Raise go version (#48)

2021-09-01 16:18:01 +03:00 · 2021-09-01 16:18:01 +03:00 · bcc7a3ad53
commit bcc7a3ad53
parent d2812476f9
6 changed files with 204 additions and 39 deletions
--- a/go.mod
+++ b/go.mod
@ -1,5 +1,5 @@
 module github.com/irlndts/go-discogs

-go 1.16
+go 1.17

-require github.com/google/go-cmp v0.5.4
+require github.com/google/go-cmp v0.5.6
--- a/go.sum
+++ b/go.sum
@ -1,4 +1,4 @@
-github.com/google/go-cmp v0.5.4 h1:L8R9j+yAqZuZjsqh/z+F1NCffTKKLShY6zXTItVIZ8M=
-github.com/google/go-cmp v0.5.4/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
+github.com/google/go-cmp v0.5.6 h1:BKbKCqvP6I+rmFHt06ZmyQtvB8xAkWdhFyr0ZUNZcxQ=
+github.com/google/go-cmp v0.5.6/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
 golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543 h1:E7g+9GITq07hpfrRu66IVDexMakfv52eLZ2CXBWiKr4=
 golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
--- a/vendor/github.com/google/go-cmp/cmp/path.go
+++ b/vendor/github.com/google/go-cmp/cmp/path.go
@ -315,7 +315,7 @@ func (tf Transform) Option() Option { return tf.trans }
 // pops the address from the stack. Thus, when traversing into a pointer from
 // reflect.Ptr, reflect.Slice element, or reflect.Map, we can detect cycles
 // by checking whether the pointer has already been visited. The cycle detection
-// uses a seperate stack for the x and y values.
+// uses a separate stack for the x and y values.
 //
 // If a cycle is detected we need to determine whether the two pointers
 // should be considered equal. The definition of equality chosen by Equal
--- a/vendor/github.com/google/go-cmp/cmp/report_compare.go
+++ b/vendor/github.com/google/go-cmp/cmp/report_compare.go
@ -79,7 +79,7 @@ func (opts formatOptions) verbosity() uint {
 	}
 }

-const maxVerbosityPreset = 3
+const maxVerbosityPreset = 6

 // verbosityPreset modifies the verbosity settings given an index
 // between 0 and maxVerbosityPreset, inclusive.
@ -100,7 +100,7 @@ func verbosityPreset(opts formatOptions, i int) formatOptions {
 func (opts formatOptions) FormatDiff(v *valueNode, ptrs *pointerReferences) (out textNode) {
 	if opts.DiffMode == diffIdentical {
 		opts = opts.WithVerbosity(1)
-	} else {
+	} else if opts.verbosity() < 3 {
 		opts = opts.WithVerbosity(3)
 	}

--- a/vendor/github.com/google/go-cmp/cmp/report_slices.go
+++ b/vendor/github.com/google/go-cmp/cmp/report_slices.go
@ -7,6 +7,7 @@ package cmp
 import (
 	"bytes"
 	"fmt"
+	"math"
 	"reflect"
 	"strconv"
 	"strings"
@ -26,8 +27,6 @@ func (opts formatOptions) CanFormatDiffSlice(v *valueNode) bool {
 		return false // No differences detected
 	case !v.ValueX.IsValid() || !v.ValueY.IsValid():
 		return false // Both values must be valid
-	case v.Type.Kind() == reflect.Slice && (v.ValueX.Len() == 0 || v.ValueY.Len() == 0):
-		return false // Both slice values have to be non-empty
 	case v.NumIgnored > 0:
 		return false // Some ignore option was used
 	case v.NumTransformed > 0:
@ -45,7 +44,16 @@ func (opts formatOptions) CanFormatDiffSlice(v *valueNode) bool {
 		return false
 	}

-	switch t := v.Type; t.Kind() {
+	// Check whether this is an interface with the same concrete types.
+	t := v.Type
+	vx, vy := v.ValueX, v.ValueY
+	if t.Kind() == reflect.Interface && !vx.IsNil() && !vy.IsNil() && vx.Elem().Type() == vy.Elem().Type() {
+		vx, vy = vx.Elem(), vy.Elem()
+		t = vx.Type()
+	}
+
+	// Check whether we provide specialized diffing for this type.
+	switch t.Kind() {
 	case reflect.String:
 	case reflect.Array, reflect.Slice:
 		// Only slices of primitive types have specialized handling.
@ -57,6 +65,11 @@ func (opts formatOptions) CanFormatDiffSlice(v *valueNode) bool {
 			return false
 		}

+		// Both slice values have to be non-empty.
+		if t.Kind() == reflect.Slice && (vx.Len() == 0 || vy.Len() == 0) {
+			return false
+		}
+
 		// If a sufficient number of elements already differ,
 		// use specialized formatting even if length requirement is not met.
 		if v.NumDiff > v.NumSame {
@ -68,7 +81,7 @@ func (opts formatOptions) CanFormatDiffSlice(v *valueNode) bool {

 	// Use specialized string diffing for longer slices or strings.
 	const minLength = 64
-	return v.ValueX.Len() >= minLength && v.ValueY.Len() >= minLength
+	return vx.Len() >= minLength && vy.Len() >= minLength
 }

 // FormatDiffSlice prints a diff for the slices (or strings) represented by v.
@ -77,17 +90,23 @@ func (opts formatOptions) CanFormatDiffSlice(v *valueNode) bool {
 func (opts formatOptions) FormatDiffSlice(v *valueNode) textNode {
 	assert(opts.DiffMode == diffUnknown)
 	t, vx, vy := v.Type, v.ValueX, v.ValueY
+	if t.Kind() == reflect.Interface {
+		vx, vy = vx.Elem(), vy.Elem()
+		t = vx.Type()
+		opts = opts.WithTypeMode(emitType)
+	}

 	// Auto-detect the type of the data.
-	var isLinedText, isText, isBinary bool
 	var sx, sy string
+	var ssx, ssy []string
+	var isString, isMostlyText, isPureLinedText, isBinary bool
 	switch {
 	case t.Kind() == reflect.String:
 		sx, sy = vx.String(), vy.String()
-		isText = true // Initial estimate, verify later
+		isString = true
 	case t.Kind() == reflect.Slice && t.Elem() == reflect.TypeOf(byte(0)):
 		sx, sy = string(vx.Bytes()), string(vy.Bytes())
-		isBinary = true // Initial estimate, verify later
+		isString = true
 	case t.Kind() == reflect.Array:
 		// Arrays need to be addressable for slice operations to work.
 		vx2, vy2 := reflect.New(t).Elem(), reflect.New(t).Elem()
@ -95,13 +114,12 @@ func (opts formatOptions) FormatDiffSlice(v *valueNode) textNode {
 		vy2.Set(vy)
 		vx, vy = vx2, vy2
 	}
-	if isText || isBinary {
-		var numLines, lastLineIdx, maxLineLen int
-		isBinary = !utf8.ValidString(sx) || !utf8.ValidString(sy)
+	if isString {
+		var numTotalRunes, numValidRunes, numLines, lastLineIdx, maxLineLen int
 		for i, r := range sx + sy {
-			if !(unicode.IsPrint(r) || unicode.IsSpace(r)) || r == utf8.RuneError {
-				isBinary = true
-				break
+			numTotalRunes++
+			if (unicode.IsPrint(r) || unicode.IsSpace(r)) && r != utf8.RuneError {
+				numValidRunes++
 			}
 			if r == '\n' {
 				if maxLineLen < i-lastLineIdx {
@ -111,8 +129,26 @@ func (opts formatOptions) FormatDiffSlice(v *valueNode) textNode {
 				numLines++
 			}
 		}
-		isText = !isBinary
-		isLinedText = isText && numLines >= 4 && maxLineLen <= 1024
+		isPureText := numValidRunes == numTotalRunes
+		isMostlyText = float64(numValidRunes) > math.Floor(0.90*float64(numTotalRunes))
+		isPureLinedText = isPureText && numLines >= 4 && maxLineLen <= 1024
+		isBinary = !isMostlyText
+
+		// Avoid diffing by lines if it produces a significantly more complex
+		// edit script than diffing by bytes.
+		if isPureLinedText {
+			ssx = strings.Split(sx, "\n")
+			ssy = strings.Split(sy, "\n")
+			esLines := diff.Difference(len(ssx), len(ssy), func(ix, iy int) diff.Result {
+				return diff.BoolResult(ssx[ix] == ssy[iy])
+			})
+			esBytes := diff.Difference(len(sx), len(sy), func(ix, iy int) diff.Result {
+				return diff.BoolResult(sx[ix] == sy[iy])
+			})
+			efficiencyLines := float64(esLines.Dist()) / float64(len(esLines))
+			efficiencyBytes := float64(esBytes.Dist()) / float64(len(esBytes))
+			isPureLinedText = efficiencyLines < 4*efficiencyBytes
+		}
 	}

 	// Format the string into printable records.
@ -121,9 +157,7 @@ func (opts formatOptions) FormatDiffSlice(v *valueNode) textNode {
 	switch {
 	// If the text appears to be multi-lined text,
 	// then perform differencing across individual lines.
-	case isLinedText:
-		ssx := strings.Split(sx, "\n")
-		ssy := strings.Split(sy, "\n")
+	case isPureLinedText:
 		list = opts.formatDiffSlice(
 			reflect.ValueOf(ssx), reflect.ValueOf(ssy), 1, "line",
 			func(v reflect.Value, d diffMode) textRecord {
@ -212,7 +246,7 @@ func (opts formatOptions) FormatDiffSlice(v *valueNode) textNode {
 	// If the text appears to be single-lined text,
 	// then perform differencing in approximately fixed-sized chunks.
 	// The output is printed as quoted strings.
-	case isText:
+	case isMostlyText:
 		list = opts.formatDiffSlice(
 			reflect.ValueOf(sx), reflect.ValueOf(sy), 64, "byte",
 			func(v reflect.Value, d diffMode) textRecord {
@ -220,7 +254,6 @@ func (opts formatOptions) FormatDiffSlice(v *valueNode) textNode {
 				return textRecord{Diff: d, Value: textLine(s)}
 			},
 		)
-		delim = ""

 	// If the text appears to be binary data,
 	// then perform differencing in approximately fixed-sized chunks.
@ -282,7 +315,7 @@ func (opts formatOptions) FormatDiffSlice(v *valueNode) textNode {

 	// Wrap the output with appropriate type information.
 	var out textNode = &textWrap{Prefix: "{", Value: list, Suffix: "}"}
-	if !isText {
+	if !isMostlyText {
 		// The "{...}" byte-sequence literal is not valid Go syntax for strings.
 		// Emit the type for extra clarity (e.g. "string{...}").
 		if t.Kind() == reflect.String {
@ -321,8 +354,11 @@ func (opts formatOptions) formatDiffSlice(
 	vx, vy reflect.Value, chunkSize int, name string,
 	makeRec func(reflect.Value, diffMode) textRecord,
 ) (list textList) {
-	es := diff.Difference(vx.Len(), vy.Len(), func(ix int, iy int) diff.Result {
-		return diff.BoolResult(vx.Index(ix).Interface() == vy.Index(iy).Interface())
+	eq := func(ix, iy int) bool {
+		return vx.Index(ix).Interface() == vy.Index(iy).Interface()
+	}
+	es := diff.Difference(vx.Len(), vy.Len(), func(ix, iy int) diff.Result {
+		return diff.BoolResult(eq(ix, iy))
 	})

 	appendChunks := func(v reflect.Value, d diffMode) int {
@ -347,6 +383,7 @@ func (opts formatOptions) formatDiffSlice(

 	groups := coalesceAdjacentEdits(name, es)
 	groups = coalesceInterveningIdentical(groups, chunkSize/4)
+	groups = cleanupSurroundingIdentical(groups, eq)
 	maxGroup := diffStats{Name: name}
 	for i, ds := range groups {
 		if maxLen >= 0 && numDiffs >= maxLen {
@ -399,25 +436,36 @@ func (opts formatOptions) formatDiffSlice(

 // coalesceAdjacentEdits coalesces the list of edits into groups of adjacent
 // equal or unequal counts.
+//
+// Example:
+//
+//	Input:  "..XXY...Y"
+//	Output: [
+//		{NumIdentical: 2},
+//		{NumRemoved: 2, NumInserted 1},
+//		{NumIdentical: 3},
+//		{NumInserted: 1},
+//	]
+//
 func coalesceAdjacentEdits(name string, es diff.EditScript) (groups []diffStats) {
-	var prevCase int // Arbitrary index into which case last occurred
-	lastStats := func(i int) *diffStats {
-		if prevCase != i {
+	var prevMode byte
+	lastStats := func(mode byte) *diffStats {
+		if prevMode != mode {
 			groups = append(groups, diffStats{Name: name})
-			prevCase = i
+			prevMode = mode
 		}
 		return &groups[len(groups)-1]
 	}
 	for _, e := range es {
 		switch e {
 		case diff.Identity:
-			lastStats(1).NumIdentical++
+			lastStats('=').NumIdentical++
 		case diff.UniqueX:
-			lastStats(2).NumRemoved++
+			lastStats('!').NumRemoved++
 		case diff.UniqueY:
-			lastStats(2).NumInserted++
+			lastStats('!').NumInserted++
 		case diff.Modified:
-			lastStats(2).NumModified++
+			lastStats('!').NumModified++
 		}
 	}
 	return groups
@ -427,6 +475,35 @@ func coalesceAdjacentEdits(name string, es diff.EditScript) (groups []diffStats)
 // equal groups into adjacent unequal groups that currently result in a
 // dual inserted/removed printout. This acts as a high-pass filter to smooth
 // out high-frequency changes within the windowSize.
+//
+// Example:
+//
+//	WindowSize: 16,
+//	Input: [
+//		{NumIdentical: 61},              // group 0
+//		{NumRemoved: 3, NumInserted: 1}, // group 1
+//		{NumIdentical: 6},               // ├── coalesce
+//		{NumInserted: 2},                // ├── coalesce
+//		{NumIdentical: 1},               // ├── coalesce
+//		{NumRemoved: 9},                 // └── coalesce
+//		{NumIdentical: 64},              // group 2
+//		{NumRemoved: 3, NumInserted: 1}, // group 3
+//		{NumIdentical: 6},               // ├── coalesce
+//		{NumInserted: 2},                // ├── coalesce
+//		{NumIdentical: 1},               // ├── coalesce
+//		{NumRemoved: 7},                 // ├── coalesce
+//		{NumIdentical: 1},               // ├── coalesce
+//		{NumRemoved: 2},                 // └── coalesce
+//		{NumIdentical: 63},              // group 4
+//	]
+//	Output: [
+//		{NumIdentical: 61},
+//		{NumIdentical: 7, NumRemoved: 12, NumInserted: 3},
+//		{NumIdentical: 64},
+//		{NumIdentical: 8, NumRemoved: 12, NumInserted: 3},
+//		{NumIdentical: 63},
+//	]
+//
 func coalesceInterveningIdentical(groups []diffStats, windowSize int) []diffStats {
 	groups, groupsOrig := groups[:0], groups
 	for i, ds := range groupsOrig {
@ -446,3 +523,91 @@ func coalesceInterveningIdentical(groups []diffStats, windowSize int) []diffStat
 	}
 	return groups
 }
+
+// cleanupSurroundingIdentical scans through all unequal groups, and
+// moves any leading sequence of equal elements to the preceding equal group and
+// moves and trailing sequence of equal elements to the succeeding equal group.
+//
+// This is necessary since coalesceInterveningIdentical may coalesce edit groups
+// together such that leading/trailing spans of equal elements becomes possible.
+// Note that this can occur even with an optimal diffing algorithm.
+//
+// Example:
+//
+//	Input: [
+//		{NumIdentical: 61},
+//		{NumIdentical: 1 , NumRemoved: 11, NumInserted: 2}, // assume 3 leading identical elements
+//		{NumIdentical: 67},
+//		{NumIdentical: 7, NumRemoved: 12, NumInserted: 3},  // assume 10 trailing identical elements
+//		{NumIdentical: 54},
+//	]
+//	Output: [
+//		{NumIdentical: 64}, // incremented by 3
+//		{NumRemoved: 9},
+//		{NumIdentical: 67},
+//		{NumRemoved: 9},
+//		{NumIdentical: 64}, // incremented by 10
+//	]
+//
+func cleanupSurroundingIdentical(groups []diffStats, eq func(i, j int) bool) []diffStats {
+	var ix, iy int // indexes into sequence x and y
+	for i, ds := range groups {
+		// Handle equal group.
+		if ds.NumDiff() == 0 {
+			ix += ds.NumIdentical
+			iy += ds.NumIdentical
+			continue
+		}
+
+		// Handle unequal group.
+		nx := ds.NumIdentical + ds.NumRemoved + ds.NumModified
+		ny := ds.NumIdentical + ds.NumInserted + ds.NumModified
+		var numLeadingIdentical, numTrailingIdentical int
+		for i := 0; i < nx && i < ny && eq(ix+i, iy+i); i++ {
+			numLeadingIdentical++
+		}
+		for i := 0; i < nx && i < ny && eq(ix+nx-1-i, iy+ny-1-i); i++ {
+			numTrailingIdentical++
+		}
+		if numIdentical := numLeadingIdentical + numTrailingIdentical; numIdentical > 0 {
+			if numLeadingIdentical > 0 {
+				// Remove leading identical span from this group and
+				// insert it into the preceding group.
+				if i-1 >= 0 {
+					groups[i-1].NumIdentical += numLeadingIdentical
+				} else {
+					// No preceding group exists, so prepend a new group,
+					// but do so after we finish iterating over all groups.
+					defer func() {
+						groups = append([]diffStats{{Name: groups[0].Name, NumIdentical: numLeadingIdentical}}, groups...)
+					}()
+				}
+				// Increment indexes since the preceding group would have handled this.
+				ix += numLeadingIdentical
+				iy += numLeadingIdentical
+			}
+			if numTrailingIdentical > 0 {
+				// Remove trailing identical span from this group and
+				// insert it into the succeeding group.
+				if i+1 < len(groups) {
+					groups[i+1].NumIdentical += numTrailingIdentical
+				} else {
+					// No succeeding group exists, so append a new group,
+					// but do so after we finish iterating over all groups.
+					defer func() {
+						groups = append(groups, diffStats{Name: groups[len(groups)-1].Name, NumIdentical: numTrailingIdentical})
+					}()
+				}
+				// Do not increment indexes since the succeeding group will handle this.
+			}
+
+			// Update this group since some identical elements were removed.
+			nx -= numIdentical
+			ny -= numIdentical
+			groups[i] = diffStats{Name: ds.Name, NumRemoved: nx, NumInserted: ny}
+		}
+		ix += nx
+		iy += ny
+	}
+	return groups
+}
--- a/vendor/modules.txt
+++ b/vendor/modules.txt
@ -1,4 +1,4 @@
-# github.com/google/go-cmp v0.5.4
+# github.com/google/go-cmp v0.5.6
 ## explicit
 github.com/google/go-cmp/cmp
 github.com/google/go-cmp/cmp/internal/diff