bug fixes

src/database.rs

@@ -1,6 +1,6 @@
 use anyhow::Result;
 
-#[derive(Debug)]
+#[derive(Debug, Clone)]
 pub struct File {
     pub path: String,
     pub hash: String,
@@ -41,7 +41,7 @@ pub fn indexed_paths(connection: &sqlite::Connection) -> Result<Vec<File>> {
     Ok(result)
 }
 
-pub fn duplicate_hashes(connection: &sqlite::Connection) -> Result<Vec<File>> {
+pub fn duplicate_hashes(connection: &sqlite::Connection, path: &String) -> Result<Vec<File>> {
     let query = format!(
         " 
             SELECT a.* FROM files a
@@ -50,8 +50,9 @@ pub fn duplicate_hashes(connection: &sqlite::Connection) -> Result<Vec<File>> {
             GROUP BY hash
             HAVING count(*) > 1 ) b
             ON a.hash = b.hash
+            WHERE a.file_identifier LIKE \"{}%\"
             ORDER BY a.file_identifier
-        "
+        ", path
     );
     let result: Vec<File> = connection
         .prepare(query)?

src/output.rs

@@ -1,10 +1,9 @@
 use crate::database::File;
-use itertools::Itertools;
-use colored::Colorize;
-use std::fs;
 use chrono::offset::Utc;
 use chrono::DateTime;
+use colored::Colorize;
 use humansize::{format_size, DECIMAL};
+use std::{collections::HashMap, fs};
 
 fn format_path(path: &String) -> String {
     let stringlen = path.len();
@@ -36,22 +35,25 @@ pub fn print(duplicates: Vec<File>) {
     );
     print_divider();
 
-    duplicates
-        .into_iter()
-        .group_by(|record| record.hash.clone())
-        .into_iter()
-        .for_each(|(_, group)| {
-            group
-                .into_iter()
-                .for_each(|file| {
-                    println!(
-                        "| {0: <16} | {1: <35} | {2: <16} | {3: <32} |",
-                        &file.hash[0..16].red(),
-                        format_path(&file.path).yellow(),
-                        file_size(&file.path).blue(),
-                        modified_time(&file.path).blue()
-                    );
-                });
-            print_divider();
+    let mut dup_index: HashMap<String, Vec<File>> = HashMap::new();
+
+    duplicates.into_iter().for_each(|file| {
+        dup_index
+            .entry(file.hash.clone())
+            .and_modify(|value| value.push(file.clone()))
+            .or_insert(vec![file]);
+    });
+
+    dup_index.into_iter().for_each(|(_, group)| {
+        group.into_iter().for_each(|file| {
+            println!(
+                "| {0: <16} | {1: <35} | {2: <16} | {3: <32} |",
+                &file.hash[0..16].red(),
+                format_path(&file.path).yellow(),
+                file_size(&file.path).blue(),
+                modified_time(&file.path).blue()
+            );
         });
+        print_divider();
+    });
 }

src/scanner.rs

@@ -1,5 +1,3 @@
-use std::path::PathBuf;
-
 use crate::database;
 use crate::{cli::App, database::File};
 use anyhow::Result;
@@ -7,11 +5,14 @@ use glob::glob;
 use itertools::Itertools;
 use rayon::prelude::*;
 use std::fs;
+use std::path::PathBuf;
 
 pub fn duplicates(app_opts: App, connection: &sqlite::Connection) -> Result<Vec<File>> {
-    let scan_results = scan(app_opts, connection)?;
+    let scan_results = scan(&app_opts, connection)?;
+    let base_path = get_directory(&app_opts)?;
+
     index_files(scan_results, connection);
-    database::duplicate_hashes(connection)
+    database::duplicate_hashes(connection, &base_path)
 }
 
 fn get_directory(opts: &App) -> Result<String> {
@@ -55,8 +56,8 @@ fn is_indexed_file(path: &String, indexed: &Vec<File>) -> bool {
         .contains(path)
 }
 
-fn scan(app_opts: App, connection: &sqlite::Connection) -> Result<Vec<String>> {
-    let directory = get_directory(&app_opts)?;
+fn scan(app_opts: &App, connection: &sqlite::Connection) -> Result<Vec<String>> {
+    let directory = get_directory(app_opts)?;
     let glob_patterns: Vec<PathBuf> = get_glob_patterns(&app_opts, &directory);
     let indexed_paths = database::indexed_paths(connection)?;
     let files: Vec<String> = glob_patterns