@@ -3,7 +3,6 @@ package gchatmeow
33import (
44 "context"
55 "encoding/base64"
6- "encoding/binary"
76 "encoding/json"
87 "errors"
98 "fmt"
@@ -17,7 +16,7 @@ import (
1716 "strconv"
1817 "strings"
1918 "time"
20- "unicode/utf8 "
19+ "unicode/utf16 "
2120
2221 "go.mau.fi/util/pblite"
2322
@@ -62,6 +61,16 @@ type Channel struct {
6261 OnReceiveArray * Event
6362}
6463
64+ type UTF16String []uint16
65+
66+ func NewUTF16String (s string ) UTF16String {
67+ return utf16 .Encode ([]rune (s ))
68+ }
69+
70+ func (u UTF16String ) String () string {
71+ return string (utf16 .Decode (u ))
72+ }
73+
6574type ChunkParser struct {
6675 buf []byte
6776}
@@ -72,91 +81,24 @@ func NewChunkParser() *ChunkParser {
7281 }
7382}
7483
75- // bestEffortDecode attempts to decode as much UTF-8 data as possible from the buffer
76- func bestEffortDecode (data []byte ) string {
77- valid := make ([]byte , 0 , len (data ))
78- for len (data ) > 0 {
79- r , size := utf8 .DecodeRune (data )
80- if r == utf8 .RuneError {
81- break
82- }
83- valid = append (valid , data [:size ]... )
84- data = data [size :]
85- }
86- return string (valid )
87- }
88-
89- // GetChunks yields chunks generated from received data.
90- // The buffer may not be decodable as UTF-8 if there's a split multi-byte
91- // character at the end. To handle this, we do a "best effort" decode of the
92- // buffer to decode as much of it as possible.
9384func (p * ChunkParser ) GetChunks (newDataBytes []byte ) []string {
9485 var chunks []string
9586 p .buf = append (p .buf , newDataBytes ... )
9687
9788 for {
98- // Decode buffer with best effort
99- bufDecoded := bestEffortDecode (p .buf )
100-
101- // Convert to UTF-16 (removing BOM)
102- var bufUtf16 []byte
103- for _ , r := range bufDecoded {
104- // Convert each rune to UTF-16
105- buf := make ([]byte , 2 )
106- binary .BigEndian .PutUint16 (buf , uint16 (r ))
107- bufUtf16 = append (bufUtf16 , buf ... )
108- }
109-
110- // Find length string match
111- matches := lenRegex .FindStringSubmatch (bufDecoded )
112- if matches == nil {
113- break
114- }
115-
116- lengthStr := matches [1 ]
117- // Both lengths are in number of bytes in UTF-16 encoding
89+ bufStr := string (p .buf )
90+ lengthStr , after , _ := strings .Cut (bufStr , "\n " )
11891 length , err := strconv .Atoi (lengthStr )
11992 if err != nil {
12093 break
12194 }
122- length *= 2 // Convert to UTF-16 byte count
123-
124- // Calculate length of the submission length and newline in UTF-16
125- lenStrAndNewline := lengthStr + "\n "
126- var lenLength int
127- for _ , r := range lenStrAndNewline {
128- lenLength += 2 // Each UTF-16 character is 2 bytes
129- _ = r
130- }
131-
132- if len (bufUtf16 )- lenLength < length {
95+ utf16Str := NewUTF16String (after )
96+ if len (utf16Str ) < length {
13397 break
13498 }
13599
136- // Extract submission
137- submission := bufUtf16 [lenLength : lenLength + length ]
138-
139- // Convert UTF-16 bytes back to string
140- var result string
141- for i := 0 ; i < len (submission ); i += 2 {
142- if i + 1 >= len (submission ) {
143- break
144- }
145- char := binary .BigEndian .Uint16 (submission [i : i + 2 ])
146- result += string (rune (char ))
147- }
148-
149- chunks = append (chunks , result )
150-
151- // Calculate how many bytes to drop from the buffer
152- dropLength := len (matches [0 ]) // length of the length string and newline
153- dropLength += len (result ) // length of the actual content in UTF-8
154-
155- if dropLength <= len (p .buf ) {
156- p .buf = p .buf [dropLength :]
157- } else {
158- p .buf = p .buf [:0 ]
159- }
100+ chunks = append (chunks , utf16Str [0 :length ].String ())
101+ p .buf = []byte (utf16Str [length :].String ())
160102 }
161103
162104 return chunks
0 commit comments