data = open(r"C:/Users/vikto/Documents/Claude/rt-4d/radio-spi-dump.bin","rb").read()

def hexdump(off, n=128):
    for i in range(0, n, 16):
        chunk = data[off+i:off+i+16]
        h = ' '.join('%02x'%b for b in chunk)
        a = ''.join(chr(b) if 32<=b<127 else '.' for b in chunk)
        print('%08x  %-47s  %s'%(off+i, h, a))

# The transition zones between 0x80 runs in BLOB2 - dump around a boundary
print("### BLOB2 boundary at ~0x352845 (0x80 run ends here)")
hexdump(0x352840-0x20, 0x120)

print("\n### BLOB2 boundary at 0x353b25")
hexdump(0x353b00, 0x80)

# check byte histogram of a single 'record' 0x352000..0x353b25
from collections import Counter
c=Counter(data[0x352000:0x353b25])
print("\nBLOB2 record0 (0x352000-0x353b25) len", 0x353b25-0x352000, "hist top:", c.most_common(10))

# BLOB1 - the 14KB dense CJK. Check for structure: is it pure 2-byte GBK codes?
print("\n### BLOB1 0x198000 first bytes stats")
seg=data[0x198000:0x19b800]
# count high-bit-set bytes
hi=sum(1 for b in seg if b>=0x80)
print("len",len(seg),"high-bit bytes",hi, "ratio", hi/len(seg))
# Interpret as GB2312: pairs where b0 in 0xA1..0xF7
pairs=0; valid=0
for i in range(0,len(seg)-1,2):
    b0,b1=seg[i],seg[i+1]
    pairs+=1
    if 0xa1<=b0<=0xf7 and 0xa1<=b1<=0xfe:
        valid+=1
print("2-byte pairs",pairs,"valid GB2312 lvl",valid, "ratio",valid/pairs)