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 // Put the selected string in the tty input buffer
74 char control_q = '\021';
75 for(char *k = string; *k; k++) {
77 ioctl(STDIN_FILENO, TIOCSTI, &control_q);
79 ioctl(STDIN_FILENO, TIOCSTI, k);
81 // Restore the old settings
82 tcsetattr(STDIN_FILENO, TCSANOW, &oldtio);
85 //////////////////////////////////////////////////////////////////////
87 void check_opt(int argc, char **argv, int n_opt, int n, const char *help) {
88 if(n_opt + n >= argc) {
89 cerr << "Missing argument for " << argv[n_opt] << "."
91 << "Expecting " << help << "."
97 int string_to_positive_integer(char *string) {
102 for(char *s = string; *s; s++) {
103 if(*s >= '0' && *s <= '9') {
104 result = result * 10 + int(*s - '0');
110 cerr << "Value `" << string << "' is not a positive integer." << endl;
117 void error_feedback() {
125 //////////////////////////////////////////////////////////////////////
126 // A quick and dirty hash table
128 // The table itself stores index of the strings in a char
129 // **table. When a string is added, if it was already in the table,
130 // the new index replaces the previous one.
132 int *new_hash_table(int hash_table_size) {
134 result = new int[hash_table_size];
135 for(int k = 0; k < hash_table_size; k++) {
141 // Adds new_string in the table, associated to new_index. If this
142 // string was not already in the table, returns -1. Otherwise, returns
143 // the previous index it had.
145 int test_and_add(char *new_string, int new_index,
146 char **strings, int *hash_table, int hash_table_size) {
148 unsigned int code = 0;
150 // This is my recipe. I checked, it seems to work (as long as
151 // hash_table_size is not a multiple of 387433 that should be okay)
153 for(int k = 0; new_string[k]; k++) {
154 code = code * 387433 + (unsigned int) (new_string[k]);
157 code = code % hash_table_size;
159 while(hash_table[code] >= 0) {
160 // There is a string with that code
161 if(strcmp(new_string, strings[hash_table[code]]) == 0) {
162 // It is the same string, we keep a copy of the stored index
163 int result = hash_table[code];
165 hash_table[code] = new_index;
166 // And return the previous one
169 // This collision was not the same string, let's move to the next
171 code = (code + 1) % hash_table_size;
174 // This string was not already in there, store the index in the
175 // table and return -1
176 hash_table[code] = new_index;
180 //////////////////////////////////////////////////////////////////////
181 // A matcher matches either with a collection of substrings, or with a
189 char *splitted_patterns, **patterns;
192 int match(char *string, matcher_t *matcher) {
193 if(matcher->nb_patterns >= 0) {
194 if(matcher->case_sensitive) {
195 for(int n = 0; n < matcher->nb_patterns; n++) {
196 if(strstr(string, matcher->patterns[n]) == 0) return 0;
199 for(int n = 0; n < matcher->nb_patterns; n++) {
200 if(strcasestr(string, matcher->patterns[n]) == 0) return 0;
205 return regexec(&matcher->preg, string, 0, 0, 0) == 0;
209 void free_matcher(matcher_t *matcher) {
210 if(matcher->nb_patterns < 0) {
211 if(!matcher->regexp_error) regfree(&matcher->preg);
213 delete[] matcher->splitted_patterns;
214 delete[] matcher->patterns;
218 void initialize_matcher(int use_regexp, int case_sensitive,
219 matcher_t *matcher, const char *pattern) {
222 matcher->nb_patterns = -1;
223 matcher->regexp_error = regcomp(&matcher->preg, pattern, case_sensitive ? 0 : REG_ICASE);
225 matcher->regexp_error = 0;
226 matcher->nb_patterns = 1;
227 matcher->case_sensitive = case_sensitive;
229 for(const char *s = pattern; *s; s++) {
230 if(*s == pattern_separator) {
231 matcher->nb_patterns++;
235 matcher->splitted_patterns = new char[strlen(pattern) + 1];
236 matcher->patterns = new char*[matcher->nb_patterns];
238 strcpy(matcher->splitted_patterns, pattern);
241 char *last_pattern_start = matcher->splitted_patterns;
242 for(char *s = matcher->splitted_patterns; n < matcher->nb_patterns; s++) {
243 if(*s == pattern_separator || *s == '\0') {
245 matcher->patterns[n++] = last_pattern_start;
246 last_pattern_start = s + 1;
252 //////////////////////////////////////////////////////////////////////
255 void delete_char(char *buffer, int *position) {
256 if(buffer[*position]) {
258 while(c < buffer_size && buffer[c]) {
259 buffer[c] = buffer[c+1];
262 } else error_feedback();
265 void backspace_char(char *buffer, int *position) {
267 if(buffer[*position]) {
268 int c = *position - 1;
270 buffer[c] = buffer[c+1];
274 buffer[*position - 1] = '\0';
278 } else error_feedback();
281 void insert_char(char *buffer, int *position, char character) {
282 if(strlen(buffer) < buffer_size - 1) {
284 char t = buffer[c], u;
293 buffer[(*position)++] = character;
294 } else error_feedback();
297 void kill_before_cursor(char *buffer, int *position) {
299 while(buffer[*position + s]) {
300 buffer[s] = buffer[*position + s];
307 void kill_after_cursor(char *buffer, int *position) {
308 buffer[*position] = '\0';
311 //////////////////////////////////////////////////////////////////////
313 int previous_visible(int current_line, int nb_lines, char **lines, matcher_t *matcher) {
314 int line = current_line - 1;
315 while(line >= 0 && !match(lines[line], matcher)) line--;
319 int next_visible(int current_line, int nb_lines, char **lines, matcher_t *matcher) {
320 int line = current_line + 1;
321 while(line < nb_lines && !match(lines[line], matcher)) line++;
329 //////////////////////////////////////////////////////////////////////
331 void update_screen(int *current_line, int *temporary_line, int motion,
332 int nb_lines, char **lines,
336 char buffer[buffer_size];
339 initialize_matcher(use_regexp, case_sensitive, &matcher, pattern);
341 // We now take care of printing the lines per se
343 int console_width = getmaxx(stdscr);
344 int console_height = getmaxy(stdscr);
346 // First, we find a visible line. In priority: The current, or the
347 // first visible after it, or the first visible before it.
349 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) {
447 // Highlight the highlighted line ...
451 attron(COLOR_PAIR(COLOR_HIGHLIGHTED_LINE));
452 addnstr(buffer, console_width);
453 attroff(COLOR_PAIR(COLOR_HIGHLIGHTED_LINE));
456 addnstr(buffer, console_width);
460 addnstr(buffer, console_width);
468 *current_line = new_line;
472 *temporary_line = new_line;
474 if(nb_printed_lines == 0) {
475 addnstr("[no selection]\n", console_width);
478 addnstr("[empty choice]\n", console_width);
486 attron(COLOR_PAIR(COLOR_MODELINE));
491 for(int k = 0; k < console_width; k++) buffer[k] = ' ';
492 buffer[console_width] = '\0';
493 addnstr(buffer, console_width);
497 // There must be a more elegant way of moving the cursor at a
498 // location met during display
505 cursor_x += strlen(title) + 1;
508 sprintf(buffer, "%d/%d ", nb_printed_lines, nb_lines);
510 cursor_x += strlen(buffer);
512 addnstr(pattern, cursor_position);
513 cursor_x += cursor_position;
515 if(pattern[cursor_position]) {
516 addstr(pattern + cursor_position);
521 if(use_regexp || case_sensitive) {
539 attroff(COLOR_PAIR(COLOR_MODELINE));
547 free_matcher(&matcher);
550 //////////////////////////////////////////////////////////////////////
552 void read_file(const char *input_filename,
553 int nb_lines_max, int *nb_lines, char **lines,
554 int hash_table_size, int *hash_table) {
556 char raw_line[buffer_size];;
558 ifstream file(input_filename);
561 cerr << "Can not open " << input_filename << endl;
565 while(*nb_lines < nb_lines_max && !file.eof()) {
567 file.getline(raw_line, buffer_size);
572 cerr << "Line too long:" << endl;
573 cerr << raw_line << endl;
581 // Remove the zsh history prefix
583 if(zsh_history && *t == ':') {
584 while(*t && *t != ';') t++;
588 // Remove the bash history prefix
591 while(*t == ' ') t++;
592 while(*t >= '0' && *t <= '9') t++;
593 while(*t == ' ') t++;
596 // Check for duplicates with the hash table and insert the line
597 // in the list if necessary
602 dup = test_and_add(t, *nb_lines, lines, hash_table, hash_table_size);
608 lines[*nb_lines] = new char[strlen(t) + 1];
609 strcpy(lines[*nb_lines], t);
611 // The string was already in there, so we do not allocate a
612 // new string but use the pointer to the first occurence of it
613 lines[*nb_lines] = lines[dup];
622 //////////////////////////////////////////////////////////////////////
624 int main(int argc, char **argv) {
626 if(!ttyname(STDIN_FILENO)) {
627 cerr << "The standard input is not a tty." << endl;
631 int color_fg_modeline, color_bg_modeline;
632 int color_fg_highlight, color_bg_highlight;
634 color_fg_modeline = COLOR_WHITE;
635 color_bg_modeline = COLOR_BLACK;
636 color_fg_highlight = COLOR_BLACK;
637 color_bg_highlight = COLOR_YELLOW;
639 setlocale(LC_ALL, "");
641 char input_filename[buffer_size], output_filename[buffer_size];
643 strcpy(input_filename, "");
644 strcpy(output_filename, "");
647 int error = 0, show_help = 0;
648 int rest_are_files = 0;
650 while(!error && !show_help && i < argc && argv[i][0] == '-' && !rest_are_files) {
652 if(strcmp(argv[i], "-o") == 0) {
653 check_opt(argc, argv, i, 1, "<output filename>");
654 strncpy(output_filename, argv[i+1], buffer_size);
658 else if(strcmp(argv[i], "-s") == 0) {
659 check_opt(argc, argv, i, 1, "<pattern separator>");
660 pattern_separator = argv[i+1][0];
664 else if(strcmp(argv[i], "-x") == 0) {
665 check_opt(argc, argv, i, 1, "<label separator>");
666 label_separator = argv[i+1][0];
670 else if(strcmp(argv[i], "-v") == 0) {
671 output_to_vt_buffer = 1;
675 else if(strcmp(argv[i], "-w") == 0) {
680 else if(strcmp(argv[i], "-m") == 0) {
685 else if(strcmp(argv[i], "-q") == 0) {
690 else if(strcmp(argv[i], "-f") == 0) {
691 check_opt(argc, argv, i, 1, "<input filename>");
692 strncpy(input_filename, argv[i+1], buffer_size);
696 else if(strcmp(argv[i], "-i") == 0) {
701 else if(strcmp(argv[i], "-b") == 0) {
706 else if(strcmp(argv[i], "-z") == 0) {
711 else if(strcmp(argv[i], "-d") == 0) {
712 remove_duplicates = 1;
716 else if(strcmp(argv[i], "-e") == 0) {
721 else if(strcmp(argv[i], "-a") == 0) {
726 else if(strcmp(argv[i], "-t") == 0) {
727 check_opt(argc, argv, i, 1, "<title>");
729 title = new char[strlen(argv[i+1]) + 1];
730 strcpy(title, argv[i+1]);
734 else if(strcmp(argv[i], "-l") == 0) {
735 check_opt(argc, argv, i, 1, "<maximum number of lines>");
736 nb_lines_max = string_to_positive_integer(argv[i+1]);
740 else if(strcmp(argv[i], "-c") == 0) {
741 check_opt(argc, argv, i, 4, "<fg modeline> <bg modeline> <fg highlight> <bg highlight>");
742 color_fg_modeline = string_to_positive_integer(argv[i + 1]);
743 color_bg_modeline = string_to_positive_integer(argv[i + 2]);
744 color_fg_highlight = string_to_positive_integer(argv[i + 3]);
745 color_bg_highlight = string_to_positive_integer(argv[i + 4]);
749 else if(strcmp(argv[i], "--") == 0) {
754 else if(strcmp(argv[i], "-h") == 0) {
760 cerr << "Unknown option " << argv[i] << "." << endl;
765 if(show_help || error) {
766 cerr << "Selector version " << VERSION << "-R" << REVISION_NUMBER
768 << "Written by Francois Fleuret <francois@fleuret.org>."
771 << "Usage: " << argv[0] << " [options] [<filename1> [<filename2> ...]]" << endl
773 << " -h show this help" << endl
774 << " -v inject the selected line in the tty" << endl
775 << " -d remove duplicated lines" << endl
776 << " -b remove the bash history line prefix" << endl
777 << " -z remove the zsh history line prefix" << endl
778 << " -i invert the order of lines" << endl
779 << " -e start in regexp mode" << endl
780 << " -a case sensitive" << endl
781 << " -m monochrome mode" << endl
782 << " -q make a flash instead of a beep on an edition error" << endl
783 << " -- rest of the arguments are filenames" << endl
784 << " -t <title>" << endl
785 << " add a title in the modeline" << endl
786 << " -c <fg modeline> <bg modeline> <fg highlight> <bg highlight>" << endl
787 << " set the display colors" << endl
788 << " -o <output filename>" << endl
789 << " set a file to write the selected line to" << endl
790 << " -s <pattern separator>" << endl
791 << " set the symbol to separate substrings in the pattern" << endl
792 << " -x <label separator>" << endl
793 << " set the symbol to terminate the label" << endl
794 << " -l <max number of lines>" << endl
795 << " set the maximum number of lines to take into account" << endl
801 char **lines = new char *[nb_lines_max];
804 int hash_table_size = nb_lines_max * 10;
807 if(remove_duplicates) {
808 hash_table = new_hash_table(hash_table_size);
811 if(input_filename[0]) {
812 read_file(input_filename,
813 nb_lines_max, &nb_lines, lines,
814 hash_table_size, hash_table);
819 nb_lines_max, &nb_lines, lines,
820 hash_table_size, hash_table);
826 // Now remove the null strings
829 for(int k = 0; k < nb_lines; k++) {
831 lines[n++] = lines[k];
838 for(int i = 0; i < nb_lines / 2; i++) {
839 char *s = lines[nb_lines - 1 - i];
840 lines[nb_lines - 1 - i] = lines[i];
845 // Build the labels from the strings, take only the part before the
846 // label_separator and transform control characters to printable
849 char **labels = new char *[nb_lines];
850 for(int l = 0; l < nb_lines; l++) {
855 while(*t && *t != label_separator) {
859 labels[l] = new char[e + 1];
862 while(*t && *t != label_separator) {
864 while(*u) { *s++ = *u++; }
869 char pattern[buffer_size];
874 //////////////////////////////////////////////////////////////////////
875 // Here we start to display with curse
881 // So that the arrow keys work
882 keypad(stdscr, TRUE);
890 if(color_fg_modeline < 0 || color_fg_modeline >= COLORS ||
891 color_bg_modeline < 0 || color_bg_modeline >= COLORS ||
892 color_fg_highlight < 0 || color_bg_highlight >= COLORS ||
893 color_bg_highlight < 0 || color_bg_highlight >= COLORS) {
896 cerr << "Color numbers have to be between 0 and " << COLORS - 1 << "." << endl;
900 init_pair(COLOR_MODELINE, color_fg_modeline, color_bg_modeline);
901 init_pair(COLOR_HIGHLIGHTED_LINE, color_fg_highlight, color_bg_highlight);
909 int current_line = 0, temporary_line = 0;
911 update_screen(¤t_line, &temporary_line, 0, nb_lines, labels, cursor_position, pattern);
919 if(key >= ' ' && key <= '~') { // Insert character
920 insert_char(pattern, &cursor_position, key);
923 else if(key == KEY_BACKSPACE ||
924 key == '\010' || // ^H
925 key == '\177') { // ^?
926 backspace_char(pattern, &cursor_position);
929 else if(key == KEY_DC ||
930 key == '\004') { // ^D
931 delete_char(pattern, &cursor_position);
934 else if(key == KEY_HOME) {
938 else if(key == KEY_END) {
939 current_line = nb_lines - 1;
942 else if(key == KEY_NPAGE) {
946 else if(key == KEY_PPAGE) {
950 else if(key == KEY_DOWN ||
951 key == '\016') { // ^N
955 else if(key == KEY_UP ||
956 key == '\020') { // ^P
960 else if(key == KEY_LEFT ||
961 key == '\002') { // ^B
962 if(cursor_position > 0) cursor_position--;
963 else error_feedback();
966 else if(key == KEY_RIGHT ||
967 key == '\006') { // ^F
968 if(pattern[cursor_position]) cursor_position++;
969 else error_feedback();
972 else if(key == '\001') { // ^A
976 else if(key == '\005') { // ^E
977 cursor_position = strlen(pattern);
980 else if(key == '\022') { // ^R
981 use_regexp = !use_regexp;
984 else if(key == '\011') { // ^I
985 case_sensitive = !case_sensitive;
988 else if(key == '\025') { // ^U
989 kill_before_cursor(pattern, &cursor_position);
992 else if(key == '\013') { // ^K
993 kill_after_cursor(pattern, &cursor_position);
996 update_screen(¤t_line, &temporary_line, motion,
997 nb_lines, labels, cursor_position, pattern);
999 } while(key != '\007' && // ^G
1000 key != '\033' && // ^[ (escape)
1007 //////////////////////////////////////////////////////////////////////
1008 // Here we come back to standard display
1010 if((key == KEY_ENTER || key == '\n')) {
1012 if(output_to_vt_buffer) {
1013 if(temporary_line >= 0 && temporary_line < nb_lines) {
1014 inject_into_tty_buffer(lines[temporary_line]);
1018 if(output_filename[0]) {
1019 ofstream out(output_filename);
1021 cerr << "Can not open " << output_filename << " for writing." << endl;
1024 if(temporary_line >= 0 && temporary_line < nb_lines) {
1025 out << lines[temporary_line] << endl;
1033 cout << "Aborted." << endl;
1036 for(int l = 0; l < nb_lines; l++) {