Free cookie consent management tool by TermsFeed Policy Generator

source: stable/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Formatters/SymbolicDataAnalysisExpressionCSharpFormatter.cs @ 17103

Last change on this file since 17103 was 17103, checked in by mkommend, 5 years ago

#2915: Merged 16675, 16802, 16905, 17028 into stable.

File size: 15.3 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 Cube) {
140          FormatPower(node, strBuilder, "3");
141        } else if (node.Symbol is CubeRoot) {
142          strBuilder.Append("Cbrt(");
143          FormatRecursively(node.GetSubtree(0), strBuilder);
144          strBuilder.Append(")");
145        } else if (node.Symbol is Power) {
146          FormatFunction(node, "Math.Pow", strBuilder);
147        } else if (node.Symbol is Root) {
148          FormatRoot(node, strBuilder);
149        } else if (node.Symbol is Absolute) {
150          FormatFunction(node, "Math.Abs", strBuilder);
151        } else if (node.Symbol is AnalyticQuotient) {
152          strBuilder.Append("(");
153          FormatRecursively(node.GetSubtree(0), strBuilder);
154          strBuilder.Append(" / Math.Sqrt(1 + Math.Pow(");
155          FormatRecursively(node.GetSubtree(1), strBuilder);
156          strBuilder.Append(" , 2) )");
157        } else {
158          throw new NotSupportedException("Formatting of symbol: " + node.Symbol + " not supported for C# symbolic expression tree formatter.");
159        }
160      } else {
161        if (node is VariableTreeNode) {
162          var varNode = node as VariableTreeNode;
163          strBuilder.AppendFormat("{0} * {1}", VariableName2Identifier(varNode.VariableName), varNode.Weight.ToString("g17", CultureInfo.InvariantCulture));
164        } else if (node is ConstantTreeNode) {
165          var constNode = node as ConstantTreeNode;
166          strBuilder.Append(constNode.Value.ToString("g17", CultureInfo.InvariantCulture));
167        } else if (node.Symbol is FactorVariable) {
168          var factorNode = node as FactorVariableTreeNode;
169          FormatFactor(factorNode, strBuilder);
170        } else if (node.Symbol is BinaryFactorVariable) {
171          var binFactorNode = node as BinaryFactorVariableTreeNode;
172          FormatBinaryFactor(binFactorNode, strBuilder);
173        } else {
174          throw new NotSupportedException("Formatting of symbol: " + node.Symbol + " not supported for C# symbolic expression tree formatter.");
175        }
176      }
177    }
178
179    private void FormatFactor(FactorVariableTreeNode node, StringBuilder strBuilder) {
180      strBuilder.AppendFormat("EvaluateFactor({0}, new [] {{ {1} }}, new [] {{ {2} }})", VariableName2Identifier(node.VariableName),
181        string.Join(",", node.Symbol.GetVariableValues(node.VariableName).Select(name => "\"" + name + "\"")), string.Join(",", node.Weights.Select(v => v.ToString(CultureInfo.InvariantCulture))));
182    }
183
184    private void FormatBinaryFactor(BinaryFactorVariableTreeNode node, StringBuilder strBuilder) {
185      strBuilder.AppendFormat(CultureInfo.InvariantCulture, "EvaluateBinaryFactor({0}, \"{1}\", {2})", VariableName2Identifier(node.VariableName), node.VariableValue, node.Weight);
186    }
187
188    private void FormatSquare(ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
189      FormatPower(node, strBuilder, "2");
190    }
191    private void FormatPower(ISymbolicExpressionTreeNode node, StringBuilder strBuilder, string exponent) {
192      strBuilder.Append("Math.Pow(");
193      FormatRecursively(node.GetSubtree(0), strBuilder);
194      strBuilder.Append($", {exponent})");
195    }
196
197    private void FormatRoot(ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
198      strBuilder.Append("Math.Pow(");
199      FormatRecursively(node.GetSubtree(0), strBuilder);
200      strBuilder.Append(", 1.0 / (");
201      FormatRecursively(node.GetSubtree(1), strBuilder);
202      strBuilder.Append("))");
203    }
204
205    private void FormatDivision(ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
206      if (node.SubtreeCount == 1) {
207        strBuilder.Append("1.0 / ");
208        FormatRecursively(node.GetSubtree(0), strBuilder);
209      } else {
210        FormatRecursively(node.GetSubtree(0), strBuilder);
211        strBuilder.Append("/ (");
212        for (int i = 1; i < node.SubtreeCount; i++) {
213          if (i > 1) strBuilder.Append(" * ");
214          FormatRecursively(node.GetSubtree(i), strBuilder);
215        }
216        strBuilder.Append(")");
217      }
218    }
219
220    private void FormatSubtraction(ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
221      if (node.SubtreeCount == 1) {
222        strBuilder.Append("-");
223        FormatRecursively(node.GetSubtree(0), strBuilder);
224        return;
225      }
226      //Default case: more than 1 child
227      FormatOperator(node, "-", strBuilder);
228    }
229
230    private void FormatOperator(ISymbolicExpressionTreeNode node, string symbol, StringBuilder strBuilder) {
231      strBuilder.Append("(");
232      foreach (var child in node.Subtrees) {
233        FormatRecursively(child, strBuilder);
234        if (child != node.Subtrees.Last())
235          strBuilder.Append(" " + symbol + " ");
236      }
237      strBuilder.Append(")");
238    }
239
240    private void FormatFunction(ISymbolicExpressionTreeNode node, string function, StringBuilder strBuilder) {
241      strBuilder.Append(function + "(");
242      foreach (var child in node.Subtrees) {
243        FormatRecursively(child, strBuilder);
244        if (child != node.Subtrees.Last())
245          strBuilder.Append(", ");
246      }
247      strBuilder.Append(")");
248    }
249
250    private void GenerateHeader(StringBuilder strBuilder, ISymbolicExpressionTree symbolicExpressionTree) {
251      strBuilder.AppendLine("using System;");
252      strBuilder.AppendLine("using System.Linq;" + Environment.NewLine);
253      strBuilder.AppendLine("namespace HeuristicLab.Models {");
254      strBuilder.AppendLine("public static class Model {");
255      GenerateAverageSource(strBuilder);
256      GenerateCbrtSource(strBuilder);
257      GenerateIfThenElseSource(strBuilder);
258      GenerateFactorSource(strBuilder);
259      GenerateBinaryFactorSource(strBuilder);
260      strBuilder.Append(Environment.NewLine + "public static double Evaluate (");
261
262      // 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
263      HashSet<string> doubleVarNames = new HashSet<string>();
264      foreach (var node in symbolicExpressionTree.IterateNodesPostfix().Where(x => x is VariableTreeNode || x is VariableConditionTreeNode)) {
265        doubleVarNames.Add(((IVariableTreeNode)node).VariableName);
266      }
267
268      HashSet<string> stringVarNames = new HashSet<string>();
269      foreach (var node in symbolicExpressionTree.IterateNodesPostfix().Where(x => x is BinaryFactorVariableTreeNode || x is FactorVariableTreeNode)) {
270        stringVarNames.Add(((IVariableTreeNode)node).VariableName);
271      }
272
273      var orderedNames = stringVarNames.OrderBy(n => n, new NaturalStringComparer()).Select(n => "string " + VariableName2Identifier(n) + " /* " + n + " */");
274      strBuilder.Append(string.Join(", ", orderedNames));
275
276      if (stringVarNames.Any() && doubleVarNames.Any())
277        strBuilder.AppendLine(",");
278      orderedNames = doubleVarNames.OrderBy(n => n, new NaturalStringComparer()).Select(n => "double " + VariableName2Identifier(n) + " /* " + n + " */");
279      strBuilder.Append(string.Join(", ", orderedNames));
280
281
282      strBuilder.AppendLine(") {");
283      strBuilder.Append("double result = ");
284    }
285
286    private void GenerateFooter(StringBuilder strBuilder) {
287      strBuilder.AppendLine(";");
288
289      strBuilder.AppendLine("return result;");
290      strBuilder.AppendLine("}");
291      strBuilder.AppendLine("}");
292      strBuilder.AppendLine("}");
293    }
294
295    private void GenerateAverageSource(StringBuilder strBuilder) {
296      strBuilder.AppendLine("private static double Average(params double[] values) {");
297      strBuilder.AppendLine("  return values.Average();");
298      strBuilder.AppendLine("}");
299    }
300    private void GenerateCbrtSource(StringBuilder strBuilder) {
301      strBuilder.AppendLine("private static double Cbrt(double x) {");
302      strBuilder.AppendLine("  return x < 0 ? -Math.Pow(-x, 1.0 / 3.0) : Math.Pow(x, 1.0 / 3.0);");
303      strBuilder.AppendLine("}");
304    }
305
306    private void GenerateIfThenElseSource(StringBuilder strBuilder) {
307      strBuilder.AppendLine("private static double EvaluateIf(bool condition, double then, double @else) {");
308      strBuilder.AppendLine("   return condition ? then : @else;");
309      strBuilder.AppendLine("}");
310    }
311
312    private void GenerateFactorSource(StringBuilder strBuilder) {
313      strBuilder.AppendLine("private static double EvaluateFactor(string factorValue, string[] factorValues, double[] constants) {");
314      strBuilder.AppendLine("   for(int i=0;i<factorValues.Length;i++) " +
315                            "      if(factorValues[i] == factorValue) return constants[i];" +
316                            "   throw new ArgumentException();");
317      strBuilder.AppendLine("}");
318    }
319
320    private void GenerateBinaryFactorSource(StringBuilder strBuilder) {
321      strBuilder.AppendLine("private static double EvaluateBinaryFactor(string factorValue, string targetValue, double weight) {");
322      strBuilder.AppendLine("  return factorValue == targetValue ? weight : 0.0;");
323      strBuilder.AppendLine("}");
324    }
325
326  }
327}
Note: See TracBrowser for help on using the repository browser.