gocr-ruby 0.0.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- checksums.yaml +7 -0
- data/.gitignore +22 -0
- data/Gemfile +4 -0
- data/LICENSE +21 -0
- data/LICENSE.txt +22 -0
- data/README.md +29 -0
- data/Rakefile +49 -0
- data/ext/gocr/Makefile +141 -0
- data/ext/gocr/Makefile.in +140 -0
- data/ext/gocr/amiga.h +31 -0
- data/ext/gocr/barcode.c +2108 -0
- data/ext/gocr/barcode.h +11 -0
- data/ext/gocr/box.c +496 -0
- data/ext/gocr/config.h +37 -0
- data/ext/gocr/config.h.in +36 -0
- data/ext/gocr/database.c +468 -0
- data/ext/gocr/detect.c +1003 -0
- data/ext/gocr/extconf.rb +6 -0
- data/ext/gocr/gocr.c +436 -0
- data/ext/gocr/gocr.h +290 -0
- data/ext/gocr/jconv.c +168 -0
- data/ext/gocr/job.c +92 -0
- data/ext/gocr/lines.c +364 -0
- data/ext/gocr/list.c +334 -0
- data/ext/gocr/list.h +91 -0
- data/ext/gocr/ocr0.c +7312 -0
- data/ext/gocr/ocr0.h +63 -0
- data/ext/gocr/ocr0n.c +1527 -0
- data/ext/gocr/ocr1.c +85 -0
- data/ext/gocr/ocr1.h +3 -0
- data/ext/gocr/otsu.c +310 -0
- data/ext/gocr/otsu.h +23 -0
- data/ext/gocr/output.c +291 -0
- data/ext/gocr/output.h +37 -0
- data/ext/gocr/pcx.c +153 -0
- data/ext/gocr/pcx.h +9 -0
- data/ext/gocr/pgm2asc.c +3259 -0
- data/ext/gocr/pgm2asc.h +105 -0
- data/ext/gocr/pixel.c +538 -0
- data/ext/gocr/pnm.c +538 -0
- data/ext/gocr/pnm.h +35 -0
- data/ext/gocr/progress.c +87 -0
- data/ext/gocr/progress.h +42 -0
- data/ext/gocr/remove.c +715 -0
- data/ext/gocr/tga.c +87 -0
- data/ext/gocr/tga.h +6 -0
- data/ext/gocr/unicode.c +1318 -0
- data/ext/gocr/unicode.h +62 -0
- data/ext/gocr/unicode_defs.h +1245 -0
- data/ext/gocr/version.h +2 -0
- data/gocr-ruby.gemspec +28 -0
- data/image.png +0 -0
- data/lib/gocr.rb +6 -0
- data/lib/gocr/image.rb +8 -0
- data/lib/gocr/version.rb +3 -0
- metadata +156 -0
data/ext/gocr/jconv.c
ADDED
@@ -0,0 +1,168 @@
|
|
1
|
+
/* OCR Aug00 JS
|
2
|
+
// PGM gray ASCII=P2 RAW=P5
|
3
|
+
// PPM RGB ASCII=P3 RAW=P6
|
4
|
+
// PBM B/W ASCII=P1 RAW=P4
|
5
|
+
// ToDo:
|
6
|
+
// - pbm-raw to pgm also for x!=0 (mod 8)
|
7
|
+
// v0.01 bug eliminated
|
8
|
+
// v0.02 convert renamed into jconv because ImageMagick uses same name
|
9
|
+
// v0.03 code review bbg
|
10
|
+
// program is not used anymore, use "convert -verbose -crop 0x0+1+1" instead
|
11
|
+
*/
|
12
|
+
|
13
|
+
// #include <iostream.h>
|
14
|
+
#include "config.h"
|
15
|
+
#include <stdio.h>
|
16
|
+
#include <stdlib.h>
|
17
|
+
#include <assert.h>
|
18
|
+
#include <string.h>
|
19
|
+
#include "pnm.h"
|
20
|
+
#ifdef HAVE_PAM_H
|
21
|
+
# include <pam.h>
|
22
|
+
#endif
|
23
|
+
#include "pcx.h"
|
24
|
+
#include "tga.h"
|
25
|
+
|
26
|
+
void help( void ) {
|
27
|
+
printf("jconv version Aug2000 JS (pnm-raw,pcx8,tga24)\n"
|
28
|
+
"use: jconv [options] ?infile.pnm? ?outfile.pgm? ?ox? ?oy? ?dx? ?dy?\n"
|
29
|
+
"options: -shrink -pbm -? -help\n"
|
30
|
+
"example: jconv -shrink -pbm font.pbm font.pbm 0 0 0 0\n");
|
31
|
+
exit(1);
|
32
|
+
}
|
33
|
+
|
34
|
+
int main(int argn, char *argv[])
|
35
|
+
{
|
36
|
+
char *inam, *onam;
|
37
|
+
pix bild;
|
38
|
+
int ox, oy, dx, dy, x, y, i, vvv = 0;
|
39
|
+
|
40
|
+
#ifdef HAVE_PAM_H
|
41
|
+
pnm_init(&argn, argv);
|
42
|
+
#endif
|
43
|
+
// skip options
|
44
|
+
for (i = 1; i < argn; i++) {
|
45
|
+
if (argv[i][0] != '-')
|
46
|
+
break;
|
47
|
+
if (!strcmp(argv[i], "-?"))
|
48
|
+
help();
|
49
|
+
else if (!strcmp(argv[i], "-help"))
|
50
|
+
help();
|
51
|
+
else if (!strcmp(argv[i], "-shrink"))
|
52
|
+
vvv |= 2;
|
53
|
+
else if (!strcmp(argv[i], "-pbm"))
|
54
|
+
vvv |= 4;
|
55
|
+
else
|
56
|
+
printf("unknown option: %s\n", argv[i]);
|
57
|
+
}
|
58
|
+
|
59
|
+
if (argn - i != 6)
|
60
|
+
help();
|
61
|
+
inam = argv[i++];
|
62
|
+
onam = argv[i++];
|
63
|
+
ox = atoi(argv[i++]);
|
64
|
+
oy = atoi(argv[i++]);
|
65
|
+
dx = atoi(argv[i++]);
|
66
|
+
dy = atoi(argv[i++]);
|
67
|
+
printf("# in=%s out=%s offs=%d,%d len=%d,%d vvv=%d\n",
|
68
|
+
inam, onam, ox, oy, dx, dy, vvv);
|
69
|
+
|
70
|
+
// ----- read picture
|
71
|
+
if (strstr(inam, ".pbm") ||
|
72
|
+
strstr(inam, ".pgm") ||
|
73
|
+
strstr(inam, ".ppm") ||
|
74
|
+
strstr(inam, ".pnm") ||
|
75
|
+
strstr(inam, ".pam"))
|
76
|
+
readpgm(inam, &bild, 1);
|
77
|
+
else if (strstr(inam, ".pcx"))
|
78
|
+
readpcx(inam, &bild, 1);
|
79
|
+
else if (strstr(inam, ".tga"))
|
80
|
+
readtga(inam, &bild, ((vvv > 1) ? 0 : 1));
|
81
|
+
else {
|
82
|
+
printf("Error: unknown suffix\n");
|
83
|
+
exit(1);
|
84
|
+
}
|
85
|
+
if (ox < 0 || ox >= bild.x)
|
86
|
+
ox = 0;
|
87
|
+
if (oy < 0 || ox >= bild.y)
|
88
|
+
oy = 0;
|
89
|
+
if (dx <= 0 || ox + dx > bild.x)
|
90
|
+
dx = bild.x - ox;
|
91
|
+
if (dy <= 0 || oy + dy > bild.y)
|
92
|
+
dy = bild.y - oy;
|
93
|
+
if ((vvv & 2) == 2 && bild.bpp == 1) { // -shrink
|
94
|
+
int x, y;
|
95
|
+
printf("# shrinking PGM: offs=%d,%d len=%d,%d\n", ox, oy, dx, dy);
|
96
|
+
for (y = 0; y < dy; y++) { // shrink upper border
|
97
|
+
for (x = 0; x < dx; x++)
|
98
|
+
if (bild.p[x + ox + (y + oy) * bild.x] < 127)
|
99
|
+
break;
|
100
|
+
if (x < dx) {
|
101
|
+
if (y > 0)
|
102
|
+
y--;
|
103
|
+
oy += y;
|
104
|
+
dy -= y;
|
105
|
+
break;
|
106
|
+
}
|
107
|
+
}
|
108
|
+
for (y = 0; y < dy; y++) { // shrink lower border
|
109
|
+
for (x = 0; x < dx; x++)
|
110
|
+
if (bild.p[ox + x + (oy + dy - y - 1) * bild.x] < 127)
|
111
|
+
break;
|
112
|
+
if (x < dx) {
|
113
|
+
if (y > 0)
|
114
|
+
y--;
|
115
|
+
dy -= y;
|
116
|
+
break;
|
117
|
+
}
|
118
|
+
}
|
119
|
+
for (x = 0; x < dx; x++) { // shrink left border
|
120
|
+
for (y = 0; y < dy; y++)
|
121
|
+
if (bild.p[x + ox + (y + oy) * bild.x] < 127)
|
122
|
+
break;
|
123
|
+
if (y < dy) {
|
124
|
+
if (x > 0)
|
125
|
+
x--;
|
126
|
+
ox += x;
|
127
|
+
dx -= x;
|
128
|
+
break;
|
129
|
+
}
|
130
|
+
}
|
131
|
+
for (x = 0; x < dx; x++) { // shrink right border
|
132
|
+
for (y = 0; y < dy; y++)
|
133
|
+
if (bild.p[ox + dx - x - 1 + (oy + y) * bild.x] < 127)
|
134
|
+
break;
|
135
|
+
if (y < dy) {
|
136
|
+
if (x > 0)
|
137
|
+
x--;
|
138
|
+
dx -= x;
|
139
|
+
break;
|
140
|
+
}
|
141
|
+
}
|
142
|
+
}
|
143
|
+
printf("# final dimension: offs=%d,%d len=%d,%d bpp=%d\n",
|
144
|
+
ox, oy, dx, dy, bild.bpp);
|
145
|
+
|
146
|
+
/* bbg: could be changed to memmoves */
|
147
|
+
// ---- new size
|
148
|
+
for (y = 0; y < dy; y++)
|
149
|
+
for (x = 0; x < dx; x++)
|
150
|
+
for (i = 0; i < 3; i++)
|
151
|
+
bild.p[i + bild.bpp * (x + dx * y)] =
|
152
|
+
bild.p[i + bild.bpp * (x + ox + (y + oy) * bild.x)];
|
153
|
+
bild.x = dx;
|
154
|
+
bild.y = dy;
|
155
|
+
// ---- write internal picture of textsite
|
156
|
+
printf("# write %s\n", onam);
|
157
|
+
if (strstr(onam, ".pbm"))
|
158
|
+
writepbm(onam, &bild);
|
159
|
+
else if (strstr(onam, ".pgm"))
|
160
|
+
writepgm(onam, &bild);
|
161
|
+
else if (strstr(onam, ".ppm"))
|
162
|
+
writeppm(onam, &bild);
|
163
|
+
else if (strstr(onam, ".pnm"))
|
164
|
+
writepgm(onam, &bild);
|
165
|
+
else
|
166
|
+
printf("Error: unknown suffix");
|
167
|
+
free( bild.p );
|
168
|
+
}
|
data/ext/gocr/job.c
ADDED
@@ -0,0 +1,92 @@
|
|
1
|
+
/*
|
2
|
+
This is a Optical-Character-Recognition program
|
3
|
+
Copyright (C) 2000-2010 Joerg Schulenburg
|
4
|
+
|
5
|
+
This program is free software; you can redistribute it and/or
|
6
|
+
modify it under the terms of the GNU General Public License
|
7
|
+
as published by the Free Software Foundation; either version 2
|
8
|
+
of the License, or (at your option) any later version.
|
9
|
+
|
10
|
+
This program is distributed in the hope that it will be useful,
|
11
|
+
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
12
|
+
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
13
|
+
GNU General Public License for more details.
|
14
|
+
|
15
|
+
You should have received a copy of the GNU General Public License
|
16
|
+
along with this program; if not, write to the Free Software
|
17
|
+
Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
|
18
|
+
|
19
|
+
see README for email address */
|
20
|
+
|
21
|
+
#include "pgm2asc.h"
|
22
|
+
#include "gocr.h"
|
23
|
+
|
24
|
+
/* initialize job structure cfg and db (for all images of a multiimage) */
|
25
|
+
void job_init(job_t *job) {
|
26
|
+
/* init source */
|
27
|
+
job->src.fname = "-";
|
28
|
+
|
29
|
+
/* init temporaries */
|
30
|
+
list_init( &job->tmp.dblist );
|
31
|
+
|
32
|
+
/* init cfg */
|
33
|
+
job->cfg.cs = 0;
|
34
|
+
job->cfg.spc = 0;
|
35
|
+
job->cfg.mode = 0;
|
36
|
+
job->cfg.dust_size = -1; /* auto detect */
|
37
|
+
job->cfg.only_numbers = 0;
|
38
|
+
job->cfg.verbose = 0;
|
39
|
+
job->cfg.out_format = UTF8; /* old: ISO8859_1; */
|
40
|
+
job->cfg.lc = "_";
|
41
|
+
job->cfg.db_path = (char*)NULL;
|
42
|
+
job->cfg.cfilter = (char*)NULL;
|
43
|
+
job->cfg.certainty = 95;
|
44
|
+
job->cfg.unrec_marker = "_";
|
45
|
+
}
|
46
|
+
|
47
|
+
/* initialize job structure for every image (multi-images) */
|
48
|
+
void job_init_image(job_t *job) {
|
49
|
+
|
50
|
+
/* FIXME jb: init pix */
|
51
|
+
job->src.p.p = NULL;
|
52
|
+
|
53
|
+
/* init results */
|
54
|
+
list_init( &job->res.boxlist );
|
55
|
+
list_init( &job->res.linelist );
|
56
|
+
job->res.avX = 5;
|
57
|
+
job->res.avY = 8;
|
58
|
+
job->res.sumX = 0;
|
59
|
+
job->res.sumY = 0;
|
60
|
+
job->res.numC = 0;
|
61
|
+
job->res.lines.dy=0;
|
62
|
+
job->res.lines.num=0;
|
63
|
+
|
64
|
+
/* init temporaries */
|
65
|
+
job->tmp.n_run = 0;
|
66
|
+
/* FIXME jb: init ppo */
|
67
|
+
job->tmp.ppo.p = NULL;
|
68
|
+
job->tmp.ppo.x = 0;
|
69
|
+
job->tmp.ppo.y = 0;
|
70
|
+
|
71
|
+
}
|
72
|
+
|
73
|
+
/* free job structure */
|
74
|
+
void job_free_image(job_t *job) {
|
75
|
+
|
76
|
+
/* if tmp is just a copy of the pointer to the original image */
|
77
|
+
if (job->tmp.ppo.p==job->src.p.p) job->tmp.ppo.p=NULL;
|
78
|
+
|
79
|
+
/* FIMXE jb: free lists
|
80
|
+
* list_free( &job->res.linelist );
|
81
|
+
* list_free( &job->tmp.dblist );
|
82
|
+
*/
|
83
|
+
|
84
|
+
list_and_data_free(&(job->res.boxlist), (void (*)(void *))free_box);
|
85
|
+
|
86
|
+
/* FIXME jb: free pix */
|
87
|
+
if (job->src.p.p) { free(job->src.p.p); job->src.p.p=NULL; }
|
88
|
+
|
89
|
+
/* FIXME jb: free pix */
|
90
|
+
if (job->tmp.ppo.p) { free(job->tmp.ppo.p); job->tmp.ppo.p=NULL; }
|
91
|
+
|
92
|
+
}
|
data/ext/gocr/lines.c
ADDED
@@ -0,0 +1,364 @@
|
|
1
|
+
/*
|
2
|
+
This is a Optical-Character-Recognition program
|
3
|
+
Copyright (C) 2000-2010 Joerg Schulenburg
|
4
|
+
|
5
|
+
This program is free software; you can redistribute it and/or
|
6
|
+
modify it under the terms of the GNU General Public License
|
7
|
+
as published by the Free Software Foundation; either version 2
|
8
|
+
of the License, or (at your option) any later version.
|
9
|
+
|
10
|
+
This program is distributed in the hope that it will be useful,
|
11
|
+
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
12
|
+
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
13
|
+
GNU General Public License for more details.
|
14
|
+
|
15
|
+
You should have received a copy of the GNU General Public License
|
16
|
+
along with this program; if not, write to the Free Software
|
17
|
+
Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
|
18
|
+
|
19
|
+
see README for EMAIL-address
|
20
|
+
*/
|
21
|
+
|
22
|
+
#include <stdlib.h>
|
23
|
+
#include <stdio.h>
|
24
|
+
#include <string.h>
|
25
|
+
#include <limits.h>
|
26
|
+
#include <assert.h>
|
27
|
+
#include "pgm2asc.h"
|
28
|
+
#include "gocr.h"
|
29
|
+
#include "unicode.h" /* decode() */
|
30
|
+
#include "unicode_defs.h" /* UNKNOWN */
|
31
|
+
|
32
|
+
const char *getTextLine (List *linelist, int line) {
|
33
|
+
int i;
|
34
|
+
Element *elem;
|
35
|
+
|
36
|
+
if (line < 0 || line > list_total(linelist))
|
37
|
+
return NULL;
|
38
|
+
|
39
|
+
for ( i = 0, elem = linelist->start.next;
|
40
|
+
i < line && elem != NULL; i++ )
|
41
|
+
elem = elem->next;
|
42
|
+
|
43
|
+
if ( elem != NULL )
|
44
|
+
return (const char *)elem->data;
|
45
|
+
|
46
|
+
return NULL;
|
47
|
+
}
|
48
|
+
|
49
|
+
void free_textlines(List *linelist) { // list.h
|
50
|
+
for_each_data(linelist) {
|
51
|
+
if (list_get_current(linelist))
|
52
|
+
free(list_get_current(linelist)); // free list element
|
53
|
+
} end_for_each(linelist);
|
54
|
+
list_free(linelist); // free list structure
|
55
|
+
}
|
56
|
+
|
57
|
+
/* append a string (s1) to the string buffer (buffer) of length (len)
|
58
|
+
* if buffer is to small or len==0 realloc buffer, len+=512
|
59
|
+
*/
|
60
|
+
char *append_to_line(char *buffer, const char *s1, int *len) {
|
61
|
+
char *temp;
|
62
|
+
int slen=0, alen;
|
63
|
+
if( s1==NULL || s1[0] == 0 ){
|
64
|
+
fprintf(stderr,"\n#BUG: appending 0 to a line makes no sense!");
|
65
|
+
return buffer;
|
66
|
+
}
|
67
|
+
if ( *len>0 ) slen= strlen(buffer); // used buffer
|
68
|
+
alen = strlen(s1);
|
69
|
+
if ( slen+alen+1 >= *len ) {
|
70
|
+
*len += (((alen+1)>>9)+1)<<9; // round up to next 512 bytes
|
71
|
+
temp = (char *)realloc(buffer, *len);
|
72
|
+
if( !temp ) { fprintf(stderr,"realloc failed!\n");
|
73
|
+
*len -= (((alen+1)>>9)+1)<<9; return buffer; } // go back
|
74
|
+
else buffer = temp; // buffer successfull enlarged
|
75
|
+
}
|
76
|
+
temp = buffer + slen; // end of buffered string
|
77
|
+
memcpy(temp,s1,alen+1); // copy including end sign '\0'
|
78
|
+
return buffer;
|
79
|
+
}
|
80
|
+
|
81
|
+
int calc_median_gap(struct tlines * lines) {
|
82
|
+
int gaps[MAXlines], l;
|
83
|
+
if (lines->num<2) return 0;
|
84
|
+
for (l = 0; l < lines->num - 1; l++)
|
85
|
+
gaps[l] = lines->m2[l + 1] - lines->m3[l];
|
86
|
+
qsort(gaps, lines->num - 1, sizeof(gaps[0]), intcompare);
|
87
|
+
return gaps[(lines->num - 1) / 2];
|
88
|
+
}
|
89
|
+
|
90
|
+
/*
|
91
|
+
* Return the indent in pixels of the least-indented line.
|
92
|
+
* Will be subtracted as base_indent to avoid negativ indent.
|
93
|
+
*
|
94
|
+
* This is adjusted to account for an angle on the page as
|
95
|
+
* a whole. For instance, if the page is rotated clockwise,
|
96
|
+
* lower lines may be physically closer to the left edge
|
97
|
+
* than higher lines that are logically less indented.
|
98
|
+
* We rotate around (0,0). Note that this rotation could
|
99
|
+
* rotate lines "off the left margin", leading to a negative
|
100
|
+
* indent.
|
101
|
+
*
|
102
|
+
* boxlist -- list of character boxes.
|
103
|
+
* dx, dy -- rotation angle as vector
|
104
|
+
*/
|
105
|
+
int get_least_line_indent(List * boxlist, int dx, int dy, int verbose) {
|
106
|
+
int min_indent = INT_MAX;
|
107
|
+
int adjusted_indent;
|
108
|
+
struct box * box2;
|
109
|
+
if (verbose)
|
110
|
+
fprintf(stderr, "get_least_line_indent: rot.vector dxdy %d %d\n",
|
111
|
+
dx, dy);
|
112
|
+
for_each_data(boxlist) {
|
113
|
+
box2 = (struct box *)list_get_current(boxlist);
|
114
|
+
/* if num == -1, indicates this is a space or newline box,
|
115
|
+
* inserted in list_insert_spaces. */
|
116
|
+
if (box2->num != -1) {
|
117
|
+
adjusted_indent = box2->x0;
|
118
|
+
if (dx) adjusted_indent += box2->y0 * dy / dx;
|
119
|
+
if (adjusted_indent < min_indent) {
|
120
|
+
min_indent = adjusted_indent;
|
121
|
+
if (dy!=0 && verbose)
|
122
|
+
fprintf(stderr,
|
123
|
+
"# Line %2d, unadjusted xy %3d %3d, adjusted x %2d\n",
|
124
|
+
box2->line, box2->x0, box2->y0, adjusted_indent);
|
125
|
+
}
|
126
|
+
}
|
127
|
+
} end_for_each(boxlist);
|
128
|
+
if (verbose)
|
129
|
+
fprintf(stderr, "# Minimum adjusted x: %d (min_indent)\n", min_indent);
|
130
|
+
return min_indent;
|
131
|
+
}
|
132
|
+
|
133
|
+
/* collect all the chars from the box tree and write them to a string buffer
|
134
|
+
mo is the mode: mode&8 means, use chars even if unsure recognized
|
135
|
+
ToDo: store full text(?), store decoded text+boxes+position chars (v0.4)
|
136
|
+
(HTML,UTF,ASCII,XML), not wchar incl. descriptions (at<95% in red)
|
137
|
+
remove decode(*c, job->cfg.out_format) from gocr.c!
|
138
|
+
XML add alternate-tags, format tags and position tags
|
139
|
+
ToDo: better output XML to stdout instead of circumstantial store to lines
|
140
|
+
not all texts/images follow the line concept?
|
141
|
+
Better use a tree of objects where leafes are chars instead of simple list.
|
142
|
+
Chars or objects are taken into account. Objects can be text strings
|
143
|
+
or XML strings.
|
144
|
+
|
145
|
+
ToDo: replacing by output_to_stream(FILE *ostream, int mo) ??
|
146
|
+
can be used via pipes (if library) ???
|
147
|
+
*/
|
148
|
+
void store_boxtree_lines(job_t *job, int mo) {
|
149
|
+
char *buffer; /* temp buffer for text */
|
150
|
+
int i = 0, j = 0;
|
151
|
+
int len = 1024; // initial buffer length for text line
|
152
|
+
struct box *box2;
|
153
|
+
int median_gap = 0;
|
154
|
+
int max_single_space_gap = 0;
|
155
|
+
struct tlines line_info;
|
156
|
+
int line, line_gap, oldline=-1;
|
157
|
+
int left_margin;
|
158
|
+
int i1=0, i2=0;
|
159
|
+
|
160
|
+
buffer = (char *)malloc(len);
|
161
|
+
if ( !buffer ) {
|
162
|
+
fprintf(stderr,"malloc failed!\n"); // ToDo: index_to_error_list
|
163
|
+
return;
|
164
|
+
}
|
165
|
+
*buffer = 0;
|
166
|
+
|
167
|
+
if ( job->cfg.verbose&1 )
|
168
|
+
fprintf(stderr,"# store boxtree to lines ...");
|
169
|
+
|
170
|
+
/* wew: calculate the median line gap, to determine line spacing
|
171
|
+
* for the text output. The line gap used is between one line's
|
172
|
+
* m3 (baseline) and the next line's m2 (height of non-rising
|
173
|
+
* lowercase). We use these lines as they are the least likely
|
174
|
+
* to vary according to actual character content of lines.
|
175
|
+
*/
|
176
|
+
median_gap = calc_median_gap(&job->res.lines);
|
177
|
+
if (median_gap <= 0) {
|
178
|
+
if ( job->cfg.verbose&1 )
|
179
|
+
fprintf(stderr, "# Warning: non-positive median line gap of %d\n",
|
180
|
+
median_gap);
|
181
|
+
median_gap = 8;
|
182
|
+
max_single_space_gap = 12; /* arbitrary */
|
183
|
+
} else {
|
184
|
+
max_single_space_gap = median_gap * 7 / 4;
|
185
|
+
}
|
186
|
+
|
187
|
+
// Will be subtracted as base_indent to avoid negativ indent.
|
188
|
+
left_margin = get_least_line_indent(&job->res.boxlist,
|
189
|
+
job->res.lines.dx,
|
190
|
+
job->res.lines.dy, job->cfg.verbose);
|
191
|
+
|
192
|
+
if (job->cfg.out_format==XML) { /* subject of change */
|
193
|
+
char s1[255]; /* ToDo: avoid potential buffer overflow !!! */
|
194
|
+
/* output lot of usefull information for XML filter */
|
195
|
+
sprintf(s1,"<page x=\"%d\" y=\"%d\" dx=\"%d\" dy=\"%d\">\n",
|
196
|
+
0,0,0,0);
|
197
|
+
buffer=append_to_line(buffer,s1,&len);
|
198
|
+
sprintf(s1,"<block x=\"%d\" y=\"%d\" dx=\"%d\" dy=\"%d\">\n",
|
199
|
+
0,0,0,0);
|
200
|
+
buffer=append_to_line(buffer,s1,&len);
|
201
|
+
}
|
202
|
+
|
203
|
+
for_each_data(&(job->res.boxlist)) {
|
204
|
+
box2 = (struct box *)list_get_current(&(job->res.boxlist));
|
205
|
+
line = box2->line;
|
206
|
+
line_info = job->res.lines;
|
207
|
+
/* reset the output char if certainty is below the limit v0.44 */
|
208
|
+
if (box2->num_ac && box2->wac[0]<job->cfg.certainty) box2->c=UNKNOWN;
|
209
|
+
if (line!=oldline) {
|
210
|
+
if (job->cfg.out_format==XML && oldline>-1) { /* subject of change */
|
211
|
+
buffer=append_to_line(buffer,"</line>\n",&len);
|
212
|
+
list_app( &(job->res.linelist), (void *)strdup(buffer) ); // wcsdup
|
213
|
+
memset(buffer, 0, len);
|
214
|
+
j=0; // reset counter for new line
|
215
|
+
}
|
216
|
+
if (job->cfg.out_format==XML) { /* subject of change */
|
217
|
+
char s1[255]; /* ToDo: avoid potential buffer overflow !!! */
|
218
|
+
/* output lot of usefull information for XML filter */
|
219
|
+
sprintf(s1,"<line x=\"%d\" y=\"%d\" dx=\"%d\" dy=\"%d\" value=\"%d\">\n",
|
220
|
+
line_info.x0[line],line_info.m1[line],
|
221
|
+
line_info.x1[line]-line_info.x0[line]+1,
|
222
|
+
line_info.m4[line]-line_info.m1[line],line);
|
223
|
+
buffer=append_to_line(buffer,s1,&len);
|
224
|
+
}
|
225
|
+
oldline=line;
|
226
|
+
}
|
227
|
+
if (box2->c > ' ' &&
|
228
|
+
box2->c <= 'z') i1++; /* count non-space chars */
|
229
|
+
if (box2->c == '\n') {
|
230
|
+
if (job->cfg.out_format!=XML) { /* subject of change */
|
231
|
+
line_info = job->res.lines;
|
232
|
+
line = box2->line;
|
233
|
+
if (line > 0) {
|
234
|
+
line_gap = line_info.m2[line] - line_info.m3[line - 1];
|
235
|
+
for (line_gap -= max_single_space_gap; line_gap > 0;
|
236
|
+
line_gap -= median_gap) {
|
237
|
+
buffer=append_to_line(buffer,"\n",&len);
|
238
|
+
j++; /* count chars in line */
|
239
|
+
}
|
240
|
+
}
|
241
|
+
list_app( &(job->res.linelist), (void *)strdup(buffer) ); // wcsdup
|
242
|
+
memset(buffer, 0, len);
|
243
|
+
j=0; // reset counter for new line
|
244
|
+
}
|
245
|
+
}
|
246
|
+
if (box2->c == ' ') // fill large gaps with spaces
|
247
|
+
{
|
248
|
+
if (job->res.avX) { /* avoid SIGFPE */
|
249
|
+
if (job->cfg.out_format==XML) { /* subject of change */
|
250
|
+
char s1[255]; /* ToDo: avoid potential buffer overflow !!! */
|
251
|
+
/* output lot of usefull information for XML filter */
|
252
|
+
sprintf(s1," <space x=\"%d\" y=\"%d\" dx=\"%d\" dy=\"%d\" />\n",
|
253
|
+
box2->x0,box2->y0,box2->x1-box2->x0+1,box2->y1-box2->y0+1);
|
254
|
+
buffer=append_to_line(buffer,s1,&len);
|
255
|
+
} else
|
256
|
+
// multi spacing is done now in pgm2asc using insert spaces 2010-09-28
|
257
|
+
// for (i = (box2->x1 - box2->x0 + 1) / (2 * job->res.avX) + 1; i > 0; i--)
|
258
|
+
{
|
259
|
+
buffer=append_to_line(buffer," ",&len);
|
260
|
+
j++; /* number of chars in line */
|
261
|
+
}
|
262
|
+
}
|
263
|
+
}
|
264
|
+
else if (box2->c != '\n') {
|
265
|
+
if (j==0 && job->res.avX) /* first char in new line? */ {
|
266
|
+
int indent = box2->x0 - job->res.lines.x0[box2->line];
|
267
|
+
/* correct for angle of page as a whole. */
|
268
|
+
if (job->res.lines.dx)
|
269
|
+
indent += box2->y0 * job->res.lines.dy / job->res.lines.dx;
|
270
|
+
/* subtract the base margin. */
|
271
|
+
indent -= left_margin;
|
272
|
+
if (job->cfg.out_format==XML) { /* subject of change */
|
273
|
+
char s1[255]; /* ToDo: avoid potential buffer overflow !!! */
|
274
|
+
/* output lot of usefull information for XML filter */
|
275
|
+
sprintf(s1," <space x=\"%d\" y=\"%d\" dx=\"%d\" dy=\"%d\" />\n",
|
276
|
+
box2->x0,box2->y0,box2->x1-box2->x0+1,box2->y1-box2->y0+1);
|
277
|
+
buffer=append_to_line(buffer,s1,&len);
|
278
|
+
} else
|
279
|
+
// ToDo: do the multi "\n" earlier in pgm2asc (like multi spacing)
|
280
|
+
for (i = indent / job->res.avX; i > 0; i--) {
|
281
|
+
buffer=append_to_line(buffer," ",&len); j++;
|
282
|
+
}
|
283
|
+
}
|
284
|
+
if (job->cfg.out_format==XML) { /* subject of change */
|
285
|
+
char s1[255]; /* ToDo: avoid potential buffer overflow !!! */
|
286
|
+
/* output lot of usefull information for XML filter */
|
287
|
+
sprintf(s1," <box x=\"%d\" y=\"%d\" dx=\"%d\" dy=\"%d\" value=\"",
|
288
|
+
box2->x0,box2->y0,box2->x1-box2->x0+1,box2->y1-box2->y0+1);
|
289
|
+
buffer=append_to_line(buffer,s1,&len);
|
290
|
+
if (box2->num_ac>1) { /* ToDo: output a list of alternatives */
|
291
|
+
}
|
292
|
+
}
|
293
|
+
if (box2->c != UNKNOWN && box2->c != 0) {
|
294
|
+
buffer=
|
295
|
+
append_to_line(buffer,decode(box2->c,job->cfg.out_format),&len);
|
296
|
+
if (box2->c > ' ' &&
|
297
|
+
box2->c <= 'z') i2++; /* count non-space chars */
|
298
|
+
} else { /* c == UNKNOWN or 0 */
|
299
|
+
wchar_t cc; cc=box2->c;
|
300
|
+
if (box2->num_ac>0 && box2->tas[0]
|
301
|
+
&& (/* job->cfg.out_format!=XML || 2010-10 */ box2->tas[0][0]!='<')
|
302
|
+
) { // 2010-10 output XML code after XML frame, see below!
|
303
|
+
/* output glued chars or ... (?) Jan08 */
|
304
|
+
buffer=append_to_line(buffer,box2->tas[0],&len);
|
305
|
+
j+=strlen(box2->tas[0]);
|
306
|
+
} // else 2010-10-07
|
307
|
+
if (box2->num_ac == 0 || box2->c == UNKNOWN) {
|
308
|
+
/* ToDo: leave string empty? set placeholder per option */
|
309
|
+
/* output dummy string to mark UNKNOWN */
|
310
|
+
if (job->cfg.unrec_marker[0])
|
311
|
+
buffer = append_to_line(buffer, job->cfg.unrec_marker, &len);
|
312
|
+
}
|
313
|
+
}
|
314
|
+
if (job->cfg.out_format==XML) {
|
315
|
+
if (box2->num_ac>0) {
|
316
|
+
/* output alist ToDo: separate <altbox ...> */
|
317
|
+
int i1; char s1[256];
|
318
|
+
sprintf(s1,"\" numac=\"%d\" weights=\"",box2->num_ac);
|
319
|
+
buffer=append_to_line(buffer,s1,&len);
|
320
|
+
for (i1=0;i1<box2->num_ac;i1++) {
|
321
|
+
sprintf(s1,"%d",box2->wac[i1]);
|
322
|
+
buffer=append_to_line(buffer,s1,&len);
|
323
|
+
if (i1+1<box2->num_ac) buffer=append_to_line(buffer,",",&len);
|
324
|
+
}
|
325
|
+
if (box2->num_ac>1)
|
326
|
+
buffer=append_to_line(buffer,"\" achars=\"",&len);
|
327
|
+
for (i1=1;i1<box2->num_ac;i1++) {
|
328
|
+
if (box2->tas[i1] && box2->tas[i1][0]!='<')
|
329
|
+
buffer=append_to_line(buffer,box2->tas[i1],&len);
|
330
|
+
else
|
331
|
+
buffer=append_to_line(buffer,
|
332
|
+
decode(box2->tac[i1],job->cfg.out_format),&len);
|
333
|
+
// ToDo: add tas[] (achars->avalues or alternate_strings?
|
334
|
+
if (i1+1<box2->num_ac) buffer=append_to_line(buffer,",",&len);
|
335
|
+
}
|
336
|
+
}
|
337
|
+
buffer=append_to_line(buffer,"\" />\n",&len);
|
338
|
+
}
|
339
|
+
if (box2->num_ac && box2->tas[0]) {
|
340
|
+
if (box2->tas[0][0]=='<') { /* output special XML object */
|
341
|
+
buffer=append_to_line(buffer,box2->tas[0],&len);
|
342
|
+
if (job->cfg.out_format==XML) // 2010-10-07
|
343
|
+
buffer=append_to_line(buffer,"\n",&len);
|
344
|
+
j+=strlen(box2->tas[0]);
|
345
|
+
}
|
346
|
+
}
|
347
|
+
j++; /* number of chars in line */
|
348
|
+
}
|
349
|
+
i++;
|
350
|
+
} end_for_each(&(job->res.boxlist));
|
351
|
+
if (job->cfg.out_format==XML && oldline>-1) { /* subject of change */
|
352
|
+
buffer=append_to_line(buffer,"</line>\n",&len);
|
353
|
+
}
|
354
|
+
if (job->cfg.out_format==XML) { /* subject of change */
|
355
|
+
buffer=append_to_line(buffer,"</block>\n</page>\n",&len);
|
356
|
+
}
|
357
|
+
|
358
|
+
/* do not forget last line */
|
359
|
+
// is there no \n in the last line? If there is, delete next line.
|
360
|
+
list_app( &(job->res.linelist), (void *)strdup(buffer) );
|
361
|
+
free(buffer);
|
362
|
+
if( job->cfg.verbose&1 )
|
363
|
+
fprintf(stderr,"... %d lines, boxes= %d, chars= %d\n",i,i1,i2);
|
364
|
+
}
|