rescribe: make go generate skip already downloaded files, checking that checksum matches expected for safety

1 files changed, 60 insertions, 9 deletions

diff --git a/cmd/rescribe/getembeds.go b/cmd/rescribe/getembeds.go
index 91cd480..6ab9fa8 100644
--- a/cmd/rescribe/getembeds.go
+++ b/cmd/rescribe/getembeds.go
@@ -9,8 +9,12 @@
 package main
 
 import (
+	"bytes"
+	"crypto/sha256"
+	"encoding/hex"
 	"fmt"
 	"io"
+	"io/ioutil"
 	"net/http"
 	"os"
 	"path"
@@ -39,19 +43,66 @@ func dl(url string) error {
 	return nil
 }
 
+// present returns true if the file is present and matches the
+// checksum, false otherwise
+func present(url string, sum string) bool {
+	fn := path.Base(url)
+	_, err := os.Stat(fn)
+	if err != nil && !os.IsExist(err) {
+		return false
+	}
+
+	b, err := ioutil.ReadFile(fn)
+	if err != nil {
+		return false
+	}
+
+	expected, err := hex.DecodeString(sum)
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "Error decoding checksum for %s: %v\n", url, err)
+		os.Exit(1)
+	}
+
+	actual := sha256.Sum256(b)
+
+	var a []byte
+	for _, v := range actual {
+		a = append(a, v)
+	}
+
+	if !bytes.Equal(a, expected) {
+		return false
+	}
+
+	return true
+}
+
 func main() {
-	urls := []string{
-		"https://rescribe.xyz/rescribe/embeds/tessdata.20211001.zip",
-		"https://rescribe.xyz/rescribe/embeds/tesseract-linux-v5.0.0-alpha.20210510.zip",
-		"https://rescribe.xyz/rescribe/embeds/tesseract-osx-v4.1.1.20191227.zip",
-		"https://rescribe.xyz/rescribe/embeds/tesseract-osx-m1-v4.1.1.20210802.zip",
-		"https://rescribe.xyz/rescribe/embeds/tesseract-w32-v5.0.0-alpha.20210506.zip",
+	urls := []struct{
+		url string
+		sum string
+	}{
+		{"https://rescribe.xyz/rescribe/embeds/tessdata.20211001.zip", "5c90ae69b9e449d85e84b4806a54d6739b572730525010483e512a62a527b030"},
+		{"https://rescribe.xyz/rescribe/embeds/tesseract-linux-v5.0.0-alpha.20210510.zip", "81cfba632b8aaf0a00180b1aa62d357d50f343b0e9bd51b941ee14c289ccd889"},
+		{"https://rescribe.xyz/rescribe/embeds/tesseract-osx-v4.1.1.20191227.zip", "5f567b95f1dea9d0581ad42ada4d1f1160a38ea22ae338f9efe190015265636b"},
+		{"https://rescribe.xyz/rescribe/embeds/tesseract-osx-m1-v4.1.1.20210802.zip", "c9a454633f7e5175e2d50dd939d30a6e5bdfb3b8c78590a08b5aa21edbf32ca4"},
+		{"https://rescribe.xyz/rescribe/embeds/tesseract-w32-v5.0.0-alpha.20210506.zip", "96734f3db4bb7c3b9a241ab6d89ab3e8436cea43b1cbbcfb13999497982f63e3"},
 	}
 	for _, v := range urls {
-		fmt.Printf("Downloading %s\n", v)
-		err := dl(v)
+		if present(v.url, v.sum) {
+			fmt.Printf("Skipping downloading of already present %s\n", path.Base(v.url))
+			continue
+		}
+
+		fmt.Printf("Downloading %s\n", v.url)
+		err := dl(v.url)
 		if err != nil {
-			fmt.Printf("Error downloading %s: %v\n", v, err)
+			fmt.Fprintf(os.Stderr, "Error downloading %s: %v\n", v.url, err)
+			os.Exit(1)
+		}
+
+		if !present(v.url, v.sum) {
+			fmt.Fprintf(os.Stderr, "Error: downloaded %s does not match expected checksum: %v\n", v.url, err)
 			os.Exit(1)
 		}
 	}