From efaedcc6b2bd83e78254897f58b86a469eadba8a Mon Sep 17 00:00:00 2001
From: antanst <antanst@antanst.com>
Date: Thu, 27 Feb 2025 09:20:22 +0200
Subject: [PATCH] Improvements in error handling & descriptions

---
 common/text/text.go  |  9 +++++++++
 gemini/errors.go     | 21 +++++++++++++++------
 gemini/network.go    |  3 ++-
 gemini/processing.go |  5 +++--
 gopher/network.go    |  9 +++------
 5 files changed, 32 insertions(+), 15 deletions(-)
 create mode 100644 common/text/text.go

diff --git a/common/text/text.go b/common/text/text.go
new file mode 100644
index 0000000..f977122
--- /dev/null
+++ b/common/text/text.go
@@ -0,0 +1,9 @@
+package text
+
+import "strings"
+
+// RemoveNullChars removes all null characters from the input string.
+func RemoveNullChars(input string) string {
+	// Replace all null characters with an empty string
+	return strings.ReplaceAll(input, "\u0000", "")
+}
diff --git a/gemini/errors.go b/gemini/errors.go
index 06f2756..29b1e70 100644
--- a/gemini/errors.go
+++ b/gemini/errors.go
@@ -20,21 +20,29 @@ func (e *GeminiError) Error() string {
 	return fmt.Sprintf("gemini error: code %d %s", e.Code, e.Msg)
 }
 
+// NewGeminiError creates a new GeminiError based on the status code and header.
+// Status codes are based on the Gemini protocol specification:
+// - 1x: Input required
+// - 2x: Success (not handled as errors)
+// - 3x: Redirect
+// - 4x: Temporary failure
+// - 5x: Permanent failure
+// - 6x: Client certificate required/rejected
 func NewGeminiError(code int, header string) error {
 	var msg string
 	switch {
 	case code >= 10 && code < 20:
-		msg = "needs input"
+		msg = fmt.Sprintf("input required: %s", header)
 	case code >= 30 && code < 40:
-		msg = "redirect"
+		msg = fmt.Sprintf("redirect: %s", header)
 	case code >= 40 && code < 50:
-		msg = "bad request"
+		msg = fmt.Sprintf("request failed: %s", header)
 	case code >= 50 && code < 60:
-		msg = "server error"
+		msg = fmt.Sprintf("server error: %s", header)
 	case code >= 60 && code < 70:
-		msg = "TLS error"
+		msg = fmt.Sprintf("TLS error: %s", header)
 	default:
-		msg = "unexpected Status code"
+		msg = fmt.Sprintf("unexpected status code %d: %s", code, header)
 	}
 	return &GeminiError{
 		Msg:    msg,
@@ -43,6 +51,7 @@ func NewGeminiError(code int, header string) error {
 	}
 }
 
+// IsGeminiError checks if the given error is a GeminiError.
 func IsGeminiError(err error) bool {
 	if err == nil {
 		return false
diff --git a/gemini/network.go b/gemini/network.go
index b2cd280..384f6b1 100644
--- a/gemini/network.go
+++ b/gemini/network.go
@@ -17,6 +17,7 @@ import (
 	_url "gemini-grc/common/url"
 	"gemini-grc/config"
 	"gemini-grc/logging"
+
 	"github.com/antanst/go_errors"
 	"github.com/guregu/null/v5"
 )
@@ -182,7 +183,7 @@ func processData(s snapshot.Snapshot, data []byte) (*snapshot.Snapshot, error) {
 	if mimeType == "text/gemini" {
 		validBody, err := BytesToValidUTF8(body)
 		if err != nil {
-			return nil, go_errors.NewError(err)
+			return nil, err
 		}
 		s.GemText = null.StringFrom(validBody)
 	} else {
diff --git a/gemini/processing.go b/gemini/processing.go
index d631778..0d86758 100644
--- a/gemini/processing.go
+++ b/gemini/processing.go
@@ -7,6 +7,7 @@ import (
 	"io"
 	"unicode/utf8"
 
+	"github.com/antanst/go_errors"
 	"golang.org/x/text/encoding/charmap"
 	"golang.org/x/text/encoding/japanese"
 	"golang.org/x/text/encoding/korean"
@@ -24,7 +25,7 @@ func BytesToValidUTF8(input []byte) (string, error) {
 	}
 	const maxSize = 10 * 1024 * 1024 // 10MB
 	if len(input) > maxSize {
-		return "", fmt.Errorf("%w: %d bytes (max %d)", ErrInputTooLarge, len(input), maxSize)
+		return "", go_errors.NewError(fmt.Errorf("%w: %d bytes (max %d)", ErrInputTooLarge, len(input), maxSize))
 	}
 	// remove NULL byte 0x00 (ReplaceAll accepts slices)
 	inputNoNull := bytes.ReplaceAll(input, []byte{byte(0)}, []byte{})
@@ -55,5 +56,5 @@ func BytesToValidUTF8(input []byte) (string, error) {
 		}
 	}
 
-	return "", fmt.Errorf("%w (tried %d encodings): %w", ErrUTF8Conversion, len(encodings), lastErr)
+	return "", go_errors.NewError(fmt.Errorf("%w (tried %d encodings): %w", ErrUTF8Conversion, len(encodings), lastErr))
 }
diff --git a/gopher/network.go b/gopher/network.go
index cd31a5f..e6c8c22 100644
--- a/gopher/network.go
+++ b/gopher/network.go
@@ -13,9 +13,11 @@ import (
 	errors2 "gemini-grc/common/errors"
 	"gemini-grc/common/linkList"
 	"gemini-grc/common/snapshot"
+	"gemini-grc/common/text"
 	_url "gemini-grc/common/url"
 	"gemini-grc/config"
 	"gemini-grc/logging"
+
 	"github.com/antanst/go_errors"
 	"github.com/guregu/null/v5"
 )
@@ -85,7 +87,7 @@ func Visit(url string) (*snapshot.Snapshot, error) {
 
 	isValidUTF8 := utf8.ValidString(string(data))
 	if isValidUTF8 {
-		s.GemText = null.StringFrom(removeNullChars(string(data)))
+		s.GemText = null.StringFrom(text.RemoveNullChars(string(data)))
 	} else {
 		s.Data = null.ValueFrom(data)
 	}
@@ -276,8 +278,3 @@ func getGopherPageLinks(content string) []string {
 
 	return links
 }
-
-func removeNullChars(input string) string {
-	// Replace all null characters with an empty string
-	return strings.ReplaceAll(input, "\u0000", "")
-}