mathjson-solver 1.4.0

Utilities for MathJSON evaluation

MathJSON Solver

MathJSON Solver is a Python module to numerically evaluate MathJSON expressions. It was created by Longenesis to add numerical evaluation capability of user generated mathematical expressions in Longenesis digital health products and later released as open source project. Its development was inspired by CortexJS Compute Engine.

Please ask questions and share feedback in our Gitter chat https://gitter.im/mathjson-solver/community.

What's new

1.4.0

We changed the behavior of Average to be more forgiving. In version 1.4.0 Average accepts arrays like [2, 4 ,"6"] and internally converts numeric strings to floats. Also, it skips values that cannot be converted to numeric. Internally Average will convert array [2, "three", 4 ,"6"] to [2.0, 4.0 ,6.0]. When given an empty array, Average now returns None instead of throwing an error.

How to use

from mathjson_solver import create_solver

parameters = {"x": 2, "y": 3}
expression = ["Add", "x", "y", 4]

solver = create_solver(parameters)
answer = solver(expression)

print(answer)
# 9, because 2+3+4=9

How to run tests

Make sure you have pytest installed. Then cd into project directory and run:

pytest

Contributing Code

We welcome your contributions in the form of pull requests.

1. Fork the repo;
2. Make improvements;
3. Make a pull request to share your improvements with the community and to include it into official release.

Currently supported constructs

Math

• Sum (alias Add)
• Subtract
• Multiply
• Divide
• Negate
• Power
• Root
• Sqrt
• Square
• Exp
• Log
• Log2
• Log10

Aggregation

• Array
• Max
• Min
• Average
• Median
• Len
• All
• Any

Conditions

• If
• Switch
• Equal
• Greater
• GreaterEqual
• Less
• LessEqual
• NotEqual

Typecasting

• Int
• Float

Additional constructs

• Constants

Examples

["Add", 2, 4, 3]                  # 2+4+3=9
["Subtract", 10, 5, 2]            # 10-5-2=3
["Add", 5, 4, ["Negate", 3]]      # 5+4+(-3)=6
["Multiply", 2, 3, 4]             # 2*3*4=24
["Divide", 10, 5]                 # 10/5=2.0
["Divide", 10, 4]                 # 10/4=2.5
["Power", 2, 3]                   # 2^3=8
["Root", 9, 2]                    # √9=3.0
["Root", 8, 3]                    # ∛8=2.0
["Sqrt", 9]                       # √9=3.0
["Square", 4]                     # 4^2=16
["Exp", 2]                        # e^2≅7.389
["Divide", 10, ["Add", 2+3]]      # 10/(2+3)=10/5=2
["Log", 2.7183]                   # ln(2.7183)≅1.0000
["Log2", 8]                       # log2(8)=3.0
["Log10", 1000]                   # log10(1000)=3.0
["Equal", 10, 10]                 # 10==10 = True
["Equal", 10, 12]                 # 10==10 = False
["Abs", -3.5]                     # |-3.5| = 3.5
["Round", -5.123456, 2]           # -5.12
["Round", -5.123456, 0]           # -5.0
["Round", -5.123456]              # -5
["Array", 1, 2]                   # ["Array", 1, 2]

["Max", ["Array", 1, 2, 3, 5, 2]] # 5
["Max", ["Array", 1, 2, ["Sum", 2, 4, 3], 5, 2]]  # 9
["Median", ["Array", 1, 2, 3, 5, 2]]              # 2
["Average", ["Array", 1, 2, 3, 5, 2]]             # 2.6
["Length", ["Array", 1, 2, 3, 5, 2, 9]]           # 6
["Length", ["Array"]]                             # 0
["Any", ["Array", 0, 0, False, 0, 0]]             # False
["Any", ["Array", 0, 1, False, 0, 0]]             # True
["All", ["Array", 0, 1, False, 0, 0]]             # False
["All", ["Array", 0, 1, False, "", 0]]            # False
["All", ["Array", 2, 1, True, "zz", 2]]           # True
["Int", "12"]                     # 12
["Int", "12.2"]                   # 12
["Float", "12.2"]                 # 12.2

Constants

[
    "Constants",
    ["constant_name1", <expression>],
    ["constant_name2", <expression>],
    ["constant_name3", <expression>],
    ...,
    <expression>
]

Constants construct consists of keyword "Constants" followed by arbitrary number of name&value pairs. The last element in Constants construct is the expression to calculate using the defined constants.

The following example has two constants defined - x=10 and y=20. Then the sum of these two constants is calculated and returned.

[
    "Constants",
    ["x", 10],
    ["y", 20],
    ["Add","x", "y"]
]

If statement

[
    "If",
    [
        <true-or-false-expression>,
        <expression-to-calculate>
    ],
    [
        <elseif-true-or-false-expression>,
        <expression-to-calculate>
    ],
    ...,
    <else-expression-to-calculate>
]

Example

[
    "If",
    [
        ["Equal", 1, 0],
        10
    ],
    [
        ["Equal", 2, 2],
        20
    ],
    9000
]

This construct translates to:

if   1 == 0 then 10
elif 2 == 2 then 20
else 9000

If expression do not need to be strictly boolean. Any value that is not false are considered true.

Switch-Case statement

["Switch", <on-expression>, <default-result-expression>, [<case1-expression>, <result-expression>], ...],

Switch construct consists of keyword "Switch" followed by expression whose value will be compared to Cases' values. Then comes the default value. Then follows arbitrary number of Cases.

Example

["Switch", "color", 100, ["red", 10], ["blue", 20], ["green", 30]],

The expression in this example will make solver to look for a constant (or a parameter) with the name "color". If "color" is "red", expression evaluates to 10, if "blue" - to 20, if "green" - to 30. Otherwise to 100. Please note that "color" here is a valid expression that evaluates to the actual value of "color" whether it is a parameter or constant.

Exception handling

A MathJSONException is raised when expression cannot be evaluated. Import MathJSONException to handle it:

from mathjson_solver import create_solver, MathJSONException

solver = create_solver({})
try:
    solver(["Divide", 1, 0])
except MathJSONException:
    pass
    # invoke your own exception logger here

Left unhandled, the exception will look like MathJSONException("Problem in Divide. ['Divide', 1, 0]. division by zero").