From ef8c02fc97b910b66726308be879fbc2335425bc Mon Sep 17 00:00:00 2001
From: Lucien Cartier-Tilet <lucien@phundrak.com>
Date: Sat, 4 Apr 2020 15:46:29 +0200
Subject: [PATCH] Good progress with applying rules, room for improvement

Rules are applied, however some bugs remain:
- The HashMap for rules should be replaced by a Vec so they can be
  stored in order and not randomly
- For some reasons, the `%` is not removed from some rules in the
  private function `update_rules` in the `Settings` struct.
- Make it so replacements between square brackets work correctly
---
 src/lib.rs          |  6 +++-
 src/settings/mod.rs | 83 ++++++++++++++++++++++++++++++++++++++++-----
 2 files changed, 80 insertions(+), 9 deletions(-)

diff --git a/src/lib.rs b/src/lib.rs
index decf7c9..c6525cb 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -18,7 +18,6 @@ extern crate simplelog;
 use log::{info, warn};
 use simplelog::*;
 
-
 pub mod settings;
 
 /// Initializes the crate
@@ -48,6 +47,11 @@ pub fn init() -> std::result::Result<(), log::SetLoggerError> {
             TerminalMode::Mixed,
         )
         .unwrap(),
+        WriteLogger::new(
+            LevelFilter::Debug,
+            Config::default(),
+            File::create("core.log").unwrap(),
+        ),
         WriteLogger::new(
             LevelFilter::Info,
             Config::default(),
diff --git a/src/settings/mod.rs b/src/settings/mod.rs
index 2284d9c..577e64a 100644
--- a/src/settings/mod.rs
+++ b/src/settings/mod.rs
@@ -4,7 +4,7 @@ extern crate serde_yaml;
 use serde::{Deserialize, Serialize};
 
 extern crate log;
-use log::{error, info};
+use log::{debug, error, info};
 
 pub mod utils;
 use utils::SettingsType;
@@ -125,7 +125,7 @@ pub struct Settings {
     /// a regex to be matched while the second represents the change
     /// to be made to the input data.
     #[serde(default)]
-    pub rules: HashMap<Regex, Regex>,
+    pub rules: HashMap<Regex, String>,
 }
 
 /// Representation inside the crate of LangEvolve’s settings.
@@ -138,9 +138,9 @@ impl Settings {
     /// let s = lang_evolve_core::settings::Settings::new();
     /// let content_yaml = r#"---
     /// version: "1"
-    /// categories: []
-    /// rules: []"#;
-    /// let content_json = r#"{"version":"1","categories":[],"rules":[]}"#;
+    /// 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());
     /// ```
@@ -283,6 +283,73 @@ impl Settings {
     fn get_ruleset_version() -> String {
         RULESET_CURRENT_VERSION.to_string()
     }
+
+    fn update_rules(&self) -> Vec<(Regex, String)> {
+        let mut rules: Vec<(Regex, String)> = Vec::new();
+        for (from, to) in &self.rules {
+            rules.push((from.clone(), to.to_string()));
+        }
+
+        for (category, content) in &self.categories {
+            let mut temp_rules: Vec<(Regex, String)> = Vec::new();
+            for (from, to) in &rules {
+                let from = Regex::new(
+                    from.to_string()
+                        .replace(
+                            category.as_str(),
+                            format!("[{}]", content).as_str(),
+                        )
+                        .replace("%", "")
+                        .as_str(),
+                );
+                let to = to.to_string().replace(
+                    category,
+                    format!("[{}]", content).replace("%", "").as_str(),
+                );
+                temp_rules.push((from, to));
+            }
+            rules = temp_rules.clone();
+        }
+        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) -> String {
+        // Replace all `%C`s by their equivalent
+        let rules = self.update_rules();
+        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);
+        }
+        s
+    }
 }
 
 use std::str::FromStr;
@@ -301,7 +368,7 @@ impl FromStr for Settings {
     ///
     /// ```
     /// # 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();
     /// ```
     fn from_str(s: &str) -> Result<Self, Self::Err> {
@@ -342,8 +409,8 @@ fn write_settings() {
     let path = std::path::Path::new("test.yaml");
     let settings = r#"---
 version: "1"
-categories: []
-rules: []"#;
+categories: {}
+rules: {}"#;
     utils::write_file(&path, &serde_yaml::to_string(&s).unwrap()).unwrap();
     assert_eq!(settings, utils::read_file(&path).unwrap());
 }