forked from nanosyzygy/ShuttlePRO
-
Notifications
You must be signed in to change notification settings - Fork 9
/
Copy pathreadconfig.c
1091 lines (991 loc) · 26.4 KB
/
readconfig.c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/*
Copyright 2013 Eric Messick (FixedImagePhoto.com/Contact)
Copyright 2018 Albert Graef <[email protected]>, various improvements
Read and process the configuration file ~/.shuttlepro
Lines starting with # are comments.
Sequence of sections defining translation classes, each section is:
[name] regex
K<1..15> output
S<-7..7> output
I<LR> output
J<LR> output
When focus is on a window whose class or title matches regex, the
following translation class is in effect. An empty regex for the last
class will always match, allowing default translations. Any output
sequences not bound in a matched section will be loaded from the
default section if they are bound there.
Each "[name] regex" line introduces the list of key and shuttle
translations for the named translation class. The name is only used
for debugging output, and needn't be unique. The following lines
with K, S, I and J labels indicate what output should be produced for
the given keypress, shuttle position, shuttle direction, or jog direction.
output is a sequence of one or more key codes with optional up/down
indicators, or strings of printable characters enclosed in double
quotes, separated by whitespace. Sequences bound to keys may have
separate press and release sequences, separated by the word RELEASE.
Examples:
K1 "qwer"
K2 XK_Right
K3 XK_Alt_L/D XK_Right
K4 "V" XK_Left XK_Page_Up "v"
K5 XK_Alt_L/D "v" XK_Alt_L/U "x" RELEASE "q"
Any keycode can be followed by an optional /D, /U, or /H, indicating
that the key is just going down (without being released), going up,
or going down and being held until the shuttlepro key is released.
So, in general, modifier key codes will be followed by /D, and
precede the keycodes they are intended to modify. If a sequence
requires different sets of modifiers for different keycodes, /U can
be used to release a modifier that was previously pressed with /D.
At the end of shuttle and jog sequences, all down keys will be
released.
Keypresses translate to separate press and release sequences.
At the end of the press sequence for key sequences, all down keys
marked by /D will be released, and the last key not marked by /D,
/U, or /H will remain pressed. The release sequence will begin by
releasing the last held key. If keys are to be pressed as part of
the release sequence, then any keys marked with /D will be repressed
before continuing the sequence. Keycodes marked with /H remain held
between the press and release sequences.
JACK MIDI support (added by [email protected] Fri Aug 3 11:01:32 CEST 2018):
The MIDI output option is useful if you want to be able to send control
messages from a Shuttle device to any kind of MIDI-capable program, such as
a synthesizer or a DAW. Also, if you put the MIDI translations into a
special "MIDI" default section of the shuttlerc file, then the target
application will be able to receive data from the device no matter which
window has the keyboard focus. (This special "MIDI" section will only be
active if Jack MIDI support is actually enabled with the -o option, see
below. This allows you to have another default section after the MIDI
section for non-MIDI operation.)
To enable MIDI output, add the -o option when invoking the program (also,
you can set the Jack client name with the -j option, and the -dj option can
be used to get verbose output from Jack if needed). This causes a Jack
client (named "shuttlepro" by default) with a single MIDI output port to be
created, and will also start up Jack if it is not already running. Any MIDI
messages in the translations will be sent on that port. You can then use any
Jack patchbay such as qjackctl to connect the output to any other Jack MIDI
client (use the a2jmidid program to connect to non-Jack ALSA MIDI
applications).
*/
#include "shuttle.h"
int default_debug_regex = 0;
int default_debug_strokes = 0;
int default_debug_keys = 0;
int debug_regex = 0;
int debug_strokes = 0;
int debug_keys = 0;
int midi_octave = 0;
char *
allocate(size_t len)
{
char *ret = (char *)malloc(len);
if (ret == NULL) {
fprintf(stderr, "Out of memory!\n");
exit(1);
}
return ret;
}
char *
alloc_strcat(char *a, char *b)
{
size_t len = 0;
char *result;
if (a != NULL) {
len += strlen(a);
}
if (b != NULL) {
len += strlen(b);
}
result = allocate(len+1);
result[0] = '\0';
if (a != NULL) {
strcpy(result, a);
}
if (b != NULL) {
strcat(result, b);
}
return result;
}
static char *read_line_buffer = NULL;
static int read_line_buffer_length = 0;
#define BUF_GROWTH_STEP 1024
// read a line of text from the given file into a managed buffer.
// returns a partial line at EOF if the file does not end with \n.
// exits with error message on read error.
char *
read_line(FILE *f, char *name)
{
int pos = 0;
char *new_buffer;
int new_buffer_length;
if (read_line_buffer == NULL) {
read_line_buffer_length = BUF_GROWTH_STEP;
read_line_buffer = allocate(read_line_buffer_length);
read_line_buffer[0] = '\0';
}
while (1) {
read_line_buffer[read_line_buffer_length-1] = '\377';
if (fgets(read_line_buffer+pos, read_line_buffer_length-pos, f) == NULL) {
if (feof(f)) {
if (pos > 0) {
// partial line at EOF
return read_line_buffer;
} else {
return NULL;
}
}
perror(name);
exit(1);
}
if (read_line_buffer[read_line_buffer_length-1] != '\0') {
return read_line_buffer;
}
if (read_line_buffer[read_line_buffer_length-2] == '\n') {
return read_line_buffer;
}
new_buffer_length = read_line_buffer_length + BUF_GROWTH_STEP;
new_buffer = allocate(new_buffer_length);
memcpy(new_buffer, read_line_buffer, read_line_buffer_length);
free(read_line_buffer);
pos = read_line_buffer_length-1;
read_line_buffer = new_buffer;
read_line_buffer_length = new_buffer_length;
}
}
static translation *first_translation_section = NULL;
static translation *last_translation_section = NULL;
translation *default_translation;
translation *
new_translation_section(char *name, char *regex)
{
translation *ret = (translation *)allocate(sizeof(translation));
int err;
int i;
if (debug_strokes) {
printf("------------------------\n[%s] %s\n\n", name, regex);
}
ret->next = NULL;
ret->name = alloc_strcat(name, NULL);
if (regex == NULL || *regex == '\0') {
ret->is_default = 1;
default_translation = ret;
} else {
ret->is_default = 0;
err = regcomp(&ret->regex, regex, REG_EXTENDED|REG_NOSUB);
if (err != 0) {
regerror(err, &ret->regex, read_line_buffer, read_line_buffer_length);
fprintf(stderr, "error compiling regex for [%s]: %s\n", name, read_line_buffer);
regfree(&ret->regex);
free(ret->name);
free(ret);
return NULL;
}
}
for (i=0; i<NUM_KEYS; i++) {
ret->key_down[i] = NULL;
ret->key_up[i] = NULL;
}
for (i=0; i<NUM_SHUTTLES; i++) {
ret->shuttle[i] = NULL;
}
for (i=0; i<NUM_SHUTTLE_INCRS; i++) {
ret->shuttle_incr[i] = NULL;
}
for (i=0; i<NUM_JOGS; i++) {
ret->jog[i] = NULL;
}
if (first_translation_section == NULL) {
first_translation_section = ret;
last_translation_section = ret;
} else {
last_translation_section->next = ret;
last_translation_section = ret;
}
return ret;
}
void
free_strokes(stroke *s)
{
stroke *next;
while (s != NULL) {
next = s->next;
free(s);
s = next;
}
}
void
free_translation_section(translation *tr)
{
int i;
if (tr != NULL) {
free(tr->name);
if (!tr->is_default) {
regfree(&tr->regex);
}
for (i=0; i<NUM_KEYS; i++) {
free_strokes(tr->key_down[i]);
free_strokes(tr->key_up[i]);
}
for (i=0; i<NUM_SHUTTLES; i++) {
free_strokes(tr->shuttle[i]);
}
for (i=0; i<NUM_SHUTTLE_INCRS; i++) {
free_strokes(tr->shuttle_incr[i]);
}
for (i=0; i<NUM_JOGS; i++) {
free_strokes(tr->jog[i]);
}
free(tr);
}
}
void
free_all_translations(void)
{
translation *tr = first_translation_section;
translation *next;
while (tr != NULL) {
next = tr->next;
free_translation_section(tr);
tr = next;
}
first_translation_section = NULL;
last_translation_section = NULL;
}
char *config_file_name = NULL;
static time_t config_file_modification_time;
static char *token_src = NULL;
// similar to strtok, but it tells us what delimiter was found at the
// end of the token, handles double quoted strings specially, and
// hardcodes the delimiter set.
char *
token(char *src, char *delim_found)
{
char *delims = " \t\n/\"";
char *d;
char *token_start;
if (src == NULL) {
src = token_src;
}
if (src == NULL) {
*delim_found = '\0';
return NULL;
}
token_start = src;
while (*src) {
d = delims;
while (*d && *src != *d) {
d++;
}
if (*d) {
if (src == token_start) {
src++;
token_start = src;
if (*d == '"') {
while (*src && *src != '"' && *src != '\n') {
src++;
}
} else {
continue;
}
}
*delim_found = *d;
if (*src) {
*src = '\0';
token_src = src+1;
} else {
token_src = NULL;
}
return token_start;
}
src++;
}
token_src = NULL;
*delim_found = '\0';
if (src == token_start) {
return NULL;
}
return token_start;
}
typedef struct _keysymmapping {
char *str;
KeySym sym;
} keysymmapping;
static keysymmapping key_sym_mapping[] = {
#include "keys.h"
{ "XK_Button_1", XK_Button_1 },
{ "XK_Button_2", XK_Button_2 },
{ "XK_Button_3", XK_Button_3 },
{ "XK_Scroll_Up", XK_Scroll_Up },
{ "XK_Scroll_Down", XK_Scroll_Down },
{ NULL, 0 }
};
KeySym
string_to_KeySym(char *str)
{
size_t len = strlen(str) + 1;
int i = 0;
while (key_sym_mapping[i].str != NULL) {
if (!strncmp(str, key_sym_mapping[i].str, len)) {
return key_sym_mapping[i].sym;
}
i++;
}
return 0;
}
char *
KeySym_to_string(KeySym ks)
{
int i = 0;
while (key_sym_mapping[i].sym != 0) {
if (key_sym_mapping[i].sym == ks) {
return key_sym_mapping[i].str;
}
i++;
}
return NULL;
}
static char *note_names[] = { "C", "C#", "D", "Eb", "E", "F", "F#", "G", "G#", "A", "Bb", "B" };
void
print_stroke(stroke *s)
{
char *str;
if (s != NULL) {
if (s->keysym) {
str = KeySym_to_string(s->keysym);
if (str == NULL) {
printf("0x%x", (int)s->keysym);
str = "???";
}
printf("%s/%c ", str, s->press ? 'D' : 'U');
} else {
int status = s->status & 0xf0;
int channel = (s->status & 0x0f) + 1;
switch (status) {
case 0x90:
printf("%s%d-%d ", note_names[s->data % 12],
s->data / 12 + midi_octave, channel);
break;
case 0xb0:
printf("CC%d-%d%s ", s->data, channel, s->incr?"~":"");
break;
case 0xc0:
printf("PC%d-%d ", s->data, channel);
break;
case 0xe0:
printf("PB-%d ", channel);
break;
default: // this can't happen
break;
}
}
}
}
void
print_stroke_sequence(char *name, char *up_or_down, stroke *s)
{
if (up_or_down && *up_or_down)
printf("%s[%s]: ", name, up_or_down);
else
printf("%s: ", name);
while (s) {
print_stroke(s);
s = s->next;
}
printf("\n");
}
stroke **first_stroke;
stroke *last_stroke;
stroke **press_first_stroke;
stroke **release_first_stroke;
int is_keystroke;
int is_midi;
char *current_translation;
char *key_name;
int first_release_stroke; // is this the first stroke of a release?
KeySym regular_key_down;
#define NUM_MODIFIERS 64
stroke modifiers_down[NUM_MODIFIERS];
int modifier_count;
int midi_channel;
void
append_stroke(KeySym sym, int press)
{
stroke *s = (stroke *)allocate(sizeof(stroke));
s->next = NULL;
s->keysym = sym;
s->press = press;
s->status = s->data = s->incr = s->dirty = 0;
if (*first_stroke) {
last_stroke->next = s;
} else {
*first_stroke = s;
}
last_stroke = s;
}
void
append_midi(int status, int data, int incr)
{
stroke *s = (stroke *)allocate(sizeof(stroke));
s->next = NULL;
s->keysym = 0;
s->press = 0;
s->status = status;
s->data = data;
s->incr = incr;
// if this is a keystroke event, for all messages but program change (which
// has no "on" and "off" states), mark the event as "dirty" so that the
// corresponding "off" event gets added later to the "release" strokes
s->dirty = is_keystroke && ((status&0xf0) != 0xc0);
if (*first_stroke) {
last_stroke->next = s;
} else {
*first_stroke = s;
}
last_stroke = s;
is_midi = 1;
}
// s->press values in modifiers_down:
// PRESS -> down
// HOLD -> held
// PRESS_RELEASE -> released, but to be re-pressed if necessary
// RELEASE -> up
void
mark_as_down(KeySym sym, int hold)
{
int i;
for (i=0; i<modifier_count; i++) {
if (modifiers_down[i].keysym == sym) {
modifiers_down[i].press = hold ? HOLD : PRESS;
return;
}
}
if (modifier_count > NUM_MODIFIERS) {
fprintf(stderr, "too many modifiers down in [%s]%s\n", current_translation, key_name);
return;
}
modifiers_down[modifier_count].keysym = sym;
modifiers_down[modifier_count].press = hold ? HOLD : PRESS;
modifier_count++;
}
void
mark_as_up(KeySym sym)
{
int i;
for (i=0; i<modifier_count; i++) {
if (modifiers_down[i].keysym == sym) {
modifiers_down[i].press = RELEASE;
return;
}
}
}
void
release_modifiers(int allkeys)
{
int i;
for (i=0; i<modifier_count; i++) {
if (modifiers_down[i].press == PRESS) {
append_stroke(modifiers_down[i].keysym, 0);
modifiers_down[i].press = PRESS_RELEASE;
} else if (allkeys && modifiers_down[i].press == HOLD) {
append_stroke(modifiers_down[i].keysym, 0);
modifiers_down[i].press = RELEASE;
}
}
}
void
re_press_temp_modifiers(void)
{
int i;
for (i=0; i<modifier_count; i++) {
if (modifiers_down[i].press == PRESS_RELEASE) {
append_stroke(modifiers_down[i].keysym, 1);
modifiers_down[i].press = PRESS;
}
}
}
int
start_translation(translation *tr, char *which_key)
{
char c;
int k;
int n;
//printf("start_translation(%s)\n", which_key);
if (tr == NULL) {
fprintf(stderr, "need to start translation section before defining key: %s\n", which_key);
return 1;
}
current_translation = tr->name;
key_name = which_key;
is_keystroke = is_midi = 0;
first_release_stroke = 0;
regular_key_down = 0;
modifier_count = 0;
midi_channel = 0;
if (tolower(which_key[0]) == 'j' &&
(tolower(which_key[1]) == 'l' || tolower(which_key[1]) == 'r') &&
which_key[2] == '\0') {
// JL, JR
k = tolower(which_key[1]) == 'l' ? 0 : 1;
first_stroke = &(tr->jog[k]);
} else if (tolower(which_key[0]) == 'i' &&
(tolower(which_key[1]) == 'l' || tolower(which_key[1]) == 'r') &&
which_key[2] == '\0') {
// IL, IR
k = tolower(which_key[1]) == 'l' ? 0 : 1;
first_stroke = &(tr->shuttle_incr[k]);
} else {
n = 0;
sscanf(which_key, "%c%d%n", &c, &k, &n);
if (n != (int)strlen(which_key)) {
fprintf(stderr, "bad key name: [%s]%s\n", current_translation, which_key);
return 1;
}
switch (c) {
case 'k':
case 'K':
// K1 .. K15
k = k - 1;
if (k < 0 || k >= NUM_KEYS) {
fprintf(stderr, "bad key name: [%s]%s\n", current_translation, which_key);
return 1;
}
first_stroke = &(tr->key_down[k]);
release_first_stroke = &(tr->key_up[k]);
is_keystroke = 1;
break;
case 's':
case 'S':
// S-7 .. S7
if (k < -7 || k > 7) {
fprintf(stderr, "bad key name: [%s]%s\n", current_translation, which_key);
return 1;
}
first_stroke = &(tr->shuttle[k+7]);
break;
default:
fprintf(stderr, "bad key name: [%s]%s\n", current_translation, which_key);
return 1;
}
}
if (*first_stroke != NULL) {
fprintf(stderr, "can't redefine key: [%s]%s\n", current_translation, which_key);
return 1;
}
press_first_stroke = first_stroke;
return 0;
}
void
add_keysym(KeySym sym, int press_release)
{
//printf("add_keysym(0x%x, %d)\n", (int)sym, press_release);
switch (press_release) {
case PRESS:
append_stroke(sym, 1);
mark_as_down(sym, 0);
break;
case RELEASE:
append_stroke(sym, 0);
mark_as_up(sym);
break;
case HOLD:
append_stroke(sym, 1);
mark_as_down(sym, 1);
break;
case PRESS_RELEASE:
default:
if (first_release_stroke) {
re_press_temp_modifiers();
}
if (regular_key_down != 0) {
append_stroke(regular_key_down, 0);
}
append_stroke(sym, 1);
regular_key_down = sym;
first_release_stroke = 0;
break;
}
}
void
add_release(int all_keys)
{
//printf("add_release(%d)\n", all_keys);
release_modifiers(all_keys);
if (!all_keys) {
first_stroke = release_first_stroke;
if (is_midi) {
// walk the list of "press" strokes, find all "dirty" (as yet unhandled)
// MIDI events in there and add them to the "release" strokes
stroke *s = *press_first_stroke;
while (s) {
if (!s->keysym && s->dirty) {
append_midi(s->status, s->data, s->incr);
s->dirty = 0;
}
s = s->next;
}
}
}
if (regular_key_down != 0) {
append_stroke(regular_key_down, 0);
}
regular_key_down = 0;
first_release_stroke = 1;
}
void
add_keystroke(char *keySymName, int press_release)
{
KeySym sym;
if (is_keystroke && !strncmp(keySymName, "RELEASE", 8)) {
add_release(0);
return;
}
sym = string_to_KeySym(keySymName);
if (sym != 0) {
add_keysym(sym, press_release);
} else {
fprintf(stderr, "unrecognized keysym: %s\n", keySymName);
}
}
void
add_string(char *str)
{
while (str && *str) {
if (*str >= ' ' && *str <= '~') {
add_keysym((KeySym)(*str), PRESS_RELEASE);
}
str++;
}
}
/* Parser for the MIDI message syntax. The syntax we actually parse here is:
tok ::= ( note | msg ) [number] [ "-" number] [ "~" ]
note ::= ( "a" | ... | "g" ) [ "#" | "b" ]
msg ::= "ch" | "pb" | "pc" | "cc"
Numbers are always in decimal. The meaning of the first number depends on
the context (octave number for notes, the actual data byte for other
messages). If present, the suffix with the second number (after the dash)
denotes the MIDI channel, otherwise the default MIDI channel is used. Also,
for "cc" messages the "~" suffix denotes a special kind of "incremental"
(endless rotary) control whose values are encoded as relative changes in
"sign bit format". Note that not all combinations are possible -- "pb" is
*not* followed by a data byte, only "cc" may have the "~" suffix on it, and
"ch" may *not* have a channel number suffix on it. (In fact, "ch" is no
real MIDI message at all; it just sets the default MIDI channel for
subsequent messages.) */
static int note_number(char c, char b, int k)
{
c = tolower(c); b = tolower(b);
if (c < 'a' || c > 'g' || (b && b != '#' && b != 'b'))
return -1; // either wrong note name or invalid accidental
else {
static int note_numbers[] = { -3, -1, 0, 2, 4, 5, 7 };
int m = note_numbers[c-'a'], a = (b=='#')?1:(b=='b')?-1:0;
if (m<0) k++;
return m + a + 12*k;
}
}
int
parse_midi(char *tok, char *s, int *status, int *data, int *incr)
{
char *p = tok, *t;
int n, m = -1, k = midi_channel;
s[0] = 0;
while (*p && !isdigit(*p)) p++;
if (p == tok || p-tok > 10) return 0; // no valid token
// the token by itself
strncpy(s, tok, p-tok); s[p-tok] = 0;
// normalize to lowercase
for (t = s; *t; t++) *t = tolower(*t);
// octave number or data byte (not permitted with 'pb', otherwise required)
if (isdigit(*p) && sscanf(p, "%d%n", &m, &n) == 1) {
if (strcmp(s, "pb") == 0) return 0;
p += n;
} else if (strcmp(s, "pb")) {
return 0;
}
if (*p == '-') {
// suffix with MIDI channel (not permitted with 'ch')
if (strcmp(s, "ch") == 0) return 0;
if (sscanf(++p, "%d%n", &k, &n) == 1) {
// check that it is a valid channel number
if (k < 1 || k > 16) return 0;
k--; // actual MIDI channel in the range 0..15
p += n;
} else {
return 0;
}
}
if (*p == '~') {
// incremental bit (only valid for 'cc')
if (strcmp(s, "cc")) return 0;
*incr = 1;
p++;
} else {
*incr = 0;
}
// check for trailing garbage
if (*p) return 0;
if (strcmp(s, "ch") == 0) {
// we return a bogus status of 0 here, along with the MIDI channel in the
// data byte; also check that the MIDI channel is in the proper range
if (m < 1 || m > 16) return 0;
*status = 0; *data = m-1;
return 1;
} else if (strcmp(s, "pb") == 0) {
// pitch bend, no data byte
*status = 0xe0 | k; *data = 0;
return 1;
} else if (strcmp(s, "pc") == 0) {
// program change
if (m < 0 || m > 127) return 0;
*status = 0xc0 | k; *data = m;
return 1;
} else if (strcmp(s, "cc") == 0) {
// control change
if (m < 0 || m > 127) return 0;
*status = 0xb0 | k; *data = m;
return 1;
} else {
// we must be looking at a MIDI note here, with m denoting the octave
// number; first character is the note name (must be a..g); optionally,
// the second character may denote an accidental (# or b)
n = note_number(s[0], s[1], m - midi_octave);
if (n < 0 || n > 127) return 0;
*status = 0x90 | k; *data = n;
return 1;
}
}
void
add_midi(char *tok)
{
int status, data, incr;
char buf[100];
if (parse_midi(tok, buf, &status, &data, &incr)) {
if (status == 0) {
// 'ch' token; this doesn't actually generate any output, it just sets
// the default MIDI channel
midi_channel = data;
} else {
append_midi(status, data, incr);
}
} else {
// inspect the token that was actually recognized (if any) to give some
// useful error message here
if (strcmp(buf, "ch"))
fprintf(stderr, "unrecognized keysym: %s\n", tok);
else
fprintf(stderr, "invalid MIDI channel: %s\n", tok);
}
}
void
finish_translation(void)
{
//printf("finish_translation()\n");
if (is_keystroke) {
add_release(0);
}
add_release(1);
if (debug_strokes) {
if (is_keystroke) {
print_stroke_sequence(key_name, "D", *press_first_stroke);
print_stroke_sequence(key_name, "U", *release_first_stroke);
} else {
print_stroke_sequence(key_name, "", *first_stroke);
}
printf("\n");
}
}
int
read_config_file(void)
{
struct stat buf;
char *home;
char *line;
char *s;
char *name = NULL;
char *regex;
char *tok;
char *which_key;
char *updown;
char delim;
translation *tr = NULL;
FILE *f;
int config_file_default = 0;
if (config_file_name == NULL) {
config_file_name = getenv("SHUTTLE_CONFIG_FILE");
if (config_file_name == NULL) {
home = getenv("HOME");
config_file_name = alloc_strcat(home, "/.shuttlerc");
config_file_default = 1;
} else {
config_file_name = alloc_strcat(config_file_name, NULL);
}
config_file_modification_time = 0;
}
if (stat(config_file_name, &buf) < 0) {
// AG: Fall back to the system-wide configuration file.
if (!config_file_default) perror(config_file_name);
config_file_name = "/etc/shuttlerc";
config_file_modification_time = 0;
}
if (stat(config_file_name, &buf) < 0) {
perror(config_file_name);
return 0;
}
if (buf.st_mtime == 0) {
buf.st_mtime = 1;
}
if (buf.st_mtime > config_file_modification_time) {
config_file_modification_time = buf.st_mtime;
if (default_debug_regex || default_debug_strokes || default_debug_keys) {
printf("Loading configuration: %s\n", config_file_name);
}
f = fopen(config_file_name, "r");
if (f == NULL) {
perror(config_file_name);
return 0;
}
free_all_translations();
debug_regex = default_debug_regex;
debug_strokes = default_debug_strokes;
debug_keys = default_debug_keys;
midi_octave = 0;
while ((line=read_line(f, config_file_name)) != NULL) {
//printf("line: %s", line);
s = line;
while (*s && isspace(*s)) {
s++;
}
if (*s == '#') {
continue;
}
if (*s == '[') {
// [name] regex\n
name = ++s;
while (*s && *s != ']') {
s++;
}
regex = NULL;
if (*s) {
*s = '\0';
s++;
while (*s && isspace(*s)) {
s++;
}
regex = s;
while (*s) {
s++;
}
s--;
while (s > regex && isspace(*s)) {
s--;
}
s[1] = '\0';
}
tr = new_translation_section(name, regex);
continue;
}
tok = token(s, &delim);
if (tok == NULL) {
continue;
}
if (!strcmp(tok, "DEBUG_REGEX")) {
debug_regex = 1;
continue;
}
if (!strcmp(tok, "DEBUG_STROKES")) {
debug_strokes = 1;
continue;
}
if (!strcmp(tok, "DEBUG_KEYS")) {
debug_keys = 1;
continue;
}
if (!strncmp(tok, "MIDI_OCTAVE", 11)) {
char *a = tok+11;
int k, n;