增加`--ad-keyword`选项过滤广告URL

#221 #204

src/N_m3u8DL-RE.Common/Resource/ResString.cs

@@ -12,6 +12,7 @@ namespace N_m3u8DL_RE.Common.Resource
         public readonly static string ReLiveTs = "<RE_LIVE_TS>";
         public static string customRangeWarn { get => GetText("customRangeWarn"); }
         public static string customRangeFound { get => GetText("customRangeFound"); }
+        public static string customAdKeywordsFound { get => GetText("customAdKeywordsFound"); }
         public static string customRangeInvalid { get => GetText("customRangeInvalid"); }
         public static string autoBinaryMerge { get => GetText("autoBinaryMerge"); }
         public static string autoBinaryMerge2 { get => GetText("autoBinaryMerge2"); }
@@ -31,6 +32,7 @@ namespace N_m3u8DL_RE.Common.Resource
         public static string cmd_ffmpegBinaryPath { get => GetText("cmd_ffmpegBinaryPath"); }
         public static string cmd_mkvmergeBinaryPath { get => GetText("cmd_mkvmergeBinaryPath"); }
         public static string cmd_baseUrl { get => GetText("cmd_baseUrl"); }
+        public static string cmd_adKeyword { get => GetText("cmd_adKeyword"); }
         public static string cmd_moreHelp { get => GetText("cmd_moreHelp"); }
         public static string cmd_header { get => GetText("cmd_header"); }
         public static string cmd_muxImport { get => GetText("cmd_muxImport"); }

src/N_m3u8DL-RE.Common/Resource/StaticText.cs

@@ -22,6 +22,12 @@ namespace N_m3u8DL_RE.Common.Resource
                 zhTW: "自定義下載範圍無效",
                 enUS: "User customed range invalid"
             ),
+            ["customAdKeywordsFound"] = new TextContainer
+            (
+                zhCN: "用户自定义广告分片URL关键字：",
+                zhTW: "用戶自定義廣告分片URL關鍵字：",
+                enUS: "User customed Ad keyword: "
+            ),
             ["customRangeFound"] = new TextContainer
             (
                 zhCN: "用户自定义下载范围：",
@@ -346,6 +352,12 @@ namespace N_m3u8DL_RE.Common.Resource
                 zhTW: "手動設置直播列表刷新間隔",
                 enUS: "Manually set the live playlist refresh interval"
             ),
+            ["cmd_adKeyword"] = new TextContainer
+            (
+                zhCN: "设置广告分片的URL关键字(正则表达式)",
+                zhTW: "設置廣告分片的URL關鍵字(正則表達式)",
+                enUS: "Set URL keywords (regular expressions) for AD segments"
+            ),
             ["cmd_liveTakeCount"] = new TextContainer
             (
                 zhCN: "手动设置录制直播时首次获取分片的数量",

src/N_m3u8DL-RE/CommandLine/CommandInvoker.cs

@@ -18,7 +18,7 @@ namespace N_m3u8DL_RE.CommandLine
 {
     internal partial class CommandInvoker
     {
-        public const string VERSION_INFO = "N_m3u8DL-RE (Beta version) 20230707";
+        public const string VERSION_INFO = "N_m3u8DL-RE (Beta version) 20230712";
 
         [GeneratedRegex("((best|worst)\\d*|all)")]
         private static partial Regex ForStrRegex();
@@ -57,6 +57,8 @@ namespace N_m3u8DL_RE.CommandLine
         private readonly static Option<string?> BaseUrl = new(new string[] { "--base-url" }, description: ResString.cmd_baseUrl);
         private readonly static Option<bool> ConcurrentDownload = new(new string[] { "-mt", "--concurrent-download" }, description: ResString.cmd_concurrentDownload, getDefaultValue: () => false);
         private readonly static Option<bool> NoLog = new(new string[] { "--no-log" }, description: ResString.cmd_noLog, getDefaultValue: () => false);
+        private readonly static Option<string[]?> AdKeywords = new(new string[] { "--ad-keyword" }, description: ResString.cmd_adKeyword) { ArgumentHelpName = "REG" };
+
 
         //代理选项
         private readonly static Option<bool> UseSystemProxy = new(new string[] { "--use-system-proxy" }, description: ResString.cmd_useSystemProxy, getDefaultValue: () => true);
@@ -499,6 +501,7 @@ namespace N_m3u8DL_RE.CommandLine
                     LiveTakeCount = bindingContext.ParseResult.GetValueForOption(LiveTakeCount),
                     NoDateInfo = bindingContext.ParseResult.GetValueForOption(NoDateInfo),
                     NoLog = bindingContext.ParseResult.GetValueForOption(NoLog),
+                    AdKeywords = bindingContext.ParseResult.GetValueForOption(AdKeywords),
                 };
 
                 if (bindingContext.ParseResult.HasOption(CustomHLSMethod)) option.CustomHLSMethod = bindingContext.ParseResult.GetValueForOption(CustomHLSMethod);
@@ -564,7 +567,7 @@ namespace N_m3u8DL_RE.CommandLine
                 MuxAfterDone,
                 CustomHLSMethod, CustomHLSKey, CustomHLSIv, UseSystemProxy, CustomProxy, CustomRange, TaskStartAt,
                 LivePerformAsVod, LiveRealTimeMerge, LiveKeepSegments, LivePipeMux, LiveFixVttByAudio, LiveRecordLimit, LiveWaitTime, LiveTakeCount,
-                MuxImports, VideoFilter, AudioFilter, SubtitleFilter, DropVideoFilter, DropAudioFilter, DropSubtitleFilter, MoreHelp
+                MuxImports, VideoFilter, AudioFilter, SubtitleFilter, DropVideoFilter, DropAudioFilter, DropSubtitleFilter, AdKeywords, MoreHelp
             };
 
             rootCommand.TreatUnmatchedTokensAsErrors = true;

src/N_m3u8DL-RE/CommandLine/MyOption.cs

@@ -17,6 +17,10 @@ namespace N_m3u8DL_RE.CommandLine
         /// </summary>
         public Dictionary<string, string> Headers { get; set; } = new Dictionary<string, string>();
         /// <summary>
+        /// See: <see cref="CommandInvoker.AdKeywords"/>.
+        /// </summary>
+        public string[]? AdKeywords { get; set; }
+        /// <summary>
         /// See: <see cref="CommandInvoker.Keys"/>.
         /// </summary>
         public string[]? Keys { get; set; }

src/N_m3u8DL-RE/Program.cs

@@ -304,6 +304,9 @@ namespace N_m3u8DL_RE
             if (!livingFlag)
                 FilterUtil.ApplyCustomRange(selectedStreams, option.CustomRange);
 
+            //应用用户自定义的广告分片关键字
+            FilterUtil.CleanAd(selectedStreams, option.AdKeywords);
+
             //记录文件
             extractor.RawFiles["meta_selected.json"] = GlobalUtil.ConvertToJson(selectedStreams);
 

src/N_m3u8DL-RE/Util/FilterUtil.cs

@@ -8,6 +8,7 @@ using System;
 using System.Collections.Generic;
 using System.Linq;
 using System.Text;
+using System.Text.RegularExpressions;
 using System.Threading.Tasks;
 
 namespace N_m3u8DL_RE.Util
@@ -227,5 +228,41 @@ namespace N_m3u8DL_RE.Util
                 stream.SkippedDuration = skippedDur;
             }
         }
+
+        /// <summary>
+        /// 根据用户输入，清除广告分片
+        /// </summary>
+        /// <param name="selectedSteams"></param>
+        /// <param name="customRange"></param>
+        public static void CleanAd(List<StreamSpec> selectedSteams, string[]? keywords)
+        {
+            if (keywords == null) return;
+            var regList = keywords.Select(s => new Regex(s));
+            foreach ( var reg in regList)
+            {
+                Logger.InfoMarkUp($"{ResString.customAdKeywordsFound}[Cyan underline]{reg}[/]");
+            }
+
+            foreach (var stream in selectedSteams)
+            {
+                if (stream.Playlist == null) continue;
+                foreach (var part in stream.Playlist.MediaParts)
+                {
+                    //没有找到广告分片
+                    if (part.MediaSegments.All(x => regList.All(reg => !reg.IsMatch(x.Url))))
+                    {
+                        continue;
+                    }
+                    //找到广告分片 清理
+                    else
+                    {
+                        part.MediaSegments = part.MediaSegments.Where(x => regList.All(reg => !reg.IsMatch(x.Url))).ToList();
+                    }
+                }
+
+                //清理已经为空的 part
+                stream.Playlist.MediaParts = stream.Playlist.MediaParts.Where(x => x.MediaSegments.Count > 0).ToList();
+            }
+        }
     }
 }