Compare commits
20 Commits
6be0f7e8f6
...
master
| Author | SHA1 | Date | |
|---|---|---|---|
|
531c9b1acc
|
|||
|
cb933d4896
|
|||
|
d33190a46d
|
|||
|
bb9d8703f3
|
|||
|
144771a2c0
|
|||
|
22fbe5aba7
|
|||
|
9cc1a52e5a
|
|||
|
78541f10ba
|
|||
|
8ddd33d76d
|
|||
|
a8bafc072b
|
|||
|
3c960eaa35
|
|||
|
de45ffc15c
|
|||
|
042fd066f0
|
|||
|
071c8b0728
|
|||
|
5f30c6d636
|
|||
|
dbbb1616dd
|
|||
|
23e3acb182
|
|||
|
6fb1c287e0
|
|||
|
ef8c02fc97
|
|||
|
bae1d86544
|
@@ -9,10 +9,15 @@ edition = "2018"
|
|||||||
[dependencies]
|
[dependencies]
|
||||||
# Logger
|
# Logger
|
||||||
log = "0.4"
|
log = "0.4"
|
||||||
simplelog = "0.7"
|
simplelog = "0.8"
|
||||||
# Struct serializing and deserializing
|
# Struct serializing and deserializing
|
||||||
serde = {version = "1.0", features = ["derive"]}
|
serde = {version = "1.0", features = ["derive"]}
|
||||||
serde_yaml = "0.7"
|
serde_yaml = "0.8"
|
||||||
serde_json = "1.0"
|
serde_json = "1.0"
|
||||||
# Regex support
|
# Regex support
|
||||||
regex = "1.3"
|
regex = "1.3"
|
||||||
|
lazy_static = "1.4"
|
||||||
|
|
||||||
|
# [dev-dependencies]
|
||||||
|
# Pretty output
|
||||||
|
prettydiff = "0.3"
|
||||||
|
|||||||
78
src/lib.rs
78
src/lib.rs
@@ -1,4 +1,5 @@
|
|||||||
#![crate_name = "lang_evolve_core"]
|
#![crate_name = "lang_evolve_core"]
|
||||||
|
#![deny(clippy::all)]
|
||||||
|
|
||||||
//! # LangEvolveCore
|
//! # LangEvolveCore
|
||||||
//!
|
//!
|
||||||
@@ -13,13 +14,12 @@
|
|||||||
//! user-defined sound changes to words and texts based on regex expressions.
|
//! user-defined sound changes to words and texts based on regex expressions.
|
||||||
|
|
||||||
use std::fs::File;
|
use std::fs::File;
|
||||||
extern crate log;
|
|
||||||
extern crate simplelog;
|
|
||||||
use log::{info, warn};
|
use log::{info, warn};
|
||||||
use simplelog::*;
|
use simplelog::*;
|
||||||
|
|
||||||
|
|
||||||
pub mod settings;
|
pub mod settings;
|
||||||
|
pub mod utils;
|
||||||
|
|
||||||
/// Initializes the crate
|
/// Initializes the crate
|
||||||
///
|
///
|
||||||
@@ -41,19 +41,39 @@ pub mod settings;
|
|||||||
/// lang_evolve_core::init();
|
/// lang_evolve_core::init();
|
||||||
/// ```
|
/// ```
|
||||||
pub fn init() -> std::result::Result<(), log::SetLoggerError> {
|
pub fn init() -> std::result::Result<(), log::SetLoggerError> {
|
||||||
match CombinedLogger::init(vec![
|
// #[cfg(debug_assertions)]
|
||||||
TermLogger::new(
|
match CombinedLogger::init(if cfg!(debug_assertions) {
|
||||||
LevelFilter::Warn,
|
vec![
|
||||||
Config::default(),
|
WriteLogger::new(
|
||||||
TerminalMode::Mixed,
|
LevelFilter::Warn,
|
||||||
)
|
Config::default(),
|
||||||
.unwrap(),
|
File::create("core.log").unwrap(),
|
||||||
WriteLogger::new(
|
),
|
||||||
LevelFilter::Info,
|
WriteLogger::new(
|
||||||
Config::default(),
|
LevelFilter::Debug,
|
||||||
File::create("core.log").unwrap(),
|
Config::default(),
|
||||||
),
|
File::create("core.log").unwrap(),
|
||||||
]) {
|
),
|
||||||
|
WriteLogger::new(
|
||||||
|
LevelFilter::Info,
|
||||||
|
Config::default(),
|
||||||
|
File::create("core.log").unwrap(),
|
||||||
|
),
|
||||||
|
]
|
||||||
|
} else {
|
||||||
|
vec![
|
||||||
|
WriteLogger::new(
|
||||||
|
LevelFilter::Warn,
|
||||||
|
Config::default(),
|
||||||
|
File::create("core.log").unwrap(),
|
||||||
|
),
|
||||||
|
WriteLogger::new(
|
||||||
|
LevelFilter::Info,
|
||||||
|
Config::default(),
|
||||||
|
File::create("core.log").unwrap(),
|
||||||
|
),
|
||||||
|
]
|
||||||
|
}) {
|
||||||
Err(why) => {
|
Err(why) => {
|
||||||
warn!("Could not initialize logger: {}", why.to_string());
|
warn!("Could not initialize logger: {}", why.to_string());
|
||||||
Err(why)
|
Err(why)
|
||||||
@@ -64,3 +84,29 @@ pub fn init() -> std::result::Result<(), log::SetLoggerError> {
|
|||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
#[cfg(test)]
|
||||||
|
mod tests {
|
||||||
|
use super::*;
|
||||||
|
|
||||||
|
#[test]
|
||||||
|
fn write_settings() {
|
||||||
|
let s = settings::Settings::new();
|
||||||
|
let path = std::path::Path::new("settings.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::Settings::new();
|
||||||
|
let path = std::path::Path::new("settings.yml");
|
||||||
|
s1.export(&path).unwrap();
|
||||||
|
let s2 = settings::Settings::import(&path).unwrap();
|
||||||
|
assert_eq!(s1, s2);
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|||||||
@@ -1,21 +1,18 @@
|
|||||||
extern crate serde;
|
|
||||||
extern crate serde_json;
|
|
||||||
extern crate serde_yaml;
|
|
||||||
use serde::{Deserialize, Serialize};
|
use serde::{Deserialize, Serialize};
|
||||||
|
|
||||||
extern crate log;
|
use log::{debug, error, info};
|
||||||
use log::{error, info};
|
|
||||||
|
|
||||||
pub mod utils;
|
use crate::utils::{self, SettingsType};
|
||||||
use utils::SettingsType;
|
|
||||||
|
|
||||||
pub mod regex_wrapper;
|
use prettydiff::diff_words;
|
||||||
use regex_wrapper::Regex;
|
|
||||||
|
mod rule;
|
||||||
|
use rule::Rule;
|
||||||
|
|
||||||
/// Current version of the ruleset. It will help determine if the ruleset is
|
/// 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
|
/// outdated or from a more recent version of the software than the one being in
|
||||||
/// use.
|
/// use.
|
||||||
pub const RULESET_CURRENT_VERSION: i32 = 1;
|
const RULESET_CURRENT_VERSION: i32 = 1;
|
||||||
|
|
||||||
/// Encode a [`Settings`] struct to a filetype, returns a
|
/// Encode a [`Settings`] struct to a filetype, returns a
|
||||||
/// `std::result::Result<std::string::String, std::io::Error>`
|
/// `std::result::Result<std::string::String, std::io::Error>`
|
||||||
@@ -27,22 +24,21 @@ pub const RULESET_CURRENT_VERSION: i32 = 1;
|
|||||||
///
|
///
|
||||||
/// # Example
|
/// # Example
|
||||||
///
|
///
|
||||||
/// ```ignore
|
/// ```no_run
|
||||||
/// # use lang_evolve_core::settings::*;
|
|
||||||
/// # use lang_evolve_core::encode_settings;
|
|
||||||
/// use std::io::{Error, ErrorKind};
|
|
||||||
/// use std::path::Path;
|
/// use std::path::Path;
|
||||||
/// let filetype = utils::get_file_type(Path::new("./path/to/file.json"));
|
/// use lang_evolve_core::utils;
|
||||||
/// let s = Settings::new();
|
///
|
||||||
|
/// let settings = Settings::new();
|
||||||
|
/// let filetype = utils::get_file_type(Path::new("settings.yml"));
|
||||||
|
///
|
||||||
/// let content = match filetype {
|
/// let content = match filetype {
|
||||||
/// utils::SettingsType::Yaml => encode_settings!(serde_yaml, &s).unwrap(),
|
/// SettingsType::Yaml => encode_settings!(serde_yaml, &settings),
|
||||||
/// utils::SettingsType::Json => encode_settings!(serde_json, &s).unwrap(),
|
/// SettingsType::Json => encode_settings!(serde_json, &settings),
|
||||||
/// _ => panic!("Could not encode settings"),
|
/// _ => String::from("Error!"),
|
||||||
/// };
|
/// };
|
||||||
/// ```
|
/// ```
|
||||||
///
|
///
|
||||||
/// [`Settings`]: ./settings/struct.Settings.html
|
/// [`Settings`]: ./settings/struct.Settings.html
|
||||||
#[macro_export(local_inner_macros)]
|
|
||||||
macro_rules! encode_settings {
|
macro_rules! encode_settings {
|
||||||
($funcrate:ident, $content:expr) => {
|
($funcrate:ident, $content:expr) => {
|
||||||
match $funcrate::to_string($content) {
|
match $funcrate::to_string($content) {
|
||||||
@@ -68,14 +64,13 @@ macro_rules! encode_settings {
|
|||||||
///
|
///
|
||||||
/// # Example
|
/// # Example
|
||||||
///
|
///
|
||||||
/// ```ignore
|
/// ```no_run
|
||||||
/// # use lang_evolve_core::decode_settings;
|
/// # use lang_evolve_core::decode_settings;
|
||||||
/// let str = r#"{"version":"1","categories":[],"rules":[]}"#;
|
/// let str = r#"{"version":"1","categories":[],"rules":[]}"#;
|
||||||
/// let settings = decode_settings!(serde_json, str);
|
/// let settings = decode_settings!(serde_json, str);
|
||||||
/// ```
|
/// ```
|
||||||
///
|
///
|
||||||
/// [`Settings`]: ./settings/struct.Settings.html
|
/// [`Settings`]: ./settings/struct.Settings.html
|
||||||
#[macro_export(local_inner_macros)]
|
|
||||||
macro_rules! decode_settings {
|
macro_rules! decode_settings {
|
||||||
($funcrate:ident, $content:expr) => {
|
($funcrate:ident, $content:expr) => {
|
||||||
match $funcrate::from_str($content) {
|
match $funcrate::from_str($content) {
|
||||||
@@ -91,6 +86,8 @@ macro_rules! decode_settings {
|
|||||||
};
|
};
|
||||||
}
|
}
|
||||||
|
|
||||||
|
use std::collections::HashMap;
|
||||||
|
|
||||||
/// Representation of the software’s settings
|
/// Representation of the software’s settings
|
||||||
///
|
///
|
||||||
/// This struct represents all the settings the software has to follow
|
/// This struct represents all the settings the software has to follow
|
||||||
@@ -104,7 +101,7 @@ pub struct Settings {
|
|||||||
/// loaded ruleset comes from a newer version of lang_evolve_core
|
/// loaded ruleset comes from a newer version of lang_evolve_core
|
||||||
/// than the one used by the user.
|
/// than the one used by the user.
|
||||||
#[serde(default = "Settings::get_ruleset_version")]
|
#[serde(default = "Settings::get_ruleset_version")]
|
||||||
pub version: String,
|
version: String,
|
||||||
|
|
||||||
/// Categories of phonemes
|
/// Categories of phonemes
|
||||||
///
|
///
|
||||||
@@ -116,7 +113,7 @@ pub struct Settings {
|
|||||||
/// phonemes. It is currently not possible to have more than one
|
/// phonemes. It is currently not possible to have more than one
|
||||||
/// character to be considered as one sound.
|
/// character to be considered as one sound.
|
||||||
#[serde(default)]
|
#[serde(default)]
|
||||||
pub categories: Vec<(String, String)>,
|
categories: HashMap<String, String>,
|
||||||
|
|
||||||
/// Soundchange rules
|
/// Soundchange rules
|
||||||
///
|
///
|
||||||
@@ -124,7 +121,7 @@ pub struct Settings {
|
|||||||
/// a regex to be matched while the second represents the change
|
/// a regex to be matched while the second represents the change
|
||||||
/// to be made to the input data.
|
/// to be made to the input data.
|
||||||
#[serde(default)]
|
#[serde(default)]
|
||||||
pub rules: Vec<(Regex, Regex)>,
|
rules: Vec<Rule>,
|
||||||
}
|
}
|
||||||
|
|
||||||
/// Representation inside the crate of LangEvolve’s settings.
|
/// Representation inside the crate of LangEvolve’s settings.
|
||||||
@@ -137,9 +134,9 @@ impl Settings {
|
|||||||
/// let s = lang_evolve_core::settings::Settings::new();
|
/// let s = lang_evolve_core::settings::Settings::new();
|
||||||
/// let content_yaml = r#"---
|
/// let content_yaml = r#"---
|
||||||
/// version: "1"
|
/// version: "1"
|
||||||
/// categories: []
|
/// categories: {}
|
||||||
/// rules: []"#;
|
/// rules: []"#;
|
||||||
/// let content_json = 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_yaml, serde_yaml::to_string(&s).unwrap());
|
||||||
/// assert_eq!(content_json, serde_json::to_string(&s).unwrap());
|
/// assert_eq!(content_json, serde_json::to_string(&s).unwrap());
|
||||||
/// ```
|
/// ```
|
||||||
@@ -148,7 +145,7 @@ impl Settings {
|
|||||||
pub fn new() -> Self {
|
pub fn new() -> Self {
|
||||||
Self {
|
Self {
|
||||||
version: Self::get_ruleset_version(),
|
version: Self::get_ruleset_version(),
|
||||||
categories: Vec::new(),
|
categories: HashMap::new(),
|
||||||
rules: Vec::new(),
|
rules: Vec::new(),
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
@@ -165,9 +162,15 @@ impl Settings {
|
|||||||
///
|
///
|
||||||
/// # Example
|
/// # Example
|
||||||
///
|
///
|
||||||
/// ```no_run
|
/// ```
|
||||||
/// use std::path::Path;
|
/// use std::path::Path;
|
||||||
/// use lang_evolve_core::settings::Settings;
|
/// # use lang_evolve_core::settings::Settings;
|
||||||
|
/// # let s = Settings::new();
|
||||||
|
/// # for path in vec!["settings.json", "settings.yaml", "settings.yml"] {
|
||||||
|
/// # let path = Path::new(path);
|
||||||
|
/// # s.export(&path).unwrap();
|
||||||
|
/// # }
|
||||||
|
///
|
||||||
/// let path_json = Path::new("settings.json");
|
/// let path_json = Path::new("settings.json");
|
||||||
/// let _s_json = Settings::import(&path_json).unwrap();
|
/// let _s_json = Settings::import(&path_json).unwrap();
|
||||||
///
|
///
|
||||||
@@ -181,10 +184,9 @@ impl Settings {
|
|||||||
/// [`utils::SettingsTYpe`]: ./utils/enum.SettingsType.html
|
/// [`utils::SettingsTYpe`]: ./utils/enum.SettingsType.html
|
||||||
/// [`Settings.version`]: ./struct.Settings.html#structfield.version
|
/// [`Settings.version`]: ./struct.Settings.html#structfield.version
|
||||||
pub fn import(path: &std::path::Path) -> std::io::Result<Self> {
|
pub fn import(path: &std::path::Path) -> std::io::Result<Self> {
|
||||||
use utils::SettingsType::{Json, Yaml};
|
use SettingsType::{Json, Yaml};
|
||||||
let file_type = utils::get_file_type(&path);
|
let file_type = utils::get_file_type(&path);
|
||||||
let content = utils::read_file(&path)?;
|
let content = utils::read_file(&path)?;
|
||||||
|
|
||||||
let settings: Settings = match file_type {
|
let settings: Settings = match file_type {
|
||||||
Yaml => decode_settings!(serde_yaml, &content),
|
Yaml => decode_settings!(serde_yaml, &content),
|
||||||
Json => decode_settings!(serde_json, &content),
|
Json => decode_settings!(serde_json, &content),
|
||||||
@@ -204,15 +206,17 @@ impl Settings {
|
|||||||
}
|
}
|
||||||
},
|
},
|
||||||
};
|
};
|
||||||
|
|
||||||
if settings.version > Self::get_ruleset_version() {
|
if settings.version > Self::get_ruleset_version() {
|
||||||
error!("Ruleset version too high!");
|
error!("Ruleset version too high!");
|
||||||
return Err(std::io::Error::new(
|
Err(std::io::Error::new(
|
||||||
std::io::ErrorKind::InvalidData,
|
std::io::ErrorKind::InvalidData,
|
||||||
"Ruleset version too high!",
|
"Ruleset version too high!",
|
||||||
));
|
))
|
||||||
|
} else {
|
||||||
|
info!("Successfuly imported {}", path.display());
|
||||||
|
Ok(settings)
|
||||||
}
|
}
|
||||||
info!("Successfuly imported {}", path.display());
|
|
||||||
Ok(settings)
|
|
||||||
}
|
}
|
||||||
|
|
||||||
/// Import settings from file path described by the argument `path`
|
/// Import settings from file path described by the argument `path`
|
||||||
@@ -223,16 +227,68 @@ impl Settings {
|
|||||||
///
|
///
|
||||||
/// # Example
|
/// # Example
|
||||||
///
|
///
|
||||||
/// ```no_run
|
|
||||||
/// let s = lang_evolve_core::settings::Settings::from("settings.yml");
|
|
||||||
/// ```
|
/// ```
|
||||||
pub fn from<S>(s: S) -> Self
|
/// # use lang_evolve_core::settings::Settings;
|
||||||
|
/// # use std::path::Path;
|
||||||
|
/// # let s = Settings::default();
|
||||||
|
/// # s.export(Path::new("settings.yml"));
|
||||||
|
/// let s = Settings::from("settings.yml");
|
||||||
|
/// ```
|
||||||
|
pub fn from<S>(s: S) -> std::io::Result<Self>
|
||||||
where
|
where
|
||||||
S: ToString,
|
S: ToString,
|
||||||
{
|
{
|
||||||
let s = s.to_string();
|
let s = s.to_string();
|
||||||
let path = std::path::Path::new(&s);
|
let path = std::path::Path::new(&s);
|
||||||
Self::import(&path).unwrap()
|
Self::import(&path)
|
||||||
|
}
|
||||||
|
|
||||||
|
/// Add a new rule to the current settings
|
||||||
|
///
|
||||||
|
/// # Arguments
|
||||||
|
///
|
||||||
|
/// * `from` - Regex that should match the text to be replaced
|
||||||
|
/// * `to` - Regex that should replace some text
|
||||||
|
///
|
||||||
|
/// # Example
|
||||||
|
///
|
||||||
|
/// ```
|
||||||
|
/// # use lang_evolve_core::settings::Settings;
|
||||||
|
/// let mut settings = Settings::default();
|
||||||
|
/// settings.add_rule("ha", "wa");
|
||||||
|
///
|
||||||
|
/// use std::str::FromStr;
|
||||||
|
/// let reference = Settings::from_str(
|
||||||
|
/// r#"{"version":"1","categories":{},"rules":[{"from":"ha","to":"wa"}]}"#)
|
||||||
|
/// .unwrap();
|
||||||
|
/// assert_eq!(reference, settings);
|
||||||
|
/// ```
|
||||||
|
pub fn add_rule(&mut self, from: &str, to: &str) {
|
||||||
|
self.rules.push(Rule::new(from, to))
|
||||||
|
}
|
||||||
|
|
||||||
|
/// Add a new category of phonemes to the current settings
|
||||||
|
///
|
||||||
|
/// # Arguments
|
||||||
|
///
|
||||||
|
/// * `name` - Name of the category
|
||||||
|
/// * `content` - Content of the category, phonemes
|
||||||
|
///
|
||||||
|
/// # Example
|
||||||
|
///
|
||||||
|
/// ```
|
||||||
|
/// # use lang_evolve_core::settings::Settings;
|
||||||
|
/// let mut settings = Settings::default();
|
||||||
|
/// settings.add_category("C", "abcde");
|
||||||
|
///
|
||||||
|
/// use std::str::FromStr;
|
||||||
|
/// let reference = Settings::from_str(
|
||||||
|
/// r#"{"version":"1","categories":{"C": "abcde"},"rules":[]}"#)
|
||||||
|
/// .unwrap();
|
||||||
|
/// assert_eq!(reference, settings);
|
||||||
|
/// ```
|
||||||
|
pub fn add_category(&mut self, name: &str, content: &str) {
|
||||||
|
self.categories.insert(String::from(name), String::from(content));
|
||||||
}
|
}
|
||||||
|
|
||||||
/// Export current settings to a file.
|
/// Export current settings to a file.
|
||||||
@@ -246,17 +302,19 @@ impl Settings {
|
|||||||
/// # Example
|
/// # Example
|
||||||
///
|
///
|
||||||
/// ```
|
/// ```
|
||||||
|
/// # use lang_evolve_core::settings::Settings;
|
||||||
/// use std::path::Path;
|
/// use std::path::Path;
|
||||||
/// let s = lang_evolve_core::settings::Settings::new();
|
///
|
||||||
|
/// let s = Settings::new();
|
||||||
///
|
///
|
||||||
/// // Export to JSON
|
/// // Export to JSON
|
||||||
/// let path_json = Path::new("./output.json");
|
/// let path_json = Path::new("settings.json");
|
||||||
/// s.export(&path_json).unwrap();
|
/// s.export(&path_json).unwrap();
|
||||||
///
|
///
|
||||||
/// // Export to Yaml, both ".yml" and ".yaml" work
|
/// // Export to Yaml, both ".yml" and ".yaml" work
|
||||||
/// let path_yaml = Path::new("./output.yaml");
|
/// let path_yaml = Path::new("settings.yaml");
|
||||||
/// s.export(&path_yaml).unwrap();
|
/// s.export(&path_yaml).unwrap();
|
||||||
/// let path_yml = Path::new("./output.yml");
|
/// let path_yml = Path::new("settings.yml");
|
||||||
/// s.export(&path_yml).unwrap();
|
/// s.export(&path_yml).unwrap();
|
||||||
/// ```
|
/// ```
|
||||||
///
|
///
|
||||||
@@ -282,6 +340,91 @@ impl Settings {
|
|||||||
fn get_ruleset_version() -> String {
|
fn get_ruleset_version() -> String {
|
||||||
RULESET_CURRENT_VERSION.to_string()
|
RULESET_CURRENT_VERSION.to_string()
|
||||||
}
|
}
|
||||||
|
|
||||||
|
/// Transform input rules into Regexes that can be understood by Rust.
|
||||||
|
fn update_rules(&self) -> std::result::Result<Vec<Rule>, String> {
|
||||||
|
let rules = self.rules.clone();
|
||||||
|
let rules: Vec<Rule> = rules
|
||||||
|
.iter()
|
||||||
|
.map(|rule| rule.update(&self.categories).unwrap())
|
||||||
|
.collect();
|
||||||
|
Ok(rules)
|
||||||
|
}
|
||||||
|
|
||||||
|
/// Apply list of rules to input
|
||||||
|
///
|
||||||
|
/// The list of rules in the struct will be applied to the input `new`. 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
|
||||||
|
///
|
||||||
|
/// * `new` - Input to modify
|
||||||
|
///
|
||||||
|
/// # Example
|
||||||
|
///
|
||||||
|
/// ```
|
||||||
|
/// # use lang_evolve_core::settings::Settings;
|
||||||
|
/// let settings = Settings::new();
|
||||||
|
/// // add some rules...
|
||||||
|
/// // set some input
|
||||||
|
/// let input = String::new();
|
||||||
|
/// 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;
|
||||||
|
debug!("===============================================");
|
||||||
|
for rule in rules {
|
||||||
|
debug!(
|
||||||
|
"from: \"{}\"\tto: \"{}\"",
|
||||||
|
rule.get_from().to_string(),
|
||||||
|
rule.get_to()
|
||||||
|
);
|
||||||
|
let old = s.clone();
|
||||||
|
let new = rule
|
||||||
|
.get_from()
|
||||||
|
.replace_all(&s, rule.get_to().as_str())
|
||||||
|
.to_string();
|
||||||
|
if cfg!(debug_assertions) {
|
||||||
|
let diffs = diff_words(&old, &new);
|
||||||
|
if diffs.diff().len() > 1 {
|
||||||
|
debug!("diff:\n{}", diff_words(&old, &new));
|
||||||
|
} else {
|
||||||
|
debug!("diff: No changes");
|
||||||
|
}
|
||||||
|
}
|
||||||
|
s = new;
|
||||||
|
}
|
||||||
|
Ok(s)
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
impl Default for Settings {
|
||||||
|
/// Creates a new empty instance of [`Settings`]
|
||||||
|
///
|
||||||
|
/// # Example
|
||||||
|
///
|
||||||
|
/// ```
|
||||||
|
/// let s = lang_evolve_core::settings::Settings::default();
|
||||||
|
/// 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
|
||||||
|
fn default() -> Self {
|
||||||
|
Self::new()
|
||||||
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
use std::str::FromStr;
|
use std::str::FromStr;
|
||||||
@@ -300,7 +443,7 @@ impl FromStr for Settings {
|
|||||||
///
|
///
|
||||||
/// ```
|
/// ```
|
||||||
/// # use std::str::FromStr;
|
/// # use std::str::FromStr;
|
||||||
/// let s = r#"{"version":"1","categories":[],"rules":[]}"#;
|
/// let s = r#"{"version":"1","categories":{},"rules":[]}"#;
|
||||||
/// let settings = lang_evolve_core::settings::Settings::from_str(s).unwrap();
|
/// let settings = lang_evolve_core::settings::Settings::from_str(s).unwrap();
|
||||||
/// ```
|
/// ```
|
||||||
fn from_str(s: &str) -> Result<Self, Self::Err> {
|
fn from_str(s: &str) -> Result<Self, Self::Err> {
|
||||||
@@ -310,7 +453,7 @@ impl FromStr for Settings {
|
|||||||
Ok(val) => Ok(val),
|
Ok(val) => Ok(val),
|
||||||
Err(e) => {
|
Err(e) => {
|
||||||
error!("Could not decode input {}: {}", s, e.to_string());
|
error!("Could not decode input {}: {}", s, e.to_string());
|
||||||
return Err(e);
|
Err(e)
|
||||||
}
|
}
|
||||||
},
|
},
|
||||||
}
|
}
|
||||||
@@ -334,24 +477,3 @@ impl PartialEq for Settings {
|
|||||||
}
|
}
|
||||||
|
|
||||||
impl Eq for Settings {}
|
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);
|
|
||||||
}
|
|
||||||
|
|||||||
179
src/settings/rule/mod.rs
Normal file
179
src/settings/rule/mod.rs
Normal file
@@ -0,0 +1,179 @@
|
|||||||
|
use std::collections::HashMap;
|
||||||
|
|
||||||
|
use lazy_static::lazy_static;
|
||||||
|
use serde::{Deserialize, Serialize};
|
||||||
|
|
||||||
|
mod regex_wrapper;
|
||||||
|
use regex_wrapper::Regex;
|
||||||
|
|
||||||
|
lazy_static! {
|
||||||
|
static ref RE: Regex = Regex::new("%([A-Z])");
|
||||||
|
}
|
||||||
|
|
||||||
|
/// Representation of a rule in LangEvolveRs
|
||||||
|
#[derive(Debug, Serialize, Deserialize, Clone)]
|
||||||
|
pub struct Rule {
|
||||||
|
/// Regex that should match the input text
|
||||||
|
from: Regex,
|
||||||
|
/// Text to replace matched text
|
||||||
|
to: String,
|
||||||
|
}
|
||||||
|
|
||||||
|
impl Rule {
|
||||||
|
/// Create new rule
|
||||||
|
///
|
||||||
|
/// # Arguments
|
||||||
|
///
|
||||||
|
/// * `from` - literal string that represents the regex that should match
|
||||||
|
/// the input text
|
||||||
|
/// * `to` - literal string that represents the regex text that should
|
||||||
|
/// replaced the text matched by `from`
|
||||||
|
///
|
||||||
|
/// # Example
|
||||||
|
/// ```
|
||||||
|
/// # use lazy_static::lazy_static;
|
||||||
|
/// # #[path = "mod.rs"]
|
||||||
|
/// # mod rule;
|
||||||
|
/// # use rule::Rule;
|
||||||
|
/// let rule = Rule::new("ab+c*", "ab");
|
||||||
|
/// ```
|
||||||
|
pub fn new(from: &str, to: &str) -> Self {
|
||||||
|
Rule {
|
||||||
|
from: Regex::new(from),
|
||||||
|
to: String::from(to),
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
/// Detect the number of categories in a rule
|
||||||
|
///
|
||||||
|
/// For a rule, this function detects the number of categories set in the
|
||||||
|
/// `from` member of a `Rule` and in its `to` member. The result is returned
|
||||||
|
/// as a tuple of `u8`s.
|
||||||
|
///
|
||||||
|
/// # Example
|
||||||
|
///
|
||||||
|
/// ```
|
||||||
|
/// # #[path = "mod.rs"]
|
||||||
|
/// # mod rule;
|
||||||
|
/// # use rule::Rule;
|
||||||
|
/// let rule = Rule::new("%Bea*i+", "a%A%C");
|
||||||
|
/// let nb_rules = rule.detect_number_categories();
|
||||||
|
/// assert_eq!((1 as u8, 2 as u8), nb_rules);
|
||||||
|
/// ```
|
||||||
|
pub fn detect_number_categories(&self) -> (u8, u8) {
|
||||||
|
let captures_from = self.from.to_string().matches('%').count() as u8;
|
||||||
|
let captures_to = self.to.matches('%').count() as u8;
|
||||||
|
(captures_from, captures_to)
|
||||||
|
}
|
||||||
|
|
||||||
|
fn simple_rewrite(&self, categories: &HashMap<String, String>) -> Self {
|
||||||
|
let mut rule = self.clone();
|
||||||
|
for (category, content) in categories {
|
||||||
|
rule.from = Regex::new(
|
||||||
|
rule.from
|
||||||
|
.to_string()
|
||||||
|
.replace(
|
||||||
|
format!("%{}", category).as_str(),
|
||||||
|
format!("[{}]", content).as_str(),
|
||||||
|
)
|
||||||
|
.as_str(),
|
||||||
|
);
|
||||||
|
}
|
||||||
|
rule
|
||||||
|
}
|
||||||
|
|
||||||
|
// TODO break categories in different rules
|
||||||
|
pub fn update(
|
||||||
|
&self,
|
||||||
|
categories: &HashMap<String, String>,
|
||||||
|
) -> std::result::Result<Rule, String> {
|
||||||
|
let mut rule = self.clone();
|
||||||
|
let (from_match, to_match) = self.detect_number_categories();
|
||||||
|
// If there are only simple rewrites to make in the from String
|
||||||
|
if from_match > 0 && to_match == 0 {
|
||||||
|
rule = self.simple_rewrite(&categories);
|
||||||
|
}
|
||||||
|
|
||||||
|
// If there are equivalences between from and to
|
||||||
|
if from_match > 0 && to_match <= from_match && to_match > 0 {}
|
||||||
|
|
||||||
|
Ok(rule)
|
||||||
|
}
|
||||||
|
|
||||||
|
pub fn get_from(&self) -> &Regex {
|
||||||
|
&self.from
|
||||||
|
}
|
||||||
|
|
||||||
|
pub fn get_to(&self) -> String {
|
||||||
|
self.to.clone()
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
impl From<&str> for Rule {
|
||||||
|
/// Allow to create a rule from a single literal string
|
||||||
|
///
|
||||||
|
/// It is possible to create a rule from a string, delimited by a `>`. This
|
||||||
|
/// means a rule like `%C>%D` will be interpreted as going from `%C` to
|
||||||
|
/// `%D`.
|
||||||
|
///
|
||||||
|
/// # Example
|
||||||
|
///
|
||||||
|
/// ```
|
||||||
|
/// # #[path = "mod.rs"]
|
||||||
|
/// # mod rule;
|
||||||
|
/// # use rule::Rule;
|
||||||
|
/// let rule1 = Rule::new("%C", "%D");
|
||||||
|
/// let rule2 = Rule::from("%C>%D");
|
||||||
|
/// assert_eq!(rule1, rule2);
|
||||||
|
/// ```
|
||||||
|
fn from(source: &str) -> Self {
|
||||||
|
let components: Vec<&str> = source.split_terminator('>').collect();
|
||||||
|
Rule::new(components[0], components[1])
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
impl From<String> for Rule {
|
||||||
|
/// Allow to create a rule from a single `String`
|
||||||
|
///
|
||||||
|
/// It is possible to create a rule from a string, delimited by a `>`. This
|
||||||
|
/// means a rule like `%C>%D` will be interpreted as going from `%C` to
|
||||||
|
/// `%D`.
|
||||||
|
///
|
||||||
|
/// # Example
|
||||||
|
///
|
||||||
|
/// ```
|
||||||
|
/// # #[path = "mod.rs"]
|
||||||
|
/// # mod rule;
|
||||||
|
/// # use rule::Rule;
|
||||||
|
/// let rule1 = Rule::new("%C", "%D");
|
||||||
|
/// let rule2 = Rule::from(String::from("%C>%D"));
|
||||||
|
/// assert_eq!(rule1, rule2);
|
||||||
|
/// ```
|
||||||
|
fn from(source: String) -> Self {
|
||||||
|
let components: Vec<&str> = source.split_terminator('>').collect();
|
||||||
|
Rule::new(components[0], components[1])
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
impl PartialEq for Rule {
|
||||||
|
/// Equality between `Rule` structs
|
||||||
|
///
|
||||||
|
/// This allows for equality comparison between two `Rule` structs.
|
||||||
|
///
|
||||||
|
/// # Example
|
||||||
|
///
|
||||||
|
/// ```
|
||||||
|
/// # #[path = "mod.rs"]
|
||||||
|
/// # mod rule;
|
||||||
|
/// use rule::Rule;
|
||||||
|
/// let rule1 = Rule::new("%C", "%D");
|
||||||
|
/// let rule2 = Rule::from("%C>%D");
|
||||||
|
/// assert!(rule1 == rule2);
|
||||||
|
/// assert!(rule2 == rule1);
|
||||||
|
/// ```
|
||||||
|
fn eq(&self, other: &Self) -> bool {
|
||||||
|
self.from == other.from && self.to == other.to
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
impl Eq for Rule {}
|
||||||
@@ -1,36 +1,38 @@
|
|||||||
// extern crate serde;
|
|
||||||
// extern crate regex;
|
|
||||||
// use serde::{Deserialize, Serialize};
|
|
||||||
|
|
||||||
// #[derive(Clone, Debug, Deserialize, Serialize)]
|
|
||||||
// #[serde(transparent)]
|
|
||||||
// pub struct Regex(regex::Regex);
|
|
||||||
|
|
||||||
// #[derive(Clone, Debug, Deserialize, Serialize)]
|
|
||||||
// #[serde(remote = "regex::Regex")]
|
|
||||||
// pub struct RegexDef{
|
|
||||||
// #[serde(getter = "regex::Regex::to_string")]
|
|
||||||
// r: String
|
|
||||||
// }
|
|
||||||
use std::{fmt, ops};
|
use std::{fmt, ops};
|
||||||
|
|
||||||
#[derive(Clone, Debug)]
|
#[derive(Clone, Debug)]
|
||||||
pub struct Regex(regex::Regex);
|
pub struct Regex(regex::Regex);
|
||||||
|
|
||||||
impl Regex {
|
impl Regex {
|
||||||
|
/// Create a new Regex wrapper around regex::Regex;
|
||||||
|
///
|
||||||
|
/// # Arguments
|
||||||
|
///
|
||||||
|
/// * `s` - string litteral from which to create the new Regex
|
||||||
pub fn new(s: &str) -> Self {
|
pub fn new(s: &str) -> Self {
|
||||||
Self(regex::Regex::new(s).unwrap())
|
Self(regex::Regex::new(s).unwrap())
|
||||||
}
|
}
|
||||||
|
|
||||||
|
/// Returns a string literal representation of the Regex
|
||||||
|
#[allow(unused)]
|
||||||
pub fn as_str(&self) -> &str {
|
pub fn as_str(&self) -> &str {
|
||||||
self.0.as_str()
|
self.0.as_str()
|
||||||
}
|
}
|
||||||
|
}
|
||||||
|
|
||||||
pub fn to_string(&self) -> String {
|
impl ToString for Regex {
|
||||||
|
fn to_string(&self) -> String {
|
||||||
self.0.to_string()
|
self.0.to_string()
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
use std::hash::{Hash, Hasher};
|
||||||
|
impl Hash for Regex {
|
||||||
|
fn hash<H: Hasher>(&self, state: &mut H) {
|
||||||
|
self.0.as_str().hash(state);
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
impl ops::Deref for Regex {
|
impl ops::Deref for Regex {
|
||||||
type Target = regex::Regex;
|
type Target = regex::Regex;
|
||||||
fn deref(&self) -> ®ex::Regex {
|
fn deref(&self) -> ®ex::Regex {
|
||||||
@@ -1,4 +1,3 @@
|
|||||||
extern crate log;
|
|
||||||
use log::{info, error};
|
use log::{info, error};
|
||||||
|
|
||||||
use std::fs::File;
|
use std::fs::File;
|
||||||
@@ -24,8 +23,9 @@ pub enum SettingsType {
|
|||||||
/// # Example
|
/// # Example
|
||||||
///
|
///
|
||||||
/// ```no_run
|
/// ```no_run
|
||||||
|
/// # use lang_evolve_core::utils;
|
||||||
/// let path = std::path::Path::new("./some/path/to/my/file.json");
|
/// let path = std::path::Path::new("./some/path/to/my/file.json");
|
||||||
/// let content = lang_evolve_core::settings::utils::read_file(&path).unwrap();
|
/// let content = utils::read_file(&path).unwrap();
|
||||||
/// ```
|
/// ```
|
||||||
pub fn read_file(path: &Path) -> Result<String> {
|
pub fn read_file(path: &Path) -> Result<String> {
|
||||||
let display = path.display();
|
let display = path.display();
|
||||||
@@ -40,7 +40,7 @@ pub fn read_file(path: &Path) -> Result<String> {
|
|||||||
match file.read_to_string(&mut content) {
|
match file.read_to_string(&mut content) {
|
||||||
Err(why) => {
|
Err(why) => {
|
||||||
error!("Could not read {}: {}", display, why.to_string());
|
error!("Could not read {}: {}", display, why.to_string());
|
||||||
return Err(why);
|
Err(why)
|
||||||
}
|
}
|
||||||
Ok(_) => {
|
Ok(_) => {
|
||||||
info!("Content of {} read", display);
|
info!("Content of {} read", display);
|
||||||
@@ -54,9 +54,10 @@ pub fn read_file(path: &Path) -> Result<String> {
|
|||||||
/// # Example
|
/// # Example
|
||||||
///
|
///
|
||||||
/// ```no_run
|
/// ```no_run
|
||||||
|
/// # use lang_evolve_core::utils;
|
||||||
/// let content = String::from("This is my content");
|
/// let content = String::from("This is my content");
|
||||||
/// let path = std::path::Path::new("./path/to/my/file.txt");
|
/// let path = std::path::Path::new("./path/to/my/file.txt");
|
||||||
/// lang_evolve_core::settings::utils::write_file(&path, &content).unwrap();
|
/// utils::write_file(&path, &content).unwrap();
|
||||||
/// ```
|
/// ```
|
||||||
pub fn write_file<S>(path: &Path, content: &S) -> Result<()>
|
pub fn write_file<S>(path: &Path, content: &S) -> Result<()>
|
||||||
where
|
where
|
||||||
@@ -99,17 +100,18 @@ where
|
|||||||
/// # Example
|
/// # Example
|
||||||
///
|
///
|
||||||
/// ```
|
/// ```
|
||||||
|
/// # use lang_evolve_core::utils;
|
||||||
/// let file_json = std::path::Path::new("file.json");
|
/// let file_json = std::path::Path::new("file.json");
|
||||||
/// assert_eq!(lang_evolve_core::settings::utils::SettingsType::Json,
|
/// assert_eq!(utils::SettingsType::Json,
|
||||||
/// lang_evolve_core::settings::utils::get_file_type(&file_json));
|
/// utils::get_file_type(&file_json));
|
||||||
///
|
///
|
||||||
/// let file_yaml = std::path::Path::new("file.yaml");
|
/// let file_yaml = std::path::Path::new("file.yaml");
|
||||||
/// assert_eq!(lang_evolve_core::settings::utils::SettingsType::Yaml,
|
/// assert_eq!(utils::SettingsType::Yaml,
|
||||||
/// lang_evolve_core::settings::utils::get_file_type(&file_yaml));
|
/// utils::get_file_type(&file_yaml));
|
||||||
///
|
///
|
||||||
/// let file_yml = std::path::Path::new("file.yml");
|
/// let file_yml = std::path::Path::new("file.yml");
|
||||||
/// assert_eq!(lang_evolve_core::settings::utils::SettingsType::Yaml,
|
/// assert_eq!(utils::SettingsType::Yaml,
|
||||||
/// lang_evolve_core::settings::utils::get_file_type(&file_yml));
|
/// utils::get_file_type(&file_yml));
|
||||||
/// ```
|
/// ```
|
||||||
pub fn get_file_type(path: &Path) -> SettingsType {
|
pub fn get_file_type(path: &Path) -> SettingsType {
|
||||||
let extension = match path.extension() {
|
let extension = match path.extension() {
|
||||||
Reference in New Issue
Block a user