--- /dev/null
+/* heap.h
+
+ An inline heap implementation
+
+ Copyright (c) 2009 Matthias Kramm <kramm@quiss.org>
+
+ This program is free software; you can redistribute it and/or modify
+ it under the terms of the GNU General Public License as published by
+ the Free Software Foundation; either version 2 of the License, or
+ (at your option) any later version.
+
+ This program 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 General Public License for more details.
+
+ You should have received a copy of the GNU General Public License
+ along with this program; if not, write to the Free Software
+ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */
+
+#define HEAP_DEFINE(name,t,lt) \
+typedef struct { \
+ t**elements; \
+ int size; \
+ int max_size; \
+} name##_t; \
+static void name##_put(name##_t*h, t*e) \
+{ \
+ int parent = h->size++; \
+ if(h->size>=h->max_size) { \
+ h->max_size = h->max_size<15?15:(h->max_size+1)*2-1; \
+ h->elements = (t**)realloc(h->elements, \
+ h->max_size*sizeof(t*)); \
+ } \
+ int node; \
+ do { \
+ node = parent; \
+ if(!node) break; \
+ parent = (node-1)/2; \
+ h->elements[node] = h->elements[parent]; \
+ } while(lt(e, h->elements[parent])); \
+ h->elements[node] = e; \
+} \
+static t* name##_get(name##_t*h) \
+{ \
+ if(!h->size) return 0; \
+ t*r = h->elements[0]; \
+ int node,child = 0; \
+ t*node_p = h->elements[--h->size]; \
+ h->elements[0] = node_p; /* for the size==1 case */ \
+ do { \
+ node = child; \
+ child = node<<1|1; \
+ if(child >= h->size) { \
+ break; \
+ } \
+ if(child+1 < h->size && lt(h->elements[child+1], \
+ h->elements[child])) \
+ child++; \
+ h->elements[node] = h->elements[child]; \
+ } while(lt(h->elements[child],node_p)); \
+ h->elements[node] = node_p; \
+ return r; \
+} \
+static name##_init(name##_t*h) \
+{ \
+ memset(h, 0, sizeof(*h)); \
+ h->max_size = 15; \
+ h->elements = malloc(h->max_size*sizeof(t*)); \
+} \
+static name##_destroy(name##_t*h) \
+{ \
+ free((h)->elements); \
+}
#include "xrow.h"
#include "wind.h"
#include "convert.h"
+#include "heap.h"
static gfxpoly_t*current_polygon = 0;
void gfxpoly_fail(char*expr, char*file, int line, const char*function)
free: point_free,
};
-typedef struct _status {
- int32_t y;
- actlist_t*actlist;
- heap_t*queue;
- xrow_t*xrow;
- windrule_t*windrule;
- windcontext_t*context;
- segment_t*ending_segments;
- polywriter_t writer;
-#ifdef CHECKS
- dict_t*seen_crossings; //list of crossing we saw so far
- dict_t*intersecting_segs; //list of segments intersecting in this scanline
- dict_t*segs_with_point; //lists of segments that received a point in this scanline
-#endif
-} status_t;
-
typedef struct _event {
eventtype_t type;
point_t p;
/* compare_events_simple differs from compare_events in that it schedules
events from left to right regardless of type. It's only used in horizontal
processing, in order to get an x-wise sorting of the current scanline */
-static int compare_events_simple(const void*_a,const void*_b)
+static inline int compare_events_simple(const void*_a,const void*_b)
{
event_t* a = (event_t*)_a;
event_t* b = (event_t*)_b;
return 0;
}
-static int compare_events(const void*_a,const void*_b)
+static inline int compare_events(const void*_a,const void*_b)
{
event_t* a = (event_t*)_a;
event_t* b = (event_t*)_b;
//return d;
}
+#define COMPARE_EVENTS(x,y) (compare_events(x,y)>0)
+#define COMPARE_EVENTS_SIMPLE(x,y) (compare_events_simple(x,y)>0)
+HEAP_DEFINE(queue,event_t,COMPARE_EVENTS);
+HEAP_DEFINE(hqueue,event_t,COMPARE_EVENTS_SIMPLE);
+
+typedef struct _status {
+ int32_t y;
+ actlist_t*actlist;
+ queue_t queue;
+ xrow_t*xrow;
+ windrule_t*windrule;
+ windcontext_t*context;
+ segment_t*ending_segments;
+ polywriter_t writer;
+#ifdef CHECKS
+ dict_t*seen_crossings; //list of crossing we saw so far
+ dict_t*intersecting_segs; //list of segments intersecting in this scanline
+ dict_t*segs_with_point; //lists of segments that received a point in this scanline
+#endif
+} status_t;
+
+
int gfxpoly_size(gfxpoly_t*poly)
{
int s,t;
fclose(fi);
}
-inline static event_t event_new()
+inline static event_t* event_new()
{
- event_t e;
- memset(&e, 0, sizeof(e));
+ event_t*e = rfx_calloc(sizeof(event_t));
return e;
}
+inline static void event_free(event_t*e)
+{
+ free(e);
+}
static void event_dump(event_t*e)
{
assert(LINE_EQ(p, s) >= 0);
#endif
- /* TODO: make this int_type */
+#ifndef DONT_REMEMBER_CROSSINGS
dict_init2(&s->scheduled_crossings, &ptr_type, 0);
+#endif
}
static segment_t* segment_new(point_t a, point_t b, int polygon_nr, segment_dir_t dir)
static void segment_clear(segment_t*s)
{
+#ifndef DONT_REMEMBER_CROSSINGS
dict_clear(&s->scheduled_crossings);
+#endif
}
static void segment_destroy(segment_t*s)
{
free(s);
}
-static void advance_stroke(heap_t*queue, gfxpolystroke_t*stroke, int polygon_nr, int pos)
+static void advance_stroke(queue_t*queue, hqueue_t*hqueue, gfxpolystroke_t*stroke, int polygon_nr, int pos)
{
if(!stroke)
return;
s->nr, s->a.x, s->a.y, s->b.x, s->b.y,
s->dir==DIR_UP?"up":"down", stroke, stroke->num_points - 1 - pos);
#endif
- event_t e = event_new();
- e.type = s->delta.y ? EVENT_START : EVENT_HORIZONTAL;
- e.p = s->a;
- e.s1 = s;
- e.s2 = 0;
- heap_put(queue, &e);
- if(e.type != EVENT_HORIZONTAL) {
+ event_t* e = event_new();
+ e->type = s->delta.y ? EVENT_START : EVENT_HORIZONTAL;
+ e->p = s->a;
+ e->s1 = s;
+ e->s2 = 0;
+
+ if(queue) queue_put(queue, e);
+ else hqueue_put(hqueue, e);
+
+ if(e->type != EVENT_HORIZONTAL) {
break;
}
}
}
}
-static void gfxpoly_enqueue(gfxpoly_t*p, heap_t*queue, int polygon_nr)
+static void gfxpoly_enqueue(gfxpoly_t*p, queue_t*queue, hqueue_t*hqueue, int polygon_nr)
{
int t;
gfxpolystroke_t*stroke = p->strokes;
assert(stroke->points[s].y <= stroke->points[s+1].y);
}
#endif
- advance_stroke(queue, stroke, polygon_nr, 0);
+ advance_stroke(queue, hqueue, stroke, polygon_nr, 0);
}
}
{
// schedule end point of segment
assert(s->b.y > status->y);
- event_t e;
- e.type = EVENT_END;
- e.p = s->b;
- e.s1 = s;
- e.s2 = 0;
- heap_put(status->queue, &e);
+ event_t*e = event_new();
+ e->type = EVENT_END;
+ e->p = s->b;
+ e->s1 = s;
+ e->s2 = 0;
+ queue_put(&status->queue, e);
}
static void schedule_crossing(status_t*status, segment_t*s1, segment_t*s2)
dict_put(&s2->scheduled_crossings, (void*)(ptroff_t)(s1->nr), 0);
#endif
- event_t e = event_new();
- e.type = EVENT_CROSS;
- e.p = p;
- e.s1 = s1;
- e.s2 = s2;
- heap_put(status->queue, &e);
+ event_t* e = event_new();
+ e->type = EVENT_CROSS;
+ e->p = p;
+ e->s1 = s1;
+ e->s2 = s2;
+ queue_put(&status->queue, e);
return;
}
event_dump(e);
#endif
intersect_with_horizontal(status, s);
- advance_stroke(status->queue, s->stroke, s->polygon_nr, s->stroke_pos);
+ advance_stroke(&status->queue, 0, s->stroke, s->polygon_nr, s->stroke_pos);
segment_destroy(s);e->s1=0;
break;
}
/* schedule segment for xrow handling */
s->left = 0; s->right = status->ending_segments;
status->ending_segments = s;
- advance_stroke(status->queue, s->stroke, s->polygon_nr, s->stroke_pos);
+ advance_stroke(&status->queue, 0, s->stroke, s->polygon_nr, s->stroke_pos);
break;
}
case EVENT_START: {
#ifdef DEBUG
fprintf(stderr, "========================================================================\n");
#endif
- heap_t* queue = heap_new(sizeof(event_t), compare_events_simple);
- gfxpoly_enqueue(poly, queue, 0);
+ hqueue_t hqueue;
+ hqueue_init(&hqueue);
+ gfxpoly_enqueue(poly, 0, &hqueue, 0);
actlist_t* actlist = actlist_new();
-
- event_t*e = heap_chopmax(queue);
+
+ event_t*e = hqueue_get(&hqueue);
while(e) {
int32_t y = e->p.y;
int32_t x = 0;
char fill = 0;
#ifdef DEBUG
- fprintf(stderr, "----------------------------------- %d\n", y);
+ fprintf(stderr, "HORIZONTALS ----------------------------------- %d\n", y);
actlist_dump(actlist, y-1);
#endif
#ifdef CHECKS
}
segment_t*left = 0;
segment_t*s = e->s1;
-
+
switch(e->type) {
case EVENT_START: {
assert(e->p.x == s->a.x && e->p.y == s->a.y);
actlist_insert(actlist, s->a, s->b, s);
- event_t e;
- e.type = EVENT_END;
- e.p = s->b;
- e.s1 = s;
- e.s2 = 0;
- heap_put(queue, &e);
+ event_t* e = event_new();
+ e->type = EVENT_END;
+ e->p = s->b;
+ e->s1 = s;
+ e->s2 = 0;
+ hqueue_put(&hqueue, e);
left = actlist_left(actlist, s);
break;
}
case EVENT_END: {
left = actlist_left(actlist, s);
actlist_delete(actlist, s);
- advance_stroke(queue, s->stroke, s->polygon_nr, s->stroke_pos);
+ advance_stroke(0, &hqueue, s->stroke, s->polygon_nr, s->stroke_pos);
break;
}
default: assert(0);
if(e->type == EVENT_END)
segment_destroy(s);
- free(e);
- e = heap_chopmax(queue);
+ event_free(e);
+ e = hqueue_get(&hqueue);
} while(e && y == e->p.y);
assert(!fill); // check that polygon is not bleeding
}
actlist_destroy(actlist);
- heap_destroy(queue);
+ hqueue_destroy(&hqueue);
}
gfxpoly_t* gfxpoly_process(gfxpoly_t*poly, windrule_t*windrule, windcontext_t*context)
{
current_polygon = poly;
- heap_t* queue = heap_new(sizeof(event_t), compare_events);
-
- gfxpoly_enqueue(poly, queue, /*polygon nr*/0);
status_t status;
memset(&status, 0, sizeof(status_t));
- status.queue = queue;
+ queue_init(&status.queue);
+ gfxpoly_enqueue(poly, &status.queue, 0, /*polygon nr*/0);
+
status.windrule = windrule;
status.context = context;
status.actlist = actlist_new();
#ifdef CHECKS
status.seen_crossings = dict_new2(&point_type);
- int lasty=heap_peek(queue)?((event_t*)heap_peek(queue))->p.y-1:0;
+ int32_t lasty=-0x80000000;
#endif
status.xrow = xrow_new();
- event_t*e = heap_chopmax(queue);
+ event_t*e = queue_get(&status.queue);
while(e) {
status.y = e->p.y;
- assert(status.y>=lasty);
#ifdef CHECKS
+ assert(status.y>=lasty);
+ lasty = status.y;
status.intersecting_segs = dict_new2(&ptr_type);
status.segs_with_point = dict_new2(&ptr_type);
#endif
do {
xrow_add(status.xrow, e->p.x);
event_apply(&status, e);
- free(e);
- e = heap_chopmax(queue);
+ event_free(e);
+ e = queue_get(&status.queue);
} while(e && status.y == e->p.y);
xrow_sort(status.xrow);
dict_destroy(status.seen_crossings);
#endif
actlist_destroy(status.actlist);
- heap_destroy(queue);
+ queue_destroy(&status.queue);
xrow_destroy(status.xrow);
gfxpoly_t*p = (gfxpoly_t*)status.writer.finish(&status.writer);