first sketch of indexer

1 files changed, 69 insertions, 0 deletions

diff --git a/src/dictionary.rs b/src/dictionary.rs
new file mode 100644
index 0000000..a8d9b28
--- /dev/null
+++ b/src/dictionary.rs
@@ -0,0 +1,69 @@
+use std::collections::HashMap;
+use crate::vector::FileVector;
+
+pub struct Dictionary {
+    last_index : usize,
+    data : HashMap<String, u64>,
+}
+
+
+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, self.last_index as u64);
+        }
+    }
+
+    pub fn get(&self, name : String) -> u64 {
+        *self.data.get(&name).unwrap()
+    }
+
+    pub fn iter(&self) -> &HashMap<String, u64> {
+        &self.data
+    }
+
+    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);
+            if !fv.contains_key(&i) {
+                fv.insert(i, 1);
+            } else {
+                let c : u64 = *fv.get(&i).unwrap();
+                fv.insert(i, c + 1);
+            }
+        }
+
+        fv
+    }
+}