fix bug about invalid utf8 in export cmd

Author: skoowoo

internal/episode/obsidian.go

@@ -101,11 +101,17 @@ var nonSafeFilenameRe = regexp.MustCompile(`[^\p{L}\p{N}\-_ ]+`)
 
 // sanitizeFilename produces a filesystem-safe name from an arbitrary string.
 func sanitizeFilename(s string) string {
+	// Strip any invalid UTF-8 sequences first so the regex and filesystem
+	// are never handed malformed bytes.
+	s = strings.ToValidUTF8(s, "")
 	s = nonSafeFilenameRe.ReplaceAllString(s, "")
 	s = strings.TrimSpace(s)
 	s = strings.ReplaceAll(s, " ", "_")
-	if len(s) > 80 {
-		s = s[:80]
+	// Truncate by rune count, not byte count, to avoid splitting multi-byte
+	// characters (e.g. Chinese/Japanese) which would produce an illegal byte
+	// sequence that macOS and other systems reject.
+	if runes := []rune(s); len(runes) > 80 {
+		s = string(runes[:80])
 	}
 	if s == "" {
 		s = "episode"

internal/episode/utils.go

@@ -10,11 +10,12 @@ import (
 
 // ParseSeq extracts the integer episode seq from a podwise episode URL.
 // Expected format: https://podwise.ai/dashboard/episodes/<seq>
+// Also accepts https://app.podwise.ai/dashboard/episodes/<seq> and beta.podwise.ai.
 func ParseSeq(input string) (int, error) {
 	const hint = "(expected https://podwise.ai/dashboard/episodes/<id>)"
 
 	u, err := url.Parse(input)
-	if err != nil || u.Scheme != "https" || (u.Host != "podwise.ai" && u.Host != "beta.podwise.ai") {
+	if err != nil || u.Scheme != "https" || !isPodwiseHost(u.Host) {
 		return 0, fmt.Errorf("%q is not a valid podwise episode URL %s", input, hint)
 	}
 
@@ -68,6 +69,16 @@ func IsXiaoyuzhouURL(rawURL string) bool {
 	return u.Hostname() == "www.xiaoyuzhoufm.com" && strings.HasPrefix(u.Path, "/episode/")
 }
 
+// isPodwiseHost reports whether host is a recognised Podwise web host.
+// Accepted: podwise.ai, app.podwise.ai, beta.podwise.ai.
+func isPodwiseHost(host string) bool {
+	switch host {
+	case "podwise.ai", "app.podwise.ai", "beta.podwise.ai":
+		return true
+	}
+	return false
+}
+
 // trimTime removes a leading "00:" hour prefix from a time string only when
 // the remainder is still a valid mm:ss[-based] string (contains at least one
 // more colon). Examples:

internal/podcast/utils.go

@@ -7,13 +7,24 @@ import (
 	"strings"
 )
 
+// isPodwiseHost reports whether host is a recognised Podwise web host.
+// Accepted: podwise.ai, app.podwise.ai, beta.podwise.ai.
+func isPodwiseHost(host string) bool {
+	switch host {
+	case "podwise.ai", "app.podwise.ai", "beta.podwise.ai":
+		return true
+	}
+	return false
+}
+
 // ParseSeq extracts the integer podcast seq from a podwise podcast URL.
 // Expected format: https://podwise.ai/dashboard/podcasts/<seq>
+// Also accepts https://app.podwise.ai/dashboard/podcasts/<seq> and beta.podwise.ai.
 func ParseSeq(input string) (int, error) {
 	const hint = "(expected https://podwise.ai/dashboard/podcasts/<id>)"
 
 	u, err := url.Parse(input)
-	if err != nil || u.Scheme != "https" || (u.Host != "podwise.ai" && u.Host != "beta.podwise.ai") {
+	if err != nil || u.Scheme != "https" || !isPodwiseHost(u.Host) {
 		return 0, fmt.Errorf("%q is not a valid podwise podcast URL %s", input, hint)
 	}