3 * selector is a simple command line utility for selection of strings
4 * with a dynamic pattern-matching.
6 * Copyright (c) 2009 Francois Fleuret
7 * Written by Francois Fleuret <francois@fleuret.org>
9 * This file is part of selector.
11 * selector is free software: you can redistribute it and/or modify
12 * it under the terms of the GNU General Public License version 3 as
13 * published by the Free Software Foundation.
15 * selector is distributed in the hope that it will be useful, but
16 * WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 * General Public License for more details.
20 * You should have received a copy of the GNU General Public License
21 * along with selector. If not, see <http://www.gnu.org/licenses/>.
25 // To use it as a super-history-search for bash:
26 // alias h='selector -d -i -b -v -f <(history)'
36 #include <sys/ioctl.h>
44 const int buffer_size = 4096;
46 // Yeah, global variables!
48 int nb_lines_max = 1000;
49 char pattern_separator = ';';
50 char label_separator = '\0';
51 int output_to_vt_buffer = 0;
52 int add_control_qs = 0;
54 int zsh_history = 0, bash_history = 0;
55 int inverse_order = 0;
56 int remove_duplicates = 0;
58 int case_sensitive = 0;
62 #define COLOR_MODELINE 1
63 #define COLOR_HIGHLIGHTED_LINE 2
65 //////////////////////////////////////////////////////////////////////
67 void inject_into_tty_buffer(char *string) {
68 struct termios oldtio, newtio;
69 tcgetattr(STDIN_FILENO, &oldtio);
70 memset(&newtio, 0, sizeof(newtio));
71 // Set input mode (non-canonical, *no echo*,...)
72 tcsetattr(STDIN_FILENO, TCSANOW, &newtio);
73 const char control_q = '\021';
74 // Put the selected string in the tty input buffer
75 for(const char *k = string; *k; k++) {
76 if(add_control_qs && !(*k >= ' ' && *k <= '~')) {
77 // Add ^Q to quote control characters
78 ioctl(STDIN_FILENO, TIOCSTI, &control_q);
80 ioctl(STDIN_FILENO, TIOCSTI, k);
82 // Restore the old settings
83 tcsetattr(STDIN_FILENO, TCSANOW, &oldtio);
86 //////////////////////////////////////////////////////////////////////
88 void check_opt(int argc, char **argv, int n_opt, int n, const char *help) {
89 if(n_opt + n >= argc) {
90 cerr << "Missing argument for " << argv[n_opt] << "."
92 << "Expecting " << help << "."
98 int string_to_positive_integer(char *string) {
103 for(char *s = string; *s; s++) {
104 if(*s >= '0' && *s <= '9') {
105 result = result * 10 + int(*s - '0');
111 cerr << "Value `" << string << "' is not a positive integer." << endl;
118 void error_feedback() {
126 //////////////////////////////////////////////////////////////////////
127 // A quick and dirty hash table
129 // The table itself stores index of the strings in a char
130 // **table. When a string is added, if it was already in the table,
131 // the new index replaces the previous one.
133 int *new_hash_table(int hash_table_size) {
135 result = new int[hash_table_size];
136 for(int k = 0; k < hash_table_size; k++) {
142 // Adds new_string in the table, associated to new_index. If this
143 // string was not already in the table, returns -1. Otherwise, returns
144 // the previous index it had.
146 int test_and_add(char *new_string, int new_index,
147 char **strings, int *hash_table, int hash_table_size) {
149 unsigned int code = 0;
151 // This is my recipe. I checked, it seems to work (as long as
152 // hash_table_size is not a multiple of 387433 that should be okay)
154 for(int k = 0; new_string[k]; k++) {
155 code = code * 387433 + (unsigned int) (new_string[k]);
158 code = code % hash_table_size;
160 while(hash_table[code] >= 0) {
161 // There is a string with that code
162 if(strcmp(new_string, strings[hash_table[code]]) == 0) {
163 // It is the same string, we keep a copy of the stored index
164 int result = hash_table[code];
166 hash_table[code] = new_index;
167 // And return the previous one
170 // This collision was not the same string, let's move to the next
172 code = (code + 1) % hash_table_size;
175 // This string was not already in there, store the index in the
176 // table and return -1
177 hash_table[code] = new_index;
181 //////////////////////////////////////////////////////////////////////
182 // A matcher matches either with a collection of substrings, or with a
190 char *splitted_patterns, **patterns;
193 int match(char *string, matcher_t *matcher) {
194 if(matcher->nb_patterns >= 0) {
195 if(matcher->case_sensitive) {
196 for(int n = 0; n < matcher->nb_patterns; n++) {
197 if(strstr(string, matcher->patterns[n]) == 0) return 0;
200 for(int n = 0; n < matcher->nb_patterns; n++) {
201 if(strcasestr(string, matcher->patterns[n]) == 0) return 0;
206 return regexec(&matcher->preg, string, 0, 0, 0) == 0;
210 void free_matcher(matcher_t *matcher) {
211 if(matcher->nb_patterns < 0) {
212 if(!matcher->regexp_error) regfree(&matcher->preg);
214 delete[] matcher->splitted_patterns;
215 delete[] matcher->patterns;
219 void initialize_matcher(int use_regexp, int case_sensitive,
220 matcher_t *matcher, const char *pattern) {
223 matcher->nb_patterns = -1;
224 matcher->regexp_error = regcomp(&matcher->preg, pattern, case_sensitive ? 0 : REG_ICASE);
226 matcher->regexp_error = 0;
227 matcher->nb_patterns = 1;
228 matcher->case_sensitive = case_sensitive;
230 for(const char *s = pattern; *s; s++) {
231 if(*s == pattern_separator) {
232 matcher->nb_patterns++;
236 matcher->splitted_patterns = new char[strlen(pattern) + 1];
237 matcher->patterns = new char*[matcher->nb_patterns];
239 strcpy(matcher->splitted_patterns, pattern);
242 char *last_pattern_start = matcher->splitted_patterns;
243 for(char *s = matcher->splitted_patterns; n < matcher->nb_patterns; s++) {
244 if(*s == pattern_separator || *s == '\0') {
246 matcher->patterns[n++] = last_pattern_start;
247 last_pattern_start = s + 1;
253 //////////////////////////////////////////////////////////////////////
256 void delete_char(char *buffer, int *position) {
257 if(buffer[*position]) {
259 while(c < buffer_size && buffer[c]) {
260 buffer[c] = buffer[c+1];
263 } else error_feedback();
266 void backspace_char(char *buffer, int *position) {
268 if(buffer[*position]) {
269 int c = *position - 1;
271 buffer[c] = buffer[c+1];
275 buffer[*position - 1] = '\0';
279 } else error_feedback();
282 void insert_char(char *buffer, int *position, char character) {
283 if(strlen(buffer) < buffer_size - 1) {
285 char t = buffer[c], u;
294 buffer[(*position)++] = character;
295 } else error_feedback();
298 void kill_before_cursor(char *buffer, int *position) {
300 while(buffer[*position + s]) {
301 buffer[s] = buffer[*position + s];
308 void kill_after_cursor(char *buffer, int *position) {
309 buffer[*position] = '\0';
312 //////////////////////////////////////////////////////////////////////
314 int previous_visible(int current_line, int nb_lines, char **lines, matcher_t *matcher) {
315 int line = current_line - 1;
316 while(line >= 0 && !match(lines[line], matcher)) line--;
320 int next_visible(int current_line, int nb_lines, char **lines, matcher_t *matcher) {
321 int line = current_line + 1;
322 while(line < nb_lines && !match(lines[line], matcher)) line++;
330 //////////////////////////////////////////////////////////////////////
332 void update_screen(int *current_line, int *temporary_line, int motion,
333 int nb_lines, char **lines,
337 char buffer[buffer_size];
340 initialize_matcher(use_regexp, case_sensitive, &matcher, pattern);
342 // We now take care of printing the lines per se
344 int console_width = getmaxx(stdscr);
345 int console_height = getmaxy(stdscr);
347 // First, we find a visible line. In priority: The current, or the
348 // first visible after it, or the first visible before it.
350 int nb_printed_lines = 0;
352 use_default_colors();
355 if(matcher.regexp_error) {
356 addstr("[regexp error]");
357 } else if(nb_lines > 0) {
359 if(match(lines[*current_line], &matcher)) {
360 new_line = *current_line;
362 new_line = next_visible(*current_line, nb_lines, lines, &matcher);
364 new_line = previous_visible(*current_line, nb_lines, lines, &matcher);
368 // If we found a visible line and we should move, let's move
370 if(new_line >= 0 && motion != 0) {
373 // We want to go down, let's find the first visible line below
374 for(int m = 0; l >= 0 && m < motion; m++) {
375 l = next_visible(l, nb_lines, lines, &matcher);
381 // We want to go up, let's find the first visible line above
382 for(int m = 0; l >= 0 && m < -motion; m++) {
383 l = previous_visible(l, nb_lines, lines, &matcher);
391 // Here new_line is either a line number matching the patterns, or -1
395 int first_line = new_line, last_line = new_line, nb_match = 1;
397 // We find the first and last line to show, so that the total of
398 // visible lines between them (them included) is console_height-1
400 while(nb_match < console_height-1 && (first_line > 0 || last_line < nb_lines - 1)) {
404 while(first_line > 0 && !match(lines[first_line], &matcher)) {
407 if(match(lines[first_line], &matcher)) {
412 if(nb_match < console_height - 1 && last_line < nb_lines - 1) {
414 while(last_line < nb_lines - 1 && !match(lines[last_line], &matcher)) {
418 if(match(lines[last_line], &matcher)) {
424 // Now we display them
426 for(int l = first_line; l <= last_line; l++) {
427 if(match(lines[l], &matcher)) {
430 while(lines[l][k] && k < buffer_size - 2 && k < console_width - 2) {
431 buffer[k] = lines[l][k];
435 // We fill the rest of the line with blanks if this is the
439 while(k < console_width) {
449 // Highlight the highlighted line ...
453 attron(COLOR_PAIR(COLOR_HIGHLIGHTED_LINE));
454 addnstr(buffer, console_width);
455 attroff(COLOR_PAIR(COLOR_HIGHLIGHTED_LINE));
458 addnstr(buffer, console_width);
462 addnstr(buffer, console_width);
470 *current_line = new_line;
474 *temporary_line = new_line;
476 if(nb_printed_lines == 0) {
477 addnstr("[no selection]\n", console_width);
480 addnstr("[empty choice]\n", console_width);
490 attron(COLOR_PAIR(COLOR_MODELINE));
495 for(int k = 0; k < console_width; k++) buffer[k] = ' ';
496 buffer[console_width] = '\0';
497 addnstr(buffer, console_width);
501 // There must be a more elegant way of moving the cursor at a
502 // location met during display
509 cursor_x += strlen(title) + 1;
512 sprintf(buffer, "%d/%d ", nb_printed_lines, nb_lines);
514 cursor_x += strlen(buffer);
516 addnstr(pattern, cursor_position);
517 cursor_x += cursor_position;
519 if(pattern[cursor_position]) {
520 addstr(pattern + cursor_position);
525 if(use_regexp || case_sensitive) {
543 attroff(COLOR_PAIR(COLOR_MODELINE));
551 free_matcher(&matcher);
554 //////////////////////////////////////////////////////////////////////
556 void read_file(const char *input_filename,
557 int nb_lines_max, int *nb_lines, char **lines,
558 int hash_table_size, int *hash_table) {
560 char raw_line[buffer_size];;
562 ifstream file(input_filename);
565 cerr << "Can not open " << input_filename << endl;
569 while(*nb_lines < nb_lines_max && !file.eof()) {
571 file.getline(raw_line, buffer_size);
576 cerr << "Line too long:" << endl;
577 cerr << raw_line << endl;
585 // Remove the zsh history prefix
587 if(zsh_history && *t == ':') {
588 while(*t && *t != ';') t++;
592 // Remove the bash history prefix
595 while(*t == ' ') t++;
596 while(*t >= '0' && *t <= '9') t++;
597 while(*t == ' ') t++;
600 // Check for duplicates with the hash table and insert the line
601 // in the list if necessary
606 dup = test_and_add(t, *nb_lines, lines, hash_table, hash_table_size);
612 lines[*nb_lines] = new char[strlen(t) + 1];
613 strcpy(lines[*nb_lines], t);
615 // The string was already in there, so we do not allocate a
616 // new string but use the pointer to the first occurence of it
617 lines[*nb_lines] = lines[dup];
626 //////////////////////////////////////////////////////////////////////
628 int main(int argc, char **argv) {
630 if(!ttyname(STDIN_FILENO)) {
631 cerr << "The standard input is not a tty." << endl;
635 int color_fg_modeline, color_bg_modeline;
636 int color_fg_highlight, color_bg_highlight;
638 color_fg_modeline = COLOR_WHITE;
639 color_bg_modeline = COLOR_BLACK;
640 color_fg_highlight = COLOR_BLACK;
641 color_bg_highlight = COLOR_YELLOW;
643 setlocale(LC_ALL, "");
645 char input_filename[buffer_size], output_filename[buffer_size];
647 strcpy(input_filename, "");
648 strcpy(output_filename, "");
651 int error = 0, show_help = 0;
652 int rest_are_files = 0;
654 while(!error && !show_help && i < argc && argv[i][0] == '-' && !rest_are_files) {
656 if(strcmp(argv[i], "-o") == 0) {
657 check_opt(argc, argv, i, 1, "<output filename>");
658 strncpy(output_filename, argv[i+1], buffer_size);
662 else if(strcmp(argv[i], "-s") == 0) {
663 check_opt(argc, argv, i, 1, "<pattern separator>");
664 pattern_separator = argv[i+1][0];
668 else if(strcmp(argv[i], "-x") == 0) {
669 check_opt(argc, argv, i, 1, "<label separator>");
670 label_separator = argv[i+1][0];
674 else if(strcmp(argv[i], "-v") == 0) {
675 output_to_vt_buffer = 1;
679 else if(strcmp(argv[i], "-w") == 0) {
684 else if(strcmp(argv[i], "-m") == 0) {
689 else if(strcmp(argv[i], "-q") == 0) {
694 else if(strcmp(argv[i], "-f") == 0) {
695 check_opt(argc, argv, i, 1, "<input filename>");
696 strncpy(input_filename, argv[i+1], buffer_size);
700 else if(strcmp(argv[i], "-i") == 0) {
705 else if(strcmp(argv[i], "-b") == 0) {
710 else if(strcmp(argv[i], "-z") == 0) {
715 else if(strcmp(argv[i], "-d") == 0) {
716 remove_duplicates = 1;
720 else if(strcmp(argv[i], "-e") == 0) {
725 else if(strcmp(argv[i], "-a") == 0) {
730 else if(strcmp(argv[i], "-t") == 0) {
731 check_opt(argc, argv, i, 1, "<title>");
733 title = new char[strlen(argv[i+1]) + 1];
734 strcpy(title, argv[i+1]);
738 else if(strcmp(argv[i], "-l") == 0) {
739 check_opt(argc, argv, i, 1, "<maximum number of lines>");
740 nb_lines_max = string_to_positive_integer(argv[i+1]);
744 else if(strcmp(argv[i], "-c") == 0) {
745 check_opt(argc, argv, i, 4, "<fg modeline> <bg modeline> <fg highlight> <bg highlight>");
746 color_fg_modeline = string_to_positive_integer(argv[i + 1]);
747 color_bg_modeline = string_to_positive_integer(argv[i + 2]);
748 color_fg_highlight = string_to_positive_integer(argv[i + 3]);
749 color_bg_highlight = string_to_positive_integer(argv[i + 4]);
753 else if(strcmp(argv[i], "--") == 0) {
758 else if(strcmp(argv[i], "-h") == 0) {
764 cerr << "Unknown option " << argv[i] << "." << endl;
769 if(show_help || error) {
770 cerr << "Selector version " << VERSION << "-R" << REVISION_NUMBER
772 << "Written by Francois Fleuret <francois@fleuret.org>."
775 << "Usage: " << argv[0] << " [options] [<filename1> [<filename2> ...]]" << endl
777 << " -h show this help" << endl
778 << " -v inject the selected line in the tty" << endl
779 << " -d remove duplicated lines" << endl
780 << " -b remove the bash history line prefix" << endl
781 << " -z remove the zsh history line prefix" << endl
782 << " -i invert the order of lines" << endl
783 << " -e start in regexp mode" << endl
784 << " -a case sensitive" << endl
785 << " -m monochrome mode" << endl
786 << " -q make a flash instead of a beep on an edition error" << endl
787 << " -- rest of the arguments are filenames" << endl
788 << " -t <title>" << endl
789 << " add a title in the modeline" << endl
790 << " -c <fg modeline> <bg modeline> <fg highlight> <bg highlight>" << endl
791 << " set the display colors" << endl
792 << " -o <output filename>" << endl
793 << " set a file to write the selected line to" << endl
794 << " -s <pattern separator>" << endl
795 << " set the symbol to separate substrings in the pattern" << endl
796 << " -x <label separator>" << endl
797 << " set the symbol to terminate the label" << endl
798 << " -l <max number of lines>" << endl
799 << " set the maximum number of lines to take into account" << endl
805 char **lines = new char *[nb_lines_max];
808 int hash_table_size = nb_lines_max * 10;
811 if(remove_duplicates) {
812 hash_table = new_hash_table(hash_table_size);
815 if(input_filename[0]) {
816 read_file(input_filename,
817 nb_lines_max, &nb_lines, lines,
818 hash_table_size, hash_table);
823 nb_lines_max, &nb_lines, lines,
824 hash_table_size, hash_table);
830 // Now remove the null strings
833 for(int k = 0; k < nb_lines; k++) {
835 lines[n++] = lines[k];
842 for(int i = 0; i < nb_lines / 2; i++) {
843 char *s = lines[nb_lines - 1 - i];
844 lines[nb_lines - 1 - i] = lines[i];
849 // Build the labels from the strings, take only the part before the
850 // label_separator and transform control characters to printable
853 char **labels = new char *[nb_lines];
854 for(int l = 0; l < nb_lines; l++) {
859 while(*t && *t != label_separator) {
863 labels[l] = new char[e + 1];
866 while(*t && *t != label_separator) {
868 while(*u) { *s++ = *u++; }
873 char pattern[buffer_size];
879 //////////////////////////////////////////////////////////////////////
880 // Here we start to display with curse
886 // So that the arrow keys work
887 keypad(stdscr, TRUE);
895 if(color_fg_modeline < 0 || color_fg_modeline >= COLORS ||
896 color_bg_modeline < 0 || color_bg_modeline >= COLORS ||
897 color_fg_highlight < 0 || color_bg_highlight >= COLORS ||
898 color_bg_highlight < 0 || color_bg_highlight >= COLORS) {
901 cerr << "Color numbers have to be between 0 and " << COLORS - 1 << "." << endl;
905 init_pair(COLOR_MODELINE, color_fg_modeline, color_bg_modeline);
906 init_pair(COLOR_HIGHLIGHTED_LINE, color_fg_highlight, color_bg_highlight);
914 int current_line = 0, temporary_line = 0;
916 update_screen(¤t_line, &temporary_line, 0,
917 nb_lines, labels, cursor_position, pattern);
925 if(key >= ' ' && key <= '~') { // Insert character
926 insert_char(pattern, &cursor_position, key);
929 else if(key == KEY_BACKSPACE ||
930 key == '\010' || // ^H
931 key == '\177') { // ^?
932 backspace_char(pattern, &cursor_position);
935 else if(key == KEY_DC ||
936 key == '\004') { // ^D
937 delete_char(pattern, &cursor_position);
940 else if(key == KEY_HOME) {
944 else if(key == KEY_END) {
945 current_line = nb_lines - 1;
948 else if(key == KEY_NPAGE) {
952 else if(key == KEY_PPAGE) {
956 else if(key == KEY_DOWN ||
957 key == '\016') { // ^N
961 else if(key == KEY_UP ||
962 key == '\020') { // ^P
966 else if(key == KEY_LEFT ||
967 key == '\002') { // ^B
968 if(cursor_position > 0) cursor_position--;
969 else error_feedback();
972 else if(key == KEY_RIGHT ||
973 key == '\006') { // ^F
974 if(pattern[cursor_position]) cursor_position++;
975 else error_feedback();
978 else if(key == '\001') { // ^A
982 else if(key == '\005') { // ^E
983 cursor_position = strlen(pattern);
986 else if(key == '\022') { // ^R
987 use_regexp = !use_regexp;
990 else if(key == '\011') { // ^I
991 case_sensitive = !case_sensitive;
994 else if(key == '\025') { // ^U
995 kill_before_cursor(pattern, &cursor_position);
998 else if(key == '\013') { // ^K
999 kill_after_cursor(pattern, &cursor_position);
1002 else if(key == '\014') { // ^L
1003 // I suspect that we may sometime mess up the display
1007 update_screen(¤t_line, &temporary_line, motion,
1008 nb_lines, labels, cursor_position, pattern);
1010 } while(key != '\007' && // ^G
1011 key != '\033' && // ^[ (escape)
1018 //////////////////////////////////////////////////////////////////////
1019 // Here we come back to standard display
1021 if((key == KEY_ENTER || key == '\n')) {
1023 if(output_to_vt_buffer) {
1024 if(temporary_line >= 0 && temporary_line < nb_lines) {
1025 inject_into_tty_buffer(lines[temporary_line]);
1029 if(output_filename[0]) {
1030 ofstream out(output_filename);
1032 cerr << "Can not open " << output_filename << " for writing." << endl;
1035 if(temporary_line >= 0 && temporary_line < nb_lines) {
1036 out << lines[temporary_line] << endl;
1044 cout << "Aborted." << endl;
1047 for(int l = 0; l < nb_lines; l++) {