429 lines
14 KiB
Rust
429 lines
14 KiB
Rust
extern crate serde;
|
||
extern crate serde_json;
|
||
extern crate serde_yaml;
|
||
use serde::{Deserialize, Serialize};
|
||
|
||
extern crate log;
|
||
use log::{debug, error, info};
|
||
|
||
pub mod utils;
|
||
use utils::SettingsType;
|
||
|
||
pub mod regex_wrapper;
|
||
use regex_wrapper::Regex;
|
||
|
||
/// Current version of the ruleset. It will help determine if the ruleset is
|
||
/// outdated or from a more recent version of the software than the one being in
|
||
/// use.
|
||
pub const RULESET_CURRENT_VERSION: i32 = 1;
|
||
|
||
/// Encode a [`Settings`] struct to a filetype, returns a
|
||
/// `std::result::Result<std::string::String, std::io::Error>`
|
||
///
|
||
/// # Arguments
|
||
///
|
||
/// * `funcrate` - `serde`-compatible crate to use, must implement `to_string`
|
||
/// * `content` - content to encode, must be `Settings` struct
|
||
///
|
||
/// # Example
|
||
///
|
||
/// ```ignore
|
||
/// # use lang_evolve_core::settings::*;
|
||
/// # use lang_evolve_core::encode_settings;
|
||
/// use std::io::{Error, ErrorKind};
|
||
/// use std::path::Path;
|
||
/// let filetype = utils::get_file_type(Path::new("./path/to/file.json"));
|
||
/// let s = Settings::new();
|
||
/// let content = match filetype {
|
||
/// utils::SettingsType::Yaml => encode_settings!(serde_yaml, &s).unwrap(),
|
||
/// utils::SettingsType::Json => encode_settings!(serde_json, &s).unwrap(),
|
||
/// _ => panic!("Could not encode settings"),
|
||
/// };
|
||
/// ```
|
||
///
|
||
/// [`Settings`]: ./settings/struct.Settings.html
|
||
#[macro_export(local_inner_macros)]
|
||
macro_rules! encode_settings {
|
||
($funcrate:ident, $content:expr) => {
|
||
match $funcrate::to_string($content) {
|
||
Err(e) => {
|
||
log::error!("Could not serialize settings: {}", e.to_string());
|
||
return Err(std::io::Error::new(
|
||
std::io::ErrorKind::InvalidData,
|
||
e,
|
||
));
|
||
}
|
||
Ok(val) => val,
|
||
}
|
||
};
|
||
}
|
||
|
||
/// Decode a [`Settings`] struct from a `std::std::String`, returns a
|
||
/// std::result::Result<lang_evolve_core::settings::Settings, std::io::Error>
|
||
///
|
||
/// # Arguments
|
||
///
|
||
/// * `funcrate` - `serde`-compatible crate to use, mus implement `from_string`
|
||
/// * `content` - `&str` to decode into a [`Settings`]
|
||
///
|
||
/// # Example
|
||
///
|
||
/// ```ignore
|
||
/// # use lang_evolve_core::decode_settings;
|
||
/// let str = r#"{"version":"1","categories":[],"rules":[]}"#;
|
||
/// let settings = decode_settings!(serde_json, str);
|
||
/// ```
|
||
///
|
||
/// [`Settings`]: ./settings/struct.Settings.html
|
||
#[macro_export(local_inner_macros)]
|
||
macro_rules! decode_settings {
|
||
($funcrate:ident, $content:expr) => {
|
||
match $funcrate::from_str($content) {
|
||
Err(e) => {
|
||
log::error!("Could not import settings: {}", e.to_string());
|
||
return Err(std::io::Error::new(
|
||
std::io::ErrorKind::InvalidInput,
|
||
e,
|
||
));
|
||
}
|
||
Ok(val) => val,
|
||
}
|
||
};
|
||
}
|
||
|
||
use std::collections::HashMap;
|
||
/// Representation of the software’s settings
|
||
///
|
||
/// This struct represents all the settings the software has to follow
|
||
/// while running, which includes the phoneme categories as well as
|
||
/// the soundchange rules to apply to the input text.
|
||
#[derive(Clone, Debug, Deserialize, Serialize)]
|
||
pub struct Settings {
|
||
/// Represents the version of the loaded ruleset.
|
||
///
|
||
/// It is used to detect obsolete ruleset representations or if a
|
||
/// loaded ruleset comes from a newer version of lang_evolve_core
|
||
/// than the one used by the user.
|
||
#[serde(default = "Settings::get_ruleset_version")]
|
||
pub version: String,
|
||
|
||
/// Categories of phonemes
|
||
///
|
||
/// This is a vector of categories of phonemes, represented
|
||
/// themselves as pairs of strings. Each pair of strings has its
|
||
/// first element represent the name of the category, which is
|
||
/// generally represented by a single capital letter. The second
|
||
/// element is a string where all its characters represent
|
||
/// phonemes. It is currently not possible to have more than one
|
||
/// character to be considered as one sound.
|
||
#[serde(default)]
|
||
pub categories: HashMap<String, String>,
|
||
|
||
/// Soundchange rules
|
||
///
|
||
/// This is a vector of pairs of strings, the first one represents
|
||
/// a regex to be matched while the second represents the change
|
||
/// to be made to the input data.
|
||
#[serde(default)]
|
||
pub rules: Vec<(Regex, String)>,
|
||
}
|
||
|
||
/// Representation inside the crate of LangEvolve’s settings.
|
||
impl Settings {
|
||
/// Creates a new empty instance of [`Settings`]
|
||
///
|
||
/// # Example
|
||
///
|
||
/// ```
|
||
/// let s = lang_evolve_core::settings::Settings::new();
|
||
/// let content_yaml = r#"---
|
||
/// version: "1"
|
||
/// categories: {}
|
||
/// rules: []"#;
|
||
/// let content_json = r#"{"version":"1","categories":{},"rules":[]}"#;
|
||
/// assert_eq!(content_yaml, serde_yaml::to_string(&s).unwrap());
|
||
/// assert_eq!(content_json, serde_json::to_string(&s).unwrap());
|
||
/// ```
|
||
///
|
||
/// [`Settings`]: ./settings/struct.Settings.html
|
||
pub fn new() -> Self {
|
||
Self {
|
||
version: Self::get_ruleset_version(),
|
||
categories: HashMap::new(),
|
||
rules: Vec::new(),
|
||
}
|
||
}
|
||
|
||
/// Import settings from an imput file.
|
||
///
|
||
/// The currently allowed file formats are described in the
|
||
/// [`utils::SettingsType`] enum. If the ruleset version is higher than the
|
||
/// current version (see [`Settings.version`]), then an error is returned.
|
||
///
|
||
/// # Arguments
|
||
///
|
||
/// * `path` - File to open and load settings from
|
||
///
|
||
/// # Example
|
||
///
|
||
/// ```no_run
|
||
/// use std::path::Path;
|
||
/// use lang_evolve_core::settings::Settings;
|
||
/// let path_json = Path::new("settings.json");
|
||
/// let _s_json = Settings::import(&path_json).unwrap();
|
||
///
|
||
/// let path_yaml = Path::new("settings.yaml");
|
||
/// let _s_yaml = Settings::import(&path_yaml).unwrap();
|
||
///
|
||
/// let path_yml = Path::new("settings.yml");
|
||
/// let _s_yml = Settings::import(&path_yml).unwrap();
|
||
/// ```
|
||
///
|
||
/// [`utils::SettingsTYpe`]: ./utils/enum.SettingsType.html
|
||
/// [`Settings.version`]: ./struct.Settings.html#structfield.version
|
||
pub fn import(path: &std::path::Path) -> std::io::Result<Self> {
|
||
use utils::SettingsType::{Json, Yaml};
|
||
let file_type = utils::get_file_type(&path);
|
||
let content = utils::read_file(&path)?;
|
||
|
||
let settings: Settings = match file_type {
|
||
Yaml => decode_settings!(serde_yaml, &content),
|
||
Json => decode_settings!(serde_json, &content),
|
||
// Attempt to decode anyway
|
||
_ => match Settings::from_str(&content.as_str()) {
|
||
Ok(val) => val,
|
||
Err(e) => {
|
||
error!(
|
||
"Could not decode input {}: {}",
|
||
content,
|
||
e.to_string()
|
||
);
|
||
return Err(std::io::Error::new(
|
||
std::io::ErrorKind::InvalidData,
|
||
e,
|
||
));
|
||
}
|
||
},
|
||
};
|
||
if settings.version > Self::get_ruleset_version() {
|
||
error!("Ruleset version too high!");
|
||
return Err(std::io::Error::new(
|
||
std::io::ErrorKind::InvalidData,
|
||
"Ruleset version too high!",
|
||
));
|
||
}
|
||
info!("Successfuly imported {}", path.display());
|
||
Ok(settings)
|
||
}
|
||
|
||
/// Import settings from file path described by the argument `path`
|
||
///
|
||
/// # Arguments
|
||
///
|
||
/// * `path` - path to the file from which settings should be imported
|
||
///
|
||
/// # Example
|
||
///
|
||
/// ```no_run
|
||
/// let s = lang_evolve_core::settings::Settings::from("settings.yml");
|
||
/// ```
|
||
pub fn from<S>(s: S) -> Self
|
||
where
|
||
S: ToString,
|
||
{
|
||
let s = s.to_string();
|
||
let path = std::path::Path::new(&s);
|
||
Self::import(&path).unwrap()
|
||
}
|
||
|
||
/// Export current settings to a file.
|
||
///
|
||
/// The allowed file formats are described in the [`SettingsType`] enum.
|
||
///
|
||
/// # Arguments
|
||
///
|
||
/// * `path` - Path to write and export settings to
|
||
///
|
||
/// # Example
|
||
///
|
||
/// ```
|
||
/// use std::path::Path;
|
||
/// let s = lang_evolve_core::settings::Settings::new();
|
||
///
|
||
/// // Export to JSON
|
||
/// let path_json = Path::new("./output.json");
|
||
/// s.export(&path_json).unwrap();
|
||
///
|
||
/// // Export to Yaml, both ".yml" and ".yaml" work
|
||
/// let path_yaml = Path::new("./output.yaml");
|
||
/// s.export(&path_yaml).unwrap();
|
||
/// let path_yml = Path::new("./output.yml");
|
||
/// s.export(&path_yml).unwrap();
|
||
/// ```
|
||
///
|
||
/// [`SettingsType`]: ./utils/enum.SettingsType.html
|
||
pub fn export(&self, path: &std::path::Path) -> std::io::Result<()> {
|
||
let filetype = utils::get_file_type(&path);
|
||
let content = match filetype {
|
||
SettingsType::Yaml => encode_settings!(serde_yaml, &self),
|
||
SettingsType::Json => encode_settings!(serde_json, &self),
|
||
_ => {
|
||
error!("Unknown filetype {}", path.to_str().unwrap());
|
||
return Err(std::io::Error::new(
|
||
std::io::ErrorKind::InvalidData,
|
||
"Unknown file type",
|
||
));
|
||
}
|
||
};
|
||
info!("Successfuly exported settings to {}", path.display());
|
||
utils::write_file(&path, &content)
|
||
}
|
||
|
||
/// Get the current ruleset version of LangEvolve.
|
||
fn get_ruleset_version() -> String {
|
||
RULESET_CURRENT_VERSION.to_string()
|
||
}
|
||
|
||
fn update_rules(&self) -> std::result::Result<Vec<(Regex, String)>, String> {
|
||
let mut rules = self.rules.clone();
|
||
|
||
// TODO break categories in different rules
|
||
for (from, to) in rules.iter_mut() {
|
||
let re = Regex::new("%\\D");
|
||
let from_match = re.is_match(from.as_str());
|
||
let to_match = re.is_match(to);
|
||
if from_match || to_match {
|
||
for (category, content) in &self.categories {
|
||
if from_match {
|
||
*from = Regex::new(
|
||
from.to_string()
|
||
.replace(
|
||
format!("%{}", category).as_str(),
|
||
format!("[{}]", content).as_str(),
|
||
)
|
||
.as_str()
|
||
);
|
||
}
|
||
if to_match {
|
||
*to = to.to_string().replace(
|
||
format!("%{}", category).as_str(),
|
||
format!("[{}]", content).as_str()
|
||
);
|
||
}
|
||
}
|
||
}
|
||
}
|
||
Ok(rules)
|
||
}
|
||
|
||
/// Apply list of rules to input
|
||
///
|
||
/// The list of rules in the struct will be applied to the input `s`. If the
|
||
/// rule contains the `%` character followed by a capital letter, this marks
|
||
/// a category of phonemes and should be replaced by them. For instance, we
|
||
/// have:
|
||
/// - the category `C` defined as `bcdfg`
|
||
/// - the rule `%Ci` to `%Cj`
|
||
/// The rule should be rewritten as `[bcdfg]` to `[bcdfg]j`
|
||
///
|
||
/// # Arguments
|
||
///
|
||
/// * `s` - Input to modify
|
||
///
|
||
/// # Example
|
||
///
|
||
/// ```
|
||
/// # use lang_evolve_core::settings::Settings;
|
||
/// let settings = Settings::new();
|
||
/// // add some rules...
|
||
/// let input = String::new();
|
||
/// // set some input
|
||
/// let _output = settings.apply(input);
|
||
/// ```
|
||
pub fn apply(&self, s: String) -> std::result::Result<String, String> {
|
||
// TODO Add Error handling
|
||
let rules = self.update_rules().unwrap();
|
||
let mut s = s.clone();
|
||
debug!("===============================================");
|
||
for (from, to) in rules {
|
||
debug!("from: {}\tto: {}", from.to_string(), to);
|
||
debug!("old: {}", s);
|
||
s = from.replace_all(&s, to.as_str()).to_string();
|
||
debug!("new: {}", s);
|
||
}
|
||
Ok(s)
|
||
}
|
||
}
|
||
|
||
use std::str::FromStr;
|
||
impl FromStr for Settings {
|
||
type Err = serde_yaml::Error;
|
||
|
||
/// Decode a litteral string into a `Settings` struct. Works only for
|
||
/// supported file types described in `SettingsType`. It will try to decode
|
||
/// the input `s` by any mean known by `SettingsType`.
|
||
///
|
||
/// # Arguments
|
||
///
|
||
/// * `s` - litteral string to decode into a `Settings` struct
|
||
///
|
||
/// # Example
|
||
///
|
||
/// ```
|
||
/// # use std::str::FromStr;
|
||
/// let s = r#"{"version":"1","categories":{},"rules":[]}"#;
|
||
/// let settings = lang_evolve_core::settings::Settings::from_str(s).unwrap();
|
||
/// ```
|
||
fn from_str(s: &str) -> Result<Self, Self::Err> {
|
||
match serde_json::from_str::<Settings>(s) {
|
||
Ok(val) => Ok(val),
|
||
Err(_) => match serde_yaml::from_str::<Settings>(s) {
|
||
Ok(val) => Ok(val),
|
||
Err(e) => {
|
||
error!("Could not decode input {}: {}", s, e.to_string());
|
||
return Err(e);
|
||
}
|
||
},
|
||
}
|
||
}
|
||
}
|
||
|
||
use std::fmt;
|
||
use std::fmt::Display;
|
||
impl Display for Settings {
|
||
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
|
||
write!(f, "{}", serde_json::to_string(&self).unwrap())
|
||
}
|
||
}
|
||
|
||
impl PartialEq for Settings {
|
||
fn eq(&self, other: &Self) -> bool {
|
||
self.version == other.version
|
||
&& self.categories == other.categories
|
||
&& self.rules == other.rules
|
||
}
|
||
}
|
||
|
||
impl Eq for Settings {}
|
||
|
||
#[test]
|
||
fn write_settings() {
|
||
let s = Settings::new();
|
||
let path = std::path::Path::new("test.yaml");
|
||
let settings = r#"---
|
||
version: "1"
|
||
categories: {}
|
||
rules: []"#;
|
||
utils::write_file(&path, &serde_yaml::to_string(&s).unwrap()).unwrap();
|
||
assert_eq!(settings, utils::read_file(&path).unwrap());
|
||
}
|
||
|
||
#[test]
|
||
fn read_settings() {
|
||
let s1 = Settings::new();
|
||
let path = std::path::Path::new("test.yml");
|
||
s1.export(&path).unwrap();
|
||
let s2 = Settings::import(&path).unwrap();
|
||
assert_eq!(s1, s2);
|
||
}
|