// Copyright 2017 The Hugo Authors. All rights reserved. // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // You may obtain a copy of the License at // http://www.apache.org/licenses/LICENSE-2.0 // // Unless required by applicable law or agreed to in writing, software // distributed under the License is distributed on an "AS IS" BASIS, // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. // See the License for the specific language governing permissions and // limitations under the License. // Package math provides template functions for mathematical operations. package math import ( "errors" "fmt" "math" "math/rand" "reflect" "sync/atomic" _math "github.com/gohugoio/hugo/common/math" "github.com/spf13/cast" ) var ( errMustTwoNumbersError = errors.New("must provide at least two numbers") errMustOneNumberError = errors.New("must provide at least one number") ) // New returns a new instance of the math-namespaced template functions. func New() *Namespace { return &Namespace{} } // Namespace provides template functions for the "math" namespace. type Namespace struct{} // Abs returns the absolute value of n. func (ns *Namespace) Abs(n any) (float64, error) { af, err := cast.ToFloat64E(n) if err != nil { return 0, errors.New("the math.Abs function requires a numeric argument") } return math.Abs(af), nil } // Add adds the multivalued addends n1 and n2 or more values. func (ns *Namespace) Add(inputs ...any) (any, error) { return ns.doArithmetic(inputs, '+') } // Ceil returns the least integer value greater than or equal to n. func (ns *Namespace) Ceil(n any) (float64, error) { xf, err := cast.ToFloat64E(n) if err != nil { return 0, errors.New("Ceil operator can't be used with non-float value") } return math.Ceil(xf), nil } // Div divides n1 by n2. func (ns *Namespace) Div(inputs ...any) (any, error) { return ns.doArithmetic(inputs, '/') } // Floor returns the greatest integer value less than or equal to n. func (ns *Namespace) Floor(n any) (float64, error) { xf, err := cast.ToFloat64E(n) if err != nil { return 0, errors.New("Floor operator can't be used with non-float value") } return math.Floor(xf), nil } // Log returns the natural logarithm of the number n. func (ns *Namespace) Log(n any) (float64, error) { af, err := cast.ToFloat64E(n) if err != nil { return 0, errors.New("Log operator can't be used with non integer or float value") } return math.Log(af), nil } // Max returns the greater of all numbers in inputs. Any slices in inputs are flattened. func (ns *Namespace) Max(inputs ...any) (maximum float64, err error) { return ns.applyOpToScalarsOrSlices("Max", math.Max, inputs...) } // Min returns the smaller of all numbers in inputs. Any slices in inputs are flattened. func (ns *Namespace) Min(inputs ...any) (minimum float64, err error) { return ns.applyOpToScalarsOrSlices("Min", math.Min, inputs...) } // Sum returns the sum of all numbers in inputs. Any slices in inputs are flattened. func (ns *Namespace) Sum(inputs ...any) (sum float64, err error) { fn := func(x, y float64) float64 { return x + y } return ns.applyOpToScalarsOrSlices("Sum", fn, inputs...) } // Product returns the product of all numbers in inputs. Any slices in inputs are flattened. func (ns *Namespace) Product(inputs ...any) (product float64, err error) { fn := func(x, y float64) float64 { return x * y } return ns.applyOpToScalarsOrSlices("Product", fn, inputs...) } // Mod returns n1 % n2. func (ns *Namespace) Mod(n1, n2 any) (int64, error) { ai, erra := cast.ToInt64E(n1) bi, errb := cast.ToInt64E(n2) if erra != nil || errb != nil { return 0, errors.New("modulo operator can't be used with non integer value") } if bi == 0 { return 0, errors.New("the number can't be divided by zero at modulo operation") } return ai % bi, nil } // ModBool returns the boolean of n1 % n2. If n1 % n2 == 0, return true. func (ns *Namespace) ModBool(n1, n2 any) (bool, error) { res, err := ns.Mod(n1, n2) if err != nil { return false, err } return res == int64(0), nil } // Mul multiplies the multivalued numbers n1 and n2 or more values. func (ns *Namespace) Mul(inputs ...any) (any, error) { return ns.doArithmetic(inputs, '*') } // Pow returns n1 raised to the power of n2. func (ns *Namespace) Pow(n1, n2 any) (float64, error) { af, erra := cast.ToFloat64E(n1) bf, errb := cast.ToFloat64E(n2) if erra != nil || errb != nil { return 0, errors.New("Pow operator can't be used with non-float value") } return math.Pow(af, bf), nil } // Rand returns, as a float64, a pseudo-random number in the half-open interval [0.0,1.0). func (ns *Namespace) Rand() float64 { return rand.Float64() } // Round returns the integer nearest to n, rounding half away from zero. func (ns *Namespace) Round(n any) (float64, error) { xf, err := cast.ToFloat64E(n) if err != nil { return 0, errors.New("Round operator can't be used with non-float value") } return _round(xf), nil } // Sqrt returns the square root of the number n. func (ns *Namespace) Sqrt(n any) (float64, error) { af, err := cast.ToFloat64E(n) if err != nil { return 0, errors.New("Sqrt operator can't be used with non integer or float value") } return math.Sqrt(af), nil } // Sub subtracts multivalued. func (ns *Namespace) Sub(inputs ...any) (any, error) { return ns.doArithmetic(inputs, '-') } func (ns *Namespace) applyOpToScalarsOrSlices(opName string, op func(x, y float64) float64, inputs ...any) (result float64, err error) { var i int var hasValue bool for _, input := range inputs { var values []float64 var isSlice bool values, isSlice, err = ns.toFloatsE(input) if err != nil { err = fmt.Errorf("%s operator can't be used with non-float values", opName) return } hasValue = hasValue || len(values) > 0 || isSlice for _, value := range values { i++ if i == 1 { result = value continue } result = op(result, value) } } if !hasValue { err = errMustOneNumberError return } return } func (ns *Namespace) toFloatsE(v any) ([]float64, bool, error) { vv := reflect.ValueOf(v) switch vv.Kind() { case reflect.Slice, reflect.Array: var floats []float64 for i := 0; i < vv.Len(); i++ { f, err := cast.ToFloat64E(vv.Index(i).Interface()) if err != nil { return nil, true, err } floats = append(floats, f) } return floats, true, nil default: f, err := cast.ToFloat64E(v) if err != nil { return nil, false, err } return []float64{f}, false, nil } } func (ns *Namespace) doArithmetic(inputs []any, operation rune) (value any, err error) { if len(inputs) < 2 { return nil, errMustTwoNumbersError } value = inputs[0] for i := 1; i < len(inputs); i++ { value, err = _math.DoArithmetic(value, inputs[i], operation) if err != nil { return } } return } var counter uint64 // Counter increments and returns a global counter. // This was originally added to be used in tests where now.UnixNano did not // have the needed precision (especially on Windows). // Note that given the parallel nature of Hugo, you cannot use this to get sequences of numbers, // and the counter will reset on new builds. // {"identifiers": ["now.UnixNano"] } func (ns *Namespace) Counter() uint64 { return atomic.AddUint64(&counter, uint64(1)) }