base_d/core/
config.rs

1use serde::Deserialize;
2use std::collections::HashMap;
3
4/// Encoding strategy for converting binary data to text.
5///
6/// Different modes offer different tradeoffs between efficiency, compatibility,
7/// and features.
8#[derive(Debug, Clone, Deserialize, PartialEq, Eq)]
9#[serde(rename_all = "snake_case")]
10pub enum EncodingMode {
11    /// Mathematical base conversion treating data as a large number.
12    /// Works with any dictionary size. Output length varies with input.
13    BaseConversion,
14    /// Fixed-size bit chunking per RFC 4648.
15    /// Requires power-of-two dictionary size. Supports padding.
16    Chunked,
17    /// Direct 1:1 byte-to-character mapping using Unicode codepoint ranges.
18    /// Zero encoding overhead. Always 256 characters.
19    ByteRange,
20}
21
22impl Default for EncodingMode {
23    fn default() -> Self {
24        EncodingMode::BaseConversion
25    }
26}
27
28/// Configuration for a single dictionary loaded from TOML.
29#[derive(Debug, Deserialize, Clone)]
30pub struct DictionaryConfig {
31    /// The characters comprising the dictionary
32    #[serde(default)]
33    pub chars: String,
34    /// The encoding mode to use
35    #[serde(default)]
36    pub mode: EncodingMode,
37    /// Optional padding character (e.g., "=" for base64)
38    #[serde(default)]
39    pub padding: Option<String>,
40    /// Starting Unicode codepoint for ByteRange mode
41    #[serde(default)]
42    pub start_codepoint: Option<u32>,
43}
44
45/// Collection of dictionary configurations loaded from TOML files.
46#[derive(Debug, Deserialize)]
47pub struct DictionariesConfig {
48    /// Map of dictionary names to their configurations
49    pub dictionaries: HashMap<String, DictionaryConfig>,
50    /// Compression algorithm configurations
51    #[serde(default)]
52    pub compression: HashMap<String, CompressionConfig>,
53    /// Global settings
54    #[serde(default)]
55    pub settings: Settings,
56}
57
58/// Configuration for a compression algorithm.
59#[derive(Debug, Deserialize, Clone)]
60pub struct CompressionConfig {
61    /// Default compression level
62    pub default_level: u32,
63}
64
65/// xxHash-specific settings.
66#[derive(Debug, Deserialize, Clone, Default)]
67pub struct XxHashSettings {
68    /// Default seed for xxHash algorithms
69    #[serde(default)]
70    pub default_seed: u64,
71    /// Path to default secret file for XXH3 variants
72    #[serde(default)]
73    pub default_secret_file: Option<String>,
74}
75
76/// Global settings for base-d.
77#[derive(Debug, Deserialize, Clone, Default)]
78pub struct Settings {
79    /// Default dictionary to use when compressing without explicit encoding
80    #[serde(default = "default_dictionary")]
81    pub default_dictionary: String,
82    /// xxHash configuration
83    #[serde(default)]
84    pub xxhash: XxHashSettings,
85}
86
87fn default_dictionary() -> String {
88    "base64".to_string()
89}
90
91impl DictionariesConfig {
92    /// Parses dictionary configurations from TOML content.
93    pub fn from_toml(content: &str) -> Result<Self, toml::de::Error> {
94        toml::from_str(content)
95    }
96    
97    /// Loads the built-in dictionary configurations.
98    ///
99    /// Returns the default alphabets bundled with the library.
100    pub fn load_default() -> Result<Self, Box<dyn std::error::Error>> {
101        let content = include_str!("../../dictionaries.toml");
102        Ok(Self::from_toml(content)?)
103    }
104    
105    /// Loads configuration from a custom file path.
106    pub fn load_from_file(path: &std::path::Path) -> Result<Self, Box<dyn std::error::Error>> {
107        let content = std::fs::read_to_string(path)?;
108        Ok(Self::from_toml(&content)?)
109    }
110    
111    /// Loads configuration with user overrides from standard locations.
112    ///
113    /// Searches in priority order:
114    /// 1. Built-in dictionaries (from library)
115    /// 2. `~/.config/base-d/dictionaries.toml` (user overrides)
116    /// 3. `./dictionaries.toml` (project-local overrides)
117    ///
118    /// Later configurations override earlier ones for matching dictionary names.
119    pub fn load_with_overrides() -> Result<Self, Box<dyn std::error::Error>> {
120        let mut config = Self::load_default()?;
121        
122        // Try to load user config from ~/.config/base-d/dictionaries.toml
123        if let Some(config_dir) = dirs::config_dir() {
124            let user_config_path = config_dir.join("base-d").join("dictionaries.toml");
125            if user_config_path.exists() {
126                match Self::load_from_file(&user_config_path) {
127                    Ok(user_config) => {
128                        config.merge(user_config);
129                    }
130                    Err(e) => {
131                        eprintln!("Warning: Failed to load user config from {:?}: {}", user_config_path, e);
132                    }
133                }
134            }
135        }
136        
137        // Try to load local config from ./dictionaries.toml
138        let local_config_path = std::path::Path::new("dictionaries.toml");
139        if local_config_path.exists() {
140            match Self::load_from_file(local_config_path) {
141                Ok(local_config) => {
142                    config.merge(local_config);
143                }
144                Err(e) => {
145                    eprintln!("Warning: Failed to load local config from {:?}: {}", local_config_path, e);
146                }
147            }
148        }
149        
150        Ok(config)
151    }
152    
153    /// Merges another configuration into this one.
154    ///
155    /// Alphabets from `other` override alphabets with the same name in `self`.
156    pub fn merge(&mut self, other: DictionariesConfig) {
157        for (name, dictionary) in other.dictionaries {
158            self.dictionaries.insert(name, dictionary);
159        }
160    }
161    
162    /// Retrieves an dictionary configuration by name.
163    pub fn get_dictionary(&self, name: &str) -> Option<&DictionaryConfig> {
164        self.dictionaries.get(name)
165    }
166}
167
168#[cfg(test)]
169mod tests {
170    use super::*;
171    
172    #[test]
173    fn test_load_default_config() {
174        let config = DictionariesConfig::load_default().unwrap();
175        assert!(config.dictionaries.contains_key("cards"));
176    }
177    
178    #[test]
179    fn test_cards_alphabet_length() {
180        let config = DictionariesConfig::load_default().unwrap();
181        let cards = config.get_dictionary("cards").unwrap();
182        assert_eq!(cards.chars.chars().count(), 52);
183    }
184    
185    #[test]
186    fn test_base64_chunked_mode() {
187        let config = DictionariesConfig::load_default().unwrap();
188        let base64 = config.get_dictionary("base64").unwrap();
189        assert_eq!(base64.mode, EncodingMode::Chunked);
190        assert_eq!(base64.padding, Some("=".to_string()));
191    }
192    
193    #[test]
194    fn test_base64_math_mode() {
195        let config = DictionariesConfig::load_default().unwrap();
196        let base64_math = config.get_dictionary("base64_math").unwrap();
197        assert_eq!(base64_math.mode, EncodingMode::BaseConversion);
198    }
199    
200    #[test]
201    fn test_merge_configs() {
202        let mut config1 = DictionariesConfig {
203            dictionaries: HashMap::new(),
204            compression: HashMap::new(),
205            settings: Settings::default(),
206        };
207        config1.dictionaries.insert("test1".to_string(), DictionaryConfig {
208            chars: "ABC".to_string(),
209            mode: EncodingMode::BaseConversion,
210            padding: None,
211            start_codepoint: None,
212        });
213        
214        let mut config2 = DictionariesConfig {
215            dictionaries: HashMap::new(),
216            compression: HashMap::new(),
217            settings: Settings::default(),
218        };
219        config2.dictionaries.insert("test2".to_string(), DictionaryConfig {
220            chars: "XYZ".to_string(),
221            mode: EncodingMode::BaseConversion,
222            padding: None,
223            start_codepoint: None,
224        });
225        config2.dictionaries.insert("test1".to_string(), DictionaryConfig {
226            chars: "DEF".to_string(),
227            mode: EncodingMode::BaseConversion,
228            padding: None,
229            start_codepoint: None,
230        });
231        
232        config1.merge(config2);
233        
234        assert_eq!(config1.dictionaries.len(), 2);
235        assert_eq!(config1.get_dictionary("test1").unwrap().chars, "DEF");
236        assert_eq!(config1.get_dictionary("test2").unwrap().chars, "XYZ");
237    }
238    
239    #[test]
240    fn test_load_from_toml_string() {
241        let toml_content = r#"
242[dictionaries.custom]
243chars = "0123456789"
244mode = "base_conversion"
245"#;
246        let config = DictionariesConfig::from_toml(toml_content).unwrap();
247        assert!(config.dictionaries.contains_key("custom"));
248        assert_eq!(config.get_dictionary("custom").unwrap().chars, "0123456789");
249    }
250}
251
base_d/core/config.rs

base_d/core/
config.rs
