refactor out reverse_dict

src/main.rs

@@ -16,4 +16,7 @@ fn main() {
     let (double_dict, triple_dict, all_token_list) = packages::parser::parse_raw(args.raw);
     packages::parser::print_dict(triple_dict);
 
+    // let's say that the cutoff is 72 for Linux2k.log.
+    // check all of the 2-grams from the less-frequent 3-grams
+
 }

src/packages/parser.rs

@@ -243,9 +243,10 @@ fn test_parse_raw() {
     assert_eq!(triple_dict, triple_dict_oracle);
 }
 
-pub fn print_dict(d: HashMap<String, i32>) {
-    let mut reverse_d : HashMap<i32, Vec<String>> = HashMap::new();
-    let mut key_set = BTreeSet::new();
+/// standard mapreduce invert map: given {<k1, v1>, <k2, v2>, <k3, v1>}, returns ([v1, v2] (sorted), {<v1, [k1, k3]>, <v2, [k2]>})
+pub fn reverse_dict(d: HashMap<String, i32>) -> (BTreeSet<i32>, HashMap<i32, Vec<String>>) {
+    let mut reverse_d: HashMap<i32, Vec<String>> = HashMap::new();
+    let mut val_set: BTreeSet<i32> = BTreeSet::new();
 
     for (key, val) in d.iter() {
         if reverse_d.contains_key(val) {
@@ -253,13 +254,18 @@ pub fn print_dict(d: HashMap<String, i32>) {
             existing_keys.push(key.to_string());
         } else {
             reverse_d.insert(*val, vec![key.to_string()]);
-            key_set.insert(val);
+            val_set.insert(*val);
         }
     }
+    return (val_set, reverse_d);
+}
+
+pub fn print_dict(d: HashMap<String, i32>) {
+    let (val_set, reverse_d) = reverse_dict(d);
 
     println!("printing dict");
-    for key in &key_set {
-        println!("{}: {:?}", key, reverse_d.get(key).unwrap());
+    for val in &val_set {
+        println!("{}: {:?}", val, reverse_d.get(val).unwrap());
     }
     println!("---");
 }