fix: Unsigned interpretation

Author: siliconlad

src/pybag/bag/record_parser.py

@@ -80,7 +80,7 @@ def _parse_header_field(cls, data: bytes, offset: int) -> tuple[int, str, bytes]
         Returns:
             Tuple of (new_offset, field_name, field_value).
         """
-        field_len = struct.unpack_from('<i', data, offset)[0]
+        field_len = struct.unpack_from('<I', data, offset)[0]
         offset += 4
         field_data = data[offset:offset + field_len]
         offset += field_len
@@ -135,7 +135,7 @@ def parse_record(cls, file: BaseReader) -> tuple[int, Any] | None:
         header_len_bytes = file.read(4)
         if len(header_len_bytes) < 4:
             return None  # EOF
-        header_len = struct.unpack('<i', header_len_bytes)[0]
+        header_len = struct.unpack('<I', header_len_bytes)[0]
         header = cls._parse_header(file, header_len)
 
         # Get the operation type
@@ -147,7 +147,7 @@ def parse_record(cls, file: BaseReader) -> tuple[int, Any] | None:
         data_len_bytes = file.read(4)
         if len(data_len_bytes) < 4:
             return None  # EOF
-        data_len = struct.unpack('<i', data_len_bytes)[0]
+        data_len = struct.unpack('<I', data_len_bytes)[0]
         data = cls._parse_data(file, data_len)
 
         return op, cls._parse_record_by_type(op, header, data)
@@ -183,9 +183,9 @@ def _parse_bag_header(
         data: bytes
     ) -> BagHeaderRecord:
         """Parse a bag header record."""
-        index_pos = struct.unpack('<q', header['index_pos'])[0]
-        conn_count = struct.unpack('<i', header['conn_count'])[0]
-        chunk_count = struct.unpack('<i', header['chunk_count'])[0]
+        index_pos = struct.unpack('<Q', header['index_pos'])[0]
+        conn_count = struct.unpack('<I', header['conn_count'])[0]
+        chunk_count = struct.unpack('<I', header['chunk_count'])[0]
         return BagHeaderRecord(index_pos, conn_count, chunk_count, data)
 
     @classmethod
@@ -196,7 +196,7 @@ def _parse_chunk(
     ) -> ChunkRecord:
         """Parse a chunk record."""
         compression = header['compression'].decode('ascii')
-        size = struct.unpack('<i', header['size'])[0]
+        size = struct.unpack('<I', header['size'])[0]
         return ChunkRecord(compression, size, data)
 
     @classmethod
@@ -210,7 +210,7 @@ def _parse_connection(
         The connection header contains basic info, and the data section
         contains the full connection header from the original publisher.
         """
-        conn = struct.unpack('<i', header['conn'])[0]
+        conn = struct.unpack('<I', header['conn'])[0]
         topic = header['topic'].decode('utf-8')
         return ConnectionRecord(conn, topic, data)
 
@@ -221,7 +221,7 @@ def _parse_message_data(
         data: bytes
     ) -> MessageDataRecord:
         """Parse a message data record."""
-        conn = struct.unpack('<i', header['conn'])[0]
+        conn = struct.unpack('<I', header['conn'])[0]
         time = _decode_ros_time(header['time'])
         return MessageDataRecord(conn, time, data)
 
@@ -232,9 +232,9 @@ def _parse_index_data(
         data: bytes
     ) -> IndexDataRecord:
         """Parse an index data record."""
-        ver = struct.unpack('<i', header['ver'])[0]
-        conn = struct.unpack('<i', header['conn'])[0]
-        count = struct.unpack('<i', header['count'])[0]
+        ver = struct.unpack('<I', header['ver'])[0]
+        conn = struct.unpack('<I', header['conn'])[0]
+        count = struct.unpack('<I', header['count'])[0]
         return IndexDataRecord(ver, conn, count, data)
 
     @classmethod
@@ -244,11 +244,11 @@ def _parse_chunk_info(
         data: bytes
     ) -> ChunkInfoRecord:
         """Parse a chunk info record."""
-        ver = struct.unpack('<i', header['ver'])[0]
-        chunk_pos = struct.unpack('<q', header['chunk_pos'])[0]
+        ver = struct.unpack('<I', header['ver'])[0]
+        chunk_pos = struct.unpack('<Q', header['chunk_pos'])[0]
         start_time = _decode_ros_time(header['start_time'])
         end_time = _decode_ros_time(header['end_time'])
-        count = struct.unpack('<i', header['count'])[0]
+        count = struct.unpack('<I', header['count'])[0]
         return ChunkInfoRecord(ver, chunk_pos, start_time, end_time, count, data)
 
     @classmethod

src/pybag/bag/records.py

@@ -83,7 +83,7 @@ def connection_header(self) -> ConnectionHeader:
         fields: dict[str, bytes] = {}
         offset = 0
         while offset < len(self.data):
-            field_len = struct.unpack_from('<i', self.data, offset)[0]
+            field_len = struct.unpack_from('<I', self.data, offset)[0]
             offset += 4
             field_data = self.data[offset:offset + field_len]
             offset += field_len
@@ -192,7 +192,7 @@ def connection_counts(self) -> dict[int, int]:
         num_entries = len(self.data) // 8
         for i in range(num_entries):
             offset = i * 8
-            conn_id, msg_count = struct.unpack_from('<ii', self.data, offset)
+            conn_id, msg_count = struct.unpack_from('<II', self.data, offset)
             counts[conn_id] = msg_count
         return counts
 
@@ -224,8 +224,8 @@ def entries(self) -> list[tuple[int, int]]:
         result: list[tuple[int, int]] = []
         for i in range(self.count):
             offset = i * 12
-            # ROS time format: two uint32 (secs, nsecs) + int32 offset
-            secs, nsecs, chunk_offset = struct.unpack_from('<IIi', self.data, offset)
+            # ROS time format: two uint32 (secs, nsecs) + uint32 offset
+            secs, nsecs, chunk_offset = struct.unpack_from('<III', self.data, offset)
             timestamp_ns = secs * NSEC_PER_SEC + nsecs
             result.append((timestamp_ns, chunk_offset))
         return result

src/pybag/bag_writer.py

@@ -146,7 +146,7 @@ def _encode_header_field(name: str, value: bytes) -> bytes:
         Format: field_len (4 bytes) | name=value
         """
         field_data = name.encode('ascii') + b'=' + value
-        return struct.pack('<i', len(field_data)) + field_data
+        return struct.pack('<I', len(field_data)) + field_data
 
     def _write_header(self) -> None:
         """Write the initial file header.
@@ -510,7 +510,7 @@ def _flush_chunk(self) -> None:
             for time_ns, offset in entries:
                 secs = time_ns // NSEC_PER_SEC
                 nsecs = time_ns % NSEC_PER_SEC
-                index_data_buffer.write(struct.pack('<IIi', secs, nsecs, offset))
+                index_data_buffer.write(struct.pack('<III', secs, nsecs, offset))
             # Write the index records
             index_record = IndexDataRecord(
                 ver=1,

tests/bag/test_bag_records.py

@@ -0,0 +1,38 @@
+import struct
+
+import pytest
+
+from pybag.bag.records import ChunkInfoRecord, IndexDataRecord
+
+
+def test_connection_counts_max_uint32(self):
+    """Test connection counts with maximum uint32 values."""
+    max_uint32 = 2**32 - 1  # 4294967295
+
+    data = struct.pack('<II', max_uint32, max_uint32)
+
+    record = ChunkInfoRecord(
+        ver=1,
+        chunk_pos=0,
+        start_time=0,
+        end_time=0,
+        count=1,
+        data=data,
+    )
+
+    counts = record.connection_counts
+    assert max_uint32 in counts
+    assert counts[max_uint32] == max_uint32
+
+
+def test_index_entries_max_offset(self):
+    """Test index entries with maximum uint32 offset."""
+    max_offset = 2**32 - 1  # 4294967295
+
+    data = struct.pack('<III', 0, 0, max_offset)
+
+    record = IndexDataRecord(ver=1, conn=0, count=1, data=data)
+    entries = record.entries
+
+    timestamp_ns, offset = entries[0]
+    assert offset == max_offset

tests/bag/test_rosmsg.py

@@ -1,7 +1,5 @@
 """Tests for ROS 1 message serialization (rosmsg format)."""
 
-import struct
-
 import pytest
 
 from pybag.encoding.rosmsg import RosMsgDecoder, RosMsgEncoder