Improve srt parsing

3 files changed, 137 insertions, 11 deletions

diff --git a/build/server.js b/build/server.js
index 18989ad..d2836d6 100644
--- a/build/server.js
+++ b/build/server.js
@@ -1,4 +1,4 @@
-// Generated by Haxe 4.3.3
+// Generated by Haxe 4.3.4
 (function ($global) { "use strict";
 var $estr = function() { return js_Boot.__string_rec(this,''); },$hxEnums = $hxEnums || {},$_;
 function $extend(from, fields) {
diff --git a/res/client.js b/res/client.js
index beb88e3..cd3157e 100644
--- a/res/client.js
+++ b/res/client.js
@@ -1,4 +1,4 @@
-// Generated by Haxe 4.3.3
+// Generated by Haxe 4.3.4
 (function ($hx_exports, $global) { "use strict";
 $hx_exports["client"] = $hx_exports["client"] || {};
 $hx_exports["client"]["JsApi"] = $hx_exports["client"]["JsApi"] || {};
@@ -105,9 +105,61 @@ EReg.prototype = {
 			throw haxe_Exception.thrown("EReg::matched");
 		}
 	}
+	,matchedPos: function() {
+		if(this.r.m == null) {
+			throw haxe_Exception.thrown("No string matched");
+		}
+		return { pos : this.r.m.index, len : this.r.m[0].length};
+	}
+	,matchSub: function(s,pos,len) {
+		if(len == null) {
+			len = -1;
+		}
+		if(this.r.global) {
+			this.r.lastIndex = pos;
+			this.r.m = this.r.exec(len < 0 ? s : HxOverrides.substr(s,0,pos + len));
+			var b = this.r.m != null;
+			if(b) {
+				this.r.s = s;
+			}
+			return b;
+		} else {
+			var b = this.match(len < 0 ? HxOverrides.substr(s,pos,null) : HxOverrides.substr(s,pos,len));
+			if(b) {
+				this.r.s = s;
+				this.r.m.index += pos;
+			}
+			return b;
+		}
+	}
 	,split: function(s) {
 		return s.replace(this.r,"#__delim__#").split("#__delim__#");
 	}
+	,map: function(s,f) {
+		var offset = 0;
+		var buf_b = "";
+		do {
+			if(offset >= s.length) {
+				break;
+			} else if(!this.matchSub(s,offset)) {
+				buf_b += Std.string(HxOverrides.substr(s,offset,null));
+				break;
+			}
+			var p = this.matchedPos();
+			buf_b += Std.string(HxOverrides.substr(s,offset,p.pos - offset));
+			buf_b += Std.string(f(this));
+			if(p.len == 0) {
+				buf_b += Std.string(HxOverrides.substr(s,p.pos,1));
+				offset = p.pos + 1;
+			} else {
+				offset = p.pos + p.len;
+			}
+		} while(this.r.global);
+		if(!this.r.global && offset > 0 && offset < s.length) {
+			buf_b += Std.string(HxOverrides.substr(s,offset,null));
+		}
+		return buf_b;
+	}
 };
 var HxOverrides = function() { };
 HxOverrides.__name__ = true;
@@ -406,6 +458,15 @@ StringTools.lpad = function(s,c,l) {
 	buf_b += s == null ? "null" : "" + s;
 	return buf_b;
 };
+StringTools.rpad = function(s,c,l) {
+	if(c.length <= 0) {
+		return s;
+	}
+	var buf_b = "";
+	buf_b = "" + (s == null ? "null" : "" + s);
+	while(buf_b.length < l) buf_b += c == null ? "null" : "" + c;
+	return buf_b;
+};
 StringTools.replace = function(s,sub,by) {
 	return s.split(sub).join(by);
 };
@@ -3244,10 +3305,12 @@ client_players_RawSubs.parseSrt = function(video,url) {
 			return;
 		}
 		var subs = [];
-		var blocks = StringTools.replace(text,"\r\n","\n").split("\n\n");
+		var blocks = client_players_RawSubs.getSrtBlocks(StringTools.replace(text,"\r\n","\n").split("\n"));
+		var badTimeReg = new EReg("(,[0-9]+)","g");
 		var _g = 0;
 		while(_g < blocks.length) {
-			var lines = blocks[_g++].split("\n");
+			var lines = blocks[_g];
+			++_g;
 			if(lines.length < 3) {
 				continue;
 			}
@@ -3255,7 +3318,14 @@ client_players_RawSubs.parseSrt = function(video,url) {
 			var _g2 = 2;
 			var _g3 = lines.length;
 			while(_g2 < _g3) _g1.push(lines[_g2++]);
-			subs.push({ counter : lines[0], time : StringTools.replace(lines[1],",","."), text : _g1.join("\n")});
+			var time = badTimeReg.map(lines[1],function(reg) {
+				var ms = reg.matched(1);
+				if(ms.length < 4) {
+					return StringTools.rpad(ms,"0",4);
+				}
+				return ms;
+			});
+			subs.push({ counter : lines[0], time : StringTools.replace(time,",","."), text : StringTools.ltrim(_g1.join("\n"))});
 		}
 		var data = "WEBVTT\n\n";
 		var _g = 0;
@@ -3270,6 +3340,32 @@ client_players_RawSubs.parseSrt = function(video,url) {
 		client_players_RawSubs.onParsed(video,"SRT subtitles",url);
 	});
 };
+client_players_RawSubs.getSrtBlocks = function(lines) {
+	var blocks = [];
+	var isNumLineReg = new EReg("^([0-9]+)$","");
+	var block = [];
+	var _g_current = 0;
+	while(_g_current < lines.length) {
+		var _g_value = lines[_g_current];
+		var _g_key = _g_current++;
+		if(blocks.length == 0 && _g_value.length == 0) {
+			continue;
+		}
+		var tmp = lines[_g_key - 1];
+		var tmp1 = lines[_g_key + 1];
+		if((tmp != null ? tmp : "").length == 0 && isNumLineReg.match(_g_value) && (tmp1 != null ? tmp1 : "").indexOf("-->") != -1) {
+			if(block.length > 0) {
+				blocks.push(block);
+				block = [];
+			}
+		}
+		block.push(_g_value);
+	}
+	if(block.length > 0) {
+		blocks.push(block);
+	}
+	return blocks;
+};
 client_players_RawSubs.parseAss = function(video,url) {
 	window.fetch(url).then(function(response) {
 		return response.text();
@@ -3369,7 +3465,7 @@ client_players_RawSubs.convertAssTime = function(time) {
 client_players_RawSubs.isProxyError = function(text) {
 	if(StringTools.startsWith(text,"Proxy error:")) {
 		client_Main.serverMessage("Failed to add subs: proxy error");
-		haxe_Log.trace("Failed to add subs: " + text,{ fileName : "src/client/players/RawSubs.hx", lineNumber : 191, className : "client.players.RawSubs", methodName : "isProxyError"});
+		haxe_Log.trace("Failed to add subs: " + text,{ fileName : "src/client/players/RawSubs.hx", lineNumber : 221, className : "client.players.RawSubs", methodName : "isProxyError"});
 		return true;
 	}
 	return false;
diff --git a/src/client/players/RawSubs.hx b/src/client/players/RawSubs.hx
index d7e32e2..8448dfd 100644
--- a/src/client/players/RawSubs.hx
+++ b/src/client/players/RawSubs.hx
@@ -51,17 +51,24 @@ class RawSubs {
 				time:String,
 				text:String
 			}> = [];
-			final blocks = text.replace("\r\n", "\n").split("\n\n");
-			for (block in blocks) {
-				final lines = block.split("\n");
+			final lines = text.replace("\r\n", "\n").split("\n");
+			final blocks = getSrtBlocks(lines);
+			final badTimeReg = ~/(,[0-9]+)/g;
+			for (lines in blocks) {
 				if (lines.length < 3) continue;
 				final textLines = [
 					for (i in 2...lines.length) lines[i]
 				];
+				// fix incomplete ms in timestamps like `00:00:02,50 --> 00:00:06,220`
+				final time = badTimeReg.map(lines[1], reg -> {
+					final ms = reg.matched(1);
+					if (ms.length < 4) return ms.rpad("0", 4);
+					return ms;
+				});
 				subs.push({
 					counter: lines[0],
-					time: lines[1].replace(",", "."),
-					text: textLines.join("\n")
+					time: time.replace(",", "."),
+					text: textLines.join("\n").ltrim()
 				});
 			}
 			var data = "WEBVTT\n\n";
@@ -76,6 +83,29 @@ class RawSubs {
 		});
 	}
 
+	static function getSrtBlocks(lines:Array<String>):Array<Array<String>> {
+		final blocks = [];
+		final isNumLineReg = ~/^([0-9]+)$/;
+		// [id, time, firstTextLine, ... lastTextLine]
+		var block = [];
+		for (i => line in lines) {
+			if (blocks.length == 0 && line.length == 0) continue;
+			final prevLine = lines[i - 1] ?? "";
+			final nextLine = lines[i + 1] ?? "";
+			// block id line
+			if (prevLine.length == 0 && isNumLineReg.match(line) && nextLine.contains("-->")) {
+				// push previously collected block and start new one
+				if (block.length > 0) {
+					blocks.push(block);
+					block = [];
+				}
+			}
+			block.push(line);
+		}
+		if (block.length > 0) blocks.push(block);
+		return blocks;
+	}
+
 	static function parseAss(video:VideoElement, url:String):Void {
 		window.fetch(url).then(response -> {
 			return response.text();