Scaffolding: ZMK module structure, Kconfig, DTS, behavior driver
- zephyr/module.yml with dts_root for dt-bindings - Kconfig: STENO_ENGINE, dict selection (Plover/Lapwing/test), MPHF toggle, Unicode modes, history size, multi-stroke timeout - DTS behavior binding (one_param, steno key index) - behavior_steno.c: chord accumulation, all-up detection, multi-stroke buffering with timeout, star undo, formatter pipeline, 3-way dict dispatch (split/MPHF/simple trie) - dict_embed.S: .incbin from generated header path - steno_keys.h: 23-key layout, bit positions matching compiler
This commit is contained in:
parent
234bc8d731
commit
1ec204b845
33 changed files with 5443 additions and 0 deletions
271
src/behavior_steno.c
Normal file
271
src/behavior_steno.c
Normal file
|
|
@ -0,0 +1,271 @@
|
|||
/*
|
||||
* Copyright (c) 2024 Afiq Zudin Hadi
|
||||
* SPDX-License-Identifier: PolyForm-Noncommercial-1.0.0
|
||||
*/
|
||||
|
||||
#define DT_DRV_COMPAT zmk_behavior_steno_engine
|
||||
|
||||
#include <zephyr/device.h>
|
||||
#include <zephyr/kernel.h>
|
||||
#include <zephyr/logging/log.h>
|
||||
|
||||
#include <drivers/behavior.h>
|
||||
#include <zmk/behavior.h>
|
||||
|
||||
#include "output.h"
|
||||
#include "undo.h"
|
||||
#include "formatter.h"
|
||||
|
||||
#if IS_ENABLED(CONFIG_STENO_DICT_MPHF)
|
||||
#include "dict_mphf.h"
|
||||
#else
|
||||
#include "trie.h"
|
||||
#endif
|
||||
|
||||
LOG_MODULE_DECLARE(zmk, CONFIG_ZMK_LOG_LEVEL);
|
||||
|
||||
extern const uint8_t _steno_dict_start[];
|
||||
extern const uint8_t _steno_dict_end[];
|
||||
|
||||
#if IS_ENABLED(CONFIG_STENO_DICT_MPHF)
|
||||
static struct dict_mphf mphf_dict;
|
||||
#endif
|
||||
|
||||
#define STENO_MAX_MULTI 8
|
||||
#define STENO_MULTI_TIMEOUT_MS CONFIG_STENO_MULTI_STROKE_TIMEOUT_MS
|
||||
|
||||
static inline const char *dict_lookup(const uint32_t *strokes, uint8_t count)
|
||||
{
|
||||
#if IS_ENABLED(CONFIG_STENO_DICT_MPHF)
|
||||
return dict_mphf_lookup(&mphf_dict, strokes, count);
|
||||
#else
|
||||
return steno_trie_lookup(strokes, count);
|
||||
#endif
|
||||
}
|
||||
|
||||
static inline bool dict_has_prefix(const uint32_t *strokes, uint8_t count)
|
||||
{
|
||||
#if IS_ENABLED(CONFIG_STENO_DICT_MPHF)
|
||||
return (count == 1) ? dict_mphf_has_prefix(&mphf_dict, strokes[0]) : false;
|
||||
#else
|
||||
return steno_trie_has_prefix(strokes, count);
|
||||
#endif
|
||||
}
|
||||
|
||||
struct steno_state {
|
||||
uint32_t current_chord;
|
||||
uint8_t keys_held;
|
||||
uint32_t pending_strokes[STENO_MAX_MULTI];
|
||||
uint8_t stroke_count;
|
||||
struct k_work_delayable multi_timeout;
|
||||
};
|
||||
|
||||
static struct steno_state state;
|
||||
static struct steno_fmt_state fmt_state;
|
||||
static struct stroke_history undo_history;
|
||||
static bool dict_ready;
|
||||
|
||||
static void flush_strokes(void);
|
||||
static void multi_timeout_handler(struct k_work *work);
|
||||
|
||||
static void emit_formatted(const char *translation,
|
||||
const uint32_t *strokes, uint8_t stroke_count)
|
||||
{
|
||||
struct steno_fmt_result result;
|
||||
steno_fmt_process(&fmt_state, translation, &result);
|
||||
if (result.backspaces > 0) {
|
||||
steno_output_backspace(result.backspaces);
|
||||
}
|
||||
if (result.len > 0) {
|
||||
steno_output_send(result.text, result.len);
|
||||
}
|
||||
if (!result.is_command_only) {
|
||||
steno_undo_push(&undo_history, strokes, stroke_count,
|
||||
result.len + result.backspaces, 0, 0);
|
||||
}
|
||||
}
|
||||
|
||||
static void process_chord(void)
|
||||
{
|
||||
if (state.current_chord == 0) {
|
||||
return;
|
||||
}
|
||||
|
||||
/* Star-only stroke (bit 9) → undo */
|
||||
if (state.current_chord == (1U << 9)) {
|
||||
state.current_chord = 0;
|
||||
struct stroke_history_entry *ue = steno_undo_pop(&undo_history);
|
||||
if (ue) {
|
||||
steno_output_backspace(ue->output_len);
|
||||
}
|
||||
return;
|
||||
}
|
||||
|
||||
k_work_cancel_delayable(&state.multi_timeout);
|
||||
|
||||
if (state.stroke_count < STENO_MAX_MULTI) {
|
||||
state.pending_strokes[state.stroke_count] = state.current_chord;
|
||||
state.stroke_count++;
|
||||
} else {
|
||||
flush_strokes();
|
||||
state.pending_strokes[0] = state.current_chord;
|
||||
state.stroke_count = 1;
|
||||
}
|
||||
state.current_chord = 0;
|
||||
|
||||
if (!dict_ready) {
|
||||
LOG_WRN("steno dict not ready, flushing");
|
||||
flush_strokes();
|
||||
return;
|
||||
}
|
||||
|
||||
const char *translation = dict_lookup(
|
||||
state.pending_strokes, state.stroke_count);
|
||||
LOG_INF("steno lookup %u strokes → %s", state.stroke_count,
|
||||
translation ? translation : "(null)");
|
||||
|
||||
if (translation) {
|
||||
emit_formatted(translation, state.pending_strokes, state.stroke_count);
|
||||
state.stroke_count = 0;
|
||||
return;
|
||||
}
|
||||
|
||||
if (dict_has_prefix(state.pending_strokes, state.stroke_count)) {
|
||||
k_work_schedule(&state.multi_timeout,
|
||||
K_MSEC(STENO_MULTI_TIMEOUT_MS));
|
||||
return;
|
||||
}
|
||||
|
||||
if (state.stroke_count > 1) {
|
||||
uint32_t last = state.pending_strokes[state.stroke_count - 1];
|
||||
state.stroke_count--;
|
||||
|
||||
const char *partial = dict_lookup(
|
||||
state.pending_strokes, state.stroke_count);
|
||||
if (partial) {
|
||||
emit_formatted(partial, state.pending_strokes, state.stroke_count);
|
||||
}
|
||||
|
||||
state.pending_strokes[0] = last;
|
||||
state.stroke_count = 1;
|
||||
|
||||
const char *rest = dict_lookup(&last, 1);
|
||||
if (rest) {
|
||||
emit_formatted(rest, &last, 1);
|
||||
state.stroke_count = 0;
|
||||
}
|
||||
return;
|
||||
}
|
||||
|
||||
flush_strokes();
|
||||
}
|
||||
|
||||
static void flush_strokes(void)
|
||||
{
|
||||
LOG_DBG("Flushing %u untranslated strokes", state.stroke_count);
|
||||
state.stroke_count = 0;
|
||||
}
|
||||
|
||||
static void multi_timeout_handler(struct k_work *work)
|
||||
{
|
||||
ARG_UNUSED(work);
|
||||
|
||||
if (state.stroke_count == 0) {
|
||||
return;
|
||||
}
|
||||
|
||||
const char *translation = dict_lookup(
|
||||
state.pending_strokes, state.stroke_count);
|
||||
|
||||
if (translation) {
|
||||
emit_formatted(translation, state.pending_strokes, state.stroke_count);
|
||||
}
|
||||
state.stroke_count = 0;
|
||||
}
|
||||
|
||||
static int on_steno_binding_pressed(struct zmk_behavior_binding *binding,
|
||||
struct zmk_behavior_binding_event event)
|
||||
{
|
||||
uint32_t key_index = binding->param1;
|
||||
|
||||
if (key_index > 35) {
|
||||
return -EINVAL;
|
||||
}
|
||||
|
||||
state.current_chord |= (1U << key_index);
|
||||
state.keys_held++;
|
||||
|
||||
LOG_INF("steno press key=%u chord=0x%06X held=%u",
|
||||
key_index, state.current_chord, state.keys_held);
|
||||
|
||||
return ZMK_BEHAVIOR_OPAQUE;
|
||||
}
|
||||
|
||||
static int on_steno_binding_released(struct zmk_behavior_binding *binding,
|
||||
struct zmk_behavior_binding_event event)
|
||||
{
|
||||
if (state.keys_held > 0) {
|
||||
state.keys_held--;
|
||||
}
|
||||
|
||||
LOG_INF("steno release held=%u chord=0x%06X", state.keys_held, state.current_chord);
|
||||
|
||||
if (state.keys_held == 0 && state.current_chord != 0) {
|
||||
LOG_INF("steno all-up → process chord 0x%06X", state.current_chord);
|
||||
process_chord();
|
||||
}
|
||||
|
||||
return ZMK_BEHAVIOR_OPAQUE;
|
||||
}
|
||||
|
||||
static int behavior_steno_init(const struct device *dev)
|
||||
{
|
||||
ARG_UNUSED(dev);
|
||||
|
||||
state.current_chord = 0;
|
||||
state.keys_held = 0;
|
||||
state.stroke_count = 0;
|
||||
|
||||
steno_fmt_init(&fmt_state);
|
||||
steno_undo_init(&undo_history);
|
||||
k_work_init_delayable(&state.multi_timeout, multi_timeout_handler);
|
||||
|
||||
size_t dict_size = _steno_dict_end - _steno_dict_start;
|
||||
if (dict_size > 4) {
|
||||
int ret;
|
||||
#if IS_ENABLED(CONFIG_STENO_DICT_MPHF)
|
||||
ret = dict_mphf_init(&mphf_dict, _steno_dict_start, dict_size);
|
||||
#else
|
||||
ret = steno_trie_init(_steno_dict_start, dict_size);
|
||||
#endif
|
||||
if (ret == 0) {
|
||||
dict_ready = true;
|
||||
LOG_INF("Steno dict loaded (%u bytes)", (unsigned)dict_size);
|
||||
} else {
|
||||
LOG_ERR("Steno dict init failed: %d", ret);
|
||||
}
|
||||
} else {
|
||||
LOG_WRN("No steno dict embedded");
|
||||
}
|
||||
|
||||
LOG_INF("Steno engine initialized");
|
||||
return 0;
|
||||
}
|
||||
|
||||
static const struct behavior_driver_api steno_driver_api = {
|
||||
.locality = BEHAVIOR_LOCALITY_CENTRAL,
|
||||
.binding_pressed = on_steno_binding_pressed,
|
||||
.binding_released = on_steno_binding_released,
|
||||
};
|
||||
|
||||
#define STENO_INST(n) \
|
||||
BEHAVIOR_DT_INST_DEFINE(n, \
|
||||
behavior_steno_init, \
|
||||
NULL, \
|
||||
NULL, \
|
||||
NULL, \
|
||||
POST_KERNEL, \
|
||||
CONFIG_KERNEL_INIT_PRIORITY_DEFAULT, \
|
||||
&steno_driver_api);
|
||||
|
||||
DT_INST_FOREACH_STATUS_OKAY(STENO_INST)
|
||||
14
src/dict_embed.S
Normal file
14
src/dict_embed.S
Normal file
|
|
@ -0,0 +1,14 @@
|
|||
#include "steno_dict_path.h"
|
||||
|
||||
.section .rodata.steno_dict, "a", %progbits
|
||||
.global _steno_dict_start
|
||||
.global _steno_dict_end
|
||||
|
||||
.balign 4
|
||||
_steno_dict_start:
|
||||
#ifdef STENO_DICT_BIN_PATH
|
||||
.incbin STENO_DICT_BIN_PATH
|
||||
#else
|
||||
.byte 0x00, 0x00, 0x00, 0x00
|
||||
#endif
|
||||
_steno_dict_end:
|
||||
323
src/dict_mphf.c
Normal file
323
src/dict_mphf.c
Normal file
|
|
@ -0,0 +1,323 @@
|
|||
/**
|
||||
* MPHF dictionary lookup engine — implementation.
|
||||
*
|
||||
* All data is read directly from flash. No heap allocation.
|
||||
* String table is block-compressed with zlib; decompressed on-demand
|
||||
* into a static 4KB buffer.
|
||||
*
|
||||
* SPDX-License-Identifier: PolyForm-Noncommercial-1.0.0
|
||||
*/
|
||||
|
||||
#include "dict_mphf.h"
|
||||
#include <string.h>
|
||||
|
||||
#ifdef __ZEPHYR__
|
||||
#include <zephyr/sys/crc.h>
|
||||
#endif
|
||||
|
||||
/* ─── Minimal inflate for non-Zephyr (native tests) ─── */
|
||||
|
||||
#ifndef __ZEPHYR__
|
||||
#include <stdlib.h>
|
||||
|
||||
/* Use zlib on host for native tests */
|
||||
#ifdef HAS_ZLIB
|
||||
#include <zlib.h>
|
||||
static int block_inflate(const uint8_t *src, size_t src_len,
|
||||
uint8_t *dst, size_t dst_cap, size_t *dst_len)
|
||||
{
|
||||
uLongf out_len = dst_cap;
|
||||
int ret = uncompress(dst, &out_len, src, src_len);
|
||||
if (ret == Z_OK) {
|
||||
*dst_len = out_len;
|
||||
return 0;
|
||||
}
|
||||
return -1;
|
||||
}
|
||||
#else
|
||||
/* Minimal tinf inflate — bundled for host-only testing.
|
||||
* On Zephyr, we use the kernel's built-in zlib. */
|
||||
#include "tinf/tinf.h"
|
||||
static int block_inflate(const uint8_t *src, size_t src_len,
|
||||
uint8_t *dst, size_t dst_cap, size_t *dst_len)
|
||||
{
|
||||
unsigned int out_len = dst_cap;
|
||||
/* Skip 2-byte zlib header, strip 4-byte adler32 checksum */
|
||||
if (src_len < 6) return -1;
|
||||
int ret = tinf_uncompress(dst, &out_len, src + 2, src_len - 6);
|
||||
if (ret == 0) {
|
||||
*dst_len = out_len;
|
||||
return 0;
|
||||
}
|
||||
return -1;
|
||||
}
|
||||
#endif /* HAS_ZLIB */
|
||||
|
||||
#else /* __ZEPHYR__ */
|
||||
|
||||
#include <zephyr/sys/util.h>
|
||||
|
||||
/* Zephyr built-in zlib decompression */
|
||||
#if __has_include(<zephyr/lib/zlib/zlib.h>)
|
||||
#include <zephyr/lib/zlib/zlib.h>
|
||||
#elif __has_include(<zlib.h>)
|
||||
#include <zlib.h>
|
||||
#endif
|
||||
|
||||
static int block_inflate(const uint8_t *src, size_t src_len,
|
||||
uint8_t *dst, size_t dst_cap, size_t *dst_len)
|
||||
{
|
||||
/* Try Zephyr's tinycrypt/miniz or fall back to raw copy */
|
||||
#if defined(CONFIG_ZLIB)
|
||||
uLongf out_len = dst_cap;
|
||||
int ret = uncompress(dst, &out_len, src, src_len);
|
||||
if (ret == Z_OK) {
|
||||
*dst_len = out_len;
|
||||
return 0;
|
||||
}
|
||||
return -1;
|
||||
#else
|
||||
/* No zlib available — strings must be uncompressed (flags bit 0 clear) */
|
||||
(void)src; (void)src_len; (void)dst; (void)dst_cap; (void)dst_len;
|
||||
return -1;
|
||||
#endif
|
||||
}
|
||||
|
||||
#endif /* __ZEPHYR__ */
|
||||
|
||||
|
||||
/* ─── FNV-1a 32-bit hash ─── */
|
||||
|
||||
static uint32_t fnv1a_32(const uint8_t *data, size_t len)
|
||||
{
|
||||
uint32_t h = 0x811c9dc5u;
|
||||
for (size_t i = 0; i < len; i++) {
|
||||
h ^= data[i];
|
||||
h *= 0x01000193u;
|
||||
}
|
||||
return h;
|
||||
}
|
||||
|
||||
static uint32_t hash_key(const uint8_t *key, size_t key_len, uint32_t seed)
|
||||
{
|
||||
uint32_t h = 0x811c9dc5u;
|
||||
uint8_t seed_bytes[4];
|
||||
seed_bytes[0] = (uint8_t)(seed);
|
||||
seed_bytes[1] = (uint8_t)(seed >> 8);
|
||||
seed_bytes[2] = (uint8_t)(seed >> 16);
|
||||
seed_bytes[3] = (uint8_t)(seed >> 24);
|
||||
for (int i = 0; i < 4; i++) {
|
||||
h ^= seed_bytes[i];
|
||||
h *= 0x01000193u;
|
||||
}
|
||||
for (size_t i = 0; i < key_len; i++) {
|
||||
h ^= key[i];
|
||||
h *= 0x01000193u;
|
||||
}
|
||||
return h;
|
||||
}
|
||||
|
||||
/* ─── Bit-packed field reading ─── */
|
||||
|
||||
static uint32_t read_bits(const uint8_t *data, uint32_t bit_pos, uint8_t n_bits)
|
||||
{
|
||||
uint32_t value = 0;
|
||||
for (uint8_t i = 0; i < n_bits; i++) {
|
||||
uint32_t byte_idx = (bit_pos + i) / 8;
|
||||
uint8_t bit_idx = (bit_pos + i) % 8;
|
||||
if (data[byte_idx] & (1u << bit_idx)) {
|
||||
value |= (1u << i);
|
||||
}
|
||||
}
|
||||
return value;
|
||||
}
|
||||
|
||||
static inline uint32_t align4(uint32_t n)
|
||||
{
|
||||
return (n + 3u) & ~3u;
|
||||
}
|
||||
|
||||
/* ─── Init ─── */
|
||||
|
||||
int dict_mphf_init(struct dict_mphf *dict, const void *data, size_t len)
|
||||
{
|
||||
if (!dict || !data) return -1;
|
||||
if (len < sizeof(struct dict_mphf_header)) return -2;
|
||||
|
||||
const struct dict_mphf_header *hdr = (const struct dict_mphf_header *)data;
|
||||
if (hdr->magic != DICT_MPHF_MAGIC) return -3;
|
||||
if (hdr->version != DICT_MPHF_VERSION) return -4;
|
||||
|
||||
dict->header = hdr;
|
||||
const uint8_t *base = (const uint8_t *)data;
|
||||
uint32_t offset = sizeof(struct dict_mphf_header);
|
||||
|
||||
/* Displacements */
|
||||
dict->displacements = base + offset;
|
||||
uint32_t disp_bits_total = (uint32_t)hdr->bucket_count * hdr->disp_bits;
|
||||
dict->disp_section_len = align4((disp_bits_total + 7) / 8);
|
||||
offset += dict->disp_section_len;
|
||||
|
||||
/* Values */
|
||||
dict->values = base + offset;
|
||||
uint32_t val_bits_total = (uint32_t)hdr->entry_count * hdr->value_bits;
|
||||
dict->val_section_len = align4((val_bits_total + 7) / 8);
|
||||
offset += dict->val_section_len;
|
||||
|
||||
/* Fingerprints */
|
||||
dict->fingerprints = base + offset;
|
||||
dict->fp_section_len = align4(hdr->entry_count);
|
||||
offset += dict->fp_section_len;
|
||||
|
||||
/* String offsets (u24 LE, 3 bytes each) */
|
||||
dict->string_offsets = base + offset;
|
||||
offset += hdr->unique_count * 3;
|
||||
|
||||
/* String data section */
|
||||
dict->str_data_start = base + offset;
|
||||
|
||||
if (hdr->flags & DICT_MPHF_FLAG_COMPRESSED) {
|
||||
/* Block directory: u16 block_count + u32[] offsets */
|
||||
dict->block_count = dict->str_data_start[0] |
|
||||
((uint16_t)dict->str_data_start[1] << 8);
|
||||
dict->block_dir = (const uint32_t *)(dict->str_data_start + 2);
|
||||
dict->blocks_start = dict->str_data_start + 2 + dict->block_count * 4;
|
||||
} else {
|
||||
dict->block_count = 0;
|
||||
dict->block_dir = NULL;
|
||||
dict->blocks_start = dict->str_data_start;
|
||||
}
|
||||
|
||||
/* Prefix table at end */
|
||||
uint32_t prefix_bytes = (uint32_t)hdr->prefix_count * 4;
|
||||
if (len >= prefix_bytes) {
|
||||
dict->prefix_table = (const uint32_t *)(base + len - prefix_bytes);
|
||||
} else {
|
||||
dict->prefix_table = NULL;
|
||||
}
|
||||
|
||||
return 0;
|
||||
}
|
||||
|
||||
/* ─── String decompression ─── */
|
||||
|
||||
static const char *resolve_string(const struct dict_mphf *dict, uint32_t val_id)
|
||||
{
|
||||
const uint8_t *off_ptr = dict->string_offsets + val_id * 3;
|
||||
uint32_t str_offset = (uint32_t)off_ptr[0]
|
||||
| ((uint32_t)off_ptr[1] << 8)
|
||||
| ((uint32_t)off_ptr[2] << 16);
|
||||
|
||||
if (!(dict->header->flags & DICT_MPHF_FLAG_COMPRESSED)) {
|
||||
return (const char *)(dict->str_data_start + str_offset);
|
||||
}
|
||||
|
||||
/* Block-compressed: decompress the right block */
|
||||
uint32_t block_idx = str_offset / DICT_MPHF_BLOCK_SIZE;
|
||||
uint32_t in_block_off = str_offset % DICT_MPHF_BLOCK_SIZE;
|
||||
|
||||
if (block_idx >= dict->block_count) {
|
||||
return NULL;
|
||||
}
|
||||
|
||||
/* Get compressed block bounds */
|
||||
uint32_t blk_start = dict->block_dir[block_idx];
|
||||
uint32_t blk_end;
|
||||
if (block_idx + 1 < dict->block_count) {
|
||||
blk_end = dict->block_dir[block_idx + 1];
|
||||
} else {
|
||||
/* Last block: extends to prefix_table or end of file */
|
||||
blk_end = (const uint8_t *)dict->prefix_table - dict->blocks_start;
|
||||
}
|
||||
|
||||
const uint8_t *compressed = dict->blocks_start + blk_start;
|
||||
uint32_t compressed_len = blk_end - blk_start;
|
||||
|
||||
static uint8_t decomp_buf[DICT_MPHF_BLOCK_SIZE];
|
||||
static uint32_t cached_block = UINT32_MAX;
|
||||
static size_t cached_len;
|
||||
|
||||
if (cached_block != block_idx) {
|
||||
size_t out_len;
|
||||
if (block_inflate(compressed, compressed_len,
|
||||
decomp_buf, sizeof(decomp_buf), &out_len) != 0) {
|
||||
return NULL;
|
||||
}
|
||||
cached_block = block_idx;
|
||||
cached_len = out_len;
|
||||
}
|
||||
|
||||
if (in_block_off >= cached_len) {
|
||||
return NULL;
|
||||
}
|
||||
|
||||
return (const char *)(decomp_buf + in_block_off);
|
||||
}
|
||||
|
||||
/* ─── Lookup ─── */
|
||||
|
||||
const char *dict_mphf_lookup(const struct dict_mphf *dict,
|
||||
const uint32_t *strokes, uint8_t count)
|
||||
{
|
||||
if (!dict || !dict->header || !strokes || count == 0) {
|
||||
return NULL;
|
||||
}
|
||||
|
||||
const struct dict_mphf_header *hdr = dict->header;
|
||||
|
||||
uint8_t key_buf[32];
|
||||
size_t key_len = (size_t)count * 4;
|
||||
if (key_len > sizeof(key_buf)) {
|
||||
return NULL;
|
||||
}
|
||||
|
||||
for (uint8_t i = 0; i < count; i++) {
|
||||
key_buf[i * 4 + 0] = (uint8_t)(strokes[i]);
|
||||
key_buf[i * 4 + 1] = (uint8_t)(strokes[i] >> 8);
|
||||
key_buf[i * 4 + 2] = (uint8_t)(strokes[i] >> 16);
|
||||
key_buf[i * 4 + 3] = (uint8_t)(strokes[i] >> 24);
|
||||
}
|
||||
|
||||
uint32_t bucket = hash_key(key_buf, key_len, 0) % hdr->bucket_count;
|
||||
uint32_t d = read_bits(dict->displacements,
|
||||
bucket * (uint32_t)hdr->disp_bits,
|
||||
hdr->disp_bits);
|
||||
uint32_t slot = hash_key(key_buf, key_len, d + 1) % hdr->entry_count;
|
||||
|
||||
uint8_t expected_fp = (uint8_t)(fnv1a_32(key_buf, key_len) & 0xFF);
|
||||
if (dict->fingerprints[slot] != expected_fp) {
|
||||
return NULL;
|
||||
}
|
||||
|
||||
uint32_t val_id = read_bits(dict->values,
|
||||
slot * (uint32_t)hdr->value_bits,
|
||||
hdr->value_bits);
|
||||
if (val_id >= hdr->unique_count) {
|
||||
return NULL;
|
||||
}
|
||||
|
||||
return resolve_string(dict, val_id);
|
||||
}
|
||||
|
||||
/* ─── has_prefix ─── */
|
||||
|
||||
bool dict_mphf_has_prefix(const struct dict_mphf *dict, uint32_t stroke)
|
||||
{
|
||||
if (!dict || !dict->header || !dict->prefix_table ||
|
||||
dict->header->prefix_count == 0) {
|
||||
return false;
|
||||
}
|
||||
|
||||
uint32_t lo = 0;
|
||||
uint32_t hi = dict->header->prefix_count;
|
||||
|
||||
while (lo < hi) {
|
||||
uint32_t mid = lo + (hi - lo) / 2;
|
||||
uint32_t val = dict->prefix_table[mid];
|
||||
if (val == stroke) return true;
|
||||
if (val < stroke) lo = mid + 1;
|
||||
else hi = mid;
|
||||
}
|
||||
|
||||
return false;
|
||||
}
|
||||
69
src/dict_mphf.h
Normal file
69
src/dict_mphf.h
Normal file
|
|
@ -0,0 +1,69 @@
|
|||
/**
|
||||
* MPHF (Minimal Perfect Hash Function) dictionary lookup engine.
|
||||
*
|
||||
* Binary format v2: CHD MPHF + bit-packed displacements/values +
|
||||
* fingerprinted verification + block-compressed string table.
|
||||
*
|
||||
* SPDX-License-Identifier: PolyForm-Noncommercial-1.0.0
|
||||
*/
|
||||
|
||||
#ifndef DICT_MPHF_H
|
||||
#define DICT_MPHF_H
|
||||
|
||||
#include <stdint.h>
|
||||
#include <stdbool.h>
|
||||
#include <stddef.h>
|
||||
|
||||
#define DICT_MPHF_MAGIC 0x4F4E5453 /* "STNO" */
|
||||
#define DICT_MPHF_VERSION 2
|
||||
|
||||
#define DICT_MPHF_FLAG_COMPRESSED 0x0001
|
||||
|
||||
#define DICT_MPHF_BLOCK_SIZE 4096
|
||||
|
||||
struct dict_mphf_header {
|
||||
uint32_t magic;
|
||||
uint16_t version;
|
||||
uint16_t flags;
|
||||
uint32_t entry_count;
|
||||
uint32_t bucket_count;
|
||||
uint32_t unique_count;
|
||||
uint8_t value_bits;
|
||||
uint8_t disp_bits;
|
||||
uint16_t prefix_count;
|
||||
uint32_t reserved0;
|
||||
uint32_t reserved1;
|
||||
} __attribute__((packed));
|
||||
|
||||
_Static_assert(sizeof(struct dict_mphf_header) == 32, "header must be 32 bytes");
|
||||
|
||||
struct dict_mphf {
|
||||
const struct dict_mphf_header *header;
|
||||
const uint8_t *displacements;
|
||||
const uint8_t *values;
|
||||
const uint8_t *fingerprints;
|
||||
const uint8_t *string_offsets;
|
||||
const uint8_t *str_data_start; /* start of string data section */
|
||||
const uint32_t *prefix_table;
|
||||
uint32_t disp_section_len;
|
||||
uint32_t val_section_len;
|
||||
uint32_t fp_section_len;
|
||||
uint16_t block_count;
|
||||
const uint32_t *block_dir; /* block offset directory */
|
||||
const uint8_t *blocks_start; /* start of compressed blocks */
|
||||
uint32_t str_data_len; /* total string data section length */
|
||||
};
|
||||
|
||||
int dict_mphf_init(struct dict_mphf *dict, const void *data, size_t len);
|
||||
|
||||
const char *dict_mphf_lookup(const struct dict_mphf *dict,
|
||||
const uint32_t *strokes, uint8_t count);
|
||||
|
||||
bool dict_mphf_has_prefix(const struct dict_mphf *dict, uint32_t stroke);
|
||||
|
||||
static inline uint32_t dict_mphf_count(const struct dict_mphf *dict)
|
||||
{
|
||||
return dict->header->entry_count;
|
||||
}
|
||||
|
||||
#endif /* DICT_MPHF_H */
|
||||
335
src/formatter.c
Normal file
335
src/formatter.c
Normal file
|
|
@ -0,0 +1,335 @@
|
|||
/*
|
||||
* Copyright (c) 2024 Afiq Zudin Hadi
|
||||
* SPDX-License-Identifier: PolyForm-Noncommercial-1.0.0
|
||||
*
|
||||
* Plover formatting engine — parses translation strings containing
|
||||
* {commands} and produces clean output with spacing/capitalization.
|
||||
* No malloc — all stack/static.
|
||||
*/
|
||||
|
||||
#include "formatter.h"
|
||||
#include <string.h>
|
||||
|
||||
/* ── helpers ────────────────────────────────────────────────────── */
|
||||
|
||||
static inline bool is_upper(char c) { return c >= 'A' && c <= 'Z'; }
|
||||
static inline bool is_lower(char c) { return c >= 'a' && c <= 'z'; }
|
||||
static inline char to_upper(char c) { return is_lower(c) ? (char)(c - 32) : c; }
|
||||
static inline char to_lower(char c) { return is_upper(c) ? (char)(c + 32) : c; }
|
||||
|
||||
static bool str_eq(const char *a, const char *b, size_t n)
|
||||
{
|
||||
for (size_t i = 0; i < n; i++) {
|
||||
if (a[i] != b[i]) return false;
|
||||
}
|
||||
return true;
|
||||
}
|
||||
|
||||
static bool starts_with(const char *s, size_t slen, const char *prefix)
|
||||
{
|
||||
size_t plen = strlen(prefix);
|
||||
if (slen < plen) return false;
|
||||
return str_eq(s, prefix, plen);
|
||||
}
|
||||
|
||||
/* ── output buffer append ───────────────────────────────────────── */
|
||||
|
||||
static void emit_char(struct steno_fmt_result *r, char c)
|
||||
{
|
||||
if (r->len < STENO_FMT_MAX_OUTPUT - 1) {
|
||||
r->text[r->len++] = c;
|
||||
}
|
||||
}
|
||||
|
||||
/* ── apply capitalization/mode transforms to a text segment ───── */
|
||||
|
||||
static void emit_text_transformed(struct steno_fmt_state *state,
|
||||
struct steno_fmt_result *r,
|
||||
const char *text, size_t len)
|
||||
{
|
||||
if (len == 0) return;
|
||||
|
||||
for (size_t i = 0; i < len; i++) {
|
||||
char c = text[i];
|
||||
|
||||
/* Mode transforms */
|
||||
if (state->mode == STENO_MODE_CAPS) {
|
||||
c = to_upper(c);
|
||||
} else if (state->mode == STENO_MODE_LOWER) {
|
||||
c = to_lower(c);
|
||||
} else if (state->mode == STENO_MODE_TITLE) {
|
||||
/* Capitalize first letter of each word */
|
||||
if (i == 0 || (i > 0 && text[i - 1] == ' ')) {
|
||||
c = to_upper(c);
|
||||
}
|
||||
}
|
||||
|
||||
/* One-shot transforms (first char only) */
|
||||
if (i == 0) {
|
||||
if (state->cap_next) {
|
||||
c = to_upper(c);
|
||||
state->cap_next = false;
|
||||
}
|
||||
if (state->upper_next) {
|
||||
/* uppercase entire word — handled below */
|
||||
}
|
||||
if (state->lower_next) {
|
||||
/* lowercase entire word — handled below */
|
||||
}
|
||||
}
|
||||
|
||||
/* upper_next: entire word */
|
||||
if (state->upper_next) {
|
||||
c = to_upper(c);
|
||||
}
|
||||
|
||||
/* lower_next: entire word */
|
||||
if (state->lower_next) {
|
||||
c = to_lower(c);
|
||||
}
|
||||
|
||||
emit_char(r, c);
|
||||
}
|
||||
|
||||
state->upper_next = false;
|
||||
state->lower_next = false;
|
||||
}
|
||||
|
||||
/* ── prepend space if needed ────────────────────────────────────── */
|
||||
|
||||
static void maybe_space(struct steno_fmt_state *state,
|
||||
struct steno_fmt_result *r)
|
||||
{
|
||||
if (state->space_pending && !state->suppress_space) {
|
||||
emit_char(r, ' ');
|
||||
}
|
||||
state->suppress_space = false;
|
||||
}
|
||||
|
||||
/* ── process a single {…} command ──────────────────────────────── */
|
||||
|
||||
static void process_command(struct steno_fmt_state *state,
|
||||
struct steno_fmt_result *r,
|
||||
const char *cmd, size_t cmd_len)
|
||||
{
|
||||
/* {^} — suppress space */
|
||||
if (cmd_len == 1 && cmd[0] == '^') {
|
||||
state->suppress_space = true;
|
||||
return;
|
||||
}
|
||||
|
||||
/* {^suffix} — attach suffix */
|
||||
if (cmd_len > 1 && cmd[0] == '^') {
|
||||
state->suppress_space = true;
|
||||
maybe_space(state, r);
|
||||
emit_text_transformed(state, r, cmd + 1, cmd_len - 1);
|
||||
state->space_pending = true;
|
||||
state->glue = false;
|
||||
return;
|
||||
}
|
||||
|
||||
/* {prefix^} — attach prefix */
|
||||
if (cmd_len > 1 && cmd[cmd_len - 1] == '^') {
|
||||
maybe_space(state, r);
|
||||
emit_text_transformed(state, r, cmd, cmd_len - 1);
|
||||
state->suppress_space = true;
|
||||
state->space_pending = false;
|
||||
state->glue = false;
|
||||
return;
|
||||
}
|
||||
|
||||
/* {-|} — capitalize next */
|
||||
if (cmd_len == 2 && cmd[0] == '-' && cmd[1] == '|') {
|
||||
state->cap_next = true;
|
||||
return;
|
||||
}
|
||||
|
||||
/* {~|} — carry capitalize */
|
||||
if (cmd_len == 2 && cmd[0] == '~' && cmd[1] == '|') {
|
||||
state->cap_next = true;
|
||||
return;
|
||||
}
|
||||
|
||||
/* {*-|} — retro capitalize (backspace + re-emit) */
|
||||
if (cmd_len == 3 && cmd[0] == '*' && cmd[1] == '-' && cmd[2] == '|') {
|
||||
/* Signal retro-capitalize — simplified: set flag, caller handles */
|
||||
r->backspaces = 1;
|
||||
return;
|
||||
}
|
||||
|
||||
/* {*!} — retro delete space */
|
||||
if (cmd_len == 2 && cmd[0] == '*' && cmd[1] == '!') {
|
||||
r->backspaces = 1;
|
||||
return;
|
||||
}
|
||||
|
||||
/* {*?} — retro insert space */
|
||||
if (cmd_len == 2 && cmd[0] == '*' && cmd[1] == '?') {
|
||||
/* Would need undo context; simplified signal */
|
||||
return;
|
||||
}
|
||||
|
||||
/* {*} — undo */
|
||||
if (cmd_len == 1 && cmd[0] == '*') {
|
||||
r->is_undo = true;
|
||||
r->is_command_only = true;
|
||||
return;
|
||||
}
|
||||
|
||||
/* {.} {,} {?} {!} {;} {:} — punctuation: attach to prev, cap next for sentence-enders */
|
||||
if (cmd_len == 1 && (cmd[0] == '.' || cmd[0] == ',' ||
|
||||
cmd[0] == '?' || cmd[0] == '!' ||
|
||||
cmd[0] == ';' || cmd[0] == ':')) {
|
||||
state->suppress_space = true;
|
||||
maybe_space(state, r);
|
||||
emit_char(r, cmd[0]);
|
||||
state->space_pending = true;
|
||||
state->suppress_space = false;
|
||||
/* Sentence-ending punctuation capitalizes next */
|
||||
if (cmd[0] == '.' || cmd[0] == '?' || cmd[0] == '!') {
|
||||
state->cap_next = true;
|
||||
}
|
||||
state->glue = false;
|
||||
return;
|
||||
}
|
||||
|
||||
/* {#...} — key combo (emit nothing for now, could extend) */
|
||||
if (cmd_len > 0 && cmd[0] == '#') {
|
||||
r->is_command_only = true;
|
||||
return;
|
||||
}
|
||||
|
||||
/* {&letter} — fingerspelling */
|
||||
if (cmd_len >= 2 && cmd[0] == '&') {
|
||||
/* Glue: no space between glue strokes */
|
||||
if (state->glue) {
|
||||
state->suppress_space = true;
|
||||
}
|
||||
maybe_space(state, r);
|
||||
emit_text_transformed(state, r, cmd + 1, cmd_len - 1);
|
||||
state->space_pending = true;
|
||||
state->glue = true;
|
||||
state->suppress_space = false;
|
||||
return;
|
||||
}
|
||||
|
||||
/* {&} — bare glue */
|
||||
if (cmd_len == 1 && cmd[0] == '&') {
|
||||
state->glue = true;
|
||||
state->suppress_space = true;
|
||||
return;
|
||||
}
|
||||
|
||||
/* {MODE:CAPS} {MODE:TITLE} {MODE:LOWER} {MODE:RESET} */
|
||||
if (starts_with(cmd, cmd_len, "MODE:")) {
|
||||
const char *mode_str = cmd + 5;
|
||||
size_t mode_len = cmd_len - 5;
|
||||
if (mode_len == 4 && str_eq(mode_str, "CAPS", 4)) {
|
||||
state->mode = STENO_MODE_CAPS;
|
||||
} else if (mode_len == 5 && str_eq(mode_str, "TITLE", 5)) {
|
||||
state->mode = STENO_MODE_TITLE;
|
||||
} else if (mode_len == 5 && str_eq(mode_str, "LOWER", 5)) {
|
||||
state->mode = STENO_MODE_LOWER;
|
||||
} else if (mode_len == 5 && str_eq(mode_str, "RESET", 5)) {
|
||||
state->mode = STENO_MODE_NORMAL;
|
||||
}
|
||||
return;
|
||||
}
|
||||
|
||||
/* {<} — uppercase next word */
|
||||
if (cmd_len == 1 && cmd[0] == '<') {
|
||||
state->upper_next = true;
|
||||
return;
|
||||
}
|
||||
|
||||
/* {>} — lowercase next word */
|
||||
if (cmd_len == 1 && cmd[0] == '>') {
|
||||
state->lower_next = true;
|
||||
return;
|
||||
}
|
||||
}
|
||||
|
||||
/* ── main entry point ──────────────────────────────────────────── */
|
||||
|
||||
void steno_fmt_init(struct steno_fmt_state *state)
|
||||
{
|
||||
memset(state, 0, sizeof(*state));
|
||||
state->space_pending = false;
|
||||
state->cap_next = false;
|
||||
state->upper_next = false;
|
||||
state->lower_next = false;
|
||||
state->suppress_space = false;
|
||||
state->glue = false;
|
||||
state->mode = STENO_MODE_NORMAL;
|
||||
}
|
||||
|
||||
void steno_fmt_process(struct steno_fmt_state *state,
|
||||
const char *translation,
|
||||
struct steno_fmt_result *result)
|
||||
{
|
||||
memset(result, 0, sizeof(*result));
|
||||
|
||||
if (!translation || !translation[0]) {
|
||||
result->is_command_only = true;
|
||||
return;
|
||||
}
|
||||
|
||||
size_t tlen = strlen(translation);
|
||||
const char *p = translation;
|
||||
const char *end = translation + tlen;
|
||||
bool emitted_text = false;
|
||||
bool had_command = false;
|
||||
bool this_glue = false;
|
||||
|
||||
while (p < end) {
|
||||
/* Escaped braces: \{ \} */
|
||||
if (p[0] == '\\' && p + 1 < end && (p[1] == '{' || p[1] == '}')) {
|
||||
maybe_space(state, result);
|
||||
emit_text_transformed(state, result, p + 1, 1);
|
||||
state->space_pending = true;
|
||||
state->glue = false;
|
||||
emitted_text = true;
|
||||
p += 2;
|
||||
continue;
|
||||
}
|
||||
|
||||
/* Command: {…} */
|
||||
if (p[0] == '{') {
|
||||
const char *close = p + 1;
|
||||
while (close < end && *close != '}') {
|
||||
close++;
|
||||
}
|
||||
if (close < end) {
|
||||
const char *cmd = p + 1;
|
||||
size_t cmd_len = (size_t)(close - cmd);
|
||||
process_command(state, result, cmd, cmd_len);
|
||||
had_command = true;
|
||||
if (state->glue) this_glue = true;
|
||||
p = close + 1;
|
||||
continue;
|
||||
}
|
||||
}
|
||||
|
||||
/* Literal text segment — find end (next '{' or '\' or end) */
|
||||
const char *seg_start = p;
|
||||
while (p < end && p[0] != '{' &&
|
||||
!(p[0] == '\\' && p + 1 < end && (p[1] == '{' || p[1] == '}'))) {
|
||||
p++;
|
||||
}
|
||||
size_t seg_len = (size_t)(p - seg_start);
|
||||
if (seg_len > 0) {
|
||||
if (!this_glue) {
|
||||
maybe_space(state, result);
|
||||
}
|
||||
emit_text_transformed(state, result, seg_start, seg_len);
|
||||
state->space_pending = true;
|
||||
if (!this_glue) {
|
||||
state->glue = false;
|
||||
}
|
||||
emitted_text = true;
|
||||
}
|
||||
}
|
||||
|
||||
result->text[result->len] = '\0';
|
||||
result->is_command_only = !emitted_text && had_command && result->len == 0;
|
||||
}
|
||||
46
src/formatter.h
Normal file
46
src/formatter.h
Normal file
|
|
@ -0,0 +1,46 @@
|
|||
#ifndef STENO_FORMATTER_H
|
||||
#define STENO_FORMATTER_H
|
||||
|
||||
#include <stdbool.h>
|
||||
#include <stddef.h>
|
||||
#include <stdint.h>
|
||||
|
||||
/* Max output length from a single translation */
|
||||
#define STENO_FMT_MAX_OUTPUT 128
|
||||
|
||||
/* Formatting mode */
|
||||
enum steno_fmt_mode {
|
||||
STENO_MODE_NORMAL = 0,
|
||||
STENO_MODE_CAPS,
|
||||
STENO_MODE_TITLE,
|
||||
STENO_MODE_LOWER,
|
||||
};
|
||||
|
||||
/* State persists between translations */
|
||||
struct steno_fmt_state {
|
||||
bool space_pending; /* insert space before next word */
|
||||
bool cap_next; /* capitalize next word */
|
||||
bool upper_next; /* uppercase next word */
|
||||
bool lower_next; /* lowercase next word */
|
||||
bool suppress_space; /* {^} suppress next space */
|
||||
bool glue; /* in fingerspelling/glue sequence */
|
||||
enum steno_fmt_mode mode;
|
||||
};
|
||||
|
||||
/* Result of formatting one translation */
|
||||
struct steno_fmt_result {
|
||||
char text[STENO_FMT_MAX_OUTPUT]; /* output text to emit */
|
||||
uint8_t len; /* length of text */
|
||||
uint8_t backspaces; /* backspaces to send BEFORE text */
|
||||
bool is_command_only; /* true if no text output (pure command) */
|
||||
bool is_undo; /* true if this is an undo command */
|
||||
};
|
||||
|
||||
void steno_fmt_init(struct steno_fmt_state *state);
|
||||
|
||||
/* Format a raw translation string. Updates state, fills result. */
|
||||
void steno_fmt_process(struct steno_fmt_state *state,
|
||||
const char *translation,
|
||||
struct steno_fmt_result *result);
|
||||
|
||||
#endif
|
||||
352
src/output.c
Normal file
352
src/output.c
Normal file
|
|
@ -0,0 +1,352 @@
|
|||
#include "output.h"
|
||||
|
||||
#include <zephyr/kernel.h>
|
||||
#include <zephyr/logging/log.h>
|
||||
|
||||
#include <zmk/hid.h>
|
||||
#include <zmk/endpoints.h>
|
||||
#include <zmk/events/keycode_state_changed.h>
|
||||
|
||||
LOG_MODULE_DECLARE(zmk, CONFIG_ZMK_LOG_LEVEL);
|
||||
|
||||
struct hid_map {
|
||||
uint8_t keycode;
|
||||
bool shift;
|
||||
};
|
||||
|
||||
static const struct hid_map ASCII_TO_HID[128] = {
|
||||
[' '] = {0x2C, false}, /* space */
|
||||
['!'] = {0x1E, true}, /* shift+1 */
|
||||
['"'] = {0x34, true}, /* shift+' */
|
||||
['#'] = {0x20, true},
|
||||
['$'] = {0x21, true},
|
||||
['%'] = {0x22, true},
|
||||
['&'] = {0x24, true},
|
||||
['\''] = {0x34, false},
|
||||
['('] = {0x26, true},
|
||||
[')'] = {0x27, true},
|
||||
['*'] = {0x25, true},
|
||||
['+'] = {0x2E, true},
|
||||
[','] = {0x36, false},
|
||||
['-'] = {0x2D, false},
|
||||
['.'] = {0x37, false},
|
||||
['/'] = {0x38, false},
|
||||
['0'] = {0x27, false},
|
||||
['1'] = {0x1E, false},
|
||||
['2'] = {0x1F, false},
|
||||
['3'] = {0x20, false},
|
||||
['4'] = {0x21, false},
|
||||
['5'] = {0x22, false},
|
||||
['6'] = {0x23, false},
|
||||
['7'] = {0x24, false},
|
||||
['8'] = {0x25, false},
|
||||
['9'] = {0x26, false},
|
||||
[':'] = {0x33, true},
|
||||
[';'] = {0x33, false},
|
||||
['<'] = {0x36, true},
|
||||
['='] = {0x2E, false},
|
||||
['>'] = {0x37, true},
|
||||
['?'] = {0x38, true},
|
||||
['@'] = {0x1F, true},
|
||||
['A'] = {0x04, true},
|
||||
['B'] = {0x05, true},
|
||||
['C'] = {0x06, true},
|
||||
['D'] = {0x07, true},
|
||||
['E'] = {0x08, true},
|
||||
['F'] = {0x09, true},
|
||||
['G'] = {0x0A, true},
|
||||
['H'] = {0x0B, true},
|
||||
['I'] = {0x0C, true},
|
||||
['J'] = {0x0D, true},
|
||||
['K'] = {0x0E, true},
|
||||
['L'] = {0x0F, true},
|
||||
['M'] = {0x10, true},
|
||||
['N'] = {0x11, true},
|
||||
['O'] = {0x12, true},
|
||||
['P'] = {0x13, true},
|
||||
['Q'] = {0x14, true},
|
||||
['R'] = {0x15, true},
|
||||
['S'] = {0x16, true},
|
||||
['T'] = {0x17, true},
|
||||
['U'] = {0x18, true},
|
||||
['V'] = {0x19, true},
|
||||
['W'] = {0x1A, true},
|
||||
['X'] = {0x1B, true},
|
||||
['Y'] = {0x1C, true},
|
||||
['Z'] = {0x1D, true},
|
||||
['['] = {0x2F, false},
|
||||
['\\'] = {0x31, false},
|
||||
[']'] = {0x30, false},
|
||||
['^'] = {0x23, true},
|
||||
['_'] = {0x2D, true},
|
||||
['`'] = {0x35, false},
|
||||
['a'] = {0x04, false},
|
||||
['b'] = {0x05, false},
|
||||
['c'] = {0x06, false},
|
||||
['d'] = {0x07, false},
|
||||
['e'] = {0x08, false},
|
||||
['f'] = {0x09, false},
|
||||
['g'] = {0x0A, false},
|
||||
['h'] = {0x0B, false},
|
||||
['i'] = {0x0C, false},
|
||||
['j'] = {0x0D, false},
|
||||
['k'] = {0x0E, false},
|
||||
['l'] = {0x0F, false},
|
||||
['m'] = {0x10, false},
|
||||
['n'] = {0x11, false},
|
||||
['o'] = {0x12, false},
|
||||
['p'] = {0x13, false},
|
||||
['q'] = {0x14, false},
|
||||
['r'] = {0x15, false},
|
||||
['s'] = {0x16, false},
|
||||
['t'] = {0x17, false},
|
||||
['u'] = {0x18, false},
|
||||
['v'] = {0x19, false},
|
||||
['w'] = {0x1A, false},
|
||||
['x'] = {0x1B, false},
|
||||
['y'] = {0x1C, false},
|
||||
['z'] = {0x1D, false},
|
||||
['{'] = {0x2F, true},
|
||||
['|'] = {0x31, true},
|
||||
['}'] = {0x30, true},
|
||||
['~'] = {0x35, true},
|
||||
};
|
||||
|
||||
#define HID_BACKSPACE 0x2A
|
||||
#define HID_RETURN 0x28
|
||||
#define HID_LSHIFT 0xE1
|
||||
#define HID_LCTRL 0xE0
|
||||
#define HID_LALT 0xE2
|
||||
#define HID_RALT 0xE6
|
||||
|
||||
static void tap_key(uint8_t keycode, bool shift)
|
||||
{
|
||||
if (shift) {
|
||||
raise_zmk_keycode_state_changed((struct zmk_keycode_state_changed){
|
||||
.usage_page = 0x07, .keycode = HID_LSHIFT,
|
||||
.implicit_modifiers = 0, .explicit_modifiers = 0,
|
||||
.state = true, .timestamp = k_uptime_get()});
|
||||
}
|
||||
|
||||
raise_zmk_keycode_state_changed((struct zmk_keycode_state_changed){
|
||||
.usage_page = 0x07, .keycode = keycode,
|
||||
.implicit_modifiers = 0, .explicit_modifiers = 0,
|
||||
.state = true, .timestamp = k_uptime_get()});
|
||||
|
||||
raise_zmk_keycode_state_changed((struct zmk_keycode_state_changed){
|
||||
.usage_page = 0x07, .keycode = keycode,
|
||||
.implicit_modifiers = 0, .explicit_modifiers = 0,
|
||||
.state = false, .timestamp = k_uptime_get()});
|
||||
|
||||
if (shift) {
|
||||
raise_zmk_keycode_state_changed((struct zmk_keycode_state_changed){
|
||||
.usage_page = 0x07, .keycode = HID_LSHIFT,
|
||||
.implicit_modifiers = 0, .explicit_modifiers = 0,
|
||||
.state = false, .timestamp = k_uptime_get()});
|
||||
}
|
||||
}
|
||||
|
||||
static void press_key(uint8_t keycode)
|
||||
{
|
||||
raise_zmk_keycode_state_changed((struct zmk_keycode_state_changed){
|
||||
.usage_page = 0x07, .keycode = keycode,
|
||||
.implicit_modifiers = 0, .explicit_modifiers = 0,
|
||||
.state = true, .timestamp = k_uptime_get()});
|
||||
}
|
||||
|
||||
static void release_key(uint8_t keycode)
|
||||
{
|
||||
raise_zmk_keycode_state_changed((struct zmk_keycode_state_changed){
|
||||
.usage_page = 0x07, .keycode = keycode,
|
||||
.implicit_modifiers = 0, .explicit_modifiers = 0,
|
||||
.state = false, .timestamp = k_uptime_get()});
|
||||
}
|
||||
|
||||
/* Map hex digit (0-15) to HID keycode */
|
||||
static uint8_t hex_to_hid(uint8_t nib)
|
||||
{
|
||||
if (nib == 0) {
|
||||
return 0x27; /* '0' */
|
||||
}
|
||||
if (nib <= 9) {
|
||||
return 0x1E + (nib - 1); /* '1'-'9': 0x1E-0x26 */
|
||||
}
|
||||
return 0x04 + (nib - 10); /* 'a'-'f': 0x04-0x09 */
|
||||
}
|
||||
|
||||
/* Tap hex digits of codepoint (variable width, skip leading zeros) */
|
||||
static void tap_hex_digits(uint32_t codepoint)
|
||||
{
|
||||
char buf[9];
|
||||
int n = 0;
|
||||
|
||||
if (codepoint == 0) {
|
||||
tap_key(hex_to_hid(0), false);
|
||||
return;
|
||||
}
|
||||
|
||||
/* Build hex digits in reverse */
|
||||
uint32_t cp = codepoint;
|
||||
while (cp > 0) {
|
||||
buf[n++] = cp & 0xF;
|
||||
cp >>= 4;
|
||||
}
|
||||
|
||||
/* Tap in forward order */
|
||||
for (int i = n - 1; i >= 0; i--) {
|
||||
tap_key(hex_to_hid(buf[i]), false);
|
||||
}
|
||||
}
|
||||
|
||||
void steno_output_unicode(uint32_t codepoint)
|
||||
{
|
||||
if (codepoint > 0x10FFFF) {
|
||||
LOG_WRN("Invalid codepoint U+%06X", codepoint);
|
||||
return;
|
||||
}
|
||||
|
||||
#if IS_ENABLED(CONFIG_STENO_UNICODE_MODE_LINUX)
|
||||
/* IBus/GTK: Ctrl+Shift+U, hex digits, Return */
|
||||
press_key(HID_LCTRL);
|
||||
press_key(HID_LSHIFT);
|
||||
tap_key(0x18, false); /* 'u' */
|
||||
release_key(HID_LSHIFT);
|
||||
release_key(HID_LCTRL);
|
||||
tap_hex_digits(codepoint);
|
||||
tap_key(HID_RETURN, false);
|
||||
|
||||
#elif IS_ENABLED(CONFIG_STENO_UNICODE_MODE_MACOS)
|
||||
/* macOS Unicode Hex Input: hold Option, type 4+ hex digits */
|
||||
press_key(HID_LALT);
|
||||
|
||||
/* Pad to at least 4 digits */
|
||||
char digits[8];
|
||||
int n = 0;
|
||||
uint32_t cp = codepoint;
|
||||
|
||||
do {
|
||||
digits[n++] = cp & 0xF;
|
||||
cp >>= 4;
|
||||
} while (cp > 0);
|
||||
|
||||
/* Pad to 4 */
|
||||
while (n < 4) {
|
||||
digits[n++] = 0;
|
||||
}
|
||||
|
||||
for (int i = n - 1; i >= 0; i--) {
|
||||
tap_key(hex_to_hid(digits[i]), false);
|
||||
}
|
||||
|
||||
release_key(HID_LALT);
|
||||
|
||||
#elif IS_ENABLED(CONFIG_STENO_UNICODE_MODE_WINC)
|
||||
/* WinCompose: tap RAlt, 'u', hex digits, Return */
|
||||
tap_key(HID_RALT, false);
|
||||
tap_key(0x18, false); /* 'u' */
|
||||
tap_hex_digits(codepoint);
|
||||
tap_key(HID_RETURN, false);
|
||||
|
||||
#else
|
||||
LOG_WRN("Unicode disabled, skipping U+%04X", codepoint);
|
||||
#endif
|
||||
}
|
||||
|
||||
/* Decode UTF-8 byte at text[i], write codepoint, return bytes consumed (0 on error) */
|
||||
static int utf8_decode(const char *text, size_t len, size_t i, uint32_t *cp)
|
||||
{
|
||||
unsigned char c = (unsigned char)text[i];
|
||||
|
||||
if (c < 0x80) {
|
||||
*cp = c;
|
||||
return 1;
|
||||
}
|
||||
|
||||
uint32_t codepoint;
|
||||
int expect; /* expected continuation bytes */
|
||||
|
||||
if ((c & 0xE0) == 0xC0) {
|
||||
codepoint = c & 0x1F;
|
||||
expect = 1;
|
||||
} else if ((c & 0xF0) == 0xE0) {
|
||||
codepoint = c & 0x0F;
|
||||
expect = 2;
|
||||
} else if ((c & 0xF8) == 0xF0) {
|
||||
codepoint = c & 0x07;
|
||||
expect = 3;
|
||||
} else {
|
||||
return 0; /* invalid lead byte */
|
||||
}
|
||||
|
||||
if (i + expect >= len) {
|
||||
return 0; /* truncated */
|
||||
}
|
||||
|
||||
for (int j = 1; j <= expect; j++) {
|
||||
unsigned char cont = (unsigned char)text[i + j];
|
||||
if ((cont & 0xC0) != 0x80) {
|
||||
return 0;
|
||||
}
|
||||
codepoint = (codepoint << 6) | (cont & 0x3F);
|
||||
}
|
||||
|
||||
/* Reject overlong encodings */
|
||||
if ((expect == 1 && codepoint < 0x80) ||
|
||||
(expect == 2 && codepoint < 0x800) ||
|
||||
(expect == 3 && codepoint < 0x10000)) {
|
||||
return 0;
|
||||
}
|
||||
|
||||
/* Reject surrogates (U+D800..U+DFFF) and beyond Unicode max */
|
||||
if ((codepoint >= 0xD800 && codepoint <= 0xDFFF) || codepoint > 0x10FFFF) {
|
||||
return 0;
|
||||
}
|
||||
|
||||
*cp = codepoint;
|
||||
return 1 + expect;
|
||||
}
|
||||
|
||||
void steno_output_send(const char *text, size_t len)
|
||||
{
|
||||
for (size_t i = 0; i < len; ) {
|
||||
unsigned char c = (unsigned char)text[i];
|
||||
|
||||
if (c == '\n') {
|
||||
tap_key(HID_RETURN, false);
|
||||
i++;
|
||||
continue;
|
||||
}
|
||||
|
||||
/* ASCII range */
|
||||
if (c < 128) {
|
||||
if (ASCII_TO_HID[c].keycode == 0) {
|
||||
LOG_WRN("Unmapped ASCII char 0x%02X", c);
|
||||
i++;
|
||||
continue;
|
||||
}
|
||||
tap_key(ASCII_TO_HID[c].keycode, ASCII_TO_HID[c].shift);
|
||||
i++;
|
||||
continue;
|
||||
}
|
||||
|
||||
/* Multi-byte UTF-8 → Unicode codepoint */
|
||||
uint32_t codepoint;
|
||||
int consumed = utf8_decode(text, len, i, &codepoint);
|
||||
|
||||
if (consumed == 0) {
|
||||
LOG_WRN("Invalid UTF-8 at offset %u (0x%02X)", (unsigned)i, c);
|
||||
i++;
|
||||
continue;
|
||||
}
|
||||
|
||||
steno_output_unicode(codepoint);
|
||||
i += consumed;
|
||||
}
|
||||
}
|
||||
|
||||
void steno_output_backspace(int count)
|
||||
{
|
||||
for (int i = 0; i < count; i++) {
|
||||
tap_key(HID_BACKSPACE, false);
|
||||
}
|
||||
}
|
||||
13
src/output.h
Normal file
13
src/output.h
Normal file
|
|
@ -0,0 +1,13 @@
|
|||
#ifndef STENO_OUTPUT_H
|
||||
#define STENO_OUTPUT_H
|
||||
|
||||
#include <stddef.h>
|
||||
#include <stdint.h>
|
||||
|
||||
void steno_output_send(const char *text, size_t len);
|
||||
|
||||
void steno_output_backspace(int count);
|
||||
|
||||
void steno_output_unicode(uint32_t codepoint);
|
||||
|
||||
#endif
|
||||
148
src/trie.c
Normal file
148
src/trie.c
Normal file
|
|
@ -0,0 +1,148 @@
|
|||
#include "trie.h"
|
||||
#include <string.h>
|
||||
|
||||
static const uint8_t *dict_data;
|
||||
static const struct steno_dict_header *dict_hdr;
|
||||
static const uint8_t *entry_array;
|
||||
static const char *string_table;
|
||||
static size_t entry_stride;
|
||||
|
||||
int steno_trie_init(const uint8_t *data, size_t len)
|
||||
{
|
||||
if (!data || len < sizeof(struct steno_dict_header)) {
|
||||
return -1;
|
||||
}
|
||||
|
||||
dict_hdr = (const struct steno_dict_header *)data;
|
||||
|
||||
if (dict_hdr->magic != STENO_DICT_MAGIC) {
|
||||
return -2;
|
||||
}
|
||||
if (dict_hdr->version != 1) {
|
||||
return -3;
|
||||
}
|
||||
if (dict_hdr->max_strokes == 0 || dict_hdr->max_strokes > 16) {
|
||||
return -4;
|
||||
}
|
||||
|
||||
entry_stride = (size_t)dict_hdr->max_strokes * 4 + 4;
|
||||
size_t entries_end = sizeof(struct steno_dict_header) +
|
||||
(size_t)dict_hdr->entry_count * entry_stride;
|
||||
|
||||
if (entries_end > len || dict_hdr->strings_offset > len) {
|
||||
return -5;
|
||||
}
|
||||
|
||||
dict_data = data;
|
||||
entry_array = data + sizeof(struct steno_dict_header);
|
||||
string_table = (const char *)(data + dict_hdr->strings_offset);
|
||||
|
||||
return 0;
|
||||
}
|
||||
|
||||
static const uint8_t *get_entry(uint32_t idx)
|
||||
{
|
||||
return entry_array + (size_t)idx * entry_stride;
|
||||
}
|
||||
|
||||
static int cmp_strokes(const uint32_t *a, uint8_t a_count,
|
||||
const uint8_t *entry_bytes, uint8_t max_s)
|
||||
{
|
||||
const uint32_t *b = (const uint32_t *)entry_bytes;
|
||||
|
||||
uint8_t b_count = 0;
|
||||
for (uint8_t i = 0; i < max_s; i++) {
|
||||
if (b[i] != 0) {
|
||||
b_count = i + 1;
|
||||
}
|
||||
}
|
||||
|
||||
uint8_t min_count = a_count < b_count ? a_count : b_count;
|
||||
for (uint8_t i = 0; i < min_count; i++) {
|
||||
if (a[i] < b[i]) return -1;
|
||||
if (a[i] > b[i]) return 1;
|
||||
}
|
||||
if (a_count < b_count) return -1;
|
||||
if (a_count > b_count) return 1;
|
||||
return 0;
|
||||
}
|
||||
|
||||
const char *steno_trie_lookup(const uint32_t *strokes, uint8_t count)
|
||||
{
|
||||
if (!dict_hdr || !strokes || count == 0 || count > dict_hdr->max_strokes) {
|
||||
return NULL;
|
||||
}
|
||||
|
||||
uint32_t lo = 0;
|
||||
uint32_t hi = dict_hdr->entry_count;
|
||||
|
||||
while (lo < hi) {
|
||||
uint32_t mid = lo + (hi - lo) / 2;
|
||||
const uint8_t *entry = get_entry(mid);
|
||||
int c = cmp_strokes(strokes, count, entry, dict_hdr->max_strokes);
|
||||
if (c == 0) {
|
||||
uint32_t str_off;
|
||||
memcpy(&str_off, entry + (size_t)dict_hdr->max_strokes * 4, 4);
|
||||
return string_table + str_off;
|
||||
}
|
||||
if (c < 0) {
|
||||
hi = mid;
|
||||
} else {
|
||||
lo = mid + 1;
|
||||
}
|
||||
}
|
||||
|
||||
return NULL;
|
||||
}
|
||||
|
||||
bool steno_trie_has_prefix(const uint32_t *strokes, uint8_t count)
|
||||
{
|
||||
if (!dict_hdr || !strokes || count == 0 || count >= dict_hdr->max_strokes) {
|
||||
return false;
|
||||
}
|
||||
|
||||
uint32_t lo = 0;
|
||||
uint32_t hi = dict_hdr->entry_count;
|
||||
uint32_t first_ge = hi;
|
||||
|
||||
while (lo < hi) {
|
||||
uint32_t mid = lo + (hi - lo) / 2;
|
||||
const uint8_t *entry = get_entry(mid);
|
||||
const uint32_t *e_strokes = (const uint32_t *)entry;
|
||||
|
||||
int cmp = 0;
|
||||
for (uint8_t i = 0; i < count; i++) {
|
||||
if (e_strokes[i] < strokes[i]) { cmp = -1; break; }
|
||||
if (e_strokes[i] > strokes[i]) { cmp = 1; break; }
|
||||
}
|
||||
|
||||
if (cmp >= 0) {
|
||||
first_ge = mid;
|
||||
hi = mid;
|
||||
} else {
|
||||
lo = mid + 1;
|
||||
}
|
||||
}
|
||||
|
||||
for (uint32_t idx = first_ge; idx < dict_hdr->entry_count; idx++) {
|
||||
const uint8_t *entry = get_entry(idx);
|
||||
const uint32_t *e_strokes = (const uint32_t *)entry;
|
||||
|
||||
bool prefix_match = true;
|
||||
for (uint8_t i = 0; i < count; i++) {
|
||||
if (e_strokes[i] != strokes[i]) {
|
||||
prefix_match = false;
|
||||
break;
|
||||
}
|
||||
}
|
||||
if (!prefix_match) {
|
||||
return false;
|
||||
}
|
||||
|
||||
if (e_strokes[count] != 0) {
|
||||
return true;
|
||||
}
|
||||
}
|
||||
|
||||
return false;
|
||||
}
|
||||
25
src/trie.h
Normal file
25
src/trie.h
Normal file
|
|
@ -0,0 +1,25 @@
|
|||
#ifndef STENO_TRIE_H
|
||||
#define STENO_TRIE_H
|
||||
|
||||
#include <stdbool.h>
|
||||
#include <stddef.h>
|
||||
#include <stdint.h>
|
||||
|
||||
#define STENO_DICT_MAGIC 0x4F4E5453 /* "STNO" */
|
||||
|
||||
struct steno_dict_header {
|
||||
uint32_t magic;
|
||||
uint16_t version;
|
||||
uint8_t max_strokes;
|
||||
uint8_t _pad;
|
||||
uint32_t entry_count;
|
||||
uint32_t strings_offset;
|
||||
} __attribute__((packed));
|
||||
|
||||
int steno_trie_init(const uint8_t *data, size_t len);
|
||||
|
||||
const char *steno_trie_lookup(const uint32_t *strokes, uint8_t count);
|
||||
|
||||
bool steno_trie_has_prefix(const uint32_t *strokes, uint8_t count);
|
||||
|
||||
#endif
|
||||
64
src/undo.c
Normal file
64
src/undo.c
Normal file
|
|
@ -0,0 +1,64 @@
|
|||
/*
|
||||
* Copyright (c) 2024 Afiq Zudin Hadi
|
||||
* SPDX-License-Identifier: PolyForm-Noncommercial-1.0.0
|
||||
*/
|
||||
|
||||
#include "undo.h"
|
||||
#include <string.h>
|
||||
|
||||
void steno_undo_init(struct stroke_history *hist)
|
||||
{
|
||||
memset(hist, 0, sizeof(*hist));
|
||||
}
|
||||
|
||||
void steno_undo_push(struct stroke_history *hist,
|
||||
const uint32_t *strokes, uint8_t stroke_count,
|
||||
uint8_t output_len, uint8_t space_before,
|
||||
uint8_t fmt_flags)
|
||||
{
|
||||
struct stroke_history_entry *e = &hist->entries[hist->head];
|
||||
|
||||
uint8_t n = stroke_count;
|
||||
if (n > STENO_MAX_MULTI_STROKE) {
|
||||
n = STENO_MAX_MULTI_STROKE;
|
||||
}
|
||||
|
||||
memcpy(e->strokes, strokes, n * sizeof(uint32_t));
|
||||
e->stroke_count = n;
|
||||
e->output_len = output_len;
|
||||
e->space_before = space_before;
|
||||
e->fmt_flags = fmt_flags;
|
||||
|
||||
hist->head = (hist->head + 1) % CONFIG_STENO_HISTORY_SIZE;
|
||||
|
||||
if (hist->count < CONFIG_STENO_HISTORY_SIZE) {
|
||||
hist->count++;
|
||||
}
|
||||
}
|
||||
|
||||
struct stroke_history_entry *steno_undo_pop(struct stroke_history *hist)
|
||||
{
|
||||
if (hist->count == 0) {
|
||||
return NULL;
|
||||
}
|
||||
|
||||
hist->head = (hist->head + CONFIG_STENO_HISTORY_SIZE - 1) % CONFIG_STENO_HISTORY_SIZE;
|
||||
hist->count--;
|
||||
|
||||
return &hist->entries[hist->head];
|
||||
}
|
||||
|
||||
const struct stroke_history_entry *steno_undo_peek(const struct stroke_history *hist)
|
||||
{
|
||||
if (hist->count == 0) {
|
||||
return NULL;
|
||||
}
|
||||
|
||||
uint16_t idx = (hist->head + CONFIG_STENO_HISTORY_SIZE - 1) % CONFIG_STENO_HISTORY_SIZE;
|
||||
return &hist->entries[idx];
|
||||
}
|
||||
|
||||
uint16_t steno_undo_count(const struct stroke_history *hist)
|
||||
{
|
||||
return hist->count;
|
||||
}
|
||||
44
src/undo.h
Normal file
44
src/undo.h
Normal file
|
|
@ -0,0 +1,44 @@
|
|||
#ifndef STENO_UNDO_H
|
||||
#define STENO_UNDO_H
|
||||
|
||||
#include <stdbool.h>
|
||||
#include <stdint.h>
|
||||
|
||||
#ifndef CONFIG_STENO_HISTORY_SIZE
|
||||
#define CONFIG_STENO_HISTORY_SIZE 100
|
||||
#endif
|
||||
|
||||
#define STENO_MAX_MULTI_STROKE 8
|
||||
|
||||
struct stroke_history_entry {
|
||||
uint32_t strokes[STENO_MAX_MULTI_STROKE];
|
||||
uint8_t stroke_count;
|
||||
uint8_t output_len; /* chars emitted (for backspace count) */
|
||||
uint8_t space_before; /* 1 if space was prepended */
|
||||
uint8_t fmt_flags; /* formatter state snapshot for restore */
|
||||
};
|
||||
|
||||
struct stroke_history {
|
||||
struct stroke_history_entry entries[CONFIG_STENO_HISTORY_SIZE];
|
||||
uint16_t head; /* next write position */
|
||||
uint16_t count; /* entries in buffer */
|
||||
};
|
||||
|
||||
void steno_undo_init(struct stroke_history *hist);
|
||||
|
||||
/* Push a new entry after successful output */
|
||||
void steno_undo_push(struct stroke_history *hist,
|
||||
const uint32_t *strokes, uint8_t stroke_count,
|
||||
uint8_t output_len, uint8_t space_before,
|
||||
uint8_t fmt_flags);
|
||||
|
||||
/* Pop most recent entry for undo. Returns NULL if empty. */
|
||||
struct stroke_history_entry *steno_undo_pop(struct stroke_history *hist);
|
||||
|
||||
/* Peek at most recent entry without removing. Returns NULL if empty. */
|
||||
const struct stroke_history_entry *steno_undo_peek(const struct stroke_history *hist);
|
||||
|
||||
/* Get current count */
|
||||
uint16_t steno_undo_count(const struct stroke_history *hist);
|
||||
|
||||
#endif
|
||||
Loading…
Add table
Add a link
Reference in a new issue