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 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 | constantCounter = original.constantCounter;
|
---|
25 | }
|
---|
26 | public InfixExpressionStringFormatter() : base() {
|
---|
27 | Name = ItemName;
|
---|
28 | Description = ItemDescription;
|
---|
29 | constants = new List<KeyValuePair<string, double>>();
|
---|
30 | }
|
---|
31 | public override IDeepCloneable Clone(Cloner cloner) {
|
---|
32 | return new InfixExpressionStringFormatter(this, cloner);
|
---|
33 | }
|
---|
34 |
|
---|
35 | public string Format(ISymbolicExpressionTree symbolicExpressionTree) {
|
---|
36 | StringBuilder strBuilder = new StringBuilder();
|
---|
37 | constants.Clear();
|
---|
38 | constantCounter = 0;
|
---|
39 | FormatRecursively(symbolicExpressionTree.Root.GetSubtree(0).GetSubtree(0), strBuilder);
|
---|
40 | strBuilder.Append($"{Environment.NewLine}{Environment.NewLine}");
|
---|
41 |
|
---|
42 | int maxDigits = GetDigits(constantCounter);
|
---|
43 | int padding = constants.Max(x => x.Value.ToString("F12", CultureInfo.InvariantCulture).Length);
|
---|
44 | foreach (var constant in constants) {
|
---|
45 | int digits = GetDigits(Int32.Parse(constant.Key.Substring(2)));
|
---|
46 | strBuilder.Append($"{constant.Key}{new String(' ', maxDigits - digits)} = " +
|
---|
47 | string.Format($"{{0,{padding}:F12}}", constant.Value, CultureInfo.InvariantCulture) + Environment.NewLine);
|
---|
48 | }
|
---|
49 | return strBuilder.ToString();
|
---|
50 | }
|
---|
51 |
|
---|
52 | private void FormatRecursively(ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
|
---|
53 | if (node.SubtreeCount > 1) {
|
---|
54 | var token = GetToken(node.Symbol);
|
---|
55 | if (token == "+" || token == "-" || token == "OR" || token == "XOR") {
|
---|
56 | strBuilder.Append("(");
|
---|
57 | FormatRecursively(node.Subtrees.First(), strBuilder);
|
---|
58 |
|
---|
59 | foreach (var subtree in node.Subtrees.Skip(1)) {
|
---|
60 | strBuilder.Append(" ").Append(token).Append(" ");
|
---|
61 | FormatRecursively(subtree, strBuilder);
|
---|
62 | }
|
---|
63 | strBuilder.Append(")");
|
---|
64 |
|
---|
65 | } else if (token == "*" || token == "/" || token == "AND") {
|
---|
66 | strBuilder.Append("(");
|
---|
67 | FormatRecursively(node.Subtrees.First(), strBuilder);
|
---|
68 |
|
---|
69 | foreach (var subtree in node.Subtrees.Skip(1)) {
|
---|
70 | strBuilder.Append(" ").Append(token).Append(" ");
|
---|
71 | FormatRecursively(subtree, strBuilder);
|
---|
72 | }
|
---|
73 | strBuilder.Append(")");
|
---|
74 | } else {
|
---|
75 | // function with multiple arguments
|
---|
76 | strBuilder.Append(token).Append("(");
|
---|
77 | FormatRecursively(node.Subtrees.First(), strBuilder);
|
---|
78 | foreach (var subtree in node.Subtrees.Skip(1)) {
|
---|
79 | strBuilder.Append(", ");
|
---|
80 | FormatRecursively(subtree, strBuilder);
|
---|
81 | }
|
---|
82 | strBuilder.Append(")");
|
---|
83 | }
|
---|
84 | } else if (node.SubtreeCount == 1) {
|
---|
85 | var token = GetToken(node.Symbol);
|
---|
86 | if (token == "-" || token == "NOT") {
|
---|
87 | strBuilder.Append("(").Append(token).Append("(");
|
---|
88 | FormatRecursively(node.GetSubtree(0), strBuilder);
|
---|
89 | strBuilder.Append("))");
|
---|
90 | } else if (token == "/") {
|
---|
91 | strBuilder.Append("1/");
|
---|
92 | FormatRecursively(node.GetSubtree(0), strBuilder);
|
---|
93 | } else if (token == "+" || token == "*") {
|
---|
94 | FormatRecursively(node.GetSubtree(0), strBuilder);
|
---|
95 | } else {
|
---|
96 | // function with only one argument
|
---|
97 | strBuilder.Append(token).Append("(");
|
---|
98 | FormatRecursively(node.GetSubtree(0), strBuilder);
|
---|
99 | strBuilder.Append(")");
|
---|
100 | }
|
---|
101 | } else {
|
---|
102 | if (node.Symbol is LaggedVariable) {
|
---|
103 | var varNode = node as LaggedVariableTreeNode;
|
---|
104 | if (!varNode.Weight.IsAlmost(1.0)) {
|
---|
105 | strBuilder.Append("(");
|
---|
106 | strBuilder.AppendFormat(CultureInfo.InvariantCulture, "{0}", varNode.Weight);
|
---|
107 | strBuilder.Append("*");
|
---|
108 | }
|
---|
109 | strBuilder.Append("LAG(");
|
---|
110 | if (varNode.VariableName.Contains("'")) {
|
---|
111 | strBuilder.AppendFormat("\"{0}\"", varNode.VariableName);
|
---|
112 | } else {
|
---|
113 | strBuilder.AppendFormat("'{0}'", varNode.VariableName);
|
---|
114 | }
|
---|
115 | strBuilder.Append(", ")
|
---|
116 | .AppendFormat(CultureInfo.InvariantCulture, "{0}", varNode.Lag)
|
---|
117 | .Append(")");
|
---|
118 | } else if (node.Symbol is Variable) {
|
---|
119 | var varNode = node as VariableTreeNode;
|
---|
120 | if (!varNode.Weight.IsAlmost(1.0)) {
|
---|
121 | string constantKey = $"c_{constantCounter}";
|
---|
122 | strBuilder.Append("(");
|
---|
123 | strBuilder.AppendFormat(CultureInfo.InvariantCulture, "{0}", constantKey);
|
---|
124 | strBuilder.Append("*");
|
---|
125 | constants.Add(new KeyValuePair<string, double>(constantKey, varNode.Weight));
|
---|
126 | constantCounter++;
|
---|
127 | }
|
---|
128 | if (varNode.VariableName.Contains("'")) {
|
---|
129 | strBuilder.AppendFormat("\"{0}\"", varNode.VariableName);
|
---|
130 | } else {
|
---|
131 | strBuilder.AppendFormat("'{0}'", varNode.VariableName);
|
---|
132 | }
|
---|
133 | if (!varNode.Weight.IsAlmost(1.0)) {
|
---|
134 | strBuilder.Append(")");
|
---|
135 | }
|
---|
136 | } else if (node.Symbol is FactorVariable) {
|
---|
137 | var factorNode = node as FactorVariableTreeNode;
|
---|
138 | if (factorNode.VariableName.Contains("'")) {
|
---|
139 | strBuilder.AppendFormat("\"{0}\"", factorNode.VariableName);
|
---|
140 | } else {
|
---|
141 | strBuilder.AppendFormat("'{0}'", factorNode.VariableName);
|
---|
142 | }
|
---|
143 | strBuilder.AppendFormat("[{0}]",
|
---|
144 | string.Join(", ", factorNode.Weights.Select(w => w.ToString(CultureInfo.InvariantCulture))));
|
---|
145 | } else if (node.Symbol is BinaryFactorVariable) {
|
---|
146 | var factorNode = node as BinaryFactorVariableTreeNode;
|
---|
147 | if (!factorNode.Weight.IsAlmost(1.0)) {
|
---|
148 | strBuilder.Append("(");
|
---|
149 | strBuilder.AppendFormat(CultureInfo.InvariantCulture, "{0}", factorNode.Weight);
|
---|
150 | strBuilder.Append("*");
|
---|
151 | }
|
---|
152 | if (factorNode.VariableName.Contains("'")) {
|
---|
153 | strBuilder.AppendFormat("\"{0}\"", factorNode.VariableName);
|
---|
154 | } else {
|
---|
155 | strBuilder.AppendFormat("'{0}'", factorNode.VariableName);
|
---|
156 | }
|
---|
157 | strBuilder.Append(" = ");
|
---|
158 | if (factorNode.VariableValue.Contains("'")) {
|
---|
159 | strBuilder.AppendFormat("\"{0}\"", factorNode.VariableValue);
|
---|
160 | } else {
|
---|
161 | strBuilder.AppendFormat("'{0}'", factorNode.VariableValue);
|
---|
162 | }
|
---|
163 |
|
---|
164 | if (!factorNode.Weight.IsAlmost(1.0)) {
|
---|
165 | strBuilder.Append(")");
|
---|
166 | }
|
---|
167 |
|
---|
168 | } else if (node.Symbol is Constant) {
|
---|
169 | var constNode = node as ConstantTreeNode;
|
---|
170 | string constantKey = $"c_{constantCounter}";
|
---|
171 |
|
---|
172 | strBuilder.AppendFormat(CultureInfo.InvariantCulture, constantKey);
|
---|
173 | constants.Add(new KeyValuePair<string, double>(constantKey, constNode.Value));
|
---|
174 | constantCounter++;
|
---|
175 | }
|
---|
176 | }
|
---|
177 | }
|
---|
178 |
|
---|
179 | private int GetDigits(int x) {
|
---|
180 | if (x < 10)
|
---|
181 | return 1;
|
---|
182 | else
|
---|
183 | return (int)Math.Floor(Math.Log10(x) + 1);
|
---|
184 | }
|
---|
185 |
|
---|
186 | private string GetToken(ISymbol symbol) {
|
---|
187 | var tok = InfixExpressionParser.knownSymbols.GetBySecond(symbol).SingleOrDefault();
|
---|
188 | if (tok == null)
|
---|
189 | throw new ArgumentException(string.Format("Unknown symbol {0} found.", symbol.Name));
|
---|
190 | return tok;
|
---|
191 | }
|
---|
192 | }
|
---|
193 | }
|
---|