Merge pull request #253 from Kevin-Lee/task/252/fix-no-title

Fix #252 - Parsing SAMI file fails when there's no `<TITLE>` tag
kevin-lee · May 16, 2023 · 7ddef9e · 7ddef9e
2 parents d3ae720 + a04fbb1
commit 7ddef9e
Show file tree

Hide file tree

Showing 4 changed files with 101 additions and 4 deletions.
diff --git a/modules/whatsub-core/src/main/scala/whatsub/parse/SmiParser.scala b/modules/whatsub-core/src/main/scala/whatsub/parse/SmiParser.scala
@@ -142,7 +142,7 @@ object SmiParser {
             case Some(ParseStatus.SamiEnd) =>
               effectOf((title, acc).asRight)
             case Some(ParseStatus.HeadStart) =>
-              parseAll(skipUntil(rest, ParseStatus.TitleStart), title, acc)
+              parseAll(skipUntil(rest, ParseStatus.TitleStart, ParseStatus.StyleStart), title, acc)
             case Some(ParseStatus.HeadEnd) =>
               parseAll(skipUntil(rest, ParseStatus.BodyStart), title, acc)
             case Some(ParseStatus.TitleStart) =>
@@ -186,14 +186,18 @@ object SmiParser {
       }
 
   @tailrec
-  private def skipUntil(lineAndLineNums: Seq[(String, Int)], target: ParseStatus): Seq[(String, Int)] =
+  private def skipUntil(
+    lineAndLineNums: Seq[(String, Int)],
+    target: ParseStatus,
+    orAnyOf: ParseStatus*
+  ): Seq[(String, Int)] =
     lineAndLineNums match {
       case (line, index) +: rest =>
         val preprocessed = line.removeEmptyChars
         parseNonLine(preprocessed)
-          .find(_ === target) match {
+          .find(parseStatus => parseStatus === target || orAnyOf.contains(parseStatus)) match {
           case None =>
-            skipUntil(rest, target)
+            skipUntil(rest, target, orAnyOf: _*)
           case Some(_) =>
             lineAndLineNums
         }

diff --git a/modules/whatsub-core/src/test/resources/golden/test-smi-parsed-no-title.smi b/modules/whatsub-core/src/test/resources/golden/test-smi-parsed-no-title.smi
@@ -0,0 +1,39 @@
+<SAMI>
+<HEAD>
+  <TITLE></TITLE>
+</HEAD>
+<BODY>
+  <SYNC Start=4561><P><font color="#ec14bd">This is a test subtitle.</Font> by <font color="Skyblue" size=10 face=HY수평선B>Kevin Lee</font>
+  <SYNC Start=7000><P>&nbsp;
+  <SYNC Start=10231><P><font color="Gold"> 자 막 </Font>: <font color="Skyblue"> 테 스 트 자 막</font><br><font color=Lime>(kevinlee.io)</font>
+  <SYNC Start=15831><P>&nbsp;
+  <SYNC Start=15831><P><font color="#80ffff">Modified </Font>by <font color="Lime">Kevin Lee<br><font color="#ec14bd">(https://kevinlee.io)</font>
+  <SYNC Start=19000><P>&nbsp;
+  <SYNC Start=87274><P><font color="Gold">헬로우 월드!</font>
+  <SYNC Start=89741><P>&nbsp;
+  <SYNC Start=90674><P>안녕
+  <SYNC Start=91741><P>&nbsp;
+  <SYNC Start=196380><P>이건 그냥 테스트를 위한 자막이에요.
+  <SYNC Start=198115><P>&nbsp;
+  <SYNC Start=199950><P>테스트를 위한 또다른 자막 입니다.
+  <SYNC Start=202119><P>&nbsp;
+  <SYNC Start=202186><P>Ok
+  <SYNC Start=204155><P>&nbsp;
+  <SYNC Start=204221><P>이제 영어도 한번 넣어 볼까요?
+  <SYNC Start=206957><P>&nbsp;
+  <SYNC Start=212029><P>짜잔!
+  <SYNC Start=213763><P>&nbsp;
+  <SYNC Start=213830><P>Lorem Ipsum is simply dummy text of the printing and typesetting industry
+  <SYNC Start=215533><P>&nbsp;
+  <SYNC Start=215598><P>영어도 잘 되는거 같군요.
+  <SYNC Start=217868><P>&nbsp;
+  <SYNC Start=222672><P>블라블라.
+  <SYNC Start=224041><P>&nbsp;
+  <SYNC Start=224108><P>이제 자막에 색깔을 입혀 볼까요?
+  <SYNC Start=225242><P>&nbsp;
+  <SYNC Start=249699><P>위에 이미 했지만...
+  <SYNC Start=250767><P>&nbsp;
+  <SYNC Start=262274><P><font color="Gold">Gold color로 입혀 봤습니다.</font>
+  <SYNC Start=8873754><P>&nbsp;
+</BODY>
+</SAMI>
diff --git a/modules/whatsub-core/src/test/resources/golden/test-src-no-title.smi b/modules/whatsub-core/src/test/resources/golden/test-src-no-title.smi
@@ -0,0 +1,52 @@
+<SAMI>
+  <HEAD>
+    <STYLE TYPE="text/css">
+      <!--
+      P { font-family: Arial; font-weight: normal; color: white; background-color: black; text-align: center; }
+      .KRCC { name: Korean; lang: ko-KR ; SAMIType: CC ; }
+      -->
+    </STYLE>
+  </HEAD>
+  <BODY>
+    <-- Open play menu, choose Captions and Subtiles, On if available -->
+    <-- Open tools menu, Security, Show local captions when present -->
+    <!-- Another test comment -->
+    <!--
+    자막 번역: 누군가 2021. 01. 01
+    -->
+    <SYNC Start=4561 End=7000><P Class=KRCC>
+    <font color="#ec14bd">This is a test subtitle.</Font> by <font color="Skyblue" size=10 face=HY수평선B>Kevin Lee</font>
+    <SYNC Start=10231><P Class=KRCC><font color="Gold"> 자 막 </Font>: <font color="Skyblue"> 테 스 트 자 막</font><br><font color=Lime>(kevinlee.io)</font>
+    <SYNC Start=15831 End=19000><P Class=KRCC><font color="#80ffff">Modified </Font>by <font color="Lime">Kevin Lee<br><font color="#ec14bd">(https://kevinlee.io)</font>
+    <SYNC Start=20754><P Class=KRCC>&nbsp;
+    <SYNC Start=87274><P Class=KRCC><font color="Gold">헬로우 월드!</font>
+    <SYNC Start=89741 End=90674><P Class=KRCC>&nbsp;
+    <SYNC Start=90674><P Class=KRCC>안녕
+    <SYNC Start=91741 End=100000><P Class=KRCC>&nbsp;
+    <SYNC Start=196380><P Class=KRCC>
+    이건 그냥 테스트를 위한 자막이에요.
+    <SYNC Start=198115 End=199950><P Class=KRCC>&nbsp;
+    <SYNC Start=199950><P Class=KRCC>테스트를 위한 또다른 자막 입니다.
+    <SYNC Start=202119><P Class=KRCC>&nbsp;
+    <SYNC Start=202186><P Class=KRCC>Ok
+    <SYNC Start=204155><P Class=KRCC>&nbsp;
+    <SYNC Start=204221><P Class=KRCC>이제 영어도 한번 넣어 볼까요?
+    <SYNC Start=206957><P Class=KRCC>&nbsp;
+    <!-- Test comment - 중간에 코멘트도 넣고 -->
+    <SYNC Start=212029><P Class=KRCC>짜잔!
+    <SYNC Start=213763><P Class=KRCC>&nbsp;
+    <SYNC Start=213830><P Class=KRCC>Lorem Ipsum is simply dummy text of the printing and typesetting industry
+    <SYNC Start=215533><P Class=KRCC>&nbsp;
+    <SYNC Start=215598><P Class=KRCC>영어도 잘 되는거 같군요.
+    <-- Test comment 2 -->
+    <SYNC Start=217868><P Class=KRCC>&nbsp;
+    <SYNC Start=222672><P Class=KRCC>블라블라.
+    <SYNC Start=224041><P Class=KRCC>&nbsp;
+    <SYNC Start=224108><P Class=KRCC>이제 자막에 색깔을 입혀 볼까요?
+    <SYNC Start=225242><P Class=KRCC>&nbsp;
+    <SYNC Start=249699><P Class=KRCC>위에 이미 했지만...
+    <SYNC Start=250767><P Class=KRCC>&nbsp;
+    <SYNC Start=262274><P Class=KRCC><font color="Gold">Gold color로 입혀 봤습니다.</font>
+    <SYNC Start=8873754><P Class=KRCC>&nbsp;
+  </BODY>
+</SAMI>
diff --git a/modules/whatsub-core/src/test/scala/whatsub/parse/SmiParserSpec.scala b/modules/whatsub-core/src/test/scala/whatsub/parse/SmiParserSpec.scala
@@ -12,6 +12,8 @@ import whatsub.*
 object SmiParserSpec extends Properties {
   override def tests: List[Test] = List(
     ParserGoldenTester.goldenTestParse[Smi]("golden/test-src.smi", "golden/test-smi-parsed.smi", SmiParser.parse[IO]),
+    ParserGoldenTester
+      .goldenTestParse[Smi]("golden/test-src-no-title.smi", "golden/test-smi-parsed-no-title.smi", SmiParser.parse[IO]),
   )
 
 }