Fix corrupt compression of some uppercase letters

This commit is contained in:
Stephan Hadinger 2020-05-25 23:21:01 +02:00
parent 4e18240532
commit de63d7088b
2 changed files with 13 additions and 6 deletions

View File

@ -57,7 +57,9 @@
typedef unsigned char byte;
// we squeeze both c_95[] and l_95[] in a sinle array.
// c_95[] uses only the 3 upper nibbles (or 12 most signifcant bits), while the last nibble encodes length (3..13)
static uint16_t cl_95[95] PROGMEM = {0x4000 + 3, 0x3F80 + 11, 0x3D80 + 11, 0x3C80 + 10, 0x3BE0 + 12, 0x3E80 + 10, 0x3F40 + 11, 0x3EC0 + 10, 0x3BA0 + 11, 0x3BC0 + 11, 0x3D60 + 11, 0x3B60 + 11, 0x3A80 + 10, 0x3AC0 + 10, 0x3A00 + 9, 0x3B00 + 10, 0x38C0 + 10, 0x3900 + 10, 0x3940 + 11, 0x3960 + 11, 0x3980 + 11, 0x39A0 + 11, 0x39C0 + 11, 0x39E0 + 12, 0x39F0 + 12, 0x3880 + 10, 0x3CC0 + 10, 0x3C00 + 9, 0x3D00 + 10, 0x3E00 + 9, 0x3F00 + 10, 0x3B40 + 11, 0x3BF0 + 12, 0x2B00 + 8, 0x21C0 + 11, 0x20C0 + 10, 0x2100 + 10, 0x2600 + 7, 0x2300 + 11, 0x21E0 + 12, 0x2140 + 11, 0x2D00 + 8, 0x2358 + 13, 0x2340 + 12, 0x2080 + 10, 0x21A0 + 11, 0x2E00 + 8, 0x2C00 + 8, 0x2180 + 11, 0x2350 + 13, 0x2F80 + 9, 0x2F00 + 9, 0x2A00 + 8, 0x2160 + 11, 0x2330 + 12, 0x21F0 + 12, 0x2360 + 13, 0x2320 + 12, 0x2368 + 13, 0x3DE0 + 12, 0x3FA0 + 11, 0x3DF0 + 12, 0x3D40 + 11, 0x3F60 + 11, 0x3FF0 + 12, 0xB000 + 4, 0x1C00 + 7, 0x0C00 + 6, 0x1000 + 6, 0x6000 + 3, 0x3000 + 7, 0x1E00 + 8, 0x1400 + 7, 0xD000 + 4, 0x3580 + 9, 0x3400 + 8, 0x0800 + 6, 0x1A00 + 7, 0xE000 + 4, 0xC000 + 4, 0x1800 + 7, 0x3500 + 9, 0xF800 + 5, 0xF000 + 5, 0xA000 + 4, 0x1600 + 7, 0x3300 + 8, 0x1F00 + 8, 0x3600 + 9, 0x3200 + 8, 0x3680 + 9, 0x3DA0 + 11, 0x3FC0 + 11, 0x3DC0 + 11, 0x3FE0 + 12 };
// static uint16_t cl_95[95] PROGMEM = {0x4000 + 3, 0x3F80 + 11, 0x3D80 + 11, 0x3C80 + 10, 0x3BE0 + 12, 0x3E80 + 10, 0x3F40 + 11, 0x3EC0 + 10, 0x3BA0 + 11, 0x3BC0 + 11, 0x3D60 + 11, 0x3B60 + 11, 0x3A80 + 10, 0x3AC0 + 10, 0x3A00 + 9, 0x3B00 + 10, 0x38C0 + 10, 0x3900 + 10, 0x3940 + 11, 0x3960 + 11, 0x3980 + 11, 0x39A0 + 11, 0x39C0 + 11, 0x39E0 + 12, 0x39F0 + 12, 0x3880 + 10, 0x3CC0 + 10, 0x3C00 + 9, 0x3D00 + 10, 0x3E00 + 9, 0x3F00 + 10, 0x3B40 + 11, 0x3BF0 + 12, 0x2B00 + 8, 0x21C0 + 11, 0x20C0 + 10, 0x2100 + 10, 0x2600 + 7, 0x2300 + 11, 0x21E0 + 12, 0x2140 + 11, 0x2D00 + 8, 0x2358 + 13, 0x2340 + 12, 0x2080 + 10, 0x21A0 + 11, 0x2E00 + 8, 0x2C00 + 8, 0x2180 + 11, 0x2350 + 13, 0x2F80 + 9, 0x2F00 + 9, 0x2A00 + 8, 0x2160 + 11, 0x2330 + 12, 0x21F0 + 12, 0x2360 + 13, 0x2320 + 12, 0x2368 + 13, 0x3DE0 + 12, 0x3FA0 + 11, 0x3DF0 + 12, 0x3D40 + 11, 0x3F60 + 11, 0x3FF0 + 12, 0xB000 + 4, 0x1C00 + 7, 0x0C00 + 6, 0x1000 + 6, 0x6000 + 3, 0x3000 + 7, 0x1E00 + 8, 0x1400 + 7, 0xD000 + 4, 0x3580 + 9, 0x3400 + 8, 0x0800 + 6, 0x1A00 + 7, 0xE000 + 4, 0xC000 + 4, 0x1800 + 7, 0x3500 + 9, 0xF800 + 5, 0xF000 + 5, 0xA000 + 4, 0x1600 + 7, 0x3300 + 8, 0x1F00 + 8, 0x3600 + 9, 0x3200 + 8, 0x3680 + 9, 0x3DA0 + 11, 0x3FC0 + 11, 0x3DC0 + 11, 0x3FE0 + 12 };
// Patched, for len == 13, shift 1 bit right
static uint16_t cl_95[95] PROGMEM = {0x4000 + 3, 0x3F80 + 11, 0x3D80 + 11, 0x3C80 + 10, 0x3BE0 + 12, 0x3E80 + 10, 0x3F40 + 11, 0x3EC0 + 10, 0x3BA0 + 11, 0x3BC0 + 11, 0x3D60 + 11, 0x3B60 + 11, 0x3A80 + 10, 0x3AC0 + 10, 0x3A00 + 9, 0x3B00 + 10, 0x38C0 + 10, 0x3900 + 10, 0x3940 + 11, 0x3960 + 11, 0x3980 + 11, 0x39A0 + 11, 0x39C0 + 11, 0x39E0 + 12, 0x39F0 + 12, 0x3880 + 10, 0x3CC0 + 10, 0x3C00 + 9, 0x3D00 + 10, 0x3E00 + 9, 0x3F00 + 10, 0x3B40 + 11, 0x3BF0 + 12, 0x2B00 + 8, 0x21C0 + 11, 0x20C0 + 10, 0x2100 + 10, 0x2600 + 7, 0x2300 + 11, 0x21E0 + 12, 0x2140 + 11, 0x2D00 + 8, 0x46B0 + 13, 0x2340 + 12, 0x2080 + 10, 0x21A0 + 11, 0x2E00 + 8, 0x2C00 + 8, 0x2180 + 11, 0x46A0 + 13, 0x2F80 + 9, 0x2F00 + 9, 0x2A00 + 8, 0x2160 + 11, 0x2330 + 12, 0x21F0 + 12, 0x46C0 + 13, 0x2320 + 12, 0x46D0 + 13, 0x3DE0 + 12, 0x3FA0 + 11, 0x3DF0 + 12, 0x3D40 + 11, 0x3F60 + 11, 0x3FF0 + 12, 0xB000 + 4, 0x1C00 + 7, 0x0C00 + 6, 0x1000 + 6, 0x6000 + 3, 0x3000 + 7, 0x1E00 + 8, 0x1400 + 7, 0xD000 + 4, 0x3580 + 9, 0x3400 + 8, 0x0800 + 6, 0x1A00 + 7, 0xE000 + 4, 0xC000 + 4, 0x1800 + 7, 0x3500 + 9, 0xF800 + 5, 0xF000 + 5, 0xA000 + 4, 0x1600 + 7, 0x3300 + 8, 0x1F00 + 8, 0x3600 + 9, 0x3200 + 8, 0x3680 + 9, 0x3DA0 + 11, 0x3FC0 + 11, 0x3DC0 + 11, 0x3FE0 + 12 };
// Original version with c/l separate
// uint16_t c_95[95] PROGMEM = {0x4000, 0x3F80, 0x3D80, 0x3C80, 0x3BE0, 0x3E80, 0x3F40, 0x3EC0, 0x3BA0, 0x3BC0, 0x3D60, 0x3B60, 0x3A80, 0x3AC0, 0x3A00, 0x3B00, 0x38C0, 0x3900, 0x3940, 0x3960, 0x3980, 0x39A0, 0x39C0, 0x39E0, 0x39F0, 0x3880, 0x3CC0, 0x3C00, 0x3D00, 0x3E00, 0x3F00, 0x3B40, 0x3BF0, 0x2B00, 0x21C0, 0x20C0, 0x2100, 0x2600, 0x2300, 0x21E0, 0x2140, 0x2D00, 0x2358, 0x2340, 0x2080, 0x21A0, 0x2E00, 0x2C00, 0x2180, 0x2350, 0x2F80, 0x2F00, 0x2A00, 0x2160, 0x2330, 0x21F0, 0x2360, 0x2320, 0x2368, 0x3DE0, 0x3FA0, 0x3DF0, 0x3D40, 0x3F60, 0x3FF0, 0xB000, 0x1C00, 0x0C00, 0x1000, 0x6000, 0x3000, 0x1E00, 0x1400, 0xD000, 0x3580, 0x3400, 0x0800, 0x1A00, 0xE000, 0xC000, 0x1800, 0x3500, 0xF800, 0xF000, 0xA000, 0x1600, 0x3300, 0x1F00, 0x3600, 0x3200, 0x3680, 0x3DA0, 0x3FC0, 0x3DC0, 0x3FE0 };
// uint8_t l_95[95] PROGMEM = { 3, 11, 11, 10, 12, 10, 11, 10, 11, 11, 11, 11, 10, 10, 9, 10, 10, 10, 11, 11, 11, 11, 11, 12, 12, 10, 10, 9, 10, 9, 10, 11, 12, 8, 11, 10, 10, 7, 11, 12, 11, 8, 13, 12, 10, 11, 8, 8, 11, 13, 9, 9, 8, 11, 12, 12, 13, 12, 13, 12, 11, 12, 11, 11, 12, 4, 7, 6, 6, 3, 7, 8, 7, 4, 9, 8, 6, 7, 4, 4, 7, 9, 5, 5, 4, 7, 8, 8, 9, 8, 9, 11, 11, 11, 12 };
@ -156,7 +158,7 @@ static const uint8_t mask[] PROGMEM = {0x80, 0xC0, 0xE0, 0xF0, 0xF8, 0xFC, 0xFE,
void Unishox::append_bits(unsigned int code, int clen) {
// Serial.printf("append_bits code = 0x%08X, clen %d\n", code, clen);
byte cur_bit;
byte blen;
unsigned char a_byte;
@ -355,8 +357,14 @@ int32_t Unishox::unishox_compress(const char *p_in, size_t p_len, char *p_out, s
if (c_in == 0 && state == SHX_STATE_2)
append_bits(ST2_SPC_CODE, ST2_SPC_CODE_LEN); // space from Set2 ionstead of Set1
else {
// Serial.printf("Encode %c %d\n", c_in + 32, c_in);
uint16_t cl = pgm_read_word(&cl_95[c_in]);
append_bits(cl & 0xFFF0, cl & 0x000F);
uint16_t cl_code = cl & 0xFFF0;
uint8_t cl_len = cl & 0x000F;
if (13 == cl_len) {
cl_code >>= 1;
}
append_bits(cl_code, cl_len);
}
} else if (c_in == 10) {
append_bits(LF_CODE, LF_CODE_LEN); // LF
@ -394,6 +402,7 @@ uint32_t Unishox::getNextBit(void) {
}
bit_no = 0;
}
// Serial.printf("getNextBit %d\n", byte_in & (0x80 >> bit_no) ? 1 : 0);
return byte_in & (0x80 >> bit_no++) ? 1 : 0;
}
@ -566,6 +575,7 @@ int32_t Unishox::unishox_decompress(const char *p_in, size_t p_len, char *p_out,
}
}
}
// Serial.printf(">>>>>>>>>>>>>>>>>>>>>> Out = %c\n", c);
out[ol++] = c;
}

View File

@ -19,9 +19,6 @@
#ifndef unishox
#define unishox
extern int32_t unishox_compress(const char *in, size_t len, char *out, size_t len_out);
//extern int32_t unishox_decompress(const char *in, size_t len, char *out, size_t len_out);
class Unishox {
public: