1 | using HeuristicLab.Core;
|
---|
2 | using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
|
---|
3 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
---|
4 | using System.Collections.Generic;
|
---|
5 | using HeuristicLab.Common;
|
---|
6 | using System.Text;
|
---|
7 | using System.Globalization;
|
---|
8 | using System.Linq;
|
---|
9 | using System;
|
---|
10 |
|
---|
11 | namespace HeuristicLab.Problems.DataAnalysis.Symbolic.Formatters {
|
---|
12 | [Item("Infix String Formater", "Formatter for symbolic expressions, which produces an infix expression " +
|
---|
13 | "as well as a list of all coefficient values")]
|
---|
14 | [StorableClass]
|
---|
15 | public sealed class InfixExpressionStringFormatter : NamedItem, ISymbolicExpressionTreeStringFormatter {
|
---|
16 | private readonly List<KeyValuePair<string, double>> constants;
|
---|
17 | private static int constantCounter = 0;
|
---|
18 |
|
---|
19 | [StorableConstructor]
|
---|
20 | private InfixExpressionStringFormatter(bool deserializing): base(deserializing) {}
|
---|
21 | private InfixExpressionStringFormatter(InfixExpressionStringFormatter original, Cloner cloner)
|
---|
22 | : base(original, cloner) {
|
---|
23 | constants = new List<KeyValuePair<string, double>>(original.constants);
|
---|
24 | }
|
---|
25 | public InfixExpressionStringFormatter() : base() {
|
---|
26 | Name = ItemName;
|
---|
27 | Description = ItemDescription;
|
---|
28 | constants = new List<KeyValuePair<string, double>>();
|
---|
29 | }
|
---|
30 | public override IDeepCloneable Clone(Cloner cloner) {
|
---|
31 | return new InfixExpressionStringFormatter(this, cloner);
|
---|
32 | }
|
---|
33 |
|
---|
34 | public string Format(ISymbolicExpressionTree symbolicExpressionTree) {
|
---|
35 | StringBuilder strBuilder = new StringBuilder();
|
---|
36 | FormatRecursively(symbolicExpressionTree.Root.GetSubtree(0).GetSubtree(0), strBuilder);
|
---|
37 | strBuilder.Append($"{Environment.NewLine}{Environment.NewLine}");
|
---|
38 | foreach (var constant in constants) {
|
---|
39 | strBuilder.Append($"{constant.Key} = {constant.Value}{Environment.NewLine}");
|
---|
40 | }
|
---|
41 | return strBuilder.ToString();
|
---|
42 | }
|
---|
43 |
|
---|
44 | private void FormatRecursively(ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
|
---|
45 | if (node.SubtreeCount > 1) {
|
---|
46 | var token = GetToken(node.Symbol);
|
---|
47 | if (token == "+" || token == "-" || token == "OR" || token == "XOR") {
|
---|
48 | strBuilder.Append("(");
|
---|
49 | FormatRecursively(node.Subtrees.First(), strBuilder);
|
---|
50 |
|
---|
51 | foreach (var subtree in node.Subtrees.Skip(1)) {
|
---|
52 | strBuilder.Append(" ").Append(token).Append(" ");
|
---|
53 | FormatRecursively(subtree, strBuilder);
|
---|
54 | }
|
---|
55 | strBuilder.Append(")");
|
---|
56 |
|
---|
57 | } else if (token == "*" || token == "/" || token == "AND") {
|
---|
58 | strBuilder.Append("(");
|
---|
59 | FormatRecursively(node.Subtrees.First(), strBuilder);
|
---|
60 |
|
---|
61 | foreach (var subtree in node.Subtrees.Skip(1)) {
|
---|
62 | strBuilder.Append(" ").Append(token).Append(" ");
|
---|
63 | FormatRecursively(subtree, strBuilder);
|
---|
64 | }
|
---|
65 | strBuilder.Append(")");
|
---|
66 | } else {
|
---|
67 | // function with multiple arguments
|
---|
68 | strBuilder.Append(token).Append("(");
|
---|
69 | FormatRecursively(node.Subtrees.First(), strBuilder);
|
---|
70 | foreach (var subtree in node.Subtrees.Skip(1)) {
|
---|
71 | strBuilder.Append(", ");
|
---|
72 | FormatRecursively(subtree, strBuilder);
|
---|
73 | }
|
---|
74 | strBuilder.Append(")");
|
---|
75 | }
|
---|
76 | } else if (node.SubtreeCount == 1) {
|
---|
77 | var token = GetToken(node.Symbol);
|
---|
78 | if (token == "-" || token == "NOT") {
|
---|
79 | strBuilder.Append("(").Append(token).Append("(");
|
---|
80 | FormatRecursively(node.GetSubtree(0), strBuilder);
|
---|
81 | strBuilder.Append("))");
|
---|
82 | } else if (token == "/") {
|
---|
83 | strBuilder.Append("1/");
|
---|
84 | FormatRecursively(node.GetSubtree(0), strBuilder);
|
---|
85 | } else if (token == "+" || token == "*") {
|
---|
86 | FormatRecursively(node.GetSubtree(0), strBuilder);
|
---|
87 | } else {
|
---|
88 | // function with only one argument
|
---|
89 | strBuilder.Append(token).Append("(");
|
---|
90 | FormatRecursively(node.GetSubtree(0), strBuilder);
|
---|
91 | strBuilder.Append(")");
|
---|
92 | }
|
---|
93 | } else {
|
---|
94 | if (node.Symbol is LaggedVariable) {
|
---|
95 | var varNode = node as LaggedVariableTreeNode;
|
---|
96 | if (!varNode.Weight.IsAlmost(1.0)) {
|
---|
97 | strBuilder.Append("(");
|
---|
98 | strBuilder.AppendFormat(CultureInfo.InvariantCulture, "{0}", varNode.Weight);
|
---|
99 | strBuilder.Append("*");
|
---|
100 | }
|
---|
101 | strBuilder.Append("LAG(");
|
---|
102 | if (varNode.VariableName.Contains("'")) {
|
---|
103 | strBuilder.AppendFormat("\"{0}\"", varNode.VariableName);
|
---|
104 | } else {
|
---|
105 | strBuilder.AppendFormat("'{0}'", varNode.VariableName);
|
---|
106 | }
|
---|
107 | strBuilder.Append(", ")
|
---|
108 | .AppendFormat(CultureInfo.InvariantCulture, "{0}", varNode.Lag)
|
---|
109 | .Append(")");
|
---|
110 | } else if (node.Symbol is Variable) {
|
---|
111 | var varNode = node as VariableTreeNode;
|
---|
112 | if (!varNode.Weight.IsAlmost(1.0)) {
|
---|
113 | string constantKey = $"c_{constantCounter}";
|
---|
114 | strBuilder.Append("(");
|
---|
115 | strBuilder.AppendFormat(CultureInfo.InvariantCulture, "{0}", constantKey);
|
---|
116 | strBuilder.Append("*");
|
---|
117 | constants.Add(new KeyValuePair<string, double>(constantKey, varNode.Weight));
|
---|
118 | constantCounter++;
|
---|
119 | }
|
---|
120 | if (varNode.VariableName.Contains("'")) {
|
---|
121 | strBuilder.AppendFormat("\"{0}\"", varNode.VariableName);
|
---|
122 | } else {
|
---|
123 | strBuilder.AppendFormat("'{0}'", varNode.VariableName);
|
---|
124 | }
|
---|
125 | if (!varNode.Weight.IsAlmost(1.0)) {
|
---|
126 | strBuilder.Append(")");
|
---|
127 | }
|
---|
128 | } else if (node.Symbol is FactorVariable) {
|
---|
129 | var factorNode = node as FactorVariableTreeNode;
|
---|
130 | if (factorNode.VariableName.Contains("'")) {
|
---|
131 | strBuilder.AppendFormat("\"{0}\"", factorNode.VariableName);
|
---|
132 | } else {
|
---|
133 | strBuilder.AppendFormat("'{0}'", factorNode.VariableName);
|
---|
134 | }
|
---|
135 | strBuilder.AppendFormat("[{0}]",
|
---|
136 | string.Join(", ", factorNode.Weights.Select(w => w.ToString(CultureInfo.InvariantCulture))));
|
---|
137 | } else if (node.Symbol is BinaryFactorVariable) {
|
---|
138 | var factorNode = node as BinaryFactorVariableTreeNode;
|
---|
139 | if (!factorNode.Weight.IsAlmost(1.0)) {
|
---|
140 | strBuilder.Append("(");
|
---|
141 | strBuilder.AppendFormat(CultureInfo.InvariantCulture, "{0}", factorNode.Weight);
|
---|
142 | strBuilder.Append("*");
|
---|
143 | }
|
---|
144 | if (factorNode.VariableName.Contains("'")) {
|
---|
145 | strBuilder.AppendFormat("\"{0}\"", factorNode.VariableName);
|
---|
146 | } else {
|
---|
147 | strBuilder.AppendFormat("'{0}'", factorNode.VariableName);
|
---|
148 | }
|
---|
149 | strBuilder.Append(" = ");
|
---|
150 | if (factorNode.VariableValue.Contains("'")) {
|
---|
151 | strBuilder.AppendFormat("\"{0}\"", factorNode.VariableValue);
|
---|
152 | } else {
|
---|
153 | strBuilder.AppendFormat("'{0}'", factorNode.VariableValue);
|
---|
154 | }
|
---|
155 |
|
---|
156 | if (!factorNode.Weight.IsAlmost(1.0)) {
|
---|
157 | strBuilder.Append(")");
|
---|
158 | }
|
---|
159 |
|
---|
160 | } else if (node.Symbol is Constant) {
|
---|
161 | var constNode = node as ConstantTreeNode;
|
---|
162 | string constantKey = $"c_{constantCounter}";
|
---|
163 |
|
---|
164 | strBuilder.AppendFormat(CultureInfo.InvariantCulture, constantKey);
|
---|
165 | constants.Add(new KeyValuePair<string, double>(constantKey, constNode.Value));
|
---|
166 | constantCounter++;
|
---|
167 | }
|
---|
168 | }
|
---|
169 | }
|
---|
170 |
|
---|
171 | private string GetToken(ISymbol symbol) {
|
---|
172 | var tok = InfixExpressionParser.knownSymbols.GetBySecond(symbol).SingleOrDefault();
|
---|
173 | if (tok == null)
|
---|
174 | throw new ArgumentException(string.Format("Unknown symbol {0} found.", symbol.Name));
|
---|
175 | return tok;
|
---|
176 | }
|
---|
177 | }
|
---|
178 | }
|
---|