Support data URLs in Google Doc exports.

claat/fetch/fetch.go

@@ -229,7 +229,7 @@ func (f *Fetcher) SlurpImages(src, dir string, n []nodes.Node, images map[string
 	for _, imageNode := range imageNodes {
 		go func(imageNode *nodes.ImageNode) {
 			url := imageNode.Src
-			file, err := f.slurpBytes(src, dir, url)
+			file, err := f.slurpBytes(src, dir, url, imageNode.Bytes)
 			if err == nil {
 				imageNode.Src = filepath.Join(util.ImgDirname, file)
 			}
@@ -251,40 +251,52 @@ func (f *Fetcher) SlurpImages(src, dir string, n []nodes.Node, images map[string
 	return nil
 }
 
-func (f *Fetcher) slurpBytes(codelabSrc, dir, imgURL string) (string, error) {
-	// images can be local in Markdown cases or remote.
+func (f *Fetcher) slurpBytes(codelabSrc, dir, imgURL string, imgBytes []byte) (string, error) {
+	// images can be data URLs, local in Markdown cases or remote.
 	// Only proceed a simple copy on local reference.
 	var b []byte
 	var ext string
-	u, err := url.Parse(imgURL)
-	if err != nil {
-		return "", err
-	}
-
-	// If the codelab source is being downloaded from the network, then we should interpret
-	// the image URL in the same way.
-	srcUrl, err := url.Parse(codelabSrc)
-	if err == nil && srcUrl.Host != "" {
-		u = srcUrl.ResolveReference(u)
-	}
+	var err error
 
-	if u.Host == "" {
-		if imgURL, err = restrictPathToParent(imgURL, filepath.Dir(codelabSrc)); err != nil {
-			return "", err
+	if len(imgBytes) > 0 {
+		// Slurp bytes from image URL data.
+		b = imgBytes
+		if ext, err = imgExtFromBytes(b); err != nil {
+			return "", fmt.Errorf("Error reading image type: %v", err)
 		}
-		if b, err = ioutil.ReadFile(imgURL); err != nil {
+	} else {
+		// Slurp bytes from local or remote URL.
+		u, err := url.Parse(imgURL)
+		if err != nil {
 			return "", err
 		}
-		ext = filepath.Ext(imgURL)
-	} else {
-		if b, err = f.slurpRemoteBytes(u.String(), 5); err != nil {
-			return "", fmt.Errorf("Error downloading image at %s: %v", u.String(), err)
+
+		// If the codelab source is being downloaded from the network, then we should interpret
+		// the image URL in the same way.
+		srcURL, err := url.Parse(codelabSrc)
+		if err == nil && srcURL.Host != "" {
+			u = srcURL.ResolveReference(u)
 		}
-		if ext, err = imgExtFromBytes(b); err != nil {
-			return "", fmt.Errorf("Error reading image type at %s: %v", u.String(), err)
+
+		if u.Host == "" {
+			if imgURL, err = restrictPathToParent(imgURL, filepath.Dir(codelabSrc)); err != nil {
+				return "", err
+			}
+			if b, err = ioutil.ReadFile(imgURL); err != nil {
+				return "", err
+			}
+			ext = filepath.Ext(imgURL)
+		} else {
+			if b, err = f.slurpRemoteBytes(u.String(), 5); err != nil {
+				return "", fmt.Errorf("Error downloading image at %s: %v", u.String(), err)
+			}
+			if ext, err = imgExtFromBytes(b); err != nil {
+				return "", fmt.Errorf("Error reading image type at %s: %v", u.String(), err)
+			}
 		}
 	}
 
+	// Generate image file from slurped bytes.
 	crc := crc64.Checksum(b, f.crcTable)
 	file := fmt.Sprintf("%x%s", crc, ext)
 	dst := filepath.Join(dir, file)

claat/nodes/image.go

@@ -7,6 +7,7 @@ type NewImageNodeOptions struct {
 	Width float32
 	Alt   string
 	Title string
+	Bytes []byte
 }
 
 // NewImageNode creates a new ImageNode with the given options.
@@ -18,6 +19,7 @@ func NewImageNode(opts NewImageNodeOptions) *ImageNode {
 		Width: opts.Width,
 		Alt:   opts.Alt,
 		Title: opts.Title,
+		Bytes: opts.Bytes,
 	}
 }
 
@@ -28,11 +30,12 @@ type ImageNode struct {
 	Width float32
 	Alt   string
 	Title string
+	Bytes []byte
 }
 
 // Empty returns true if its Src is zero, excluding space runes.
 func (in *ImageNode) Empty() bool {
-	return strings.TrimSpace(in.Src) == ""
+	return strings.TrimSpace(in.Src) == "" && len(in.Bytes) == 0
 }
 
 // ImageNodes extracts everything except NodeImage nodes, recursively.

claat/nodes/image_test.go

@@ -1,11 +1,14 @@
 package nodes
 
 import (
+	"encoding/base64"
 	"testing"
 
 	"github.com/google/go-cmp/cmp"
 )
 
+var testBytes, _ = base64.StdEncoding.DecodeString("R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7")
+
 func TestNewImageNode(t *testing.T) {
 	tests := []struct {
 		name   string
@@ -19,7 +22,7 @@ func TestNewImageNode(t *testing.T) {
 			},
 		},
 		{
-			name: "NonEmpty",
+			name: "StandardURL",
 			inOpts: NewImageNodeOptions{
 				Src:   "https://www.google.com/images/branding/googlelogo/1x/googlelogo_color_272x92dp.png",
 				Width: 1.0,
@@ -34,6 +37,22 @@ func TestNewImageNode(t *testing.T) {
 				Alt:   "bar",
 			},
 		},
+		{
+			name: "DataURL",
+			inOpts: NewImageNodeOptions{
+				Width: 1.0,
+				Title: "foo",
+				Alt:   "bar",
+				Bytes: testBytes,
+			},
+			out: &ImageNode{
+				node:  node{typ: NodeImage},
+				Width: 1.0,
+				Title: "foo",
+				Alt:   "bar",
+				Bytes: testBytes,
+			},
+		},
 	}
 	for _, tc := range tests {
 		t.Run(tc.name, func(t *testing.T) {

claat/parser/gdoc/parse.go

@@ -16,6 +16,7 @@ package gdoc
 
 import (
 	"bytes"
+	"encoding/base64"
 	"fmt"
 	"io"
 	"net/url"
@@ -697,12 +698,32 @@ func image(ds *docState) nodes.Node {
 		errorAlt = "The domain of the requested iframe (" + u.Hostname() + ") has not been whitelisted."
 		fmt.Fprint(os.Stderr, errorAlt+"\n")
 	}
+
+	var imageBytes []byte
+	var imageSrc string
 	s := nodeAttr(ds.cur, "src")
 	if s == "" {
 		return nil
+	} else if strings.HasPrefix(s, "data:") {
+		_, data, ok := strings.Cut(s, ",")
+		if !ok {
+			fmt.Fprint(os.Stderr, "Failed to decode data URL: "+s+" \n")
+			return nil
+		}
+		b, err := base64.StdEncoding.DecodeString(data)
+		if err != nil {
+			fmt.Fprint(os.Stderr, "Failed to decode data URL: "+s+"\n"+err.Error()+"\n")
+			return nil
+		}
+		imageSrc = ""
+		imageBytes = b
+	} else {
+		imageSrc = s
+		imageBytes = []byte{}
 	}
 	n := nodes.NewImageNode(nodes.NewImageNodeOptions{
-		Src:   s,
+		Src:   imageSrc,
+		Bytes: imageBytes,
 		Width: styleFloatValue(ds.cur, "width"),
 	})
 	n.MutateBlock(findBlockParent(ds.cur))

claat/parser/gdoc/parse_test.go

@@ -16,6 +16,7 @@ package gdoc
 
 import (
 	"bytes"
+	"encoding/base64"
 	"io"
 	"reflect"
 	"strings"
@@ -295,6 +296,9 @@ func TestParseDoc(t *testing.T) {
 		<p><span>[[</span><span class="bold">import</span><span>&nbsp;</span><span><a href="https://example.com/import">shared</a></span><span>]]</span></p>
 
 		<img src="https://host/image.png" alt="alt text" title="title text">
+		<p><img alt="JPEG" src="data:image/jpeg;base64,/9j/2wBDAP//////////////////////////////////////////////////////////////////////////////////////wAALCAABAAEBAREA/8QAFAABAAAAAAAAAAAAAAAAAAAAA//EABQQAQAAAAAAAAAAAAAAAAAAAAD/2gAIAQEAAD8AN//Z"></p>
+		<p><img alt="GIF" src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7"></p>
+		<p><img alt="PNG" src="data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVQYV2NgYAAAAAMAAWgmWQ0AAAAASUVORK5CYII="></p>
 		<p><img src="https://host/small.png" style="height: 10px; width: 25.5px"> icon.</p>
 
 		<p><img alt="https://www.youtube.com/watch?v=vid" src="https://yt.com/vid.jpg"></p>
@@ -405,6 +409,33 @@ func TestParseDoc(t *testing.T) {
 	para.MutateBlock(true)
 	content.Append(para)
 
+	bytes, _ := base64.StdEncoding.DecodeString("/9j/2wBDAP//////////////////////////////////////////////////////////////////////////////////////wAALCAABAAEBAREA/8QAFAABAAAAAAAAAAAAAAAAAAAAA//EABQQAQAAAAAAAAAAAAAAAAAAAAD/2gAIAQEAAD8AN//Z")
+	img = nodes.NewImageNode(nodes.NewImageNodeOptions{
+		Bytes: bytes,
+		Alt:   "JPEG",
+	})
+	para = nodes.NewListNode(img)
+	para.MutateBlock(true)
+	content.Append(para)
+
+	bytes, _ = base64.StdEncoding.DecodeString("R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7")
+	img = nodes.NewImageNode(nodes.NewImageNodeOptions{
+		Bytes: bytes,
+		Alt:   "GIF",
+	})
+	para = nodes.NewListNode(img)
+	para.MutateBlock(true)
+	content.Append(para)
+
+	bytes, _ = base64.StdEncoding.DecodeString("iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVQYV2NgYAAAAAMAAWgmWQ0AAAAASUVORK5CYII=")
+	img = nodes.NewImageNode(nodes.NewImageNodeOptions{
+		Bytes: bytes,
+		Alt:   "PNG",
+	})
+	para = nodes.NewListNode(img)
+	para.MutateBlock(true)
+	content.Append(para)
+
 	img = nodes.NewImageNode(nodes.NewImageNodeOptions{
 		Src:   "https://host/small.png",
 		Width: 25.5,