Switch EXIF library

Closes #10855
Closes #8586
Closes #8996

common/hugio/readers.go

@@ -23,6 +23,7 @@ import (
 type ReadSeeker interface {
 	io.Reader
 	io.Seeker
+	io.ReaderAt
 }
 
 // ReadSeekCloser is implemented by afero.File. We use this as the common type for

go.mod

@@ -15,6 +15,7 @@ require (
 	github.com/bep/golibsass v1.1.1
 	github.com/bep/gowebp v0.3.0
 	github.com/bep/helpers v0.4.0
+	github.com/bep/imagemeta v0.3.0
 	github.com/bep/lazycache v0.4.0
 	github.com/bep/logg v0.4.0
 	github.com/bep/mclib v1.20400.20402
@@ -60,7 +61,6 @@ require (
 	github.com/pbnjay/memory v0.0.0-20210728143218-7b4eea64cf58
 	github.com/pelletier/go-toml/v2 v2.2.2
 	github.com/rogpeppe/go-internal v1.12.0
-	github.com/rwcarlsen/goexif v0.0.0-20190401172101-9e8deecbddbd
 	github.com/sanity-io/litter v1.5.5
 	github.com/spf13/afero v1.11.0
 	github.com/spf13/cast v1.6.0

go.sum

@@ -132,6 +132,12 @@ github.com/bep/gowebp v0.3.0 h1:MhmMrcf88pUY7/PsEhMgEP0T6fDUnRTMpN8OclDrbrY=
 github.com/bep/gowebp v0.3.0/go.mod h1:ZhFodwdiFp8ehGJpF4LdPl6unxZm9lLFjxD3z2h2AgI=
 github.com/bep/helpers v0.4.0 h1:ab9veaAiWY4ST48Oxp5usaqivDmYdB744fz+tcZ3Ifs=
 github.com/bep/helpers v0.4.0/go.mod h1:/QpHdmcPagDw7+RjkLFCvnlUc8lQ5kg4KDrEkb2Yyco=
+github.com/bep/imagemeta v0.1.0 h1:Y/bE6q5HYgF3tEYL1dF2oprEYiuyCiADokp1Yc+pvDo=
+github.com/bep/imagemeta v0.1.0/go.mod h1:FAcTxuAx8JxWVelC6VTK4GKxsHUzXEtR5lwKZG3Oupk=
+github.com/bep/imagemeta v0.2.0 h1:938HCN0vtTpFW/Rs+HYiqkjFbLDojeR2VJPp/S+hIRU=
+github.com/bep/imagemeta v0.2.0/go.mod h1:FAcTxuAx8JxWVelC6VTK4GKxsHUzXEtR5lwKZG3Oupk=
+github.com/bep/imagemeta v0.3.0 h1:2dSTgp0RA8s8KUA8PKlnczVyOuvJ5bd553V1VijUXTo=
+github.com/bep/imagemeta v0.3.0/go.mod h1:FAcTxuAx8JxWVelC6VTK4GKxsHUzXEtR5lwKZG3Oupk=
 github.com/bep/lazycache v0.4.0 h1:X8yVyWNVupPd4e1jV7efi3zb7ZV/qcjKQgIQ5aPbkYI=
 github.com/bep/lazycache v0.4.0/go.mod h1:NmRm7Dexh3pmR1EignYR8PjO2cWybFQ68+QgY3VMCSc=
 github.com/bep/logg v0.4.0 h1:luAo5mO4ZkhA5M1iDVDqDqnBBnlHjmtZF6VAyTp+nCQ=
@@ -405,7 +411,6 @@ github.com/rogpeppe/go-internal v1.12.0/go.mod h1:E+RYuTGaKKdloAfM02xzb0FW3Paa99
 github.com/russross/blackfriday/v2 v2.1.0 h1:JIOH55/0cWyOuilr9/qlrm0BSXldqnqwMsf35Ld67mk=
 github.com/russross/blackfriday/v2 v2.1.0/go.mod h1:+Rmxgy9KzJVeS9/2gXHxylqXiyQDYRxCVz55jmeOWTM=
 github.com/rwcarlsen/goexif v0.0.0-20190401172101-9e8deecbddbd h1:CmH9+J6ZSsIjUK3dcGsnCnO41eRBOnY12zwkn5qVwgc=
-github.com/rwcarlsen/goexif v0.0.0-20190401172101-9e8deecbddbd/go.mod h1:hPqNNc0+uJM6H+SuU8sEs5K5IQeKccPqeSjfgcKGgPk=
 github.com/sanity-io/litter v1.5.5 h1:iE+sBxPBzoK6uaEP5Lt3fHNgpKcHXc/A2HGETy0uJQo=
 github.com/sanity-io/litter v1.5.5/go.mod h1:9gzJgR2i4ZpjZHsKvUXIRQVk7P+yM3e+jAF7bU2UI5U=
 github.com/shogo82148/go-shuffle v0.0.0-20180218125048-27e6095f230d/go.mod h1:2htx6lmL0NGLHlO8ZCf+lQBGBHIbEujyywxJArf+2Yc=

hugolib/image_test.go

@@ -83,11 +83,11 @@ SUNSET2: {{ $resized2.RelPermalink }}/{{ $resized2.Width }}/Lat: {{ $resized2.Ex
 	b.AssertImage(200, 200, "resources/_gen/images/bundle/sunset_hu59e56ffff1bc1d8d122b1403d34e039f_90587_200x200_resize_q75_box.jpg")
 
 	b.AssertFileContent("resources/_gen/images/bundle/sunset_3166614710256882113.json",
-		"DateTimeDigitized|time.Time", "PENTAX")
+		"FocalLengthIn35mmFormat|uint16", "PENTAX")
 
 	b.AssertImage(123, 234, "resources/_gen/images/images/sunset_hu59e56ffff1bc1d8d122b1403d34e039f_90587_123x234_resize_q75_box.jpg")
 	b.AssertFileContent("resources/_gen/images/images/sunset_3166614710256882113.json",
-		"DateTimeDigitized|time.Time", "PENTAX")
+		"FocalLengthIn35mmFormat|uint16", "PENTAX")
 
 	b.AssertNoDuplicateWrites()
 }

resources/image.go

@@ -82,8 +82,9 @@ func (i *imageResource) Exif() *exif.ExifInfo {
 
 func (i *imageResource) getExif() *exif.ExifInfo {
 	i.metaInit.Do(func() {
-		supportsExif := i.Format == images.JPEG || i.Format == images.TIFF
-		if !supportsExif {
+		mf := i.Format.ToImageMetaImageFormatFormat()
+		if mf == -1 {
+			// No Exif support for this format.
 			return
 		}
 
@@ -114,7 +115,7 @@ func (i *imageResource) getExif() *exif.ExifInfo {
 			}
 			defer f.Close()
 
-			x, err := i.getSpec().imaging.DecodeExif(f)
+			x, err := i.getSpec().imaging.DecodeExif(mf, f)
 			if err != nil {
 				i.getSpec().Logger.Warnf("Unable to decode Exif metadata from image: %s", i.Key())
 				return nil

resources/image_test.go

@@ -19,7 +19,6 @@ import (
 	"image"
 	"image/gif"
 	"io/fs"
-	"math/big"
 	"math/rand"
 	"os"
 	"path/filepath"
@@ -30,6 +29,7 @@ import (
 	"testing"
 	"time"
 
+	"github.com/bep/imagemeta"
 	"github.com/gohugoio/hugo/htesting"
 	"github.com/gohugoio/hugo/resources/images/webp"
 
@@ -67,8 +67,13 @@ var eq = qt.CmpEquals(
 		return m1.Type == m2.Type
 	}),
 	cmp.Comparer(
-		func(v1, v2 *big.Rat) bool {
-			return v1.RatString() == v2.RatString()
+		func(v1, v2 imagemeta.Rat[uint32]) bool {
+			return v1.String() == v2.String()
+		},
+	),
+	cmp.Comparer(
+		func(v1, v2 imagemeta.Rat[int32]) bool {
+			return v1.String() == v2.String()
 		},
 	),
 	cmp.Comparer(func(v1, v2 time.Time) bool {
@@ -392,7 +397,7 @@ func TestImageResize8BitPNG(t *testing.T) {
 	c.Assert(image.MediaType().Type, qt.Equals, "image/png")
 	c.Assert(image.RelPermalink(), qt.Equals, "/a/gohugoio.png")
 	c.Assert(image.ResourceType(), qt.Equals, "image")
-	c.Assert(image.Exif(), qt.IsNil)
+	c.Assert(image.Exif(), qt.IsNotNil)
 
 	resized, err := image.Resize("800x")
 	c.Assert(err, qt.IsNil)

resources/images/exif/exif.go

@@ -14,21 +14,15 @@
 package exif
 
 import (
-	"bytes"
 	"fmt"
 	"io"
-	"math"
-	"math/big"
 	"regexp"
+	"strconv"
 	"strings"
 	"time"
-	"unicode"
-	"unicode/utf8"
 
+	"github.com/bep/imagemeta"
 	"github.com/bep/tmc"
-
-	_exif "github.com/rwcarlsen/goexif/exif"
-	"github.com/rwcarlsen/goexif/tiff"
 )
 
 const exifTimeLayout = "2006:01:02 15:04:05"
@@ -55,6 +49,14 @@ type Decoder struct {
 	noLatLong        bool
 }
 
+func (d *Decoder) shouldInclude(s string) bool {
+	return (d.includeFieldsRe == nil || d.includeFieldsRe.MatchString(s))
+}
+
+func (d *Decoder) shouldExclude(s string) bool {
+	return d.excludeFieldsrRe != nil && d.excludeFieldsrRe.MatchString(s)
+}
+
 func IncludeFields(expression string) func(*Decoder) error {
 	return func(d *Decoder) error {
 		re, err := compileRegexp(expression)
@@ -115,148 +117,120 @@ func NewDecoder(options ...func(*Decoder) error) (*Decoder, error) {
 	return d, nil
 }
 
-func (d *Decoder) Decode(r io.Reader) (ex *ExifInfo, err error) {
+var (
+	isTimeTag = func(s string) bool {
+		return strings.Contains(s, "Time")
+	}
+	isGPSTag = func(s string) bool {
+		return strings.HasPrefix(s, "GPS")
+	}
+)
+
+// TODO1 bump the exif cache version number.
+func (d *Decoder) Decode(format imagemeta.ImageFormat, r io.Reader) (ex *ExifInfo, err error) {
 	defer func() {
 		if r := recover(); r != nil {
 			err = fmt.Errorf("exif failed: %v", r)
 		}
 	}()
 
-	var x *_exif.Exif
-	x, err = _exif.Decode(r)
-	if err != nil {
-		if err.Error() == "EOF" {
-			// Found no Exif
-			return nil, nil
-		}
-		return
+	var tagInfos imagemeta.Tags
+	handleTag := func(ti imagemeta.TagInfo) error {
+		tagInfos.Add(ti)
+		return nil
 	}
 
-	var tm time.Time
-	var lat, long float64
-
-	if !d.noDate {
-		tm, _ = x.DateTime()
-	}
+	shouldInclude := func(ti imagemeta.TagInfo) bool {
+		if ti.Source == imagemeta.EXIF {
+			if !d.noDate {
+				// We need the time tags to calculate the date.
+				if isTimeTag(ti.Tag) {
+					return true
+				}
+			}
+			if !d.noLatLong {
+				// We need to GPS tags to calculate the lat/long.
+				if isGPSTag(ti.Tag) {
+					return true
+				}
+			}
 
-	if !d.noLatLong {
-		lat, long, _ = x.LatLong()
-		if math.IsNaN(lat) {
-			lat = 0
+			if !strings.HasPrefix(ti.Namespace, "IFD0") {
+				// Drop thumbnail tags.
+				return false
+			}
 		}
-		if math.IsNaN(long) {
-			long = 0
+
+		if d.shouldExclude(ti.Tag) {
+			return false
 		}
-	}
 
-	walker := &exifWalker{x: x, vals: make(map[string]any), includeMatcher: d.includeFieldsRe, excludeMatcher: d.excludeFieldsrRe}
-	if err = x.Walk(walker); err != nil {
-		return
+		return d.shouldInclude(ti.Tag)
 	}
 
-	ex = &ExifInfo{Lat: lat, Long: long, Date: tm, Tags: walker.vals}
+	err = imagemeta.Decode(
+		imagemeta.Options{
+			R:               r.(imagemeta.Reader),
+			ImageFormat:     format,
+			ShouldHandleTag: shouldInclude,
+			HandleTag:       handleTag,
+			Sources:         imagemeta.EXIF, // For now. TODO1.
+		},
+	)
 
-	return
-}
+	var tm time.Time
+	var lat, long float64
 
-func decodeTag(x *_exif.Exif, f _exif.FieldName, t *tiff.Tag) (any, error) {
-	switch t.Format() {
-	case tiff.StringVal, tiff.UndefVal:
-		s := nullString(t.Val)
-		if strings.Contains(string(f), "DateTime") {
-			if d, err := tryParseDate(x, s); err == nil {
-				return d, nil
-			}
-		}
-		return s, nil
-	case tiff.OtherVal:
-		return "unknown", nil
+	if !d.noDate {
+		tm, _ = tagInfos.GetDateTime()
 	}
 
-	var rv []any
-
-	for i := 0; i < int(t.Count); i++ {
-		switch t.Format() {
-		case tiff.RatVal:
-			n, d, _ := t.Rat2(i)
-			rat := big.NewRat(n, d)
-			// if t is int or t > 1, use float64
-			if rat.IsInt() || rat.Cmp(big.NewRat(1, 1)) == 1 {
-				f, _ := rat.Float64()
-				rv = append(rv, f)
-			} else {
-				rv = append(rv, rat)
-			}
-
-		case tiff.FloatVal:
-			v, _ := t.Float(i)
-			rv = append(rv, v)
-		case tiff.IntVal:
-			v, _ := t.Int(i)
-			rv = append(rv, v)
-		}
+	if !d.noLatLong {
+		lat, long, _ = tagInfos.GetLatLong()
 	}
 
-	if t.Count == 1 {
-		if len(rv) == 1 {
-			return rv[0], nil
+	tags := make(map[string]any)
+	for k, v := range tagInfos.All() {
+		if d.shouldExclude(k) {
+			continue
 		}
+		if !d.shouldInclude(k) {
+			continue
+		}
+		tags[k] = v.Value
 	}
 
-	return rv, nil
-}
+	ex = &ExifInfo{Lat: lat, Long: long, Date: tm, Tags: tags}
 
-// Code borrowed from exif.DateTime and adjusted.
-func tryParseDate(x *_exif.Exif, s string) (time.Time, error) {
-	dateStr := strings.TrimRight(s, "\x00")
-	// TODO(bep): look for timezone offset, GPS time, etc.
-	timeZone := time.Local
-	if tz, _ := x.TimeZone(); tz != nil {
-		timeZone = tz
-	}
-	return time.ParseInLocation(exifTimeLayout, dateStr, timeZone)
+	return
 }
 
-type exifWalker struct {
-	x              *_exif.Exif
-	vals           map[string]any
-	includeMatcher *regexp.Regexp
-	excludeMatcher *regexp.Regexp
+func getDateTime(tags map[string]imagemeta.TagInfo) time.Time {
+	return time.Time{}
 }
 
-func (e *exifWalker) Walk(f _exif.FieldName, tag *tiff.Tag) error {
-	name := string(f)
-	if e.excludeMatcher != nil && e.excludeMatcher.MatchString(name) {
-		return nil
-	}
-	if e.includeMatcher != nil && !e.includeMatcher.MatchString(name) {
-		return nil
-	}
-	val, err := decodeTag(e.x, f, tag)
-	if err != nil {
-		return err
-	}
-	e.vals[name] = val
-	return nil
-}
+var tcodec *tmc.Codec
 
-func nullString(in []byte) string {
-	var rv bytes.Buffer
-	for len(in) > 0 {
-		r, size := utf8.DecodeRune(in)
-		if unicode.IsGraphic(r) {
-			rv.WriteRune(r)
-		}
-		in = in[size:]
+func init() {
+	tmcAdapters := []tmc.Adapter{
+		tmc.NewAdapter(imagemeta.NewRat[uint32](1, 2), nil, nil),
+		tmc.NewAdapter(imagemeta.NewRat[int32](1, 2), nil, nil),
+		// TODO1, if we don't remove the cache, this needs to be expanded to support other
+		tmc.NewAdapter(uint16(1),
+			func(s string) (interface{}, error) {
+				i, err := strconv.ParseUint(s, 10, 16)
+				return uint16(i), err
+			},
+			func(v interface{}) (string, error) {
+				vv := v.(uint16)
+				return strconv.Itoa(int(vv)), nil
+			}),
 	}
-	return rv.String()
-}
 
-var tcodec *tmc.Codec
+	tmcAdapters = append(tmc.DefaultTypeAdapters, tmcAdapters...)
 
-func init() {
 	var err error
-	tcodec, err = tmc.New()
+	tcodec, err = tmc.New(tmc.WithTypeAdapters(tmcAdapters))
 	if err != nil {
 		panic(err)
 	}