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 // selector -q -b -i -d -v -w -l 10000 <(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;
56 int inverse_order = 0;
57 int remove_duplicates = 0;
59 int case_sensitive = 0;
63 #define COLOR_MODELINE 1
64 #define COLOR_HIGHLIGHTED_LINE 2
67 //////////////////////////////////////////////////////////////////////
69 void inject_into_tty_buffer(char *string) {
70 struct termios oldtio, newtio;
71 tcgetattr(STDIN_FILENO, &oldtio);
72 memset(&newtio, 0, sizeof(newtio));
73 // Set input mode (non-canonical, *no echo*,...)
74 tcsetattr(STDIN_FILENO, TCSANOW, &newtio);
75 const char control_q = '\021';
76 // Put the selected string in the tty input buffer
77 for(const char *k = string; *k; k++) {
78 if(add_control_qs && !(*k >= ' ' && *k <= '~')) {
79 // Add ^Q to quote control characters
80 ioctl(STDIN_FILENO, TIOCSTI, &control_q);
82 ioctl(STDIN_FILENO, TIOCSTI, k);
84 // Restore the old settings
85 tcsetattr(STDIN_FILENO, TCSANOW, &oldtio);
88 //////////////////////////////////////////////////////////////////////
90 void check_opt(int argc, char **argv, int n_opt, int n, const char *help) {
91 if(n_opt + n >= argc) {
92 cerr << "Missing argument for " << argv[n_opt] << "."
94 << "Expecting " << help << "."
100 int string_to_positive_integer(char *string) {
105 for(char *s = string; *s; s++) {
106 if(*s >= '0' && *s <= '9') {
107 result = result * 10 + int(*s - '0');
113 cerr << "Value `" << string << "' is not a positive integer." << endl;
120 void error_feedback() {
128 void print_error_message(const char *message, int width) {
130 attron(COLOR_PAIR(COLOR_ERROR));
131 addnstr(message, width);
132 attroff(COLOR_PAIR(COLOR_ERROR));
135 addnstr(message, width);
140 //////////////////////////////////////////////////////////////////////
141 // A quick and dirty hash table
143 // The table itself stores index of the strings in a char
144 // **table. When a string is added, if it was already in the table,
145 // the new index replaces the previous one.
147 int *new_hash_table(int hash_table_size) {
149 result = new int[hash_table_size];
150 for(int k = 0; k < hash_table_size; k++) {
156 // Adds new_string in the table, associated to new_index. If this
157 // string was not already in the table, returns -1. Otherwise, returns
158 // the previous index it had.
160 int test_and_add(char *new_string, int new_index,
161 char **strings, int *hash_table, int hash_table_size) {
163 unsigned int code = 0;
165 // This is my recipe. I checked, it seems to work (as long as
166 // hash_table_size is not a multiple of 387433 that should be okay)
168 for(int k = 0; new_string[k]; k++) {
169 code = code * 387433 + (unsigned int) (new_string[k]);
172 code = code % hash_table_size;
174 while(hash_table[code] >= 0) {
175 // There is a string with that code
176 if(strcmp(new_string, strings[hash_table[code]]) == 0) {
177 // It is the same string, we keep a copy of the stored index
178 int result = hash_table[code];
180 hash_table[code] = new_index;
181 // And return the previous one
184 // This collision was not the same string, let's move to the next
186 code = (code + 1) % hash_table_size;
189 // This string was not already in there, store the index in the
190 // table and return -1
191 hash_table[code] = new_index;
195 //////////////////////////////////////////////////////////////////////
196 // A matcher matches either with a collection of substrings, or with a
204 char *splitted_patterns, **patterns;
207 int match(char *string, matcher_t *matcher) {
208 if(matcher->nb_patterns >= 0) {
209 if(matcher->case_sensitive) {
210 for(int n = 0; n < matcher->nb_patterns; n++) {
211 if(strstr(string, matcher->patterns[n]) == 0) return 0;
214 for(int n = 0; n < matcher->nb_patterns; n++) {
215 if(strcasestr(string, matcher->patterns[n]) == 0) return 0;
220 return regexec(&matcher->preg, string, 0, 0, 0) == 0;
224 void free_matcher(matcher_t *matcher) {
225 if(matcher->nb_patterns < 0) {
226 if(!matcher->regexp_error) regfree(&matcher->preg);
228 delete[] matcher->splitted_patterns;
229 delete[] matcher->patterns;
233 void initialize_matcher(int use_regexp, int case_sensitive,
234 matcher_t *matcher, const char *pattern) {
237 matcher->nb_patterns = -1;
238 matcher->regexp_error = regcomp(&matcher->preg, pattern, case_sensitive ? 0 : REG_ICASE);
240 matcher->regexp_error = 0;
241 matcher->nb_patterns = 1;
242 matcher->case_sensitive = case_sensitive;
244 for(const char *s = pattern; *s; s++) {
245 if(*s == pattern_separator) {
246 matcher->nb_patterns++;
250 matcher->splitted_patterns = new char[strlen(pattern) + 1];
251 matcher->patterns = new char*[matcher->nb_patterns];
253 strcpy(matcher->splitted_patterns, pattern);
256 char *last_pattern_start = matcher->splitted_patterns;
257 for(char *s = matcher->splitted_patterns; n < matcher->nb_patterns; s++) {
258 if(*s == pattern_separator || *s == '\0') {
260 matcher->patterns[n++] = last_pattern_start;
261 last_pattern_start = s + 1;
267 //////////////////////////////////////////////////////////////////////
270 void delete_char(char *buffer, int *position) {
271 if(buffer[*position]) {
273 while(c < buffer_size && buffer[c]) {
274 buffer[c] = buffer[c+1];
277 } else error_feedback();
280 void backspace_char(char *buffer, int *position) {
282 if(buffer[*position]) {
283 int c = *position - 1;
285 buffer[c] = buffer[c+1];
289 buffer[*position - 1] = '\0';
293 } else error_feedback();
296 void insert_char(char *buffer, int *position, char character) {
297 if(strlen(buffer) < buffer_size - 1) {
299 char t = buffer[c], u;
308 buffer[(*position)++] = character;
309 } else error_feedback();
312 void kill_before_cursor(char *buffer, int *position) {
314 while(buffer[*position + s]) {
315 buffer[s] = buffer[*position + s];
322 void kill_after_cursor(char *buffer, int *position) {
323 buffer[*position] = '\0';
326 //////////////////////////////////////////////////////////////////////
328 int previous_visible(int current_line, int nb_lines, char **lines, matcher_t *matcher) {
329 int line = current_line - 1;
330 while(line >= 0 && !match(lines[line], matcher)) line--;
334 int next_visible(int current_line, int nb_lines, char **lines, matcher_t *matcher) {
335 int line = current_line + 1;
336 while(line < nb_lines && !match(lines[line], matcher)) line++;
344 //////////////////////////////////////////////////////////////////////
346 void update_screen(int *current_line, int *temporary_line, int motion,
347 int nb_lines, char **lines,
351 char buffer[buffer_size];
354 initialize_matcher(use_regexp, case_sensitive, &matcher, pattern);
356 // We now take care of printing the lines per se
358 int console_width = getmaxx(stdscr);
359 int console_height = getmaxy(stdscr);
361 // First, we find a visible line. In priority: The current, or the
362 // first visible after it, or the first visible before it.
364 int nb_printed_lines = 0;
366 use_default_colors();
369 if(matcher.regexp_error) {
370 print_error_message("[regexp error]", console_width);
371 } else if(nb_lines > 0) {
373 if(match(lines[*current_line], &matcher)) {
374 new_line = *current_line;
376 new_line = next_visible(*current_line, nb_lines, lines, &matcher);
378 new_line = previous_visible(*current_line, nb_lines, lines, &matcher);
382 // If we found a visible line and we should move, let's move
384 if(new_line >= 0 && motion != 0) {
387 // We want to go down, let's find the first visible line below
388 for(int m = 0; l >= 0 && m < motion; m++) {
389 l = next_visible(l, nb_lines, lines, &matcher);
395 // We want to go up, let's find the first visible line above
396 for(int m = 0; l >= 0 && m < -motion; m++) {
397 l = previous_visible(l, nb_lines, lines, &matcher);
405 // Here new_line is either a line number matching the patterns, or -1
409 int first_line = new_line, last_line = new_line, nb_match = 1;
411 // We find the first and last line to show, so that the total of
412 // visible lines between them (them included) is console_height-1
414 while(nb_match < console_height-1 && (first_line > 0 || last_line < nb_lines - 1)) {
418 while(first_line > 0 && !match(lines[first_line], &matcher)) {
421 if(match(lines[first_line], &matcher)) {
426 if(nb_match < console_height - 1 && last_line < nb_lines - 1) {
428 while(last_line < nb_lines - 1 && !match(lines[last_line], &matcher)) {
432 if(match(lines[last_line], &matcher)) {
438 // Now we display them
440 for(int l = first_line; l <= last_line; l++) {
441 if(match(lines[l], &matcher)) {
444 while(lines[l][k] && k < buffer_size - 2 && k < console_width - 2) {
445 buffer[k] = lines[l][k];
449 // We fill the rest of the line with blanks if this is the
453 while(k < console_width) {
463 // Highlight the highlighted line ...
467 attron(COLOR_PAIR(COLOR_HIGHLIGHTED_LINE));
468 addnstr(buffer, console_width);
469 attroff(COLOR_PAIR(COLOR_HIGHLIGHTED_LINE));
472 addnstr(buffer, console_width);
476 addnstr(buffer, console_width);
484 *current_line = new_line;
488 *temporary_line = new_line;
490 if(nb_printed_lines == 0) {
491 print_error_message("[no selection]\n", console_width);
494 print_error_message("[empty choice]\n", console_width);
504 attron(COLOR_PAIR(COLOR_MODELINE));
509 for(int k = 0; k < console_width; k++) buffer[k] = ' ';
510 buffer[console_width] = '\0';
511 addnstr(buffer, console_width);
515 // There must be a more elegant way of moving the cursor at a
516 // location met during display
523 cursor_x += strlen(title) + 1;
526 sprintf(buffer, "%d/%d ", nb_printed_lines, nb_lines);
528 cursor_x += strlen(buffer);
530 addnstr(pattern, cursor_position);
531 cursor_x += cursor_position;
533 if(pattern[cursor_position]) {
534 addstr(pattern + cursor_position);
539 if(use_regexp || case_sensitive) {
557 attroff(COLOR_PAIR(COLOR_MODELINE));
565 free_matcher(&matcher);
568 //////////////////////////////////////////////////////////////////////
570 void read_file(const char *input_filename,
571 int nb_lines_max, int *nb_lines, char **lines,
572 int hash_table_size, int *hash_table) {
574 char raw_line[buffer_size];;
576 ifstream file(input_filename);
579 cerr << "Can not open " << input_filename << endl;
583 while(*nb_lines < nb_lines_max && !file.eof()) {
585 file.getline(raw_line, buffer_size);
590 cerr << "Line too long:" << endl;
591 cerr << raw_line << endl;
599 // Remove the zsh history prefix
601 if(zsh_history && *t == ':') {
602 while(*t && *t != ';') t++;
606 // Remove the bash history prefix
609 while(*t == ' ') t++;
610 while(*t >= '0' && *t <= '9') t++;
611 while(*t == ' ') t++;
614 // Check for duplicates with the hash table and insert the line
615 // in the list if necessary
620 dup = test_and_add(t, *nb_lines, lines, hash_table, hash_table_size);
626 lines[*nb_lines] = new char[strlen(t) + 1];
627 strcpy(lines[*nb_lines], t);
629 // The string was already in there, so we do not allocate a
630 // new string but use the pointer to the first occurence of it
631 lines[*nb_lines] = lines[dup];
640 //////////////////////////////////////////////////////////////////////
642 int main(int argc, char **argv) {
644 if(!ttyname(STDIN_FILENO)) {
645 cerr << "The standard input is not a tty." << endl;
649 int color_fg_modeline, color_bg_modeline;
650 int color_fg_highlight, color_bg_highlight;
652 color_fg_modeline = COLOR_WHITE;
653 color_bg_modeline = COLOR_BLACK;
654 color_fg_highlight = COLOR_BLACK;
655 color_bg_highlight = COLOR_YELLOW;
657 setlocale(LC_ALL, "");
659 char input_filename[buffer_size], output_filename[buffer_size];
661 strcpy(input_filename, "");
662 strcpy(output_filename, "");
665 int error = 0, show_help = 0;
666 int rest_are_files = 0;
668 while(!error && !show_help && i < argc && argv[i][0] == '-' && !rest_are_files) {
670 if(strcmp(argv[i], "-o") == 0) {
671 check_opt(argc, argv, i, 1, "<output filename>");
672 strncpy(output_filename, argv[i+1], buffer_size);
676 else if(strcmp(argv[i], "-s") == 0) {
677 check_opt(argc, argv, i, 1, "<pattern separator>");
678 pattern_separator = argv[i+1][0];
682 else if(strcmp(argv[i], "-x") == 0) {
683 check_opt(argc, argv, i, 1, "<label separator>");
684 label_separator = argv[i+1][0];
688 else if(strcmp(argv[i], "-v") == 0) {
689 output_to_vt_buffer = 1;
693 else if(strcmp(argv[i], "-w") == 0) {
698 else if(strcmp(argv[i], "-m") == 0) {
703 else if(strcmp(argv[i], "-q") == 0) {
708 else if(strcmp(argv[i], "-f") == 0) {
709 check_opt(argc, argv, i, 1, "<input filename>");
710 strncpy(input_filename, argv[i+1], buffer_size);
714 else if(strcmp(argv[i], "-i") == 0) {
719 else if(strcmp(argv[i], "-b") == 0) {
724 else if(strcmp(argv[i], "-z") == 0) {
729 else if(strcmp(argv[i], "-d") == 0) {
730 remove_duplicates = 1;
734 else if(strcmp(argv[i], "-e") == 0) {
739 else if(strcmp(argv[i], "-a") == 0) {
744 else if(strcmp(argv[i], "-t") == 0) {
745 check_opt(argc, argv, i, 1, "<title>");
747 title = new char[strlen(argv[i+1]) + 1];
748 strcpy(title, argv[i+1]);
752 else if(strcmp(argv[i], "-l") == 0) {
753 check_opt(argc, argv, i, 1, "<maximum number of lines>");
754 nb_lines_max = string_to_positive_integer(argv[i+1]);
758 else if(strcmp(argv[i], "-c") == 0) {
759 check_opt(argc, argv, i, 4, "<fg modeline> <bg modeline> <fg highlight> <bg highlight>");
760 color_fg_modeline = string_to_positive_integer(argv[i + 1]);
761 color_bg_modeline = string_to_positive_integer(argv[i + 2]);
762 color_fg_highlight = string_to_positive_integer(argv[i + 3]);
763 color_bg_highlight = string_to_positive_integer(argv[i + 4]);
767 else if(strcmp(argv[i], "--") == 0) {
772 else if(strcmp(argv[i], "-h") == 0) {
778 cerr << "Unknown option " << argv[i] << "." << endl;
783 if(show_help || error) {
784 cerr << "Selector version " << VERSION << "-R" << REVISION_NUMBER
786 << "Written by Francois Fleuret <francois@fleuret.org>."
789 << "Usage: " << argv[0] << " [options] [<filename1> [<filename2> ...]]" << endl
791 << " -h show this help" << endl
792 << " -v inject the selected line in the tty" << endl
793 << " -w quote control characters with ^Qs when using -v" << endl
794 << " -d remove duplicated lines" << endl
795 << " -b remove the bash history line prefix" << endl
796 << " -z remove the zsh history line prefix" << endl
797 << " -i invert the order of lines" << endl
798 << " -e start in regexp mode" << endl
799 << " -a case sensitive" << endl
800 << " -m monochrome mode" << endl
801 << " -q make a flash instead of a beep on an edition error" << endl
802 << " -- all following arguments are filenames" << endl
803 << " -t <title>" << endl
804 << " add a title in the modeline" << endl
805 << " -c <fg modeline> <bg modeline> <fg highlight> <bg highlight>" << endl
806 << " set the display colors" << endl
807 << " -o <output filename>" << endl
808 << " set a file to write the selected line to" << endl
809 << " -s <pattern separator>" << endl
810 << " set the symbol to separate substrings in the pattern" << endl
811 << " -x <label separator>" << endl
812 << " set the symbol to terminate the label" << endl
813 << " -l <max number of lines>" << endl
814 << " set the maximum number of lines to take into account" << endl
820 char **lines = new char *[nb_lines_max];
823 int hash_table_size = nb_lines_max * 10;
826 if(remove_duplicates) {
827 hash_table = new_hash_table(hash_table_size);
830 if(input_filename[0]) {
831 read_file(input_filename,
832 nb_lines_max, &nb_lines, lines,
833 hash_table_size, hash_table);
838 nb_lines_max, &nb_lines, lines,
839 hash_table_size, hash_table);
845 // Now remove the null strings
848 for(int k = 0; k < nb_lines; k++) {
850 lines[n++] = lines[k];
857 for(int i = 0; i < nb_lines / 2; i++) {
858 char *s = lines[nb_lines - 1 - i];
859 lines[nb_lines - 1 - i] = lines[i];
864 // Build the labels from the strings, take only the part before the
865 // label_separator and transform control characters to printable
868 char **labels = new char *[nb_lines];
869 for(int l = 0; l < nb_lines; l++) {
874 while(*t && *t != label_separator) {
878 labels[l] = new char[e + 1];
881 while(*t && *t != label_separator) {
883 while(*u) { *s++ = *u++; }
888 char pattern[buffer_size];
894 //////////////////////////////////////////////////////////////////////
895 // Here we start to display with curse
901 // So that the arrow keys work
902 keypad(stdscr, TRUE);
910 if(color_fg_modeline < 0 || color_fg_modeline >= COLORS ||
911 color_bg_modeline < 0 || color_bg_modeline >= COLORS ||
912 color_fg_highlight < 0 || color_bg_highlight >= COLORS ||
913 color_bg_highlight < 0 || color_bg_highlight >= COLORS) {
916 cerr << "Color numbers have to be between 0 and " << COLORS - 1 << "." << endl;
920 init_pair(COLOR_MODELINE, color_fg_modeline, color_bg_modeline);
921 init_pair(COLOR_HIGHLIGHTED_LINE, color_fg_highlight, color_bg_highlight);
922 init_pair(COLOR_ERROR, COLOR_WHITE, COLOR_RED);
930 int current_line = 0, temporary_line = 0;
932 update_screen(¤t_line, &temporary_line, 0,
933 nb_lines, labels, cursor_position, pattern);
941 if(key >= ' ' && key <= '~') { // Insert character
942 insert_char(pattern, &cursor_position, key);
945 else if(key == KEY_BACKSPACE ||
946 key == '\010' || // ^H
947 key == '\177') { // ^?
948 backspace_char(pattern, &cursor_position);
951 else if(key == KEY_DC ||
952 key == '\004') { // ^D
953 delete_char(pattern, &cursor_position);
956 else if(key == KEY_HOME) {
960 else if(key == KEY_END) {
961 current_line = nb_lines - 1;
964 else if(key == KEY_NPAGE) {
968 else if(key == KEY_PPAGE) {
972 else if(key == KEY_DOWN ||
973 key == '\016') { // ^N
977 else if(key == KEY_UP ||
978 key == '\020') { // ^P
982 else if(key == KEY_LEFT ||
983 key == '\002') { // ^B
984 if(cursor_position > 0) cursor_position--;
985 else error_feedback();
988 else if(key == KEY_RIGHT ||
989 key == '\006') { // ^F
990 if(pattern[cursor_position]) cursor_position++;
991 else error_feedback();
994 else if(key == '\001') { // ^A
998 else if(key == '\005') { // ^E
999 cursor_position = strlen(pattern);
1002 else if(key == '\022') { // ^R
1003 use_regexp = !use_regexp;
1006 else if(key == '\011') { // ^I
1007 case_sensitive = !case_sensitive;
1010 else if(key == '\025') { // ^U
1011 kill_before_cursor(pattern, &cursor_position);
1014 else if(key == '\013') { // ^K
1015 kill_after_cursor(pattern, &cursor_position);
1018 else if(key == '\014') { // ^L
1019 // I suspect that we may sometime mess up the display
1023 update_screen(¤t_line, &temporary_line, motion,
1024 nb_lines, labels, cursor_position, pattern);
1026 } while(key != '\007' && // ^G
1027 key != '\033' && // ^[ (escape)
1034 //////////////////////////////////////////////////////////////////////
1035 // Here we come back to standard display
1037 if((key == KEY_ENTER || key == '\n')) {
1039 if(output_to_vt_buffer) {
1040 if(temporary_line >= 0 && temporary_line < nb_lines) {
1041 inject_into_tty_buffer(lines[temporary_line]);
1045 if(output_filename[0]) {
1046 ofstream out(output_filename);
1048 cerr << "Can not open " << output_filename << " for writing." << endl;
1051 if(temporary_line >= 0 && temporary_line < nb_lines) {
1052 out << lines[temporary_line] << endl;
1060 cout << "Aborted." << endl;
1063 for(int l = 0; l < nb_lines; l++) {