Finish implementing multi-segment ODS support

Related #13 Signed-off-by: Ethan Dye <[email protected]>
ecdye · Sep 30, 2024 · aa2d22d · aa2d22d
1 parent 0ea3870
commit aa2d22d
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 10 deletions.
diff --git a/Sources/macSubtitleOCR/PGS/PGSError.swift b/Sources/macSubtitleOCR/PGS/PGSError.swift
@@ -9,4 +9,5 @@
 enum PGSError: Error {
     case invalidFormat
     case fileReadError
+    case invalidObjectDataSegment
 }
diff --git a/Sources/macSubtitleOCR/PGS/Parsers/ODS.swift b/Sources/macSubtitleOCR/PGS/Parsers/ODS.swift
@@ -13,7 +13,6 @@ class ODS {
     // MARK: - Properties
 
     private let logger = Logger(subsystem: "github.ecdye.macSubtitleOCR", category: "ODS")
-    private var objectDataLength: Int = 0
     private var objectWidth: Int = 0
     private var objectHeight: Int = 0
     private var imageData: Data = .init()
@@ -51,8 +50,8 @@ class ODS {
     //   0x17: Segment Type; already checked by the caller
     //   2 bytes: Object ID (unused by us)
     //   1 byte: Version number (unused by us)
-    //   1 byte: Sequence flag (should be 0x80 for new object, 0x00 for continuation) (unused by us)
-    //   3 bytes: Object data length
+    //   1 byte: Sequence flag (should be 0x80 for multi-segment object, 0xC0 for single segment, 0x40 for last segment)
+    //   3 bytes: Object data length (unused by us)
     //   2 bytes: Object width
     //   2 bytes: Object height
     //   Rest: Image data (run-length encoded, RLE)
@@ -61,16 +60,18 @@ class ODS {
         let sequenceFlag = data[3]
 
         if sequenceFlag == 0xC0 {
-            logger.debug("Sequence flag is 0xC0, indicating a new object fully contained in this segment")
+            logger.debug("Sequence flag indicates an object fully contained in this segment")
         } else if sequenceFlag == 0x80 {
-            logger.debug("Sequence flag is 0x80, indicating a new object with more data in subsequent segments")
+            logger.debug("Sequence flag indicates an object contained in multiple segments")
+        } else if sequenceFlag == 0x40 {
+            logger.debug("Sequence flag indicates the last segment of a multi-segment object")
+        } else {
+            logger.warning("Unknown sequence flag: \(sequenceFlag)")
         }
 
-        objectDataLength = Int(data[4]) << 16 | Int(data[5]) << 8 | Int(data[6])
-
         // PGS includes the width and height as part of the image data length calculations
-        guard objectDataLength <= data.count - 7 else {
-            throw macSubtitleOCRError.invalidFormat
+        guard data.count > 7 else {
+            throw PGSError.invalidObjectDataSegment
         }
 
         objectWidth = Int(data[7]) << 8 | Int(data[8])

diff --git a/Sources/macSubtitleOCR/macSubtitleOCRError.swift b/Sources/macSubtitleOCR/macSubtitleOCRError.swift
@@ -7,7 +7,6 @@
 //
 
 enum macSubtitleOCRError: Error {
-    case invalidFormat
     case fileReadError
     case fileCreationError
     case fileWriteError