Release v0.6.10: preserve typing during correction and add personal dictionaries
This commit is contained in:
+1
-1
@@ -1,5 +1,5 @@
|
||||
cmake_minimum_required(VERSION 3.24)
|
||||
project(KSwitcher VERSION 0.6.9 LANGUAGES CXX RC)
|
||||
project(KSwitcher VERSION 0.6.10 LANGUAGES CXX RC)
|
||||
if(NOT WIN32)
|
||||
message(FATAL_ERROR "KSwitcher uses native Windows APIs")
|
||||
endif()
|
||||
|
||||
@@ -343,14 +343,23 @@ after Space, not immediately after a punctuation key. Save Settings to apply cha
|
||||
**Frequency rank limit** defaults to **0 (entire dictionary)**. A positive value allows corrections
|
||||
only to words with a rank at or below that number (for example, 10000). All dictionary words still
|
||||
protect already valid input, and ambiguous alternatives remain unchanged regardless of this limit.
|
||||
**My dictionary** provides separate English, Ukrainian and Russian fields for correct words missing
|
||||
from the bundled dictionaries. Enter words separated by commas or new lines, then click OK and Save
|
||||
in Settings. Case and duplicates are normalized automatically; no file editing or ranks are needed.
|
||||
These entries are stored in settings and survive dictionary updates. They protect correct input and
|
||||
are eligible as correction targets at any frequency limit. The language and target layout must still
|
||||
be enabled. **Never correct these words** is a separate exception list and takes precedence.
|
||||
**Open dictionary folder** opens the files beside the running executable. After editing, click OK
|
||||
in the options window and Save in Settings to validate and reload the dictionaries without restarting.
|
||||
Use UTF-8, one lowercase word followed by a Tab and a positive rank per line; keep words sorted and
|
||||
unique. For a personal addition, rank 1 makes it eligible at every positive limit. Back up your edited
|
||||
Use UTF-8, one lowercase word followed by a Tab and a positive rank per line. Entries can be appended;
|
||||
loading sorts them and merges duplicates using the lowest rank. Back up your edited
|
||||
files before replacing dictionaries with a new distribution.
|
||||
|
||||
After Space and a 250 ms typing pause, KSwitcher checks the last physically typed word against local
|
||||
dictionaries. For example, `ghbdsn ` can become `привіт `. Continuing to type cancels pending work.
|
||||
dictionaries. For example, `ghbdsn ` can become `привіт `. Continuing to type before selection cancels
|
||||
pending work. During the replacement itself, key presses and releases are queued and replayed after
|
||||
clipboard restoration and layout switching, preserving typing order, modifiers and Backspace.
|
||||
If the active window changes, queued typing is not sent to another window and a warning is shown.
|
||||
It only corrects unknown words when exactly one alternative is a dictionary word.
|
||||
**Minimum word length** defaults to 3 and can be set from 1 to 32 letters. Surrounding punctuation
|
||||
does not count toward this minimum; the maximum word length is 32 letters.
|
||||
|
||||
Binary file not shown.
+7
-9
@@ -1,26 +1,24 @@
|
||||
# Latest Windows release
|
||||
|
||||
## v0.6.9 — 2026-10-29
|
||||
## v0.6.10 — 2026-10-29
|
||||
|
||||
## Changes
|
||||
- Show automatic update announcements using the same native Windows notification channel as screenshots. Review update opens the update check and confirmation; Later dismisses the notification.
|
||||
- Retry notification delivery after five minutes if Windows rejects it, and log update checks and delivery results. Successfully submitted notifications are deduplicated for the current version during the running session.
|
||||
- Fix Settings flicker when using the mouse wheel: skip layout work at scroll boundaries, move only changed controls in one batch, and avoid erasing the entire window background.
|
||||
- Preserve typing during automatic layout correction: queue key presses and releases until replacement, clipboard restoration and layout switching finish. Preserve modifiers and Backspace; collapse temporary selection before replay when a correction is rejected.
|
||||
- Add My dictionary fields for English, Ukrainian and Russian in Auto-correction options. Enter correct words separated by commas or new lines; save them with settings without editing dictionary files, sorting entries or assigning ranks.
|
||||
- Accept unsorted dictionary files and duplicate entries, retaining the lowest rank for duplicates. Include Ukrainian `реліз` in the bundled vocabulary so `релыз` can be corrected when Russian and Ukrainian are enabled.
|
||||
|
||||
## Validation
|
||||
203 native checks, 456 input/layout checks, screenshot editor UI integration, isolated notification registration and activation checks, and 45 updater checks. Regression coverage includes notification retry/deduplication, review-only activation, partial wheel steps, scroll boundaries and batched control movement.
|
||||
Native regression checks cover personal words, settings persistence, language validation, dictionary ordering and duplicate ranks. Input/UI checks cover personal-word entry and normalization. Native editor integration covers typing with Shift and Backspace during successful and rejected corrections. Release checks also cover screenshot UI, isolated notification registration and updater validation.
|
||||
|
||||
**Upgrade:** choose Update in the tray menu. Legacy v0.3.0–v0.4.1 clients first receive the v0.4.2 compatibility bridge, then the current full release. v0.2.3 and earlier require manual installation.
|
||||
For manual installation, exit KSwitcher and extract the entire ZIP into its installation folder. Keep personal `data/` and customized dictionaries.
|
||||
For manual installation, exit KSwitcher and extract the entire ZIP into its installation folder. Keep personal `data/` and customized dictionaries. Existing customized dictionary files may be preserved by the updater; use My dictionary to add missing words without replacing those files.
|
||||
|
||||
Download **KSwitcher-win-x64.zip** with **SHA256SUMS.txt**. The package contains all three executables, a release manifest, offline dictionaries, documentation and license notices. It contains no personal settings, clipboard history, Tesseract installation or language models.
|
||||
|
||||
Windows notification banners follow system notification and Do not disturb settings. The new update notification behavior takes effect after installing v0.6.9.
|
||||
|
||||
Build and package:
|
||||
|
||||
```powershell
|
||||
.\build.ps1 -Test
|
||||
.\build.ps1 -Stage -Test -Integration
|
||||
.\package.ps1 -PublishRelease
|
||||
```
|
||||
|
||||
|
||||
@@ -1 +1 @@
|
||||
cd6b3dd62065e50b68cf77344fbad8d7be41467142f31527933342f00e3c0f63 KSwitcher-win-x64.zip
|
||||
36966eb0b2b0af47c245aec6337a16703e0fbdca4d3eda7a8f25fa14a20f347e KSwitcher-win-x64.zip
|
||||
|
||||
+12
-2
@@ -17,9 +17,19 @@ replace counts with original frequency ranks, sort lexically. Reproduce with
|
||||
hashes. The lists contain word forms from subtitles, not a complete spelling
|
||||
or grammar dictionary. Recognition can miss names, jargon and uncommon forms.
|
||||
|
||||
File format: UTF-8 word, TAB, positive original frequency rank, newline.
|
||||
Supplemental Ukrainian vocabulary: `реліз`. These additions use rank 50001,
|
||||
after the upstream 50k list, rather than claiming a measured corpus frequency.
|
||||
They are included with the default unlimited rank setting and recorded in sources.json.
|
||||
|
||||
File format: UTF-8 word, TAB, positive frequency rank (or supplemental rank), newline.
|
||||
All entries protect known words. The frequency rank limit in Auto-correction options defaults to 0
|
||||
(entire dictionary); a positive limit only permits correction candidates at or below that rank.
|
||||
Personal words may use rank 1. Keep entries lowercase, unique and sorted, with a Tab separator.
|
||||
Prefer Settings > Auto-correction options > My dictionary for personal words.
|
||||
Enter correct spellings in the matching language field, separated by commas or new lines.
|
||||
These words are stored in settings, survive dictionary updates, and need no ranks or sorting.
|
||||
Click OK, then Save in Settings to apply them.
|
||||
|
||||
For advanced file edits, personal words may use rank 1. Keep entries lowercase with a Tab separator.
|
||||
Words may be appended: loading sorts entries and merges duplicates using the lowest rank.
|
||||
Save the UTF-8 file, then click OK in Auto-correction options and Save in Settings to reload it.
|
||||
Back up edited files before replacing dictionaries with a new distribution.
|
||||
|
||||
@@ -9,9 +9,12 @@
|
||||
"uk": {
|
||||
"source": "https://raw.githubusercontent.com/hermitdave/FrequencyWords/525f9b560de45753a5ea01069454e72e9aa541c6/content/2016/uk/uk_50k.txt",
|
||||
"sourceSha256": "9548b79a608d33f39c5546037a69117d679730b1fda08dd0ce67c043dbf3335a",
|
||||
"words": 42549,
|
||||
"bytes": 898210,
|
||||
"sha256": "b4104898dcccbc1331ec01e9cd690e30285fe58481460e47438e973964d09802"
|
||||
"words": 42550,
|
||||
"bytes": 898227,
|
||||
"sha256": "c94605e6be676d2d64f89a3bc16f92b9128f9977d226a97b103cf22a3e6835e3",
|
||||
"supplementalWords": {
|
||||
"\u0440\u0435\u043b\u0456\u0437": 50001
|
||||
}
|
||||
},
|
||||
"ru": {
|
||||
"source": "https://raw.githubusercontent.com/hermitdave/FrequencyWords/525f9b560de45753a5ea01069454e72e9aa541c6/content/2016/ru/ru_50k.txt",
|
||||
|
||||
@@ -31207,6 +31207,7 @@
|
||||
релігію 26138
|
||||
релігія 22202
|
||||
релігії 23427
|
||||
реліз 50001
|
||||
рембо 28603
|
||||
ремень 23872
|
||||
ремені 28385
|
||||
|
||||
+3
-3
@@ -9,8 +9,8 @@ IDC_SCREENSHOT_PENCIL CURSOR "icons/lucide/pencil.cur"
|
||||
1 RT_MANIFEST "app.manifest"
|
||||
201 RCDATA "update.ps1"
|
||||
1 VERSIONINFO
|
||||
FILEVERSION 0,6,9,0
|
||||
PRODUCTVERSION 0,6,9,0
|
||||
FILEVERSION 0,6,10,0
|
||||
PRODUCTVERSION 0,6,10,0
|
||||
FILEOS VOS_NT_WINDOWS32
|
||||
FILETYPE VFT_APP
|
||||
BEGIN
|
||||
@@ -20,7 +20,7 @@ BEGIN
|
||||
BEGIN
|
||||
VALUE "FileDescription", "KSwitcher — native keyboard and clipboard utility"
|
||||
VALUE "ProductName", "KSwitcher"
|
||||
VALUE "FileVersion", "0.6.9"
|
||||
VALUE "FileVersion", "0.6.10"
|
||||
END
|
||||
END
|
||||
BLOCK "VarFileInfo"
|
||||
|
||||
+86
-1
@@ -598,6 +598,27 @@ bool App::handleKeyboard(const KBDLLHOOKSTRUCT &key, bool down, UINT mods)
|
||||
// down AND up events use exactly the same path as physical keyboard input.
|
||||
if (ownKeyboardInput(key) || installingUpdate)
|
||||
return false;
|
||||
{
|
||||
std::lock_guard lock(stateMutex);
|
||||
if (typingTarget && GetForegroundWindow() == typingTarget)
|
||||
{
|
||||
INPUT input{};
|
||||
input.type = INPUT_KEYBOARD;
|
||||
input.ki.wVk = static_cast<WORD>(key.vkCode);
|
||||
input.ki.wScan = static_cast<WORD>(key.scanCode);
|
||||
input.ki.dwFlags = (down ? 0 : KEYEVENTF_KEYUP) |
|
||||
((key.flags & LLKHF_EXTENDED) ? KEYEVENTF_EXTENDEDKEY : 0);
|
||||
if (key.vkCode == VK_PACKET)
|
||||
{
|
||||
input.ki.wVk = 0;
|
||||
input.ki.dwFlags |= KEYEVENTF_UNICODE;
|
||||
}
|
||||
// Replay as ordinary input so tracking and shortcuts see it once,
|
||||
// after the replacement and layout switch have finished.
|
||||
deferredTyping.push_back(input);
|
||||
return true;
|
||||
}
|
||||
}
|
||||
DWORD captureProcess{};
|
||||
if (hotkeyCapture) GetWindowThreadProcessId(hotkeyCapture, &captureProcess);
|
||||
if (hotkeyCapture && (!captureProcess || captureProcess != hotkeyCaptureProcess))
|
||||
@@ -640,6 +661,41 @@ bool App::handleKeyboard(const KBDLLHOOKSTRUCT &key, bool down, UINT mods)
|
||||
return true;
|
||||
return false;
|
||||
}
|
||||
bool App::beginTypingDeferral(HWND target, uint64_t version)
|
||||
{
|
||||
std::lock_guard lock(stateMutex);
|
||||
if (inputVersion != version || GetForegroundWindow() != target || !modifiersReleased())
|
||||
return false;
|
||||
typingTarget = target;
|
||||
return true;
|
||||
}
|
||||
void App::replayTyping()
|
||||
{
|
||||
std::vector<INPUT> inputs;
|
||||
HWND target;
|
||||
{
|
||||
std::lock_guard lock(stateMutex);
|
||||
target = typingTarget;
|
||||
typingTarget = nullptr;
|
||||
inputs.swap(deferredTyping);
|
||||
}
|
||||
busy = false;
|
||||
if (inputs.empty())
|
||||
return;
|
||||
if (GetForegroundWindow() != target)
|
||||
{
|
||||
resetInput();
|
||||
report(L"Typing during correction was not replayed because the active window changed.");
|
||||
return;
|
||||
}
|
||||
// Runs on the hook thread: no physical key can pass between releasing the
|
||||
// gate and submitting the entire ordered batch to the Windows input queue.
|
||||
if (SendInput(static_cast<UINT>(inputs.size()), inputs.data(), sizeof(INPUT)) != inputs.size())
|
||||
{
|
||||
resetInput();
|
||||
report(L"Windows could not replay all typing buffered during correction.");
|
||||
}
|
||||
}
|
||||
LRESULT CALLBACK App::keyProc(int code, WPARAM wp, LPARAM lp)
|
||||
{
|
||||
if (code == HC_ACTION && current &&
|
||||
@@ -762,11 +818,16 @@ void App::transform(int command, int snippet)
|
||||
struct BusyReset
|
||||
{
|
||||
std::atomic<bool> &b;
|
||||
HWND window;
|
||||
bool deferred{};
|
||||
~BusyReset()
|
||||
{
|
||||
if (deferred)
|
||||
PostMessageW(window, ReplayTypingMessage, 0, 0);
|
||||
else
|
||||
b = false;
|
||||
}
|
||||
} reset{busy};
|
||||
} reset{busy, window};
|
||||
auto start = GetTickCount64();
|
||||
std::wstring reason = L"cancelled-before-selection";
|
||||
std::wstring text, result, expected;
|
||||
@@ -849,6 +910,21 @@ void App::transform(int command, int snippet)
|
||||
return;
|
||||
}
|
||||
reason = L"clipboard-snapshot";
|
||||
if (command == AutoCorrect)
|
||||
{
|
||||
if (!beginTypingDeferral(target, version))
|
||||
return;
|
||||
reset.deferred = true;
|
||||
}
|
||||
// Collapse a temporary selection on every failed auto-correction,
|
||||
// including clipboard exceptions, before queued typing is replayed.
|
||||
bool autoSelected = false;
|
||||
struct SelectionReset
|
||||
{
|
||||
bool &selected;
|
||||
std::function<bool()> valid;
|
||||
~SelectionReset() { if (selected && valid()) chord(VK_RIGHT); }
|
||||
} selectionReset{autoSelected, valid};
|
||||
Transaction tx(*this, target, version);
|
||||
reason = L"selection-unavailable";
|
||||
diagnosticEvent(L"Transform target=" + processName(target) + L" command=" + std::to_wstring(command) +
|
||||
@@ -865,6 +941,7 @@ void App::transform(int command, int snippet)
|
||||
// conversion, automatic correction must never fall back to a suffix.
|
||||
if (!tx.valid() || !chord(VK_LEFT, MOD_CONTROL | MOD_SHIFT))
|
||||
return;
|
||||
autoSelected = true;
|
||||
Sleep(25);
|
||||
text = tx.copy();
|
||||
// Some editors first select only ") ". Extend by whole-word
|
||||
@@ -881,7 +958,10 @@ void App::transform(int command, int snippet)
|
||||
if (!suggestion || !tx.valid() || !autoCorrectionAllowed(target, config))
|
||||
{
|
||||
if (tx.valid())
|
||||
{
|
||||
chord(VK_RIGHT);
|
||||
autoSelected = false;
|
||||
}
|
||||
return;
|
||||
}
|
||||
to = suggestion->target;
|
||||
@@ -1018,6 +1098,8 @@ void App::transform(int command, int snippet)
|
||||
sent = tx.paste(textClip(result));
|
||||
else if (automatic && tx.valid())
|
||||
chord(VK_RIGHT);
|
||||
if (sent || result == text)
|
||||
autoSelected = false;
|
||||
reason = result == text ? L"conversion-unchanged" :
|
||||
sent ? L"paste-sent-unverified" : L"paste-cancelled";
|
||||
if (result == text &&
|
||||
@@ -1377,6 +1459,9 @@ LRESULT CALLBACK App::windowProc(HWND hwnd, UINT msg, WPARAM wp, LPARAM lp)
|
||||
case WM_HOTKEY:
|
||||
app->dispatch((int)wp);
|
||||
return 0;
|
||||
case ReplayTypingMessage:
|
||||
app->replayTyping();
|
||||
return 0;
|
||||
case TrayMessage:
|
||||
if (LOWORD(lp) == WM_CONTEXTMENU)
|
||||
app->trayMenu();
|
||||
|
||||
@@ -16,6 +16,7 @@ inline constexpr UINT TrayMessage = WM_APP + 1, ErrorMessage = WM_APP + 2, Histo
|
||||
inline constexpr int Toggle = 1, Case = 2, Translit = 3, Undo = 4, ShowHistory = 5, AutoCorrect = 6,
|
||||
Screenshot = 7, Direct = 100;
|
||||
inline constexpr UINT AutoCorrectMessage = WM_APP + 6;
|
||||
inline constexpr UINT ReplayTypingMessage = WM_APP + 30;
|
||||
inline constexpr UINT CaptureHotkeyMessage = WM_APP + 5;
|
||||
struct LastTransform
|
||||
{
|
||||
@@ -105,6 +106,10 @@ class App
|
||||
int previousAttemptCommand{}, repeatAttempts{};
|
||||
std::vector<std::pair<int, Hotkey>> hookBindings;
|
||||
HotkeyHook hotkeyHook;
|
||||
HWND typingTarget{};
|
||||
std::vector<INPUT> deferredTyping;
|
||||
bool beginTypingDeferral(HWND, uint64_t);
|
||||
void replayTyping();
|
||||
static LRESULT CALLBACK windowProc(HWND, UINT, WPARAM, LPARAM);
|
||||
static LRESULT CALLBACK keyProc(int, WPARAM, LPARAM);
|
||||
static LRESULT CALLBACK mouseProc(int, WPARAM, LPARAM);
|
||||
|
||||
+21
-5
@@ -70,9 +70,18 @@ AutoCorrector::Dictionary::Dictionary(const std::filesystem::path &path)
|
||||
}
|
||||
if (offsets.empty())
|
||||
throw std::runtime_error("Empty auto-correction dictionary");
|
||||
for (size_t i = 1; i < offsets.size(); ++i)
|
||||
if (word(i - 1) >= word(i))
|
||||
throw std::runtime_error("Dictionary must be uniquely sorted");
|
||||
auto entry = [&](uint32_t offset) { return std::string_view(bytes).substr(offset, bytes.find('\t', offset) - offset); };
|
||||
auto frequency = [&](uint32_t offset)
|
||||
{
|
||||
int rank{};
|
||||
auto start = offset + entry(offset).size() + 1;
|
||||
std::from_chars(bytes.data() + start, bytes.data() + bytes.size(), rank);
|
||||
return rank;
|
||||
};
|
||||
std::sort(offsets.begin(), offsets.end(), [&](uint32_t a, uint32_t b)
|
||||
{ return entry(a) == entry(b) ? frequency(a) < frequency(b) : entry(a) < entry(b); });
|
||||
offsets.erase(std::unique(offsets.begin(), offsets.end(), [&](uint32_t a, uint32_t b)
|
||||
{ return entry(a) == entry(b); }), offsets.end());
|
||||
offsets.shrink_to_fit();
|
||||
bytes.shrink_to_fit();
|
||||
}
|
||||
@@ -151,9 +160,16 @@ std::optional<AutoSuggestion> AutoCorrector::suggest(const std::wstring &typed,
|
||||
auto sourceLanguage = layoutLanguage(layouts[source]);
|
||||
if (!dictionaries.contains(sourceLanguage))
|
||||
return {};
|
||||
auto personal = [&](const std::wstring &language, const std::wstring &candidate)
|
||||
{
|
||||
auto found = settings.autoWords.find(language);
|
||||
return found != settings.autoWords.end() &&
|
||||
std::any_of(found->second.begin(), found->second.end(), [&](const std::wstring &known)
|
||||
{ return lower(known) == lower(candidate); });
|
||||
};
|
||||
// Valid words from either selected language are never changed.
|
||||
for (auto &[language, d] : dictionaries)
|
||||
if (d.rank(word))
|
||||
if (d.rank(word) || personal(language, word))
|
||||
return {};
|
||||
std::optional<AutoSuggestion> candidate;
|
||||
int matches = 0;
|
||||
@@ -168,7 +184,7 @@ std::optional<AutoSuggestion> AutoCorrector::suggest(const std::wstring &typed,
|
||||
auto converted = convert(word, layouts[source], layouts[i], true);
|
||||
if (converted == word || !safeAutoWord(converted, settings.autoMinLength))
|
||||
continue;
|
||||
int rank = dictionary->second.rank(converted);
|
||||
int rank = personal(dictionary->first, converted) ? 1 : dictionary->second.rank(converted);
|
||||
if (!rank)
|
||||
continue;
|
||||
if (++matches > 1)
|
||||
|
||||
@@ -1,6 +1,7 @@
|
||||
#pragma once
|
||||
#include <cstdint>
|
||||
#include <functional>
|
||||
#include <map>
|
||||
#include <optional>
|
||||
#include <string>
|
||||
#include <vector>
|
||||
@@ -58,6 +59,7 @@ struct Settings
|
||||
std::wstring autoSuffixChars = L")]}»”’\".,!?:;";
|
||||
std::vector<std::wstring> autoLanguages{L"en", L"uk"};
|
||||
std::vector<std::wstring> autoNever;
|
||||
std::map<std::wstring, std::vector<std::wstring>> autoWords;
|
||||
std::vector<std::wstring> autoExcluded{
|
||||
L"Code", L"rider64", L"devenv", L"WindowsTerminal", L"cmd", L"powershell", L"pwsh",
|
||||
L"Far", L"ConEmu64", L"putty", L"KeePass", L"KeePassXC", L"1Password", L"Bitwarden"};
|
||||
|
||||
@@ -92,6 +92,17 @@ void validateSettings(const Settings &s)
|
||||
throw std::runtime_error("Unsupported auto-correction language");
|
||||
if (s.autoNever.size() > 10000 || s.autoExcluded.size() > 1000)
|
||||
throw std::runtime_error("Too many auto-correction exceptions");
|
||||
for (const auto &[language, words] : s.autoWords)
|
||||
{
|
||||
std::wstring alphabet = language == L"en" ? L"abcdefghijklmnopqrstuvwxyz" :
|
||||
language == L"uk" ? L"абвгґдеєжзиіїйклмнопрстуфхцчшщьюя" :
|
||||
language == L"ru" ? L"абвгдеёжзийклмнопрстуфхцчшщъыьэюя" : L"";
|
||||
if (alphabet.empty() || words.size() > 10000)
|
||||
throw std::runtime_error("Invalid personal dictionary language or too many words");
|
||||
for (const auto &word : words)
|
||||
if (word.empty() || word.size() > 32 || lower(word).find_first_not_of(alphabet) != word.npos)
|
||||
throw std::runtime_error("Personal words must contain 1-32 letters of their selected language");
|
||||
}
|
||||
if (s.maxItems < 10 || s.maxItems > 10000 || s.maxMiB < 1 || s.maxMiB > 2048 || s.maxImageMiB < 1 ||
|
||||
s.maxImageMiB > 64)
|
||||
throw std::runtime_error("History limits: 10–10000 items, 1–2048 MiB total, 1–64 MiB per image");
|
||||
@@ -143,6 +154,10 @@ Settings decodeSettings(const Bytes &bytes, const std::vector<Layout> &layouts)
|
||||
s.autoMinLength = j.value("autoCorrectMinLength", 3);
|
||||
s.autoPrefixChars = str(j, "autoCorrectPrefixChars", s.autoPrefixChars);
|
||||
s.autoSuffixChars = str(j, "autoCorrectSuffixChars", s.autoSuffixChars);
|
||||
if (j.contains("autoCorrectWords"))
|
||||
for (auto &[language, words] : j.at("autoCorrectWords").items())
|
||||
for (auto &word : words)
|
||||
s.autoWords[wide(language)].push_back(lower(wide(word.get<std::string>())));
|
||||
for (auto pair : {std::pair{"autoCorrectLanguages", &s.autoLanguages},
|
||||
{"autoCorrectNever", &s.autoNever},
|
||||
{"autoCorrectExcludedApps", &s.autoExcluded}})
|
||||
@@ -229,6 +244,13 @@ Bytes encodeSettings(const Settings &s)
|
||||
j["autoCorrectMinLength"] = s.autoMinLength;
|
||||
j["autoCorrectPrefixChars"] = utf8(s.autoPrefixChars);
|
||||
j["autoCorrectSuffixChars"] = utf8(s.autoSuffixChars);
|
||||
j["autoCorrectWords"] = Json::object();
|
||||
for (const auto &[language, words] : s.autoWords)
|
||||
{
|
||||
auto &items = j["autoCorrectWords"][utf8(language)] = Json::array();
|
||||
for (const auto &word : words)
|
||||
items.push_back(utf8(lower(word)));
|
||||
}
|
||||
for (auto pair : {std::pair{"autoCorrectLanguages", &s.autoLanguages},
|
||||
{"autoCorrectNever", &s.autoNever},
|
||||
{"autoCorrectExcludedApps", &s.autoExcluded}})
|
||||
|
||||
+22
-5
@@ -1009,6 +1009,15 @@ LRESULT CALLBACK autoOptionsProc(HWND w, UINT msg, WPARAM wp, LPARAM lp)
|
||||
if (checked(w, 403))
|
||||
draft.autoLanguages.push_back(L"ru");
|
||||
draft.autoNever = parseCommaList(value(w, 405));
|
||||
for (auto [id, language] : {std::pair{411, L"en"}, {412, L"uk"}, {413, L"ru"}})
|
||||
{
|
||||
auto text = lower(value(w, id));
|
||||
std::replace(text.begin(), text.end(), L'\n', L',');
|
||||
auto words = parseCommaList(text);
|
||||
std::sort(words.begin(), words.end());
|
||||
words.erase(std::unique(words.begin(), words.end()), words.end());
|
||||
draft.autoWords[language] = std::move(words);
|
||||
}
|
||||
draft.autoExcluded = parseCommaList(value(w, 404));
|
||||
draft.autoPrefixChars = value(w, 408);
|
||||
draft.autoSuffixChars = value(w, 409);
|
||||
@@ -1017,6 +1026,7 @@ LRESULT CALLBACK autoOptionsProc(HWND w, UINT msg, WPARAM wp, LPARAM lp)
|
||||
validateSettings(draft);
|
||||
view->settings->draft.autoLanguages = draft.autoLanguages;
|
||||
view->settings->draft.autoNever = draft.autoNever;
|
||||
view->settings->draft.autoWords = draft.autoWords;
|
||||
view->settings->draft.autoExcluded = draft.autoExcluded;
|
||||
view->settings->draft.autoMaxRank = draft.autoMaxRank;
|
||||
view->settings->draft.autoMinLength = draft.autoMinLength;
|
||||
@@ -1063,13 +1073,20 @@ void showAutoOptions(HWND parent, SettingsView *v)
|
||||
for (auto &lang : v->draft.autoLanguages)
|
||||
check(w, lang == L"en" ? 401 : lang == L"uk" ? 402 : 403, true);
|
||||
label(w, L"Excluded applications (process names, comma-separated)", 18, 94, 525);
|
||||
edit(w, 404, 18, 122, 524, 84, ES_MULTILINE | WS_VSCROLL | ES_AUTOVSCROLL);
|
||||
edit(w, 404, 18, 122, 524, 52, ES_MULTILINE | WS_VSCROLL | ES_AUTOVSCROLL);
|
||||
value(w, 404, commaList(v->draft.autoExcluded));
|
||||
label(w, L"Personal dictionary: never correct these words (comma-separated)", 18, 220, 525);
|
||||
edit(w, 405, 18, 248, 524, 84, ES_MULTILINE | WS_VSCROLL | ES_AUTOVSCROLL);
|
||||
label(w, L"Never correct these words (comma-separated)", 18, 184, 525);
|
||||
edit(w, 405, 18, 212, 524, 52, ES_MULTILINE | WS_VSCROLL | ES_AUTOVSCROLL);
|
||||
value(w, 405, commaList(v->draft.autoNever));
|
||||
label(w, L"Corrects after Space and a short pause.", 18, 344, 525);
|
||||
label(w, L"Undo skips this word until KSwitcher restarts.", 18, 366, 525);
|
||||
label(w, L"My dictionary: correct words, separated by commas or new lines", 18, 276, 525);
|
||||
for (auto [id, language] : {std::pair{411, L"en"}, {412, L"uk"}, {413, L"ru"}})
|
||||
{
|
||||
int x = 18 + (id - 411) * 178;
|
||||
label(w, id == 411 ? L"English" : id == 412 ? L"Ukrainian" : L"Russian", x, 302, 168);
|
||||
edit(w, id, x, 328, 168, 56, ES_MULTILINE | WS_VSCROLL | ES_AUTOVSCROLL | ES_WANTRETURN);
|
||||
auto words = v->draft.autoWords.find(language);
|
||||
if (words != v->draft.autoWords.end()) value(w, id, commaList(words->second));
|
||||
}
|
||||
label(w, L"Frequency rank limit (0 = entire dictionary)", 18, 400, 390);
|
||||
edit(w, 406, 420, 396, 122, 26, ES_NUMBER);
|
||||
value(w, 406, std::to_wstring(v->draft.autoMaxRank));
|
||||
|
||||
+1
-1
@@ -2,7 +2,7 @@
|
||||
#include "platform.h"
|
||||
namespace ks
|
||||
{
|
||||
inline constexpr wchar_t AppVersion[] = L"0.6.9";
|
||||
inline constexpr wchar_t AppVersion[] = L"0.6.10";
|
||||
inline constexpr UINT UpdateMessage = WM_APP + 7;
|
||||
inline constexpr UINT UpdateCheckIntervalMs = 6 * 60 * 60 * 1000;
|
||||
inline constexpr UINT UpdateNotificationRetryMs = 5 * 60 * 1000;
|
||||
|
||||
@@ -586,6 +586,29 @@ int main(int argc, char **argv) {
|
||||
fitted=ks::fitWindowRect({2000,2000,2600,2400},{-1920,0,0,1080});
|
||||
require(fitted.left==-600&&fitted.top==680&&fitted.right==0,"negative-coordinate monitor supported");
|
||||
ks::showSettings(app,false);
|
||||
SendMessageW(app.settingsWindow,WM_COMMAND,255,0);
|
||||
auto autoOptions=FindWindowW(L"KSwitcher.AutoOptions",nullptr);
|
||||
require(autoOptions!=nullptr,"auto-correction options open");
|
||||
ShowWindow(autoOptions,SW_HIDE);
|
||||
for(int id:{411,412,413}) {
|
||||
RECT control{},client{};
|
||||
GetWindowRect(GetDlgItem(autoOptions,id),&control);
|
||||
MapWindowPoints(nullptr,autoOptions,reinterpret_cast<POINT*>(&control),2);
|
||||
GetClientRect(autoOptions,&client);
|
||||
require(control.left>=0&&control.top>=0&&control.right<=client.right&&control.bottom<=client.bottom,
|
||||
"personal dictionary field fits options window");
|
||||
}
|
||||
SetDlgItemTextW(autoOptions,412,L"Релізик\r\nрелізик, словечко");
|
||||
SendMessageW(autoOptions,WM_COMMAND,IDOK,0);
|
||||
require(!IsWindow(autoOptions),"personal words accepted without file editing");
|
||||
SendMessageW(app.settingsWindow,WM_COMMAND,255,0);
|
||||
autoOptions=FindWindowW(L"KSwitcher.AutoOptions",nullptr);
|
||||
ShowWindow(autoOptions,SW_HIDE);
|
||||
wchar_t personalWords[128]{};
|
||||
GetDlgItemTextW(autoOptions,412,personalWords,128);
|
||||
require(std::wstring(personalWords)==L"релізик, словечко",
|
||||
"personal dictionary normalizes case, newlines and duplicate words");
|
||||
SendMessageW(autoOptions,WM_COMMAND,IDCANCEL,0);
|
||||
if (!app.history) app.history=std::make_unique<ks::History>(temp/L"history-layout");
|
||||
ks::showHistory(app,false);
|
||||
auto history=app.historyWindow;
|
||||
|
||||
@@ -3,6 +3,17 @@
|
||||
#include <fstream>
|
||||
#include <commctrl.h>
|
||||
|
||||
namespace ks
|
||||
{
|
||||
struct InputRegressionAccess
|
||||
{
|
||||
static bool deferring(App &app)
|
||||
{
|
||||
std::lock_guard lock(app.stateMutex);
|
||||
return app.typingTarget != nullptr;
|
||||
}
|
||||
};
|
||||
}
|
||||
namespace
|
||||
{
|
||||
ks::App *app{};
|
||||
@@ -11,6 +22,7 @@ ks::Clip saved;
|
||||
bool captured = false, failed = false;
|
||||
int stage = 0, ru = -1, en = -1;
|
||||
size_t punctuationCase = 0;
|
||||
bool typeDuringCorrection = false, typedDuringCorrection = false;
|
||||
const std::vector<std::pair<std::wstring, std::wstring>> punctuationCases{
|
||||
{L"(ljvf) ", L"(дома) "}, {L"[ljvf] ", L"[дома] "}, {L"{ljvf} ", L"{дома} "},
|
||||
{L"ljvf: ", L"дома: "}, {L"ljvf; ", L"дома; "}, {L"#ljvf%; ", L"#дома%; "}};
|
||||
@@ -25,7 +37,14 @@ size_t diagnosticStart{};
|
||||
void require(bool ok, const char *name)
|
||||
{
|
||||
if (!ok)
|
||||
{
|
||||
wchar_t observed[2048]{};
|
||||
GetWindowTextW(field, observed, 2048);
|
||||
std::cerr << "Editor at failure: " << ks::utf8(observed)
|
||||
<< "; foregroundSame=" << (GetForegroundWindow() == host)
|
||||
<< "; inputVersion=" << app->inputVersion.load() << '\n';
|
||||
throw std::runtime_error(name);
|
||||
}
|
||||
std::cout << "PASS " << name << '\n';
|
||||
}
|
||||
std::wstring text()
|
||||
@@ -80,7 +99,26 @@ LRESULT CALLBACK procedure(HWND w, UINT msg, WPARAM wp, LPARAM lp)
|
||||
if (GetTickCount64() > deadline)
|
||||
throw std::runtime_error("Native integration timed out");
|
||||
if (app->busy)
|
||||
{
|
||||
if (typeDuringCorrection && ks::InputRegressionAccess::deferring(*app))
|
||||
{
|
||||
typeDuringCorrection = false;
|
||||
typedDuringCorrection = true;
|
||||
std::vector<INPUT> inputs;
|
||||
for (auto [key, up] : std::vector<std::pair<WORD, bool>>{
|
||||
{'F', false}, {'F', true}, {VK_LSHIFT, false},
|
||||
{'F', false}, {'F', true}, {VK_LSHIFT, true},
|
||||
{'B', false}, {'B', true}, {VK_BACK, false}, {VK_BACK, true}})
|
||||
{
|
||||
auto input = ks::keyboardInput(key, up ? KEYEVENTF_KEYUP : 0);
|
||||
input.ki.dwExtraInfo = 123; // External input must traverse the actual hook.
|
||||
inputs.push_back(input);
|
||||
}
|
||||
require(SendInput((UINT)inputs.size(), inputs.data(), sizeof(INPUT)) == inputs.size(),
|
||||
"concurrent typing submitted during selection/paste transaction");
|
||||
}
|
||||
return 0;
|
||||
}
|
||||
switch (stage++)
|
||||
{
|
||||
case 0:
|
||||
@@ -242,6 +280,41 @@ LRESULT CALLBACK procedure(HWND w, UINT msg, WPARAM wp, LPARAM lp)
|
||||
break;
|
||||
case 17:
|
||||
require(text() == punctuationCases.back().first, "undo preserves custom punctuation");
|
||||
[[fallthrough]];
|
||||
case 20:
|
||||
if (!captured)
|
||||
{
|
||||
saved = ks::readClipboard(app->window, 64 * 1024 * 1024);
|
||||
captured = true;
|
||||
app->ownSequence = ks::writeClipboard(app->window, ks::textClip(sentinel));
|
||||
}
|
||||
app->settings.autoCorrect = true;
|
||||
app->settings.switchLayout = true;
|
||||
app->settings.autoLanguages = {L"en", L"ru"};
|
||||
app->settings.cycle = {app->layouts[en].id, app->layouts[ru].id};
|
||||
app->autoCorrector = ks::AutoCorrector{};
|
||||
ActivateKeyboardLayout(app->layouts[en].handle, 0);
|
||||
prepare(L"ghbdtn ", false);
|
||||
record(L"ghbdtn ");
|
||||
typeDuringCorrection = true;
|
||||
app->transform(ks::AutoCorrect);
|
||||
stage = 18;
|
||||
break;
|
||||
case 18:
|
||||
require(typedDuringCorrection && text() == L"привет аА",
|
||||
"typing during correction replays in order with shift and backspace in new layout");
|
||||
require(ks::readClipboard(app->window).text() == sentinel,
|
||||
"concurrent typing preserves clipboard restoration");
|
||||
ActivateKeyboardLayout(app->layouts[en].handle, 0);
|
||||
prepare(L"OLDghbdtn ", false);
|
||||
record(L"ghbdtn ");
|
||||
typedDuringCorrection = false;
|
||||
typeDuringCorrection = true;
|
||||
app->transform(ks::AutoCorrect);
|
||||
break;
|
||||
case 19:
|
||||
require(typedDuringCorrection && text() == L"OLDghbdtn fF",
|
||||
"rejected correction collapses selection before replaying typing in original layout");
|
||||
finish();
|
||||
break;
|
||||
}
|
||||
@@ -366,6 +439,8 @@ int main(int argc, char **argv)
|
||||
CreateWindowW(L"EDIT", L"", WS_CHILD | WS_VISIBLE | WS_BORDER | ES_MULTILINE | ES_WANTRETURN,
|
||||
20, 20, 590, 230, host, nullptr, cls.hInstance, nullptr);
|
||||
deadline = GetTickCount64() + 30000;
|
||||
if (argc > 1 && std::string(argv[1]) == "--typing-only")
|
||||
stage = 20;
|
||||
SetTimer(host, 1, 100, nullptr);
|
||||
app->execute();
|
||||
}
|
||||
|
||||
@@ -241,6 +241,37 @@ int main(int argc, char **argv)
|
||||
require(corrector.memoryBytes() < 3 * 1024 * 1024, "two dictionaries fit within 3 MiB");
|
||||
require(!ks::autoCorrectionAllowed(nullptr, autoSettings), "missing focused editor fails closed");
|
||||
ks::AutoCorrector missing;
|
||||
{
|
||||
ks::AutoCorrector supplemental;
|
||||
auto options = autoSettings;
|
||||
options.autoLanguages = {L"en", L"uk", L"ru"};
|
||||
options.cycle = {L"en", L"uk", L"ru"};
|
||||
supplemental.configure(L"dictionaries", options.autoLanguages);
|
||||
auto release = supplemental.suggest(L"релыз ", 1, layouts, options);
|
||||
require(release && release->text == L"реліз " && release->target == 2,
|
||||
"supplemental Ukrainian release word corrects from Russian layout");
|
||||
require(!supplemental.suggest(L"реліз ", 2, layouts, options),
|
||||
"correct Ukrainian release word stays unchanged");
|
||||
require(!supplemental.suggest(L"релиз ", 1, layouts, options),
|
||||
"valid Russian release word stays unchanged");
|
||||
options.autoMaxRank = 50000;
|
||||
require(!supplemental.suggest(L"релыз ", 1, layouts, options),
|
||||
"supplemental vocabulary respects explicit frequency limit");
|
||||
options.autoWords[L"uk"] = {L"релізик"};
|
||||
options.autoMaxRank = 1;
|
||||
auto custom = supplemental.suggest(L"релызик ", 1, layouts, options);
|
||||
require(custom && custom->text == L"релізик " && custom->target == 2,
|
||||
"personal word enables correction without editing dictionary files");
|
||||
require(!supplemental.suggest(L"релізик ", 2, layouts, options),
|
||||
"personal word is protected from correction");
|
||||
options.autoNever = {L"релызик"};
|
||||
require(!supplemental.suggest(L"релызик ", 1, layouts, options),
|
||||
"never-correct list takes precedence over personal word");
|
||||
options.autoNever.clear();
|
||||
options.cycle = {L"en", L"ru"};
|
||||
require(!supplemental.suggest(L"релызик ", 1, layouts, options),
|
||||
"personal dictionary respects disabled target layouts");
|
||||
}
|
||||
rejects([&] { missing.configure(L"missing-dictionaries", {L"en"}); }, "missing dictionary rejected");
|
||||
require(ks::convertTo(L"ghbdtn", layouts, 1, 0) == L"привет", "hint conversion");
|
||||
require(ks::convertTo(L"привет", layouts, 1, 1) == L"привет", "same source is unchanged");
|
||||
@@ -335,6 +366,25 @@ int main(int argc, char **argv)
|
||||
auto temp = std::filesystem::temp_directory_path() /
|
||||
(L"KSwitcher-tests-" + std::to_wstring(GetCurrentProcessId()));
|
||||
std::filesystem::create_directories(temp);
|
||||
{
|
||||
auto folder = temp / L"dictionaries";
|
||||
std::filesystem::create_directories(folder);
|
||||
auto put = [&](const wchar_t *name, const std::wstring &text)
|
||||
{
|
||||
auto data = ks::utf8(text);
|
||||
ks::atomicWrite(folder / name, ks::Bytes(data.begin(), data.end()));
|
||||
};
|
||||
put(L"en.txt", L"zoo\t3\nhello\t99\nhello\t1\n");
|
||||
put(L"uk.txt", L"дім\t3\n");
|
||||
ks::AutoCorrector edited;
|
||||
edited.configure(folder, {L"en", L"uk"});
|
||||
auto options = ks::Settings{};
|
||||
options.cycle = {L"en", L"uk"};
|
||||
options.autoMaxRank = 1;
|
||||
auto correction = edited.suggest(L"руддщ ", 2, layouts, options);
|
||||
require(correction && correction->text == L"hello ",
|
||||
"unsorted dictionaries and duplicates load using lowest duplicate rank");
|
||||
}
|
||||
ks::Pixels pixels{2, 1, {10, 20, 30, 255, 40, 50, 60, 255}};
|
||||
ks::Clip firstImage{{ks::imageDib(pixels), {0, L"Ole Private Data", {1, 2, 3}}}};
|
||||
auto thumbnail = ks::decodeImage(firstImage, 1, 1);
|
||||
@@ -384,11 +434,16 @@ int main(int argc, char **argv)
|
||||
settings.autoPrefixChars = L"(#";
|
||||
settings.autoSuffixChars = L");%";
|
||||
settings.autoNever = {L"myWord"};
|
||||
settings.autoWords = {{L"uk", {L"релізик"}}, {L"en", {L"kswitcher"}}};
|
||||
settings.autoExcluded = {L"Code.exe"};
|
||||
settings.cycle = {L"en", L"uk"};
|
||||
settings.snippets = {{L";address", L"Line 1\r\n{cursor}Line 2"}};
|
||||
ks::saveSettings(temp / L"settings.json", settings);
|
||||
auto restored = ks::loadSettings(temp / L"settings.json", {});
|
||||
require(restored.autoWords == settings.autoWords, "personal dictionaries survive settings roundtrip");
|
||||
auto invalidWords = settings;
|
||||
invalidWords.autoWords[L"uk"] = {L"hello"};
|
||||
rejects([&] { ks::validateSettings(invalidWords); }, "personal dictionary rejects wrong language letters");
|
||||
require(ks::encodeSettings(ks::decodeSettings(ks::encodeSettings(settings), {})) == ks::encodeSettings(settings),
|
||||
"configurator snapshot survives serialization without changing the concurrency token");
|
||||
rejects([&] { ks::decodeSettings(ks::Bytes{'{', '}'}, {}); },
|
||||
|
||||
@@ -3,6 +3,8 @@ from pathlib import Path
|
||||
import urllib.request, hashlib, json
|
||||
ROOT=Path(__file__).resolve().parents[1]/'dictionaries'
|
||||
REV='525f9b560de45753a5ea01069454e72e9aa541c6'
|
||||
# Supplemental vocabulary has no corpus frequency: rank after the 50k source list.
|
||||
SUPPLEMENTAL={'uk': {'реліз': 50001}}
|
||||
ROOT.mkdir(exist_ok=True)
|
||||
manifest={}
|
||||
for lang in ('en','uk','ru'):
|
||||
@@ -15,8 +17,12 @@ for lang in ('en','uk','ru'):
|
||||
if not (2<=len(word)<=40 and all(c in alphabet for c in word)):continue
|
||||
if int(count)<5:continue
|
||||
rows.setdefault(word,rank)
|
||||
for word,rank in SUPPLEMENTAL.get(lang,{}).items():
|
||||
rows.setdefault(word,rank)
|
||||
out=''.join(f'{word}\t{rows[word]}\n' for word in sorted(rows))
|
||||
(ROOT/f'{lang}.txt').write_bytes(out.encode('utf-8'))
|
||||
manifest[lang]={'source':url,'sourceSha256':hashlib.sha256(data).hexdigest(),'words':len(rows),'bytes':len(out.encode()),'sha256':hashlib.sha256(out.encode()).hexdigest()}
|
||||
if lang in SUPPLEMENTAL:
|
||||
manifest[lang]['supplementalWords']=SUPPLEMENTAL[lang]
|
||||
(ROOT/'sources.json').write_text(json.dumps(manifest,indent=2)+'\n',encoding='utf-8')
|
||||
print(json.dumps(manifest,indent=2))
|
||||
|
||||
Reference in New Issue
Block a user