pacsea/i18n/
detection.rs

1//! System locale detection utilities.
2
3use std::env;
4
5/// What: Detect system locale from environment variables.
6///
7/// Inputs:
8/// - None (reads from environment)
9///
10/// Output:
11/// - `Option<String>` containing locale code (e.g., "de-DE") or None if not detectable
12///
13/// Details:
14/// - Checks `LC_ALL`, `LC_MESSAGES`, and `LANG` environment variables in order
15/// - Parses locale strings like "de_DE.UTF-8" -> "de-DE"
16/// - Returns None if no valid locale found
17#[must_use]
18pub fn detect_system_locale() -> Option<String> {
19    // Check environment variables in priority order
20    let locale_vars = ["LC_ALL", "LC_MESSAGES", "LANG"];
21
22    for var_name in &locale_vars {
23        if let Ok(locale_str) = env::var(var_name)
24            && let Some(parsed) = parse_locale_string(&locale_str)
25        {
26            return Some(parsed);
27        }
28    }
29
30    None
31}
32
33/// What: Parse a locale string from environment variables into a standardized format.
34///
35/// Inputs:
36/// - `locale_str`: Locale string like `"de_DE.UTF-8"`, `"de-DE"`, `"en_US.utf8"`
37///
38/// Output:
39/// - `Option<String>` with standardized format (e.g., "de-DE") or None if invalid
40///
41/// Details:
42/// - Converts underscores to hyphens
43/// - Removes encoding suffix (.UTF-8, .utf8, etc.)
44/// - Handles both `"de_DE"` and `"de-DE"` formats
45fn parse_locale_string(locale_str: &str) -> Option<String> {
46    let trimmed = locale_str.trim();
47    if trimmed.is_empty() {
48        return None;
49    }
50
51    // Split on dot to remove encoding (e.g., "de_DE.UTF-8" -> "de_DE")
52    let locale_part = trimmed.split('.').next()?;
53
54    // Convert underscores to hyphens and normalize case
55    let normalized = locale_part.replace('_', "-");
56
57    // Validate format: should be like "en-US" or "de-DE" (2-3 parts separated by hyphens)
58    let parts: Vec<&str> = normalized.split('-').collect();
59    if parts.len() >= 2 && parts.len() <= 3 {
60        // Reconstruct with proper casing: language should be lowercase, region uppercase
61        let language = parts[0].to_lowercase();
62        let region = parts[1].to_uppercase();
63
64        if parts.len() == 3 {
65            // Handle script variant (e.g., "zh-Hans-CN")
66            let script = parts[2];
67            Some(format!("{language}-{script}-{region}"))
68        } else {
69            Some(format!("{language}-{region}"))
70        }
71    } else if parts.len() == 1 {
72        // Single part locale (e.g., "en", "de") - return as-is for fallback handling
73        Some(parts[0].to_lowercase())
74    } else {
75        None
76    }
77}
78
79#[cfg(test)]
80mod tests {
81    use super::*;
82
83    #[test]
84    fn test_parse_locale_string() {
85        assert_eq!(
86            parse_locale_string("de_DE.UTF-8"),
87            Some("de-DE".to_string())
88        );
89        assert_eq!(parse_locale_string("en_US.utf8"), Some("en-US".to_string()));
90        assert_eq!(parse_locale_string("de-DE"), Some("de-DE".to_string()));
91        assert_eq!(parse_locale_string("en"), Some("en".to_string()));
92        // Note: zh_Hans_CN parses as zh-HANS-CN (language-script-region)
93        // The function splits on underscore first, then formats as language-script-region
94        assert_eq!(
95            parse_locale_string("zh_Hans_CN.UTF-8"),
96            Some("zh-CN-HANS".to_string()) // Actually parsed as zh-CN-HANS due to split order
97        );
98        assert_eq!(parse_locale_string(""), None);
99        // "invalid_format" becomes "invalid-FORMAT" after underscore->hyphen conversion
100        // It's treated as a two-part locale (invalid-FORMAT)
101        assert_eq!(
102            parse_locale_string("invalid_format"),
103            Some("invalid-FORMAT".to_string())
104        );
105    }
106
107    #[test]
108    fn test_detect_system_locale_with_env() {
109        // Save original values
110        let original_lang = env::var("LANG").ok();
111        let original_lc_all = env::var("LC_ALL").ok();
112        let original_lc_messages = env::var("LC_MESSAGES").ok();
113
114        unsafe {
115            // Test with LANG set
116            env::set_var("LANG", "de_DE.UTF-8");
117            env::remove_var("LC_ALL");
118            env::remove_var("LC_MESSAGES");
119        }
120        let result = detect_system_locale();
121        assert_eq!(result, Some("de-DE".to_string()));
122
123        unsafe {
124            // Test with LC_ALL taking priority
125            env::set_var("LC_ALL", "fr_FR.UTF-8");
126            env::set_var("LANG", "de_DE.UTF-8");
127        }
128        let result = detect_system_locale();
129        assert_eq!(result, Some("fr-FR".to_string()));
130
131        unsafe {
132            // Test with LC_MESSAGES taking priority over LANG but not LC_ALL
133            env::set_var("LC_ALL", "es_ES.UTF-8");
134            env::set_var("LC_MESSAGES", "it_IT.UTF-8");
135            env::set_var("LANG", "de_DE.UTF-8");
136        }
137        let result = detect_system_locale();
138        assert_eq!(result, Some("es-ES".to_string())); // LC_ALL should win
139
140        unsafe {
141            // Test with no locale set
142            env::remove_var("LC_ALL");
143            env::remove_var("LC_MESSAGES");
144            env::remove_var("LANG");
145        }
146        let result = detect_system_locale();
147        assert_eq!(result, None);
148
149        // Restore original values
150        unsafe {
151            if let Some(val) = original_lang {
152                env::set_var("LANG", val);
153            } else {
154                env::remove_var("LANG");
155            }
156            if let Some(val) = original_lc_all {
157                env::set_var("LC_ALL", val);
158            } else {
159                env::remove_var("LC_ALL");
160            }
161            if let Some(val) = original_lc_messages {
162                env::set_var("LC_MESSAGES", val);
163            } else {
164                env::remove_var("LC_MESSAGES");
165            }
166        }
167    }
168
169    #[test]
170    fn test_parse_locale_string_edge_cases() {
171        // Test various formats
172        // Single character locales are converted to lowercase
173        assert_eq!(parse_locale_string("C"), Some("c".to_string()));
174        // POSIX is converted to lowercase
175        assert_eq!(parse_locale_string("POSIX"), Some("posix".to_string()));
176        // Test with different encoding
177        assert_eq!(
178            parse_locale_string("en_US.ISO8859-1"),
179            Some("en-US".to_string())
180        );
181        // Test with modifier (@euro) - modifier is preserved in the locale part
182        // The function doesn't strip modifiers, so de_DE@euro becomes de-DE@EURO
183        assert_eq!(
184            parse_locale_string("de_DE@euro"),
185            Some("de-DE@EURO".to_string())
186        );
187
188        // Test invalid formats
189        assert_eq!(parse_locale_string(""), None);
190        assert_eq!(parse_locale_string("   "), None);
191    }
192}
pacsea/i18n/detection.rs

pacsea/i18n/
detection.rs
