fix: Align CRC-16 calculation with ETSI EN 300 706 and improve retrieval

src/teletext/models.py

@@ -77,18 +77,20 @@ class Page:
 
     def calculate_crc(self) -> int:
         """
-        Calculates the CRC-16 (CCITT) checksum for the page.
-        According to ETSI EN 300 706 (Section 9.4.1.2 & Figure 13):
-        - Covers Row 0 columns 8-31 (Bytes 14-37 in packet, excluding header and clock).
-        - Covers Rows 1-25 columns 0-39 (Bytes 6-45 in packet).
-        - Total 1024 bytes (8192 bits).
-        - Data is 7-bit (stripped parity).
+        Calculates the CRC-16 checksum for the page.
+        According to ETSI EN 300 706 (Section 9.6.1 & Figure 13):
+        - G(x) = x^16 + x^12 + x^9 + x^7 + 1 (Poly 0x1281)
         - Initial value: 0.
+        - Processed bits b8 to b1 (MSB first for stored bytes).
+        - Total 1024 bytes (32 packets * 32 bytes).
+        - Packet X/0: Bytes 14 to 37 (24 bytes) + 8 spaces.
+        - Packets X/1 to X/25: Bytes 14 to 45 (32 bytes).
+        - Packets X/26 to X/31: 32 spaces each.
         """
         crc = 0
-        poly = 0x1021
+        poly = 0x1281
         
-        # Helper to update CRC with a byte
+        # Helper to update CRC with a byte (MSB first)
         def update_crc(c, val):
             v = (val << 8) & 0xFFFF
             for _ in range(8):
@@ -101,33 +103,39 @@ class Page:
             return c
 
         # Organize packets by row
-        rows = {}
-        for p in self.packets:
-            rows[p.row] = p
+        rows = {p.row: p for p in self.packets}
+            
+        for r in range(32): # Process 32 slots (0-31)
+            start, end = 0, 0
+            padding = 0
             
-        for r in range(26): # 0 to 25
-            # Determine column range
             if r == 0:
-                start_col, end_col = 8, 32 # Cols 8-31 (24 bytes)
+                # Row 0: Bytes 14-37 (24 bytes)
+                start, end = 8, 32 # data[8..31]
+                padding = 8 
+            elif 1 <= r <= 25:
+                # Rows 1-25: Bytes 14-45 (32 bytes)
+                start, end = 8, 40 # data[8..39]
+                padding = 0
             else:
-                start_col, end_col = 0, 40 # Cols 0-39 (40 bytes)
+                # Rows 26-31: 32 spaces each
+                padding = 32
                 
-            if r in rows:
-                data = rows[r].data
-                # Ensure data is long enough (should be 40)
-                d_len = len(data)
-                
-                for i in range(start_col, end_col):
-                    if i < d_len:
-                        byte_val = data[i] & 0x7F # Strip parity
-                    else:
-                        byte_val = 0x20 # Pad with space if short
+            # Process packet data if available
+            if r in rows and start < end:
+                p_data = rows[r].data
+                for i in range(start, end):
+                    byte_val = (p_data[i] & 0x7F) if i < len(p_data) else 0x20
                     crc = update_crc(crc, byte_val)
-            else:
-                # Missing row treated as spaces
-                for i in range(start_col, end_col):
+            elif start < end:
+                # Missing packet but slot exists
+                for _ in range(start, end):
                     crc = update_crc(crc, 0x20)
             
+            # Add padding for this slot
+            for _ in range(padding):
+                crc = update_crc(crc, 0x20)
+                    
         return crc
 
     def get_stored_crc(self) -> Optional[int]:
@@ -140,15 +148,16 @@ class Page:
             if p.row == 27:
                 # Check Designation Code (Byte 0)
                 try:
-                    b0 = p.data[0]
-                    # Decode Hamming 8/4
-                    designation = decode_hamming_8_4(b0)
+                    if len(p.data) >= 40:
+                        b0 = p.data[0]
+                        # Decode Hamming 8/4
+                        designation = decode_hamming_8_4(b0)
                         
-                    if designation == 0:
-                        # Packet 27/0
-                        # Checksum is in bytes 38 and 39.
-                        # ETSI EN 300 706: 8 bits of each byte are used.
-                        if len(p.data) >= 40:
+                        # Packets X/27/0 to X/27/3 exist, but only X/27/0 has the CRC.
+                        # We also check if b0 is raw 0 as a fallback for some captures.
+                        if designation == 0 or b0 == 0:
+                            # Packet 27/0
+                            # Checksum is in bytes 38 and 39 (TBytes 44 and 45).
                             hi = p.data[38]
                             lo = p.data[39]