git_object/commit/message/
body.rs

1use std::ops::Deref;
2
3use nom::{
4    bytes::complete::{tag, take_until1},
5    combinator::all_consuming,
6    error::{ErrorKind, ParseError},
7    sequence::terminated,
8    IResult,
9};
10
11use crate::{
12    bstr::{BStr, ByteSlice},
13    commit::message::BodyRef,
14};
15
16/// An iterator over trailers as parsed from a commit message body.
17///
18/// lines with parsing failures will be skipped
19pub struct Trailers<'a> {
20    pub(crate) cursor: &'a [u8],
21}
22
23/// A trailer as parsed from the commit message body.
24#[derive(PartialEq, Eq, Debug, Hash, Ord, PartialOrd, Clone, Copy)]
25#[cfg_attr(feature = "serde1", derive(serde::Serialize, serde::Deserialize))]
26pub struct TrailerRef<'a> {
27    /// The name of the trailer, like "Signed-off-by", up to the separator ": "
28    #[cfg_attr(feature = "serde1", serde(borrow))]
29    pub token: &'a BStr,
30    /// The value right after the separator ": ", with leading and trailing whitespace trimmed.
31    /// Note that multi-line values aren't currently supported.
32    pub value: &'a BStr,
33}
34
35fn parse_single_line_trailer<'a, E: ParseError<&'a [u8]>>(i: &'a [u8]) -> IResult<&'a [u8], (&'a BStr, &'a BStr), E> {
36    let (value, token) = terminated(take_until1(b":".as_ref()), tag(b": "))(i.trim_end())?;
37    if token.trim_end().len() != token.len() || value.trim_start().len() != value.len() {
38        Err(nom::Err::Failure(E::from_error_kind(i, ErrorKind::Fail)))
39    } else {
40        Ok((&[], (token.as_bstr(), value.as_bstr())))
41    }
42}
43
44impl<'a> Iterator for Trailers<'a> {
45    type Item = TrailerRef<'a>;
46
47    fn next(&mut self) -> Option<Self::Item> {
48        if self.cursor.is_empty() {
49            return None;
50        }
51        for line in self.cursor.lines_with_terminator() {
52            self.cursor = &self.cursor[line.len()..];
53            if let Some(trailer) =
54                all_consuming(parse_single_line_trailer::<()>)(line)
55                    .ok()
56                    .map(|(_, (token, value))| TrailerRef {
57                        token: token.trim().as_bstr(),
58                        value: value.trim().as_bstr(),
59                    })
60            {
61                return Some(trailer);
62            }
63        }
64        None
65    }
66}
67
68impl<'a> BodyRef<'a> {
69    /// Parse `body` bytes into the trailer and the actual body.
70    pub fn from_bytes(body: &'a [u8]) -> Self {
71        body.rfind(b"\n\n")
72            .map(|pos| (2, pos))
73            .or_else(|| body.rfind(b"\r\n\r\n").map(|pos| (4, pos)))
74            .and_then(|(sep_len, pos)| {
75                let trailer = &body[pos + sep_len..];
76                let body = &body[..pos];
77                Trailers { cursor: trailer }.next().map(|_| BodyRef {
78                    body_without_trailer: body.as_bstr(),
79                    start_of_trailer: trailer,
80                })
81            })
82            .unwrap_or_else(|| BodyRef {
83                body_without_trailer: body.as_bstr(),
84                start_of_trailer: &[],
85            })
86    }
87
88    /// Returns the body with the trailers stripped.
89    ///
90    /// You can iterate trailers with the [`trailers()`][BodyRef::trailers()] method.
91    pub fn without_trailer(&self) -> &'a BStr {
92        self.body_without_trailer
93    }
94
95    /// Return an iterator over the trailers parsed from the last paragraph of the body. May be empty.
96    pub fn trailers(&self) -> Trailers<'a> {
97        Trailers {
98            cursor: self.start_of_trailer,
99        }
100    }
101}
102
103impl<'a> AsRef<BStr> for BodyRef<'a> {
104    fn as_ref(&self) -> &BStr {
105        self.body_without_trailer
106    }
107}
108
109impl<'a> Deref for BodyRef<'a> {
110    type Target = BStr;
111
112    fn deref(&self) -> &Self::Target {
113        self.body_without_trailer
114    }
115}
116#[cfg(test)]
117mod test_parse_trailer {
118    use super::*;
119
120    fn parse(input: &str) -> (&BStr, &BStr) {
121        parse_single_line_trailer::<()>(input.as_bytes()).unwrap().1
122    }
123
124    #[test]
125    fn simple_newline() {
126        assert_eq!(parse("foo: bar\n"), ("foo".into(), "bar".into()));
127    }
128
129    #[test]
130    fn simple_non_ascii_no_newline() {
131        assert_eq!(parse("🤗: 🎉"), ("🤗".into(), "🎉".into()));
132    }
133
134    #[test]
135    fn with_lots_of_whitespace_newline() {
136        assert_eq!(
137            parse("hello foo: bar there   \n"),
138            ("hello foo".into(), "bar there".into())
139        );
140    }
141
142    #[test]
143    fn extra_whitespace_before_token_or_value_is_error() {
144        assert!(parse_single_line_trailer::<()>(b"foo : bar").is_err());
145        assert!(parse_single_line_trailer::<()>(b"foo:  bar").is_err())
146    }
147
148    #[test]
149    fn simple_newline_windows() {
150        assert_eq!(parse("foo: bar\r\n"), ("foo".into(), "bar".into()));
151    }
152}
git_object/commit/message/body.rs

git_object/commit/message/
body.rs
