delta_lib/utils/
regex_replacement.rs

1use std::borrow::Cow;
2
3use regex::{Regex, RegexBuilder};
4
5#[derive(Clone, Debug)]
6pub struct RegexReplacement {
7    regex: Regex,
8    replacement: String,
9    replace_all: bool,
10}
11
12impl RegexReplacement {
13    pub fn from_sed_command(sed_command: &str) -> Option<Self> {
14        let sep = sed_command.chars().nth(1)?;
15        let mut parts = sed_command[2..].split(sep);
16        let regex = parts.next()?;
17        let replacement = parts.next()?.to_string();
18        let flags = parts.next()?;
19        let mut re_builder = RegexBuilder::new(regex);
20        let mut replace_all = false;
21        for flag in flags.chars() {
22            match flag {
23                'g' => {
24                    replace_all = true;
25                }
26                'i' => {
27                    re_builder.case_insensitive(true);
28                }
29                'm' => {
30                    re_builder.multi_line(true);
31                }
32                's' => {
33                    re_builder.dot_matches_new_line(true);
34                }
35                'U' => {
36                    re_builder.swap_greed(true);
37                }
38                'x' => {
39                    re_builder.ignore_whitespace(true);
40                }
41                _ => {}
42            }
43        }
44        let regex = re_builder.build().ok()?;
45        Some(RegexReplacement {
46            regex,
47            replacement,
48            replace_all,
49        })
50    }
51
52    pub fn execute<'t>(&self, s: &'t str) -> Cow<'t, str> {
53        if self.replace_all {
54            self.regex.replace_all(s, &self.replacement)
55        } else {
56            self.regex.replace(s, &self.replacement)
57        }
58    }
59}
60
61#[cfg(test)]
62mod tests {
63    use super::*;
64
65    #[test]
66    fn test_sed_command() {
67        let command = "s,foo,bar,";
68        let rr = RegexReplacement::from_sed_command(command).unwrap();
69        assert_eq!(rr.regex.as_str(), "foo");
70        assert_eq!(rr.replacement, "bar");
71        assert_eq!(rr.replace_all, false);
72        assert_eq!(rr.execute("foo"), "bar");
73    }
74
75    #[test]
76    fn test_sed_command_i_flag() {
77        let command = "s,FOO,bar,";
78        let rr = RegexReplacement::from_sed_command(command).unwrap();
79        assert_eq!(rr.execute("foo"), "foo");
80        let command = "s,FOO,bar,i";
81        let rr = RegexReplacement::from_sed_command(command).unwrap();
82        assert_eq!(rr.execute("foo"), "bar");
83    }
84
85    #[test]
86    fn test_sed_command_g_flag() {
87        let command = "s,foo,bar,";
88        let rr = RegexReplacement::from_sed_command(command).unwrap();
89        assert_eq!(rr.execute("foofoo"), "barfoo");
90        let command = "s,foo,bar,g";
91        let rr = RegexReplacement::from_sed_command(command).unwrap();
92        assert_eq!(rr.execute("foofoo"), "barbar");
93    }
94
95    #[test]
96    fn test_sed_command_with_named_captures() {
97        let command = r"s/(?P<last>[^,\s]+),\s+(?P<first>\S+)/$first $last/";
98        let rr = RegexReplacement::from_sed_command(command).unwrap();
99        assert_eq!(rr.execute("Springsteen, Bruce"), "Bruce Springsteen");
100    }
101
102    #[test]
103    fn test_sed_command_invalid() {
104        assert!(RegexReplacement::from_sed_command("").is_none());
105        assert!(RegexReplacement::from_sed_command("s").is_none());
106        assert!(RegexReplacement::from_sed_command("s,").is_none());
107        assert!(RegexReplacement::from_sed_command("s,,").is_none());
108        assert!(RegexReplacement::from_sed_command("s,,i").is_none());
109        assert!(RegexReplacement::from_sed_command("s,,,").is_some());
110        assert!(RegexReplacement::from_sed_command("s,,,i").is_some());
111    }
112}