Initial commit: Core Teletext Editor functionality

src/teletext/models.py

@@ -0,0 +1,93 @@
+from dataclasses import dataclass, field
+from typing import List, Optional
+
+@dataclass
+class Packet:
+    """
+    Represents a single Teletext packet (Row).
+    In T42, we have 42 bytes:
+    Byte 0-1: Magazine and Row Address (Hamming 8/4 encoding)
+    Byte 2-41: Data bytes (40 bytes)
+    """
+    original_data: bytes
+    magazine: int = field(init=False)
+    row: int = field(init=False)
+    data: bytearray = field(init=False)
+
+    def __post_init__(self):
+        if len(self.original_data) != 42:
+            raise ValueError(f"Packet must be 42 bytes, got {len(self.original_data)}")
+        
+        # Parse Magazine and Row from the first 2 bytes (Hamming 8/4)
+        # MRAG (Magazine + Row Address Group)
+        # Byte 0: P1 D1 P2 D2 P3 D3 P4 D4
+        # Byte 1: P1 D1 P2 D2 P3 D3 P4 D4
+        
+        b1 = self.original_data[0]
+        b2 = self.original_data[1]
+        
+        # De-interleave Hamming bits to get M (3 bits) and R (5 bits)
+        # This is the "basic" interpretation. 
+        # For a robust editor we assume the input T42 is valid or we just store bytes.
+        # But we need Mag/Row to organize pages.
+        
+        # Decode Hamming 8/4 logic is complex to implementation from scratch correctly 
+        # without a reference, but usually D1, D2, D3, D4 are at bit positions 1, 3, 5, 7 
+        # (0-indexed, where 0 is LSB). 
+        # Let's perform a simple extraction assuming no bit errors for now.
+        
+        def decode_hamming_8_4(byte_val):
+            # Extract data bits: bits 1, 3, 5, 7
+            return ((byte_val >> 1) & 1) | \
+                   (((byte_val >> 3) & 1) << 1) | \
+                   (((byte_val >> 5) & 1) << 2) | \
+                   (((byte_val >> 7) & 1) << 3)
+
+        d1 = decode_hamming_8_4(b1)
+        d2 = decode_hamming_8_4(b2)
+        
+        # Magazine is 3 bits (Logic is specific: Mag 8 is encoded as 0)
+        # Row is 5 bits.
+        
+        # According to Spec (ETSI EN 300 706):
+        # b1 encoded: M1 M2 M3 R1
+        # b2 encoded: R2 R3 R4 R5
+        
+        self.magazine = (d1 & 0b0111) 
+        if self.magazine == 0:
+            self.magazine = 8
+            
+        row_low_bit = (d1 >> 3) & 1
+        row_high_bits = d2
+        self.row = (row_high_bits << 1) | row_low_bit
+        
+        self.data = bytearray(self.original_data[2:])
+
+    @property
+    def is_header(self):
+        return self.row == 0
+
+@dataclass
+class Page:
+    """
+    Represents a Teletext Page (e.g., 100).
+    Can have multiple subpages.
+    """
+    magazine: int
+    page_number: int  # 00-99
+    sub_code: int = 0 # Subpage code (0000 to 3F7F hex usually, simplest is 0-99 equivalent)
+    packets: List[Packet] = field(default_factory=list)
+    
+    @property
+    def full_page_number(self):
+        return f"{self.magazine}{self.page_number:02d}"
+
+@dataclass
+class TeletextService:
+    """
+    Container for all pages.
+    """
+    pages: List[Page] = field(default_factory=list)
+    # We also keep a flat list of all packets to preserve order on save
+    all_packets: List[Packet] = field(default_factory=list)
+