skip duplicate cowfile images

go.mod

@@ -1,6 +1,6 @@
 module github.com/tmck-code/pokesay
 
-go 1.20
+go 1.23
 
 require (
 	github.com/fatih/color v1.18.0

src/bin/convert/png_convert.go

@@ -5,6 +5,8 @@ import (
 	"flag"
 	"fmt"
 	"os"
+	"path/filepath"
+	"strings"
 
 	"github.com/tmck-code/pokesay/src/bin"
 	"github.com/tmck-code/pokesay/src/pokedex"
@@ -52,9 +54,44 @@ func main() {
 
 	fmt.Println("Converting PNGs -> cowfiles")
 	pbar := bin.NewProgressBar(len(fpaths))
+
+	allData := make([]string, 0, len(fpaths))
+	nDuplicates, nFailures := 0, 0
+
 	for _, f := range fpaths {
-		pokedex.ConvertPngToCow(args.FromDir, f, args.ToDir, args.Padding)
+		data, err := pokedex.ConvertPngToCow(args.FromDir, f, args.ToDir, args.Padding)
+		if err != nil {
+			nFailures++
+			continue
+		}
+
+		// check if this cawfile is a duplicate of one that has already been written
+		found := false
+		for _, existingData := range allData {
+			if existingData == data {
+				found = true
+				break
+			}
+		}
+		if found {
+			fmt.Println("Skipping duplicate data for", f)
+			nDuplicates++
+			pbar.Add(1)
+			continue
+		}
+		allData = append(allData, data)
+
+		destDirpath := filepath.Join(
+			args.ToDir,
+			// strip the root "source dirpath" from the source path
+			// e.g. fpath: /a/b/c.txt sourceDir: /a/ -> b/c.txt
+			filepath.Dir(strings.ReplaceAll(f, args.FromDir, "")),
+		)
+		destFpath := filepath.Join(destDirpath, strings.ReplaceAll(filepath.Base(f), ".png", ".cow"))
+
+		pokedex.WriteToCowfile(data, destDirpath, destFpath)
 		pbar.Add(1)
 	}
-	fmt.Println("Finished converting", len(fpaths), "pokesprite PNGs", "-> cowfiles")
+	fmt.Println("Finished converting", len(fpaths), "pokesprite PNGs -> cowfiles")
+	fmt.Println("(skipped", nDuplicates, "duplicates and", nFailures, "failures)")
 }

src/pokedex/convert.go

@@ -107,15 +107,7 @@ func padLeft(cowfile []byte, n int) []string {
 	return converted
 }
 
-func ConvertPngToCow(sourceDirpath string, sourceFpath string, destDirpath string, extraPadding int) {
-	destDir := filepath.Join(
-		destDirpath,
-		// strip the root "source dirpath" from the source path
-		// e.g. fpath: /a/b/c.txt sourceDir: /a/ -> b/c.txt
-		filepath.Dir(strings.ReplaceAll(sourceFpath, sourceDirpath, "")),
-	)
-	// Ensure that the destination dir exists
-	os.MkdirAll(destDir, 0755)
+func ConvertPngToCow(sourceDirpath string, sourceFpath string, destDirpath string, extraPadding int) (string, error) {
 
 	// Trim the whitespace from the edges of the images. This helps with the conversion
 	autoCrop(sourceFpath)
@@ -125,20 +117,24 @@ func ConvertPngToCow(sourceDirpath string, sourceFpath string, destDirpath strin
 
 	if len(converted) == 0 {
 		failures = append(failures, sourceFpath)
-		return
+		return "", fmt.Errorf("failed to convert %s", sourceFpath)
 	}
+	final := stripEmptyLines(padLeft(converted, extraPadding))
+	return strings.Join(final, "\n") + COLOUR_RESET, nil
+}
+
+func WriteToCowfile(data string, destDirpath string, destFpath string) {
+	// Ensure that the destination dir exists
+	os.MkdirAll(destDirpath, 0755)
 
-	destFpath := filepath.Join(destDir, strings.ReplaceAll(filepath.Base(sourceFpath), ".png", ".cow"))
 	ostream, err := os.Create(destFpath)
 	Check(err)
 	defer ostream.Close()
 	writer := bufio.NewWriter(ostream)
 
-	final := stripEmptyLines(padLeft(converted, extraPadding))
-
-	// Join all of the lines back together, add colour reset sequence at the end
-	_, err = writer.WriteString(strings.Join(final, "\n") + COLOUR_RESET)
+	_, err = writer.WriteString(data)
 	Check(err)
 
-	writer.Flush()
+	err = writer.Flush()
+	Check(err)
 }