202
203    private void FormatSubtraction(ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
204      if (node.SubtreeCount == 1) {
205        strBuilder.Append("-");
206        FormatRecursively(node.GetSubtree(0), strBuilder);
207        return;
208      }
209      //Default case: more than 1 child
210      FormatOperator(node, "-", strBuilder);
211    }
212
213    private void FormatOperator(ISymbolicExpressionTreeNode node, string symbol, StringBuilder strBuilder) {
214      strBuilder.Append("(");
215      foreach (var child in node.Subtrees) {
216        FormatRecursively(child, strBuilder);
217        if (child != node.Subtrees.Last())
218          strBuilder.Append(" " + symbol + " ");
219      }
220      strBuilder.Append(")");
221    }
222
223    private void FormatFunction(ISymbolicExpressionTreeNode node, string function, StringBuilder strBuilder) {
224      strBuilder.Append(function + "(");
225      foreach (var child in node.Subtrees) {
226        FormatRecursively(child, strBuilder);
227        if (child != node.Subtrees.Last())
228          strBuilder.Append(", ");
229      }
230      strBuilder.Append(")");
231    }
232
233    private void GenerateHeader(StringBuilder strBuilder, ISymbolicExpressionTree symbolicExpressionTree) {
234      strBuilder.AppendLine("using System;");
235      strBuilder.AppendLine("using System.Linq;" + Environment.NewLine);
236      strBuilder.AppendLine("namespace HeuristicLab.Models {");
237      strBuilder.AppendLine("public static class Model {");
238      GenerateAverageSource(strBuilder);
239      GenerateIfThenElseSource(strBuilder);
240      GenerateFactorSource(strBuilder);
241      GenerateBinaryFactorSource(strBuilder);
242      strBuilder.Append(Environment.NewLine + "public static double Evaluate (");
243
244      // 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
245      HashSet<string> doubleVarNames = new HashSet<string>();
246      foreach (var node in symbolicExpressionTree.IterateNodesPostfix().Where(x => x is VariableTreeNode || x is VariableConditionTreeNode)) {
247        doubleVarNames.Add(((IVariableTreeNode)node).VariableName);
248      }
249
250      HashSet<string> stringVarNames = new HashSet<string>();
251      foreach (var node in symbolicExpressionTree.IterateNodesPostfix().Where(x => x is BinaryFactorVariableTreeNode || x is FactorVariableTreeNode)) {
252        stringVarNames.Add(((IVariableTreeNode)node).VariableName);
253      }
254
255      var orderedNames = stringVarNames.OrderBy(n => n, new NaturalStringComparer()).Select(n => "string " + VariableName2Identifier(n) + " /* " + n + " */");
256      strBuilder.Append(string.Join(", ", orderedNames));
257
258      if (stringVarNames.Any() && doubleVarNames.Any())
259        strBuilder.AppendLine(",");
260      orderedNames = doubleVarNames.OrderBy(n => n, new NaturalStringComparer()).Select(n => "double " + VariableName2Identifier(n) + " /* " + n + " */");
261      strBuilder.Append(string.Join(", ", orderedNames));
262
263
264      strBuilder.AppendLine(") {");
265      strBuilder.Append("double result = ");
266    }
267
268    private void GenerateFooter(StringBuilder strBuilder) {
269      strBuilder.AppendLine(";");
270
271      strBuilder.AppendLine("return result;");
272      strBuilder.AppendLine("}");
273      strBuilder.AppendLine("}");
274      strBuilder.AppendLine("}");
275    }
276
277    private void GenerateAverageSource(StringBuilder strBuilder) {
278      strBuilder.AppendLine("private static double Average(params double[] values) {");
279      strBuilder.AppendLine("  return values.Average();");
280      strBuilder.AppendLine("}");
281    }
282
283    private void GenerateIfThenElseSource(StringBuilder strBuilder) {
284      strBuilder.AppendLine("private static double EvaluateIf(bool condition, double then, double @else) {");
285      strBuilder.AppendLine("   return condition ? then : @else;");
286      strBuilder.AppendLine("}");
287    }
288
289    private void GenerateFactorSource(StringBuilder strBuilder) {
290      strBuilder.AppendLine("private static double EvaluateFactor(string factorValue, string[] factorValues, double[] constants) {");
291      strBuilder.AppendLine("   for(int i=0;i<factorValues.Length;i++) " +
292                            "      if(factorValues[i] == factorValue) return constants[i];" +
293                            "   throw new ArgumentException();");
294      strBuilder.AppendLine("}");
295    }
296
297    private void GenerateBinaryFactorSource(StringBuilder strBuilder) {
298      strBuilder.AppendLine("private static double EvaluateBinaryFactor(string factorValue, string targetValue, double weight) {");
299      strBuilder.AppendLine("  return factorValue == targetValue ? weight : 0.0;");
300      strBuilder.AppendLine("}");
301    }
302
303  }
304}
Note: See TracBrowser for help on using the repository browser.