xml/
util.rs

Help
1use std::fmt;
2use std::io::{self, Read};
3use std::str::{self, FromStr};
4
5#[derive(Debug)]
6pub enum CharReadError {
7    UnexpectedEof,
8    Utf8(str::Utf8Error),
9    Io(io::Error),
10}
11
12impl From<str::Utf8Error> for CharReadError {
13    #[cold]
14    fn from(e: str::Utf8Error) -> Self {
15        Self::Utf8(e)
16    }
17}
18
19impl From<io::Error> for CharReadError {
20    #[cold]
21    fn from(e: io::Error) -> Self {
22        Self::Io(e)
23    }
24}
25
26impl fmt::Display for CharReadError {
27    #[cold]
28    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
29        use self::CharReadError::{Io, UnexpectedEof, Utf8};
30        match *self {
31            UnexpectedEof => write!(f, "unexpected end of stream"),
32            Utf8(ref e) => write!(f, "UTF-8 decoding error: {e}"),
33            Io(ref e) => write!(f, "I/O error: {e}"),
34        }
35    }
36}
37
38/// Character encoding used for parsing
39#[derive(Debug, Copy, Clone, Eq, PartialEq)]
40#[non_exhaustive]
41pub enum Encoding {
42    /// Explicitly UTF-8 only
43    Utf8,
44    /// UTF-8 fallback, but can be any 8-bit encoding
45    Default,
46    /// ISO-8859-1
47    Latin1,
48    /// US-ASCII
49    Ascii,
50    /// Big-Endian
51    Utf16Be,
52    /// Little-Endian
53    Utf16Le,
54    /// Unknown endianness yet, will be sniffed
55    Utf16,
56    /// Not determined yet, may be sniffed to be anything
57    Unknown,
58}
59
60// Rustc inlines eq_ignore_ascii_case and creates kilobytes of code!
61#[inline(never)]
62fn icmp(lower: &str, varcase: &str) -> bool {
63    lower.bytes().zip(varcase.bytes()).all(|(l, v)| l == v.to_ascii_lowercase())
64}
65
66impl FromStr for Encoding {
67    type Err = &'static str;
68
69    fn from_str(val: &str) -> Result<Self, Self::Err> {
70        if ["utf-8", "utf8"].into_iter().any(move |label| icmp(label, val)) {
71            Ok(Self::Utf8)
72        } else if ["iso-8859-1", "latin1"].into_iter().any(move |label| icmp(label, val)) {
73            Ok(Self::Latin1)
74        } else if ["utf-16", "utf16"].into_iter().any(move |label| icmp(label, val)) {
75            Ok(Self::Utf16)
76        } else if ["ascii", "us-ascii"].into_iter().any(move |label| icmp(label, val)) {
77            Ok(Self::Ascii)
78        } else {
79            Err("unknown encoding name")
80        }
81    }
82}
83
84impl fmt::Display for Encoding {
85    #[cold]
86    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
87        f.write_str(match self {
88            Self::Utf8 |
89            Self::Default => "UTF-8",
90            Self::Latin1 => "ISO-8859-1",
91            Self::Ascii => "US-ASCII",
92            Self::Utf16Be |
93            Self::Utf16Le |
94            Self::Utf16 => "UTF-16",
95            Self::Unknown => "(unknown)",
96        })
97    }
98}
99
100pub(crate) struct CharReader {
101    pub encoding: Encoding,
102}
103
104impl CharReader {
105    pub const fn new() -> Self {
106        Self { encoding: Encoding::Unknown }
107    }
108
109    pub fn next_char_from<R: Read>(&mut self, source: &mut R) -> Result<Option<char>, CharReadError> {
110        let mut bytes = source.bytes();
111        const MAX_CODEPOINT_LEN: usize = 4;
112
113        let mut buf = [0u8; MAX_CODEPOINT_LEN];
114        let mut pos = 0;
115        loop {
116            let next = match bytes.next() {
117                Some(Ok(b)) => b,
118                Some(Err(e)) => return Err(e.into()),
119                None if pos == 0 => return Ok(None),
120                None => return Err(CharReadError::UnexpectedEof),
121            };
122
123            match self.encoding {
124                Encoding::Utf8 | Encoding::Default => {
125                    // fast path for ASCII subset
126                    if pos == 0 && next.is_ascii() {
127                        return Ok(Some(next.into()));
128                    }
129
130                    buf[pos] = next;
131                    pos += 1;
132
133                    match str::from_utf8(&buf[..pos]) {
134                        Ok(s) => return Ok(s.chars().next()), // always Some(..)
135                        Err(_) if pos < MAX_CODEPOINT_LEN => continue,
136                        Err(e) => return Err(e.into()),
137                    }
138                },
139                Encoding::Latin1 => {
140                    return Ok(Some(next.into()));
141                },
142                Encoding::Ascii => {
143                    return if next.is_ascii() {
144                        Ok(Some(next.into()))
145                    } else {
146                        Err(CharReadError::Io(io::Error::new(io::ErrorKind::InvalidData, "char is not ASCII")))
147                    };
148                },
149                Encoding::Unknown | Encoding::Utf16 => {
150                    buf[pos] = next;
151                    pos += 1;
152                    if let Some(value) = self.sniff_bom(&buf[..pos], &mut pos) {
153                        return value;
154                    }
155                },
156                Encoding::Utf16Be => {
157                    buf[pos] = next;
158                    pos += 1;
159                    if pos == 2 {
160                        if let Some(Ok(c)) = char::decode_utf16([u16::from_be_bytes(buf[..2].try_into().unwrap())]).next() {
161                            return Ok(Some(c));
162                        }
163                    } else if pos == 4 { // surrogate
164                        return char::decode_utf16([u16::from_be_bytes(buf[..2].try_into().unwrap()), u16::from_be_bytes(buf[2..4].try_into().unwrap())])
165                            .next().transpose()
166                            .map_err(|e| CharReadError::Io(io::Error::new(io::ErrorKind::InvalidData, e)));
167                    }
168                },
169                Encoding::Utf16Le => {
170                    buf[pos] = next;
171                    pos += 1;
172                    if pos == 2 {
173                        if let Some(Ok(c)) = char::decode_utf16([u16::from_le_bytes(buf[..2].try_into().unwrap())]).next() {
174                            return Ok(Some(c));
175                        }
176                    } else if pos == 4 { // surrogate
177                        return char::decode_utf16([u16::from_le_bytes(buf[..2].try_into().unwrap()), u16::from_le_bytes(buf[2..4].try_into().unwrap())])
178                            .next().transpose()
179                            .map_err(|e| CharReadError::Io(io::Error::new(io::ErrorKind::InvalidData, e)));
180                    }
181                },
182            }
183        }
184    }
185
186    #[cold]
187    fn sniff_bom(&mut self, buf: &[u8], pos: &mut usize) -> Option<Result<Option<char>, CharReadError>> {
188        // sniff BOM
189        if buf.len() <= 3 && [0xEF, 0xBB, 0xBF].starts_with(buf) {
190            if buf.len() == 3 && self.encoding != Encoding::Utf16 {
191                *pos = 0;
192                self.encoding = Encoding::Utf8;
193            }
194        } else if buf.len() <= 2 && [0xFE, 0xFF].starts_with(buf) {
195            if buf.len() == 2 {
196                *pos = 0;
197                self.encoding = Encoding::Utf16Be;
198            }
199        } else if buf.len() <= 2 && [0xFF, 0xFE].starts_with(buf) {
200            if buf.len() == 2 {
201                *pos = 0;
202                self.encoding = Encoding::Utf16Le;
203            }
204        } else if buf.len() == 1 && self.encoding == Encoding::Utf16 {
205            // sniff ASCII char in UTF-16
206            self.encoding = if buf[0] == 0 { Encoding::Utf16Be } else { Encoding::Utf16Le };
207        } else {
208            // UTF-8 is the default, but XML decl can change it to other 8-bit encoding
209            self.encoding = Encoding::Default;
210            if buf.len() == 1 && buf[0].is_ascii() {
211                return Some(Ok(Some(buf[0].into())));
212            }
213        }
214        None
215    }
216}
217
218#[cfg(test)]
219mod tests {
220    use super::{CharReadError, CharReader, Encoding};
221
222    #[test]
223    fn test_next_char_from() {
224        use std::io;
225
226        let mut bytes: &[u8] = b"correct";    // correct ASCII
227        assert_eq!(CharReader::new().next_char_from(&mut bytes).unwrap(), Some('c'));
228
229        let mut bytes: &[u8] = b"\xEF\xBB\xBF\xE2\x80\xA2!";  // BOM
230        assert_eq!(CharReader::new().next_char_from(&mut bytes).unwrap(), Some('•'));
231
232        let mut bytes: &[u8] = b"\xEF\xBB\xBFx123";  // BOM
233        assert_eq!(CharReader::new().next_char_from(&mut bytes).unwrap(), Some('x'));
234
235        let mut bytes: &[u8] = b"\xEF\xBB\xBF";  // Nothing after BOM
236        assert_eq!(CharReader::new().next_char_from(&mut bytes).unwrap(), None);
237
238        let mut bytes: &[u8] = b"\xEF\xBB";  // Nothing after BO
239        assert!(matches!(CharReader::new().next_char_from(&mut bytes), Err(CharReadError::UnexpectedEof)));
240
241        let mut bytes: &[u8] = b"\xEF\xBB\x42";  // Nothing after BO
242        assert!(CharReader::new().next_char_from(&mut bytes).is_err());
243
244        let mut bytes: &[u8] = b"\xFE\xFF\x00\x42";  // UTF-16
245        assert_eq!(CharReader::new().next_char_from(&mut bytes).unwrap(), Some('B'));
246
247        let mut bytes: &[u8] = b"\xFF\xFE\x42\x00";  // UTF-16
248        assert_eq!(CharReader::new().next_char_from(&mut bytes).unwrap(), Some('B'));
249
250        let mut bytes: &[u8] = b"\xFF\xFE";  // UTF-16
251        assert_eq!(CharReader::new().next_char_from(&mut bytes).unwrap(), None);
252
253        let mut bytes: &[u8] = b"\xFF\xFE\x00";  // UTF-16
254        assert!(matches!(CharReader::new().next_char_from(&mut bytes), Err(CharReadError::UnexpectedEof)));
255
256        let mut bytes: &[u8] = "правильно".as_bytes();  // correct BMP
257        assert_eq!(CharReader::new().next_char_from(&mut bytes).unwrap(), Some('п'));
258
259        let mut bytes: &[u8] = "правильно".as_bytes();
260        assert_eq!(CharReader { encoding: Encoding::Utf16Be }.next_char_from(&mut bytes).unwrap(), Some('킿'));
261
262        let mut bytes: &[u8] = "правильно".as_bytes();
263        assert_eq!(CharReader { encoding: Encoding::Utf16Le }.next_char_from(&mut bytes).unwrap(), Some('뿐'));
264
265        let mut bytes: &[u8] = b"\xD8\xD8\x80";
266        assert!(CharReader { encoding: Encoding::Utf16 }.next_char_from(&mut bytes).is_err());
267
268        let mut bytes: &[u8] = b"\x00\x42";
269        assert_eq!(CharReader { encoding: Encoding::Utf16 }.next_char_from(&mut bytes).unwrap(), Some('B'));
270
271        let mut bytes: &[u8] = b"\x42\x00";
272        assert_eq!(CharReader { encoding: Encoding::Utf16 }.next_char_from(&mut bytes).unwrap(), Some('B'));
273
274        let mut bytes: &[u8] = b"\x00";
275        assert!(CharReader { encoding: Encoding::Utf16Be }.next_char_from(&mut bytes).is_err());
276
277        let mut bytes: &[u8] = "😊".as_bytes();          // correct non-BMP
278        assert_eq!(CharReader::new().next_char_from(&mut bytes).unwrap(), Some('😊'));
279
280        let mut bytes: &[u8] = b"";                     // empty
281        assert_eq!(CharReader::new().next_char_from(&mut bytes).unwrap(), None);
282
283        let mut bytes: &[u8] = b"\xf0\x9f\x98";         // incomplete code point
284        match CharReader::new().next_char_from(&mut bytes).unwrap_err() {
285            super::CharReadError::UnexpectedEof => {},
286            e => panic!("Unexpected result: {e:?}")
287        };
288
289        let mut bytes: &[u8] = b"\xff\x9f\x98\x32";     // invalid code point
290        match CharReader::new().next_char_from(&mut bytes).unwrap_err() {
291            super::CharReadError::Utf8(_) => {},
292            e => panic!("Unexpected result: {e:?}")
293        };
294
295        // error during read
296        struct ErrorReader;
297        impl io::Read for ErrorReader {
298            fn read(&mut self, _: &mut [u8]) -> io::Result<usize> {
299                Err(io::Error::new(io::ErrorKind::Other, "test error"))
300            }
301        }
302
303        let mut r = ErrorReader;
304        match CharReader::new().next_char_from(&mut r).unwrap_err() {
305            super::CharReadError::Io(ref e) if e.kind() == io::ErrorKind::Other &&
306                                               e.to_string().contains("test error") => {},
307            e => panic!("Unexpected result: {e:?}")
308        }
309    }
310}
xml/util.rs

xml/
util.rs
