foot/search.c

1110 lines
32 KiB
C
Raw Normal View History

#include "search.h"
2019-12-01 19:22:45 +01:00
#include <string.h>
#include <wayland-client.h>
#include <xkbcommon/xkbcommon-compose.h>
#define LOG_MODULE "search"
2019-08-27 21:11:29 +02:00
#define LOG_ENABLE_DBG 0
#include "log.h"
#include "char32.h"
#include "config.h"
#include "extract.h"
2019-08-27 21:11:40 +02:00
#include "grid.h"
#include "input.h"
key-binding: new API, for handling sets of key bindings Up until now, our Wayland seats have been tracking key bindings. This makes sense, since the seat’s keymap determines how the key bindings are resolved. However, tying bindings to the seat/keymap alone isn’t enough, since we also depend on the current configuration (i.e. user settings) when resolving a key binding. This means configurations that doesn’t match the wayland object’s configuration, currently don’t resolve key bindings correctly. This applies to footclients where the user has overridden key bindings on the command line (e.g. --override key-bindings.foo=bar). Thus, to correctly resolve key bindings, each set of key bindings must be tied *both* to a seat/keymap, *and* a configuration. This patch introduces a key-binding manager, with an API to add/remove/lookup, and load/unload keymaps from sets of key bindings. In the API, sets are tied to a seat and terminal instance, since this makes the most sense (we need to instantiate, or incref a set whenever a new terminal instance is created). Internally, the set is tied to a seat and the terminal’s configuration. Sets are *added* when a new seat is added, and when a new terminal instance is created. Since there can only be one instance of each seat, sets are always removed when a seat is removed. Terminals on the other hand can re-use the same configuration (and typically do). Thus, sets ref-count the configuration. In other words, when instantiating a new terminal, we may not have to instantiate a new set of key bindings, but can often be incref:ed instead. Whenever the keymap changes on a seat, all key bindings sets associated with that seat reloads (re-resolves) their key bindings. Closes #931
2022-04-17 15:39:51 +02:00
#include "key-binding.h"
#include "misc.h"
#include "render.h"
#include "selection.h"
#include "shm.h"
#include "util.h"
#include "xmalloc.h"
/*
* Ensures a "new" viewport doesn't contain any unallocated rows.
*
* This is done by first checking if the *first* row is NULL. If so,
* we move the viewport *forward*, until the first row is non-NULL. At
* this point, the entire viewport should be allocated rows only.
*
* If the first row already was non-NULL, we instead check the *last*
* row, and if it is NULL, we move the viewport *backward* until the
* last row is non-NULL.
*/
static int
ensure_view_is_allocated(struct terminal *term, int new_view)
{
struct grid *grid = term->grid;
int view_end = (new_view + term->rows - 1) & (grid->num_rows - 1);
if (grid->rows[new_view] == NULL) {
while (grid->rows[new_view] == NULL)
new_view = (new_view + 1) & (grid->num_rows - 1);
}
else if (grid->rows[view_end] == NULL) {
while (grid->rows[view_end] == NULL) {
new_view--;
if (new_view < 0)
new_view += grid->num_rows;
view_end = (new_view + term->rows - 1) & (grid->num_rows - 1);
}
}
#if defined(_DEBUG)
for (size_t r = 0; r < term->rows; r++)
xassert(grid->rows[(new_view + r) & (grid->num_rows - 1)] != NULL);
#endif
return new_view;
}
static bool
search_ensure_size(struct terminal *term, size_t wanted_size)
{
while (wanted_size >= term->search.sz) {
size_t new_sz = term->search.sz == 0 ? 64 : term->search.sz * 2;
char32_t *new_buf = realloc(term->search.buf, new_sz * sizeof(term->search.buf[0]));
if (new_buf == NULL) {
LOG_ERRNO("failed to resize search buffer");
return false;
}
term->search.buf = new_buf;
term->search.sz = new_sz;
}
return true;
}
static bool
has_wrapped_around(const struct terminal *term, int abs_row_no)
{
const struct grid *grid = term->grid;
int scrollback_start = grid->offset + term->rows;
int rebased_row = abs_row_no - scrollback_start + grid->num_rows;
rebased_row &= grid->num_rows - 1;
return rebased_row == 0;
}
static void
search_cancel_keep_selection(struct terminal *term)
{
struct wl_window *win = term->window;
wayl_win_subsurface_destroy(&win->search);
if (term->search.len > 0) {
free(term->search.last.buf);
term->search.last.buf = term->search.buf;
term->search.last.len = term->search.len;
} else
free(term->search.buf);
term->search.buf = NULL;
term->search.len = term->search.sz = 0;
term->search.cursor = 0;
term->search.match = (struct coord){-1, -1};
term->search.match_len = 0;
term->is_searching = false;
term->render.search_glyph_offset = 0;
/* Reset IME state */
if (term_ime_is_enabled(term)) {
term_ime_disable(term);
term_ime_enable(term);
}
term_xcursor_update(term);
render_refresh(term);
}
void
search_begin(struct terminal *term)
{
LOG_DBG("search: begin");
search_cancel_keep_selection(term);
selection_cancel(term);
/* Reset IME state */
if (term_ime_is_enabled(term)) {
term_ime_disable(term);
term_ime_enable(term);
}
/* On-demand instantiate wayland surface */
bool ret = wayl_win_subsurface_new(
term->window, &term->window->search, false);
xassert(ret);
const struct grid *grid = term->grid;
term->search.original_view = grid->view;
term->search.view_followed_offset = grid->view == grid->offset;
term->is_searching = true;
term->search.len = 0;
term->search.sz = 64;
term->search.buf = xmalloc(term->search.sz * sizeof(term->search.buf[0]));
term->search.buf[0] = U'\0';
term_xcursor_update(term);
render_refresh_search(term);
}
void
search_cancel(struct terminal *term)
{
if (!term->is_searching)
return;
search_cancel_keep_selection(term);
selection_cancel(term);
}
void
search_selection_cancelled(struct terminal *term)
{
term->search.match = (struct coord){-1, -1};
term->search.match_len = 0;
render_refresh_search(term);
}
static void
search_update_selection(struct terminal *term, const struct range *match)
{
struct grid *grid = term->grid;
int start_row = match->start.row;
int start_col = match->start.col;
int end_row = match->end.row;
int end_col = match->end.col;
bool move_viewport = true;
int view_end = (grid->view + term->rows - 1) & (grid->num_rows - 1);
if (view_end >= grid->view) {
/* Viewport does *not* wrap around */
if (start_row >= grid->view && end_row <= view_end)
move_viewport = false;
} else {
/* Viewport wraps */
if (start_row >= grid->view || end_row <= view_end)
move_viewport = false;
}
if (move_viewport) {
int old_view = grid->view;
int new_view = start_row - term->rows / 2;
while (new_view < 0)
new_view += grid->num_rows;
new_view = ensure_view_is_allocated(term, new_view);
/* Don't scroll past scrollback history */
int end = (grid->offset + term->rows - 1) & (grid->num_rows - 1);
if (end >= grid->offset) {
/* Not wrapped */
if (new_view >= grid->offset && new_view <= end)
new_view = grid->offset;
} else {
if (new_view >= grid->offset || new_view <= end)
new_view = grid->offset;
}
#if defined(_DEBUG)
/* Verify all to-be-visible rows have been allocated */
for (int r = 0; r < term->rows; r++)
xassert(grid->rows[(new_view + r) & (grid->num_rows - 1)] != NULL);
#endif
/* Update view */
grid->view = new_view;
if (new_view != old_view)
term_damage_view(term);
}
#if 0
/* Selection endpoint is inclusive */
if (--end_col < 0) {
end_col = term->cols - 1;
end_row--;
}
#endif
/*
* Begin a new selection if the start coords changed
*
* Note: check column against selection.coords, since our old
* start column isnt reliable - we modify it when searching
* next or prev.
*/
if (start_row != term->search.match.row ||
start_col != term->selection.coords.start.col)
{
int selection_row = start_row - grid->view + grid->num_rows;
selection_row &= grid->num_rows - 1;
selection_start(
term, start_col, selection_row, SELECTION_CHAR_WISE, false);
}
/* Update selection endpoint */
{
int selection_row = end_row - grid->view + grid->num_rows;
selection_row &= grid->num_rows - 1;
selection_update(term, end_col, selection_row);
}
}
static ssize_t
matches_cell(const struct terminal *term, const struct cell *cell, size_t search_ofs)
{
assert(search_ofs < term->search.len);
char32_t base = cell->wc;
const struct composed *composed = NULL;
composed: store compose chains in a binary search tree The previous implementation stored compose chains in a dynamically allocated array. Adding a chain was easy: resize the array and append the new chain at the end. Looking up a compose chain given a compose chain key/index was also easy: just index into the array. However, searching for a pre-existing chain given a codepoint sequence was very slow. Since the array wasn’t sorted, we typically had to scan through the entire array, just to realize that there is no pre-existing chain, and that we need to add a new one. Since this happens for *each* codepoint in a grapheme cluster, things quickly became really slow. Things were ok:ish as long as the compose chain struct was small, as that made it possible to hold all the chains in the cache. Once the number of chains reached a certain point, or when we were forced to bump maximum number of allowed codepoints in a chain, we started thrashing the cache and things got much much worse. So what can we do? We can’t sort the array, because a) that would invalidate all existing chain keys in the grid (and iterating the entire scrollback and updating compose keys is *not* an option). b) inserting a chain becomes slow as we need to first find _where_ to insert it, and then memmove() the rest of the array. This patch uses a binary search tree to store the chains instead of a simple array. The tree is sorted on a “key”, which is the XOR of all codepoints, truncated to the CELL_COMB_CHARS_HI-CELL_COMB_CHARS_LO range. The grid now stores CELL_COMB_CHARS_LO+key, instead of CELL_COMB_CHARS_LO+index. Since the key is truncated, collisions may occur. This is handled by incrementing the key by 1. Lookup is of course slower than before, O(log n) instead of O(1). Insertion is slightly slower as well: technically it’s O(log n) instead of O(1). However, we also need to take into account the re-allocating the array will occasionally force a full copy of the array when it cannot simply be growed. But finding a pre-existing chain is now *much* faster: O(log n) instead of O(n). In most cases, the first lookup will either succeed (return a true match), or fail (return NULL). However, since key collisions are possible, it may also return false matches. This means we need to verify the contents of the chain before deciding to use it instead of inserting a new chain. But remember that this comparison was being done for each and every chain in the previous implementation. With lookups being much faster, and in particular, no longer requiring us to check the chain contents for every singlec chain, we can now use a dynamically allocated ‘chars’ array in the chain. This was previously a hardcoded array of 10 chars. Using a dynamic allocated array means looking in the array is slower, since we now need two loads: one to load the pointer, and a second to load _from_ the pointer. As a result, the base size of a compose chain (i.e. an “empty” chain) has now been reduced from 48 bytes to 32. A chain with two codepoints is 40 bytes. This means we have up to 4 codepoints while still using less, or the same amount, of memory as before. Furthermore, the Unicode random test (i.e. write random “unicode” chars) is now **faster** than current master (i.e. before text-shaping support was added), **with** test-shaping enabled. With text-shaping disabled, we’re _even_ faster.
2021-06-24 13:17:07 +02:00
if (base >= CELL_COMB_CHARS_LO && base <= CELL_COMB_CHARS_HI)
{
composed: store compose chains in a binary search tree The previous implementation stored compose chains in a dynamically allocated array. Adding a chain was easy: resize the array and append the new chain at the end. Looking up a compose chain given a compose chain key/index was also easy: just index into the array. However, searching for a pre-existing chain given a codepoint sequence was very slow. Since the array wasn’t sorted, we typically had to scan through the entire array, just to realize that there is no pre-existing chain, and that we need to add a new one. Since this happens for *each* codepoint in a grapheme cluster, things quickly became really slow. Things were ok:ish as long as the compose chain struct was small, as that made it possible to hold all the chains in the cache. Once the number of chains reached a certain point, or when we were forced to bump maximum number of allowed codepoints in a chain, we started thrashing the cache and things got much much worse. So what can we do? We can’t sort the array, because a) that would invalidate all existing chain keys in the grid (and iterating the entire scrollback and updating compose keys is *not* an option). b) inserting a chain becomes slow as we need to first find _where_ to insert it, and then memmove() the rest of the array. This patch uses a binary search tree to store the chains instead of a simple array. The tree is sorted on a “key”, which is the XOR of all codepoints, truncated to the CELL_COMB_CHARS_HI-CELL_COMB_CHARS_LO range. The grid now stores CELL_COMB_CHARS_LO+key, instead of CELL_COMB_CHARS_LO+index. Since the key is truncated, collisions may occur. This is handled by incrementing the key by 1. Lookup is of course slower than before, O(log n) instead of O(1). Insertion is slightly slower as well: technically it’s O(log n) instead of O(1). However, we also need to take into account the re-allocating the array will occasionally force a full copy of the array when it cannot simply be growed. But finding a pre-existing chain is now *much* faster: O(log n) instead of O(n). In most cases, the first lookup will either succeed (return a true match), or fail (return NULL). However, since key collisions are possible, it may also return false matches. This means we need to verify the contents of the chain before deciding to use it instead of inserting a new chain. But remember that this comparison was being done for each and every chain in the previous implementation. With lookups being much faster, and in particular, no longer requiring us to check the chain contents for every singlec chain, we can now use a dynamically allocated ‘chars’ array in the chain. This was previously a hardcoded array of 10 chars. Using a dynamic allocated array means looking in the array is slower, since we now need two loads: one to load the pointer, and a second to load _from_ the pointer. As a result, the base size of a compose chain (i.e. an “empty” chain) has now been reduced from 48 bytes to 32. A chain with two codepoints is 40 bytes. This means we have up to 4 codepoints while still using less, or the same amount, of memory as before. Furthermore, the Unicode random test (i.e. write random “unicode” chars) is now **faster** than current master (i.e. before text-shaping support was added), **with** test-shaping enabled. With text-shaping disabled, we’re _even_ faster.
2021-06-24 13:17:07 +02:00
composed = composed_lookup(term->composed, base - CELL_COMB_CHARS_LO);
2020-08-20 19:25:35 +02:00
base = composed->chars[0];
}
if (composed == NULL && base == 0 && term->search.buf[search_ofs] == U' ')
return 1;
if (c32ncasecmp(&base, &term->search.buf[search_ofs], 1) != 0)
return -1;
if (composed != NULL) {
if (search_ofs + 1 + composed->count > term->search.len)
return -1;
2020-08-20 19:25:35 +02:00
for (size_t j = 1; j < composed->count; j++) {
if (composed->chars[j] != term->search.buf[search_ofs + 1 + j])
return -1;
}
}
return composed != NULL ? 1 + composed->count : 1;
}
static bool
find_next(struct terminal *term, enum search_direction direction,
struct coord abs_start, struct coord abs_end, struct range *match)
{
#define ROW_DEC(_r) ((_r) = ((_r) - 1 + grid->num_rows) & (grid->num_rows - 1))
#define ROW_INC(_r) ((_r) = ((_r) + 1) & (grid->num_rows - 1))
struct grid *grid = term->grid;
const bool backward = direction != SEARCH_FORWARD;
LOG_DBG("%s: start: %dx%d, end: %dx%d", backward ? "backward" : "forward",
abs_start.row, abs_start.col, abs_end.row, abs_end.col);
xassert(abs_start.row >= 0);
xassert(abs_start.row < grid->num_rows);
xassert(abs_start.col >= 0);
xassert(abs_start.col < term->cols);
xassert(abs_end.row >= 0);
xassert(abs_end.row < grid->num_rows);
xassert(abs_end.col >= 0);
xassert(abs_end.col < term->cols);
for (int match_start_row = abs_start.row, match_start_col = abs_start.col;
;
backward ? ROW_DEC(match_start_row) : ROW_INC(match_start_row)) {
const struct row *row = grid->rows[match_start_row];
if (row == NULL) {
if (match_start_row == abs_end.row)
break;
continue;
}
for (;
backward ? match_start_col >= 0 : match_start_col < term->cols;
backward ? match_start_col-- : match_start_col++)
{
if (matches_cell(term, &row->cells[match_start_col], 0) < 0) {
if (match_start_row == abs_end.row &&
match_start_col == abs_end.col)
{
break;
}
continue;
}
/*
* Got a match on the first letter. Now we'll see if the
* rest of the search buffer matches.
*/
LOG_DBG("search: initial match at row=%d, col=%d",
match_start_row, match_start_col);
int match_end_row = match_start_row;
int match_end_col = match_start_col;
const struct row *match_row = row;
size_t match_len = 0;
for (size_t i = 0; i < term->search.len;) {
if (match_end_col >= term->cols) {
ROW_INC(match_end_row);
match_end_col = 0;
match_row = grid->rows[match_end_row];
if (match_row == NULL)
break;
}
if (match_row->cells[match_end_col].wc >= CELL_SPACER) {
match_end_col++;
continue;
}
ssize_t additional_chars = matches_cell(
term, &match_row->cells[match_end_col], i);
if (additional_chars < 0)
break;
i += additional_chars;
match_len += additional_chars;
match_end_col++;
}
if (match_len != term->search.len) {
/* Didn't match (completely) */
continue;
}
*match = (struct range){
.start = {match_start_col, match_start_row},
.end = {match_end_col - 1, match_end_row},
};
return true;
}
if (match_start_row == abs_end.row && match_start_col == abs_end.col)
break;
match_start_col = backward ? term->cols - 1 : 0;
}
return false;
}
static void
search_find_next(struct terminal *term, enum search_direction direction)
{
struct grid *grid = term->grid;
if (term->search.len == 0) {
term->search.match = (struct coord){-1, -1};
term->search.match_len = 0;
selection_cancel(term);
return;
}
struct coord start = term->search.match;
size_t len = term->search.match_len;
xassert((len == 0 && start.row == -1 && start.col == -1) ||
(len > 0 && start.row >= 0 && start.col >= 0));
if (len == 0) {
/* No previous match, start from the top, or bottom, of the scrollback */
switch (direction) {
case SEARCH_FORWARD:
start.row = grid_row_absolute_in_view(grid, 0);
start.col = 0;
break;
case SEARCH_BACKWARD:
case SEARCH_BACKWARD_SAME_POSITION:
start.row = grid_row_absolute_in_view(grid, term->rows - 1);
start.col = term->cols - 1;
break;
}
} else {
/* Continue from last match */
xassert(start.row >= 0);
xassert(start.col >= 0);
switch (direction) {
case SEARCH_BACKWARD_SAME_POSITION:
break;
case SEARCH_BACKWARD:
if (--start.col < 0) {
start.col = term->cols - 1;
start.row += grid->num_rows - 1;
start.row &= grid->num_rows - 1;
}
break;
case SEARCH_FORWARD:
if (++start.col >= term->cols) {
start.col = 0;
start.row++;
start.row &= grid->num_rows - 1;
}
break;
}
xassert(start.row >= 0);
xassert(start.row < grid->num_rows);
xassert(start.col >= 0);
xassert(start.col < term->cols);
}
LOG_DBG(
"update: %s: starting at row=%d col=%d "
"(offset = %d, view = %d)",
direction != SEARCH_FORWARD ? "backward" : "forward",
start.row, start.col,
grid->offset, grid->view);
struct coord end = start;
switch (direction) {
case SEARCH_FORWARD:
/* Search forward, until we reach the cell *before* current start */
if (--end.col < 0) {
end.col = term->cols - 1;
end.row += grid->num_rows - 1;
end.row &= grid->num_rows - 1;
}
break;
case SEARCH_BACKWARD:
case SEARCH_BACKWARD_SAME_POSITION:
/* Search backwards, until we reach the cell *after* current start */
if (++end.col >= term->cols) {
end.col = 0;
end.row++;
end.row &= grid->num_rows - 1;
}
break;
}
struct range match;
bool found = find_next(term, direction, start, end, &match);
if (found) {
LOG_DBG("primary match found at %dx%d",
match.start.row, match.start.col);
search_update_selection(term, &match);
term->search.match = match.start;
term->search.match_len = term->search.len;
} else {
LOG_DBG("no match");
term->search.match = (struct coord){-1, -1};
term->search.match_len = 0;
selection_cancel(term);
}
#undef ROW_DEC
}
struct search_match_iterator
search_matches_new_iter(struct terminal *term)
{
return (struct search_match_iterator){
.term = term,
.start = {-2, -2},
};
}
struct range
search_matches_next(struct search_match_iterator *iter)
{
struct terminal *term = iter->term;
struct grid *grid = term->grid;
if (term->search.match_len == 0)
goto no_match;
struct range match;
bool found;
const bool return_primary_match =
iter->start.row == -2 && term->selection.coords.end.row >= 0;
if (return_primary_match) {
/* First, return the primary match */
match = term->selection.coords;
found = true;
}
else if (iter->start.row >= term->rows) {
goto no_match;
}
else {
xassert(iter->start.row >= 0);
xassert(iter->start.row < term->rows);
xassert(iter->start.col >= 0);
xassert(iter->start.col < term->cols);
struct coord abs_start = iter->start;
abs_start.row = grid_row_absolute_in_view(grid, abs_start.row);
struct coord abs_end = {
term->cols - 1,
grid_row_absolute_in_view(grid, term->rows - 1)};
found = find_next(term, SEARCH_FORWARD, abs_start, abs_end, &match);
}
if (found) {
LOG_DBG("match at %dx%d-%dx%d",
match.start.row, match.start.col,
match.end.row, match.end.col);
/* Convert absolute row numbers back to view relative */
match.start.row = match.start.row - grid->view + grid->num_rows;
match.start.row &= grid->num_rows - 1;
match.end.row = match.end.row - grid->view + grid->num_rows;
match.end.row &= grid->num_rows - 1;
xassert(match.start.row >= 0);
xassert(match.start.row < term->rows);
xassert(match.end.row >= 0);
xassert(match.end.row < term->rows);
xassert(match.end.row > match.start.row ||
(match.end.row == match.start.row &&
match.end.col >= match.start.col));
if (return_primary_match) {
iter->start.row = 0;
iter->start.col = 0;
} else {
/* Continue at next column, next time */
iter->start.row = match.start.row;
iter->start.col = match.start.col + 1;
if (iter->start.col >= term->cols) {
iter->start.col = 0;
iter->start.row++; /* Overflow is caught in next iteration */
}
xassert(iter->start.row >= 0);
xassert(iter->start.row <= term->rows);
xassert(iter->start.col >= 0);
xassert(iter->start.col < term->cols);
if (match.start.row == term->search.match.row &&
match.start.col == term->search.match.col)
{
/* Primary match is handled explicitly */
LOG_DBG("primary match: skipping");
return search_matches_next(iter);
}
}
return match;
}
no_match:
iter->start.row = -1;
iter->start.col = -1;
return (struct range){{-1, -1}, {-1, -1}};
}
static void
add_wchars(struct terminal *term, char32_t *src, size_t count)
{
/* Strip non-printable characters */
for (size_t i = 0, j = 0, orig_count = count; i < orig_count; i++) {
if (isc32print(src[i]))
src[j++] = src[i];
else
count--;
}
if (!search_ensure_size(term, term->search.len + count))
return;
xassert(term->search.len + count < term->search.sz);
memmove(&term->search.buf[term->search.cursor + count],
&term->search.buf[term->search.cursor],
(term->search.len - term->search.cursor) * sizeof(char32_t));
memcpy(&term->search.buf[term->search.cursor], src, count * sizeof(char32_t));
term->search.len += count;
term->search.cursor += count;
term->search.buf[term->search.len] = U'\0';
}
void
search_add_chars(struct terminal *term, const char *src, size_t count)
{
size_t chars = mbsntoc32(NULL, src, count, 0);
if (chars == (size_t)-1) {
LOG_ERRNO("failed to convert %.*s to Unicode", (int)count, src);
return;
}
char32_t c32s[chars + 1];
mbsntoc32(c32s, src, count, chars);
add_wchars(term, c32s, chars);
}
static void
search_match_to_end_of_word(struct terminal *term, bool spaces_only)
{
if (term->search.match_len == 0)
return;
xassert(term->selection.coords.end.row != -1);
struct grid *grid = term->grid;
const bool move_cursor = term->search.cursor == term->search.len;
const struct coord old_end = term->selection.coords.end;
struct coord new_end = old_end;
struct row *row = NULL;
/* Advances a coordinate by one column, to the right. Returns
* false if weve reached the scrollback wrap-around */
#define advance_pos(coord) __extension__ \
({ \
bool wrapped_around = false; \
if (++(coord).col >= term->cols) { \
(coord).row = ((coord).row + 1) & (grid->num_rows - 1); \
(coord).col = 0; \
row = grid->rows[(coord).row]; \
if (has_wrapped_around(term, (coord.row))) \
wrapped_around = true; \
} \
!wrapped_around; \
})
/* First character to consider is the *next* character */
if (!advance_pos(new_end))
return;
xassert(grid->rows[new_end.row] != NULL);
/* Find next word boundary */
new_end.row -= grid->view;
selection_find_word_boundary_right(term, &new_end, spaces_only);
new_end.row += grid->view;
struct coord pos = old_end;
row = grid->rows[pos.row];
struct extraction_context *ctx = extract_begin(SELECTION_NONE, false);
if (ctx == NULL)
return;
do {
if (!advance_pos(pos))
break;
if (!extract_one(term, row, &row->cells[pos.col], pos.col, ctx))
break;
} while (pos.col != new_end.col || pos.row != new_end.row);
char32_t *new_text;
size_t new_len;
if (!extract_finish_wide(ctx, &new_text, &new_len))
return;
if (!search_ensure_size(term, term->search.len + new_len))
return;
for (size_t i = 0; i < new_len; i++) {
if (new_text[i] == U'\n') {
/* extract() adds newlines, which we never match against */
continue;
}
term->search.buf[term->search.len++] = new_text[i];
}
term->search.buf[term->search.len] = U'\0';
free(new_text);
if (move_cursor)
term->search.cursor = term->search.len;
struct range match = {.start = term->search.match, .end = new_end};
search_update_selection(term, &match);
term->search.match_len = term->search.len;
#undef advance_pos
}
static size_t
distance_next_word(const struct terminal *term)
{
size_t cursor = term->search.cursor;
/* First eat non-whitespace. This is the word we're skipping past */
while (cursor < term->search.len) {
if (isc32space(term->search.buf[cursor++]))
break;
}
xassert(cursor == term->search.len || isc32space(term->search.buf[cursor - 1]));
/* Now skip past whitespace, so that we end up at the beginning of
* the next word */
while (cursor < term->search.len) {
if (!isc32space(term->search.buf[cursor++]))
break;
}
xassert(cursor == term->search.len || !isc32space(term->search.buf[cursor - 1]));
if (cursor < term->search.len && !isc32space(term->search.buf[cursor]))
cursor--;
return cursor - term->search.cursor;
}
static size_t
distance_prev_word(const struct terminal *term)
{
int cursor = term->search.cursor;
/* First, eat whitespace prefix */
while (cursor > 0) {
if (!isc32space(term->search.buf[--cursor]))
break;
}
xassert(cursor == 0 || !isc32space(term->search.buf[cursor]));
/* Now eat non-whitespace. This is the word we're skipping past */
while (cursor > 0) {
if (isc32space(term->search.buf[--cursor]))
break;
}
xassert(cursor == 0 || isc32space(term->search.buf[cursor]));
if (cursor > 0 && isc32space(term->search.buf[cursor]))
cursor++;
return term->search.cursor - cursor;
}
static void
from_clipboard_cb(char *text, size_t size, void *user)
{
struct terminal *term = user;
search_add_chars(term, text, size);
}
static void
from_clipboard_done(void *user)
{
struct terminal *term = user;
LOG_DBG("search: buffer: %ls", (const wchar_t *)term->search.buf);
search_find_next(term, SEARCH_BACKWARD_SAME_POSITION);
render_refresh_search(term);
}
static bool
execute_binding(struct seat *seat, struct terminal *term,
const struct key_binding *binding, uint32_t serial,
bool *update_search_result, enum search_direction *direction,
bool *redraw)
{
*update_search_result = *redraw = false;
const enum bind_action_search action = binding->action;
struct grid *grid = term->grid;
switch (action) {
case BIND_ACTION_SEARCH_NONE:
return false;
case BIND_ACTION_SEARCH_CANCEL:
if (term->search.view_followed_offset)
grid->view = grid->offset;
else {
grid->view = ensure_view_is_allocated(
term, term->search.original_view);
}
term_damage_view(term);
search_cancel(term);
return true;
case BIND_ACTION_SEARCH_COMMIT:
selection_finalize(seat, term, serial);
search_cancel_keep_selection(term);
return true;
case BIND_ACTION_SEARCH_FIND_PREV:
if (term->search.last.buf != NULL && term->search.len == 0) {
add_wchars(term, term->search.last.buf, term->search.last.len);
free(term->search.last.buf);
term->search.last.buf = NULL;
term->search.last.len = 0;
}
*direction = SEARCH_BACKWARD;
*update_search_result = *redraw = true;
return true;
case BIND_ACTION_SEARCH_FIND_NEXT:
if (term->search.last.buf != NULL && term->search.len == 0) {
add_wchars(term, term->search.last.buf, term->search.last.len);
free(term->search.last.buf);
term->search.last.buf = NULL;
term->search.last.len = 0;
}
*direction = SEARCH_FORWARD;
*update_search_result = *redraw = true;
return true;
2019-08-30 20:15:36 +02:00
case BIND_ACTION_SEARCH_EDIT_LEFT:
if (term->search.cursor > 0) {
term->search.cursor--;
*redraw = true;
}
return true;
case BIND_ACTION_SEARCH_EDIT_LEFT_WORD: {
size_t diff = distance_prev_word(term);
term->search.cursor -= diff;
xassert(term->search.cursor <= term->search.len);
if (diff > 0)
*redraw = true;
return true;
}
case BIND_ACTION_SEARCH_EDIT_RIGHT:
if (term->search.cursor < term->search.len) {
term->search.cursor++;
*redraw = true;
}
return true;
case BIND_ACTION_SEARCH_EDIT_RIGHT_WORD: {
size_t diff = distance_next_word(term);
term->search.cursor += diff;
xassert(term->search.cursor <= term->search.len);
if (diff > 0)
*redraw = true;
return true;
}
case BIND_ACTION_SEARCH_EDIT_HOME:
if (term->search.cursor != 0) {
term->search.cursor = 0;
*redraw = true;
}
return true;
case BIND_ACTION_SEARCH_EDIT_END:
if (term->search.cursor != term->search.len) {
term->search.cursor = term->search.len;
*redraw = true;
}
return true;
case BIND_ACTION_SEARCH_DELETE_PREV:
if (term->search.cursor > 0) {
memmove(
&term->search.buf[term->search.cursor - 1],
&term->search.buf[term->search.cursor],
(term->search.len - term->search.cursor) * sizeof(char32_t));
term->search.cursor--;
term->search.buf[--term->search.len] = U'\0';
*update_search_result = *redraw = true;
}
return true;
case BIND_ACTION_SEARCH_DELETE_PREV_WORD: {
size_t diff = distance_prev_word(term);
size_t old_cursor = term->search.cursor;
size_t new_cursor = old_cursor - diff;
if (diff > 0) {
memmove(&term->search.buf[new_cursor],
&term->search.buf[old_cursor],
(term->search.len - old_cursor) * sizeof(char32_t));
term->search.len -= diff;
term->search.cursor = new_cursor;
*update_search_result = *redraw = true;
}
return true;
}
case BIND_ACTION_SEARCH_DELETE_NEXT:
if (term->search.cursor < term->search.len) {
memmove(
&term->search.buf[term->search.cursor],
&term->search.buf[term->search.cursor + 1],
(term->search.len - term->search.cursor - 1) * sizeof(char32_t));
term->search.buf[--term->search.len] = U'\0';
*update_search_result = *redraw = true;
}
return true;
case BIND_ACTION_SEARCH_DELETE_NEXT_WORD: {
size_t diff = distance_next_word(term);
size_t cursor = term->search.cursor;
if (diff > 0) {
memmove(&term->search.buf[cursor],
&term->search.buf[cursor + diff],
(term->search.len - (cursor + diff)) * sizeof(char32_t));
term->search.len -= diff;
*update_search_result = *redraw = true;
}
return true;
}
case BIND_ACTION_SEARCH_EXTEND_WORD:
search_match_to_end_of_word(term, false);
*update_search_result = false;
*redraw = true;
return true;
case BIND_ACTION_SEARCH_EXTEND_WORD_WS:
search_match_to_end_of_word(term, true);
*update_search_result = false;
*redraw = true;
return true;
case BIND_ACTION_SEARCH_CLIPBOARD_PASTE:
text_from_clipboard(
seat, term, &from_clipboard_cb, &from_clipboard_done, term);
*update_search_result = *redraw = true;
return true;
case BIND_ACTION_SEARCH_PRIMARY_PASTE:
text_from_primary(
seat, term, &from_clipboard_cb, &from_clipboard_done, term);
*update_search_result = *redraw = true;
return true;
case BIND_ACTION_SEARCH_COUNT:
BUG("Invalid action type");
return true;
}
BUG("Unhandled action type");
return false;
}
void
key-binding: new API, for handling sets of key bindings Up until now, our Wayland seats have been tracking key bindings. This makes sense, since the seat’s keymap determines how the key bindings are resolved. However, tying bindings to the seat/keymap alone isn’t enough, since we also depend on the current configuration (i.e. user settings) when resolving a key binding. This means configurations that doesn’t match the wayland object’s configuration, currently don’t resolve key bindings correctly. This applies to footclients where the user has overridden key bindings on the command line (e.g. --override key-bindings.foo=bar). Thus, to correctly resolve key bindings, each set of key bindings must be tied *both* to a seat/keymap, *and* a configuration. This patch introduces a key-binding manager, with an API to add/remove/lookup, and load/unload keymaps from sets of key bindings. In the API, sets are tied to a seat and terminal instance, since this makes the most sense (we need to instantiate, or incref a set whenever a new terminal instance is created). Internally, the set is tied to a seat and the terminal’s configuration. Sets are *added* when a new seat is added, and when a new terminal instance is created. Since there can only be one instance of each seat, sets are always removed when a seat is removed. Terminals on the other hand can re-use the same configuration (and typically do). Thus, sets ref-count the configuration. In other words, when instantiating a new terminal, we may not have to instantiate a new set of key bindings, but can often be incref:ed instead. Whenever the keymap changes on a seat, all key bindings sets associated with that seat reloads (re-resolves) their key bindings. Closes #931
2022-04-17 15:39:51 +02:00
search_input(struct seat *seat, struct terminal *term,
const struct key_binding_set *bindings, uint32_t key,
input: rewrite of how we match foot’s own key bindings Bindings are matched in one out of three ways: * By translated (by XKB) symbols * By untranslated symbols * By raw key codes A translated symbol is affected by pressed modifiers, some of which can be “consumed”. Consumed modifiers to not partake in the comparison with the binding’s modifiers. In this mode, ctrl+shift+2 maps to ctrl+@ on a US layout. Untranslated symbols, or un-shifted symbols refer to the “base” symbol of the pressed key, i.e. it’s unaffected by modifiers. In this mode, consumed modifiers *do* partake in the comparison with the binding’s modifiers, and ctrl+shift+2 maps to ctrl+shift+2 on a US layout. More examples: ctrl+shift+u maps to ctrl+U in the translated lookup, while ctrl+shift+u maps to ctrl+shift+u in the untranslated lookup. Finally, we also match raw key codes. This allows our bindings to work using the same physical keys when the user switches between latin and non-latin layouts. This means key bindings in foot.ini *must* not include both +shift+ and a *shifted* key. I.e. ctrl+shift+U is not a valid combo as it cannot be triggered. Unfortunately, this was how you were supposed to write bindings up until now... so, we try to detect such bindings, log a deprecation warning and then “fix” the binding for the user. When specifying bindings in foot.ini, both ctrl+U and ctrl+shift+u are valid, and will work. The latter is preferred though, since we cannot detect the raw key code for the former variant. Personally, I also prefer the latter one because it is more explicit; it’s more obvious which keys are involved. However, in some cases it makes more sense to use the other variant. Typically for non-letter combos.
2021-02-27 20:42:31 +01:00
xkb_keysym_t sym, xkb_mod_mask_t mods, xkb_mod_mask_t consumed,
xkb_mod_mask_t locked,
input: rewrite of how we match foot’s own key bindings Bindings are matched in one out of three ways: * By translated (by XKB) symbols * By untranslated symbols * By raw key codes A translated symbol is affected by pressed modifiers, some of which can be “consumed”. Consumed modifiers to not partake in the comparison with the binding’s modifiers. In this mode, ctrl+shift+2 maps to ctrl+@ on a US layout. Untranslated symbols, or un-shifted symbols refer to the “base” symbol of the pressed key, i.e. it’s unaffected by modifiers. In this mode, consumed modifiers *do* partake in the comparison with the binding’s modifiers, and ctrl+shift+2 maps to ctrl+shift+2 on a US layout. More examples: ctrl+shift+u maps to ctrl+U in the translated lookup, while ctrl+shift+u maps to ctrl+shift+u in the untranslated lookup. Finally, we also match raw key codes. This allows our bindings to work using the same physical keys when the user switches between latin and non-latin layouts. This means key bindings in foot.ini *must* not include both +shift+ and a *shifted* key. I.e. ctrl+shift+U is not a valid combo as it cannot be triggered. Unfortunately, this was how you were supposed to write bindings up until now... so, we try to detect such bindings, log a deprecation warning and then “fix” the binding for the user. When specifying bindings in foot.ini, both ctrl+U and ctrl+shift+u are valid, and will work. The latter is preferred though, since we cannot detect the raw key code for the former variant. Personally, I also prefer the latter one because it is more explicit; it’s more obvious which keys are involved. However, in some cases it makes more sense to use the other variant. Typically for non-letter combos.
2021-02-27 20:42:31 +01:00
const xkb_keysym_t *raw_syms, size_t raw_count,
uint32_t serial)
{
2022-04-18 14:57:13 +02:00
LOG_DBG("search: input: sym=%d/0x%x, mods=0x%08x, consumed=0x%08x",
sym, sym, mods, consumed);
const xkb_mod_mask_t bind_mods =
mods & seat->kbd.bind_significant & ~locked;
const xkb_mod_mask_t bind_consumed =
consumed & seat->kbd.bind_significant & ~locked;
enum xkb_compose_status compose_status = seat->kbd.xkb_compose_state != NULL
? xkb_compose_state_get_status(seat->kbd.xkb_compose_state)
: XKB_COMPOSE_NOTHING;
enum search_direction search_direction = SEARCH_BACKWARD_SAME_POSITION;
bool update_search_result = false;
bool redraw = false;
/* Key bindings */
key-binding: new API, for handling sets of key bindings Up until now, our Wayland seats have been tracking key bindings. This makes sense, since the seat’s keymap determines how the key bindings are resolved. However, tying bindings to the seat/keymap alone isn’t enough, since we also depend on the current configuration (i.e. user settings) when resolving a key binding. This means configurations that doesn’t match the wayland object’s configuration, currently don’t resolve key bindings correctly. This applies to footclients where the user has overridden key bindings on the command line (e.g. --override key-bindings.foo=bar). Thus, to correctly resolve key bindings, each set of key bindings must be tied *both* to a seat/keymap, *and* a configuration. This patch introduces a key-binding manager, with an API to add/remove/lookup, and load/unload keymaps from sets of key bindings. In the API, sets are tied to a seat and terminal instance, since this makes the most sense (we need to instantiate, or incref a set whenever a new terminal instance is created). Internally, the set is tied to a seat and the terminal’s configuration. Sets are *added* when a new seat is added, and when a new terminal instance is created. Since there can only be one instance of each seat, sets are always removed when a seat is removed. Terminals on the other hand can re-use the same configuration (and typically do). Thus, sets ref-count the configuration. In other words, when instantiating a new terminal, we may not have to instantiate a new set of key bindings, but can often be incref:ed instead. Whenever the keymap changes on a seat, all key bindings sets associated with that seat reloads (re-resolves) their key bindings. Closes #931
2022-04-17 15:39:51 +02:00
tll_foreach(bindings->search, it) {
input: rewrite of how we match foot’s own key bindings Bindings are matched in one out of three ways: * By translated (by XKB) symbols * By untranslated symbols * By raw key codes A translated symbol is affected by pressed modifiers, some of which can be “consumed”. Consumed modifiers to not partake in the comparison with the binding’s modifiers. In this mode, ctrl+shift+2 maps to ctrl+@ on a US layout. Untranslated symbols, or un-shifted symbols refer to the “base” symbol of the pressed key, i.e. it’s unaffected by modifiers. In this mode, consumed modifiers *do* partake in the comparison with the binding’s modifiers, and ctrl+shift+2 maps to ctrl+shift+2 on a US layout. More examples: ctrl+shift+u maps to ctrl+U in the translated lookup, while ctrl+shift+u maps to ctrl+shift+u in the untranslated lookup. Finally, we also match raw key codes. This allows our bindings to work using the same physical keys when the user switches between latin and non-latin layouts. This means key bindings in foot.ini *must* not include both +shift+ and a *shifted* key. I.e. ctrl+shift+U is not a valid combo as it cannot be triggered. Unfortunately, this was how you were supposed to write bindings up until now... so, we try to detect such bindings, log a deprecation warning and then “fix” the binding for the user. When specifying bindings in foot.ini, both ctrl+U and ctrl+shift+u are valid, and will work. The latter is preferred though, since we cannot detect the raw key code for the former variant. Personally, I also prefer the latter one because it is more explicit; it’s more obvious which keys are involved. However, in some cases it makes more sense to use the other variant. Typically for non-letter combos.
2021-02-27 20:42:31 +01:00
const struct key_binding *bind = &it->item;
/* Match translated symbol */
if (bind->k.sym == sym &&
bind->mods == (bind_mods & ~bind_consumed)) {
if (execute_binding(seat, term, bind, serial,
&update_search_result, &search_direction,
&redraw))
{
goto update_search;
}
return;
}
if (bind->mods != bind_mods || bind_mods != (mods & ~locked))
input: rewrite of how we match foot’s own key bindings Bindings are matched in one out of three ways: * By translated (by XKB) symbols * By untranslated symbols * By raw key codes A translated symbol is affected by pressed modifiers, some of which can be “consumed”. Consumed modifiers to not partake in the comparison with the binding’s modifiers. In this mode, ctrl+shift+2 maps to ctrl+@ on a US layout. Untranslated symbols, or un-shifted symbols refer to the “base” symbol of the pressed key, i.e. it’s unaffected by modifiers. In this mode, consumed modifiers *do* partake in the comparison with the binding’s modifiers, and ctrl+shift+2 maps to ctrl+shift+2 on a US layout. More examples: ctrl+shift+u maps to ctrl+U in the translated lookup, while ctrl+shift+u maps to ctrl+shift+u in the untranslated lookup. Finally, we also match raw key codes. This allows our bindings to work using the same physical keys when the user switches between latin and non-latin layouts. This means key bindings in foot.ini *must* not include both +shift+ and a *shifted* key. I.e. ctrl+shift+U is not a valid combo as it cannot be triggered. Unfortunately, this was how you were supposed to write bindings up until now... so, we try to detect such bindings, log a deprecation warning and then “fix” the binding for the user. When specifying bindings in foot.ini, both ctrl+U and ctrl+shift+u are valid, and will work. The latter is preferred though, since we cannot detect the raw key code for the former variant. Personally, I also prefer the latter one because it is more explicit; it’s more obvious which keys are involved. However, in some cases it makes more sense to use the other variant. Typically for non-letter combos.
2021-02-27 20:42:31 +01:00
continue;
/* Match untranslated symbols */
for (size_t i = 0; i < raw_count; i++) {
if (bind->k.sym == raw_syms[i]) {
if (execute_binding(seat, term, bind, serial,
&update_search_result, &search_direction,
&redraw))
input: rewrite of how we match foot’s own key bindings Bindings are matched in one out of three ways: * By translated (by XKB) symbols * By untranslated symbols * By raw key codes A translated symbol is affected by pressed modifiers, some of which can be “consumed”. Consumed modifiers to not partake in the comparison with the binding’s modifiers. In this mode, ctrl+shift+2 maps to ctrl+@ on a US layout. Untranslated symbols, or un-shifted symbols refer to the “base” symbol of the pressed key, i.e. it’s unaffected by modifiers. In this mode, consumed modifiers *do* partake in the comparison with the binding’s modifiers, and ctrl+shift+2 maps to ctrl+shift+2 on a US layout. More examples: ctrl+shift+u maps to ctrl+U in the translated lookup, while ctrl+shift+u maps to ctrl+shift+u in the untranslated lookup. Finally, we also match raw key codes. This allows our bindings to work using the same physical keys when the user switches between latin and non-latin layouts. This means key bindings in foot.ini *must* not include both +shift+ and a *shifted* key. I.e. ctrl+shift+U is not a valid combo as it cannot be triggered. Unfortunately, this was how you were supposed to write bindings up until now... so, we try to detect such bindings, log a deprecation warning and then “fix” the binding for the user. When specifying bindings in foot.ini, both ctrl+U and ctrl+shift+u are valid, and will work. The latter is preferred though, since we cannot detect the raw key code for the former variant. Personally, I also prefer the latter one because it is more explicit; it’s more obvious which keys are involved. However, in some cases it makes more sense to use the other variant. Typically for non-letter combos.
2021-02-27 20:42:31 +01:00
{
goto update_search;
}
return;
}
}
/* Match raw key code */
tll_foreach(bind->k.key_codes, code) {
if (code->item == key) {
if (execute_binding(seat, term, bind, serial,
&update_search_result, &search_direction,
&redraw))
{
goto update_search;
}
return;
}
}
}
2020-08-23 07:42:20 +02:00
uint8_t buf[64] = {0};
int count = 0;
if (compose_status == XKB_COMPOSE_COMPOSED) {
count = xkb_compose_state_get_utf8(
seat->kbd.xkb_compose_state, (char *)buf, sizeof(buf));
xkb_compose_state_reset(seat->kbd.xkb_compose_state);
} else if (compose_status == XKB_COMPOSE_CANCELLED) {
count = 0;
} else {
count = xkb_state_key_get_utf8(
seat->kbd.xkb_state, key, (char *)buf, sizeof(buf));
}
update_search_result = redraw = count > 0;
search_direction = SEARCH_BACKWARD_SAME_POSITION;
if (count == 0)
return;
search_add_chars(term, (const char *)buf, count);
update_search:
LOG_DBG("search: buffer: %ls", (const wchar_t *)term->search.buf);
if (update_search_result)
search_find_next(term, search_direction);
if (redraw)
render_refresh_search(term);
}