gtk/gsk/gskpath.c

830 lines
22 KiB
C
Raw Normal View History

/*
* Copyright © 2020 Benjamin Otte
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library. If not, see <http://www.gnu.org/licenses/>.
*
* Authors: Benjamin Otte <otte@gnome.org>
*/
#include "config.h"
#include "gskpathprivate.h"
#include "gskcurveprivate.h"
#include "gskpathbuilder.h"
#include "gskpathpoint.h"
#include "gskcontourprivate.h"
/**
* GskPath:
*
* A `GskPath` describes lines and curves that are more complex
* than simple rectangles.
*
* Paths can used for rendering (filling or stroking) and for animations
* (e.g. as trajectories).
*
* `GskPath` is an immutable, opaque, reference-counted struct.
* After creation, you cannot change the types it represents. Instead,
* new `GskPath` objects have to be created. The [struct@Gsk.PathBuilder]
* structure is meant to help in this endeavor.
*
* Conceptually, a path consists of zero or more contours (continuous, connected
* curves), each of which may or may not be closed. Contours are typically
* constructed from Bézier segments.
*
* <picture>
* <source srcset="path-dark.png" media="(prefers-color-scheme: dark)">
* <img alt="A Path" src="path-light.png">
* </picture>
*
* Since: 4.14
*/
struct _GskPath
{
/*< private >*/
guint ref_count;
GskPathFlags flags;
gsize n_contours;
GskContour *contours[];
/* followed by the contours data */
};
G_DEFINE_BOXED_TYPE (GskPath, gsk_path, gsk_path_ref, gsk_path_unref)
/* {{{ Private API */
GskPath *
gsk_path_new_from_contours (const GSList *contours)
{
GskPath *path;
const GSList *l;
gsize size;
gsize n_contours;
guint8 *contour_data;
GskPathFlags flags;
flags = GSK_PATH_CLOSED | GSK_PATH_FLAT;
size = 0;
n_contours = 0;
for (l = contours; l; l = l->next)
{
GskContour *contour = l->data;
n_contours++;
size += sizeof (GskContour *);
size += gsk_contour_get_size (contour);
flags &= gsk_contour_get_flags (contour);
}
path = g_malloc0 (sizeof (GskPath) + size);
path->ref_count = 1;
path->flags = flags;
path->n_contours = n_contours;
contour_data = (guint8 *) &path->contours[n_contours];
n_contours = 0;
for (l = contours; l; l = l->next)
{
GskContour *contour = l->data;
path->contours[n_contours] = (GskContour *) contour_data;
gsk_contour_copy ((GskContour *) contour_data, contour);
size = gsk_contour_get_size (contour);
contour_data += size;
n_contours++;
}
return path;
}
const GskContour *
gsk_path_get_contour (const GskPath *self,
gsize i)
{
if (i < self->n_contours)
return self->contours[i];
else
return NULL;
}
GskPathFlags
gsk_path_get_flags (const GskPath *self)
{
return self->flags;
}
gsize
gsk_path_get_n_contours (const GskPath *self)
{
return self->n_contours;
}
/* }}} */
/* {{{ Public API */
/**
* gsk_path_ref:
* @self: a `GskPath`
*
* Increases the reference count of a `GskPath` by one.
*
* Returns: the passed in `GskPath`.
*
* Since: 4.14
*/
GskPath *
gsk_path_ref (GskPath *self)
{
g_return_val_if_fail (self != NULL, NULL);
self->ref_count++;
return self;
}
/**
* gsk_path_unref:
* @self: a `GskPath`
*
* Decreases the reference count of a `GskPath` by one.
*
* If the resulting reference count is zero, frees the path.
*
* Since: 4.14
*/
void
gsk_path_unref (GskPath *self)
{
g_return_if_fail (self != NULL);
g_return_if_fail (self->ref_count > 0);
self->ref_count--;
if (self->ref_count > 0)
return;
g_free (self);
}
/**
* gsk_path_print:
* @self: a `GskPath`
* @string: The string to print into
*
* Converts @self into a human-readable string representation suitable
* for printing.
*
2023-08-26 02:21:07 +00:00
* The string is compatible with (a superset of)
* [SVG path syntax](https://www.w3.org/TR/SVG11/paths.html#PathData),
* see [func@Gsk.Path.parse] for a summary of the syntax.
*
* Since: 4.14
*/
void
gsk_path_print (GskPath *self,
GString *string)
{
gsize i;
g_return_if_fail (self != NULL);
g_return_if_fail (string != NULL);
for (i = 0; i < self->n_contours; i++)
{
if (i > 0)
g_string_append_c (string, ' ');
gsk_contour_print (self->contours[i], string);
}
}
/**
* gsk_path_to_string:
* @self: a `GskPath`
*
* Converts the path into a string that is suitable for printing.
*
* You can use this function in a debugger to get a quick overview
* of the path.
*
* This is a wrapper around [method@Gsk.Path.print], see that function
* for details.
*
* Returns: A new string for @self
*
* Since: 4.14
*/
char *
gsk_path_to_string (GskPath *self)
{
GString *string;
g_return_val_if_fail (self != NULL, NULL);
string = g_string_new ("");
gsk_path_print (self, string);
return g_string_free (string, FALSE);
}
static gboolean
gsk_path_to_cairo_add_op (GskPathOperation op,
const graphene_point_t *pts,
gsize n_pts,
float weight,
gpointer cr)
{
switch (op)
{
case GSK_PATH_MOVE:
cairo_move_to (cr, pts[0].x, pts[0].y);
break;
case GSK_PATH_CLOSE:
cairo_close_path (cr);
break;
case GSK_PATH_LINE:
cairo_line_to (cr, pts[1].x, pts[1].y);
break;
case GSK_PATH_CUBIC:
cairo_curve_to (cr, pts[1].x, pts[1].y, pts[2].x, pts[2].y, pts[3].x, pts[3].y);
break;
case GSK_PATH_QUAD:
case GSK_PATH_CONIC:
default:
g_assert_not_reached ();
return FALSE;
}
return TRUE;
}
/**
* gsk_path_to_cairo:
* @self: a `GskPath`
* @cr: a cairo context
*
* Appends the given @path to the given cairo context for drawing
* with Cairo.
*
* This may cause some suboptimal conversions to be performed as
* Cairo does not support all features of `GskPath`.
*
* This function does not clear the existing Cairo path. Call
* cairo_new_path() if you want this.
*
* Since: 4.14
*/
void
gsk_path_to_cairo (GskPath *self,
cairo_t *cr)
{
g_return_if_fail (self != NULL);
g_return_if_fail (cr != NULL);
gsk_path_foreach_with_tolerance (self,
GSK_PATH_FOREACH_ALLOW_CUBIC,
cairo_get_tolerance (cr),
gsk_path_to_cairo_add_op,
cr);
}
/**
* gsk_path_is_empty:
* @self: a `GskPath`
*
* Checks if the path is empty, i.e. contains no lines or curves.
*
* Returns: `TRUE` if the path is empty
*
* Since: 4.14
*/
gboolean
gsk_path_is_empty (GskPath *self)
{
g_return_val_if_fail (self != NULL, FALSE);
return self->n_contours == 0;
}
/**
* gsk_path_is_closed:
* @self: a `GskPath`
*
* Returns if the path represents a single closed
* contour.
*
* Returns: `TRUE` if the path is closed
*
* Since: 4.14
*/
gboolean
gsk_path_is_closed (GskPath *self)
{
g_return_val_if_fail (self != NULL, FALSE);
/* XXX: is the empty path closed? Currently it's not */
if (self->n_contours != 1)
return FALSE;
return gsk_contour_get_flags (self->contours[0]) & GSK_PATH_CLOSED ? TRUE : FALSE;
}
/**
* gsk_path_get_bounds:
* @self: a `GskPath`
* @bounds: (out caller-allocates): the bounds of the given path
*
* Computes the bounds of the given path.
*
* The returned bounds may be larger than necessary, because this
* function aims to be fast, not accurate. The bounds are guaranteed
* to contain the path.
*
* It is possible that the returned rectangle has 0 width and/or height.
* This can happen when the path only describes a point or an
* axis-aligned line.
*
* If the path is empty, `FALSE` is returned and @bounds are set to
* graphene_rect_zero(). This is different from the case where the path
* is a single point at the origin, where the @bounds will also be set to
2023-08-24 18:32:38 +00:00
* the zero rectangle but `TRUE` will be returned.
*
* Returns: `TRUE` if the path has bounds, `FALSE` if the path is known
* to be empty and have no bounds.
*
* Since: 4.14
*/
gboolean
gsk_path_get_bounds (GskPath *self,
graphene_rect_t *bounds)
{
GskBoundingBox b;
g_return_val_if_fail (self != NULL, FALSE);
g_return_val_if_fail (bounds != NULL, FALSE);
if (self->n_contours == 0)
{
graphene_rect_init_from_rect (bounds, graphene_rect_zero ());
return FALSE;
}
gsk_contour_get_bounds (self->contours[0], &b);
for (gsize i = 1; i < self->n_contours; i++)
{
GskBoundingBox tmp;
gsk_contour_get_bounds (self->contours[i], &tmp);
gsk_bounding_box_union (&b, &tmp, &b);
}
gsk_bounding_box_to_rect (&b, bounds);
return TRUE;
}
/**
* gsk_path_get_stroke_bounds:
* @self: a #GtkPath
* @stroke: stroke parameters
* @bounds: (out caller-allocates): the bounds to fill in
*
* Computes the bounds for stroking the given path with the
* parameters in @stroke.
*
* The returned bounds may be larger than necessary, because this
* function aims to be fast, not accurate. The bounds are guaranteed
* to contain the area affected by the stroke, including protrusions
* like miters.
*
* Returns: `TRUE` if the path has bounds, `FALSE` if the path is known
* to be empty and have no bounds.
2023-08-26 02:21:07 +00:00
*
* Since: 4.14
*/
gboolean
gsk_path_get_stroke_bounds (GskPath *self,
const GskStroke *stroke,
graphene_rect_t *bounds)
{
GskBoundingBox b;
g_return_val_if_fail (self != NULL, FALSE);
g_return_val_if_fail (bounds != NULL, FALSE);
if (self->n_contours == 0)
{
graphene_rect_init_from_rect (bounds, graphene_rect_zero ());
return FALSE;
}
gsk_contour_get_stroke_bounds (self->contours[0], stroke, &b);
for (gsize i = 1; i < self->n_contours; i++)
{
GskBoundingBox tmp;
if (gsk_contour_get_stroke_bounds (self->contours[i], stroke, &tmp))
gsk_bounding_box_union (&b, &tmp, &b);
}
gsk_bounding_box_to_rect (&b, bounds);
return TRUE;
}
/**
* gsk_path_in_fill:
* @self: a `GskPath`
* @point: the point to test
* @fill_rule: the fill rule to follow
*
* Returns whether the given point is inside the area
* that would be affected if the path was filled according
* to @fill_rule.
*
* Note that this function assumes that filling a contour
* implicitly closes it.
*
* Returns: `TRUE` if @point is inside
*
* Since: 4.14
*/
gboolean
gsk_path_in_fill (GskPath *self,
const graphene_point_t *point,
GskFillRule fill_rule)
{
int winding = 0;
for (int i = 0; i < self->n_contours; i++)
winding += gsk_contour_get_winding (self->contours[i], point);
switch (fill_rule)
{
case GSK_FILL_RULE_EVEN_ODD:
return winding & 1;
case GSK_FILL_RULE_WINDING:
return winding != 0;
default:
g_assert_not_reached ();
}
}
/**
* gsk_path_get_start_point:
* @self: a `GskPath`
* @result: (out caller-allocates): return location for point
*
* Gets the start point of the path.
*
* An empty path has no points, so `FALSE`
* is returned in this case.
*
* Returns: `TRUE` if @result was filled
*
* Since: 4.14
*/
gboolean
gsk_path_get_start_point (GskPath *self,
GskPathPoint *result)
{
g_return_val_if_fail (self != NULL, FALSE);
g_return_val_if_fail (result != NULL, FALSE);
if (self->n_contours == 0)
return FALSE;
2023-08-27 14:10:08 +00:00
/* Conceptually, there is always a move at the
* beginning, which jumps from where to the start
* point of the contour, so we use idx == 1 here.
*/
result->contour = 0;
result->idx = 1;
result->t = 0;
return TRUE;
}
/**
* gsk_path_get_end_point:
* @self: a `GskPath`
* @result: (out caller-allocates): return location for point
*
* Gets the end point of the path.
*
* An empty path has no points, so `FALSE`
* is returned in this case.
*
* Returns: `TRUE` if @result was filled
*
* Since: 4.14
*/
gboolean
gsk_path_get_end_point (GskPath *self,
GskPathPoint *result)
{
g_return_val_if_fail (self != NULL, FALSE);
g_return_val_if_fail (result != NULL, FALSE);
if (self->n_contours == 0)
return FALSE;
result->contour = self->n_contours - 1;
result->idx = gsk_contour_get_n_ops (self->contours[self->n_contours - 1]) - 1;
result->t = 1;
return TRUE;
}
/**
* gsk_path_get_closest_point:
* @self: a `GskPath`
* @point: the point
* @threshold: maximum allowed distance
* @result: (out caller-allocates): return location for the closest point
* @distance: (out) (optional): return location for the distance
*
* Computes the closest point on the path to the given point
* and sets the @result to it.
*
* If there is no point closer than the given threshold,
* `FALSE` is returned.
*
* Returns: `TRUE` if @point was set to the closest point
* on @self, `FALSE` if no point is closer than @threshold
*
* Since: 4.14
*/
gboolean
gsk_path_get_closest_point (GskPath *self,
const graphene_point_t *point,
float threshold,
GskPathPoint *result,
float *distance)
{
gboolean found;
g_return_val_if_fail (self != NULL, FALSE);
g_return_val_if_fail (point != NULL, FALSE);
g_return_val_if_fail (threshold >= 0, FALSE);
g_return_val_if_fail (result != NULL, FALSE);
found = FALSE;
for (int i = 0; i < self->n_contours; i++)
{
float dist;
if (gsk_contour_get_closest_point (self->contours[i], point, threshold, result, &dist))
{
found = TRUE;
g_assert (0 <= result->t && result->t <= 1);
result->contour = i;
threshold = dist;
if (distance)
*distance = dist;
}
}
return found;
}
/* }}} */
/* {{{ Foreach and decomposition */
/**
* gsk_path_foreach:
* @self: a `GskPath`
* @flags: flags to pass to the foreach function. See [flags@Gsk.PathForeachFlags]
* for details about flags
* @func: (scope call) (closure user_data): the function to call for operations
* @user_data: (nullable): user data passed to @func
*
* Calls @func for every operation of the path.
*
2023-08-24 18:32:38 +00:00
* Note that this may only approximate @self, because paths can contain
* optimizations for various specialized contours, and depending on the
* @flags, the path may be decomposed into simpler curves than the ones
* that it contained originally.
*
* This function serves two purposes:
*
* - When the @flags allow everything, it provides access to the raw,
* unmodified data of the path.
* - When the @flags disallow certain operations, it provides
* an approximation of the path using just the allowed operations.
*
* Returns: `FALSE` if @func returned FALSE`, `TRUE` otherwise.
*
* Since: 4.14
*/
gboolean
gsk_path_foreach (GskPath *self,
GskPathForeachFlags flags,
GskPathForeachFunc func,
gpointer user_data)
{
g_return_val_if_fail (self != NULL, FALSE);
g_return_val_if_fail (func, FALSE);
return gsk_path_foreach_with_tolerance (self,
flags,
GSK_PATH_TOLERANCE_DEFAULT,
func,
user_data);
}
typedef struct _GskPathForeachTrampoline GskPathForeachTrampoline;
struct _GskPathForeachTrampoline
{
GskPathForeachFlags flags;
double tolerance;
GskPathForeachFunc func;
gpointer user_data;
};
static gboolean
gsk_path_foreach_trampoline_add_line (const graphene_point_t *from,
const graphene_point_t *to,
float from_progress,
float to_progress,
GskCurveLineReason reason,
gpointer data)
{
GskPathForeachTrampoline *trampoline = data;
return trampoline->func (GSK_PATH_LINE,
(graphene_point_t[2]) { *from, *to },
2,
0.f,
trampoline->user_data);
}
static gboolean
gsk_path_foreach_trampoline_add_curve (GskPathOperation op,
const graphene_point_t *pts,
gsize n_pts,
float weight,
gpointer data)
{
GskPathForeachTrampoline *trampoline = data;
return trampoline->func (op, pts, n_pts, weight, trampoline->user_data);
}
static gboolean
gsk_path_foreach_trampoline (GskPathOperation op,
const graphene_point_t *pts,
gsize n_pts,
float weight,
gpointer data)
{
GskPathForeachTrampoline *trampoline = data;
gskpathop: Introduce a type to represent an aligned graphene_point_t When we allocate a graphene_point_t on the stack, there's no guarantee that it will be aligned at an 8-byte boundary, which is an assumption made by gsk_pathop_encode() (which wants to use the lowest 3 bits to encode the operation). In the places where it matters, force the points on the stack and embedded in structs to be nicely aligned. By using a distinct type for this (a union with a suitable size and alignment), we ensure that the compiler will warn or error whenever we can't prove that a particular point is, in fact, suitably aligned. We can go from a `GskAlignedPoint *` to a `graphene_point_t *` (which is always valid, because the `GskAlignedPoint` is aligned) via &aligned_points[0].pt, but we cannot go back the other way (which is not always valid, because the `graphene_point_t` is not necessarily aligned nicely) without a cast. In practice, it seems that a graphene_point_t on x86_64 *is* usually placed at an 8-byte boundary, but this is not the case on 32-bit architectures or on s390x. In many cases we can avoid needing an explicit reference to the more complicated type by making use of a transparent union. There's already at least one transparent union in GSK's public API, so it's presumably portable enough to match GTK's requirements. Increasing the alignment of GskAlignedPoint also requires adjusting how a GskStandardContour is allocated and initialized. This data structure allocates extra memory to hold an array of GskAlignedPoint outside the bounds of the struct itself, and that array now needs to be aligned suitably. Previously the array started with at next byte after the flexible array of gskpathop, but the alignment of a gskpathop is only 4 bytes on 32-bit architectures, so depending on the number of gskpathop in the trailing flexible array, that pointer might be an unsuitable location to allocate a GskAlignedPoint. Resolves: https://gitlab.gnome.org/GNOME/gtk/-/issues/6395 Signed-off-by: Simon McVittie <smcv@debian.org>
2024-07-28 13:42:00 +00:00
GskAlignedPoint *aligned = g_alloca (sizeof (graphene_point_t) * n_pts);
/* We can't necessarily guarantee that pts is 8-byte aligned
* (probably it is, but we've been through too many layers of
* indirection to be sure) so copy it into a buffer that is
* definitely suitably-aligned. */
memcpy (aligned, pts, sizeof (graphene_point_t) * n_pts);
switch (op)
{
case GSK_PATH_MOVE:
case GSK_PATH_CLOSE:
case GSK_PATH_LINE:
return trampoline->func (op, pts, n_pts, weight, trampoline->user_data);
case GSK_PATH_QUAD:
{
GskCurve curve;
if (trampoline->flags & GSK_PATH_FOREACH_ALLOW_QUAD)
return trampoline->func (op, pts, n_pts, weight, trampoline->user_data);
else if (trampoline->flags & GSK_PATH_FOREACH_ALLOW_CUBIC)
{
return trampoline->func (GSK_PATH_CUBIC,
(graphene_point_t[4]) {
pts[0],
GRAPHENE_POINT_INIT ((pts[0].x + 2 * pts[1].x) / 3,
(pts[0].y + 2 * pts[1].y) / 3),
GRAPHENE_POINT_INIT ((pts[2].x + 2 * pts[1].x) / 3,
(pts[2].y + 2 * pts[1].y) / 3),
pts[2],
},
4,
weight,
trampoline->user_data);
}
gskpathop: Introduce a type to represent an aligned graphene_point_t When we allocate a graphene_point_t on the stack, there's no guarantee that it will be aligned at an 8-byte boundary, which is an assumption made by gsk_pathop_encode() (which wants to use the lowest 3 bits to encode the operation). In the places where it matters, force the points on the stack and embedded in structs to be nicely aligned. By using a distinct type for this (a union with a suitable size and alignment), we ensure that the compiler will warn or error whenever we can't prove that a particular point is, in fact, suitably aligned. We can go from a `GskAlignedPoint *` to a `graphene_point_t *` (which is always valid, because the `GskAlignedPoint` is aligned) via &aligned_points[0].pt, but we cannot go back the other way (which is not always valid, because the `graphene_point_t` is not necessarily aligned nicely) without a cast. In practice, it seems that a graphene_point_t on x86_64 *is* usually placed at an 8-byte boundary, but this is not the case on 32-bit architectures or on s390x. In many cases we can avoid needing an explicit reference to the more complicated type by making use of a transparent union. There's already at least one transparent union in GSK's public API, so it's presumably portable enough to match GTK's requirements. Increasing the alignment of GskAlignedPoint also requires adjusting how a GskStandardContour is allocated and initialized. This data structure allocates extra memory to hold an array of GskAlignedPoint outside the bounds of the struct itself, and that array now needs to be aligned suitably. Previously the array started with at next byte after the flexible array of gskpathop, but the alignment of a gskpathop is only 4 bytes on 32-bit architectures, so depending on the number of gskpathop in the trailing flexible array, that pointer might be an unsuitable location to allocate a GskAlignedPoint. Resolves: https://gitlab.gnome.org/GNOME/gtk/-/issues/6395 Signed-off-by: Simon McVittie <smcv@debian.org>
2024-07-28 13:42:00 +00:00
gsk_curve_init (&curve, gsk_pathop_encode (GSK_PATH_QUAD, aligned));
return gsk_curve_decompose (&curve,
trampoline->tolerance,
gsk_path_foreach_trampoline_add_line,
trampoline);
}
case GSK_PATH_CUBIC:
{
GskCurve curve;
if (trampoline->flags & GSK_PATH_FOREACH_ALLOW_CUBIC)
return trampoline->func (op, pts, n_pts, weight, trampoline->user_data);
gskpathop: Introduce a type to represent an aligned graphene_point_t When we allocate a graphene_point_t on the stack, there's no guarantee that it will be aligned at an 8-byte boundary, which is an assumption made by gsk_pathop_encode() (which wants to use the lowest 3 bits to encode the operation). In the places where it matters, force the points on the stack and embedded in structs to be nicely aligned. By using a distinct type for this (a union with a suitable size and alignment), we ensure that the compiler will warn or error whenever we can't prove that a particular point is, in fact, suitably aligned. We can go from a `GskAlignedPoint *` to a `graphene_point_t *` (which is always valid, because the `GskAlignedPoint` is aligned) via &aligned_points[0].pt, but we cannot go back the other way (which is not always valid, because the `graphene_point_t` is not necessarily aligned nicely) without a cast. In practice, it seems that a graphene_point_t on x86_64 *is* usually placed at an 8-byte boundary, but this is not the case on 32-bit architectures or on s390x. In many cases we can avoid needing an explicit reference to the more complicated type by making use of a transparent union. There's already at least one transparent union in GSK's public API, so it's presumably portable enough to match GTK's requirements. Increasing the alignment of GskAlignedPoint also requires adjusting how a GskStandardContour is allocated and initialized. This data structure allocates extra memory to hold an array of GskAlignedPoint outside the bounds of the struct itself, and that array now needs to be aligned suitably. Previously the array started with at next byte after the flexible array of gskpathop, but the alignment of a gskpathop is only 4 bytes on 32-bit architectures, so depending on the number of gskpathop in the trailing flexible array, that pointer might be an unsuitable location to allocate a GskAlignedPoint. Resolves: https://gitlab.gnome.org/GNOME/gtk/-/issues/6395 Signed-off-by: Simon McVittie <smcv@debian.org>
2024-07-28 13:42:00 +00:00
gsk_curve_init (&curve, gsk_pathop_encode (GSK_PATH_CUBIC, aligned));
if (trampoline->flags & (GSK_PATH_FOREACH_ALLOW_QUAD|GSK_PATH_FOREACH_ALLOW_CONIC))
return gsk_curve_decompose_curve (&curve,
trampoline->flags,
trampoline->tolerance,
gsk_path_foreach_trampoline_add_curve,
trampoline);
return gsk_curve_decompose (&curve,
trampoline->tolerance,
gsk_path_foreach_trampoline_add_line,
trampoline);
}
case GSK_PATH_CONIC:
{
GskCurve curve;
if (trampoline->flags & GSK_PATH_FOREACH_ALLOW_CONIC)
return trampoline->func (op, pts, n_pts, weight, trampoline->user_data);
gskpathop: Introduce a type to represent an aligned graphene_point_t When we allocate a graphene_point_t on the stack, there's no guarantee that it will be aligned at an 8-byte boundary, which is an assumption made by gsk_pathop_encode() (which wants to use the lowest 3 bits to encode the operation). In the places where it matters, force the points on the stack and embedded in structs to be nicely aligned. By using a distinct type for this (a union with a suitable size and alignment), we ensure that the compiler will warn or error whenever we can't prove that a particular point is, in fact, suitably aligned. We can go from a `GskAlignedPoint *` to a `graphene_point_t *` (which is always valid, because the `GskAlignedPoint` is aligned) via &aligned_points[0].pt, but we cannot go back the other way (which is not always valid, because the `graphene_point_t` is not necessarily aligned nicely) without a cast. In practice, it seems that a graphene_point_t on x86_64 *is* usually placed at an 8-byte boundary, but this is not the case on 32-bit architectures or on s390x. In many cases we can avoid needing an explicit reference to the more complicated type by making use of a transparent union. There's already at least one transparent union in GSK's public API, so it's presumably portable enough to match GTK's requirements. Increasing the alignment of GskAlignedPoint also requires adjusting how a GskStandardContour is allocated and initialized. This data structure allocates extra memory to hold an array of GskAlignedPoint outside the bounds of the struct itself, and that array now needs to be aligned suitably. Previously the array started with at next byte after the flexible array of gskpathop, but the alignment of a gskpathop is only 4 bytes on 32-bit architectures, so depending on the number of gskpathop in the trailing flexible array, that pointer might be an unsuitable location to allocate a GskAlignedPoint. Resolves: https://gitlab.gnome.org/GNOME/gtk/-/issues/6395 Signed-off-by: Simon McVittie <smcv@debian.org>
2024-07-28 13:42:00 +00:00
gsk_curve_init (&curve, gsk_pathop_encode (GSK_PATH_CONIC, (GskAlignedPoint[4]) { { pts[0] }, { pts[1] }, { { weight, 0.f } }, { pts[2] } } ));
if (trampoline->flags & (GSK_PATH_FOREACH_ALLOW_QUAD|GSK_PATH_FOREACH_ALLOW_CUBIC))
return gsk_curve_decompose_curve (&curve,
trampoline->flags,
trampoline->tolerance,
gsk_path_foreach_trampoline_add_curve,
trampoline);
return gsk_curve_decompose (&curve,
trampoline->tolerance,
gsk_path_foreach_trampoline_add_line,
trampoline);
}
default:
g_assert_not_reached ();
return FALSE;
}
}
#define ALLOW_ANY (GSK_PATH_FOREACH_ALLOW_QUAD | \
GSK_PATH_FOREACH_ALLOW_CUBIC | \
GSK_PATH_FOREACH_ALLOW_CONIC)
gboolean
gsk_path_foreach_with_tolerance (GskPath *self,
GskPathForeachFlags flags,
double tolerance,
GskPathForeachFunc func,
gpointer user_data)
{
GskPathForeachTrampoline trampoline;
gsize i;
/* If we need to massage the data, set up a trampoline here */
if ((flags & ALLOW_ANY) != ALLOW_ANY)
{
trampoline = (GskPathForeachTrampoline) { flags, tolerance, func, user_data };
func = gsk_path_foreach_trampoline;
user_data = &trampoline;
}
for (i = 0; i < self->n_contours; i++)
{
if (!gsk_contour_foreach (self->contours[i], func, user_data))
return FALSE;
}
return TRUE;
}
/* }}} */
/* vim:set foldmethod=marker expandtab: */