fix: parser frame logic now tracks DLE state.

parsers/s3_parser.py

@@ -102,68 +102,113 @@ CRC_FUNCS = {
 
 def parse_frames(blob: bytes, trailer_len: int, etx_mode: str = ETX_MODE_AUTO) -> List[Frame]:
     frames: List[Frame] = []
-    i = 0
-    idx = 0
     n = len(blob)
 
-    def is_dle_seq(pos: int, second: int) -> bool:
-        return pos + 1 < n and blob[pos] == DLE and blob[pos + 1] == second
+    STATE_IDLE = 0
+    STATE_IN_FRAME = 1
+    STATE_AFTER_DLE = 2
 
-    # Auto-detect whether ETX is bare (logger-stripped) or DLE+ETX (raw wire).
+    state = STATE_IDLE
+    payload_raw = bytearray()
+    start_offset = 0
+    idx = 0
+    i = 0
+
+    # Auto-detect ETX style once
     if etx_mode == ETX_MODE_AUTO:
-        raw_etx = sum(1 for p in range(n - 1) if is_dle_seq(p, ETX))
-        stx_count = sum(1 for p in range(n - 1) if is_dle_seq(p, STX))
-        # Heuristic: the logger-stripped .bin files have plenty of STX but
-        # almost no DLE+ETX. If ETX count is far below STX count, assume stripped.
+        raw_etx = sum(1 for p in range(n - 1) if blob[p] == DLE and blob[p + 1] == ETX)
+        stx_count = sum(1 for p in range(n - 1) if blob[p] == DLE and blob[p + 1] == STX)
         if raw_etx and raw_etx >= max(1, int(0.8 * stx_count)):
             etx_mode = ETX_MODE_RAW
         else:
             etx_mode = ETX_MODE_STRIPPED
 
-    while i < n - 1:
-        if is_dle_seq(i, STX):
-            start = i
-            i += 2  # move past DLE STX
-            payload_start = i
+    while i < n:
+        b = blob[i]
 
-            # find end-of-frame marker
-            while i < n:
-                if etx_mode == ETX_MODE_RAW and is_dle_seq(i, ETX):
-                    payload_end = i  # up to (but not including) DLE ETX
-                    i += 2  # skip DLE ETX
-                    end = i
-                    break
-                if etx_mode == ETX_MODE_STRIPPED and blob[i] == ETX:
-                    payload_end = i  # up to (but not including) bare ETX
-                    i += 1  # skip ETX
-                    end = i
-                    break
-                i += 1
-
-            else:
-                # Ran off the end without finding ETX
-                break
-
-            payload_raw = blob[payload_start:payload_end]
-            payload = unescape_dle(payload_raw)
-
-            trailer = blob[i:i + trailer_len] if trailer_len > 0 else b""
-            i += trailer_len
-
-            frames.append(Frame(
-                index=idx,
-                start_offset=start,
-                end_offset=end,
-                payload_raw=payload_raw,
-                payload=payload,
-                trailer=trailer,
-                crc_match=None
-            ))
-            idx += 1
+        if state == STATE_IDLE:
+            if b == DLE and i + 1 < n and blob[i + 1] == STX:
+                start_offset = i
+                payload_raw = bytearray()
+                i += 2
+                state = STATE_IN_FRAME
+                continue
+            i += 1
             continue
 
-        # optional: you can also detect DLE EOT boundaries if useful later
-        i += 1
+        elif state == STATE_IN_FRAME:
+
+            # RAW mode: look for DLE+ETX
+            if etx_mode == ETX_MODE_RAW and b == DLE:
+                state = STATE_AFTER_DLE
+                i += 1
+                continue
+
+            # STRIPPED mode: bare ETX ends frame
+            if etx_mode == ETX_MODE_STRIPPED and b == ETX:
+                payload_end = i
+                i += 1
+                end_offset = i
+
+                payload = unescape_dle(bytes(payload_raw))
+                trailer = blob[i:i + trailer_len] if trailer_len > 0 else b""
+                i += trailer_len
+
+                frames.append(Frame(
+                    index=idx,
+                    start_offset=start_offset,
+                    end_offset=end_offset,
+                    payload_raw=bytes(payload_raw),
+                    payload=payload,
+                    trailer=trailer,
+                    crc_match=None
+                ))
+                idx += 1
+                state = STATE_IDLE
+                continue
+
+            payload_raw.append(b)
+            i += 1
+            continue
+
+        elif state == STATE_AFTER_DLE:
+
+            if etx_mode == ETX_MODE_RAW and b == ETX:
+                # Proper DLE ETX terminator
+                end_offset = i + 1
+                i += 1
+
+                payload = unescape_dle(bytes(payload_raw))
+                trailer = blob[i:i + trailer_len] if trailer_len > 0 else b""
+                i += trailer_len
+
+                frames.append(Frame(
+                    index=idx,
+                    start_offset=start_offset,
+                    end_offset=end_offset,
+                    payload_raw=bytes(payload_raw),
+                    payload=payload,
+                    trailer=trailer,
+                    crc_match=None
+                ))
+                idx += 1
+                state = STATE_IDLE
+                continue
+
+            elif b == DLE:
+                # Escaped DLE (10 10)
+                payload_raw.append(DLE)
+                state = STATE_IN_FRAME
+                i += 1
+                continue
+
+            else:
+                # False alarm — previous 10 was data
+                payload_raw.append(DLE)
+                payload_raw.append(b)
+                state = STATE_IN_FRAME
+                i += 1
+                continue
 
     return frames