GetFileSizeGroup

PowerShellで作成したCSVファイルを元に
ファイル名が近いものを探してグループ化して
合計サイズを降順で表示するプログラムです。

まだ、結果がいまいちかもしれません。
もう少し改良できそうです。

2017-07-25:
コマンドライン引数の取り込みを修正。
(ワイルドカード(*)が指定されたら展開する)
時間の掛かる比較処理を並列化。

	/**
	* Copyright (c) 2017 Yuki Ono
	* Licensed under the MIT License.
	*/
	using System;
	using System.Collections.Generic;
	using System.Linq;
	using System.Text;
	using System.Text.RegularExpressions;
	using System.Threading.Tasks;
	using System.IO;
	using Microsoft.VisualBasic.FileIO;
	namespace GetFileSizeGroup {
	class Program {
	// https://discuss.leetcode.com/topic/17639/20ms-detailed-explained-c-solutions-o-n-space
	static int MinDistance(string w1, string w2) {
	var M = w1.Length; var N = w2.Length;
	var cur = Enumerable.Range(0, M + 1).ToArray();
	foreach (var j in Enumerable.Range(1, N)) {
	var pre = cur.First();
	cur[0] = j;
	foreach (var i in Enumerable.Range(1, M)) {
	var temp = cur[i];
	cur[i] = (w1[i - 1] == w2[j - 1]) ? pre
	: Math.Min(pre + 1, Math.Min(cur[i] + 1, cur[i - 1] + 1));
	pre = temp;
	}
	}
	return cur.Last();
	}
	static IEnumerable<string[]> ReadCSV(string csvPath) {
	using (var tfp = new TextFieldParser(csvPath)) {
	tfp.CommentTokens = new string[] { "#" };
	tfp.SetDelimiters(new string[] { "," });
	tfp.HasFieldsEnclosedInQuotes = true;
	tfp.ReadLine();
	tfp.ReadLine();
	while (!tfp.EndOfData)
	yield return tfp.ReadFields();
	}
	}
	static string GetSizeMB(long size) =>
	(size / 1024 / 1024) + "MB";
	static long GetGroupSize(List<(string, long)> fileList) =>
	fileList.Select(f => f.Item2).Sum();
	static readonly Regex SPACE_REG = new Regex(@"[_-]", RegexOptions.Compiled);
	// without file extension
	static string GetCompName(string path) {
	var name = Path.GetFileNameWithoutExtension(path);
	return SPACE_REG.Replace(name.ToLower(), " ");
	}
	// https://stackoverflow.com/questions/381366/is-there-a-wildcard-expansion-option-for-net-apps
	static string[] ExpandPath(string path) {
	var dir = Path.GetDirectoryName(path);
	dir = String.IsNullOrEmpty(dir) ? "./" : dir;
	var file = Path.GetFileName(path);
	return Directory.GetFiles(dir, file);
	}
	static string[] GetCSVFiles(string[] args) {
	return args.SelectMany(ExpandPath)
	.Where(p => p.EndsWith(".csv")).ToArray();
	}
	static bool CompareName(string[] x, string[] y) {
	var w1 = GetCompName(x[0]);
	var w2 = GetCompName(y[0]);
	return (MinDistance(w1, w2) < 5);
	}
	static void Main(string[] args) {
	var csvFiles = GetCSVFiles(args);
	if (csvFiles.Length == 0) {
	Console.WriteLine("Usage:");
	Console.WriteLine(@"Get-ChildItem C:\ -rec | where {!$_.PSIsContainer} | select-object FullName, Length | export-csv -encoding utf8 -path C:\file-list.csv");
	Console.WriteLine("GetFileSizeGroup *.csv > fsg-list.txt");
	return;
	}
	LinkedList<string[]> fileList = new LinkedList<string[]>(csvFiles.SelectMany(fn => ReadCSV(fn)));
	var fileGroups = new List<(List<(string, long)>, long)>();
	while (fileList.Count != 0) {
	var x = fileList.First.Value;
	fileList.RemoveFirst();
	(string, long) makeTuple(string[] fields) =>
	(fields[0], long.Parse(fields[1]));
	var group = new List<(string, long)> { makeTuple(x) };
	foreach (var f in fileList.AsParallel().Where(y => CompareName(x, y)).ToArray()) {
	group.Add(makeTuple(f));
	fileList.Remove(f);
	}
	fileGroups.Add((group, GetGroupSize(group)));
	}
	foreach (var (fileGroup, groupSize, index) in fileGroups.OrderByDescending(fg => fg.Item2).Select((f, i) => (f.Item1, f.Item2, i))) {
	Console.WriteLine((index + 1) + ": " + GetSizeMB(groupSize));
	foreach (var fg in fileGroup)
	Console.WriteLine(fg.Item1 + ": " + fg.Item2);
	Console.WriteLine();
	Console.WriteLine();
	}
	//Console.ReadKey();
	}
	}
	}

view raw Program.cs hosted with ❤ by GitHub