foot/terminal.c

4292 lines
126 KiB
C
Raw Normal View History

2019-06-29 21:03:28 +02:00
#include "terminal.h"
#if defined(__GLIBC__)
#include <malloc.h>
#endif
#include <signal.h>
2019-06-29 21:03:28 +02:00
#include <string.h>
#include <unistd.h>
#include <errno.h>
#include <limits.h>
2019-06-29 21:03:28 +02:00
#include <sys/stat.h>
#include <sys/wait.h>
#include <sys/ioctl.h>
2019-10-28 18:35:16 +01:00
#include <sys/epoll.h>
#include <sys/eventfd.h>
#include <sys/timerfd.h>
#include <fcntl.h>
2019-07-09 09:17:24 +02:00
#include <linux/input-event-codes.h>
#include <xdg-shell.h>
2019-07-09 09:17:24 +02:00
2019-06-29 21:03:28 +02:00
#define LOG_MODULE "terminal"
2021-08-05 18:24:52 +02:00
#define LOG_ENABLE_DBG 0
2019-06-29 21:03:28 +02:00
#include "log.h"
#include "async.h"
#include "commands.h"
2020-02-22 00:23:19 +01:00
#include "config.h"
#include "debug.h"
#include "extract.h"
#include "grid.h"
#include "ime.h"
#include "input.h"
#include "notify.h"
#include "quirks.h"
#include "reaper.h"
#include "render.h"
2019-08-01 20:51:11 +02:00
#include "selection.h"
#include "shm.h"
2020-02-22 00:23:19 +01:00
#include "sixel.h"
#include "slave.h"
#include "spawn.h"
#include "url-mode.h"
#include "util.h"
2020-02-22 00:23:19 +01:00
#include "vt.h"
#include "xmalloc.h"
#include "xsnprintf.h"
2019-06-29 21:03:28 +02:00
#define PTMX_TIMING 0
static void
enqueue_data_for_slave(const void *data, size_t len, size_t offset,
ptmx_buffer_list_t *buffer_list)
{
2020-08-22 09:25:25 +02:00
struct ptmx_buffer queued = {
.data = xmemdup(data, len),
2020-08-22 09:25:25 +02:00
.len = len,
.idx = offset,
};
tll_push_back(*buffer_list, queued);
}
static bool
data_to_slave(struct terminal *term, const void *data, size_t len,
ptmx_buffer_list_t *buffer_list)
{
2019-11-03 12:13:51 +01:00
/*
* Try a synchronous write first. If we fail to write everything,
* switch to asynchronous.
*/
size_t async_idx = 0;
switch (async_write(term->ptmx, data, len, &async_idx)) {
case ASYNC_WRITE_REMAIN:
/* Switch to asynchronous mode; let FDM write the remaining data */
if (!fdm_event_add(term->fdm, term->ptmx, EPOLLOUT))
return false;
enqueue_data_for_slave(data, len, async_idx, buffer_list);
return true;
case ASYNC_WRITE_DONE:
return true;
case ASYNC_WRITE_ERR:
LOG_ERRNO("failed to synchronously write %zu bytes to slave", len);
return false;
}
BUG("Unexpected async_write() return value");
2019-11-03 12:13:51 +01:00
return false;
}
2019-11-03 12:13:51 +01:00
bool
term_paste_data_to_slave(struct terminal *term, const void *data, size_t len)
{
xassert(term->is_sending_paste_data);
if (term->ptmx < 0) {
/* We're probably in "hold" */
return false;
}
if (tll_length(term->ptmx_paste_buffers) > 0) {
/* Don't even try to send data *now* if there's queued up
* data, since that would result in events arriving out of
* order. */
enqueue_data_for_slave(data, len, 0, &term->ptmx_paste_buffers);
return true;
}
return data_to_slave(term, data, len, &term->ptmx_paste_buffers);
}
bool
term_to_slave(struct terminal *term, const void *data, size_t len)
{
if (term->ptmx < 0) {
/* We're probably in "hold" */
return false;
}
if (tll_length(term->ptmx_buffers) > 0 || term->is_sending_paste_data) {
/*
* Don't even try to send data *now* if there's queued up
* data, since that would result in events arriving out of
* order.
*
* Furthermore, if we're currently sending paste data to the
* client, do *not* mix that stream with other events
* (https://codeberg.org/dnkl/foot/issues/101).
*/
enqueue_data_for_slave(data, len, 0, &term->ptmx_buffers);
return true;
}
return data_to_slave(term, data, len, &term->ptmx_buffers);
}
static bool
fdm_ptmx_out(struct fdm *fdm, int fd, int events, void *data)
{
struct terminal *term = data;
/* If there is no queued data, then we shouldn't be in asynchronous mode */
xassert(tll_length(term->ptmx_buffers) > 0 ||
tll_length(term->ptmx_paste_buffers) > 0);
/* Writes a single buffer, returns if not all of it could be written */
#define write_one_buffer(buffer_list) \
{ \
switch (async_write(term->ptmx, it->item.data, it->item.len, &it->item.idx)) { \
case ASYNC_WRITE_DONE: \
free(it->item.data); \
tll_remove(buffer_list, it); \
break; \
case ASYNC_WRITE_REMAIN: \
/* to_slave() updated it->item.idx */ \
return true; \
case ASYNC_WRITE_ERR: \
LOG_ERRNO("failed to asynchronously write %zu bytes to slave", \
it->item.len - it->item.idx); \
return false; \
} \
}
tll_foreach(term->ptmx_paste_buffers, it)
write_one_buffer(term->ptmx_paste_buffers);
/* If we get here, *all* paste data buffers were successfully
* flushed */
if (!term->is_sending_paste_data) {
tll_foreach(term->ptmx_buffers, it)
write_one_buffer(term->ptmx_buffers);
}
/*
* If we get here, *all* buffers were successfully flushed.
*
* Or, we're still sending paste data, in which case we do *not*
* want to send the "normal" queued up data
*
* In both cases, we want to *disable* the FDM callback since
* otherwise we'd just be called right away again, with nothing to
* write.
*/
fdm_event_del(term->fdm, term->ptmx, EPOLLOUT);
return true;
}
static bool
add_utmp_record(const struct config *conf, struct reaper *reaper, int ptmx)
{
#if defined(UTMP_ADD)
if (ptmx < 0)
return true;
if (conf->utmp_helper_path == NULL)
return true;
char *const argv[] = {conf->utmp_helper_path, UTMP_ADD, getenv("WAYLAND_DISPLAY"), NULL};
return spawn(reaper, NULL, argv, ptmx, ptmx, -1, NULL);
#else
return true;
#endif
}
static bool
del_utmp_record(const struct config *conf, struct reaper *reaper, int ptmx)
{
#if defined(UTMP_DEL)
if (ptmx < 0)
return true;
if (conf->utmp_helper_path == NULL)
return true;
char *del_argument =
#if defined(UTMP_DEL_HAVE_ARGUMENT)
getenv("WAYLAND_DISPLAY")
#else
NULL
#endif
;
char *const argv[] = {conf->utmp_helper_path, UTMP_DEL, del_argument, NULL};
return spawn(reaper, NULL, argv, ptmx, ptmx, -1, NULL);
#else
return true;
#endif
}
#if PTMX_TIMING
2020-08-23 07:42:20 +02:00
static struct timespec last = {0};
#endif
static bool cursor_blink_rearm_timer(struct terminal *term);
/* Externally visible, but not declared in terminal.h, to enable pgo
* to call this function directly */
bool
2019-10-28 18:35:16 +01:00
fdm_ptmx(struct fdm *fdm, int fd, int events, void *data)
{
struct terminal *term = data;
const bool pollin = events & EPOLLIN;
const bool pollout = events & EPOLLOUT;
const bool hup = events & EPOLLHUP;
if (pollout) {
if (!fdm_ptmx_out(fdm, fd, events, data))
return false;
}
/* Prevent blinking while typing */
if (term->cursor_blink.fd >= 0) {
term->cursor_blink.state = CURSOR_BLINK_ON;
cursor_blink_rearm_timer(term);
}
if (unlikely(term->interactive_resizing.grid != NULL)) {
/*
* Don't consume PTMX while we're doing an interactive resize,
* since the 'normal' grid we're currently using is a
* temporary one - all changes done to it will be lost when
* the interactive resize ends.
*/
return true;
}
uint8_t buf[24 * 1024];
const size_t max_iterations = !hup ? 10 : SIZE_MAX;
2019-10-28 18:35:16 +01:00
for (size_t i = 0; i < max_iterations && pollin; i++) {
xassert(pollin);
2021-09-05 12:39:25 +02:00
ssize_t count = read(term->ptmx, buf, sizeof(buf));
if (count < 0) {
if (errno == EAGAIN || errno == EIO) {
/*
* EAGAIN: no more to read - FDM will trigger us again
* EIO: assume PTY was closed - we already have, or will get, a EPOLLHUP
*/
break;
}
LOG_ERRNO("failed to read from pseudo terminal");
return false;
2021-04-18 13:56:56 +03:00
} else if (count == 0) {
2021-04-18 14:05:28 +03:00
/* Reached end-of-file */
2021-04-18 13:56:56 +03:00
break;
}
xassert(term->interactive_resizing.grid == NULL);
vt_from_slave(term, buf, count);
}
2019-12-19 07:23:58 +01:00
if (!term->render.app_sync_updates.enabled) {
/*
* We likely need to re-render. But, we don't want to do it
* immediately. Often, a single client update is done through
* multiple writes. This could lead to us rendering one frame with
* "intermediate" state.
*
* For example, we might end up rendering a frame
* where the client just erased a line, while in the
* next frame, the client wrote to the same line. This
* causes screen "flickering".
*
* Mitigate by always incuring a small delay before
* rendering the next frame. This gives the client
* some time to finish the operation (and thus gives
* us time to receive the last writes before doing any
* actual rendering).
*
* We incur this delay *every* time we receive
* input. To ensure we don't delay rendering
* indefinitely, we start a second timer that is only
* reset when we render.
*
* Note that when the client is producing data at a
* very high pace, we're rate limited by the wayland
* compositor anyway. The delay we introduce here only
* has any effect when the renderer is idle.
*/
uint64_t lower_ns = term->conf->tweak.delayed_render_lower_ns;
uint64_t upper_ns = term->conf->tweak.delayed_render_upper_ns;
if (lower_ns > 0 && upper_ns > 0) {
#if PTMX_TIMING
struct timespec now;
clock_gettime(CLOCK_MONOTONIC, &now);
if (last.tv_sec > 0 || last.tv_nsec > 0) {
struct timespec diff;
timespec_sub(&now, &last, &diff);
LOG_INFO("waited %lds %ldns for more input",
(long)diff.tv_sec, diff.tv_nsec);
}
last = now;
#endif
xassert(lower_ns < 1000000000);
xassert(upper_ns < 1000000000);
xassert(upper_ns > lower_ns);
2019-10-28 18:35:16 +01:00
timerfd_settime(
term->delayed_render_timer.lower_fd, 0,
&(struct itimerspec){.it_value = {.tv_nsec = lower_ns}},
2019-10-28 18:35:16 +01:00
NULL);
/* Second timeout - only reset when we render. Set to one
* frame (assuming 60Hz) */
if (!term->delayed_render_timer.is_armed) {
timerfd_settime(
term->delayed_render_timer.upper_fd, 0,
&(struct itimerspec){.it_value = {.tv_nsec = upper_ns}},
NULL);
term->delayed_render_timer.is_armed = true;
}
} else
render_refresh(term);
}
2019-10-28 18:35:16 +01:00
if (hup) {
del_utmp_record(term->conf, term->reaper, term->ptmx);
fdm_del(fdm, fd);
term->ptmx = -1;
/*
* Normally, we do *not* want to shutdown when the PTY is
* closed. Instead, we want to wait for the client application
* to exit.
*
* However, when we're using a pre-existing PTY (the --pty
* option), there _is_ no client application. That is, foot
* does *not* fork+exec anything, and thus the only way to
* shutdown is to wait for the PTY to be closed.
*/
if (term->slave < 0 && !term->conf->hold_at_exit) {
term_shutdown(term);
}
}
return true;
2019-10-28 18:35:16 +01:00
}
bool
term_ptmx_pause(struct terminal *term)
{
return fdm_event_del(term->fdm, term->ptmx, EPOLLIN);
}
bool
term_ptmx_resume(struct terminal *term)
{
return fdm_event_add(term->fdm, term->ptmx, EPOLLIN);
}
2019-10-28 18:35:16 +01:00
static bool
fdm_flash(struct fdm *fdm, int fd, int events, void *data)
{
if (events & EPOLLHUP)
return false;
struct terminal *term = data;
uint64_t expiration_count;
ssize_t ret = read(
term->flash.fd, &expiration_count, sizeof(expiration_count));
if (ret < 0) {
if (errno == EAGAIN)
return true;
2019-10-28 18:35:16 +01:00
LOG_ERRNO("failed to read flash timer");
return false;
}
LOG_DBG("flash timer expired %llu times",
(unsigned long long)expiration_count);
term->flash.active = false;
render_refresh(term);
return true;
}
static bool
fdm_blink(struct fdm *fdm, int fd, int events, void *data)
{
if (events & EPOLLHUP)
return false;
struct terminal *term = data;
uint64_t expiration_count;
ssize_t ret = read(
term->blink.fd, &expiration_count, sizeof(expiration_count));
if (ret < 0) {
if (errno == EAGAIN)
return true;
2019-10-28 18:35:16 +01:00
LOG_ERRNO("failed to read blink timer");
return false;
}
LOG_DBG("blink timer expired %llu times",
(unsigned long long)expiration_count);
/* Invert blink state */
2019-10-28 18:35:16 +01:00
term->blink.state = term->blink.state == BLINK_ON
? BLINK_OFF : BLINK_ON;
/* Scan all visible cells and mark rows with blinking cells dirty */
bool no_blinking_cells = true;
2019-10-28 18:35:16 +01:00
for (int r = 0; r < term->rows; r++) {
struct row *row = grid_row_in_view(term->grid, r);
for (int col = 0; col < term->cols; col++) {
struct cell *cell = &row->cells[col];
if (cell->attrs.blink) {
cell->attrs.clean = 0;
row->dirty = true;
no_blinking_cells = false;
2019-10-28 18:35:16 +01:00
}
}
}
if (no_blinking_cells) {
LOG_DBG("disarming blink timer");
term->blink.state = BLINK_ON;
fdm_del(term->fdm, term->blink.fd);
term->blink.fd = -1;
} else
render_refresh(term);
2019-10-28 18:35:16 +01:00
return true;
}
void
term_arm_blink_timer(struct terminal *term)
{
if (term->blink.fd >= 0)
return;
LOG_DBG("arming blink timer");
int fd = timerfd_create(CLOCK_MONOTONIC, TFD_CLOEXEC | TFD_NONBLOCK);
if (fd < 0) {
LOG_ERRNO("failed to create blink timer FD");
return;
}
if (!fdm_add(term->fdm, fd, EPOLLIN, &fdm_blink, term)) {
close(fd);
return;
}
struct itimerspec alarm = {
.it_value = {.tv_sec = 0, .tv_nsec = 500 * 1000000},
.it_interval = {.tv_sec = 0, .tv_nsec = 500 * 1000000},
};
if (timerfd_settime(fd, 0, &alarm, NULL) < 0) {
LOG_ERRNO("failed to arm blink timer");
fdm_del(term->fdm, fd);
}
term->blink.fd = fd;
}
static void
cursor_refresh(struct terminal *term)
{
term->grid->cur_row->cells[term->grid->cursor.point.col].attrs.clean = 0;
term->grid->cur_row->dirty = true;
render_refresh(term);
}
static bool
fdm_cursor_blink(struct fdm *fdm, int fd, int events, void *data)
{
if (events & EPOLLHUP)
return false;
struct terminal *term = data;
uint64_t expiration_count;
ssize_t ret = read(
term->cursor_blink.fd, &expiration_count, sizeof(expiration_count));
if (ret < 0) {
if (errno == EAGAIN)
return true;
LOG_ERRNO("failed to read cursor blink timer");
return false;
}
LOG_DBG("cursor blink timer expired %llu times",
(unsigned long long)expiration_count);
/* Invert blink state */
term->cursor_blink.state = term->cursor_blink.state == CURSOR_BLINK_ON
? CURSOR_BLINK_OFF : CURSOR_BLINK_ON;
cursor_refresh(term);
return true;
}
2019-10-28 18:35:16 +01:00
static bool
fdm_delayed_render(struct fdm *fdm, int fd, int events, void *data)
{
if (events & EPOLLHUP)
return false;
struct terminal *term = data;
uint64_t unused;
ssize_t ret1 = 0;
ssize_t ret2 = 0;
if (fd == term->delayed_render_timer.lower_fd)
ret1 = read(term->delayed_render_timer.lower_fd, &unused, sizeof(unused));
if (fd == term->delayed_render_timer.upper_fd)
ret2 = read(term->delayed_render_timer.upper_fd, &unused, sizeof(unused));
if ((ret1 < 0 || ret2 < 0)) {
if (errno == EAGAIN)
return true;
2019-10-28 18:35:16 +01:00
LOG_ERRNO("failed to read timeout timer");
return false;
}
if (ret1 > 0)
LOG_DBG("lower delay timer expired");
else if (ret2 > 0)
LOG_DBG("upper delay timer expired");
if (ret1 == 0 && ret2 == 0)
return true;
#if PTMX_TIMING
2020-08-23 07:42:20 +02:00
last = (struct timespec){0};
#endif
/* Reset timers */
2020-08-23 07:42:20 +02:00
struct itimerspec reset = {{0}};
timerfd_settime(term->delayed_render_timer.lower_fd, 0, &reset, NULL);
timerfd_settime(term->delayed_render_timer.upper_fd, 0, &reset, NULL);
term->delayed_render_timer.is_armed = false;
render_refresh(term);
2019-10-28 18:35:16 +01:00
return true;
}
static bool
fdm_app_sync_updates_timeout(
struct fdm *fdm, int fd, int events, void *data)
{
if (events & EPOLLHUP)
return false;
struct terminal *term = data;
uint64_t unused;
ssize_t ret = read(term->render.app_sync_updates.timer_fd,
&unused, sizeof(unused));
if (ret < 0) {
if (errno == EAGAIN)
return true;
LOG_ERRNO("failed to read application synchronized updates timeout timer");
return false;
}
term_disable_app_sync_updates(term);
return true;
}
static bool
fdm_title_update_timeout(struct fdm *fdm, int fd, int events, void *data)
{
if (events & EPOLLHUP)
return false;
struct terminal *term = data;
uint64_t unused;
ssize_t ret = read(term->render.title.timer_fd, &unused, sizeof(unused));
if (ret < 0) {
if (errno == EAGAIN)
return true;
LOG_ERRNO("failed to read title update throttle timer");
return false;
}
struct itimerspec reset = {{0}};
timerfd_settime(term->render.title.timer_fd, 0, &reset, NULL);
render_refresh_title(term);
return true;
}
static bool
fdm_app_id_update_timeout(struct fdm *fdm, int fd, int events, void *data)
{
if (events & EPOLLHUP)
return false;
struct terminal *term = data;
uint64_t unused;
ssize_t ret = read(term->render.app_id.timer_fd, &unused, sizeof(unused));
if (ret < 0) {
if (errno == EAGAIN)
return true;
LOG_ERRNO("failed to read app ID update throttle timer");
return false;
}
struct itimerspec reset = {{0}};
timerfd_settime(term->render.app_id.timer_fd, 0, &reset, NULL);
render_refresh_app_id(term);
return true;
}
static bool
initialize_render_workers(struct terminal *term)
{
LOG_INFO("using %hu rendering threads", term->render.workers.count);
if (sem_init(&term->render.workers.start, 0, 0) < 0 ||
sem_init(&term->render.workers.done, 0, 0) < 0)
{
LOG_ERRNO("failed to instantiate render worker semaphores");
return false;
}
int err;
if ((err = mtx_init(&term->render.workers.lock, mtx_plain)) != thrd_success) {
LOG_ERR("failed to instantiate render worker mutex: %s (%d)",
thrd_err_as_string(err), err);
goto err_sem_destroy;
}
term->render.workers.threads = xcalloc(
term->render.workers.count, sizeof(term->render.workers.threads[0]));
for (size_t i = 0; i < term->render.workers.count; i++) {
struct render_worker_context *ctx = xmalloc(sizeof(*ctx));
*ctx = (struct render_worker_context) {
.term = term,
.my_id = 1 + i,
};
int ret = thrd_create(
&term->render.workers.threads[i], &render_worker_thread, ctx);
2020-05-03 14:17:54 +02:00
if (ret != thrd_success) {
LOG_ERR("failed to create render worker thread: %s (%d)",
thrd_err_as_string(ret), ret);
term->render.workers.threads[i] = 0;
return false;
}
}
return true;
err_sem_destroy:
sem_destroy(&term->render.workers.start);
sem_destroy(&term->render.workers.done);
return false;
}
static void
free_custom_glyph(struct fcft_glyph **glyph)
{
if (*glyph == NULL)
return;
free(pixman_image_get_data((*glyph)->pix));
pixman_image_unref((*glyph)->pix);
free(*glyph);
*glyph = NULL;
}
static void
free_custom_glyphs(struct fcft_glyph ***glyphs, size_t count)
{
if (*glyphs == NULL)
return;
for (size_t i = 0; i < count; i++)
free_custom_glyph(&(*glyphs)[i]);
free(*glyphs);
*glyphs = NULL;
}
static void
term_line_height_update(struct terminal *term)
{
const struct config *conf = term->conf;
if (term->conf->line_height.px < 0) {
term->font_line_height.pt = 0;
term->font_line_height.px = -1;
return;
}
const float dpi = term->font_is_sized_by_dpi ? term->font_dpi : 96.;
const float font_original_pt_size =
conf->fonts[0].arr[0].px_size > 0
? conf->fonts[0].arr[0].px_size * 72. / dpi
: conf->fonts[0].arr[0].pt_size;
const float font_current_pt_size =
term->font_sizes[0][0].px_size > 0
? term->font_sizes[0][0].px_size * 72. / dpi
: term->font_sizes[0][0].pt_size;
const float change = font_current_pt_size / font_original_pt_size;
const float line_original_pt_size = conf->line_height.px > 0
? conf->line_height.px * 72. / dpi
: conf->line_height.pt;
term->font_line_height.px = 0;
term->font_line_height.pt = fmaxf(line_original_pt_size * change, 0.);
}
static bool
terminal: break out scaling factor updating, and reduce number of calls to render_resize() Break out the logic that updates the terminal’s scaling factor value, from render_resize(), to a new function, term_update_scale(). This allows us to update the scaling factor without a full grid resize. We also change how we pick the scaling factor (when fractional scaling is not in use). Before, we’d use the highest scaling factor from all monitors we were mapped on. Now, we use the scaling factor from the monitor we were *last* mapped on. Then, add a boolean parameter to term_set_fonts(), and when false, *don’t* call render_resize_force(). Also change term_font_dpi_changed() to only return true if the font was changed in any way. Finally, rewrite update_term_for_output_change() to: * Call term_update_scale() before doing anything else * Call render_resize{,_force} *last*, and *only* if either the scale or the fonts were updated. This fixes several things: * A bug where we failed to update the fonts when fractional scaling was in use, and we guessed the initial scale/DPI wrong. The bug happened because updated the internal "preferred" scale value, and a later call to render_resize() updated the terminal’s scale value, but since that code path didn’t call term_font_dpi_changed() (and it shouldn’t), the fonts weren’t resized properly. * It ensures we only resize the grid *once* when the scaling factor, or DPI is changed. Before this, we’d resize it twice. And this happened when e.g. dragging the window between monitors.
2023-07-17 16:21:16 +02:00
term_set_fonts(struct terminal *term, struct fcft_font *fonts[static 4],
bool resize_grid)
{
for (size_t i = 0; i < 4; i++) {
xassert(fonts[i] != NULL);
fcft_destroy(term->fonts[i]);
term->fonts[i] = fonts[i];
}
free_custom_glyphs(
&term->custom_glyphs.box_drawing, GLYPH_BOX_DRAWING_COUNT);
free_custom_glyphs(
&term->custom_glyphs.braille, GLYPH_BRAILLE_COUNT);
free_custom_glyphs(
&term->custom_glyphs.legacy, GLYPH_LEGACY_COUNT);
const struct config *conf = term->conf;
const struct fcft_glyph *M = fcft_rasterize_char_utf32(
fonts[0], U'M', term->font_subpixel);
int advance = M != NULL ? M->advance.x : term->fonts[0]->max_advance.x;
term_line_height_update(term);
term->cell_width = advance +
term_pt_or_px_as_pixels(term, &conf->letter_spacing);
term->cell_height = term->font_line_height.px >= 0
? term_pt_or_px_as_pixels(term, &term->font_line_height)
: max(term->fonts[0]->height,
term->fonts[0]->ascent + term->fonts[0]->descent);
if (term->cell_width <= 0)
term->cell_width = 1;
if (term->cell_height <= 0)
term->cell_height = 1;
term->font_x_ofs = term_pt_or_px_as_pixels(term, &conf->horizontal_letter_offset);
term->font_y_ofs = term_pt_or_px_as_pixels(term, &conf->vertical_letter_offset);
term->font_baseline = term_font_baseline(term);
LOG_INFO("cell width=%d, height=%d", term->cell_width, term->cell_height);
sixel_cell_size_changed(term);
terminal: break out scaling factor updating, and reduce number of calls to render_resize() Break out the logic that updates the terminal’s scaling factor value, from render_resize(), to a new function, term_update_scale(). This allows us to update the scaling factor without a full grid resize. We also change how we pick the scaling factor (when fractional scaling is not in use). Before, we’d use the highest scaling factor from all monitors we were mapped on. Now, we use the scaling factor from the monitor we were *last* mapped on. Then, add a boolean parameter to term_set_fonts(), and when false, *don’t* call render_resize_force(). Also change term_font_dpi_changed() to only return true if the font was changed in any way. Finally, rewrite update_term_for_output_change() to: * Call term_update_scale() before doing anything else * Call render_resize{,_force} *last*, and *only* if either the scale or the fonts were updated. This fixes several things: * A bug where we failed to update the fonts when fractional scaling was in use, and we guessed the initial scale/DPI wrong. The bug happened because updated the internal "preferred" scale value, and a later call to render_resize() updated the terminal’s scale value, but since that code path didn’t call term_font_dpi_changed() (and it shouldn’t), the fonts weren’t resized properly. * It ensures we only resize the grid *once* when the scaling factor, or DPI is changed. Before this, we’d resize it twice. And this happened when e.g. dragging the window between monitors.
2023-07-17 16:21:16 +02:00
/* Optimization - some code paths (are forced to) call
* render_resize() after this function */
if (resize_grid) {
/* Use force, since cell-width/height may have changed */
render_resize(
terminal: break out scaling factor updating, and reduce number of calls to render_resize() Break out the logic that updates the terminal’s scaling factor value, from render_resize(), to a new function, term_update_scale(). This allows us to update the scaling factor without a full grid resize. We also change how we pick the scaling factor (when fractional scaling is not in use). Before, we’d use the highest scaling factor from all monitors we were mapped on. Now, we use the scaling factor from the monitor we were *last* mapped on. Then, add a boolean parameter to term_set_fonts(), and when false, *don’t* call render_resize_force(). Also change term_font_dpi_changed() to only return true if the font was changed in any way. Finally, rewrite update_term_for_output_change() to: * Call term_update_scale() before doing anything else * Call render_resize{,_force} *last*, and *only* if either the scale or the fonts were updated. This fixes several things: * A bug where we failed to update the fonts when fractional scaling was in use, and we guessed the initial scale/DPI wrong. The bug happened because updated the internal "preferred" scale value, and a later call to render_resize() updated the terminal’s scale value, but since that code path didn’t call term_font_dpi_changed() (and it shouldn’t), the fonts weren’t resized properly. * It ensures we only resize the grid *once* when the scaling factor, or DPI is changed. Before this, we’d resize it twice. And this happened when e.g. dragging the window between monitors.
2023-07-17 16:21:16 +02:00
term,
(int)roundf(term->width / term->scale),
(int)roundf(term->height / term->scale),
RESIZE_FORCE | RESIZE_KEEP_GRID);
terminal: break out scaling factor updating, and reduce number of calls to render_resize() Break out the logic that updates the terminal’s scaling factor value, from render_resize(), to a new function, term_update_scale(). This allows us to update the scaling factor without a full grid resize. We also change how we pick the scaling factor (when fractional scaling is not in use). Before, we’d use the highest scaling factor from all monitors we were mapped on. Now, we use the scaling factor from the monitor we were *last* mapped on. Then, add a boolean parameter to term_set_fonts(), and when false, *don’t* call render_resize_force(). Also change term_font_dpi_changed() to only return true if the font was changed in any way. Finally, rewrite update_term_for_output_change() to: * Call term_update_scale() before doing anything else * Call render_resize{,_force} *last*, and *only* if either the scale or the fonts were updated. This fixes several things: * A bug where we failed to update the fonts when fractional scaling was in use, and we guessed the initial scale/DPI wrong. The bug happened because updated the internal "preferred" scale value, and a later call to render_resize() updated the terminal’s scale value, but since that code path didn’t call term_font_dpi_changed() (and it shouldn’t), the fonts weren’t resized properly. * It ensures we only resize the grid *once* when the scaling factor, or DPI is changed. Before this, we’d resize it twice. And this happened when e.g. dragging the window between monitors.
2023-07-17 16:21:16 +02:00
}
return true;
}
static float
get_font_dpi(const struct terminal *term)
{
/*
* Use output's DPI to scale font. This is to ensure the font has
* the same physical height (if measured by a ruler) regardless of
* monitor.
*
* Conceptually, we use the physical monitor specs to calculate
* the DPI, and we ignore the output's scaling factor.
*
* However, to deal with legacy fractional scaling, where we're
* told to render at e.g. 2x, but are then downscaled by the
* compositor to e.g. 1.25, we use the scaled DPI value multiplied
* by the scale factor instead.
*
* For integral scaling factors the resulting DPI is the same as
* if we had used the physical DPI.
*
* For legacy fractional scaling factors we'll get a DPI *larger*
* than the physical DPI, that ends up being right when later
* downscaled by the compositor.
*
* With the newer fractional-scale-v1 protocol, we use the
* monitor's real DPI, since we scale everything to the correct
* scaling factor (no downscaling done by the compositor).
*/
xassert(tll_length(term->wl->monitors) > 0);
const struct wl_window *win = term->window;
const struct monitor *mon = NULL;
if (tll_length(win->on_outputs) > 0)
mon = tll_back(win->on_outputs);
else {
if (term->font_dpi_before_unmap > 0.) {
/*
* Use last known "good" DPI
*
* This avoids flickering when window is unmapped/mapped
* (some compositors do this when a window is minimized),
* on a multi-monitor setup with different monitor DPIs.
*/
return term->font_dpi_before_unmap;
}
if (tll_length(term->wl->monitors) > 0)
mon = &tll_front(term->wl->monitors);
}
const float monitor_dpi = mon != NULL
? term_fractional_scaling(term)
? mon->dpi.physical
: mon->dpi.scaled
: 96.;
return monitor_dpi > 0. ? monitor_dpi : 96.;
}
static enum fcft_subpixel
get_font_subpixel(const struct terminal *term)
{
if (term->colors.alpha != 0xffff) {
/* Can't do subpixel rendering on transparent background */
return FCFT_SUBPIXEL_NONE;
}
enum wl_output_subpixel wl_subpixel;
/*
* Wayland doesn't tell us *which* part of the surface that goes
* on a specific output, only whether the surface is mapped to an
* output or not.
*
* Thus, when determining which subpixel mode to use, we can't do
* much but select *an* output. So, we pick the one we were most
* recently mapped on.
*
* If we're not mapped at all, we pick the first available
* monitor, and hope that's where we'll eventually get mapped.
*
* If there aren't any monitors we use the "default" subpixel
* mode.
*/
if (tll_length(term->window->on_outputs) > 0)
wl_subpixel = tll_back(term->window->on_outputs)->subpixel;
else if (tll_length(term->wl->monitors) > 0)
wl_subpixel = tll_front(term->wl->monitors).subpixel;
else
wl_subpixel = WL_OUTPUT_SUBPIXEL_UNKNOWN;
switch (wl_subpixel) {
case WL_OUTPUT_SUBPIXEL_UNKNOWN: return FCFT_SUBPIXEL_DEFAULT;
case WL_OUTPUT_SUBPIXEL_NONE: return FCFT_SUBPIXEL_NONE;
case WL_OUTPUT_SUBPIXEL_HORIZONTAL_RGB: return FCFT_SUBPIXEL_HORIZONTAL_RGB;
case WL_OUTPUT_SUBPIXEL_HORIZONTAL_BGR: return FCFT_SUBPIXEL_HORIZONTAL_BGR;
case WL_OUTPUT_SUBPIXEL_VERTICAL_RGB: return FCFT_SUBPIXEL_VERTICAL_RGB;
case WL_OUTPUT_SUBPIXEL_VERTICAL_BGR: return FCFT_SUBPIXEL_VERTICAL_BGR;
}
return FCFT_SUBPIXEL_DEFAULT;
}
int
term_pt_or_px_as_pixels(const struct terminal *term,
const struct pt_or_px *pt_or_px)
{
float scale = !term->font_is_sized_by_dpi ? term->scale : 1.;
float dpi = term->font_is_sized_by_dpi ? term->font_dpi : 96.;
return pt_or_px->px == 0
? (int)roundf(pt_or_px->pt * scale * dpi / 72)
: (int)roundf(pt_or_px->px * scale);
}
struct font_load_data {
size_t count;
const char **names;
const char *attrs;
struct fcft_font **font;
};
static int
font_loader_thread(void *_data)
{
struct font_load_data *data = _data;
*data->font = fcft_from_name(data->count, data->names, data->attrs);
return *data->font != NULL;
}
static bool
terminal: break out scaling factor updating, and reduce number of calls to render_resize() Break out the logic that updates the terminal’s scaling factor value, from render_resize(), to a new function, term_update_scale(). This allows us to update the scaling factor without a full grid resize. We also change how we pick the scaling factor (when fractional scaling is not in use). Before, we’d use the highest scaling factor from all monitors we were mapped on. Now, we use the scaling factor from the monitor we were *last* mapped on. Then, add a boolean parameter to term_set_fonts(), and when false, *don’t* call render_resize_force(). Also change term_font_dpi_changed() to only return true if the font was changed in any way. Finally, rewrite update_term_for_output_change() to: * Call term_update_scale() before doing anything else * Call render_resize{,_force} *last*, and *only* if either the scale or the fonts were updated. This fixes several things: * A bug where we failed to update the fonts when fractional scaling was in use, and we guessed the initial scale/DPI wrong. The bug happened because updated the internal "preferred" scale value, and a later call to render_resize() updated the terminal’s scale value, but since that code path didn’t call term_font_dpi_changed() (and it shouldn’t), the fonts weren’t resized properly. * It ensures we only resize the grid *once* when the scaling factor, or DPI is changed. Before this, we’d resize it twice. And this happened when e.g. dragging the window between monitors.
2023-07-17 16:21:16 +02:00
reload_fonts(struct terminal *term, bool resize_grid)
{
const struct config *conf = term->conf;
const size_t counts[4] = {
conf->fonts[0].count,
conf->fonts[1].count,
conf->fonts[2].count,
conf->fonts[3].count,
};
/* Configure size (which may have been changed run-time) */
char **names[4];
for (size_t i = 0; i < 4; i++) {
names[i] = xmalloc(counts[i] * sizeof(names[i][0]));
const struct config_font_list *font_list = &conf->fonts[i];
for (size_t j = 0; j < font_list->count; j++) {
const struct config_font *font = &font_list->arr[j];
bool use_px_size = term->font_sizes[i][j].px_size > 0;
char size[64];
const float scale = term->font_is_sized_by_dpi ? 1. : term->scale;
if (use_px_size)
snprintf(size, sizeof(size), ":pixelsize=%d",
(int)roundf(term->font_sizes[i][j].px_size * scale));
else
snprintf(size, sizeof(size), ":size=%.2f",
term->font_sizes[i][j].pt_size * scale);
names[i][j] = xstrjoin(font->pattern, size);
}
}
/* Did user configure custom bold/italic fonts?
* Or should we use the regular font, with weight/slant attributes? */
const bool custom_bold = counts[1] > 0;
const bool custom_italic = counts[2] > 0;
const bool custom_bold_italic = counts[3] > 0;
const size_t count_regular = counts[0];
const char **names_regular = (const char **)names[0];
const size_t count_bold = custom_bold ? counts[1] : counts[0];
const char **names_bold = (const char **)(custom_bold ? names[1] : names[0]);
const size_t count_italic = custom_italic ? counts[2] : counts[0];
const char **names_italic = (const char **)(custom_italic ? names[2] : names[0]);
const size_t count_bold_italic = custom_bold_italic ? counts[3] : counts[0];
const char **names_bold_italic = (const char **)(custom_bold_italic ? names[3] : names[0]);
const bool use_dpi = term->font_is_sized_by_dpi;
char *dpi = xasprintf("dpi=%.2f", use_dpi ? term->font_dpi : 96.);
char *attrs[4] = {
[0] = dpi, /* Takes ownership */
[1] = xstrjoin(dpi, !custom_bold ? ":weight=bold" : ""),
[2] = xstrjoin(dpi, !custom_italic ? ":slant=italic" : ""),
[3] = xstrjoin(dpi, !custom_bold_italic ? ":weight=bold:slant=italic" : ""),
};
struct fcft_font *fonts[4];
struct font_load_data data[4] = {
{count_regular, names_regular, attrs[0], &fonts[0]},
{count_bold, names_bold, attrs[1], &fonts[1]},
{count_italic, names_italic, attrs[2], &fonts[2]},
{count_bold_italic, names_bold_italic, attrs[3], &fonts[3]},
};
2020-08-23 07:42:20 +02:00
thrd_t tids[4] = {0};
for (size_t i = 0; i < 4; i++) {
int ret = thrd_create(&tids[i], &font_loader_thread, &data[i]);
2020-05-03 14:17:54 +02:00
if (ret != thrd_success) {
LOG_ERR("failed to create font loader thread: %s (%d)",
thrd_err_as_string(ret), ret);
break;
}
}
bool success = true;
for (size_t i = 0; i < 4; i++) {
if (tids[i] != 0) {
int ret;
thrd_join(tids[i], &ret);
success = success && ret;
} else
success = false;
}
2020-02-08 17:57:50 +01:00
for (size_t i = 0; i < 4; i++) {
for (size_t j = 0; j < counts[i]; j++)
free(names[i][j]);
free(names[i]);
free(attrs[i]);
}
if (!success) {
LOG_ERR("failed to load primary fonts");
2020-02-08 17:57:50 +01:00
for (size_t i = 0; i < 4; i++) {
fcft_destroy(fonts[i]);
2020-02-08 17:57:50 +01:00
fonts[i] = NULL;
}
}
terminal: break out scaling factor updating, and reduce number of calls to render_resize() Break out the logic that updates the terminal’s scaling factor value, from render_resize(), to a new function, term_update_scale(). This allows us to update the scaling factor without a full grid resize. We also change how we pick the scaling factor (when fractional scaling is not in use). Before, we’d use the highest scaling factor from all monitors we were mapped on. Now, we use the scaling factor from the monitor we were *last* mapped on. Then, add a boolean parameter to term_set_fonts(), and when false, *don’t* call render_resize_force(). Also change term_font_dpi_changed() to only return true if the font was changed in any way. Finally, rewrite update_term_for_output_change() to: * Call term_update_scale() before doing anything else * Call render_resize{,_force} *last*, and *only* if either the scale or the fonts were updated. This fixes several things: * A bug where we failed to update the fonts when fractional scaling was in use, and we guessed the initial scale/DPI wrong. The bug happened because updated the internal "preferred" scale value, and a later call to render_resize() updated the terminal’s scale value, but since that code path didn’t call term_font_dpi_changed() (and it shouldn’t), the fonts weren’t resized properly. * It ensures we only resize the grid *once* when the scaling factor, or DPI is changed. Before this, we’d resize it twice. And this happened when e.g. dragging the window between monitors.
2023-07-17 16:21:16 +02:00
return success ? term_set_fonts(term, fonts, resize_grid) : success;
}
static bool
load_fonts_from_conf(struct terminal *term)
{
const struct config *conf = term->conf;
for (size_t i = 0; i < 4; i++) {
const struct config_font_list *font_list = &conf->fonts[i];
for (size_t j = 0; j < font_list->count; j++) {
const struct config_font *font = &font_list->arr[j];
term->font_sizes[i][j] = (struct config_font){
.pt_size = font->pt_size, .px_size = font->px_size};
}
}
terminal: break out scaling factor updating, and reduce number of calls to render_resize() Break out the logic that updates the terminal’s scaling factor value, from render_resize(), to a new function, term_update_scale(). This allows us to update the scaling factor without a full grid resize. We also change how we pick the scaling factor (when fractional scaling is not in use). Before, we’d use the highest scaling factor from all monitors we were mapped on. Now, we use the scaling factor from the monitor we were *last* mapped on. Then, add a boolean parameter to term_set_fonts(), and when false, *don’t* call render_resize_force(). Also change term_font_dpi_changed() to only return true if the font was changed in any way. Finally, rewrite update_term_for_output_change() to: * Call term_update_scale() before doing anything else * Call render_resize{,_force} *last*, and *only* if either the scale or the fonts were updated. This fixes several things: * A bug where we failed to update the fonts when fractional scaling was in use, and we guessed the initial scale/DPI wrong. The bug happened because updated the internal "preferred" scale value, and a later call to render_resize() updated the terminal’s scale value, but since that code path didn’t call term_font_dpi_changed() (and it shouldn’t), the fonts weren’t resized properly. * It ensures we only resize the grid *once* when the scaling factor, or DPI is changed. Before this, we’d resize it twice. And this happened when e.g. dragging the window between monitors.
2023-07-17 16:21:16 +02:00
return reload_fonts(term, true);
}
static void fdm_client_terminated(
struct reaper *reaper, pid_t pid, int status, void *data);
static const int PTY_OPEN_FLAGS = O_RDWR | O_NOCTTY;
struct terminal *
term_init(const struct config *conf, struct fdm *fdm, struct reaper *reaper,
struct wayland *wayl, const char *foot_exe, const char *cwd,
const char *token, const char *pty_path,
int argc, char *const *argv, const char *const *envp,
void (*shutdown_cb)(void *data, int exit_code), void *shutdown_data)
{
int ptmx = -1;
int flash_fd = -1;
int delay_lower_fd = -1;
int delay_upper_fd = -1;
int app_sync_updates_fd = -1;
int title_update_fd = -1;
int app_id_update_fd = -1;
struct terminal *term = malloc(sizeof(*term));
if (unlikely(term == NULL)) {
LOG_ERRNO("malloc() failed");
return NULL;
}
ptmx = pty_path ? open(pty_path, PTY_OPEN_FLAGS) : posix_openpt(PTY_OPEN_FLAGS);
if (ptmx < 0) {
LOG_ERRNO("failed to open PTY");
goto close_fds;
}
if ((flash_fd = timerfd_create(CLOCK_MONOTONIC, TFD_CLOEXEC | TFD_NONBLOCK)) < 0) {
LOG_ERRNO("failed to create flash timer FD");
goto close_fds;
}
if ((delay_lower_fd = timerfd_create(CLOCK_MONOTONIC, TFD_CLOEXEC | TFD_NONBLOCK)) < 0 ||
(delay_upper_fd = timerfd_create(CLOCK_MONOTONIC, TFD_CLOEXEC | TFD_NONBLOCK)) < 0)
{
LOG_ERRNO("failed to create delayed rendering timer FDs");
goto close_fds;
}
if ((app_sync_updates_fd = timerfd_create(CLOCK_MONOTONIC, TFD_CLOEXEC | TFD_NONBLOCK)) < 0)
{
LOG_ERRNO("failed to create application synchronized updates timer FD");
goto close_fds;
}
if ((title_update_fd = timerfd_create(CLOCK_MONOTONIC, TFD_CLOEXEC | TFD_NONBLOCK)) < 0)
{
LOG_ERRNO("failed to create title update throttle timer FD");
goto close_fds;
}
if ((app_id_update_fd = timerfd_create(CLOCK_MONOTONIC, TFD_CLOEXEC | TFD_NONBLOCK)) < 0)
{
LOG_ERRNO("failed to create app ID update throttle timer FD");
goto close_fds;
}
if (ioctl(ptmx, (unsigned int)TIOCSWINSZ,
&(struct winsize){.ws_row = 24, .ws_col = 80}) < 0)
{
LOG_ERRNO("failed to set initial TIOCSWINSZ");
goto close_fds;
}
/* Need to register *very* early (before the first "goto err"), to
* ensure term_destroy() doesn't unref a key-binding we haven't
* yet ref:d */
key_binding_new_for_conf(wayl->key_binding_manager, wayl, conf);
int ptmx_flags;
if ((ptmx_flags = fcntl(ptmx, F_GETFL)) < 0 ||
fcntl(ptmx, F_SETFL, ptmx_flags | O_NONBLOCK) < 0)
{
LOG_ERRNO("failed to configure ptmx as non-blocking");
goto err;
}
/*
* Enable all FDM callbackes *except* ptmx - we can't do that
* until the window has been 'configured' since we don't have a
* size (and thus no grid) before then.
*/
if (!fdm_add(fdm, flash_fd, EPOLLIN, &fdm_flash, term) ||
2019-11-03 00:25:17 +01:00
!fdm_add(fdm, delay_lower_fd, EPOLLIN, &fdm_delayed_render, term) ||
!fdm_add(fdm, delay_upper_fd, EPOLLIN, &fdm_delayed_render, term) ||
!fdm_add(fdm, app_sync_updates_fd, EPOLLIN, &fdm_app_sync_updates_timeout, term) ||
!fdm_add(fdm, title_update_fd, EPOLLIN, &fdm_title_update_timeout, term) ||
!fdm_add(fdm, app_id_update_fd, EPOLLIN, &fdm_app_id_update_timeout, term))
{
goto err;
}
/* Initialize configure-based terminal attributes */
*term = (struct terminal) {
.fdm = fdm,
.reaper = reaper,
.conf = conf,
.slave = -1,
.ptmx = ptmx,
.ptmx_buffers = tll_init(),
.ptmx_paste_buffers = tll_init(),
.font_sizes = {
xmalloc(sizeof(term->font_sizes[0][0]) * conf->fonts[0].count),
xmalloc(sizeof(term->font_sizes[1][0]) * conf->fonts[1].count),
xmalloc(sizeof(term->font_sizes[2][0]) * conf->fonts[2].count),
xmalloc(sizeof(term->font_sizes[3][0]) * conf->fonts[3].count),
},
.font_dpi = 0.,
.font_dpi_before_unmap = -1.,
.font_subpixel = (conf->colors.alpha == 0xffff /* Can't do subpixel rendering on transparent background */
? FCFT_SUBPIXEL_DEFAULT
: FCFT_SUBPIXEL_NONE),
.cursor_keys_mode = CURSOR_KEYS_NORMAL,
.keypad_keys_mode = KEYPAD_NUMERICAL,
.reverse_wrap = true,
.auto_margin = true,
.window_title_stack = tll_init(),
.scale = 1.,
.scale_before_unmap = -1,
.flash = {.fd = flash_fd},
.blink = {.fd = -1},
.vt = {
2020-01-20 18:35:13 +01:00
.state = 0, /* STATE_GROUND */
},
.colors = {
.fg = conf->colors.fg,
.bg = conf->colors.bg,
.alpha = conf->colors.alpha,
.cursor_fg = conf->cursor.color.text,
.cursor_bg = conf->cursor.color.cursor,
.selection_fg = conf->colors.selection_fg,
.selection_bg = conf->colors.selection_bg,
.use_custom_selection = conf->colors.use_custom.selection,
},
csi: implement XTPUSHCOLORS+XTPOPCOLORS+XTREPORTCOLORS The documentation of these sequences are vague and lacking, as is often the case with XTerm invented control sequences. I've tried to replicate what XTerm does (as of xterm-392). The stack represents *stashed/stored* palettes. The currently active palette is *not* stored on the stack. The stack is dynamically allocated, and starts out with zero elements. Now, XTerm has a somewhat weird definition of "pushing" and "popping" in this context, and the documentation is somewhat misleading. What a push does is this: it stores the current palette to the stack at the specified slot. If the specified slot number (Pm) is 0, the slot used is the current slot index incremented by 1. The "current" slot index is then set to the specified slot (which is current slot + 1 if Pm == 0). Thus, "push" (i.e. when Pm == 0 is used) means store to the "next" slot. This is true even if the current slot index points into the middle of stack. Pop works in a similar way. The palette is restored from the specified slot index. If the specified slot number is 0, we use the current slot index. The "current" slot index is then set to the specified slot - 1 (current slot - 1 if Pm == 0). XTREPORTCOLORS return the current slot index, and the number of palettes stored on the stack, on the format CSI ? <slot index> ; <palette count> # Q When XTPUSHCOLORS grows the stack with more than one element (i.e. via a 'CSI N # P' sequence), make sure *all* new slots are initialized (to the current color palette). This avoids uninitialized slots, that could then be popped with XTPOPCOLORS. Closes #856
2024-07-01 17:40:45 +02:00
.color_stack = {
.stack = NULL,
.size = 0,
.idx = 0,
},
.origin = ORIGIN_ABSOLUTE,
.cursor_style = conf->cursor.style,
.cursor_blink = {
.decset = false,
.deccsusr = conf->cursor.blink.enabled,
.state = CURSOR_BLINK_ON,
.fd = -1,
},
.selection = {
.coords = {
.start = {-1, -1},
.end = {-1, -1},
},
.pivot = {
.start = {-1, -1},
.end = {-1, -1},
},
.auto_scroll = {
.fd = -1,
},
},
2020-08-04 18:07:22 +02:00
.normal = {.scroll_damage = tll_init(), .sixel_images = tll_init()},
.alt = {.scroll_damage = tll_init(), .sixel_images = tll_init()},
.grid = &term->normal,
.composed = NULL,
.alt_scrolling = conf->mouse.alternate_scroll_mode,
.meta = {
.esc_prefix = true,
.eight_bit = true,
},
.num_lock_modifier = true,
.bell_action_enabled = true,
.tab_stops = tll_init(),
.wl = wayl,
.render = {
.chains = {
.grid = shm_chain_new(wayl->shm, true, 1 + conf->render_worker_count),
.search = shm_chain_new(wayl->shm, false, 1),
.scrollback_indicator = shm_chain_new(wayl->shm, false, 1),
.render_timer = shm_chain_new(wayl->shm, false, 1),
.url = shm_chain_new(wayl->shm, false, 1),
.csd = shm_chain_new(wayl->shm, false, 1),
render: implement ‘flash’ and search mode’s ‘dimming’ with a sub-surface Search mode and ‘flash’ (OSC-555) both achieves similar visual effects: flash tints the entire window yellow, and search mode dims it (except the search match). But, they do so in completely different ways. Search mode is detected in render_cell(), and the colors are then dimmed there. Flash is implemented by blending a yellow, semi-transparent color on top of the rendered grid. This patch replaces those two implementations with a single one. We add a new sub-surface, called the ‘overlay’. In normal mode, it’s unmapped. When either search mode, or flash, is enabled, we enable it, and fill it with a semi-transparent color. Yellow for ‘flash’, and “black” (i.e. no color) for search mode. The compositor then blends it with the grid. Hopefully on the GPU, meaning it’ll be faster than if we blend in software. There are more performance benefits however. By using a separate surface, we can do much better damage tracking. The normal grid rendering code no longer have to care about neither search mode, nor flash. Thus, we get rid of a couple of ‘if’ statements in render_cell(), which is nice. But more importantly, we can drop full grid repaints in a couple of circumstances: * Entering/exiting search mode * Every frame while flash is active Now, when rendering the search mode overlay, we do want to do some damage tracking, also of the overlay. This, since search mode doesn’t dim the *entire* window. The search match is *not* dimmed. This is implemented by punching a hole in the overlay sub-surface. That is, we make part of it *fully* transparent. The basic idea is to set a clip region that excludes the search match, and then dim the rest of the overlay. It’s slightly more complicated than that however, if we want to reuse the last frame’s overlay buffer (i.e we don’t want to re-render the *entire* overlay every frame). In short, we need to: * Clear (punch hole) in areas that are part of this frame’s search match, but not the last frame’s (since those parts are _already_ cleared). * Dim the areas that were part of the last frame’s search match, but aren’t anymore (the rest of the overlay should already be dimmed). To do this, we save the last frame’s “holes” (as a pixman region). Then, when rendering the next frame, we first calculate the new frame’s “holes” region. The region to clear is “this frame’s holes minus last frame’s holes” The region to dim is “last frame’s holes minus this frames holes”. Finally, we compute the bounding box of all modified cells by taking the union of the two diff regions mentioned above. This allows us to limit the buffer damage sent to the compositor.
2022-04-16 17:49:46 +02:00
.overlay = shm_chain_new(wayl->shm, false, 1),
},
.scrollback_lines = conf->scrollback.lines,
.app_sync_updates.timer_fd = app_sync_updates_fd,
.title = {
.timer_fd = title_update_fd,
},
.app_id = {
.timer_fd = app_id_update_fd,
},
.workers = {
.count = conf->render_worker_count,
.queue = tll_init(),
},
},
.delayed_render_timer = {
.is_armed = false,
.lower_fd = delay_lower_fd,
.upper_fd = delay_upper_fd,
},
.sixel = {
.scrolling = true,
.use_private_palette = true,
.palette_size = SIXEL_MAX_COLORS,
.max_width = SIXEL_MAX_WIDTH,
.max_height = SIXEL_MAX_HEIGHT,
},
.shutdown = {
.terminate_timeout_fd = -1,
.cb = shutdown_cb,
.cb_data = shutdown_data,
},
.foot_exe = xstrdup(foot_exe),
.cwd = xstrdup(cwd),
.grapheme_shaping = conf->tweak.grapheme_shaping,
#if defined(FOOT_IME_ENABLED) && FOOT_IME_ENABLED
.ime_enabled = true,
#endif
.kitty_notifications = tll_init(),
};
render: implement ‘flash’ and search mode’s ‘dimming’ with a sub-surface Search mode and ‘flash’ (OSC-555) both achieves similar visual effects: flash tints the entire window yellow, and search mode dims it (except the search match). But, they do so in completely different ways. Search mode is detected in render_cell(), and the colors are then dimmed there. Flash is implemented by blending a yellow, semi-transparent color on top of the rendered grid. This patch replaces those two implementations with a single one. We add a new sub-surface, called the ‘overlay’. In normal mode, it’s unmapped. When either search mode, or flash, is enabled, we enable it, and fill it with a semi-transparent color. Yellow for ‘flash’, and “black” (i.e. no color) for search mode. The compositor then blends it with the grid. Hopefully on the GPU, meaning it’ll be faster than if we blend in software. There are more performance benefits however. By using a separate surface, we can do much better damage tracking. The normal grid rendering code no longer have to care about neither search mode, nor flash. Thus, we get rid of a couple of ‘if’ statements in render_cell(), which is nice. But more importantly, we can drop full grid repaints in a couple of circumstances: * Entering/exiting search mode * Every frame while flash is active Now, when rendering the search mode overlay, we do want to do some damage tracking, also of the overlay. This, since search mode doesn’t dim the *entire* window. The search match is *not* dimmed. This is implemented by punching a hole in the overlay sub-surface. That is, we make part of it *fully* transparent. The basic idea is to set a clip region that excludes the search match, and then dim the rest of the overlay. It’s slightly more complicated than that however, if we want to reuse the last frame’s overlay buffer (i.e we don’t want to re-render the *entire* overlay every frame). In short, we need to: * Clear (punch hole) in areas that are part of this frame’s search match, but not the last frame’s (since those parts are _already_ cleared). * Dim the areas that were part of the last frame’s search match, but aren’t anymore (the rest of the overlay should already be dimmed). To do this, we save the last frame’s “holes” (as a pixman region). Then, when rendering the next frame, we first calculate the new frame’s “holes” region. The region to clear is “this frame’s holes minus last frame’s holes” The region to dim is “last frame’s holes minus this frames holes”. Finally, we compute the bounding box of all modified cells by taking the union of the two diff regions mentioned above. This allows us to limit the buffer damage sent to the compositor.
2022-04-16 17:49:46 +02:00
pixman_region32_init(&term->render.last_overlay_clip);
term_update_ascii_printer(term);
for (size_t i = 0; i < 4; i++) {
const struct config_font_list *font_list = &conf->fonts[i];
for (size_t j = 0; j < font_list->count; j++) {
const struct config_font *font = &font_list->arr[j];
term->font_sizes[i][j] = (struct config_font){
.pt_size = font->pt_size, .px_size = font->px_size};
}
}
add_utmp_record(conf, reaper, ptmx);
if (!pty_path) {
/* Start the slave/client */
if ((term->slave = slave_spawn(
term->ptmx, argc, term->cwd, argv, envp, &conf->env_vars,
conf->term, conf->shell, conf->login_shell,
&conf->notifications)) == -1)
{
goto err;
}
reaper_add(term->reaper, term->slave, &fdm_client_terminated, term);
}
/* Guess scale; we're not mapped yet, so we don't know on which
* output we'll be. Use scaling factor from first monitor */
xassert(tll_length(term->wl->monitors) > 0);
term->scale = tll_front(term->wl->monitors).scale;
memcpy(term->colors.table, term->conf->colors.table, sizeof(term->colors.table));
2020-01-03 11:15:35 +01:00
/* Initialize the Wayland window backend */
if ((term->window = wayl_win_init(term, token)) == NULL)
goto err;
/* Load fonts */
if (!term_font_dpi_changed(term, 0.))
goto err;
term->font_subpixel = get_font_subpixel(term);
term_set_window_title(term, conf->title);
2020-03-09 18:46:50 +01:00
/* Let the Wayland backend know we exist */
tll_push_back(wayl->terms, term);
switch (conf->startup_mode) {
case STARTUP_WINDOWED:
break;
case STARTUP_MAXIMIZED:
xdg_toplevel_set_maximized(term->window->xdg_toplevel);
break;
case STARTUP_FULLSCREEN:
xdg_toplevel_set_fullscreen(term->window->xdg_toplevel, NULL);
break;
}
if (!initialize_render_workers(term))
goto err;
return term;
err:
term->shutdown.in_progress = true;
term_destroy(term);
return NULL;
close_fds:
close(ptmx);
2019-11-01 20:29:16 +01:00
fdm_del(fdm, flash_fd);
fdm_del(fdm, delay_lower_fd);
fdm_del(fdm, delay_upper_fd);
fdm_del(fdm, app_sync_updates_fd);
fdm_del(fdm, title_update_fd);
fdm_del(fdm, app_id_update_fd);
2019-11-01 20:29:16 +01:00
free(term);
return NULL;
}
void
term_window_configured(struct terminal *term)
{
/* Enable ptmx FDM callback */
if (!term->shutdown.in_progress) {
xassert(term->window->is_configured);
fdm_add(term->fdm, term->ptmx, EPOLLIN, &fdm_ptmx, term);
}
}
/*
* Shutdown logic
*
* A foot instance can be terminated in two ways:
*
* - the client application terminates (user types 'exit', or pressed C-d in the
* shell, etc)
* - the foot window is closed
*
* Both variants need to trigger to "other" action. I.e. if the client
* application is terminated, then we need to close the window. If the window is
* closed, we need to terminate the client application.
*
* Only when *both* tasks have completed do we consider ourselves fully
* shutdown. This is when we can call term_destroy(), and the user provided
* shutdown callback.
*
* The functions involved with this are:
*
* - shutdown_maybe_done(): called after any of the two tasks above have
* completed. When it determines that *both* tasks are done, it calls
* term_destroy() and the user provided shutdown callback.
*
* - fdm_client_terminated(): reaper callback, called when the client
* application has terminated.
*
* + Kills the "terminate" timeout timer
* + Calls shutdown_maybe_done() if the shutdown procedure has already
* started (i.e. the window being closed initiated the shutdown)
* -OR-
* Initiates the shutdown itself, by calling term_shutdown() (client
* application termination initiated the shutdown).
*
* - term_shutdown(): unregisters all FDM callbacks, sends SIGTERM to the client
* application and installs a "terminate" timeout timer (if it hasn't already
* terminated). Finally registers an event FD with the FDM, which is
* immediately triggered. This is done to ensure any pending FDM events are
* handled before shutting down.
*
* - fdm_shutdown(): FDM callback, triggered by the event FD in
* term_shutdown(). Unmaps and destroys the window resources, and ensures the
* seats' focused pointers don't reference us. Finally calls
* shutdown_maybe_done().
*
* - fdm_terminate_timeout(): FDM callback for the "terminate" timeout
* timer. This function is called when the client application hasn't
* terminated after 60 seconds (after the SIGTERM). Sends SIGKILL to the
* client application.
*
* - term_destroy(): normally called from shutdown_maybe_done(), when both the
* window has been unmapped, and the client application has terminated. In
* this case, it simply destroys all resources.
*
* It may however also be called without term_shutdown() having been called
* (typically in error code paths - for example, when the Wayland connection
* is closed by the compositor). In this case, the client application is
* typically still running, and we can't assume the FDM is running. To handle
* this, we install configure a 60 second SIGALRM, send SIGTERM to the client
* application, and then enter a blocking waitpid().
*
* If the alarm triggers, we send SIGKILL and once again enter a blocking
* waitpid().
*/
static void
shutdown_maybe_done(struct terminal *term)
{
bool shutdown_done =
term->window == NULL && term->shutdown.client_has_terminated;
LOG_DBG("window=%p, slave-has-been-reaped=%d --> %s",
(void *)term->window, term->shutdown.client_has_terminated,
(shutdown_done
? "shutdown done, calling term_destroy()"
: "no action"));
if (!shutdown_done)
return;
void (*cb)(void *, int) = term->shutdown.cb;
void *cb_data = term->shutdown.cb_data;
int exit_code = term_destroy(term);
if (cb != NULL)
cb(cb_data, exit_code);
}
static void
fdm_client_terminated(struct reaper *reaper, pid_t pid, int status, void *data)
{
struct terminal *term = data;
LOG_DBG("slave (PID=%u) died", pid);
term->shutdown.client_has_terminated = true;
term->shutdown.exit_status = status;
if (term->shutdown.terminate_timeout_fd >= 0) {
fdm_del(term->fdm, term->shutdown.terminate_timeout_fd);
term->shutdown.terminate_timeout_fd = -1;
}
if (term->shutdown.in_progress)
shutdown_maybe_done(term);
else if (!term->conf->hold_at_exit)
term_shutdown(term);
}
static bool
fdm_shutdown(struct fdm *fdm, int fd, int events, void *data)
{
struct terminal *term = data;
/* Kill the event FD */
fdm_del(term->fdm, fd);
wayl_win_destroy(term->window);
term->window = NULL;
struct wayland *wayl = term->wl;
/*
* Normally we'd get unmapped when we destroy the Wayland
* above.
*
* However, it appears that under certain conditions, those events
* are deferred (for example, when a screen locker is active), and
* thus we can get here without having been unmapped.
*/
tll_foreach(wayl->seats, it) {
if (it->item.kbd_focus == term)
it->item.kbd_focus = NULL;
if (it->item.mouse_focus == term)
it->item.mouse_focus = NULL;
}
shutdown_maybe_done(term);
return true;
}
static bool
fdm_terminate_timeout(struct fdm *fdm, int fd, int events, void *data)
{
uint64_t unused;
ssize_t bytes = read(fd, &unused, sizeof(unused));
if (bytes < 0) {
LOG_ERRNO("failed to read from slave terminate timeout FD");
return false;
}
struct terminal *term = data;
xassert(!term->shutdown.client_has_terminated);
LOG_DBG("slave (PID=%u) has not terminated, sending %s (%d)",
term->slave,
term->shutdown.next_signal == SIGTERM ? "SIGTERM"
: term->shutdown.next_signal == SIGKILL ? "SIGKILL"
: "<unknown>",
term->shutdown.next_signal);
kill(-term->slave, term->shutdown.next_signal);
switch (term->shutdown.next_signal) {
case SIGTERM:
term->shutdown.next_signal = SIGKILL;
break;
case SIGKILL:
/* Disarm. Shouldn't be necessary, as we should be able to
shutdown completely after sending SIGKILL, before the next
timeout occurs). But lets play it safe... */
if (term->shutdown.terminate_timeout_fd >= 0) {
timerfd_settime(
term->shutdown.terminate_timeout_fd, 0,
&(const struct itimerspec){0}, NULL);
}
break;
default:
BUG("can only handle SIGTERM and SIGKILL");
return false;
}
return true;
}
bool
term_shutdown(struct terminal *term)
{
if (term->shutdown.in_progress)
return true;
term->shutdown.in_progress = true;
/*
* Close FDs then postpone self-destruction to the next poll
* iteration, by creating an event FD that we trigger immediately.
*/
term_cursor_blink_update(term);
xassert(term->cursor_blink.fd < 0);
fdm_del(term->fdm, term->selection.auto_scroll.fd);
fdm_del(term->fdm, term->render.app_sync_updates.timer_fd);
fdm_del(term->fdm, term->render.app_id.timer_fd);
fdm_del(term->fdm, term->render.title.timer_fd);
fdm_del(term->fdm, term->delayed_render_timer.lower_fd);
fdm_del(term->fdm, term->delayed_render_timer.upper_fd);
fdm_del(term->fdm, term->blink.fd);
fdm_del(term->fdm, term->flash.fd);
del_utmp_record(term->conf, term->reaper, term->ptmx);
if (term->window != NULL && term->window->is_configured)
fdm_del(term->fdm, term->ptmx);
else
close(term->ptmx);
if (!term->shutdown.client_has_terminated) {
if (term->slave <= 0) {
term->shutdown.client_has_terminated = true;
} else {
LOG_DBG("initiating asynchronous terminate of slave; "
"sending SIGHUP to PID=%u", term->slave);
kill(-term->slave, SIGHUP);
/*
* Set up a timer, with an interval - on the first timeout
* we'll send SIGTERM. If the the client application still
* isn't terminating, we'll wait an additional interval,
* and then send SIGKILL.
*/
const struct itimerspec timeout = {.it_value = {.tv_sec = 30},
.it_interval = {.tv_sec = 30}};
int timeout_fd = timerfd_create(CLOCK_MONOTONIC, TFD_CLOEXEC | TFD_NONBLOCK);
if (timeout_fd < 0 ||
timerfd_settime(timeout_fd, 0, &timeout, NULL) < 0 ||
!fdm_add(term->fdm, timeout_fd, EPOLLIN, &fdm_terminate_timeout, term))
{
if (timeout_fd >= 0)
close(timeout_fd);
LOG_ERRNO("failed to create slave terminate timeout FD");
return false;
}
xassert(term->shutdown.terminate_timeout_fd < 0);
term->shutdown.terminate_timeout_fd = timeout_fd;
term->shutdown.next_signal = SIGTERM;
}
}
term->selection.auto_scroll.fd = -1;
term->render.app_sync_updates.timer_fd = -1;
term->render.app_id.timer_fd = -1;
term->render.title.timer_fd = -1;
term->delayed_render_timer.lower_fd = -1;
term->delayed_render_timer.upper_fd = -1;
term->blink.fd = -1;
term->flash.fd = -1;
term->ptmx = -1;
int event_fd = eventfd(0, EFD_CLOEXEC | EFD_NONBLOCK);
if (event_fd == -1) {
LOG_ERRNO("failed to create terminal shutdown event FD");
return false;
}
if (!fdm_add(term->fdm, event_fd, EPOLLIN, &fdm_shutdown, term)) {
close(event_fd);
return false;
2019-10-28 18:35:16 +01:00
}
if (write(event_fd, &(uint64_t){1}, sizeof(uint64_t)) != sizeof(uint64_t)) {
LOG_ERRNO("failed to send terminal shutdown event");
fdm_del(term->fdm, event_fd);
return false;
2019-10-28 18:35:16 +01:00
}
return true;
}
static volatile sig_atomic_t alarm_raised;
static void
sig_alarm(int signo)
{
LOG_DBG("SIGALRM");
alarm_raised = 1;
}
int
term_destroy(struct terminal *term)
{
if (term == NULL)
return 0;
tll_foreach(term->wl->terms, it) {
if (it->item == term) {
tll_remove(term->wl->terms, it);
break;
}
}
del_utmp_record(term->conf, term->reaper, term->ptmx);
fdm_del(term->fdm, term->selection.auto_scroll.fd);
fdm_del(term->fdm, term->render.app_sync_updates.timer_fd);
fdm_del(term->fdm, term->render.app_id.timer_fd);
fdm_del(term->fdm, term->render.title.timer_fd);
fdm_del(term->fdm, term->delayed_render_timer.lower_fd);
fdm_del(term->fdm, term->delayed_render_timer.upper_fd);
fdm_del(term->fdm, term->cursor_blink.fd);
fdm_del(term->fdm, term->blink.fd);
fdm_del(term->fdm, term->flash.fd);
fdm_del(term->fdm, term->ptmx);
if (term->shutdown.terminate_timeout_fd >= 0)
fdm_del(term->fdm, term->shutdown.terminate_timeout_fd);
if (term->window != NULL) {
wayl_win_destroy(term->window);
term->window = NULL;
}
mtx_lock(&term->render.workers.lock);
xassert(tll_length(term->render.workers.queue) == 0);
/* Count livinig threads - we may get here when only some of the
* threads have been successfully started */
size_t worker_count = 0;
if (term->render.workers.threads != NULL) {
for (size_t i = 0; i < term->render.workers.count; i++, worker_count++) {
if (term->render.workers.threads[i] == 0)
break;
}
for (size_t i = 0; i < worker_count; i++) {
sem_post(&term->render.workers.start);
tll_push_back(term->render.workers.queue, -2);
}
}
mtx_unlock(&term->render.workers.lock);
key_binding_unref(term->wl->key_binding_manager, term->conf);
urls_reset(term);
free(term->vt.osc.data);
free(term->vt.osc8.uri);
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_free(term->composed);
free(term->app_id);
free(term->window_title);
tll_free_and_free(term->window_title_stack, free);
for (size_t i = 0; i < sizeof(term->fonts) / sizeof(term->fonts[0]); i++)
fcft_destroy(term->fonts[i]);
for (size_t i = 0; i < 4; i++)
free(term->font_sizes[i]);
free_custom_glyphs(
&term->custom_glyphs.box_drawing, GLYPH_BOX_DRAWING_COUNT);
free_custom_glyphs(
&term->custom_glyphs.braille, GLYPH_BRAILLE_COUNT);
free_custom_glyphs(
&term->custom_glyphs.legacy, GLYPH_LEGACY_COUNT);
free(term->search.buf);
free(term->search.last.buf);
if (term->render.workers.threads != NULL) {
for (size_t i = 0; i < term->render.workers.count; i++) {
if (term->render.workers.threads[i] != 0)
thrd_join(term->render.workers.threads[i], NULL);
}
}
free(term->render.workers.threads);
mtx_destroy(&term->render.workers.lock);
sem_destroy(&term->render.workers.start);
sem_destroy(&term->render.workers.done);
xassert(tll_length(term->render.workers.queue) == 0);
tll_free(term->render.workers.queue);
shm_unref(term->render.last_buf);
shm_chain_free(term->render.chains.grid);
shm_chain_free(term->render.chains.search);
shm_chain_free(term->render.chains.scrollback_indicator);
shm_chain_free(term->render.chains.render_timer);
shm_chain_free(term->render.chains.url);
shm_chain_free(term->render.chains.csd);
render: implement ‘flash’ and search mode’s ‘dimming’ with a sub-surface Search mode and ‘flash’ (OSC-555) both achieves similar visual effects: flash tints the entire window yellow, and search mode dims it (except the search match). But, they do so in completely different ways. Search mode is detected in render_cell(), and the colors are then dimmed there. Flash is implemented by blending a yellow, semi-transparent color on top of the rendered grid. This patch replaces those two implementations with a single one. We add a new sub-surface, called the ‘overlay’. In normal mode, it’s unmapped. When either search mode, or flash, is enabled, we enable it, and fill it with a semi-transparent color. Yellow for ‘flash’, and “black” (i.e. no color) for search mode. The compositor then blends it with the grid. Hopefully on the GPU, meaning it’ll be faster than if we blend in software. There are more performance benefits however. By using a separate surface, we can do much better damage tracking. The normal grid rendering code no longer have to care about neither search mode, nor flash. Thus, we get rid of a couple of ‘if’ statements in render_cell(), which is nice. But more importantly, we can drop full grid repaints in a couple of circumstances: * Entering/exiting search mode * Every frame while flash is active Now, when rendering the search mode overlay, we do want to do some damage tracking, also of the overlay. This, since search mode doesn’t dim the *entire* window. The search match is *not* dimmed. This is implemented by punching a hole in the overlay sub-surface. That is, we make part of it *fully* transparent. The basic idea is to set a clip region that excludes the search match, and then dim the rest of the overlay. It’s slightly more complicated than that however, if we want to reuse the last frame’s overlay buffer (i.e we don’t want to re-render the *entire* overlay every frame). In short, we need to: * Clear (punch hole) in areas that are part of this frame’s search match, but not the last frame’s (since those parts are _already_ cleared). * Dim the areas that were part of the last frame’s search match, but aren’t anymore (the rest of the overlay should already be dimmed). To do this, we save the last frame’s “holes” (as a pixman region). Then, when rendering the next frame, we first calculate the new frame’s “holes” region. The region to clear is “this frame’s holes minus last frame’s holes” The region to dim is “last frame’s holes minus this frames holes”. Finally, we compute the bounding box of all modified cells by taking the union of the two diff regions mentioned above. This allows us to limit the buffer damage sent to the compositor.
2022-04-16 17:49:46 +02:00
shm_chain_free(term->render.chains.overlay);
pixman_region32_fini(&term->render.last_overlay_clip);
tll_free(term->tab_stops);
tll_foreach(term->ptmx_buffers, it) {
free(it->item.data);
tll_remove(term->ptmx_buffers, it);
}
tll_foreach(term->ptmx_paste_buffers, it) {
free(it->item.data);
tll_remove(term->ptmx_paste_buffers, it);
}
tll_foreach(term->kitty_notifications, it) {
free(it->item.id);
free(it->item.body);
free(it->item.title);
tll_remove(term->kitty_notifications, it);
}
sixel_fini(term);
term_ime_reset(term);
grid_free(&term->normal);
grid_free(&term->alt);
grid_free(term->interactive_resizing.grid);
free(term->interactive_resizing.grid);
free(term->foot_exe);
free(term->cwd);
free(term->mouse_user_cursor);
csi: implement XTPUSHCOLORS+XTPOPCOLORS+XTREPORTCOLORS The documentation of these sequences are vague and lacking, as is often the case with XTerm invented control sequences. I've tried to replicate what XTerm does (as of xterm-392). The stack represents *stashed/stored* palettes. The currently active palette is *not* stored on the stack. The stack is dynamically allocated, and starts out with zero elements. Now, XTerm has a somewhat weird definition of "pushing" and "popping" in this context, and the documentation is somewhat misleading. What a push does is this: it stores the current palette to the stack at the specified slot. If the specified slot number (Pm) is 0, the slot used is the current slot index incremented by 1. The "current" slot index is then set to the specified slot (which is current slot + 1 if Pm == 0). Thus, "push" (i.e. when Pm == 0 is used) means store to the "next" slot. This is true even if the current slot index points into the middle of stack. Pop works in a similar way. The palette is restored from the specified slot index. If the specified slot number is 0, we use the current slot index. The "current" slot index is then set to the specified slot - 1 (current slot - 1 if Pm == 0). XTREPORTCOLORS return the current slot index, and the number of palettes stored on the stack, on the format CSI ? <slot index> ; <palette count> # Q When XTPUSHCOLORS grows the stack with more than one element (i.e. via a 'CSI N # P' sequence), make sure *all* new slots are initialized (to the current color palette). This avoids uninitialized slots, that could then be popped with XTPOPCOLORS. Closes #856
2024-07-01 17:40:45 +02:00
free(term->color_stack.stack);
int ret = EXIT_SUCCESS;
if (term->slave > 0) {
/* We'll deal with this explicitly */
reaper_del(term->reaper, term->slave);
int exit_status;
if (term->shutdown.client_has_terminated)
exit_status = term->shutdown.exit_status;
else {
LOG_DBG("initiating blocking terminate of slave; "
"sending SIGHUP to PID=%u", term->slave);
kill(-term->slave, SIGHUP);
/*
* we've closed the ptxm, and sent SIGTERM to the client
* application. It *should* exit...
*
* But, since it is possible to write clients that ignore
* this, we need to handle it in *some* way.
*
* So, what we do is register a SIGALRM handler, and configure a 30
* second alarm. If the slave hasn't died after this time, we send
* it a SIGKILL,
*
* Note that this solution is *not* asynchronous, and any
* other events etc will be ignored during this time. This of
* course only applies to a 'foot --server' instance, where
* there might be other terminals running.
*/
struct sigaction action = {.sa_handler = &sig_alarm};
sigemptyset(&action.sa_mask);
sigaction(SIGALRM, &action, NULL);
/* Wait, then send SIGTERM, wait again, then send SIGKILL */
int next_signal = SIGTERM;
alarm_raised = 0;
alarm(30);
while (true) {
int r = waitpid(term->slave, &exit_status, 0);
if (r == term->slave)
break;
if (r == -1) {
xassert(errno == EINTR);
if (alarm_raised) {
LOG_DBG("slave (PID=%u) has not terminated yet, "
"sending: %s (%d)", term->slave,
next_signal == SIGTERM ? "SIGTERM" : "SIGKILL",
next_signal);
kill(-term->slave, next_signal);
next_signal = SIGKILL;
alarm_raised = 0;
alarm(30);
}
}
}
/* Cancel alarm */
alarm(0);
action.sa_handler = SIG_DFL;
sigaction(SIGALRM, &action, NULL);
}
ret = EXIT_FAILURE;
if (WIFEXITED(exit_status)) {
ret = WEXITSTATUS(exit_status);
LOG_DBG("slave exited with code %d", ret);
} else if (WIFSIGNALED(exit_status)) {
ret = WTERMSIG(exit_status);
LOG_WARN("slave exited with signal %d (%s)", ret, strsignal(ret));
} else {
LOG_WARN("slave exited for unknown reason (status = 0x%08x)",
exit_status);
}
}
free(term);
#if defined(__GLIBC__)
if (!malloc_trim(0))
LOG_WARN("failed to trim memory");
#endif
return ret;
}
static inline void
erase_cell_range(struct terminal *term, struct row *row, int start, int end)
{
xassert(start < term->cols);
xassert(end < term->cols);
row->dirty = true;
const enum color_source bg_src = term->vt.attrs.bg_src;
if (unlikely(bg_src != COLOR_DEFAULT)) {
for (int col = start; col <= end; col++) {
struct cell *c = &row->cells[col];
c->wc = 0;
c->attrs = (struct attributes){.bg_src = bg_src, .bg = term->vt.attrs.bg};
}
} else
memset(&row->cells[start], 0, (end - start + 1) * sizeof(row->cells[0]));
if (unlikely(row->extra != NULL)) {
grid_row_uri_range_erase(row, start, end);
grid_row_underline_range_erase(row, start, end);
}
}
static inline void
erase_line(struct terminal *term, struct row *row)
{
erase_cell_range(term, row, 0, term->cols - 1);
row->linebreak = false;
row->shell_integration.prompt_marker = false;
row->shell_integration.cmd_start = -1;
row->shell_integration.cmd_end = -1;
}
2019-08-01 20:51:11 +02:00
void
term_reset(struct terminal *term, bool hard)
{
2023-10-04 08:23:27 +02:00
LOG_INFO("%s resetting the terminal", hard ? "hard" : "soft");
2019-08-01 20:51:11 +02:00
term->cursor_keys_mode = CURSOR_KEYS_NORMAL;
term->keypad_keys_mode = KEYPAD_NUMERICAL;
term->reverse = false;
term->hide_cursor = false;
term->reverse_wrap = true;
2019-08-01 20:51:11 +02:00
term->auto_margin = true;
term->insert_mode = false;
term->bracketed_paste = false;
term->focus_events = false;
term->num_lock_modifier = true;
term->bell_action_enabled = true;
2019-08-01 20:51:11 +02:00
term->mouse_tracking = MOUSE_NONE;
term->mouse_reporting = MOUSE_NORMAL;
term->charsets.selected = G0;
term->charsets.set[G0] = CHARSET_ASCII;
term->charsets.set[G1] = CHARSET_ASCII;
term->charsets.set[G2] = CHARSET_ASCII;
term->charsets.set[G3] = CHARSET_ASCII;
term->saved_charsets = term->charsets;
2019-08-01 20:51:11 +02:00
tll_free_and_free(term->window_title_stack, free);
term_set_window_title(term, term->conf->title);
2019-08-01 20:51:11 +02:00
term_set_user_mouse_cursor(term, NULL);
term->modify_other_keys_2 = false;
memset(term->normal.kitty_kbd.flags, 0, sizeof(term->normal.kitty_kbd.flags));
memset(term->alt.kitty_kbd.flags, 0, sizeof(term->alt.kitty_kbd.flags));
term->normal.kitty_kbd.idx = term->alt.kitty_kbd.idx = 0;
2019-08-01 20:51:11 +02:00
term->scroll_region.start = 0;
term->scroll_region.end = term->rows;
free(term->vt.osc8.uri);
2019-08-01 20:51:11 +02:00
free(term->vt.osc.data);
term->vt = (struct vt){
.state = 0, /* STATE_GROUND */
};
2019-08-01 20:51:11 +02:00
if (term->grid == &term->alt) {
term->grid = &term->normal;
selection_cancel(term);
}
term->meta.esc_prefix = true;
term->meta.eight_bit = true;
tll_foreach(term->normal.sixel_images, it) {
sixel_destroy(&it->item);
tll_remove(term->normal.sixel_images, it);
}
tll_foreach(term->alt.sixel_images, it) {
2020-02-22 10:47:16 +01:00
sixel_destroy(&it->item);
tll_remove(term->alt.sixel_images, it);
}
2020-02-22 10:47:16 +01:00
tll_foreach(term->kitty_notifications, it) {
free(it->item.id);
free(it->item.title);
free(it->item.body);
tll_remove(term->kitty_notifications, it);
}
term->grapheme_shaping = term->conf->tweak.grapheme_shaping;
#if defined(FOOT_IME_ENABLED) && FOOT_IME_ENABLED
term_ime_enable(term);
#endif
term->bits_affecting_ascii_printer.value = 0;
term_update_ascii_printer(term);
2019-08-01 20:51:11 +02:00
if (!hard)
return;
term->flash.active = false;
term->blink.state = BLINK_ON;
2020-10-13 19:28:42 +02:00
fdm_del(term->fdm, term->blink.fd); term->blink.fd = -1;
term->colors.fg = term->conf->colors.fg;
term->colors.bg = term->conf->colors.bg;
term->colors.alpha = term->conf->colors.alpha;
term->colors.cursor_fg = term->conf->cursor.color.text;
term->colors.cursor_bg = term->conf->cursor.color.cursor;
term->colors.selection_fg = term->conf->colors.selection_fg;
term->colors.selection_bg = term->conf->colors.selection_bg;
term->colors.use_custom_selection = term->conf->colors.use_custom.selection;
memcpy(term->colors.table, term->conf->colors.table,
sizeof(term->colors.table));
csi: implement XTPUSHCOLORS+XTPOPCOLORS+XTREPORTCOLORS The documentation of these sequences are vague and lacking, as is often the case with XTerm invented control sequences. I've tried to replicate what XTerm does (as of xterm-392). The stack represents *stashed/stored* palettes. The currently active palette is *not* stored on the stack. The stack is dynamically allocated, and starts out with zero elements. Now, XTerm has a somewhat weird definition of "pushing" and "popping" in this context, and the documentation is somewhat misleading. What a push does is this: it stores the current palette to the stack at the specified slot. If the specified slot number (Pm) is 0, the slot used is the current slot index incremented by 1. The "current" slot index is then set to the specified slot (which is current slot + 1 if Pm == 0). Thus, "push" (i.e. when Pm == 0 is used) means store to the "next" slot. This is true even if the current slot index points into the middle of stack. Pop works in a similar way. The palette is restored from the specified slot index. If the specified slot number is 0, we use the current slot index. The "current" slot index is then set to the specified slot - 1 (current slot - 1 if Pm == 0). XTREPORTCOLORS return the current slot index, and the number of palettes stored on the stack, on the format CSI ? <slot index> ; <palette count> # Q When XTPUSHCOLORS grows the stack with more than one element (i.e. via a 'CSI N # P' sequence), make sure *all* new slots are initialized (to the current color palette). This avoids uninitialized slots, that could then be popped with XTPOPCOLORS. Closes #856
2024-07-01 17:40:45 +02:00
free(term->color_stack.stack);
term->color_stack.stack = NULL;
term->color_stack.size = 0;
term->color_stack.idx = 0;
term->origin = ORIGIN_ABSOLUTE;
term->normal.cursor.lcf = false;
term->alt.cursor.lcf = false;
term->normal.cursor = (struct cursor){.point = {0, 0}};
term->normal.saved_cursor = (struct cursor){.point = {0, 0}};
term->alt.cursor = (struct cursor){.point = {0, 0}};
term->alt.saved_cursor = (struct cursor){.point = {0, 0}};
term->cursor_style = term->conf->cursor.style;
term->cursor_blink.decset = false;
term->cursor_blink.deccsusr = term->conf->cursor.blink.enabled;
term_cursor_blink_update(term);
2019-08-01 20:51:11 +02:00
selection_cancel(term);
term->normal.offset = term->normal.view = 0;
term->alt.offset = term->alt.view = 0;
for (size_t i = 0; i < term->rows; i++) {
struct row *r = grid_row_and_alloc(&term->normal, i);
erase_line(term, r);
}
for (size_t i = 0; i < term->rows; i++) {
struct row *r = grid_row_and_alloc(&term->alt, i);
erase_line(term, r);
2019-08-01 20:51:11 +02:00
}
for (size_t i = term->rows; i < term->normal.num_rows; i++) {
grid_row_free(term->normal.rows[i]);
term->normal.rows[i] = NULL;
}
for (size_t i = term->rows; i < term->alt.num_rows; i++) {
grid_row_free(term->alt.rows[i]);
term->alt.rows[i] = NULL;
}
term->normal.cur_row = term->normal.rows[0];
term->alt.cur_row = term->alt.rows[0];
tll_free(term->normal.scroll_damage);
tll_free(term->alt.scroll_damage);
term->render.last_cursor.row = NULL;
2019-08-01 20:51:11 +02:00
term_damage_all(term);
term->sixel.scrolling = true;
term->sixel.cursor_right_of_graphics = false;
term->sixel.use_private_palette = true;
term->sixel.max_width = SIXEL_MAX_WIDTH;
term->sixel.max_height = SIXEL_MAX_HEIGHT;
term->sixel.palette_size = SIXEL_MAX_COLORS;
free(term->sixel.private_palette);
free(term->sixel.shared_palette);
term->sixel.private_palette = term->sixel.shared_palette = NULL;
2019-08-01 20:51:11 +02:00
}
static bool
term_font_size_adjust_by_points(struct terminal *term, float amount)
{
const struct config *conf = term->conf;
const float dpi = term->font_is_sized_by_dpi ? term->font_dpi : 96.;
for (size_t i = 0; i < 4; i++) {
const struct config_font_list *font_list = &conf->fonts[i];
for (size_t j = 0; j < font_list->count; j++) {
struct config_font *font = &term->font_sizes[i][j];
float old_pt_size = font->pt_size;
if (font->px_size > 0)
old_pt_size = font->px_size * 72. / dpi;
font->pt_size = fmaxf(old_pt_size + amount, 0.);
font->px_size = -1;
}
}
terminal: break out scaling factor updating, and reduce number of calls to render_resize() Break out the logic that updates the terminal’s scaling factor value, from render_resize(), to a new function, term_update_scale(). This allows us to update the scaling factor without a full grid resize. We also change how we pick the scaling factor (when fractional scaling is not in use). Before, we’d use the highest scaling factor from all monitors we were mapped on. Now, we use the scaling factor from the monitor we were *last* mapped on. Then, add a boolean parameter to term_set_fonts(), and when false, *don’t* call render_resize_force(). Also change term_font_dpi_changed() to only return true if the font was changed in any way. Finally, rewrite update_term_for_output_change() to: * Call term_update_scale() before doing anything else * Call render_resize{,_force} *last*, and *only* if either the scale or the fonts were updated. This fixes several things: * A bug where we failed to update the fonts when fractional scaling was in use, and we guessed the initial scale/DPI wrong. The bug happened because updated the internal "preferred" scale value, and a later call to render_resize() updated the terminal’s scale value, but since that code path didn’t call term_font_dpi_changed() (and it shouldn’t), the fonts weren’t resized properly. * It ensures we only resize the grid *once* when the scaling factor, or DPI is changed. Before this, we’d resize it twice. And this happened when e.g. dragging the window between monitors.
2023-07-17 16:21:16 +02:00
return reload_fonts(term, true);
}
static bool
term_font_size_adjust_by_pixels(struct terminal *term, int amount)
{
const struct config *conf = term->conf;
const float dpi = term->font_is_sized_by_dpi ? term->font_dpi : 96.;
for (size_t i = 0; i < 4; i++) {
const struct config_font_list *font_list = &conf->fonts[i];
for (size_t j = 0; j < font_list->count; j++) {
struct config_font *font = &term->font_sizes[i][j];
int old_px_size = font->px_size;
if (font->px_size <= 0)
old_px_size = font->pt_size * dpi / 72.;
font->px_size = max(old_px_size + amount, 1);
}
}
terminal: break out scaling factor updating, and reduce number of calls to render_resize() Break out the logic that updates the terminal’s scaling factor value, from render_resize(), to a new function, term_update_scale(). This allows us to update the scaling factor without a full grid resize. We also change how we pick the scaling factor (when fractional scaling is not in use). Before, we’d use the highest scaling factor from all monitors we were mapped on. Now, we use the scaling factor from the monitor we were *last* mapped on. Then, add a boolean parameter to term_set_fonts(), and when false, *don’t* call render_resize_force(). Also change term_font_dpi_changed() to only return true if the font was changed in any way. Finally, rewrite update_term_for_output_change() to: * Call term_update_scale() before doing anything else * Call render_resize{,_force} *last*, and *only* if either the scale or the fonts were updated. This fixes several things: * A bug where we failed to update the fonts when fractional scaling was in use, and we guessed the initial scale/DPI wrong. The bug happened because updated the internal "preferred" scale value, and a later call to render_resize() updated the terminal’s scale value, but since that code path didn’t call term_font_dpi_changed() (and it shouldn’t), the fonts weren’t resized properly. * It ensures we only resize the grid *once* when the scaling factor, or DPI is changed. Before this, we’d resize it twice. And this happened when e.g. dragging the window between monitors.
2023-07-17 16:21:16 +02:00
return reload_fonts(term, true);
}
static bool
term_font_size_adjust_by_percent(struct terminal *term, bool increment, float percent)
{
const struct config *conf = term->conf;
const float multiplier = increment
? 1. + percent
: 1. / (1. + percent);
for (size_t i = 0; i < 4; i++) {
const struct config_font_list *font_list = &conf->fonts[i];
for (size_t j = 0; j < font_list->count; j++) {
struct config_font *font = &term->font_sizes[i][j];
if (font->px_size > 0)
font->px_size = max(font->px_size * multiplier, 1);
else
font->pt_size = fmax(font->pt_size * multiplier, 0);
}
}
terminal: break out scaling factor updating, and reduce number of calls to render_resize() Break out the logic that updates the terminal’s scaling factor value, from render_resize(), to a new function, term_update_scale(). This allows us to update the scaling factor without a full grid resize. We also change how we pick the scaling factor (when fractional scaling is not in use). Before, we’d use the highest scaling factor from all monitors we were mapped on. Now, we use the scaling factor from the monitor we were *last* mapped on. Then, add a boolean parameter to term_set_fonts(), and when false, *don’t* call render_resize_force(). Also change term_font_dpi_changed() to only return true if the font was changed in any way. Finally, rewrite update_term_for_output_change() to: * Call term_update_scale() before doing anything else * Call render_resize{,_force} *last*, and *only* if either the scale or the fonts were updated. This fixes several things: * A bug where we failed to update the fonts when fractional scaling was in use, and we guessed the initial scale/DPI wrong. The bug happened because updated the internal "preferred" scale value, and a later call to render_resize() updated the terminal’s scale value, but since that code path didn’t call term_font_dpi_changed() (and it shouldn’t), the fonts weren’t resized properly. * It ensures we only resize the grid *once* when the scaling factor, or DPI is changed. Before this, we’d resize it twice. And this happened when e.g. dragging the window between monitors.
2023-07-17 16:21:16 +02:00
return reload_fonts(term, true);
}
bool
term_font_size_increase(struct terminal *term)
{
const struct config *conf = term->conf;
const struct font_size_adjustment *inc_dec = &conf->font_size_adjustment;
if (inc_dec->percent > 0.)
return term_font_size_adjust_by_percent(term, true, inc_dec->percent);
else if (inc_dec->pt_or_px.px > 0)
return term_font_size_adjust_by_pixels(term, inc_dec->pt_or_px.px);
else
return term_font_size_adjust_by_points(term, inc_dec->pt_or_px.pt);
}
bool
term_font_size_decrease(struct terminal *term)
{
const struct config *conf = term->conf;
const struct font_size_adjustment *inc_dec = &conf->font_size_adjustment;
if (inc_dec->percent > 0.)
return term_font_size_adjust_by_percent(term, false, inc_dec->percent);
else if (inc_dec->pt_or_px.px > 0)
return term_font_size_adjust_by_pixels(term, -inc_dec->pt_or_px.px);
else
return term_font_size_adjust_by_points(term, -inc_dec->pt_or_px.pt);
}
bool
2020-02-08 17:57:50 +01:00
term_font_size_reset(struct terminal *term)
{
return load_fonts_from_conf(term);
}
bool
term_fractional_scaling(const struct terminal *term)
{
return term->wl->fractional_scale_manager != NULL &&
term->wl->viewporter != NULL &&
term->window->scale > 0.;
}
bool
term_preferred_buffer_scale(const struct terminal *term)
{
return term->window->preferred_buffer_scale > 0;
}
terminal: break out scaling factor updating, and reduce number of calls to render_resize() Break out the logic that updates the terminal’s scaling factor value, from render_resize(), to a new function, term_update_scale(). This allows us to update the scaling factor without a full grid resize. We also change how we pick the scaling factor (when fractional scaling is not in use). Before, we’d use the highest scaling factor from all monitors we were mapped on. Now, we use the scaling factor from the monitor we were *last* mapped on. Then, add a boolean parameter to term_set_fonts(), and when false, *don’t* call render_resize_force(). Also change term_font_dpi_changed() to only return true if the font was changed in any way. Finally, rewrite update_term_for_output_change() to: * Call term_update_scale() before doing anything else * Call render_resize{,_force} *last*, and *only* if either the scale or the fonts were updated. This fixes several things: * A bug where we failed to update the fonts when fractional scaling was in use, and we guessed the initial scale/DPI wrong. The bug happened because updated the internal "preferred" scale value, and a later call to render_resize() updated the terminal’s scale value, but since that code path didn’t call term_font_dpi_changed() (and it shouldn’t), the fonts weren’t resized properly. * It ensures we only resize the grid *once* when the scaling factor, or DPI is changed. Before this, we’d resize it twice. And this happened when e.g. dragging the window between monitors.
2023-07-17 16:21:16 +02:00
bool
term_update_scale(struct terminal *term)
{
const struct wl_window *win = term->window;
/*
* We have a number of "sources" we can use as scale. We choose
terminal: break out scaling factor updating, and reduce number of calls to render_resize() Break out the logic that updates the terminal’s scaling factor value, from render_resize(), to a new function, term_update_scale(). This allows us to update the scaling factor without a full grid resize. We also change how we pick the scaling factor (when fractional scaling is not in use). Before, we’d use the highest scaling factor from all monitors we were mapped on. Now, we use the scaling factor from the monitor we were *last* mapped on. Then, add a boolean parameter to term_set_fonts(), and when false, *don’t* call render_resize_force(). Also change term_font_dpi_changed() to only return true if the font was changed in any way. Finally, rewrite update_term_for_output_change() to: * Call term_update_scale() before doing anything else * Call render_resize{,_force} *last*, and *only* if either the scale or the fonts were updated. This fixes several things: * A bug where we failed to update the fonts when fractional scaling was in use, and we guessed the initial scale/DPI wrong. The bug happened because updated the internal "preferred" scale value, and a later call to render_resize() updated the terminal’s scale value, but since that code path didn’t call term_font_dpi_changed() (and it shouldn’t), the fonts weren’t resized properly. * It ensures we only resize the grid *once* when the scaling factor, or DPI is changed. Before this, we’d resize it twice. And this happened when e.g. dragging the window between monitors.
2023-07-17 16:21:16 +02:00
* the scale in the following order:
*
* - "preferred" scale, from the fractional-scale-v1 protocol
* - "preferred" scale, from wl_compositor version 6.
NOTE: if the compositor advertises version 6 we must use 1.0
until wl_surface.preferred_buffer_scale is sent
terminal: break out scaling factor updating, and reduce number of calls to render_resize() Break out the logic that updates the terminal’s scaling factor value, from render_resize(), to a new function, term_update_scale(). This allows us to update the scaling factor without a full grid resize. We also change how we pick the scaling factor (when fractional scaling is not in use). Before, we’d use the highest scaling factor from all monitors we were mapped on. Now, we use the scaling factor from the monitor we were *last* mapped on. Then, add a boolean parameter to term_set_fonts(), and when false, *don’t* call render_resize_force(). Also change term_font_dpi_changed() to only return true if the font was changed in any way. Finally, rewrite update_term_for_output_change() to: * Call term_update_scale() before doing anything else * Call render_resize{,_force} *last*, and *only* if either the scale or the fonts were updated. This fixes several things: * A bug where we failed to update the fonts when fractional scaling was in use, and we guessed the initial scale/DPI wrong. The bug happened because updated the internal "preferred" scale value, and a later call to render_resize() updated the terminal’s scale value, but since that code path didn’t call term_font_dpi_changed() (and it shouldn’t), the fonts weren’t resized properly. * It ensures we only resize the grid *once* when the scaling factor, or DPI is changed. Before this, we’d resize it twice. And this happened when e.g. dragging the window between monitors.
2023-07-17 16:21:16 +02:00
* - scaling factor of output we most recently were mapped on
* - if we're not mapped, use the last known scaling factor
* - if we're not mapped, and we don't have a last known scaling
* factor, use the scaling factor from the first available
* output.
* - if there aren't any outputs available, use 1.0
terminal: break out scaling factor updating, and reduce number of calls to render_resize() Break out the logic that updates the terminal’s scaling factor value, from render_resize(), to a new function, term_update_scale(). This allows us to update the scaling factor without a full grid resize. We also change how we pick the scaling factor (when fractional scaling is not in use). Before, we’d use the highest scaling factor from all monitors we were mapped on. Now, we use the scaling factor from the monitor we were *last* mapped on. Then, add a boolean parameter to term_set_fonts(), and when false, *don’t* call render_resize_force(). Also change term_font_dpi_changed() to only return true if the font was changed in any way. Finally, rewrite update_term_for_output_change() to: * Call term_update_scale() before doing anything else * Call render_resize{,_force} *last*, and *only* if either the scale or the fonts were updated. This fixes several things: * A bug where we failed to update the fonts when fractional scaling was in use, and we guessed the initial scale/DPI wrong. The bug happened because updated the internal "preferred" scale value, and a later call to render_resize() updated the terminal’s scale value, but since that code path didn’t call term_font_dpi_changed() (and it shouldn’t), the fonts weren’t resized properly. * It ensures we only resize the grid *once* when the scaling factor, or DPI is changed. Before this, we’d resize it twice. And this happened when e.g. dragging the window between monitors.
2023-07-17 16:21:16 +02:00
*/
const float new_scale = (term_fractional_scaling(term)
? win->scale
: term_preferred_buffer_scale(term)
? win->preferred_buffer_scale
: tll_length(win->on_outputs) > 0
? tll_back(win->on_outputs)->scale
: term->scale_before_unmap > 0.
? term->scale_before_unmap
: tll_length(term->wl->monitors) > 0
? tll_front(term->wl->monitors).scale
: 1.);
terminal: break out scaling factor updating, and reduce number of calls to render_resize() Break out the logic that updates the terminal’s scaling factor value, from render_resize(), to a new function, term_update_scale(). This allows us to update the scaling factor without a full grid resize. We also change how we pick the scaling factor (when fractional scaling is not in use). Before, we’d use the highest scaling factor from all monitors we were mapped on. Now, we use the scaling factor from the monitor we were *last* mapped on. Then, add a boolean parameter to term_set_fonts(), and when false, *don’t* call render_resize_force(). Also change term_font_dpi_changed() to only return true if the font was changed in any way. Finally, rewrite update_term_for_output_change() to: * Call term_update_scale() before doing anything else * Call render_resize{,_force} *last*, and *only* if either the scale or the fonts were updated. This fixes several things: * A bug where we failed to update the fonts when fractional scaling was in use, and we guessed the initial scale/DPI wrong. The bug happened because updated the internal "preferred" scale value, and a later call to render_resize() updated the terminal’s scale value, but since that code path didn’t call term_font_dpi_changed() (and it shouldn’t), the fonts weren’t resized properly. * It ensures we only resize the grid *once* when the scaling factor, or DPI is changed. Before this, we’d resize it twice. And this happened when e.g. dragging the window between monitors.
2023-07-17 16:21:16 +02:00
if (new_scale == term->scale)
return false;
LOG_DBG("scaling factor changed: %.2f -> %.2f", term->scale, new_scale);
term->scale_before_unmap = new_scale;
terminal: break out scaling factor updating, and reduce number of calls to render_resize() Break out the logic that updates the terminal’s scaling factor value, from render_resize(), to a new function, term_update_scale(). This allows us to update the scaling factor without a full grid resize. We also change how we pick the scaling factor (when fractional scaling is not in use). Before, we’d use the highest scaling factor from all monitors we were mapped on. Now, we use the scaling factor from the monitor we were *last* mapped on. Then, add a boolean parameter to term_set_fonts(), and when false, *don’t* call render_resize_force(). Also change term_font_dpi_changed() to only return true if the font was changed in any way. Finally, rewrite update_term_for_output_change() to: * Call term_update_scale() before doing anything else * Call render_resize{,_force} *last*, and *only* if either the scale or the fonts were updated. This fixes several things: * A bug where we failed to update the fonts when fractional scaling was in use, and we guessed the initial scale/DPI wrong. The bug happened because updated the internal "preferred" scale value, and a later call to render_resize() updated the terminal’s scale value, but since that code path didn’t call term_font_dpi_changed() (and it shouldn’t), the fonts weren’t resized properly. * It ensures we only resize the grid *once* when the scaling factor, or DPI is changed. Before this, we’d resize it twice. And this happened when e.g. dragging the window between monitors.
2023-07-17 16:21:16 +02:00
term->scale = new_scale;
return true;
}
bool
term_font_dpi_changed(struct terminal *term, float old_scale)
{
float dpi = get_font_dpi(term);
xassert(term->scale > 0.);
bool was_scaled_using_dpi = term->font_is_sized_by_dpi;
bool will_scale_using_dpi = term->conf->dpi_aware;
bool need_font_reload =
was_scaled_using_dpi != will_scale_using_dpi ||
(will_scale_using_dpi
? term->font_dpi != dpi
: old_scale != term->scale);
if (need_font_reload) {
LOG_DBG("DPI/scale change: DPI-aware=%s, "
"DPI: %.2f -> %.2f, scale: %.2f -> %.2f, "
"sizing font based on monitor's %s",
term->conf->dpi_aware ? "yes" : "no",
term->font_dpi, dpi, old_scale, term->scale,
will_scale_using_dpi ? "DPI" : "scaling factor");
}
term->font_dpi = dpi;
term->font_dpi_before_unmap = dpi;
term->font_is_sized_by_dpi = will_scale_using_dpi;
if (!need_font_reload)
terminal: break out scaling factor updating, and reduce number of calls to render_resize() Break out the logic that updates the terminal’s scaling factor value, from render_resize(), to a new function, term_update_scale(). This allows us to update the scaling factor without a full grid resize. We also change how we pick the scaling factor (when fractional scaling is not in use). Before, we’d use the highest scaling factor from all monitors we were mapped on. Now, we use the scaling factor from the monitor we were *last* mapped on. Then, add a boolean parameter to term_set_fonts(), and when false, *don’t* call render_resize_force(). Also change term_font_dpi_changed() to only return true if the font was changed in any way. Finally, rewrite update_term_for_output_change() to: * Call term_update_scale() before doing anything else * Call render_resize{,_force} *last*, and *only* if either the scale or the fonts were updated. This fixes several things: * A bug where we failed to update the fonts when fractional scaling was in use, and we guessed the initial scale/DPI wrong. The bug happened because updated the internal "preferred" scale value, and a later call to render_resize() updated the terminal’s scale value, but since that code path didn’t call term_font_dpi_changed() (and it shouldn’t), the fonts weren’t resized properly. * It ensures we only resize the grid *once* when the scaling factor, or DPI is changed. Before this, we’d resize it twice. And this happened when e.g. dragging the window between monitors.
2023-07-17 16:21:16 +02:00
return false;
terminal: break out scaling factor updating, and reduce number of calls to render_resize() Break out the logic that updates the terminal’s scaling factor value, from render_resize(), to a new function, term_update_scale(). This allows us to update the scaling factor without a full grid resize. We also change how we pick the scaling factor (when fractional scaling is not in use). Before, we’d use the highest scaling factor from all monitors we were mapped on. Now, we use the scaling factor from the monitor we were *last* mapped on. Then, add a boolean parameter to term_set_fonts(), and when false, *don’t* call render_resize_force(). Also change term_font_dpi_changed() to only return true if the font was changed in any way. Finally, rewrite update_term_for_output_change() to: * Call term_update_scale() before doing anything else * Call render_resize{,_force} *last*, and *only* if either the scale or the fonts were updated. This fixes several things: * A bug where we failed to update the fonts when fractional scaling was in use, and we guessed the initial scale/DPI wrong. The bug happened because updated the internal "preferred" scale value, and a later call to render_resize() updated the terminal’s scale value, but since that code path didn’t call term_font_dpi_changed() (and it shouldn’t), the fonts weren’t resized properly. * It ensures we only resize the grid *once* when the scaling factor, or DPI is changed. Before this, we’d resize it twice. And this happened when e.g. dragging the window between monitors.
2023-07-17 16:21:16 +02:00
return reload_fonts(term, false);
2020-02-08 17:57:50 +01:00
}
void
term_font_subpixel_changed(struct terminal *term)
{
enum fcft_subpixel subpixel = get_font_subpixel(term);
if (term->font_subpixel == subpixel)
return;
#if defined(_DEBUG) && LOG_ENABLE_DBG
static const char *const str[] = {
[FCFT_SUBPIXEL_DEFAULT] = "default",
[FCFT_SUBPIXEL_NONE] = "disabled",
[FCFT_SUBPIXEL_HORIZONTAL_RGB] = "RGB",
[FCFT_SUBPIXEL_HORIZONTAL_BGR] = "BGR",
[FCFT_SUBPIXEL_VERTICAL_RGB] = "V-RGB",
[FCFT_SUBPIXEL_VERTICAL_BGR] = "V-BGR",
};
LOG_DBG("subpixel mode changed: %s -> %s", str[term->font_subpixel], str[subpixel]);
#endif
term->font_subpixel = subpixel;
term_damage_view(term);
render_refresh(term);
}
int
term_font_baseline(const struct terminal *term)
{
const struct fcft_font *font = term->fonts[0];
const int line_height = term->cell_height;
const int font_height = font->ascent + font->descent;
/*
* Center glyph on the line *if* using a custom line height,
* otherwise the baseline is simply 'descent' pixels above the
* bottom of the cell
*/
const int glyph_top_y = term->font_line_height.px >= 0
? round((line_height - font_height) / 2.)
: 0;
return term->font_y_ofs + line_height - glyph_top_y - font->descent;
}
2019-07-11 09:51:51 +02:00
void
term_damage_rows(struct terminal *term, int start, int end)
{
xassert(start <= end);
for (int r = start; r <= end; r++) {
struct row *row = grid_row(term->grid, r);
row->dirty = true;
for (int c = 0; c < term->grid->num_cols; c++)
row->cells[c].attrs.clean = 0;
}
2019-07-11 09:51:51 +02:00
}
void
term_damage_rows_in_view(struct terminal *term, int start, int end)
{
xassert(start <= end);
for (int r = start; r <= end; r++) {
struct row *row = grid_row_in_view(term->grid, r);
row->dirty = true;
for (int c = 0; c < term->grid->num_cols; c++)
row->cells[c].attrs.clean = 0;
}
2019-07-11 09:51:51 +02:00
}
2019-06-29 21:03:28 +02:00
void
term_damage_all(struct terminal *term)
{
term_damage_rows(term, 0, term->rows - 1);
2019-06-29 21:03:28 +02:00
}
void
term_damage_view(struct terminal *term)
{
term_damage_rows_in_view(term, 0, term->rows - 1);
}
void
term_damage_cursor(struct terminal *term)
{
term->grid->cur_row->cells[term->grid->cursor.point.col].attrs.clean = 0;
term->grid->cur_row->dirty = true;
}
void
term_damage_margins(struct terminal *term)
{
term->render.margins = true;
}
void
term_damage_color(struct terminal *term, enum color_source src, int idx)
{
xassert(src == COLOR_DEFAULT || src == COLOR_BASE256);
for (int r = 0; r < term->rows; r++) {
struct row *row = grid_row_in_view(term->grid, r);
struct cell *cell = &row->cells[0];
const struct cell *end = &row->cells[term->cols];
for (; cell < end; cell++) {
bool dirty = false;
switch (cell->attrs.fg_src) {
case COLOR_BASE16:
case COLOR_BASE256:
if (src == COLOR_BASE256 && cell->attrs.fg == idx)
dirty = true;
break;
case COLOR_DEFAULT:
if (src == COLOR_DEFAULT) {
/* Doesn't matter whether we've updated the
default foreground, or background, we still
want to dirty this cell, to be sure we handle
all cases of color inversion/reversal */
dirty = true;
}
break;
case COLOR_RGB:
/* Not affected */
break;
}
switch (cell->attrs.bg_src) {
case COLOR_BASE16:
case COLOR_BASE256:
if (src == COLOR_BASE256 && cell->attrs.bg == idx)
dirty = true;
break;
case COLOR_DEFAULT:
if (src == COLOR_DEFAULT) {
/* Doesn't matter whether we've updated the
default foreground, or background, we still
want to dirty this cell, to be sure we handle
all cases of color inversion/reversal */
dirty = true;
}
break;
case COLOR_RGB:
/* Not affected */
break;
}
if (dirty) {
cell->attrs.clean = 0;
row->dirty = true;
}
}
/* Colored underlines */
if (row->extra != NULL) {
const struct row_ranges *underlines = &row->extra->underline_ranges;
for (int i = 0; i < underlines->count; i++) {
const struct row_range *range = &underlines->v[i];
/* Underline colors are either default, or
BASE256/RGB, but never BASE16 */
xassert(range->underline.color_src == COLOR_DEFAULT ||
range->underline.color_src == COLOR_BASE256 ||
range->underline.color_src == COLOR_RGB);
if (range->underline.color_src == src) {
struct cell *c = &row->cells[range->start];
const struct cell *e = &row->cells[range->end + 1];
for (; c < e; c++)
c->attrs.clean = 0;
row->dirty = true;
}
}
}
}
}
2019-06-29 21:03:28 +02:00
void
term_damage_scroll(struct terminal *term, enum damage_type damage_type,
struct scroll_region region, int lines)
{
if (likely(tll_length(term->grid->scroll_damage) > 0)) {
2019-06-29 21:23:36 +02:00
struct damage *dmg = &tll_back(term->grid->scroll_damage);
2019-06-29 21:03:28 +02:00
if (likely(
dmg->type == damage_type &&
dmg->region.start == region.start &&
dmg->region.end == region.end))
2019-06-29 21:03:28 +02:00
{
/* Make sure we don't overflow... */
int new_line_count = (int)dmg->lines + lines;
if (likely(new_line_count <= UINT16_MAX)) {
dmg->lines = new_line_count;
return;
}
2019-06-29 21:03:28 +02:00
}
}
struct damage dmg = {
.type = damage_type,
.region = region,
.lines = lines,
2019-06-29 21:03:28 +02:00
};
2019-06-29 21:23:36 +02:00
tll_push_back(term->grid->scroll_damage, dmg);
2019-06-29 21:03:28 +02:00
}
void
term_erase(struct terminal *term, int start_row, int start_col,
int end_row, int end_col)
2019-06-29 21:03:28 +02:00
{
xassert(start_row <= end_row);
xassert(start_col <= end_col || start_row < end_row);
if (start_row == end_row) {
struct row *row = grid_row(term->grid, start_row);
erase_cell_range(term, row, start_col, end_col);
sixel_overwrite_by_row(term, start_row, start_col, end_col - start_col + 1);
return;
}
xassert(end_row > start_row);
erase_cell_range(
term, grid_row(term->grid, start_row), start_col, term->cols - 1);
sixel_overwrite_by_row(term, start_row, start_col, term->cols - start_col);
for (int r = start_row + 1; r < end_row; r++)
erase_line(term, grid_row(term->grid, r));
2020-06-27 15:29:47 +02:00
sixel_overwrite_by_rectangle(
term, start_row + 1, 0, end_row - start_row, term->cols);
erase_cell_range(term, grid_row(term->grid, end_row), 0, end_col);
sixel_overwrite_by_row(term, end_row, 0, end_col + 1);
2019-06-29 21:03:28 +02:00
}
void
term_erase_scrollback(struct terminal *term)
{
const struct grid *grid = term->grid;
const int num_rows = grid->num_rows;
const int mask = num_rows - 1;
const int scrollback_history_size = num_rows - term->rows;
if (scrollback_history_size == 0)
return;
const int start = (grid->offset + term->rows) & mask;
const int end = (grid->offset - 1) & mask;
const int rel_start = grid_row_abs_to_sb(grid, term->rows, start);
const int rel_end = grid_row_abs_to_sb(grid, term->rows, end);
const int sel_start = selection_get_start(term).row;
const int sel_end = selection_get_end(term).row;
if (sel_end >= 0) {
/*
* Cancel selection if it touches any of the rows in the
* scrollback, since we can't have the selection reference
* soon-to-be deleted rows.
*
* This is done by range checking the selection range against
* the scrollback range.
*
* To make this comparison simpler, the start/end absolute row
* numbers are "rebased" against the scrollback start, where
* row 0 is the *first* row in the scrollback. A high number
* thus means the row is further *down* in the scrollback,
* closer to the screen bottom.
*/
const int rel_sel_start = grid_row_abs_to_sb(grid, term->rows, sel_start);
const int rel_sel_end = grid_row_abs_to_sb(grid, term->rows, sel_end);
if ((rel_sel_start <= rel_start && rel_sel_end >= rel_start) ||
(rel_sel_start <= rel_end && rel_sel_end >= rel_end) ||
(rel_sel_start >= rel_start && rel_sel_end <= rel_end))
{
selection_cancel(term);
}
}
tll_foreach(term->grid->sixel_images, it) {
struct sixel *six = &it->item;
const int six_start = grid_row_abs_to_sb(grid, term->rows, six->pos.row);
const int six_end = grid_row_abs_to_sb(
grid, term->rows, six->pos.row + six->rows - 1);
if ((six_start <= rel_start && six_end >= rel_start) ||
(six_start <= rel_end && six_end >= rel_end) ||
(six_start >= rel_start && six_end <= rel_end))
{
sixel_destroy(six);
tll_remove(term->grid->sixel_images, it);
}
}
for (int i = start;; i = (i + 1) & mask) {
struct row *row = term->grid->rows[i];
if (row != NULL) {
if (term->render.last_cursor.row == row)
term->render.last_cursor.row = NULL;
grid_row_free(row);
term->grid->rows[i] = NULL;
}
if (i == end)
break;
}
term->grid->view = term->grid->offset;
#if defined(_DEBUG)
for (int i = 0; i < term->rows; i++) {
xassert(grid_row_in_view(term->grid, i) != NULL);
}
#endif
term_damage_view(term);
}
UNITTEST
{
const int scrollback_rows = 16;
const int term_rows = 5;
const int cols = 5;
struct fdm *fdm = fdm_init();
xassert(fdm != NULL);
struct terminal term = {
.fdm = fdm,
.rows = term_rows,
.cols = cols,
.normal = {
.rows = xcalloc(scrollback_rows, sizeof(term.normal.rows[0])),
.num_rows = scrollback_rows,
.num_cols = cols,
},
.grid = &term.normal,
.selection = {
.coords = {
.start = {-1, -1},
.end = {-1, -1},
},
.kind = SELECTION_NONE,
.auto_scroll = {
.fd = timerfd_create(CLOCK_MONOTONIC, TFD_CLOEXEC | TFD_NONBLOCK),
},
},
};
xassert(term.selection.auto_scroll.fd >= 0);
#define populate_scrollback() do { \
for (int i = 0; i < scrollback_rows; i++) { \
if (term.normal.rows[i] == NULL) { \
struct row *r = xcalloc(1, sizeof(*term.normal.rows[i])); \
r->cells = xcalloc(cols, sizeof(r->cells[0])); \
term.normal.rows[i] = r; \
} \
} \
} while (0)
/*
* Test case 1 - no selection, just verify all rows except those
* on screen have been deleted.
*/
populate_scrollback();
term.normal.offset = 11;
term_erase_scrollback(&term);
for (int i = 0; i < scrollback_rows; i++) {
if (i >= term.normal.offset && i < term.normal.offset + term_rows)
xassert(term.normal.rows[i] != NULL);
else
xassert(term.normal.rows[i] == NULL);
}
/*
* Test case 2 - selection that touches the scrollback. Verify the
* selection is cancelled.
*/
term.normal.offset = 14; /* Screen covers rows 14,15,0,1,2 */
/* Selection covers rows 15,0,1,2,3 */
term.selection.coords.start = (struct coord){.row = 15};
term.selection.coords.end = (struct coord){.row = 19};
term.selection.kind = SELECTION_CHAR_WISE;
populate_scrollback();
term_erase_scrollback(&term);
xassert(term.selection.coords.start.row < 0);
xassert(term.selection.coords.end.row < 0);
xassert(term.selection.kind == SELECTION_NONE);
/*
* Test case 3 - selection that does *not* touch the
* scrollback. Verify the selection is *not* cancelled.
*/
/* Selection covers rows 15,0 */
term.selection.coords.start = (struct coord){.row = 15};
term.selection.coords.end = (struct coord){.row = 16};
term.selection.kind = SELECTION_CHAR_WISE;
populate_scrollback();
term_erase_scrollback(&term);
xassert(term.selection.coords.start.row == 15);
xassert(term.selection.coords.end.row == 16);
xassert(term.selection.kind == SELECTION_CHAR_WISE);
term.selection.coords.start = (struct coord){-1, -1};
term.selection.coords.end = (struct coord){-1, -1};
term.selection.kind = SELECTION_NONE;
/*
* Test case 4 - sixel that touch the scrollback
*/
struct sixel six = {
.rows = 5,
.pos = {
.row = 15,
},
};
tll_push_back(term.normal.sixel_images, six);
populate_scrollback();
term_erase_scrollback(&term);
xassert(tll_length(term.normal.sixel_images) == 0);
/*
* Test case 5 - sixel that does *not* touch the scrollback
*/
six.rows = 3;
tll_push_back(term.normal.sixel_images, six);
populate_scrollback();
term_erase_scrollback(&term);
xassert(tll_length(term.normal.sixel_images) == 1);
/* Cleanup */
tll_free(term.normal.sixel_images);
close(term.selection.auto_scroll.fd);
for (int i = 0; i < scrollback_rows; i++)
grid_row_free(term.normal.rows[i]);
free(term.normal.rows);
fdm_destroy(fdm);
}
int
term_row_rel_to_abs(const struct terminal *term, int row)
{
switch (term->origin) {
case ORIGIN_ABSOLUTE:
return min(row, term->rows - 1);
case ORIGIN_RELATIVE:
return min(row + term->scroll_region.start, term->scroll_region.end - 1);
}
BUG("Invalid cursor_origin value");
return -1;
}
2019-06-29 21:03:28 +02:00
void
term_cursor_to(struct terminal *term, int row, int col)
{
xassert(row < term->rows);
xassert(col < term->cols);
2019-06-29 21:03:28 +02:00
term->grid->cursor.lcf = false;
2019-06-29 21:03:28 +02:00
term->grid->cursor.point.col = col;
term->grid->cursor.point.row = row;
term->grid->cur_row = grid_row(term->grid, row);
2019-06-29 21:03:28 +02:00
}
void
term_cursor_home(struct terminal *term)
{
term_cursor_to(term, term_row_rel_to_abs(term, 0), 0);
}
void
term_cursor_col(struct terminal *term, int col)
{
xassert(col < term->cols);
term->grid->cursor.lcf = false;
term->grid->cursor.point.col = col;
}
2019-06-29 21:03:28 +02:00
void
term_cursor_left(struct terminal *term, int count)
{
int move_amount = min(term->grid->cursor.point.col, count);
term->grid->cursor.point.col -= move_amount;
xassert(term->grid->cursor.point.col >= 0);
term->grid->cursor.lcf = false;
2019-06-29 21:03:28 +02:00
}
void
term_cursor_right(struct terminal *term, int count)
{
int move_amount = min(term->cols - term->grid->cursor.point.col - 1, count);
term->grid->cursor.point.col += move_amount;
xassert(term->grid->cursor.point.col < term->cols);
term->grid->cursor.lcf = false;
2019-06-29 21:03:28 +02:00
}
void
term_cursor_up(struct terminal *term, int count)
{
int top = term->origin == ORIGIN_ABSOLUTE ? 0 : term->scroll_region.start;
xassert(term->grid->cursor.point.row >= top);
int move_amount = min(term->grid->cursor.point.row - top, count);
term_cursor_to(term, term->grid->cursor.point.row - move_amount, term->grid->cursor.point.col);
2019-06-29 21:03:28 +02:00
}
void
term_cursor_down(struct terminal *term, int count)
{
int bottom = term->origin == ORIGIN_ABSOLUTE ? term->rows : term->scroll_region.end;
xassert(bottom >= term->grid->cursor.point.row);
int move_amount = min(bottom - term->grid->cursor.point.row - 1, count);
term_cursor_to(term, term->grid->cursor.point.row + move_amount, term->grid->cursor.point.col);
2019-06-29 21:03:28 +02:00
}
static bool
cursor_blink_rearm_timer(struct terminal *term)
{
if (term->cursor_blink.fd < 0) {
int fd = timerfd_create(CLOCK_MONOTONIC, TFD_CLOEXEC | TFD_NONBLOCK);
if (fd < 0) {
LOG_ERRNO("failed to create cursor blink timer FD");
return false;
}
if (!fdm_add(term->fdm, fd, EPOLLIN, &fdm_cursor_blink, term)) {
close(fd);
return false;
}
term->cursor_blink.fd = fd;
}
const int rate_ms = term->conf->cursor.blink.rate_ms;
const long secs = rate_ms / 1000;
const long nsecs = (rate_ms % 1000) * 1000000;
const struct itimerspec timer = {
.it_value = {.tv_sec = secs, .tv_nsec = nsecs},
.it_interval = {.tv_sec = secs, .tv_nsec = nsecs},
};
if (timerfd_settime(term->cursor_blink.fd, 0, &timer, NULL) < 0) {
LOG_ERRNO("failed to arm cursor blink timer");
fdm_del(term->fdm, term->cursor_blink.fd);
term->cursor_blink.fd = -1;
return false;
}
return true;
}
static bool
cursor_blink_disarm_timer(struct terminal *term)
{
fdm_del(term->fdm, term->cursor_blink.fd);
term->cursor_blink.fd = -1;
return true;
}
void
term_cursor_blink_update(struct terminal *term)
{
bool enable = term->cursor_blink.decset || term->cursor_blink.deccsusr;
bool activate = !term->shutdown.in_progress && enable && term->visual_focus;
LOG_DBG("decset=%d, deccsrusr=%d, focus=%d, shutting-down=%d, enable=%d, activate=%d",
term->cursor_blink.decset, term->cursor_blink.deccsusr,
term->visual_focus, term->shutdown.in_progress,
enable, activate);
if (activate && term->cursor_blink.fd < 0) {
term->cursor_blink.state = CURSOR_BLINK_ON;
cursor_blink_rearm_timer(term);
} else if (!activate && term->cursor_blink.fd >= 0)
cursor_blink_disarm_timer(term);
}
term: scrolling: hopefully fix all selection/scrolling related crashes When scrolling, there are a couple of cases where an existing selection must be canceled because we cannot meaningfully represent it after scrolling. These are when the selection is (partly) inside: * The top scrolling region * The bottom scrolling region * The new lines scrolled in. I.e. re-used lines For the scrolling regions, the real problem is when the selection crosses the scrolling region boundary; a selection that is completely inside a scrolling regions _might_ be possible to keep, but we would need to translate the selection coordinates to the new scrolling region lines. For simplicity, we cancel the selection if it touches the scrolling region. Period. The last item, newly scrolled in lines is when the selection covers very old lines and we're now wrapping around the scrollback history. Then there's a fourth problem case: when the user has started a selection, but hasn't yet moved the cursor. In this case, we have no end point. What's more problematic is that when the user (after scrolling) moves the cursor, we try to create a huge selection that covers mostly empty (NULL) rows, causing us to crash. This can happen e.g. when reverse scrolling in such a way that we wrap around the scrollback history. The actual viewport in this case is something like `-n - m`. But the selection we'll end up trying to create will be `m - (rows - n)`. This range may very well contain NULL rows. To deal with this, we simply cancel the selection.
2020-05-17 15:34:49 +02:00
static bool
selection_on_top_region(const struct terminal *term,
struct scroll_region region)
{
return region.start > 0 &&
selection_on_rows(term, 0, region.start - 1);
}
static bool
selection_on_bottom_region(const struct terminal *term,
struct scroll_region region)
{
return region.end < term->rows &&
selection_on_rows(term, region.end, term->rows - 1);
}
2019-06-29 21:03:28 +02:00
void
term_scroll_partial(struct terminal *term, struct scroll_region region, int rows)
{
LOG_DBG("scroll: rows=%d, region.start=%d, region.end=%d",
rows, region.start, region.end);
2019-06-29 21:03:28 +02:00
/* Verify scroll amount has been clamped */
xassert(rows <= region.end - region.start);
2019-06-29 21:03:28 +02:00
term: scrolling: hopefully fix all selection/scrolling related crashes When scrolling, there are a couple of cases where an existing selection must be canceled because we cannot meaningfully represent it after scrolling. These are when the selection is (partly) inside: * The top scrolling region * The bottom scrolling region * The new lines scrolled in. I.e. re-used lines For the scrolling regions, the real problem is when the selection crosses the scrolling region boundary; a selection that is completely inside a scrolling regions _might_ be possible to keep, but we would need to translate the selection coordinates to the new scrolling region lines. For simplicity, we cancel the selection if it touches the scrolling region. Period. The last item, newly scrolled in lines is when the selection covers very old lines and we're now wrapping around the scrollback history. Then there's a fourth problem case: when the user has started a selection, but hasn't yet moved the cursor. In this case, we have no end point. What's more problematic is that when the user (after scrolling) moves the cursor, we try to create a huge selection that covers mostly empty (NULL) rows, causing us to crash. This can happen e.g. when reverse scrolling in such a way that we wrap around the scrollback history. The actual viewport in this case is something like `-n - m`. But the selection we'll end up trying to create will be `m - (rows - n)`. This range may very well contain NULL rows. To deal with this, we simply cancel the selection.
2020-05-17 15:34:49 +02:00
/* Cancel selections that cannot be scrolled */
if (unlikely(term->selection.coords.end.row >= 0)) {
/*
* Selection is (partly) inside either the top or bottom
* scrolling regions, or on (at least one) of the lines
2023-10-03 14:11:55 +02:00
* scrolled in (i.e. reused lines).
*/
if (selection_on_top_region(term, region) ||
selection_on_bottom_region(term, region))
{
term: scrolling: hopefully fix all selection/scrolling related crashes When scrolling, there are a couple of cases where an existing selection must be canceled because we cannot meaningfully represent it after scrolling. These are when the selection is (partly) inside: * The top scrolling region * The bottom scrolling region * The new lines scrolled in. I.e. re-used lines For the scrolling regions, the real problem is when the selection crosses the scrolling region boundary; a selection that is completely inside a scrolling regions _might_ be possible to keep, but we would need to translate the selection coordinates to the new scrolling region lines. For simplicity, we cancel the selection if it touches the scrolling region. Period. The last item, newly scrolled in lines is when the selection covers very old lines and we're now wrapping around the scrollback history. Then there's a fourth problem case: when the user has started a selection, but hasn't yet moved the cursor. In this case, we have no end point. What's more problematic is that when the user (after scrolling) moves the cursor, we try to create a huge selection that covers mostly empty (NULL) rows, causing us to crash. This can happen e.g. when reverse scrolling in such a way that we wrap around the scrollback history. The actual viewport in this case is something like `-n - m`. But the selection we'll end up trying to create will be `m - (rows - n)`. This range may very well contain NULL rows. To deal with this, we simply cancel the selection.
2020-05-17 15:34:49 +02:00
selection_cancel(term);
} else
selection_scroll_up(term, rows);
term: scrolling: hopefully fix all selection/scrolling related crashes When scrolling, there are a couple of cases where an existing selection must be canceled because we cannot meaningfully represent it after scrolling. These are when the selection is (partly) inside: * The top scrolling region * The bottom scrolling region * The new lines scrolled in. I.e. re-used lines For the scrolling regions, the real problem is when the selection crosses the scrolling region boundary; a selection that is completely inside a scrolling regions _might_ be possible to keep, but we would need to translate the selection coordinates to the new scrolling region lines. For simplicity, we cancel the selection if it touches the scrolling region. Period. The last item, newly scrolled in lines is when the selection covers very old lines and we're now wrapping around the scrollback history. Then there's a fourth problem case: when the user has started a selection, but hasn't yet moved the cursor. In this case, we have no end point. What's more problematic is that when the user (after scrolling) moves the cursor, we try to create a huge selection that covers mostly empty (NULL) rows, causing us to crash. This can happen e.g. when reverse scrolling in such a way that we wrap around the scrollback history. The actual viewport in this case is something like `-n - m`. But the selection we'll end up trying to create will be `m - (rows - n)`. This range may very well contain NULL rows. To deal with this, we simply cancel the selection.
2020-05-17 15:34:49 +02:00
}
sixel_scroll_up(term, rows);
/* How many lines from the scrollback start is the current viewport? */
int view_sb_start_distance = grid_row_abs_to_sb(
term->grid, term->rows, term->grid->view);
bool view_follows = term->grid->view == term->grid->offset;
term->grid->offset += rows;
term->grid->offset &= term->grid->num_rows - 1;
2019-07-01 19:20:21 +02:00
if (likely(view_follows)) {
term: scroll: only record scroll damage when viewport is at the bottom We don’t need to record scroll damage if the viewport isn’t at the bottom, since in this case, the renderer ignores the scroll damage anyway. This fixes a performance corner case, when the viewport is at the top of the scrollback history. When application scrolls the terminal contents, and the scrollback history is full, and the viewport is at top of the history, then the viewport needs to be moved (the scrollback history is a circular buffer, and thus the top of the history “moves” when we’re scrolling in new contents). Moving the viewport typically results in another type of scroll damage (DAMAGE_SCROLL_IN_VIEW, instead of the “normal” DAMAGE_SCROLL). Thus, each application triggered scroll, will result in two scroll damage records: one DAMAGE_SCROLL, and one DAMAGE_SCROLL_IN_VIEW. These two are incompatible, meaning they can’t be merged. What’s worse, it also means the DAMAGE_SCROLL records from two application triggered scrolls cannot be merged (since there’s a DAMAGE_SCROLL_IN_VIEW in between). As a result, the renderer will not see one, or “a few” scroll damage events, but a *ton*. _Each_ one typically a single line, or so. And each one resulting in lots of traffic on the wayland socket, as we create and destroy new buffer pools, when doing “shm scrolling”. This eventually leads to the socket not being able to keep up, and the socket is closed on us, forcing us to exit. The fix is really simple: don’t record “normal” scroll damage when scrolling, _unless_ the viewport is at the bottom (and thus “follows” the application output). As soon as the user scrolls up in the history, we’ll stop emitting normal scroll damage records. This is just fine, since, as mentioned above, the renderer ignores them when the viewport isn’t at the bottom. What if the viewport is moved back down again, before the next frame has been rendered? Wont there be “missing” scroll damage records? No, because moving the viewport results in scroll damage records by itself. Closes #1380
2023-06-23 20:20:01 +02:00
term_damage_scroll(term, DAMAGE_SCROLL, region, rows);
selection_view_down(term, term->grid->offset);
term->grid->view = term->grid->offset;
} else if (unlikely(rows > view_sb_start_distance)) {
/* Part of current view is being scrolled out */
int new_view = grid_row_sb_to_abs(term->grid, term->rows, 0);
selection_view_down(term, new_view);
cmd_scrollback_down(term, rows - view_sb_start_distance);
}
/* Top non-scrolling region. */
for (int i = region.start - 1; i >= 0; i--)
grid_swap_row(term->grid, i - rows, i);
/* Bottom non-scrolling region */
for (int i = term->rows - 1; i >= region.end; i--)
grid_swap_row(term->grid, i - rows, i);
/* Erase scrolled in lines */
for (int r = region.end - rows; r < region.end; r++) {
struct row *row = grid_row_and_alloc(term->grid, r);
erase_line(term, row);
}
term: scroll: only record scroll damage when viewport is at the bottom We don’t need to record scroll damage if the viewport isn’t at the bottom, since in this case, the renderer ignores the scroll damage anyway. This fixes a performance corner case, when the viewport is at the top of the scrollback history. When application scrolls the terminal contents, and the scrollback history is full, and the viewport is at top of the history, then the viewport needs to be moved (the scrollback history is a circular buffer, and thus the top of the history “moves” when we’re scrolling in new contents). Moving the viewport typically results in another type of scroll damage (DAMAGE_SCROLL_IN_VIEW, instead of the “normal” DAMAGE_SCROLL). Thus, each application triggered scroll, will result in two scroll damage records: one DAMAGE_SCROLL, and one DAMAGE_SCROLL_IN_VIEW. These two are incompatible, meaning they can’t be merged. What’s worse, it also means the DAMAGE_SCROLL records from two application triggered scrolls cannot be merged (since there’s a DAMAGE_SCROLL_IN_VIEW in between). As a result, the renderer will not see one, or “a few” scroll damage events, but a *ton*. _Each_ one typically a single line, or so. And each one resulting in lots of traffic on the wayland socket, as we create and destroy new buffer pools, when doing “shm scrolling”. This eventually leads to the socket not being able to keep up, and the socket is closed on us, forcing us to exit. The fix is really simple: don’t record “normal” scroll damage when scrolling, _unless_ the viewport is at the bottom (and thus “follows” the application output). As soon as the user scrolls up in the history, we’ll stop emitting normal scroll damage records. This is just fine, since, as mentioned above, the renderer ignores them when the viewport isn’t at the bottom. What if the viewport is moved back down again, before the next frame has been rendered? Wont there be “missing” scroll damage records? No, because moving the viewport results in scroll damage records by itself. Closes #1380
2023-06-23 20:20:01 +02:00
term->grid->cur_row = grid_row(term->grid, term->grid->cursor.point.row);
#if defined(_DEBUG)
for (int r = 0; r < term->rows; r++)
xassert(grid_row(term->grid, r) != NULL);
#endif
2019-06-29 21:03:28 +02:00
}
void
term_scroll(struct terminal *term, int rows)
{
term_scroll_partial(term, term->scroll_region, rows);
2019-06-29 21:03:28 +02:00
}
void
term_scroll_reverse_partial(struct terminal *term,
struct scroll_region region, int rows)
{
LOG_DBG("scroll reverse: rows=%d, region.start=%d, region.end=%d",
rows, region.start, region.end);
/* Verify scroll amount has been clamped */
xassert(rows <= region.end - region.start);
term: scrolling: hopefully fix all selection/scrolling related crashes When scrolling, there are a couple of cases where an existing selection must be canceled because we cannot meaningfully represent it after scrolling. These are when the selection is (partly) inside: * The top scrolling region * The bottom scrolling region * The new lines scrolled in. I.e. re-used lines For the scrolling regions, the real problem is when the selection crosses the scrolling region boundary; a selection that is completely inside a scrolling regions _might_ be possible to keep, but we would need to translate the selection coordinates to the new scrolling region lines. For simplicity, we cancel the selection if it touches the scrolling region. Period. The last item, newly scrolled in lines is when the selection covers very old lines and we're now wrapping around the scrollback history. Then there's a fourth problem case: when the user has started a selection, but hasn't yet moved the cursor. In this case, we have no end point. What's more problematic is that when the user (after scrolling) moves the cursor, we try to create a huge selection that covers mostly empty (NULL) rows, causing us to crash. This can happen e.g. when reverse scrolling in such a way that we wrap around the scrollback history. The actual viewport in this case is something like `-n - m`. But the selection we'll end up trying to create will be `m - (rows - n)`. This range may very well contain NULL rows. To deal with this, we simply cancel the selection.
2020-05-17 15:34:49 +02:00
/* Cancel selections that cannot be scrolled */
if (unlikely(term->selection.coords.end.row >= 0)) {
/*
* Selection is (partly) inside either the top or bottom
* scrolling regions, or on (at least one) of the lines
2023-10-03 14:11:55 +02:00
* scrolled in (i.e. reused lines).
*/
if (selection_on_top_region(term, region) ||
selection_on_bottom_region(term, region))
{
term: scrolling: hopefully fix all selection/scrolling related crashes When scrolling, there are a couple of cases where an existing selection must be canceled because we cannot meaningfully represent it after scrolling. These are when the selection is (partly) inside: * The top scrolling region * The bottom scrolling region * The new lines scrolled in. I.e. re-used lines For the scrolling regions, the real problem is when the selection crosses the scrolling region boundary; a selection that is completely inside a scrolling regions _might_ be possible to keep, but we would need to translate the selection coordinates to the new scrolling region lines. For simplicity, we cancel the selection if it touches the scrolling region. Period. The last item, newly scrolled in lines is when the selection covers very old lines and we're now wrapping around the scrollback history. Then there's a fourth problem case: when the user has started a selection, but hasn't yet moved the cursor. In this case, we have no end point. What's more problematic is that when the user (after scrolling) moves the cursor, we try to create a huge selection that covers mostly empty (NULL) rows, causing us to crash. This can happen e.g. when reverse scrolling in such a way that we wrap around the scrollback history. The actual viewport in this case is something like `-n - m`. But the selection we'll end up trying to create will be `m - (rows - n)`. This range may very well contain NULL rows. To deal with this, we simply cancel the selection.
2020-05-17 15:34:49 +02:00
selection_cancel(term);
} else
selection_scroll_down(term, rows);
term: scrolling: hopefully fix all selection/scrolling related crashes When scrolling, there are a couple of cases where an existing selection must be canceled because we cannot meaningfully represent it after scrolling. These are when the selection is (partly) inside: * The top scrolling region * The bottom scrolling region * The new lines scrolled in. I.e. re-used lines For the scrolling regions, the real problem is when the selection crosses the scrolling region boundary; a selection that is completely inside a scrolling regions _might_ be possible to keep, but we would need to translate the selection coordinates to the new scrolling region lines. For simplicity, we cancel the selection if it touches the scrolling region. Period. The last item, newly scrolled in lines is when the selection covers very old lines and we're now wrapping around the scrollback history. Then there's a fourth problem case: when the user has started a selection, but hasn't yet moved the cursor. In this case, we have no end point. What's more problematic is that when the user (after scrolling) moves the cursor, we try to create a huge selection that covers mostly empty (NULL) rows, causing us to crash. This can happen e.g. when reverse scrolling in such a way that we wrap around the scrollback history. The actual viewport in this case is something like `-n - m`. But the selection we'll end up trying to create will be `m - (rows - n)`. This range may very well contain NULL rows. To deal with this, we simply cancel the selection.
2020-05-17 15:34:49 +02:00
}
/* Unallocate scrolled out lines */
for (int r = region.end - rows; r < region.end; r++) {
const int abs_r = grid_row_absolute(term->grid, r);
struct row *row = term->grid->rows[abs_r];
grid_row_free(row);
term->grid->rows[abs_r] = NULL;
if (term->render.last_cursor.row == row)
term->render.last_cursor.row = NULL;
}
sixel_scroll_down(term, rows);
bool view_follows = term->grid->view == term->grid->offset;
term->grid->offset -= rows;
term->grid->offset += term->grid->num_rows;
term->grid->offset &= term->grid->num_rows - 1;
xassert(term->grid->offset >= 0);
xassert(term->grid->offset < term->grid->num_rows);
2019-06-29 21:03:28 +02:00
if (view_follows) {
term: scroll: only record scroll damage when viewport is at the bottom We don’t need to record scroll damage if the viewport isn’t at the bottom, since in this case, the renderer ignores the scroll damage anyway. This fixes a performance corner case, when the viewport is at the top of the scrollback history. When application scrolls the terminal contents, and the scrollback history is full, and the viewport is at top of the history, then the viewport needs to be moved (the scrollback history is a circular buffer, and thus the top of the history “moves” when we’re scrolling in new contents). Moving the viewport typically results in another type of scroll damage (DAMAGE_SCROLL_IN_VIEW, instead of the “normal” DAMAGE_SCROLL). Thus, each application triggered scroll, will result in two scroll damage records: one DAMAGE_SCROLL, and one DAMAGE_SCROLL_IN_VIEW. These two are incompatible, meaning they can’t be merged. What’s worse, it also means the DAMAGE_SCROLL records from two application triggered scrolls cannot be merged (since there’s a DAMAGE_SCROLL_IN_VIEW in between). As a result, the renderer will not see one, or “a few” scroll damage events, but a *ton*. _Each_ one typically a single line, or so. And each one resulting in lots of traffic on the wayland socket, as we create and destroy new buffer pools, when doing “shm scrolling”. This eventually leads to the socket not being able to keep up, and the socket is closed on us, forcing us to exit. The fix is really simple: don’t record “normal” scroll damage when scrolling, _unless_ the viewport is at the bottom (and thus “follows” the application output). As soon as the user scrolls up in the history, we’ll stop emitting normal scroll damage records. This is just fine, since, as mentioned above, the renderer ignores them when the viewport isn’t at the bottom. What if the viewport is moved back down again, before the next frame has been rendered? Wont there be “missing” scroll damage records? No, because moving the viewport results in scroll damage records by itself. Closes #1380
2023-06-23 20:20:01 +02:00
term_damage_scroll(term, DAMAGE_SCROLL_REVERSE, region, rows);
selection_view_up(term, term->grid->offset);
term->grid->view = term->grid->offset;
}
2019-07-09 09:17:24 +02:00
/* Bottom non-scrolling region */
for (int i = region.end + rows; i < term->rows + rows; i++)
grid_swap_row(term->grid, i, i - rows);
2019-07-09 09:17:24 +02:00
/* Top non-scrolling region */
for (int i = 0 + rows; i < region.start + rows; i++)
grid_swap_row(term->grid, i, i - rows);
2019-06-29 21:03:28 +02:00
/* Erase scrolled in lines */
for (int r = region.start; r < region.start + rows; r++) {
struct row *row = grid_row_and_alloc(term->grid, r);
erase_line(term, row);
}
term->grid->cur_row = grid_row(term->grid, term->grid->cursor.point.row);
#if defined(_DEBUG)
for (int r = 0; r < term->rows; r++)
xassert(grid_row(term->grid, r) != NULL);
#endif
2019-06-29 21:03:28 +02:00
}
void
term_scroll_reverse(struct terminal *term, int rows)
{
term_scroll_reverse_partial(term, term->scroll_region, rows);
2019-06-29 21:03:28 +02:00
}
void
term_carriage_return(struct terminal *term)
{
term_cursor_left(term, term->grid->cursor.point.col);
}
void
term_linefeed(struct terminal *term)
{
term->grid->cur_row->linebreak = true;
term->grid->cursor.lcf = false;
if (term->grid->cursor.point.row == term->scroll_region.end - 1)
term_scroll(term, 1);
else
term_cursor_down(term, 1);
}
void
term_reverse_index(struct terminal *term)
{
if (term->grid->cursor.point.row == term->scroll_region.start)
term_scroll_reverse(term, 1);
else
term_cursor_up(term, 1);
}
void
term_reset_view(struct terminal *term)
{
if (term->grid->view == term->grid->offset)
return;
term->grid->view = term->grid->offset;
term_damage_view(term);
}
void
term_save_cursor(struct terminal *term)
{
term->grid->saved_cursor = term->grid->cursor;
term->vt.saved_attrs = term->vt.attrs;
term->saved_charsets = term->charsets;
}
void
term_restore_cursor(struct terminal *term, const struct cursor *cursor)
{
int row = min(cursor->point.row, term->rows - 1);
int col = min(cursor->point.col, term->cols - 1);
term_cursor_to(term, row, col);
term->grid->cursor.lcf = cursor->lcf;
term->vt.attrs = term->vt.saved_attrs;
term->charsets = term->saved_charsets;
term->bits_affecting_ascii_printer.charset =
term->charsets.set[term->charsets.selected] != CHARSET_ASCII;
term_update_ascii_printer(term);
}
2019-07-16 10:34:08 +02:00
void
term_visual_focus_in(struct terminal *term)
2019-07-16 10:34:08 +02:00
{
if (term->visual_focus)
return;
term->visual_focus = true;
term_cursor_blink_update(term);
render_refresh_csd(term);
2019-07-16 10:34:08 +02:00
}
void
term_visual_focus_out(struct terminal *term)
2019-07-16 10:34:08 +02:00
{
if (!term->visual_focus)
return;
term->visual_focus = false;
term_cursor_blink_update(term);
render_refresh_csd(term);
}
void
term_kbd_focus_in(struct terminal *term)
{
if (term->kbd_focus)
return;
term->kbd_focus = true;
if (term->render.urgency) {
term->render.urgency = false;
term_damage_margins(term);
}
cursor_refresh(term);
if (term->focus_events)
term_to_slave(term, "\033[I", 3);
}
void
term_kbd_focus_out(struct terminal *term)
{
if (!term->kbd_focus)
return;
tll_foreach(term->wl->seats, it)
if (it->item.kbd_focus == term)
return;
2020-12-03 18:36:56 +01:00
#if defined(FOOT_IME_ENABLED) && FOOT_IME_ENABLED
if (term_ime_reset(term))
render_refresh(term);
2020-12-03 18:36:56 +01:00
#endif
term->kbd_focus = false;
cursor_refresh(term);
if (term->focus_events)
term_to_slave(term, "\033[O", 3);
2019-07-16 10:34:08 +02:00
}
static int
linux_mouse_button_to_x(int button)
{
/* Note: on X11, scroll events where reported as buttons. Not so
* on Wayland. We manually map scroll events to custom "button"
* defines (BTN_WHEEL_*).
*/
switch (button) {
case BTN_LEFT: return 1;
case BTN_MIDDLE: return 2;
case BTN_RIGHT: return 3;
case BTN_WHEEL_BACK: return 4; /* Foot custom define */
case BTN_WHEEL_FORWARD: return 5; /* Foot custom define */
case BTN_WHEEL_LEFT: return 6; /* Foot custom define */
case BTN_WHEEL_RIGHT: return 7; /* Foot custom define */
case BTN_SIDE: return 8;
case BTN_EXTRA: return 9;
case BTN_FORWARD: return 10;
case BTN_BACK: return 11;
case BTN_TASK: return 12; /* Guessing... */
default:
LOG_WARN("unrecognized mouse button: %d (0x%x)", button, button);
return -1;
}
}
static int
encode_xbutton(int xbutton)
{
switch (xbutton) {
case 1: case 2: case 3:
return xbutton - 1;
case 4: case 5: case 6: case 7:
/* Like button 1 and 2, but with 64 added */
return xbutton - 4 + 64;
case 8: case 9: case 10: case 11:
/* Similar to 4 and 5, but adding 128 instead of 64 */
return xbutton - 8 + 128;
default:
LOG_ERR("cannot encode X mouse button: %d", xbutton);
return -1;
}
}
static void
report_mouse_click(struct terminal *term, int encoded_button, int row, int col,
2021-12-30 05:13:45 -06:00
int row_pixels, int col_pixels, bool release)
{
char response[128];
switch (term->mouse_reporting) {
case MOUSE_NORMAL: {
int encoded_col = 32 + col + 1;
int encoded_row = 32 + row + 1;
if (encoded_col > 255 || encoded_row > 255)
return;
snprintf(response, sizeof(response), "\033[M%c%c%c",
32 + (release ? 3 : encoded_button), encoded_col, encoded_row);
break;
}
case MOUSE_SGR:
snprintf(response, sizeof(response), "\033[<%d;%d;%d%c",
encoded_button, col + 1, row + 1, release ? 'm' : 'M');
break;
2021-12-30 05:13:45 -06:00
case MOUSE_SGR_PIXELS:
snprintf(response, sizeof(response), "\033[<%d;%d;%d%c",
encoded_button, col_pixels + 1, row_pixels + 1, release ? 'm' : 'M');
break;
case MOUSE_URXVT:
snprintf(response, sizeof(response), "\033[%d;%d;%dM",
32 + (release ? 3 : encoded_button), col + 1, row + 1);
break;
case MOUSE_UTF8:
/* Unimplemented */
return;
}
term_to_slave(term, response, strlen(response));
}
static void
2021-12-30 05:13:45 -06:00
report_mouse_motion(struct terminal *term, int encoded_button, int row, int col, int row_pixels, int col_pixels)
{
2021-12-30 05:13:45 -06:00
report_mouse_click(term, encoded_button, row, col, row_pixels, col_pixels, false);
}
bool
term_mouse_grabbed(const struct terminal *term, const struct seat *seat)
{
/*
* Mouse is grabbed by us, regardless of whether mouse tracking
* has been enabled or not.
*/
xkb_mod_mask_t mods;
get_current_modifiers(seat, &mods, NULL, 0, true);
const struct key_binding_set *bindings =
key_binding_for(term->wl->key_binding_manager, term->conf, seat);
const xkb_mod_mask_t override_modmask = bindings->selection_overrides;
bool override_mods_pressed = (mods & override_modmask) == override_modmask;
return term->mouse_tracking == MOUSE_NONE ||
(seat->kbd_focus == term && override_mods_pressed);
}
void
term_mouse_down(struct terminal *term, int button, int row, int col,
2021-12-30 05:13:45 -06:00
int row_pixels, int col_pixels,
bool _shift, bool _alt, bool _ctrl)
{
/* Map libevent button event code to X button number */
int xbutton = linux_mouse_button_to_x(button);
if (xbutton == -1)
return;
int encoded = encode_xbutton(xbutton);
if (encoded == -1)
return;
bool has_focus = term->kbd_focus;
bool shift = has_focus ? _shift : false;
bool alt = has_focus ? _alt : false;
bool ctrl = has_focus ? _ctrl : false;
encoded += (shift ? 4 : 0) + (alt ? 8 : 0) + (ctrl ? 16 : 0);
switch (term->mouse_tracking) {
case MOUSE_NONE:
break;
case MOUSE_CLICK:
case MOUSE_DRAG:
case MOUSE_MOTION:
2021-12-30 05:13:45 -06:00
report_mouse_click(term, encoded, row, col, row_pixels, col_pixels, false);
break;
case MOUSE_X10:
/* Never enabled */
BUG("X10 mouse mode not implemented");
break;
}
}
void
term_mouse_up(struct terminal *term, int button, int row, int col,
2021-12-30 05:13:45 -06:00
int row_pixels, int col_pixels,
bool _shift, bool _alt, bool _ctrl)
{
/* Map libevent button event code to X button number */
int xbutton = linux_mouse_button_to_x(button);
if (xbutton == -1)
return;
if (xbutton == 4 || xbutton == 5) {
/* No release events for vertical scroll wheel buttons */
return;
}
int encoded = encode_xbutton(xbutton);
if (encoded == -1)
return;
bool has_focus = term->kbd_focus;
bool shift = has_focus ? _shift : false;
bool alt = has_focus ? _alt : false;
bool ctrl = has_focus ? _ctrl : false;
encoded += (shift ? 4 : 0) + (alt ? 8 : 0) + (ctrl ? 16 : 0);
switch (term->mouse_tracking) {
case MOUSE_NONE:
break;
case MOUSE_CLICK:
case MOUSE_DRAG:
case MOUSE_MOTION:
2021-12-30 05:13:45 -06:00
report_mouse_click(term, encoded, row, col, row_pixels, col_pixels, true);
break;
case MOUSE_X10:
/* Never enabled */
BUG("X10 mouse mode not implemented");
break;
}
}
void
term_mouse_motion(struct terminal *term, int button, int row, int col,
2021-12-30 05:13:45 -06:00
int row_pixels, int col_pixels,
bool _shift, bool _alt, bool _ctrl)
{
int encoded = 0;
if (button != 0) {
/* Map libevent button event code to X button number */
int xbutton = linux_mouse_button_to_x(button);
if (xbutton == -1)
return;
encoded = encode_xbutton(xbutton);
if (encoded == -1)
return;
} else
encoded = 3; /* "released" */
bool has_focus = term->kbd_focus;
bool shift = has_focus ? _shift : false;
bool alt = has_focus ? _alt : false;
bool ctrl = has_focus ? _ctrl : false;
encoded += 32; /* Motion event */
encoded += (shift ? 4 : 0) + (alt ? 8 : 0) + (ctrl ? 16 : 0);
switch (term->mouse_tracking) {
case MOUSE_NONE:
case MOUSE_CLICK:
return;
case MOUSE_DRAG:
if (button == 0)
return;
/* FALLTHROUGH */
case MOUSE_MOTION:
2021-12-30 05:13:45 -06:00
report_mouse_motion(term, encoded, row, col, row_pixels, col_pixels);
break;
case MOUSE_X10:
/* Never enabled */
BUG("X10 mouse mode not implemented");
break;
}
}
void
term_xcursor_update_for_seat(struct terminal *term, struct seat *seat)
{
enum cursor_shape shape = CURSOR_SHAPE_NONE;
switch (term->active_surface) {
case TERM_SURF_GRID:
if (seat->pointer.hidden)
shape = CURSOR_SHAPE_HIDDEN;
else if (cursor_string_to_server_shape(term->mouse_user_cursor) != 0 ||
render_xcursor_is_valid(seat, term->mouse_user_cursor))
{
shape = CURSOR_SHAPE_CUSTOM;
}
input: allow mouse selections to start inside the margins Before this, margins were special cased: * The mouse cursor was always a pointer, and never an I-beam (thus signaling selections cannot be made). * The internal mouse coords where set to -1 when the cursor was inside the margins, causing: - text selections from being made - mouse events being passed to mouse grabbing applications In particular, even with a one-pixel margin, making selections was unnecessarily hard in e.g. fullscreen mode, where you'd expect to be able to throw the cursor into the corner of the screen and then start a selection. With this patch, the cursor is treated as if it was in the first/last column/row, when inside the margin(s). An unintended side-effect of this, initially, was that auto-scrolling selections where way too easy to trigger, since part of its logic is checking if the cursor is inside the margins. That problem has been reduced by two things: * auto-scrolling does not occur unless a selection has been started. That is, just holding down the mouse in the margins and moving up/down doesn't cause scrolling. You have to first select at least one cell in the visible viewport. * A selection isn't fully started (i.e. a cell is actually selected) unless the cursor is inside the actual grid, and *not* in the margins. What does the last point mean? We now allow a selection to be _started_ when clicking in the margin. What this means internally is we set a start coordinate for a selection, but *not* and end coordinate. At this point, we don't have an actual selection. Nothing is selected, and no cells are highlighted, graphically. This happens when we set an end coordinate. Without the last bullet point, that would happen as soon as the cursor was _moved_, even if still inside the margins. Now, we require the cursor to leave the margins and touch an actual cell before we set an end coordinate. Closes #1702
2024-07-18 08:08:44 +02:00
else if (term_mouse_grabbed(term, seat)) {
shape = CURSOR_SHAPE_TEXT;
}
else
shape = CURSOR_SHAPE_LEFT_PTR;
break;
case TERM_SURF_TITLE:
case TERM_SURF_BUTTON_MINIMIZE:
case TERM_SURF_BUTTON_MAXIMIZE:
case TERM_SURF_BUTTON_CLOSE:
shape = CURSOR_SHAPE_LEFT_PTR;
break;
case TERM_SURF_BORDER_LEFT:
case TERM_SURF_BORDER_RIGHT:
case TERM_SURF_BORDER_TOP:
case TERM_SURF_BORDER_BOTTOM:
shape = xcursor_for_csd_border(term, seat->mouse.x, seat->mouse.y);
break;
case TERM_SURF_NONE:
return;
}
if (shape == CURSOR_SHAPE_NONE)
BUG("xcursor not set");
render_xcursor_set(seat, term, shape);
}
void
term_xcursor_update(struct terminal *term)
{
tll_foreach(term->wl->seats, it)
term_xcursor_update_for_seat(term, &it->item);
}
void
term_set_window_title(struct terminal *term, const char *title)
{
if (term->conf->locked_title && term->window_title_has_been_set)
return;
if (term->window_title != NULL && streq(term->window_title, title))
return;
if (mbsntoc32(NULL, title, strlen(title), 0) == (char32_t)-1) {
/* It's an xdg_toplevel::set_title() protocol violation to set
a title with an invalid UTF-8 sequence */
LOG_WARN("%s: title is not valid UTF-8, ignoring", title);
return;
}
free(term->window_title);
term->window_title = xstrdup(title);
render_refresh_title(term);
term->window_title_has_been_set = true;
}
void
term_set_app_id(struct terminal *term, const char *app_id)
{
if (app_id != NULL && *app_id == '\0')
app_id = NULL;
if (term->app_id == NULL && app_id == NULL)
return;
if (term->app_id != NULL && app_id != NULL && strcmp(term->app_id, app_id) == 0)
return;
free(term->app_id);
if (app_id != NULL) {
term->app_id = xstrdup(app_id);
} else {
term->app_id = NULL;
}
render_refresh_app_id(term);
}
void
term_flash(struct terminal *term, unsigned duration_ms)
{
LOG_DBG("FLASH for %ums", duration_ms);
struct itimerspec alarm = {
.it_value = {.tv_sec = 0, .tv_nsec = duration_ms * 1000000},
};
if (timerfd_settime(term->flash.fd, 0, &alarm, NULL) < 0)
LOG_ERRNO("failed to arm flash timer");
else {
term->flash.active = true;
}
}
void
term_bell(struct terminal *term)
{
if (!term->bell_action_enabled)
return;
if (term->conf->bell.urgent && !term->kbd_focus) {
if (!wayl_win_set_urgent(term->window)) {
/*
* Urgency (xdg-activation) is relatively new in
* Wayland. Fallback to our old, "faked", urgency -
* rendering our window margins in red
*/
term->render.urgency = true;
term_damage_margins(term);
}
}
if (term->conf->bell.notify) {
notify_notify(term, "Bell", "Bell in terminal",
NOTIFY_ALWAYS, NOTIFY_URGENCY_NORMAL);
}
if (term->conf->bell.flash)
term_flash(term, 100);
if ((term->conf->bell.command.argv.args != NULL) &&
(!term->kbd_focus || term->conf->bell.command_focused))
{
int devnull = open("/dev/null", O_RDONLY);
spawn(term->reaper, NULL, term->conf->bell.command.argv.args,
devnull, -1, -1, NULL);
if (devnull >= 0)
close(devnull);
}
}
bool
term_spawn_new(const struct terminal *term)
{
return spawn(
term->reaper, term->cwd, (char *const []){term->foot_exe, NULL},
-1, -1, -1, NULL);
}
void
term_enable_app_sync_updates(struct terminal *term)
{
term->render.app_sync_updates.enabled = true;
if (timerfd_settime(
term->render.app_sync_updates.timer_fd, 0,
&(struct itimerspec){.it_value = {.tv_sec = 1}}, NULL) < 0)
{
LOG_ERR("failed to arm timer for application synchronized updates");
}
/* Disable pending refresh *iff* the grid is the *only* thing
* scheduled to be re-rendered */
if (!term->render.refresh.csd && !term->render.refresh.search &&
!term->render.pending.csd && !term->render.pending.search)
{
term->render.refresh.grid = false;
term->render.pending.grid = false;
}
/* Disarm delayed rendering timers */
timerfd_settime(
term->delayed_render_timer.lower_fd, 0,
2020-08-23 07:42:20 +02:00
&(struct itimerspec){{0}}, NULL);
timerfd_settime(
term->delayed_render_timer.upper_fd, 0,
2020-08-23 07:42:20 +02:00
&(struct itimerspec){{0}}, NULL);
term->delayed_render_timer.is_armed = false;
}
void
term_disable_app_sync_updates(struct terminal *term)
{
if (!term->render.app_sync_updates.enabled)
return;
term->render.app_sync_updates.enabled = false;
render_refresh(term);
/* Reset timers */
timerfd_settime(
term->render.app_sync_updates.timer_fd, 0,
2020-08-23 07:42:20 +02:00
&(struct itimerspec){{0}}, NULL);
}
static inline void
print_linewrap(struct terminal *term)
{
if (likely(!term->grid->cursor.lcf)) {
/* Not and end of line */
return;
}
if (unlikely(!term->auto_margin)) {
/* Auto-wrap disabled */
return;
}
term->grid->cur_row->linebreak = false;
term->grid->cursor.lcf = false;
const int row = term->grid->cursor.point.row;
if (row == term->scroll_region.end - 1)
term_scroll(term, 1);
else {
const int new_row = min(row + 1, term->rows - 1);
term->grid->cursor.point.row = new_row;
term->grid->cur_row = grid_row(term->grid, new_row);
}
term->grid->cursor.point.col = 0;
}
static inline void
print_insert(struct terminal *term, int width)
{
2020-06-09 17:31:28 +02:00
if (likely(!term->insert_mode))
return;
xassert(width > 0);
2020-06-09 17:31:28 +02:00
struct row *row = term->grid->cur_row;
const size_t move_count = max(0, term->cols - term->grid->cursor.point.col - width);
2020-06-09 17:31:28 +02:00
memmove(
&row->cells[term->grid->cursor.point.col + width],
&row->cells[term->grid->cursor.point.col],
move_count * sizeof(struct cell));
/* Mark moved cells as dirty */
for (size_t i = term->grid->cursor.point.col + width; i < term->cols; i++)
row->cells[i].attrs.clean = 0;
}
static void
print_spacer(struct terminal *term, int col, int remaining)
{
struct grid *grid = term->grid;
struct row *row = grid->cur_row;
struct cell *cell = &row->cells[col];
cell->wc = CELL_SPACER + remaining;
cell->attrs = term->vt.attrs;
}
/*
* Puts a character on the grid. Coordinates are in screen coordinates
* (i.e. cursor coordinates).
*
* Does NOT:
* - update the cursor
* - linewrap
* - erase sixels
*
* Limitations:
* - double width characters not supported
*/
void
term_fill(struct terminal *term, int r, int c, uint8_t data, size_t count,
bool use_sgr_attrs)
{
struct row *row = grid_row(term->grid, r);
row->dirty = true;
xassert(c + count <= term->cols);
struct attributes attrs = use_sgr_attrs
? term->vt.attrs
: (struct attributes){0};
const struct cell *last = &row->cells[c + count];
for (struct cell *cell = &row->cells[c]; cell < last; cell++) {
cell->wc = data;
cell->attrs = attrs;
/* TODO: why do we print the URI here, and then erase it below? */
if (unlikely(use_sgr_attrs && term->vt.osc8.uri != NULL)) {
grid_row_uri_range_put(row, c, term->vt.osc8.uri, term->vt.osc8.id);
switch (term->conf->url.osc8_underline) {
case OSC8_UNDERLINE_ALWAYS:
cell->attrs.url = true;
break;
case OSC8_UNDERLINE_URL_MODE:
break;
}
}
if (unlikely(use_sgr_attrs &&
(term->vt.underline.style > UNDERLINE_SINGLE ||
term->vt.underline.color_src != COLOR_DEFAULT)))
{
grid_row_underline_range_put(row, c, term->vt.underline);
}
}
if (unlikely(row->extra != NULL)) {
if (likely(term->vt.osc8.uri != NULL))
grid_row_uri_range_erase(row, c, c + count - 1);
if (likely(term->vt.underline.style <= UNDERLINE_SINGLE &&
term->vt.underline.color_src == COLOR_DEFAULT))
{
/* No extended/styled underlines active, so erase any such
attributes at the target columns */
grid_row_underline_range_erase(row, c, c + count - 1);
}
}
}
void
term_print(struct terminal *term, char32_t wc, int width)
{
xassert(width > 0);
struct grid *grid = term->grid;
if (unlikely(term->charsets.set[term->charsets.selected] == CHARSET_GRAPHIC) &&
wc >= 0x60 && wc <= 0x7e)
{
/* 0x60 - 0x7e */
static const char32_t vt100_0[] = {
U'', U'', U'', U'', U'', U'', U'°', U'±', /* ` - g */
U'', U'', U'', U'', U'', U'', U'', U'', /* h - o */
U'', U'', U'', U'', U'', U'', U'', U'', /* p - w */
U'', U'', U'', U'π', U'', U'£', U'·', /* x - ~ */
};
xassert(width == 1);
wc = vt100_0[wc - 0x60];
}
print_linewrap(term);
print_insert(term, width);
int col = grid->cursor.point.col;
if (unlikely(width > 1) && likely(term->auto_margin) &&
col + width > term->cols)
{
/* Multi-column character that doesn't fit on current line -
* pad with spacers */
for (size_t i = col; i < term->cols; i++)
print_spacer(term, i, 0);
/* And force a line-wrap */
grid->cursor.lcf = 1;
print_linewrap(term);
col = 0;
}
sixel_overwrite_at_cursor(term, width);
/* *Must* get current cell *after* linewrap+insert */
struct row *row = grid->cur_row;
row->dirty = true;
row->linebreak = true;
struct cell *cell = &row->cells[col];
cell->wc = term->vt.last_printed = wc;
cell->attrs = term->vt.attrs;
if (term->vt.osc8.uri != NULL) {
grid_row_uri_range_put(
row, col, term->vt.osc8.uri, term->vt.osc8.id);
switch (term->conf->url.osc8_underline) {
case OSC8_UNDERLINE_ALWAYS:
cell->attrs.url = true;
break;
case OSC8_UNDERLINE_URL_MODE:
break;
}
} else if (row->extra != NULL)
grid_row_uri_range_erase(row, col, col + width - 1);
if (unlikely(term->vt.underline.style > UNDERLINE_SINGLE ||
term->vt.underline.color_src != COLOR_DEFAULT))
{
grid_row_underline_range_put(row, col, term->vt.underline);
} else if (row->extra != NULL)
grid_row_underline_range_erase(row, col, col + width - 1);
/* Advance cursor the 'additional' columns while dirty:ing the cells */
for (int i = 1; i < width && (col + 1) < term->cols; i++) {
col++;
print_spacer(term, col, width - i);
}
xassert(col < term->cols);
/* Advance cursor */
if (unlikely(++col >= term->cols)) {
grid->cursor.lcf = true;
col--;
} else
xassert(!grid->cursor.lcf);
grid->cursor.point.col = col;
}
static void
ascii_printer_generic(struct terminal *term, char32_t wc)
{
term_print(term, wc, 1);
}
static void
ascii_printer_fast(struct terminal *term, char32_t wc)
{
struct grid *grid = term->grid;
xassert(term->charsets.set[term->charsets.selected] == CHARSET_ASCII);
xassert(!term->insert_mode);
xassert(tll_length(grid->sixel_images) == 0);
print_linewrap(term);
/* *Must* get current cell *after* linewrap+insert */
int col = grid->cursor.point.col;
const int uri_start = col;
struct row *row = grid->cur_row;
row->dirty = true;
row->linebreak = true;
struct cell *cell = &row->cells[col];
cell->wc = term->vt.last_printed = wc;
cell->attrs = term->vt.attrs;
/* Advance cursor */
if (unlikely(++col >= term->cols)) {
xassert(col == term->cols);
grid->cursor.lcf = true;
col--;
} else
xassert(!grid->cursor.lcf);
grid->cursor.point.col = col;
if (unlikely(row->extra != NULL)) {
grid_row_uri_range_erase(row, uri_start, uri_start);
grid_row_underline_range_erase(row, uri_start, uri_start);
}
}
static void
ascii_printer_single_shift(struct terminal *term, char32_t wc)
{
ascii_printer_generic(term, wc);
term->charsets.selected = term->charsets.saved;
term->bits_affecting_ascii_printer.charset =
term->charsets.set[term->charsets.selected] != CHARSET_ASCII;
term_update_ascii_printer(term);
}
void
term_update_ascii_printer(struct terminal *term)
{
_Static_assert(sizeof(term->bits_affecting_ascii_printer) == sizeof(uint8_t), "bad size");
void (*new_printer)(struct terminal *term, char32_t wc) =
unlikely(term->bits_affecting_ascii_printer.value != 0)
? &ascii_printer_generic
: &ascii_printer_fast;
#if defined(_DEBUG) && LOG_ENABLE_DBG
if (term->ascii_printer != new_printer) {
terminal: break out scaling factor updating, and reduce number of calls to render_resize() Break out the logic that updates the terminal’s scaling factor value, from render_resize(), to a new function, term_update_scale(). This allows us to update the scaling factor without a full grid resize. We also change how we pick the scaling factor (when fractional scaling is not in use). Before, we’d use the highest scaling factor from all monitors we were mapped on. Now, we use the scaling factor from the monitor we were *last* mapped on. Then, add a boolean parameter to term_set_fonts(), and when false, *don’t* call render_resize_force(). Also change term_font_dpi_changed() to only return true if the font was changed in any way. Finally, rewrite update_term_for_output_change() to: * Call term_update_scale() before doing anything else * Call render_resize{,_force} *last*, and *only* if either the scale or the fonts were updated. This fixes several things: * A bug where we failed to update the fonts when fractional scaling was in use, and we guessed the initial scale/DPI wrong. The bug happened because updated the internal "preferred" scale value, and a later call to render_resize() updated the terminal’s scale value, but since that code path didn’t call term_font_dpi_changed() (and it shouldn’t), the fonts weren’t resized properly. * It ensures we only resize the grid *once* when the scaling factor, or DPI is changed. Before this, we’d resize it twice. And this happened when e.g. dragging the window between monitors.
2023-07-17 16:21:16 +02:00
LOG_DBG("switching ASCII printer %s -> %s",
term->ascii_printer == &ascii_printer_fast ? "fast" : "generic",
new_printer == &ascii_printer_fast ? "fast" : "generic");
}
#endif
term->ascii_printer = new_printer;
}
void
term_single_shift(struct terminal *term, enum charset_designator idx)
{
term->charsets.saved = term->charsets.selected;
term->charsets.selected = idx;
term->ascii_printer = &ascii_printer_single_shift;
}
enum term_surface
term_surface_kind(const struct terminal *term, const struct wl_surface *surface)
{
if (likely(surface == term->window->surface.surf))
return TERM_SURF_GRID;
else if (surface == term->window->csd.surface[CSD_SURF_TITLE].surface.surf)
return TERM_SURF_TITLE;
else if (surface == term->window->csd.surface[CSD_SURF_LEFT].surface.surf)
return TERM_SURF_BORDER_LEFT;
else if (surface == term->window->csd.surface[CSD_SURF_RIGHT].surface.surf)
return TERM_SURF_BORDER_RIGHT;
else if (surface == term->window->csd.surface[CSD_SURF_TOP].surface.surf)
return TERM_SURF_BORDER_TOP;
else if (surface == term->window->csd.surface[CSD_SURF_BOTTOM].surface.surf)
return TERM_SURF_BORDER_BOTTOM;
else if (surface == term->window->csd.surface[CSD_SURF_MINIMIZE].surface.surf)
return TERM_SURF_BUTTON_MINIMIZE;
else if (surface == term->window->csd.surface[CSD_SURF_MAXIMIZE].surface.surf)
return TERM_SURF_BUTTON_MAXIMIZE;
else if (surface == term->window->csd.surface[CSD_SURF_CLOSE].surface.surf)
return TERM_SURF_BUTTON_CLOSE;
else
return TERM_SURF_NONE;
}
static bool
rows_to_text(const struct terminal *term, int start, int end,
int col_start, int col_end, char **text, size_t *len)
{
struct extraction_context *ctx = extract_begin(SELECTION_NONE, true);
if (ctx == NULL)
return false;
const int grid_rows = term->grid->num_rows;
int r = start;
while (true) {
const struct row *row = term->grid->rows[r];
xassert(row != NULL);
const int c_end = r == end ? col_end : term->cols;
for (int c = col_start; c < c_end; c++) {
if (!extract_one(term, row, &row->cells[c], c, ctx))
goto out;
}
if (r == end)
break;
r++;
r &= grid_rows - 1;
col_start = 0;
}
out:
return extract_finish(ctx, text, len);
}
bool
term_scrollback_to_text(const struct terminal *term, char **text, size_t *len)
{
const int grid_rows = term->grid->num_rows;
int start = (term->grid->offset + term->rows) & (grid_rows - 1);
int end = (term->grid->offset + term->rows - 1) & (grid_rows - 1);
xassert(start >= 0);
xassert(start < grid_rows);
xassert(end >= 0);
xassert(end < grid_rows);
/* If scrollback isn't full yet, this may be NULL, so scan forward
* until we find the first non-NULL row */
while (term->grid->rows[start] == NULL) {
start++;
start &= grid_rows - 1;
}
while (term->grid->rows[end] == NULL) {
end--;
if (end < 0)
end += term->grid->num_rows;
}
return rows_to_text(term, start, end, 0, term->cols, text, len);
}
bool
term_view_to_text(const struct terminal *term, char **text, size_t *len)
{
int start = grid_row_absolute_in_view(term->grid, 0);
int end = grid_row_absolute_in_view(term->grid, term->rows - 1);
return rows_to_text(term, start, end, 0, term->cols, text, len);
}
bool
term_command_output_to_text(const struct terminal *term, char **text, size_t *len)
{
int start_row = -1;
int end_row = -1;
int start_col = -1;
int end_col = -1;
const struct grid *grid = term->grid;
const int sb_end = grid_row_absolute(grid, term->rows - 1);
const int sb_start = (sb_end + 1) & (grid->num_rows - 1);
int r = sb_end;
while (start_row < 0) {
const struct row *row = grid->rows[r];
if (row == NULL)
break;
if (row->shell_integration.cmd_end >= 0) {
end_row = r;
end_col = row->shell_integration.cmd_end;
}
if (end_row >= 0 && row->shell_integration.cmd_start >= 0) {
start_row = r;
start_col = row->shell_integration.cmd_start;
}
if (r == sb_start)
break;
r = (r - 1 + grid->num_rows) & (grid->num_rows - 1);
}
if (start_row < 0)
return false;
bool ret = rows_to_text(term, start_row, end_row, start_col, end_col, text, len);
if (!ret)
return false;
/*
* If the FTCS_COMMAND_FINISHED marker was emitted at the *first*
* column, then the *entire* previous line is part of the command
* output. *Including* the newline, if any.
*
* Since rows_to_text() doesn't extract the column
* FTCS_COMMAND_FINISHED was emitted at (that would be wrong -
* FTCS_COMMAND_FINISHED is emitted *after* the command output,
* not at its last character), the extraction logic will not see
* the last newline (this is true for all non-line-wise selection
* types), and the extracted text will *not* end with a newline.
*
* Here we try to compensate for that. Note that if 'end_col' is
* not 0, then the command output only covers a partial row, and
* thus we do *not* want to append a newline.
*/
if (end_col > 0) {
/* Command output covers partial row - don't append newline */
return true;
}
int next_to_last_row = (end_row - 1 + grid->num_rows) & (grid->num_rows - 1);
const struct row *row = grid->rows[next_to_last_row];
/* Add newline if last row has a hard linebreak */
if (row->linebreak) {
char *new_text = xrealloc(*text, *len + 1 + 1);
if (new_text == NULL) {
/* Ignore failure - use text as is (without inserting newline) */
return true;
}
*text = new_text;
(*len)++;
(*text)[*len - 1] = '\n';
(*text)[*len] = '\0';
}
return true;
}
2020-12-03 18:36:56 +01:00
bool
term_ime_is_enabled(const struct terminal *term)
{
#if defined(FOOT_IME_ENABLED) && FOOT_IME_ENABLED
return term->ime_enabled;
#else
return false;
#endif
}
2020-12-03 18:36:56 +01:00
void
term_ime_enable(struct terminal *term)
2020-12-03 18:36:56 +01:00
{
#if defined(FOOT_IME_ENABLED) && FOOT_IME_ENABLED
if (term->ime_enabled)
return;
LOG_DBG("IME enabled");
term->ime_enabled = true;
2020-12-04 20:08:22 +01:00
/* IME is per seat - enable on all seat currently focusing us */
tll_foreach(term->wl->seats, it) {
if (it->item.kbd_focus == term)
ime_enable(&it->item);
}
#endif
}
void
term_ime_disable(struct terminal *term)
{
#if defined(FOOT_IME_ENABLED) && FOOT_IME_ENABLED
if (!term->ime_enabled)
return;
LOG_DBG("IME disabled");
term->ime_enabled = false;
/* IME is per seat - disable on all seat currently focusing us */
tll_foreach(term->wl->seats, it) {
if (it->item.kbd_focus == term)
ime_disable(&it->item);
}
#endif
}
bool
term_ime_reset(struct terminal *term)
{
bool at_least_one_seat_was_reset = false;
#if defined(FOOT_IME_ENABLED) && FOOT_IME_ENABLED
tll_foreach(term->wl->seats, it) {
struct seat *seat = &it->item;
if (seat->kbd_focus != term)
continue;
ime_reset_preedit(seat);
at_least_one_seat_was_reset = true;
}
2020-12-03 18:36:56 +01:00
#endif
return at_least_one_seat_was_reset;
2020-12-03 18:36:56 +01:00
}
void
term_ime_set_cursor_rect(struct terminal *term, int x, int y, int width,
int height)
{
#if defined(FOOT_IME_ENABLED) && FOOT_IME_ENABLED
tll_foreach(term->wl->seats, it) {
if (it->item.kbd_focus == term) {
it->item.ime.cursor_rect.pending.x = x;
it->item.ime.cursor_rect.pending.y = y;
it->item.ime.cursor_rect.pending.width = width;
it->item.ime.cursor_rect.pending.height = height;
}
}
#endif
}
void
term_osc8_open(struct terminal *term, uint64_t id, const char *uri)
{
term_osc8_close(term);
xassert(term->vt.osc8.uri == NULL);
term->vt.osc8.id = id;
term->vt.osc8.uri = xstrdup(uri);
term->bits_affecting_ascii_printer.osc8 = true;
term_update_ascii_printer(term);
}
void
term_osc8_close(struct terminal *term)
{
free(term->vt.osc8.uri);
term->vt.osc8.uri = NULL;
term->vt.osc8.id = 0;
term->bits_affecting_ascii_printer.osc8 = false;
term_update_ascii_printer(term);
}
void
term_set_user_mouse_cursor(struct terminal *term, const char *cursor)
{
free(term->mouse_user_cursor);
term->mouse_user_cursor = cursor != NULL && strlen(cursor) > 0
? xstrdup(cursor)
: NULL;
term_xcursor_update(term);
}
void
term_enable_size_notifications(struct terminal *term)
{
/* Note: always send current size upon activation, regardless of
previous state */
term->size_notifications = true;
term_send_size_notification(term);
}
void
term_disable_size_notifications(struct terminal *term)
{
if (!term->size_notifications)
return;
term->size_notifications = false;
}
void
term_send_size_notification(struct terminal *term)
{
if (!term->size_notifications)
return;
const int height = term->height - term->margins.top - term->margins.bottom;
const int width = term->width - term->margins.left - term->margins.right;
char buf[128];
const int n = xsnprintf(
buf, sizeof(buf), "\033[48;%d;%d;%d;%dt",
term->rows, term->cols, height, width);
term_to_slave(term, buf, n);
}