parse xml using the encoding declared in xml header

mayswind · Oct 30, 2024 · 5006a96 · 5006a96
1 parent ce8c020
commit 5006a96
Show file tree

Hide file tree

Showing 4 changed files with 11 additions and 12 deletions.
diff --git a/pkg/converters/gnucash/gnucash_data_reader.go b/pkg/converters/gnucash/gnucash_data_reader.go
@@ -5,9 +5,10 @@ import (
 	"compress/gzip"
 	"encoding/xml"
 
+	"golang.org/x/net/html/charset"
+
 	"github.com/mayswind/ezbookkeeping/pkg/core"
 	"github.com/mayswind/ezbookkeeping/pkg/errs"
-	"github.com/mayswind/ezbookkeeping/pkg/utils"
 )
 
 // gnucashDatabaseReader defines the structure of gnucash database reader
@@ -37,14 +38,14 @@ func createNewGnuCashDatabaseReader(data []byte) (*gnucashDatabaseReader, error)
 		}
 
 		xmlDecoder := xml.NewDecoder(gzipReader)
-		xmlDecoder.CharsetReader = utils.IdentReader
+		xmlDecoder.CharsetReader = charset.NewReaderLabel
 
 		return &gnucashDatabaseReader{
 			xmlDecoder: xmlDecoder,
 		}, nil
 	} else if len(data) > 5 && data[0] == 0x3C && data[1] == 0x3F && data[2] == 0x78 && data[3] == 0x6D && data[4] == 0x6C { // <?xml
 		xmlDecoder := xml.NewDecoder(bytes.NewReader(data))
-		xmlDecoder.CharsetReader = utils.IdentReader
+		xmlDecoder.CharsetReader = charset.NewReaderLabel
 
 		return &gnucashDatabaseReader{
 			xmlDecoder: xmlDecoder,

diff --git a/pkg/converters/ofx/ofx_data_reader.go b/pkg/converters/ofx/ofx_data_reader.go
@@ -4,9 +4,10 @@ import (
 	"bytes"
 	"encoding/xml"
 
+	"golang.org/x/net/html/charset"
+
 	"github.com/mayswind/ezbookkeeping/pkg/core"
 	"github.com/mayswind/ezbookkeeping/pkg/errs"
-	"github.com/mayswind/ezbookkeeping/pkg/utils"
 )
 
 // ofxFileReader defines the structure of open financial exchange (ofx) file reader
@@ -30,7 +31,7 @@ func (r *ofxFileReader) read(ctx core.Context) (*ofxFile, error) {
 func createNewOFXFileReader(data []byte) (*ofxFileReader, error) {
 	if len(data) > 5 && data[0] == 0x3C && data[1] == 0x3F && data[2] == 0x78 && data[3] == 0x6D && data[4] == 0x6C { // ofx 2.x starts with <?xml
 		xmlDecoder := xml.NewDecoder(bytes.NewReader(data))
-		xmlDecoder.CharsetReader = utils.IdentReader
+		xmlDecoder.CharsetReader = charset.NewReaderLabel
 
 		return &ofxFileReader{
 			xmlDecoder: xmlDecoder,
@@ -39,7 +40,7 @@ func createNewOFXFileReader(data []byte) (*ofxFileReader, error) {
 
 	} else if len(data) > 5 && string(data[0:5]) == "<OFX>" { // no ofx header
 		xmlDecoder := xml.NewDecoder(bytes.NewReader(data))
-		xmlDecoder.CharsetReader = utils.IdentReader
+		xmlDecoder.CharsetReader = charset.NewReaderLabel
 
 		return &ofxFileReader{
 			xmlDecoder: xmlDecoder,

diff --git a/pkg/exchangerates/national_bank_of_poland_datasource.go b/pkg/exchangerates/national_bank_of_poland_datasource.go
@@ -6,6 +6,8 @@ import (
 	"math"
 	"time"
 
+	"golang.org/x/net/html/charset"
+
 	"github.com/mayswind/ezbookkeeping/pkg/core"
 	"github.com/mayswind/ezbookkeeping/pkg/errs"
 	"github.com/mayswind/ezbookkeeping/pkg/log"
@@ -128,7 +130,7 @@ func (e *NationalBankOfPolandDataSource) Parse(c core.Context, content []byte) (
 	nationalBankOfPolandData := &NationalBankOfPolandExchangeRateData{}
 
 	xmlDecoder := xml.NewDecoder(bytes.NewReader(content))
-	xmlDecoder.CharsetReader = utils.IdentReader
+	xmlDecoder.CharsetReader = charset.NewReaderLabel
 	err := xmlDecoder.Decode(&nationalBankOfPolandData)
 
 	if err != nil {

diff --git a/pkg/utils/io.go b/pkg/utils/io.go
@@ -125,8 +125,3 @@ func GetFileNameExtension(path string) string {
 
 	return extension[1:]
 }
-
-// IdentReader returns the original io reader
-func IdentReader(encoding string, input io.Reader) (io.Reader, error) {
-	return input, nil
-}