Patrik Lundin, patriklundin.info
Math expression parser, evaluates a string mathematical expression and returns a double value.
ExpressionParser parser = new ExpressionParser(); parser.Values.Add("x", 2); parser.Values.Add("y", 10); double result = parser.Parse("x^3+5x^2-3");
Updating variable values
ExpressionParser parser = new ExpressionParser(); // Create value instances DoubleValue xval = new DoubleValue(); DoubleValue yval = new DoubleValue(); // Add values for variables x and y parser.Values.Add("x", xval); parser.Values.Add("y", yval); xval.Value = 2; // Update value of "x" yval.Value = 10; // Update value of "y" double result = parser.Parse("x^3+5x^2-3");Updating values using the SetValue method:
parser.Values["x"].SetValue(2); // x previously added with Add parser.Values["y"].SetValue(10); // y previously added with AddUpdating by casting and setting the Value property:
((DoubleValue)parser.Values["x"]).Value = 2; // x previously added with Add ((DoubleValue)parser.Values["y"]).Value = 10; // y previously added with Add
Faster evaluation by keeping the expression tree
ExpressionParser parser = new ExpressionParser(); parser.Values.Add("x", 2); parser.Values.Add("y", 10); // Parse once string func = "x^3+5x^2-3"; parser.Parse(func); // Fetch expression Expression expression = parser.Expressions[func]; // Evaluate saved expression double result = parser.EvalExpression(expression);
Serializing parsed expressions
To save some initial parsing time you can serialize the expressions to a Stream for storage and later use.
Both the Expression (see above) and the ExpressionDictionary accessable by the property parser.Expressions have Save and Load methods that take a Stream as parameter.
Serializing should only be needed with large sets of expressions.
Please note that serializing expressions and then deserializing using a different culture setting may
Supported operators and functions
+, -, *, /, ^, %
^ is raised to (power) for example 3^2
sqrt, sin, cos, tan, atan, acos, asin, acotan, exp, ln, log, sinh, cosh, tanh, abs, ceil, floor, fac, sfac, round, fpart
These functions mostly map to the System.Math functions except fac, sfac which is the factorial and semi-factorial functions and fpart which returns the decimal part of a value.
!, ==, !=, ||, &&, >, < , >=, <= Logical operators, 1.0 means true, 0.0 means false. If an expression evaluates to anything other than 1.0 it is considered false.
There is currently no support for adding additional operators or functions.
PI (value of System.Math.PI)
Variable naming restrictions
All variable names must start with an alphabetic letter (a-z) and may contain digits at the end but not inside the variable name.
Examples: x, y, z, var1, var2, myverylongvariablename
Variable names can contain function keywords ONLY when ImplicitMultiplication is set to false and RequireParentheses is set to true.
Examples: glucose, sincosexp
Properties and settings
The following properties can be set on the ExpressionParser:
RequireParentheses - toggles the requirement to use parentheses around function arguments (default true)
Cultures and decimal separators
The default culture is set to CultureInfo.InvariantCulture which uses a single dot (.) as the decimal separator and comma (,) as grouping (thousands) separator.
You can change the culture by setting the property Culture in the parser, if you do change the culture the number decimal separator in your expressions must match the culture. It is recommended that you use the default invariant culture and use a dot as a decimal separator especially if you decide to save parsed expressions as serialized data.
Culture fa, fa-IR uses the division operator (/) as the decimal separator, this is not allowed in the parser and the recommended solution is to use the invariant culture.
Currency symbols are not allowed. Thousands separators are supported but not recommended to use.
Cultures using the same decimal and grouping separators are not allowed.
Limitations of using double values
If you use the parser for financial calculations please make sure you understand how IEEE754 floating point values work and the rounding errors that may result.
It is generally not recommended to use float or double values for money applications since these are floating point values with a binary representation, for languages that have them a decimal representation is recommended, however this parser uses the default System.Math library and only operate with double values.
The provided tests should not be considered to fully test every scenario.
Remember it is your responsibility to verify that the parser works correctly for the intended use, it is recommended that you write some additional unit tests that tests with your specific data.
License and disclaimer
Patrik Lundin, patrik (at) lundin.info, http://www.lundin.info Copyright 2002-2018 Patrik Lundin
Library and Source code released under the Microsoft Public License (Ms-PL)
© 1997-2018 Patrik Lundin, patrik (at) lundin.info