Detect encoding and BOM in content (#6727) (#6765)

Detect and remove a decoded BOM when showing content.
Restore the previous encoding and BOM when updating content.
On error keep as UTF-8 encoding.

Signed-off-by: Andrew Thornton <art27@cantab.net>
This commit is contained in:
zeripath 2019-04-28 00:36:12 +01:00 committed by techknowlogick
parent 8b3aad940e
commit 21fb791747
3 changed files with 127 additions and 7 deletions

View file

@ -5,6 +5,7 @@
package base package base
import ( import (
"bytes"
"crypto/md5" "crypto/md5"
"crypto/rand" "crypto/rand"
"crypto/sha1" "crypto/sha1"
@ -32,6 +33,9 @@ import (
"github.com/gogits/chardet" "github.com/gogits/chardet"
) )
// UTF8BOM is the utf-8 byte-order marker
var UTF8BOM = []byte{'\xef', '\xbb', '\xbf'}
// EncodeMD5 encodes string to md5 hex value. // EncodeMD5 encodes string to md5 hex value.
func EncodeMD5(str string) string { func EncodeMD5(str string) string {
m := md5.New() m := md5.New()
@ -87,6 +91,14 @@ func DetectEncoding(content []byte) (string, error) {
return result.Charset, err return result.Charset, err
} }
// RemoveBOMIfPresent removes a UTF-8 BOM from a []byte
func RemoveBOMIfPresent(content []byte) []byte {
if len(content) > 2 && bytes.Equal(content[0:3], UTF8BOM) {
return content[3:]
}
return content
}
// BasicAuthDecode decode basic auth string // BasicAuthDecode decode basic auth string
func BasicAuthDecode(encoded string) (string, string, error) { func BasicAuthDecode(encoded string) (string, string, error) {
s, err := base64.StdEncoding.DecodeString(encoded) s, err := base64.StdEncoding.DecodeString(encoded)

View file

@ -267,7 +267,7 @@ func ToUTF8WithErr(content []byte) (string, error) {
if err != nil { if err != nil {
return "", err return "", err
} else if charsetLabel == "UTF-8" { } else if charsetLabel == "UTF-8" {
return string(content), nil return string(base.RemoveBOMIfPresent(content)), nil
} }
encoding, _ := charset.Lookup(charsetLabel) encoding, _ := charset.Lookup(charsetLabel)
@ -277,19 +277,21 @@ func ToUTF8WithErr(content []byte) (string, error) {
// If there is an error, we concatenate the nicely decoded part and the // If there is an error, we concatenate the nicely decoded part and the
// original left over. This way we won't lose data. // original left over. This way we won't lose data.
result, n, err := transform.String(encoding.NewDecoder(), string(content)) result, n, err := transform.Bytes(encoding.NewDecoder(), content)
if err != nil { if err != nil {
result = result + string(content[n:]) result = append(result, content[n:]...)
} }
return result, err result = base.RemoveBOMIfPresent(result)
return string(result), err
} }
// ToUTF8WithFallback detects the encoding of content and coverts to UTF-8 if possible // ToUTF8WithFallback detects the encoding of content and coverts to UTF-8 if possible
func ToUTF8WithFallback(content []byte) []byte { func ToUTF8WithFallback(content []byte) []byte {
charsetLabel, err := base.DetectEncoding(content) charsetLabel, err := base.DetectEncoding(content)
if err != nil || charsetLabel == "UTF-8" { if err != nil || charsetLabel == "UTF-8" {
return content return base.RemoveBOMIfPresent(content)
} }
encoding, _ := charset.Lookup(charsetLabel) encoding, _ := charset.Lookup(charsetLabel)
@ -304,7 +306,7 @@ func ToUTF8WithFallback(content []byte) []byte {
return append(result, content[n:]...) return append(result, content[n:]...)
} }
return result return base.RemoveBOMIfPresent(result)
} }
// ToUTF8 converts content to UTF8 encoding and ignore error // ToUTF8 converts content to UTF8 encoding and ignore error

View file

@ -5,15 +5,85 @@
package uploader package uploader
import ( import (
"bytes"
"fmt" "fmt"
"strings" "strings"
"golang.org/x/net/html/charset"
"golang.org/x/text/transform"
"code.gitea.io/git" "code.gitea.io/git"
"code.gitea.io/gitea/models" "code.gitea.io/gitea/models"
"code.gitea.io/gitea/modules/base"
"code.gitea.io/gitea/modules/lfs" "code.gitea.io/gitea/modules/lfs"
"code.gitea.io/gitea/modules/log"
"code.gitea.io/gitea/modules/setting" "code.gitea.io/gitea/modules/setting"
) )
func detectEncodingAndBOM(entry *git.TreeEntry, repo *models.Repository) (string, bool) {
reader, err := entry.Blob().DataAsync()
if err != nil {
// return default
return "UTF-8", false
}
defer reader.Close()
buf := make([]byte, 1024)
n, err := reader.Read(buf)
if err != nil {
// return default
return "UTF-8", false
}
buf = buf[:n]
if setting.LFS.StartServer {
meta := lfs.IsPointerFile(&buf)
if meta != nil {
meta, err = repo.GetLFSMetaObjectByOid(meta.Oid)
if err != nil && err != models.ErrLFSObjectNotExist {
// return default
return "UTF-8", false
}
}
if meta != nil {
dataRc, err := lfs.ReadMetaObject(meta)
if err != nil {
// return default
return "UTF-8", false
}
defer dataRc.Close()
buf = make([]byte, 1024)
n, err = dataRc.Read(buf)
if err != nil {
// return default
return "UTF-8", false
}
buf = buf[:n]
}
}
encoding, err := base.DetectEncoding(buf)
if err != nil {
// just default to utf-8 and no bom
return "UTF-8", false
}
if encoding == "UTF-8" {
return encoding, bytes.Equal(buf[0:3], base.UTF8BOM)
}
charsetEncoding, _ := charset.Lookup(encoding)
if charsetEncoding == nil {
return "UTF-8", false
}
result, n, err := transform.String(charsetEncoding.NewDecoder(), string(buf))
if n > 2 {
return encoding, bytes.Equal([]byte(result)[0:3], base.UTF8BOM)
}
return encoding, false
}
// UpdateRepoFileOptions holds the repository file update options // UpdateRepoFileOptions holds the repository file update options
type UpdateRepoFileOptions struct { type UpdateRepoFileOptions struct {
LastCommitID string LastCommitID string
@ -45,12 +115,29 @@ func UpdateRepoFile(repo *models.Repository, doer *models.User, opts *UpdateRepo
return fmt.Errorf("UpdateRepoFile: %v", err) return fmt.Errorf("UpdateRepoFile: %v", err)
} }
encoding := "UTF-8"
bom := false
if opts.IsNewFile { if opts.IsNewFile {
for _, file := range filesInIndex { for _, file := range filesInIndex {
if file == opts.NewTreeName { if file == opts.NewTreeName {
return models.ErrRepoFileAlreadyExist{FileName: opts.NewTreeName} return models.ErrRepoFileAlreadyExist{FileName: opts.NewTreeName}
} }
} }
} else {
gitRepo, err := git.OpenRepository(t.basePath)
if err != nil {
return err
}
tree, err := gitRepo.GetTree("HEAD")
if err != nil {
return err
}
entry, err := tree.GetTreeEntryByPath(opts.OldTreeName)
if err != nil {
return err
}
encoding, bom = detectEncodingAndBOM(entry, repo)
} }
//var stdout string //var stdout string
@ -72,9 +159,28 @@ func UpdateRepoFile(repo *models.Repository, doer *models.User, opts *UpdateRepo
} }
content := opts.Content content := opts.Content
if bom {
content = string(base.UTF8BOM) + content
}
if encoding != "UTF-8" {
charsetEncoding, _ := charset.Lookup(encoding)
if charsetEncoding != nil {
result, _, err := transform.String(charsetEncoding.NewEncoder(), string(content))
if err != nil {
// Look if we can't encode back in to the original we should just stick with utf-8
log.Error(4, "Error re-encoding %s (%s) as %s - will stay as UTF-8: %v", opts.NewTreeName, opts.OldTreeName, encoding, err)
result = content
}
content = result
} else {
log.Error(4, "Unknown encoding: %s", encoding)
}
}
// Reset the opts.Content with the re-encoded and BOM'd content
opts.Content = content
var lfsMetaObject *models.LFSMetaObject var lfsMetaObject *models.LFSMetaObject
if filename2attribute2info[opts.NewTreeName] != nil && filename2attribute2info[opts.NewTreeName]["filter"] == "lfs" { if setting.LFS.StartServer && filename2attribute2info[opts.NewTreeName] != nil && filename2attribute2info[opts.NewTreeName]["filter"] == "lfs" {
// OK so we are supposed to LFS this data! // OK so we are supposed to LFS this data!
oid, err := models.GenerateLFSOid(strings.NewReader(opts.Content)) oid, err := models.GenerateLFSOid(strings.NewReader(opts.Content))
if err != nil { if err != nil {