ucd_parse/
emoji_properties.rs

1use std::path::{Path, PathBuf};
2
3use crate::{
4    common::{
5        parse_codepoint_association, CodepointIter, Codepoints, UcdFile,
6        UcdFileByCodepoint,
7    },
8    error::Error,
9};
10
11/// A single row in the `emoji-data.txt` file.
12///
13/// The `emoji-data.txt` file is the source of truth on several Emoji-related
14/// Unicode properties.
15///
16/// Note that `emoji-data.txt` is not formally part of the Unicode Character
17/// Database. You can download the Emoji data files separately here:
18/// https://unicode.org/Public/emoji/
19#[derive(Clone, Debug, Default, Eq, PartialEq)]
20pub struct EmojiProperty {
21    /// The codepoint or codepoint range for this entry.
22    pub codepoints: Codepoints,
23    /// The property name assigned to the codepoints in this entry.
24    pub property: String,
25}
26
27impl UcdFile for EmojiProperty {
28    fn relative_file_path() -> &'static Path {
29        Path::new("emoji/emoji-data.txt")
30    }
31
32    fn file_path<P: AsRef<Path>>(ucd_dir: P) -> PathBuf {
33        let ucd_dir = ucd_dir.as_ref();
34        // The standard location, but only on UCDs from 13.0.0 and up.
35        let std = ucd_dir.join(Self::relative_file_path());
36        if std.exists() {
37            std
38        } else {
39            // If the old location does exist, use it.
40            let legacy = ucd_dir.join("emoji-data.txt");
41            if legacy.exists() {
42                legacy
43            } else {
44                // This might end up in an error message, so use the standard
45                // one if forced to choose. Arguably we could do something like
46                // peek
47                std
48            }
49        }
50    }
51}
52
53impl UcdFileByCodepoint for EmojiProperty {
54    fn codepoints(&self) -> CodepointIter {
55        self.codepoints.into_iter()
56    }
57}
58
59impl std::str::FromStr for EmojiProperty {
60    type Err = Error;
61
62    fn from_str(line: &str) -> Result<EmojiProperty, Error> {
63        let (codepoints, property) = parse_codepoint_association(line)?;
64        Ok(EmojiProperty { codepoints, property: property.to_string() })
65    }
66}
67
68#[cfg(test)]
69mod tests {
70    use super::EmojiProperty;
71
72    #[test]
73    fn parse_single() {
74        let line = "24C2          ; Emoji                #  1.1  [1] (Ⓜ️)       circled M\n";
75        let row: EmojiProperty = line.parse().unwrap();
76        assert_eq!(row.codepoints, 0x24C2);
77        assert_eq!(row.property, "Emoji");
78    }
79
80    #[test]
81    fn parse_range() {
82        let line = "1FA6E..1FFFD  ; Extended_Pictographic#   NA[1424] (🩮️..🿽️)   <reserved-1FA6E>..<reserved-1FFFD>\n";
83        let row: EmojiProperty = line.parse().unwrap();
84        assert_eq!(row.codepoints, (0x1FA6E, 0x1FFFD));
85        assert_eq!(row.property, "Extended_Pictographic");
86    }
87}
ucd_parse/emoji_properties.rs

ucd_parse/
emoji_properties.rs
