Tasmota/lib/default/Unishox-Tasmota-1.0/src/unishox.cpp
2025-06-16 17:46:00 +02:00

574 lines
24 KiB
C++

/*
* Copyright (C) 2019 Siara Logics (cc)
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*
* @author Arundale R.
*
*/
/*
*
* This is a highly modified and optimized version of Unishox
* for Tasmota, aimed at compressing `Rules` which are typically
* short strings from 50 to 500 bytes.
*
* - moved to C++ (but still C-style)
* - c_95[] and l_95[] are pre-computed
* - all arrays in PROGMEM
* - removed all Unicode specific code to get code smaller, Unicode is rare in rules and encoded as pure binary
* - removed prev_lines management to reduce code size, we don't track previous encodings
* - using C++ const instead of #define
* - reusing the Unicode market to encode pure binary, which is 3 bits instead of 9
* - reverse binary encoding to 255-byte, favoring short encoding for values above 127, typical of Unicode
* - remove 2 bits encoding for Counts, since it could lead to a series of more than 8 consecutive 0-bits and output NULL char.
* Minimum encoding is 5 bits, which means spending 3+1=4 more bits for values in the range 0..3
* - removed CRLF encoding and reusing entry for RPT, saving 3 bits for repeats. Note: any CR will be binary encded
* - add safeguard to the output size (len_out), note that the compress buffer needs to be 4 bytes larger than actual compressed output.
* This is needed to avoid crash, since output can have ~30 bits
* - combined c_95[] and l_95[] to a single array to save space
* - Changed mapping of some characters in Set3, Set4 and Set4A, favoring frequent characters in rules and javascript
* - Added escape mechanism to ensure we never output NULL char. The marker is 0x2A which looked rare in preliminary tests
*
* @author Stephan Hadinger
*
*/
#include <time.h>
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <ctype.h>
#include <stdint.h>
#include <pgmspace.h>
#include "unishox.h"
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 };
// 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 };
enum {SHX_STATE_1 = 1, SHX_STATE_2}; // removed Unicode state
enum {SHX_SET1 = 0, SHX_SET1A, SHX_SET1B, SHX_SET2, SHX_SET3, SHX_SET4, SHX_SET4A};
// changed mapping in Set3, Set4, Set4A to accomodate frequencies in Rules and Javascript
static char sets[][11] PROGMEM =
{{ 0, ' ', 'e', 0, 't', 'a', 'o', 'i', 'n', 's', 'r'},
{ 0, 'l', 'c', 'd', 'h', 'u', 'p', 'm', 'b', 'g', 'w'},
{'f', 'y', 'v', 'k', 'q', 'j', 'x', 'z', 0, 0, 0},
{ 0, '9', '0', '1', '2', '3', '4', '5', '6', '7', '8'},
{'.', ',', '-', '/', '?', '+', ' ', '(', ')', '$', '@'},
{';', '#', ':', '<', '^', '*', '"', '{', '}', '[', ']'},
{'=', '%', '\'', '>', '&', '_', '!', '\\', '|', '~', '`'}};
// Decoder is designed for using less memory, not speed
// Decode lookup table for code index and length
// First 2 bits 00, Next 3 bits indicate index of code from 0,
// last 3 bits indicate code length in bits
// 0, 1, 2, 3, 4,
static char us_vcode[32] PROGMEM =
{2 + (0 << 3), 3 + (3 << 3), 3 + (1 << 3), 4 + (6 << 3), 0,
// 5, 6, 7, 8, 9, 10
4 + (4 << 3), 3 + (2 << 3), 4 + (8 << 3), 0, 0, 0,
// 11, 12, 13, 14, 15
4 + (7 << 3), 0, 4 + (5 << 3), 0, 5 + (9 << 3),
// 16, 17, 18, 19, 20, 21, 22, 23
0, 0, 0, 0, 0, 0, 0, 0,
// 24, 25, 26, 27, 28, 29, 30, 31
0, 0, 0, 0, 0, 0, 0, 5 + (10 << 3)};
// 0, 1, 2, 3, 4, 5, 6, 7,
static char us_hcode[32] PROGMEM =
{1 + (1 << 3), 2 + (0 << 3), 0, 3 + (2 << 3), 0, 0, 0, 5 + (3 << 3),
// 8, 9, 10, 11, 12, 13, 14, 15,
0, 0, 0, 0, 0, 0, 0, 5 + (5 << 3),
// 16, 17, 18, 19, 20, 21, 22, 23
0, 0, 0, 0, 0, 0, 0, 5 + (4 << 3),
// 24, 25, 26, 27, 28, 29, 30, 31
0, 0, 0, 0, 0, 0, 0, 5 + (6 << 3)};
static const char ESCAPE_MARKER = 0x2A; // Escape any null char
static const uint16_t TERM_CODE = 0x37C0; // 0b0011011111000000
static const uint16_t TERM_CODE_LEN = 10;
static const uint16_t DICT_CODE = 0x0000;
static const uint16_t DICT_CODE_LEN = 5;
static const uint16_t DICT_OTHER_CODE = 0x0000; // not used
static const uint16_t DICT_OTHER_CODE_LEN = 6;
static const uint16_t RPT_CODE_TASMOTA = 0x3780;
static const uint16_t RPT_CODE_TASMOTA_LEN = 10;
static const uint16_t BACK2_STATE1_CODE = 0x2000; // 0010 = back to lower case
static const uint16_t BACK2_STATE1_CODE_LEN = 4;
static const uint16_t BACK_FROM_UNI_CODE = 0xFE00;
static const uint16_t BACK_FROM_UNI_CODE_LEN = 8;
static const uint16_t LF_CODE = 0x3700;
static const uint16_t LF_CODE_LEN = 9;
static const uint16_t TAB_CODE = 0x2400;
static const uint16_t TAB_CODE_LEN = 7;
static const uint16_t ALL_UPPER_CODE = 0x2200;
static const uint16_t ALL_UPPER_CODE_LEN = 8;
static const uint16_t SW2_STATE2_CODE = 0x3800;
static const uint16_t SW2_STATE2_CODE_LEN = 7;
static const uint16_t ST2_SPC_CODE = 0x3B80;
static const uint16_t ST2_SPC_CODE_LEN = 11;
static const uint16_t BIN_CODE_TASMOTA = 0x8000;
static const uint16_t BIN_CODE_TASMOTA_LEN = 3;
#define NICE_LEN 5
// uint16_t mask[] PROGMEM = {0x8000, 0xC000, 0xE000, 0xF000, 0xF800, 0xFC00, 0xFE00, 0xFF00};
static const uint8_t mask[] PROGMEM = {0x80, 0xC0, 0xE0, 0xF0, 0xF8, 0xFC, 0xFE, 0xFF};
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;
if (state == SHX_STATE_2) {
// remove change state prefix
if ((code >> 9) == 0x1C) {
code <<= 7;
clen -= 7;
}
}
while (clen > 0) {
cur_bit = ol % 8;
blen = (clen > 8 ? 8 : clen);
a_byte = (code >> 8) & pgm_read_word(&mask[blen - 1]);
a_byte >>= cur_bit;
if (blen + cur_bit > 8)
blen = (8 - cur_bit);
if (out) { // if out == nullptr, then we are in dry-run mode
if (cur_bit == 0)
out[ol >> 3] = a_byte;
else
out[ol >> 3] |= a_byte;
}
code <<= blen;
ol += blen;
if ((out) && (0 == ol % 8)) { // if out == nullptr, dry-run mode. We miss the escaping of characters in the length
// we completed a full byte
char last_c = out[(ol / 8) - 1];
if ((0 == last_c) || (ESCAPE_MARKER == last_c)) {
out[ol >> 3] = 1 + last_c; // increment to 0x01 or 0x2B
out[(ol >>3) -1] = ESCAPE_MARKER; // replace old value with marker
ol += 8; // add one full byte
}
}
clen -= blen;
}
}
// First five bits are code and Last three bits of codes represent length
// removing last 2 bytes, unused, we will never have values above 600 bytes
// const byte codes[7] = {0x01, 0x82, 0xC3, 0xE5, 0xED, 0xF5, 0xFD};
// const byte bit_len[7] = {2, 5, 7, 9, 12, 16, 17};
// const uint16_t adder[7] = {0, 4, 36, 164, 676, 4772, 0};
byte codes[] PROGMEM = { 0x82, 0xC3, 0xE5, 0xED, 0xF5 };
byte bit_len[] PROGMEM = { 5, 7, 9, 12, 16 };
// uint16_t adder[7] PROGMEM = { 0, 32, 160, 672, 4768 }; // no more used
void Unishox::encodeCount(int32_t count) {
int till = 0;
int base = 0;
for (uint32_t i = 0; i < sizeof(bit_len); i++) {
uint32_t bit_len_i = pgm_read_byte(&bit_len[i]);
till += (1 << bit_len_i);
if (count < till) {
byte codes_i = pgm_read_byte(&codes[i]);
append_bits((codes_i & 0xF8) << 8, codes_i & 0x07);
// ol = append_bits(out, ol, (count - pgm_read_word(&adder[i])) << (16 - bit_len_i), bit_len_i, 1);
append_bits((count - base) << (16 - bit_len_i), bit_len_i);
return;
}
base = till;
}
return;
}
bool Unishox::matchOccurance(void) {
int32_t j, k;
uint32_t longest_dist = 0;
uint32_t longest_len = 0;
for (j = l - NICE_LEN; j >= 0; j--) {
for (k = l; k < len && j + k - l < l; k++) {
if (in[k] != in[j + k - l])
break;
}
if (k - l > NICE_LEN - 1) {
uint32_t match_len = k - l - NICE_LEN;
uint32_t match_dist = l - j - NICE_LEN + 1;
if (match_len > longest_len) {
longest_len = match_len;
longest_dist = match_dist;
}
}
}
if (longest_len) {
if (state == SHX_STATE_2 || is_all_upper) {
is_all_upper = 0;
state = SHX_STATE_1;
append_bits(BACK2_STATE1_CODE, BACK2_STATE1_CODE_LEN);
}
append_bits(DICT_CODE, DICT_CODE_LEN);
encodeCount(longest_len);
encodeCount(longest_dist);
l += longest_len + NICE_LEN - 1;
return true;
}
return false;
}
// Compress a buffer.
// Inputs:
// - in: non-null pointer to a buffer of bytes to be compressed. Progmem is not valid. Null bytes are valid.
// - len: size of the input buffer. 0 is valid for empty buffer
// - out: pointer to output buffer. out is nullptr, the compressor does a dry-run and reports the compressed size without writing bytes
// - len_out: length in bytes of the output buffer.
// Output:
// - if >= 0: size of the compressed buffer. The output buffer does not contain NULL bytes, and it is not NULL terminated
// - if < 0: an error occured, most certainly the output buffer was not large enough
int32_t Unishox::unishox_compress(const char *p_in, size_t p_len, char *p_out, size_t p_len_out) {
in = p_in;
len = p_len;
out = p_out;
len_out = p_len_out;
byte bits;
int ll;
char c_in, c_next;
byte is_upper;
ol = 0;
state = SHX_STATE_1;
is_all_upper = 0;
for (l=0; l<len; l++) {
c_in = in[l];
if (l && l < len - 4) {
if (c_in == in[l - 1] && c_in == in[l + 1] && c_in == in[l + 2] && c_in == in[l + 3]) { // check for repeat
int rpt_count = l + 4;
while (rpt_count < len && in[rpt_count] == c_in)
rpt_count++;
rpt_count -= l;
if (state == SHX_STATE_2 || is_all_upper) {
is_all_upper = 0;
state = SHX_STATE_1;
append_bits(BACK2_STATE1_CODE, BACK2_STATE1_CODE_LEN); // back to lower case and Set1
}
// ol = append_bits(out, ol, RPT_CODE, RPT_CODE_LEN, 1);
append_bits(RPT_CODE_TASMOTA, RPT_CODE_TASMOTA_LEN); // reusing CRLF for RPT
encodeCount(rpt_count - 4);
l += rpt_count - 1;
continue;
}
}
if (l < (len - NICE_LEN + 1)) {
if (matchOccurance()) {
continue;
}
}
if (state == SHX_STATE_2) { // if Set2
if ((c_in >= ' ' && c_in <= '@') ||
(c_in >= '[' && c_in <= '`') ||
(c_in >= '{' && c_in <= '~')) {
} else {
state = SHX_STATE_1; // back to Set1 and lower case
append_bits(BACK2_STATE1_CODE, BACK2_STATE1_CODE_LEN);
}
}
is_upper = 0;
if (c_in >= 'A' && c_in <= 'Z')
is_upper = 1;
else {
if (is_all_upper) {
is_all_upper = 0;
append_bits(BACK2_STATE1_CODE, BACK2_STATE1_CODE_LEN);
}
}
c_next = 0;
if (l+1 < len)
c_next = in[l+1];
if (c_in >= 32 && c_in <= 126) {
if (is_upper && !is_all_upper) {
for (ll=l+5; ll>=l && ll<len; ll--) {
if (in[ll] < 'A' || in[ll] > 'Z')
break;
}
if (ll == l-1) {
append_bits(ALL_UPPER_CODE, ALL_UPPER_CODE_LEN); // CapsLock
is_all_upper = 1;
}
}
if (state == SHX_STATE_1 && c_in >= '0' && c_in <= '9') {
append_bits(SW2_STATE2_CODE, SW2_STATE2_CODE_LEN); // Switch to sticky Set2
state = SHX_STATE_2;
}
c_in -= 32;
if (is_all_upper && is_upper)
c_in += 32;
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]);
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
} else if (c_in == '\t') {
append_bits(TAB_CODE, TAB_CODE_LEN); // TAB
} else {
append_bits(BIN_CODE_TASMOTA, BIN_CODE_TASMOTA_LEN); // Binary, we reuse the Unicode marker which 3 bits instead of 9
encodeCount((unsigned char) 255 - c_in);
}
// check that we have some headroom in the output buffer
if (ol / 8 >= len_out - 4) {
return -1; // we risk overflow and crash
}
}
bits = ol % 8;
if (bits) {
state = SHX_STATE_1;
append_bits(TERM_CODE, 8 - bits); // 0011 0111 1100 0000 TERM = 0011 0111 11
}
return ol / 8; // we already arrived to a byte boundary
// return ol/8+(ol%8?1:0);
}
uint32_t Unishox::getNextBit(void) {
if (8 == bit_no) {
if (byte_no >= len) {
in_eof = true;
return 1; // return only 1s, which appends 'r' in worst case
}
byte_in = pgm_read_byte(&in[byte_no++]);
if (ESCAPE_MARKER == byte_in) {
byte_in = pgm_read_byte(&in[byte_no++]) - 1; // we shouldn't need to test if byte_no >= len, because it should not be possible to end with ESCAPE_MARKER
}
bit_no = 0;
}
// Serial.printf("getNextBit %d\n", byte_in & (0x80 >> bit_no) ? 1 : 0);
return byte_in & (0x80 >> bit_no++) ? 1 : 0;
}
// Returns:
// 0..11
// or -1 if end of stream
int32_t Unishox::getCodeIdx(const char *code_type) {
int32_t code = 0;
int32_t count = 0;
do {
if (in_eof) return -1; // invalid state
code += getNextBit() << count;
count++;
uint8_t code_type_code = pgm_read_byte(&code_type[code]);
if (code_type_code && (code_type_code & 0x07) == count) {
return code_type_code >> 3;
}
} while (count < 5);
return -1; // skip if code not found
}
int32_t Unishox::getNumFromBits(uint32_t count) {
int ret = 0;
while (count--) {
ret += getNextBit() << count;
}
if (in_eof) return 0;
return ret;
}
// const byte bit_len[7] = {5, 2, 7, 9, 12, 16, 17};
// const uint16_t adder[7] = {4, 0, 36, 164, 676, 4772, 0};
// byte bit_len[7] PROGMEM = { 5, 7, 9, 12, 16 };
// byte bit_len_read[7] PROGMEM = {5, 2, 7, 9, 12, 16 };
// uint16_t adder_read[7] PROGMEM = {4, 0, 36, 164, 676, 4772, 0};
// uint16_t adder_read[] PROGMEM = {0, 0, 32, 160, 672, 4768 };
// byte bit_len[7] PROGMEM = { 5, 7, 9, 12, 16 };
// uint16_t adder_read[] PROGMEM = {0, 32, 160, 672, 4768 };
// Code size optimized, recalculate adder[] like in encodeCount
uint32_t Unishox::readCount(void) {
int32_t idx = getCodeIdx(us_hcode);
if ((1 == idx) || (idx >= sizeof(bit_len)) || (idx < 0)) return 0; // unsupported or end of stream
if (idx >= 1) idx--; // we skip v = 1 (code '0') since we no more accept 2 bits encoding
int base;
int till = 0;
byte bit_len_idx; // bit_len[0]
for (uint32_t i = 0; i <= idx; i++) {
base = till;
bit_len_idx = pgm_read_byte(&bit_len[i]);
till += (1 << bit_len_idx);
}
int count = getNumFromBits(bit_len_idx) + base;
return count;
}
void Unishox::decodeRepeat(void) {
uint32_t dict_len = readCount() + NICE_LEN;
uint32_t dist = readCount() + NICE_LEN - 1;
if (ol + dict_len <= len_out) {
memcpy(out + ol, out + ol - dist, dict_len);
ol += dict_len;
}
}
int32_t Unishox::unishox_decompress(const char *p_in, size_t p_len, char *p_out, size_t p_len_out) {
in = p_in;
len = p_len;
out = p_out;
len_out = p_len_out;
in_eof = false;
ol = 0;
bit_no = 8; // force load of first byte, pretending we expired the last one
byte_no = 0;
dstate = SHX_SET1;
is_all_upper = 0;
if (out) out[ol] = 0;
// while ((byte_no << 3) + bit_no - 8 < len) {
while (!in_eof) {
if (out && ol >= len_out) {
break;
}
int32_t h, v;
char c = 0;
byte is_upper = is_all_upper;
v = getCodeIdx(us_vcode); // read vCode
if (v < 0) break; // end of stream
h = dstate; // Set1 or Set2
if (v == 0) { // Switch which is common to Set1 and Set2, first entry
h = getCodeIdx(us_hcode); // read hCode
if (h < 0) break; // end of stream
if (h == SHX_SET1) { // target is Set1
if (dstate == SHX_SET1) { // Switch from Set1 to Set1 us UpperCase
if (is_all_upper) { // if CapsLock, then back to LowerCase
is_upper = is_all_upper = 0;
continue;
}
v = getCodeIdx(us_vcode); // read again vCode
if (v < 0) break; // end of stream
if (v == 0) {
h = getCodeIdx(us_hcode); // read second hCode
if (h < 0) break; // end of stream
if (h == SHX_SET1) { // If double Switch Set1, the CapsLock
is_all_upper = 1;
continue;
}
}
is_upper = 1; // anyways, still uppercase
} else {
dstate = SHX_SET1; // if Set was not Set1, switch to Set1
continue;
}
} else
if (h == SHX_SET2) { // If Set2, switch dstate to Set2
if (dstate == SHX_SET1) // TODO: is this test useful, there are only 2 states possible
dstate = SHX_SET2;
continue;
}
if (h != SHX_SET1) { // all other Sets (why not else)
v = getCodeIdx(us_vcode); // we changed set, now read vCode for char
if (v < 0) break; // end of stream
}
}
if (v == 0 && h == SHX_SET1A) {
if (is_upper) {
if (out) out[ol] = 255 - readCount(); // binary
ol++;
} else {
decodeRepeat(); // dist
}
continue;
}
if (h == SHX_SET1 && v == 3) {
// was Unicode, will do Binary instead
if (out) out[ol] = 255 - readCount(); // binary
ol++;
continue;
}
if (h < 7 && v < 11) // TODO: are these the actual limits? Not 11x7 ?
c = pgm_read_byte(&sets[h][v]);
if (c >= 'a' && c <= 'z') {
if (is_upper)
c -= 32; // go to UpperCase for letters
} else { // handle all other cases
if (is_upper && dstate == SHX_SET1 && v == 1)
c = '\t'; // If UpperCase Space, change to TAB
if (h == SHX_SET1B) {
if (8 == v) { // was LF or RPT, now only LF
if (out) out[ol] = '\n';
ol++;
continue;
}
if (9 == v) { // was CRLF, now RPT
uint32_t count = readCount() + 4;
if (out && ol + count >= len_out) {
return -1; // overflow
}
if (out) {
char rpt_c = out[ol - 1];
while (count--)
out[ol++] = rpt_c;
} else {
ol += count;
}
continue;
}
if (10 == v) {
break; // TERM, stop decoding
}
}
}
// Serial.printf(">>>>>>>>>>>>>>>>>>>>>> Out = %c\n", c);
if (out) out[ol] = c;
ol++;
}
if (out && ol > len_out) {
return -1; // overflow
} else {
return ol;
}
}