crawdad 0.0.1
Sign up to get free protection for your applications and to get access to all the features.
- data/Rakefile +56 -0
- data/ext/crawdad/Makefile +25 -0
- data/ext/crawdad/breakpoint.h +53 -0
- data/ext/crawdad/paragraph.c +275 -0
- data/ext/crawdad/paragraph.h +29 -0
- data/ext/crawdad/tokens.c +57 -0
- data/ext/crawdad/tokens.h +41 -0
- data/lib/crawdad.rb +18 -0
- data/lib/crawdad/breakpoint.rb +82 -0
- data/lib/crawdad/compatibility.rb +12 -0
- data/lib/crawdad/ffi.rb +7 -0
- data/lib/crawdad/ffi/breakpoint_node.rb +36 -0
- data/lib/crawdad/ffi/paragraph.rb +58 -0
- data/lib/crawdad/ffi/tokens.rb +71 -0
- data/lib/crawdad/native.rb +11 -0
- data/lib/crawdad/paragraph.rb +293 -0
- data/lib/crawdad/prawn_tokenizer.rb +139 -0
- data/lib/crawdad/tokens.rb +48 -0
- metadata +75 -0
data/Rakefile
ADDED
@@ -0,0 +1,56 @@
|
|
1
|
+
require 'rubygems'
|
2
|
+
require 'rake'
|
3
|
+
require 'rake/testtask'
|
4
|
+
require 'rake/rdoctask'
|
5
|
+
require 'rake/gempackagetask'
|
6
|
+
|
7
|
+
CRAWDAD_VERSION = '0.0.1'
|
8
|
+
|
9
|
+
task :default => [:build]
|
10
|
+
|
11
|
+
task :build do
|
12
|
+
system "make -Cext/crawdad"
|
13
|
+
end
|
14
|
+
|
15
|
+
desc "Run all tests, test-spec required"
|
16
|
+
Rake::TestTask.new do |test|
|
17
|
+
test.libs << "spec"
|
18
|
+
test.test_files = Dir[ "spec/*_spec.rb" ]
|
19
|
+
test.verbose = true
|
20
|
+
end
|
21
|
+
|
22
|
+
desc "Generate documentation"
|
23
|
+
Rake::RDocTask.new do |rdoc|
|
24
|
+
rdoc.rdoc_files.include("README", "lib/")
|
25
|
+
rdoc.main = "README"
|
26
|
+
rdoc.rdoc_dir = "doc/html"
|
27
|
+
rdoc.title = "Crawdad Documentation"
|
28
|
+
end
|
29
|
+
|
30
|
+
spec = Gem::Specification.new do |spec|
|
31
|
+
spec.name = 'crawdad'
|
32
|
+
spec.version = CRAWDAD_VERSION
|
33
|
+
spec.platform = Gem::Platform::RUBY
|
34
|
+
spec.summary = "Knuth-Plass linebreaking for Ruby"
|
35
|
+
spec.files = FileList["lib/**/**/*"] + FileList["ext/crawdad/*"]
|
36
|
+
spec.require_paths << 'ext'
|
37
|
+
|
38
|
+
binaries = FileList['ext/crawdad/*.bundle', 'ext/crawdad/*.so']
|
39
|
+
spec.extensions << 'Rakefile'
|
40
|
+
spec.files += binaries.to_a
|
41
|
+
|
42
|
+
spec.has_rdoc = true
|
43
|
+
spec.rdoc_options << '--title' << 'Crawdad Documentation' << '-q'
|
44
|
+
spec.author = 'Brad Ediger'
|
45
|
+
spec.email = 'brad.ediger@madriska.com'
|
46
|
+
spec.homepage = 'http://github.com/madriska/crawdad'
|
47
|
+
spec.description = <<END_DESC
|
48
|
+
Crawdad is an implementation of Knuth-Plass linebreaking (justification)
|
49
|
+
for Ruby.
|
50
|
+
END_DESC
|
51
|
+
end
|
52
|
+
|
53
|
+
Rake::GemPackageTask.new(spec) do |pkg|
|
54
|
+
pkg.need_tar = true
|
55
|
+
end
|
56
|
+
|
@@ -0,0 +1,25 @@
|
|
1
|
+
OS:=$(shell uname | sed 's/[-_].*//')
|
2
|
+
CFLAGS=-Wall -O2 -fPIC
|
3
|
+
#CFLAGS=-Wall -fPIC -g
|
4
|
+
SHARED=-shared
|
5
|
+
SOEXT:=.so
|
6
|
+
|
7
|
+
objects = tokens.o paragraph.o
|
8
|
+
headers = tokens.h paragraph.h breakpoint.h
|
9
|
+
|
10
|
+
ifeq (${OS},Darwin)
|
11
|
+
SHARED = -dynamiclib
|
12
|
+
SOEXT:=.bundle
|
13
|
+
endif
|
14
|
+
|
15
|
+
all: crawdad$(SOEXT)
|
16
|
+
|
17
|
+
crawdad$(SOEXT): $(objects)
|
18
|
+
$(CC) $(SHARED) -o crawdad$(SOEXT) $(objects)
|
19
|
+
|
20
|
+
%.o: %.c $(headers)
|
21
|
+
$(CC) -c -o $@ $< $(CFLAGS)
|
22
|
+
|
23
|
+
clean:
|
24
|
+
rm *.o *$(SOEXT)
|
25
|
+
|
@@ -0,0 +1,53 @@
|
|
1
|
+
#ifndef _BREAKPOINT_H_
|
2
|
+
#define _BREAKPOINT_H_
|
3
|
+
|
4
|
+
typedef struct breakpoint {
|
5
|
+
int position;
|
6
|
+
int line;
|
7
|
+
int fitness_class;
|
8
|
+
|
9
|
+
float total_width;
|
10
|
+
float total_stretch;
|
11
|
+
float total_shrink;
|
12
|
+
float total_demerits;
|
13
|
+
|
14
|
+
float ratio;
|
15
|
+
|
16
|
+
struct breakpoint *previous;
|
17
|
+
struct breakpoint *link;
|
18
|
+
} breakpoint;
|
19
|
+
|
20
|
+
struct breakpoint *active_nodes;
|
21
|
+
|
22
|
+
breakpoint *make_starting_breakpoint() {
|
23
|
+
breakpoint *bp;
|
24
|
+
|
25
|
+
bp = malloc(sizeof(breakpoint));
|
26
|
+
|
27
|
+
bp->position = 0;
|
28
|
+
bp->line = 0;
|
29
|
+
bp->fitness_class = 1;
|
30
|
+
|
31
|
+
bp->total_width = 0.0;
|
32
|
+
bp->total_stretch = 0.0;
|
33
|
+
bp->total_shrink = 0.0;
|
34
|
+
bp->total_demerits = 0.0;
|
35
|
+
|
36
|
+
bp->ratio = 0.0;
|
37
|
+
|
38
|
+
bp->previous = NULL;
|
39
|
+
bp->link = NULL;
|
40
|
+
|
41
|
+
return bp;
|
42
|
+
}
|
43
|
+
|
44
|
+
/* Holds information about the best breakpoint found so far for a particular
|
45
|
+
* fitness class. */
|
46
|
+
typedef struct best_breakpoint {
|
47
|
+
breakpoint *bp;
|
48
|
+
float demerits;
|
49
|
+
float ratio;
|
50
|
+
} best_breakpoint;
|
51
|
+
|
52
|
+
#endif
|
53
|
+
|
@@ -0,0 +1,275 @@
|
|
1
|
+
#include <stdio.h>
|
2
|
+
#include <stdlib.h>
|
3
|
+
#include <math.h>
|
4
|
+
|
5
|
+
#include "tokens.h"
|
6
|
+
#include "paragraph.h"
|
7
|
+
#include "breakpoint.h"
|
8
|
+
|
9
|
+
#define FLAGGED_PENALTY 3000
|
10
|
+
#define FITNESS_PENALTY 100
|
11
|
+
|
12
|
+
#define GAMMA INFINITY
|
13
|
+
|
14
|
+
void inspect_token(token *t) {
|
15
|
+
printf("(0x%02lX) ", (unsigned long)t);
|
16
|
+
switch(t->box.type){
|
17
|
+
case BOX:
|
18
|
+
printf("BOX %f \"%s\"\n", t->box.width, t->box.content);
|
19
|
+
break;
|
20
|
+
case GLUE:
|
21
|
+
printf("GLUE %f %f %f\n", t->glue.width, t->glue.stretch,
|
22
|
+
t->glue.shrink);
|
23
|
+
break;
|
24
|
+
case PENALTY:
|
25
|
+
printf("PENALTY %f %f %s\n", t->penalty.penalty, t->penalty.width,
|
26
|
+
(t->penalty.flagged ? "F" : "-"));
|
27
|
+
break;
|
28
|
+
default:
|
29
|
+
printf("UNKNOWN %d\n", t->box.type);
|
30
|
+
}
|
31
|
+
}
|
32
|
+
|
33
|
+
float calculate_demerits(token *stream[], int old_i, token *new_item,
|
34
|
+
float r) {
|
35
|
+
token *old_item = stream[old_i];
|
36
|
+
float d;
|
37
|
+
|
38
|
+
if((new_item->penalty.type == PENALTY) &&
|
39
|
+
(new_item->penalty.penalty >= 0)) {
|
40
|
+
d = pow(1 + 100*(pow(abs(r), 3) + new_item->penalty.penalty), 2);
|
41
|
+
} else if((new_item->penalty.type == PENALTY) &&
|
42
|
+
(new_item->penalty.penalty != -INFINITY)) {
|
43
|
+
d = pow((1 + 100*(pow(abs(r), 3))), 2) -
|
44
|
+
pow(new_item->penalty.penalty, 2);
|
45
|
+
} else {
|
46
|
+
d = pow(1 + 100*(pow(abs(r), 3)), 2);
|
47
|
+
}
|
48
|
+
|
49
|
+
if(old_item->penalty.type == PENALTY && old_item->penalty.flagged &&
|
50
|
+
new_item->penalty.type == PENALTY && new_item->penalty.flagged)
|
51
|
+
d += FLAGGED_PENALTY;
|
52
|
+
|
53
|
+
return d;
|
54
|
+
}
|
55
|
+
|
56
|
+
float adjustment_ratio(float tw, float ty, float tz,
|
57
|
+
float aw, float ay, float az,
|
58
|
+
float target_width, token *stream[], int b) {
|
59
|
+
float w, y, z; /* w=width y=stretch z=shrink */
|
60
|
+
token *item_b = stream[b];
|
61
|
+
|
62
|
+
w = tw - aw; /* Non-adjusted width of the line. */
|
63
|
+
|
64
|
+
/* Add the penalty width (hyphen) if we are breaking at a penalty. */
|
65
|
+
if(item_b->penalty.type == PENALTY)
|
66
|
+
w += item_b->penalty.width;
|
67
|
+
|
68
|
+
if(w < target_width) {
|
69
|
+
y = ty - ay;
|
70
|
+
return (y > 0) ? (target_width - w) / y : INFINITY;
|
71
|
+
} else if(w > target_width) {
|
72
|
+
z = tz - az;
|
73
|
+
return (z > 0) ? (target_width - w) / z : INFINITY;
|
74
|
+
} else {
|
75
|
+
return 0.0;
|
76
|
+
}
|
77
|
+
}
|
78
|
+
|
79
|
+
void calculate_widths(token *stream[], float *tw, float *ty, float *tz){
|
80
|
+
int i;
|
81
|
+
token *p;
|
82
|
+
for(i=0; (p = stream[i]); i++) {
|
83
|
+
switch(p->box.type) {
|
84
|
+
case BOX:
|
85
|
+
return;
|
86
|
+
case GLUE:
|
87
|
+
*tw += p->glue.width;
|
88
|
+
*ty += p->glue.stretch;
|
89
|
+
*tz += p->glue.shrink;
|
90
|
+
break;
|
91
|
+
case PENALTY:
|
92
|
+
if((p->penalty.penalty == -INFINITY) && (i > 0))
|
93
|
+
return;
|
94
|
+
}
|
95
|
+
}
|
96
|
+
}
|
97
|
+
|
98
|
+
void foreach_legal_breakpoint(token *stream[], float width, float threshold,
|
99
|
+
void (*fn)(token **, int, float, float, float, float, float)) {
|
100
|
+
float tw=0, ty=0, tz=0;
|
101
|
+
int i;
|
102
|
+
token *t;
|
103
|
+
|
104
|
+
for(i=0; (t = stream[i]); i++) {
|
105
|
+
switch(t->box.type) {
|
106
|
+
case BOX:
|
107
|
+
tw += t->box.width;
|
108
|
+
break;
|
109
|
+
case GLUE:
|
110
|
+
if(stream[i-1]->box.type == BOX)
|
111
|
+
fn(stream, i, tw, ty, tz, width, threshold);
|
112
|
+
tw += t->glue.width;
|
113
|
+
ty += t->glue.stretch;
|
114
|
+
tz += t->glue.shrink;
|
115
|
+
break;
|
116
|
+
case PENALTY:
|
117
|
+
if(t->penalty.penalty != INFINITY)
|
118
|
+
fn(stream, i, tw, ty, tz, width, threshold);
|
119
|
+
break;
|
120
|
+
}
|
121
|
+
}
|
122
|
+
}
|
123
|
+
|
124
|
+
int fitness_class(float ratio) {
|
125
|
+
if(ratio < -0.5)
|
126
|
+
return 0;
|
127
|
+
if(ratio < 0.5)
|
128
|
+
return 1;
|
129
|
+
if(ratio < 1)
|
130
|
+
return 2;
|
131
|
+
return 3;
|
132
|
+
}
|
133
|
+
|
134
|
+
void concat_new_active_nodes(token *stream[], float total_width, float
|
135
|
+
total_stretch, float total_shrink, best_breakpoint best[4], int i,
|
136
|
+
breakpoint *active, breakpoint **p_previous_node) {
|
137
|
+
float lowest_demerits = INFINITY;
|
138
|
+
float tw = total_width, ty = total_stretch, tz = total_shrink;
|
139
|
+
int fclass;
|
140
|
+
breakpoint *bp;
|
141
|
+
|
142
|
+
for(fclass=0; fclass<4; fclass++)
|
143
|
+
if(best[fclass].demerits < lowest_demerits)
|
144
|
+
lowest_demerits = best[fclass].demerits;
|
145
|
+
|
146
|
+
calculate_widths(stream + i, &tw, &ty, &tz);
|
147
|
+
|
148
|
+
for(fclass=0; fclass<4; fclass++) {
|
149
|
+
if((best[fclass].demerits == INFINITY) ||
|
150
|
+
(best[fclass].demerits > lowest_demerits + GAMMA))
|
151
|
+
continue;
|
152
|
+
|
153
|
+
/* Create and activate node */
|
154
|
+
bp = malloc(sizeof(breakpoint));
|
155
|
+
|
156
|
+
bp->position = i;
|
157
|
+
bp->line = best[fclass].bp->line + 1;
|
158
|
+
bp->fitness_class = fclass;
|
159
|
+
|
160
|
+
bp->total_width = tw;
|
161
|
+
bp->total_stretch = ty;
|
162
|
+
bp->total_shrink = tz;
|
163
|
+
|
164
|
+
bp->total_demerits = best[fclass].demerits;
|
165
|
+
bp->ratio = best[fclass].ratio;
|
166
|
+
|
167
|
+
bp->previous = best[fclass].bp;
|
168
|
+
bp->link = active;
|
169
|
+
|
170
|
+
if(*p_previous_node)
|
171
|
+
(*p_previous_node)->link = bp;
|
172
|
+
else
|
173
|
+
active_nodes = bp;
|
174
|
+
|
175
|
+
*p_previous_node = bp;
|
176
|
+
}
|
177
|
+
}
|
178
|
+
|
179
|
+
void main_loop(token *stream[], int i, float tw, float ty, float tz,
|
180
|
+
float width, float threshold) {
|
181
|
+
breakpoint *active, *next_node, *previous_node;
|
182
|
+
best_breakpoint best[4];
|
183
|
+
int current_line;
|
184
|
+
float ratio;
|
185
|
+
float demerits;
|
186
|
+
int fclass;
|
187
|
+
|
188
|
+
if(active_nodes == NULL) {
|
189
|
+
/* TODO: be nicer */
|
190
|
+
printf("No feasible solution. Try relaxing threshold.");
|
191
|
+
exit(1);
|
192
|
+
}
|
193
|
+
|
194
|
+
active = active_nodes;
|
195
|
+
previous_node = NULL;
|
196
|
+
next_node = NULL;
|
197
|
+
|
198
|
+
while(active) {
|
199
|
+
best[0].demerits = INFINITY;
|
200
|
+
best[1].demerits = INFINITY;
|
201
|
+
best[2].demerits = INFINITY;
|
202
|
+
best[3].demerits = INFINITY;
|
203
|
+
|
204
|
+
while(active) {
|
205
|
+
current_line = active->line + 1;
|
206
|
+
next_node = active->link;
|
207
|
+
|
208
|
+
/* TODO: width can be replaced by a line-specific width for line j */
|
209
|
+
ratio = adjustment_ratio(tw, ty, tz, active->total_width,
|
210
|
+
active->total_stretch, active->total_shrink, width,
|
211
|
+
stream, i);
|
212
|
+
|
213
|
+
if((ratio < -1) || (is_penalty(stream[i]) &&
|
214
|
+
(stream[i]->penalty.penalty == -INFINITY))) {
|
215
|
+
/* Remove active node from the list */
|
216
|
+
if(previous_node)
|
217
|
+
previous_node->link = next_node;
|
218
|
+
else
|
219
|
+
active_nodes = next_node;
|
220
|
+
/* TODO: put active on the passive list or free? */
|
221
|
+
} else {
|
222
|
+
previous_node = active;
|
223
|
+
}
|
224
|
+
|
225
|
+
if((ratio >= -1) && (ratio <= threshold)) {
|
226
|
+
demerits = calculate_demerits(stream, active->position, stream[i],
|
227
|
+
ratio) + active->total_demerits;
|
228
|
+
fclass = fitness_class(ratio);
|
229
|
+
|
230
|
+
/* Penalize consecutive lines more than one fitness class away from
|
231
|
+
* each other. */
|
232
|
+
if(abs(fclass - active->fitness_class) > 1)
|
233
|
+
demerits += FITNESS_PENALTY;
|
234
|
+
|
235
|
+
/* Update high scores if this is a new best. */
|
236
|
+
if(demerits < best[fclass].demerits) {
|
237
|
+
best[fclass].bp = active;
|
238
|
+
best[fclass].demerits = demerits;
|
239
|
+
best[fclass].ratio = ratio;
|
240
|
+
}
|
241
|
+
}
|
242
|
+
|
243
|
+
/* Add nodes to the active list before moving to the next line. */
|
244
|
+
active = next_node;
|
245
|
+
if(!active)
|
246
|
+
break;
|
247
|
+
if(active->line >= current_line)
|
248
|
+
break;
|
249
|
+
}
|
250
|
+
|
251
|
+
/* If we found any best nodes, add them to the active list. */
|
252
|
+
concat_new_active_nodes(stream, tw, ty, tz, best, i, active,
|
253
|
+
&previous_node);
|
254
|
+
|
255
|
+
active = next_node;
|
256
|
+
}
|
257
|
+
}
|
258
|
+
|
259
|
+
breakpoint *populate_active_nodes(token *stream[], float width,
|
260
|
+
float threshold) {
|
261
|
+
breakpoint *bp, *min_node;
|
262
|
+
|
263
|
+
active_nodes = make_starting_breakpoint();
|
264
|
+
foreach_legal_breakpoint(stream, width, threshold, main_loop);
|
265
|
+
|
266
|
+
/* Find node with minimum demerits */
|
267
|
+
min_node = NULL;
|
268
|
+
for(bp = active_nodes; bp; bp = bp->link)
|
269
|
+
if(!min_node || (bp->total_demerits < min_node->total_demerits))
|
270
|
+
min_node = bp;
|
271
|
+
|
272
|
+
return min_node;
|
273
|
+
}
|
274
|
+
|
275
|
+
|
@@ -0,0 +1,29 @@
|
|
1
|
+
#ifndef _PARAGRAPH_H_
|
2
|
+
#define _PARAGRAPH_H_
|
3
|
+
|
4
|
+
#include "breakpoint.h"
|
5
|
+
|
6
|
+
void inspect_token(token *t);
|
7
|
+
|
8
|
+
float calculate_demerits(token *stream[], int old_i, token *new_item,
|
9
|
+
float r);
|
10
|
+
|
11
|
+
float adjustment_ratio(float tw, float ty, float tz,
|
12
|
+
float aw, float ay, float az,
|
13
|
+
float target_width, token *stream[], int b);
|
14
|
+
|
15
|
+
void calculate_widths(token *stream[], float *tw, float *ty, float *tz);
|
16
|
+
|
17
|
+
void foreach_legal_breakpoint(token *stream[], float width, float threshold,
|
18
|
+
void (*fn)(token **, int, float, float, float, float, float));
|
19
|
+
|
20
|
+
int fitness_class(float ratio);
|
21
|
+
|
22
|
+
void main_loop(token *stream[], int i, float tw, float ty, float tz,
|
23
|
+
float width, float threshold);
|
24
|
+
|
25
|
+
breakpoint *populate_active_nodes(token *stream[], float width,
|
26
|
+
float threshold);
|
27
|
+
|
28
|
+
#endif
|
29
|
+
|
@@ -0,0 +1,57 @@
|
|
1
|
+
#include <stdio.h>
|
2
|
+
#include <stdlib.h>
|
3
|
+
#include <math.h>
|
4
|
+
#include <string.h>
|
5
|
+
|
6
|
+
#include "tokens.h"
|
7
|
+
|
8
|
+
struct box *make_box(float width, char *content) {
|
9
|
+
int len;
|
10
|
+
struct box *t;
|
11
|
+
|
12
|
+
t = malloc(sizeof(struct box));
|
13
|
+
t->type = BOX;
|
14
|
+
t->width = width;
|
15
|
+
|
16
|
+
len = strlen(content);
|
17
|
+
t->content = malloc(len+1);
|
18
|
+
strncpy(t->content, content, len);
|
19
|
+
t->content[len] = '\0';
|
20
|
+
|
21
|
+
return t;
|
22
|
+
}
|
23
|
+
|
24
|
+
struct glue *make_glue(float width, float stretch, float shrink) {
|
25
|
+
struct glue *t = malloc(sizeof(struct glue));
|
26
|
+
t->type = GLUE;
|
27
|
+
t->width = width;
|
28
|
+
t->stretch = stretch;
|
29
|
+
t->shrink = shrink;
|
30
|
+
return t;
|
31
|
+
}
|
32
|
+
|
33
|
+
struct penalty *make_penalty(float width, float penalty, int flagged) {
|
34
|
+
struct penalty *t = malloc(sizeof(struct penalty));
|
35
|
+
t->type = PENALTY;
|
36
|
+
t->width = width;
|
37
|
+
t->penalty = penalty;
|
38
|
+
t->flagged = flagged;
|
39
|
+
return t;
|
40
|
+
}
|
41
|
+
|
42
|
+
int token_type(token *t) {
|
43
|
+
return t->box.type;
|
44
|
+
}
|
45
|
+
|
46
|
+
int is_box(token *t) {
|
47
|
+
return (t->box.type == BOX);
|
48
|
+
}
|
49
|
+
|
50
|
+
int is_penalty(token *t) {
|
51
|
+
return (t->penalty.type == PENALTY);
|
52
|
+
}
|
53
|
+
|
54
|
+
int is_glue(token *t) {
|
55
|
+
return (t->glue.type == GLUE);
|
56
|
+
}
|
57
|
+
|