Better error handling, many fixes all around

This commit is contained in:
2024-12-09 19:53:15 +02:00
parent b52d4f6532
commit 7a36614232
15 changed files with 520 additions and 233 deletions

View File

@@ -4,15 +4,18 @@ import (
"crypto/tls"
"errors"
"fmt"
"gemini-grc/logging"
"io"
"net"
gourl "net/url"
"regexp"
"slices"
"strconv"
"strings"
"time"
"gemini-grc/config"
"github.com/guregu/null/v5"
)
@@ -55,8 +58,7 @@ func ConnectAndGetData(url string) ([]byte, error) {
host := fmt.Sprintf("%s:%s", hostname, port)
// Establish the underlying TCP connection.
dialer := &net.Dialer{
Timeout: time.Duration(config.CONFIG.ResponseTimeout) * time.Second,
KeepAlive: 10 * time.Second,
Timeout: time.Duration(config.CONFIG.ResponseTimeout) * time.Second,
}
conn, err := dialer.Dial("tcp", host)
if err != nil {
@@ -95,7 +97,11 @@ func ConnectAndGetData(url string) ([]byte, error) {
var data []byte
// Send Gemini request to trigger server response.
_, err = tlsConn.Write([]byte(fmt.Sprintf("%s\r\n", url)))
// Fix for stupid server bug:
// Some servers return 'Header: 53 No proxying to other hosts or ports!'
// when the port is 1965 and is still specified explicitely in the URL.
_url, _ := ParseURL(url, "")
_, err = tlsConn.Write([]byte(fmt.Sprintf("%s\r\n", _url.StringNoDefaultPort())))
if err != nil {
return nil, fmt.Errorf("%w: %w", ErrNetworkCannotWrite, err)
}
@@ -111,9 +117,8 @@ func ConnectAndGetData(url string) ([]byte, error) {
if err != nil {
if errors.Is(err, io.EOF) {
break
} else {
return nil, fmt.Errorf("%w: %w", ErrNetwork, err)
}
return nil, fmt.Errorf("%w: %w", ErrNetwork, err)
}
}
return data, nil
@@ -121,7 +126,19 @@ func ConnectAndGetData(url string) ([]byte, error) {
// Visit given URL, using the Gemini protocol.
// Mutates given Snapshot with the data.
func Visit(s *Snapshot) error {
// In case of error, we store the error string
// inside snapshot and return the error.
func Visit(s *Snapshot) (err error) {
// Don't forget to also store error
// response code (if we have one)
defer func() {
if err != nil {
s.Error = null.StringFrom(err.Error())
if errors.As(err, new(*GeminiError)) {
s.ResponseCode = null.IntFrom(int64(err.(*GeminiError).Code))
}
}
}()
data, err := ConnectAndGetData(s.URL.String())
if err != nil {
return err
@@ -130,6 +147,9 @@ func Visit(s *Snapshot) error {
if err != nil {
return err
}
//marshalled, _ := json.MarshalIndent(pageData, "", " ")
//fmt.Printf("%s\n", marshalled)
s.Header = null.StringFrom(pageData.ResponseHeader)
s.ResponseCode = null.IntFrom(int64(pageData.ResponseCode))
s.MimeType = null.StringFrom(pageData.MimeType)
s.Lang = null.StringFrom(pageData.Lang)
@@ -142,24 +162,21 @@ func Visit(s *Snapshot) error {
return nil
}
// Update given snapshot with the
// Gemini header data: response code,
// mime type and lang (optional)
// processData returne results from
// parsing Gemini header data:
// Code, mime type and lang (optional)
// Returns error if header was invalid
func processData(data []byte) (*PageData, error) {
header, body, err := getHeadersAndData(data)
if err != nil {
return nil, err
}
code, mimeType, lang := getMimeTypeAndLang(header)
var geminiError error
logging.LogDebug("Header: %s", strings.TrimSpace(header))
if code != 20 {
geminiError = NewErrGeminiStatusCode(code, header)
return nil, NewErrGeminiStatusCode(code, header)
}
fmt.Printf("%v\n", header)
if geminiError != nil {
return nil, geminiError
}
pageData := PageData{
ResponseCode: code,
ResponseHeader: header,
@@ -201,11 +218,11 @@ func getHeadersAndData(data []byte) (string, []byte, error) {
// `20 text/gemini` (code, mimetype)
// `31 gemini://redirected.to/other/site` (code)
func getMimeTypeAndLang(headers string) (int, string, string) {
// Regex that parses code, mimetype & lang
re := regexp.MustCompile(`^(\d+)\s+([a-zA-Z0-9/\-+]+)(?:[;\s]+(lang=([a-zA-Z0-9-]+)))?\s*$`)
// Regex that parses code, mimetype & optional charset/lang parameters
re := regexp.MustCompile(`^(\d+)\s+([a-zA-Z0-9/\-+]+)(?:[;\s]+(?:(?:charset|lang)=([a-zA-Z0-9-]+)))?\s*$`)
matches := re.FindStringSubmatch(headers)
if matches == nil || len(matches) <= 1 {
// Try to get code at least.
// Try to get code at least
re := regexp.MustCompile(`^(\d+)\s+`)
matches := re.FindStringSubmatch(headers)
if matches == nil || len(matches) <= 1 {
@@ -222,6 +239,6 @@ func getMimeTypeAndLang(headers string) (int, string, string) {
return 0, "", ""
}
mimeType := matches[2]
lang := matches[4]
return code, mimeType, lang
param := matches[3] // This will capture either charset or lang value
return code, mimeType, param
}