| 1 | /** |
|---|
| 2 | * Parse a TimedText XML and return an array of captions. |
|---|
| 3 | **/ |
|---|
| 4 | package com.jeroenwijering.parsers { |
|---|
| 5 | |
|---|
| 6 | |
|---|
| 7 | import com.jeroenwijering.utils.Strings; |
|---|
| 8 | |
|---|
| 9 | |
|---|
| 10 | public class TTParser { |
|---|
| 11 | |
|---|
| 12 | |
|---|
| 13 | /** |
|---|
| 14 | * Parse the captions XML. |
|---|
| 15 | * |
|---|
| 16 | * @param dat The loaded XML, which must be in W3C TimedText format. |
|---|
| 17 | * @return An array with captions. |
|---|
| 18 | * Each caption is an object with 'begin' and 'text' parameters. |
|---|
| 19 | **/ |
|---|
| 20 | public static function parseCaptions(dat:XML):Array { |
|---|
| 21 | var arr:Array = new Array({begin:0,text:''}); |
|---|
| 22 | for each (var i:XML in dat.children()) { |
|---|
| 23 | if(i.localName() == "body") { |
|---|
| 24 | for each (var j:XML in i.children()) { |
|---|
| 25 | for each (var k:XML in j.children()) { |
|---|
| 26 | if(k.localName() == 'p') { |
|---|
| 27 | var obj:Object = TTParser.parseCaption(k); |
|---|
| 28 | arr.push(obj); |
|---|
| 29 | if(obj['end']) { |
|---|
| 30 | arr.push({begin:obj['end'],text:''}); |
|---|
| 31 | delete obj['end']; |
|---|
| 32 | } else if (obj['dur']) { |
|---|
| 33 | arr.push({begin:obj['begin']+obj['dur'],text:''}); |
|---|
| 34 | delete obj['dur']; |
|---|
| 35 | } |
|---|
| 36 | } |
|---|
| 37 | } |
|---|
| 38 | } |
|---|
| 39 | } |
|---|
| 40 | } |
|---|
| 41 | return arr; |
|---|
| 42 | }; |
|---|
| 43 | |
|---|
| 44 | |
|---|
| 45 | /** Parse a single captions entry. **/ |
|---|
| 46 | private static function parseCaption(dat:XML):Object { |
|---|
| 47 | var ptn:RegExp = /(\n<br.*>\n)+/; |
|---|
| 48 | var obj:Object = { |
|---|
| 49 | begin:Strings.seconds(dat.@begin), |
|---|
| 50 | dur:Strings.seconds(dat.@dur), |
|---|
| 51 | end:Strings.seconds(dat.@end), |
|---|
| 52 | text:dat.children().toString().replace(ptn,'\n').replace(ptn,'\n') |
|---|
| 53 | }; |
|---|
| 54 | return obj; |
|---|
| 55 | }; |
|---|
| 56 | |
|---|
| 57 | |
|---|
| 58 | } |
|---|
| 59 | |
|---|
| 60 | |
|---|
| 61 | } |
|---|