vendor: add all of our dependencies

vendor/github.com/disintegration/imaging/utils.go

@@ -0,0 +1,77 @@
+package imaging
+
+import (
+	"math"
+	"runtime"
+	"sync"
+	"sync/atomic"
+)
+
+var parallelizationEnabled = true
+
+// if GOMAXPROCS = 1: no goroutines used
+// if GOMAXPROCS > 1: spawn N=GOMAXPROCS workers in separate goroutines
+func parallel(dataSize int, fn func(partStart, partEnd int)) {
+	numGoroutines := 1
+	partSize := dataSize
+
+	if parallelizationEnabled {
+		numProcs := runtime.GOMAXPROCS(0)
+		if numProcs > 1 {
+			numGoroutines = numProcs
+			partSize = dataSize / (numGoroutines * 10)
+			if partSize < 1 {
+				partSize = 1
+			}
+		}
+	}
+
+	if numGoroutines == 1 {
+		fn(0, dataSize)
+	} else {
+		var wg sync.WaitGroup
+		wg.Add(numGoroutines)
+		idx := uint64(0)
+
+		for p := 0; p < numGoroutines; p++ {
+			go func() {
+				defer wg.Done()
+				for {
+					partStart := int(atomic.AddUint64(&idx, uint64(partSize))) - partSize
+					if partStart >= dataSize {
+						break
+					}
+					partEnd := partStart + partSize
+					if partEnd > dataSize {
+						partEnd = dataSize
+					}
+					fn(partStart, partEnd)
+				}
+			}()
+		}
+
+		wg.Wait()
+	}
+}
+
+func absint(i int) int {
+	if i < 0 {
+		return -i
+	}
+	return i
+}
+
+// clamp & round float64 to uint8 (0..255)
+func clamp(v float64) uint8 {
+	return uint8(math.Min(math.Max(v, 0.0), 255.0) + 0.5)
+}
+
+// clamp int32 to uint8 (0..255)
+func clampint32(v int32) uint8 {
+	if v < 0 {
+		return 0
+	} else if v > 255 {
+		return 255
+	}
+	return uint8(v)
+}