From 149e0b6ae9871515be21f23b492f5ef7355e2ca4 Mon Sep 17 00:00:00 2001
From: Nathan Reiner <nathan@nathanreiner.xyz>
Date: Fri, 14 Jul 2023 00:22:39 +0200
Subject: make fast using hash instead of dictionary

---
 src/dictionary.rs | 105 ------------------------------------------------------
 1 file changed, 105 deletions(-)
 delete mode 100644 src/dictionary.rs

(limited to 'src/dictionary.rs')

diff --git a/src/dictionary.rs b/src/dictionary.rs
deleted file mode 100644
index 3e05b91..0000000
--- a/src/dictionary.rs
+++ /dev/null
@@ -1,105 +0,0 @@
-use std::collections::HashMap;
-use crate::vector::FileVector;
-
-/// The dictionary is used to cache to words ids.
-/// It also provides a function to convert it to
-/// a vector and generate a FileVector from a word list
-/// with the current directory.
-#[derive(Clone, Debug)]
-pub struct Dictionary {
-    last_index : usize,
-    data : HashMap<String, u64>,
-}
-
-impl Default for Dictionary {
-    fn default() -> Self {
-        Self::new()
-    }
-}
-
-impl Dictionary {
-    pub fn new() -> Self {
-        Self { last_index : 0, data : HashMap::new() }
-    }
-
-    pub fn from_line(line : &str) -> Self {
-        let mut data : HashMap<String, u64> = HashMap::new();
-        let mut i : usize = 0;
-
-        for word in line.split(',') {
-            data.insert(word.to_string(), i as u64);
-            i += 1;
-        }
-
-        Self { last_index : i - 1, data }
-    }
-
-    pub fn set(&mut self, name : &String) {
-        if !self.data.contains_key(name) {
-            self.last_index += 1;
-            self.data.insert(name.clone(), self.last_index as u64);
-        }
-    }
-
-    pub fn set_and_get(&mut self, name : &String) -> u64 {
-        if !self.data.contains_key(name) {
-            self.last_index += 1;
-            self.data.insert(name.clone(), self.last_index as u64);
-            self.last_index as u64
-        } else {
-            *self.data.get(name).unwrap()
-        }
-    }
-
-    pub fn get(&self, name : &String) -> Option<&u64> {
-        self.data.get(name)
-    }
-
-    pub fn iter(&self) -> std::collections::hash_map::Iter<'_, String, u64> {
-        self.data.iter()
-    }
-
-    pub fn to_list(&self) -> Vec<String> {
-        let mut v = Vec::with_capacity(self.last_index + 1);
-
-        v.resize(self.last_index + 1, "".to_string());
-
-        for (word, id) in self.iter() {
-            v[(*id) as usize] = word.clone();
-        }
-
-        v
-    }
-
-    pub fn vectorize_word_list(&self, words : &Vec<&String>) -> FileVector {
-        let mut fv = FileVector::new();
-
-        for word in words {
-            let i = *self.get(word).unwrap();
-            if !fv.contains_key(&i) {
-                fv.insert(i, 1);
-            } else {
-                let c = *fv.get(&i).unwrap();
-                fv.insert(i, c + 1);
-            }
-        }
-
-        fv
-    }
-
-    pub fn insert_words_and_vectorize_word_list(&mut self, words : &Vec<&String>) -> FileVector {
-        let mut fv = FileVector::new();
-
-        for word in words {
-            let i = self.set_and_get(word);
-            if !fv.contains_key(&i) {
-                fv.insert(i, 1);
-            } else {
-                let c = *fv.get(&i).unwrap();
-                fv.insert(i, c + 1);
-            }
-        }
-
-        fv
-    }
-}
-- 
cgit v1.2.3-70-g09d2